diff options
Diffstat (limited to 'drivers/crypto/intel')
196 files changed, 51226 insertions, 0 deletions
diff --git a/drivers/crypto/intel/Kconfig b/drivers/crypto/intel/Kconfig new file mode 100644 index 000000000000..f38cd62a3f67 --- /dev/null +++ b/drivers/crypto/intel/Kconfig @@ -0,0 +1,6 @@ +# SPDX-License-Identifier: GPL-2.0 + +source "drivers/crypto/intel/keembay/Kconfig" +source "drivers/crypto/intel/ixp4xx/Kconfig" +source "drivers/crypto/intel/qat/Kconfig" +source "drivers/crypto/intel/iaa/Kconfig" diff --git a/drivers/crypto/intel/Makefile b/drivers/crypto/intel/Makefile new file mode 100644 index 000000000000..2f56f6d34cf0 --- /dev/null +++ b/drivers/crypto/intel/Makefile @@ -0,0 +1,6 @@ +# SPDX-License-Identifier: GPL-2.0 + +obj-y += keembay/ +obj-y += ixp4xx/ +obj-$(CONFIG_CRYPTO_DEV_QAT) += qat/ +obj-$(CONFIG_CRYPTO_DEV_IAA_CRYPTO) += iaa/ diff --git a/drivers/crypto/intel/iaa/Kconfig b/drivers/crypto/intel/iaa/Kconfig new file mode 100644 index 000000000000..d53f4b1d494f --- /dev/null +++ b/drivers/crypto/intel/iaa/Kconfig @@ -0,0 +1,19 @@ +config CRYPTO_DEV_IAA_CRYPTO + tristate "Support for Intel(R) IAA Compression Accelerator" + depends on CRYPTO_DEFLATE + depends on INTEL_IDXD + default n + help + This driver supports acceleration for compression and + decompression with the Intel Analytics Accelerator (IAA) + hardware using the cryptographic API. If you choose 'M' + here, the module will be called iaa_crypto. + +config CRYPTO_DEV_IAA_CRYPTO_STATS + bool "Enable Intel(R) IAA Compression Accelerator Statistics" + depends on CRYPTO_DEV_IAA_CRYPTO + default n + help + Enable statistics for the IAA compression accelerator. + These include per-device and per-workqueue statistics in + addition to global driver statistics. diff --git a/drivers/crypto/intel/iaa/Makefile b/drivers/crypto/intel/iaa/Makefile new file mode 100644 index 000000000000..55bda7770fac --- /dev/null +++ b/drivers/crypto/intel/iaa/Makefile @@ -0,0 +1,12 @@ +# SPDX-License-Identifier: GPL-2.0 +# +# Makefile for IAA crypto device drivers +# + +ccflags-y += -I $(srctree)/drivers/dma/idxd -DDEFAULT_SYMBOL_NAMESPACE='"IDXD"' + +obj-$(CONFIG_CRYPTO_DEV_IAA_CRYPTO) := iaa_crypto.o + +iaa_crypto-y := iaa_crypto_main.o iaa_crypto_comp_fixed.o + +iaa_crypto-$(CONFIG_CRYPTO_DEV_IAA_CRYPTO_STATS) += iaa_crypto_stats.o diff --git a/drivers/crypto/intel/iaa/iaa_crypto.h b/drivers/crypto/intel/iaa/iaa_crypto.h new file mode 100644 index 000000000000..56985e395263 --- /dev/null +++ b/drivers/crypto/intel/iaa/iaa_crypto.h @@ -0,0 +1,148 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Copyright(c) 2021 Intel Corporation. All rights rsvd. */ + +#ifndef __IAA_CRYPTO_H__ +#define __IAA_CRYPTO_H__ + +#include <linux/crypto.h> +#include <linux/idxd.h> +#include <uapi/linux/idxd.h> + +#define IDXD_SUBDRIVER_NAME "crypto" + +#define IAA_DECOMP_ENABLE BIT(0) +#define IAA_DECOMP_FLUSH_OUTPUT BIT(1) +#define IAA_DECOMP_CHECK_FOR_EOB BIT(2) +#define IAA_DECOMP_STOP_ON_EOB BIT(3) +#define IAA_DECOMP_SUPPRESS_OUTPUT BIT(9) + +#define IAA_COMP_FLUSH_OUTPUT BIT(1) +#define IAA_COMP_APPEND_EOB BIT(2) + +#define IAA_COMPLETION_TIMEOUT 1000000 + +#define IAA_ANALYTICS_ERROR 0x0a +#define IAA_ERROR_DECOMP_BUF_OVERFLOW 0x0b +#define IAA_ERROR_COMP_BUF_OVERFLOW 0x19 +#define IAA_ERROR_WATCHDOG_EXPIRED 0x24 + +#define IAA_COMP_MODES_MAX 2 + +#define FIXED_HDR 0x2 +#define FIXED_HDR_SIZE 3 + +#define IAA_COMP_FLAGS (IAA_COMP_FLUSH_OUTPUT | \ + IAA_COMP_APPEND_EOB) + +#define IAA_DECOMP_FLAGS (IAA_DECOMP_ENABLE | \ + IAA_DECOMP_FLUSH_OUTPUT | \ + IAA_DECOMP_CHECK_FOR_EOB | \ + IAA_DECOMP_STOP_ON_EOB) + +/* Representation of IAA workqueue */ +struct iaa_wq { + struct list_head list; + + struct idxd_wq *wq; + int ref; + bool remove; + + struct iaa_device *iaa_device; + + atomic64_t comp_calls; + atomic64_t comp_bytes; + atomic64_t decomp_calls; + atomic64_t decomp_bytes; +}; + +struct iaa_device_compression_mode { + const char *name; + + struct aecs_comp_table_record *aecs_comp_table; + + dma_addr_t aecs_comp_table_dma_addr; +}; + +/* Representation of IAA device with wqs, populated by probe */ +struct iaa_device { + struct list_head list; + struct idxd_device *idxd; + + struct iaa_device_compression_mode *compression_modes[IAA_COMP_MODES_MAX]; + + int n_wq; + struct list_head wqs; + + atomic64_t comp_calls; + atomic64_t comp_bytes; + atomic64_t decomp_calls; + atomic64_t decomp_bytes; +}; + +struct wq_table_entry { + struct idxd_wq **wqs; + int max_wqs; + int n_wqs; + int cur_wq; +}; + +#define IAA_AECS_ALIGN 32 + +/* + * Analytics Engine Configuration and State (AECS) contains parameters and + * internal state of the analytics engine. + */ +struct aecs_comp_table_record { + u32 crc; + u32 xor_checksum; + u32 reserved0[5]; + u32 num_output_accum_bits; + u8 output_accum[256]; + u32 ll_sym[286]; + u32 reserved1; + u32 reserved2; + u32 d_sym[30]; + u32 reserved_padding[2]; +} __packed; + +int iaa_aecs_init_fixed(void); +void iaa_aecs_cleanup_fixed(void); + +typedef int (*iaa_dev_comp_init_fn_t) (struct iaa_device_compression_mode *mode); +typedef int (*iaa_dev_comp_free_fn_t) (struct iaa_device_compression_mode *mode); + +struct iaa_compression_mode { + const char *name; + u32 *ll_table; + int ll_table_size; + u32 *d_table; + int d_table_size; + iaa_dev_comp_init_fn_t init; + iaa_dev_comp_free_fn_t free; +}; + +int add_iaa_compression_mode(const char *name, + const u32 *ll_table, + int ll_table_size, + const u32 *d_table, + int d_table_size, + iaa_dev_comp_init_fn_t init, + iaa_dev_comp_free_fn_t free); + +void remove_iaa_compression_mode(const char *name); + +enum iaa_mode { + IAA_MODE_FIXED, +}; + +struct iaa_compression_ctx { + enum iaa_mode mode; + bool verify_compress; + bool async_mode; + bool use_irq; +}; + +extern struct list_head iaa_devices; +extern struct mutex iaa_devices_lock; + +#endif diff --git a/drivers/crypto/intel/iaa/iaa_crypto_comp_fixed.c b/drivers/crypto/intel/iaa/iaa_crypto_comp_fixed.c new file mode 100644 index 000000000000..19d9a333ac49 --- /dev/null +++ b/drivers/crypto/intel/iaa/iaa_crypto_comp_fixed.c @@ -0,0 +1,91 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright(c) 2021 Intel Corporation. All rights rsvd. */ + +#include "idxd.h" +#include "iaa_crypto.h" + +/* + * Fixed Huffman tables the IAA hardware requires to implement RFC-1951. + */ +static const u32 fixed_ll_sym[286] = { + 0x40030, 0x40031, 0x40032, 0x40033, 0x40034, 0x40035, 0x40036, 0x40037, + 0x40038, 0x40039, 0x4003A, 0x4003B, 0x4003C, 0x4003D, 0x4003E, 0x4003F, + 0x40040, 0x40041, 0x40042, 0x40043, 0x40044, 0x40045, 0x40046, 0x40047, + 0x40048, 0x40049, 0x4004A, 0x4004B, 0x4004C, 0x4004D, 0x4004E, 0x4004F, + 0x40050, 0x40051, 0x40052, 0x40053, 0x40054, 0x40055, 0x40056, 0x40057, + 0x40058, 0x40059, 0x4005A, 0x4005B, 0x4005C, 0x4005D, 0x4005E, 0x4005F, + 0x40060, 0x40061, 0x40062, 0x40063, 0x40064, 0x40065, 0x40066, 0x40067, + 0x40068, 0x40069, 0x4006A, 0x4006B, 0x4006C, 0x4006D, 0x4006E, 0x4006F, + 0x40070, 0x40071, 0x40072, 0x40073, 0x40074, 0x40075, 0x40076, 0x40077, + 0x40078, 0x40079, 0x4007A, 0x4007B, 0x4007C, 0x4007D, 0x4007E, 0x4007F, + 0x40080, 0x40081, 0x40082, 0x40083, 0x40084, 0x40085, 0x40086, 0x40087, + 0x40088, 0x40089, 0x4008A, 0x4008B, 0x4008C, 0x4008D, 0x4008E, 0x4008F, + 0x40090, 0x40091, 0x40092, 0x40093, 0x40094, 0x40095, 0x40096, 0x40097, + 0x40098, 0x40099, 0x4009A, 0x4009B, 0x4009C, 0x4009D, 0x4009E, 0x4009F, + 0x400A0, 0x400A1, 0x400A2, 0x400A3, 0x400A4, 0x400A5, 0x400A6, 0x400A7, + 0x400A8, 0x400A9, 0x400AA, 0x400AB, 0x400AC, 0x400AD, 0x400AE, 0x400AF, + 0x400B0, 0x400B1, 0x400B2, 0x400B3, 0x400B4, 0x400B5, 0x400B6, 0x400B7, + 0x400B8, 0x400B9, 0x400BA, 0x400BB, 0x400BC, 0x400BD, 0x400BE, 0x400BF, + 0x48190, 0x48191, 0x48192, 0x48193, 0x48194, 0x48195, 0x48196, 0x48197, + 0x48198, 0x48199, 0x4819A, 0x4819B, 0x4819C, 0x4819D, 0x4819E, 0x4819F, + 0x481A0, 0x481A1, 0x481A2, 0x481A3, 0x481A4, 0x481A5, 0x481A6, 0x481A7, + 0x481A8, 0x481A9, 0x481AA, 0x481AB, 0x481AC, 0x481AD, 0x481AE, 0x481AF, + 0x481B0, 0x481B1, 0x481B2, 0x481B3, 0x481B4, 0x481B5, 0x481B6, 0x481B7, + 0x481B8, 0x481B9, 0x481BA, 0x481BB, 0x481BC, 0x481BD, 0x481BE, 0x481BF, + 0x481C0, 0x481C1, 0x481C2, 0x481C3, 0x481C4, 0x481C5, 0x481C6, 0x481C7, + 0x481C8, 0x481C9, 0x481CA, 0x481CB, 0x481CC, 0x481CD, 0x481CE, 0x481CF, + 0x481D0, 0x481D1, 0x481D2, 0x481D3, 0x481D4, 0x481D5, 0x481D6, 0x481D7, + 0x481D8, 0x481D9, 0x481DA, 0x481DB, 0x481DC, 0x481DD, 0x481DE, 0x481DF, + 0x481E0, 0x481E1, 0x481E2, 0x481E3, 0x481E4, 0x481E5, 0x481E6, 0x481E7, + 0x481E8, 0x481E9, 0x481EA, 0x481EB, 0x481EC, 0x481ED, 0x481EE, 0x481EF, + 0x481F0, 0x481F1, 0x481F2, 0x481F3, 0x481F4, 0x481F5, 0x481F6, 0x481F7, + 0x481F8, 0x481F9, 0x481FA, 0x481FB, 0x481FC, 0x481FD, 0x481FE, 0x481FF, + 0x38000, 0x38001, 0x38002, 0x38003, 0x38004, 0x38005, 0x38006, 0x38007, + 0x38008, 0x38009, 0x3800A, 0x3800B, 0x3800C, 0x3800D, 0x3800E, 0x3800F, + 0x38010, 0x38011, 0x38012, 0x38013, 0x38014, 0x38015, 0x38016, 0x38017, + 0x400C0, 0x400C1, 0x400C2, 0x400C3, 0x400C4, 0x400C5 +}; + +static const u32 fixed_d_sym[30] = { + 0x28000, 0x28001, 0x28002, 0x28003, 0x28004, 0x28005, 0x28006, 0x28007, + 0x28008, 0x28009, 0x2800A, 0x2800B, 0x2800C, 0x2800D, 0x2800E, 0x2800F, + 0x28010, 0x28011, 0x28012, 0x28013, 0x28014, 0x28015, 0x28016, 0x28017, + 0x28018, 0x28019, 0x2801A, 0x2801B, 0x2801C, 0x2801D +}; + +static int init_fixed_mode(struct iaa_device_compression_mode *mode) +{ + struct aecs_comp_table_record *comp_table = mode->aecs_comp_table; + u32 bfinal = 1; + u32 offset; + + /* Configure aecs table using fixed Huffman table */ + comp_table->crc = 0; + comp_table->xor_checksum = 0; + offset = comp_table->num_output_accum_bits / 8; + comp_table->output_accum[offset] = FIXED_HDR | bfinal; + comp_table->num_output_accum_bits = FIXED_HDR_SIZE; + + return 0; +} + +int iaa_aecs_init_fixed(void) +{ + int ret; + + ret = add_iaa_compression_mode("fixed", + fixed_ll_sym, + sizeof(fixed_ll_sym), + fixed_d_sym, + sizeof(fixed_d_sym), + init_fixed_mode, NULL); + if (!ret) + pr_debug("IAA fixed compression mode initialized\n"); + + return ret; +} + +void iaa_aecs_cleanup_fixed(void) +{ + remove_iaa_compression_mode("fixed"); +} diff --git a/drivers/crypto/intel/iaa/iaa_crypto_main.c b/drivers/crypto/intel/iaa/iaa_crypto_main.c new file mode 100644 index 000000000000..d0058757b000 --- /dev/null +++ b/drivers/crypto/intel/iaa/iaa_crypto_main.c @@ -0,0 +1,1949 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright(c) 2021 Intel Corporation. All rights rsvd. */ + +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/device.h> +#include <linux/iommu.h> +#include <uapi/linux/idxd.h> +#include <linux/highmem.h> +#include <linux/sched/smt.h> +#include <crypto/internal/acompress.h> + +#include "idxd.h" +#include "iaa_crypto.h" +#include "iaa_crypto_stats.h" + +#ifdef pr_fmt +#undef pr_fmt +#endif + +#define pr_fmt(fmt) "idxd: " IDXD_SUBDRIVER_NAME ": " fmt + +#define IAA_ALG_PRIORITY 300 + +/* number of iaa instances probed */ +static unsigned int nr_iaa; +static unsigned int nr_cpus; +static unsigned int nr_nodes; +static unsigned int nr_cpus_per_node; + +/* Number of physical cpus sharing each iaa instance */ +static unsigned int cpus_per_iaa; + +/* Per-cpu lookup table for balanced wqs */ +static struct wq_table_entry __percpu *wq_table; + +static struct idxd_wq *wq_table_next_wq(int cpu) +{ + struct wq_table_entry *entry = per_cpu_ptr(wq_table, cpu); + + if (++entry->cur_wq >= entry->n_wqs) + entry->cur_wq = 0; + + if (!entry->wqs[entry->cur_wq]) + return NULL; + + pr_debug("%s: returning wq at idx %d (iaa wq %d.%d) from cpu %d\n", __func__, + entry->cur_wq, entry->wqs[entry->cur_wq]->idxd->id, + entry->wqs[entry->cur_wq]->id, cpu); + + return entry->wqs[entry->cur_wq]; +} + +static void wq_table_add(int cpu, struct idxd_wq *wq) +{ + struct wq_table_entry *entry = per_cpu_ptr(wq_table, cpu); + + if (WARN_ON(entry->n_wqs == entry->max_wqs)) + return; + + entry->wqs[entry->n_wqs++] = wq; + + pr_debug("%s: added iaa wq %d.%d to idx %d of cpu %d\n", __func__, + entry->wqs[entry->n_wqs - 1]->idxd->id, + entry->wqs[entry->n_wqs - 1]->id, entry->n_wqs - 1, cpu); +} + +static void wq_table_free_entry(int cpu) +{ + struct wq_table_entry *entry = per_cpu_ptr(wq_table, cpu); + + kfree(entry->wqs); + memset(entry, 0, sizeof(*entry)); +} + +static void wq_table_clear_entry(int cpu) +{ + struct wq_table_entry *entry = per_cpu_ptr(wq_table, cpu); + + entry->n_wqs = 0; + entry->cur_wq = 0; + memset(entry->wqs, 0, entry->max_wqs * sizeof(struct idxd_wq *)); +} + +LIST_HEAD(iaa_devices); +DEFINE_MUTEX(iaa_devices_lock); + +/* If enabled, IAA hw crypto algos are registered, unavailable otherwise */ +static bool iaa_crypto_enabled; +static bool iaa_crypto_registered; + +/* Verify results of IAA compress or not */ +static bool iaa_verify_compress = true; + +static ssize_t verify_compress_show(struct device_driver *driver, char *buf) +{ + return sprintf(buf, "%d\n", iaa_verify_compress); +} + +static ssize_t verify_compress_store(struct device_driver *driver, + const char *buf, size_t count) +{ + int ret = -EBUSY; + + mutex_lock(&iaa_devices_lock); + + if (iaa_crypto_enabled) + goto out; + + ret = kstrtobool(buf, &iaa_verify_compress); + if (ret) + goto out; + + ret = count; +out: + mutex_unlock(&iaa_devices_lock); + + return ret; +} +static DRIVER_ATTR_RW(verify_compress); + +/* + * The iaa crypto driver supports three 'sync' methods determining how + * compressions and decompressions are performed: + * + * - sync: the compression or decompression completes before + * returning. This is the mode used by the async crypto + * interface when the sync mode is set to 'sync' and by + * the sync crypto interface regardless of setting. + * + * - async: the compression or decompression is submitted and returns + * immediately. Completion interrupts are not used so + * the caller is responsible for polling the descriptor + * for completion. This mode is applicable to only the + * async crypto interface and is ignored for anything + * else. + * + * - async_irq: the compression or decompression is submitted and + * returns immediately. Completion interrupts are + * enabled so the caller can wait for the completion and + * yield to other threads. When the compression or + * decompression completes, the completion is signaled + * and the caller awakened. This mode is applicable to + * only the async crypto interface and is ignored for + * anything else. + * + * These modes can be set using the iaa_crypto sync_mode driver + * attribute. + */ + +/* Use async mode */ +static bool async_mode; +/* Use interrupts */ +static bool use_irq; + +/** + * set_iaa_sync_mode - Set IAA sync mode + * @name: The name of the sync mode + * + * Make the IAA sync mode named @name the current sync mode used by + * compression/decompression. + */ + +static int set_iaa_sync_mode(const char *name) +{ + int ret = 0; + + if (sysfs_streq(name, "sync")) { + async_mode = false; + use_irq = false; + } else if (sysfs_streq(name, "async")) { + async_mode = false; + use_irq = false; + } else if (sysfs_streq(name, "async_irq")) { + async_mode = true; + use_irq = true; + } else { + ret = -EINVAL; + } + + return ret; +} + +static ssize_t sync_mode_show(struct device_driver *driver, char *buf) +{ + int ret = 0; + + if (!async_mode && !use_irq) + ret = sprintf(buf, "%s\n", "sync"); + else if (async_mode && !use_irq) + ret = sprintf(buf, "%s\n", "async"); + else if (async_mode && use_irq) + ret = sprintf(buf, "%s\n", "async_irq"); + + return ret; +} + +static ssize_t sync_mode_store(struct device_driver *driver, + const char *buf, size_t count) +{ + int ret = -EBUSY; + + mutex_lock(&iaa_devices_lock); + + if (iaa_crypto_enabled) + goto out; + + ret = set_iaa_sync_mode(buf); + if (ret == 0) + ret = count; +out: + mutex_unlock(&iaa_devices_lock); + + return ret; +} +static DRIVER_ATTR_RW(sync_mode); + +static struct iaa_compression_mode *iaa_compression_modes[IAA_COMP_MODES_MAX]; + +static int find_empty_iaa_compression_mode(void) +{ + int i = -EINVAL; + + for (i = 0; i < IAA_COMP_MODES_MAX; i++) { + if (iaa_compression_modes[i]) + continue; + break; + } + + return i; +} + +static struct iaa_compression_mode *find_iaa_compression_mode(const char *name, int *idx) +{ + struct iaa_compression_mode *mode; + int i; + + for (i = 0; i < IAA_COMP_MODES_MAX; i++) { + mode = iaa_compression_modes[i]; + if (!mode) + continue; + + if (!strcmp(mode->name, name)) { + *idx = i; + return iaa_compression_modes[i]; + } + } + + return NULL; +} + +static void free_iaa_compression_mode(struct iaa_compression_mode *mode) +{ + kfree(mode->name); + kfree(mode->ll_table); + kfree(mode->d_table); + + kfree(mode); +} + +/* + * IAA Compression modes are defined by an ll_table and a d_table. + * These tables are typically generated and captured using statistics + * collected from running actual compress/decompress workloads. + * + * A module or other kernel code can add and remove compression modes + * with a given name using the exported @add_iaa_compression_mode() + * and @remove_iaa_compression_mode functions. + * + * When a new compression mode is added, the tables are saved in a + * global compression mode list. When IAA devices are added, a + * per-IAA device dma mapping is created for each IAA device, for each + * compression mode. These are the tables used to do the actual + * compression/deccompression and are unmapped if/when the devices are + * removed. Currently, compression modes must be added before any + * device is added, and removed after all devices have been removed. + */ + +/** + * remove_iaa_compression_mode - Remove an IAA compression mode + * @name: The name the compression mode will be known as + * + * Remove the IAA compression mode named @name. + */ +void remove_iaa_compression_mode(const char *name) +{ + struct iaa_compression_mode *mode; + int idx; + + mutex_lock(&iaa_devices_lock); + + if (!list_empty(&iaa_devices)) + goto out; + + mode = find_iaa_compression_mode(name, &idx); + if (mode) { + free_iaa_compression_mode(mode); + iaa_compression_modes[idx] = NULL; + } +out: + mutex_unlock(&iaa_devices_lock); +} +EXPORT_SYMBOL_GPL(remove_iaa_compression_mode); + +/** + * add_iaa_compression_mode - Add an IAA compression mode + * @name: The name the compression mode will be known as + * @ll_table: The ll table + * @ll_table_size: The ll table size in bytes + * @d_table: The d table + * @d_table_size: The d table size in bytes + * @init: Optional callback function to init the compression mode data + * @free: Optional callback function to free the compression mode data + * + * Add a new IAA compression mode named @name. + * + * Returns 0 if successful, errcode otherwise. + */ +int add_iaa_compression_mode(const char *name, + const u32 *ll_table, + int ll_table_size, + const u32 *d_table, + int d_table_size, + iaa_dev_comp_init_fn_t init, + iaa_dev_comp_free_fn_t free) +{ + struct iaa_compression_mode *mode; + int idx, ret = -ENOMEM; + + mutex_lock(&iaa_devices_lock); + + if (!list_empty(&iaa_devices)) { + ret = -EBUSY; + goto out; + } + + mode = kzalloc(sizeof(*mode), GFP_KERNEL); + if (!mode) + goto out; + + mode->name = kstrdup(name, GFP_KERNEL); + if (!mode->name) + goto free; + + if (ll_table) { + mode->ll_table = kmemdup(ll_table, ll_table_size, GFP_KERNEL); + if (!mode->ll_table) + goto free; + mode->ll_table_size = ll_table_size; + } + + if (d_table) { + mode->d_table = kmemdup(d_table, d_table_size, GFP_KERNEL); + if (!mode->d_table) + goto free; + mode->d_table_size = d_table_size; + } + + mode->init = init; + mode->free = free; + + idx = find_empty_iaa_compression_mode(); + if (idx < 0) + goto free; + + pr_debug("IAA compression mode %s added at idx %d\n", + mode->name, idx); + + iaa_compression_modes[idx] = mode; + + ret = 0; +out: + mutex_unlock(&iaa_devices_lock); + + return ret; +free: + free_iaa_compression_mode(mode); + goto out; +} +EXPORT_SYMBOL_GPL(add_iaa_compression_mode); + +static struct iaa_device_compression_mode * +get_iaa_device_compression_mode(struct iaa_device *iaa_device, int idx) +{ + return iaa_device->compression_modes[idx]; +} + +static void free_device_compression_mode(struct iaa_device *iaa_device, + struct iaa_device_compression_mode *device_mode) +{ + size_t size = sizeof(struct aecs_comp_table_record) + IAA_AECS_ALIGN; + struct device *dev = &iaa_device->idxd->pdev->dev; + + kfree(device_mode->name); + + if (device_mode->aecs_comp_table) + dma_free_coherent(dev, size, device_mode->aecs_comp_table, + device_mode->aecs_comp_table_dma_addr); + kfree(device_mode); +} + +#define IDXD_OP_FLAG_AECS_RW_TGLS 0x400000 +#define IAX_AECS_DEFAULT_FLAG (IDXD_OP_FLAG_CRAV | IDXD_OP_FLAG_RCR | IDXD_OP_FLAG_CC) +#define IAX_AECS_COMPRESS_FLAG (IAX_AECS_DEFAULT_FLAG | IDXD_OP_FLAG_RD_SRC2_AECS) +#define IAX_AECS_DECOMPRESS_FLAG (IAX_AECS_DEFAULT_FLAG | IDXD_OP_FLAG_RD_SRC2_AECS) +#define IAX_AECS_GEN_FLAG (IAX_AECS_DEFAULT_FLAG | \ + IDXD_OP_FLAG_WR_SRC2_AECS_COMP | \ + IDXD_OP_FLAG_AECS_RW_TGLS) + +static int check_completion(struct device *dev, + struct iax_completion_record *comp, + bool compress, + bool only_once); + +static int init_device_compression_mode(struct iaa_device *iaa_device, + struct iaa_compression_mode *mode, + int idx, struct idxd_wq *wq) +{ + size_t size = sizeof(struct aecs_comp_table_record) + IAA_AECS_ALIGN; + struct device *dev = &iaa_device->idxd->pdev->dev; + struct iaa_device_compression_mode *device_mode; + int ret = -ENOMEM; + + device_mode = kzalloc(sizeof(*device_mode), GFP_KERNEL); + if (!device_mode) + return -ENOMEM; + + device_mode->name = kstrdup(mode->name, GFP_KERNEL); + if (!device_mode->name) + goto free; + + device_mode->aecs_comp_table = dma_alloc_coherent(dev, size, + &device_mode->aecs_comp_table_dma_addr, GFP_KERNEL); + if (!device_mode->aecs_comp_table) + goto free; + + /* Add Huffman table to aecs */ + memset(device_mode->aecs_comp_table, 0, sizeof(*device_mode->aecs_comp_table)); + memcpy(device_mode->aecs_comp_table->ll_sym, mode->ll_table, mode->ll_table_size); + memcpy(device_mode->aecs_comp_table->d_sym, mode->d_table, mode->d_table_size); + + if (mode->init) { + ret = mode->init(device_mode); + if (ret) + goto free; + } + + /* mode index should match iaa_compression_modes idx */ + iaa_device->compression_modes[idx] = device_mode; + + pr_debug("IAA %s compression mode initialized for iaa device %d\n", + mode->name, iaa_device->idxd->id); + + ret = 0; +out: + return ret; +free: + pr_debug("IAA %s compression mode initialization failed for iaa device %d\n", + mode->name, iaa_device->idxd->id); + + free_device_compression_mode(iaa_device, device_mode); + goto out; +} + +static int init_device_compression_modes(struct iaa_device *iaa_device, + struct idxd_wq *wq) +{ + struct iaa_compression_mode *mode; + int i, ret = 0; + + for (i = 0; i < IAA_COMP_MODES_MAX; i++) { + mode = iaa_compression_modes[i]; + if (!mode) + continue; + + ret = init_device_compression_mode(iaa_device, mode, i, wq); + if (ret) + break; + } + + return ret; +} + +static void remove_device_compression_modes(struct iaa_device *iaa_device) +{ + struct iaa_device_compression_mode *device_mode; + int i; + + for (i = 0; i < IAA_COMP_MODES_MAX; i++) { + device_mode = iaa_device->compression_modes[i]; + if (!device_mode) + continue; + + if (iaa_compression_modes[i]->free) + iaa_compression_modes[i]->free(device_mode); + free_device_compression_mode(iaa_device, device_mode); + iaa_device->compression_modes[i] = NULL; + } +} + +static struct iaa_device *iaa_device_alloc(void) +{ + struct iaa_device *iaa_device; + + iaa_device = kzalloc(sizeof(*iaa_device), GFP_KERNEL); + if (!iaa_device) + return NULL; + + INIT_LIST_HEAD(&iaa_device->wqs); + + return iaa_device; +} + +static bool iaa_has_wq(struct iaa_device *iaa_device, struct idxd_wq *wq) +{ + struct iaa_wq *iaa_wq; + + list_for_each_entry(iaa_wq, &iaa_device->wqs, list) { + if (iaa_wq->wq == wq) + return true; + } + + return false; +} + +static struct iaa_device *add_iaa_device(struct idxd_device *idxd) +{ + struct iaa_device *iaa_device; + + iaa_device = iaa_device_alloc(); + if (!iaa_device) + return NULL; + + iaa_device->idxd = idxd; + + list_add_tail(&iaa_device->list, &iaa_devices); + + nr_iaa++; + + return iaa_device; +} + +static int init_iaa_device(struct iaa_device *iaa_device, struct iaa_wq *iaa_wq) +{ + int ret = 0; + + ret = init_device_compression_modes(iaa_device, iaa_wq->wq); + if (ret) + return ret; + + return ret; +} + +static void del_iaa_device(struct iaa_device *iaa_device) +{ + list_del(&iaa_device->list); + + nr_iaa--; +} + +static int add_iaa_wq(struct iaa_device *iaa_device, struct idxd_wq *wq, + struct iaa_wq **new_wq) +{ + struct idxd_device *idxd = iaa_device->idxd; + struct pci_dev *pdev = idxd->pdev; + struct device *dev = &pdev->dev; + struct iaa_wq *iaa_wq; + + iaa_wq = kzalloc(sizeof(*iaa_wq), GFP_KERNEL); + if (!iaa_wq) + return -ENOMEM; + + iaa_wq->wq = wq; + iaa_wq->iaa_device = iaa_device; + idxd_wq_set_private(wq, iaa_wq); + + list_add_tail(&iaa_wq->list, &iaa_device->wqs); + + iaa_device->n_wq++; + + if (new_wq) + *new_wq = iaa_wq; + + dev_dbg(dev, "added wq %d to iaa device %d, n_wq %d\n", + wq->id, iaa_device->idxd->id, iaa_device->n_wq); + + return 0; +} + +static void del_iaa_wq(struct iaa_device *iaa_device, struct idxd_wq *wq) +{ + struct idxd_device *idxd = iaa_device->idxd; + struct pci_dev *pdev = idxd->pdev; + struct device *dev = &pdev->dev; + struct iaa_wq *iaa_wq; + + list_for_each_entry(iaa_wq, &iaa_device->wqs, list) { + if (iaa_wq->wq == wq) { + list_del(&iaa_wq->list); + iaa_device->n_wq--; + + dev_dbg(dev, "removed wq %d from iaa_device %d, n_wq %d, nr_iaa %d\n", + wq->id, iaa_device->idxd->id, + iaa_device->n_wq, nr_iaa); + + if (iaa_device->n_wq == 0) + del_iaa_device(iaa_device); + break; + } + } +} + +static void clear_wq_table(void) +{ + int cpu; + + for (cpu = 0; cpu < nr_cpus; cpu++) + wq_table_clear_entry(cpu); + + pr_debug("cleared wq table\n"); +} + +static void free_iaa_device(struct iaa_device *iaa_device) +{ + if (!iaa_device) + return; + + remove_device_compression_modes(iaa_device); + kfree(iaa_device); +} + +static void __free_iaa_wq(struct iaa_wq *iaa_wq) +{ + struct iaa_device *iaa_device; + + if (!iaa_wq) + return; + + iaa_device = iaa_wq->iaa_device; + if (iaa_device->n_wq == 0) + free_iaa_device(iaa_wq->iaa_device); +} + +static void free_iaa_wq(struct iaa_wq *iaa_wq) +{ + struct idxd_wq *wq; + + __free_iaa_wq(iaa_wq); + + wq = iaa_wq->wq; + + kfree(iaa_wq); + idxd_wq_set_private(wq, NULL); +} + +static int iaa_wq_get(struct idxd_wq *wq) +{ + struct idxd_device *idxd = wq->idxd; + struct iaa_wq *iaa_wq; + int ret = 0; + + spin_lock(&idxd->dev_lock); + iaa_wq = idxd_wq_get_private(wq); + if (iaa_wq && !iaa_wq->remove) { + iaa_wq->ref++; + idxd_wq_get(wq); + } else { + ret = -ENODEV; + } + spin_unlock(&idxd->dev_lock); + + return ret; +} + +static int iaa_wq_put(struct idxd_wq *wq) +{ + struct idxd_device *idxd = wq->idxd; + struct iaa_wq *iaa_wq; + bool free = false; + int ret = 0; + + spin_lock(&idxd->dev_lock); + iaa_wq = idxd_wq_get_private(wq); + if (iaa_wq) { + iaa_wq->ref--; + if (iaa_wq->ref == 0 && iaa_wq->remove) { + idxd_wq_set_private(wq, NULL); + free = true; + } + idxd_wq_put(wq); + } else { + ret = -ENODEV; + } + spin_unlock(&idxd->dev_lock); + if (free) { + __free_iaa_wq(iaa_wq); + kfree(iaa_wq); + } + + return ret; +} + +static void free_wq_table(void) +{ + int cpu; + + for (cpu = 0; cpu < nr_cpus; cpu++) + wq_table_free_entry(cpu); + + free_percpu(wq_table); + + pr_debug("freed wq table\n"); +} + +static int alloc_wq_table(int max_wqs) +{ + struct wq_table_entry *entry; + int cpu; + + wq_table = alloc_percpu(struct wq_table_entry); + if (!wq_table) + return -ENOMEM; + + for (cpu = 0; cpu < nr_cpus; cpu++) { + entry = per_cpu_ptr(wq_table, cpu); + entry->wqs = kcalloc(max_wqs, sizeof(*entry->wqs), GFP_KERNEL); + if (!entry->wqs) { + free_wq_table(); + return -ENOMEM; + } + + entry->max_wqs = max_wqs; + } + + pr_debug("initialized wq table\n"); + + return 0; +} + +static int save_iaa_wq(struct idxd_wq *wq) +{ + struct iaa_device *iaa_device, *found = NULL; + struct idxd_device *idxd; + struct pci_dev *pdev; + struct device *dev; + int ret = 0; + + list_for_each_entry(iaa_device, &iaa_devices, list) { + if (iaa_device->idxd == wq->idxd) { + idxd = iaa_device->idxd; + pdev = idxd->pdev; + dev = &pdev->dev; + /* + * Check to see that we don't already have this wq. + * Shouldn't happen but we don't control probing. + */ + if (iaa_has_wq(iaa_device, wq)) { + dev_dbg(dev, "same wq probed multiple times for iaa_device %p\n", + iaa_device); + goto out; + } + + found = iaa_device; + + ret = add_iaa_wq(iaa_device, wq, NULL); + if (ret) + goto out; + + break; + } + } + + if (!found) { + struct iaa_device *new_device; + struct iaa_wq *new_wq; + + new_device = add_iaa_device(wq->idxd); + if (!new_device) { + ret = -ENOMEM; + goto out; + } + + ret = add_iaa_wq(new_device, wq, &new_wq); + if (ret) { + del_iaa_device(new_device); + free_iaa_device(new_device); + goto out; + } + + ret = init_iaa_device(new_device, new_wq); + if (ret) { + del_iaa_wq(new_device, new_wq->wq); + del_iaa_device(new_device); + free_iaa_wq(new_wq); + goto out; + } + } + + if (WARN_ON(nr_iaa == 0)) + return -EINVAL; + + cpus_per_iaa = (nr_nodes * nr_cpus_per_node) / nr_iaa; + if (!cpus_per_iaa) + cpus_per_iaa = 1; +out: + return ret; +} + +static void remove_iaa_wq(struct idxd_wq *wq) +{ + struct iaa_device *iaa_device; + + list_for_each_entry(iaa_device, &iaa_devices, list) { + if (iaa_has_wq(iaa_device, wq)) { + del_iaa_wq(iaa_device, wq); + break; + } + } + + if (nr_iaa) { + cpus_per_iaa = (nr_nodes * nr_cpus_per_node) / nr_iaa; + if (!cpus_per_iaa) + cpus_per_iaa = 1; + } else + cpus_per_iaa = 1; +} + +static int wq_table_add_wqs(int iaa, int cpu) +{ + struct iaa_device *iaa_device, *found_device = NULL; + int ret = 0, cur_iaa = 0, n_wqs_added = 0; + struct idxd_device *idxd; + struct iaa_wq *iaa_wq; + struct pci_dev *pdev; + struct device *dev; + + list_for_each_entry(iaa_device, &iaa_devices, list) { + idxd = iaa_device->idxd; + pdev = idxd->pdev; + dev = &pdev->dev; + + if (cur_iaa != iaa) { + cur_iaa++; + continue; + } + + found_device = iaa_device; + dev_dbg(dev, "getting wq from iaa_device %d, cur_iaa %d\n", + found_device->idxd->id, cur_iaa); + break; + } + + if (!found_device) { + found_device = list_first_entry_or_null(&iaa_devices, + struct iaa_device, list); + if (!found_device) { + pr_debug("couldn't find any iaa devices with wqs!\n"); + ret = -EINVAL; + goto out; + } + cur_iaa = 0; + + idxd = found_device->idxd; + pdev = idxd->pdev; + dev = &pdev->dev; + dev_dbg(dev, "getting wq from only iaa_device %d, cur_iaa %d\n", + found_device->idxd->id, cur_iaa); + } + + list_for_each_entry(iaa_wq, &found_device->wqs, list) { + wq_table_add(cpu, iaa_wq->wq); + pr_debug("rebalance: added wq for cpu=%d: iaa wq %d.%d\n", + cpu, iaa_wq->wq->idxd->id, iaa_wq->wq->id); + n_wqs_added++; + } + + if (!n_wqs_added) { + pr_debug("couldn't find any iaa wqs!\n"); + ret = -EINVAL; + goto out; + } +out: + return ret; +} + +/* + * Rebalance the wq table so that given a cpu, it's easy to find the + * closest IAA instance. The idea is to try to choose the most + * appropriate IAA instance for a caller and spread available + * workqueues around to clients. + */ +static void rebalance_wq_table(void) +{ + const struct cpumask *node_cpus; + int node_cpu, node, cpu, iaa = 0; + + if (nr_iaa == 0) + return; + + pr_debug("rebalance: nr_nodes=%d, nr_cpus %d, nr_iaa %d, cpus_per_iaa %d\n", + nr_nodes, nr_cpus, nr_iaa, cpus_per_iaa); + + clear_wq_table(); + + if (nr_iaa == 1) { + for_each_possible_cpu(cpu) { + if (WARN_ON(wq_table_add_wqs(0, cpu))) + goto err; + } + + return; + } + + for_each_node_with_cpus(node) { + cpu = 0; + node_cpus = cpumask_of_node(node); + + for_each_cpu(node_cpu, node_cpus) { + iaa = cpu / cpus_per_iaa; + if (WARN_ON(wq_table_add_wqs(iaa, node_cpu))) + goto err; + cpu++; + } + } + + return; +err: + pr_debug("could not add any wqs for iaa %d to cpu %d!\n", iaa, cpu); +} + +static inline int check_completion(struct device *dev, + struct iax_completion_record *comp, + bool compress, + bool only_once) +{ + char *op_str = compress ? "compress" : "decompress"; + int status_checks = 0; + int ret = 0; + + while (!comp->status) { + if (only_once) + return -EAGAIN; + cpu_relax(); + if (status_checks++ >= IAA_COMPLETION_TIMEOUT) { + /* Something is wrong with the hw, disable it. */ + dev_err(dev, "%s completion timed out - " + "assuming broken hw, iaa_crypto now DISABLED\n", + op_str); + iaa_crypto_enabled = false; + ret = -ETIMEDOUT; + goto out; + } + } + + if (comp->status != IAX_COMP_SUCCESS) { + if (comp->status == IAA_ERROR_WATCHDOG_EXPIRED) { + ret = -ETIMEDOUT; + dev_dbg(dev, "%s timed out, size=0x%x\n", + op_str, comp->output_size); + update_completion_timeout_errs(); + goto out; + } + + if (comp->status == IAA_ANALYTICS_ERROR && + comp->error_code == IAA_ERROR_COMP_BUF_OVERFLOW && compress) { + ret = -E2BIG; + dev_dbg(dev, "compressed > uncompressed size," + " not compressing, size=0x%x\n", + comp->output_size); + update_completion_comp_buf_overflow_errs(); + goto out; + } + + if (comp->status == IAA_ERROR_DECOMP_BUF_OVERFLOW) { + ret = -EOVERFLOW; + goto out; + } + + ret = -EINVAL; + dev_dbg(dev, "iaa %s status=0x%x, error=0x%x, size=0x%x\n", + op_str, comp->status, comp->error_code, comp->output_size); + print_hex_dump(KERN_INFO, "cmp-rec: ", DUMP_PREFIX_OFFSET, 8, 1, comp, 64, 0); + update_completion_einval_errs(); + + goto out; + } +out: + return ret; +} + +static int deflate_generic_decompress(struct acomp_req *req) +{ + ACOMP_FBREQ_ON_STACK(fbreq, req); + int ret; + + ret = crypto_acomp_decompress(fbreq); + req->dlen = fbreq->dlen; + + update_total_sw_decomp_calls(); + + return ret; +} + +static int iaa_remap_for_verify(struct device *dev, struct iaa_wq *iaa_wq, + struct acomp_req *req, + dma_addr_t *src_addr, dma_addr_t *dst_addr); + +static int iaa_compress_verify(struct crypto_tfm *tfm, struct acomp_req *req, + struct idxd_wq *wq, + dma_addr_t src_addr, unsigned int slen, + dma_addr_t dst_addr, unsigned int *dlen); + +static void iaa_desc_complete(struct idxd_desc *idxd_desc, + enum idxd_complete_type comp_type, + bool free_desc, void *__ctx, + u32 *status) +{ + struct iaa_device_compression_mode *active_compression_mode; + struct iaa_compression_ctx *compression_ctx; + struct crypto_ctx *ctx = __ctx; + struct iaa_device *iaa_device; + struct idxd_device *idxd; + struct iaa_wq *iaa_wq; + struct pci_dev *pdev; + struct device *dev; + int ret, err = 0; + + compression_ctx = crypto_tfm_ctx(ctx->tfm); + + iaa_wq = idxd_wq_get_private(idxd_desc->wq); + iaa_device = iaa_wq->iaa_device; + idxd = iaa_device->idxd; + pdev = idxd->pdev; + dev = &pdev->dev; + + active_compression_mode = get_iaa_device_compression_mode(iaa_device, + compression_ctx->mode); + dev_dbg(dev, "%s: compression mode %s," + " ctx->src_addr %llx, ctx->dst_addr %llx\n", __func__, + active_compression_mode->name, + ctx->src_addr, ctx->dst_addr); + + ret = check_completion(dev, idxd_desc->iax_completion, + ctx->compress, false); + if (ret) { + dev_dbg(dev, "%s: check_completion failed ret=%d\n", __func__, ret); + if (!ctx->compress && + idxd_desc->iax_completion->status == IAA_ANALYTICS_ERROR) { + pr_warn("%s: falling back to deflate-generic decompress, " + "analytics error code %x\n", __func__, + idxd_desc->iax_completion->error_code); + ret = deflate_generic_decompress(ctx->req); + if (ret) { + dev_dbg(dev, "%s: deflate-generic failed ret=%d\n", + __func__, ret); + err = -EIO; + goto err; + } + } else { + err = -EIO; + goto err; + } + } else { + ctx->req->dlen = idxd_desc->iax_completion->output_size; + } + + /* Update stats */ + if (ctx->compress) { + update_total_comp_bytes_out(ctx->req->dlen); + update_wq_comp_bytes(iaa_wq->wq, ctx->req->dlen); + } else { + update_total_decomp_bytes_in(ctx->req->slen); + update_wq_decomp_bytes(iaa_wq->wq, ctx->req->slen); + } + + if (ctx->compress && compression_ctx->verify_compress) { + u32 *compression_crc = acomp_request_ctx(ctx->req); + dma_addr_t src_addr, dst_addr; + + *compression_crc = idxd_desc->iax_completion->crc; + + ret = iaa_remap_for_verify(dev, iaa_wq, ctx->req, &src_addr, &dst_addr); + if (ret) { + dev_dbg(dev, "%s: compress verify remap failed ret=%d\n", __func__, ret); + err = -EIO; + goto out; + } + + ret = iaa_compress_verify(ctx->tfm, ctx->req, iaa_wq->wq, src_addr, + ctx->req->slen, dst_addr, &ctx->req->dlen); + if (ret) { + dev_dbg(dev, "%s: compress verify failed ret=%d\n", __func__, ret); + err = -EIO; + } + + dma_unmap_sg(dev, ctx->req->dst, sg_nents(ctx->req->dst), DMA_TO_DEVICE); + dma_unmap_sg(dev, ctx->req->src, sg_nents(ctx->req->src), DMA_FROM_DEVICE); + + goto out; + } +err: + dma_unmap_sg(dev, ctx->req->dst, sg_nents(ctx->req->dst), DMA_FROM_DEVICE); + dma_unmap_sg(dev, ctx->req->src, sg_nents(ctx->req->src), DMA_TO_DEVICE); +out: + if (ret != 0) + dev_dbg(dev, "asynchronous compress failed ret=%d\n", ret); + + if (ctx->req->base.complete) + acomp_request_complete(ctx->req, err); + + if (free_desc) + idxd_free_desc(idxd_desc->wq, idxd_desc); + iaa_wq_put(idxd_desc->wq); +} + +static int iaa_compress(struct crypto_tfm *tfm, struct acomp_req *req, + struct idxd_wq *wq, + dma_addr_t src_addr, unsigned int slen, + dma_addr_t dst_addr, unsigned int *dlen) +{ + struct iaa_device_compression_mode *active_compression_mode; + struct iaa_compression_ctx *ctx = crypto_tfm_ctx(tfm); + u32 *compression_crc = acomp_request_ctx(req); + struct iaa_device *iaa_device; + struct idxd_desc *idxd_desc; + struct iax_hw_desc *desc; + struct idxd_device *idxd; + struct iaa_wq *iaa_wq; + struct pci_dev *pdev; + struct device *dev; + int ret = 0; + + iaa_wq = idxd_wq_get_private(wq); + iaa_device = iaa_wq->iaa_device; + idxd = iaa_device->idxd; + pdev = idxd->pdev; + dev = &pdev->dev; + + active_compression_mode = get_iaa_device_compression_mode(iaa_device, ctx->mode); + + idxd_desc = idxd_alloc_desc(wq, IDXD_OP_BLOCK); + if (IS_ERR(idxd_desc)) { + dev_dbg(dev, "idxd descriptor allocation failed\n"); + dev_dbg(dev, "iaa compress failed: ret=%ld\n", PTR_ERR(idxd_desc)); + return PTR_ERR(idxd_desc); + } + desc = idxd_desc->iax_hw; + + desc->flags = IDXD_OP_FLAG_CRAV | IDXD_OP_FLAG_RCR | + IDXD_OP_FLAG_RD_SRC2_AECS | IDXD_OP_FLAG_CC; + desc->opcode = IAX_OPCODE_COMPRESS; + desc->compr_flags = IAA_COMP_FLAGS; + desc->priv = 0; + + desc->src1_addr = (u64)src_addr; + desc->src1_size = slen; + desc->dst_addr = (u64)dst_addr; + desc->max_dst_size = *dlen; + desc->src2_addr = active_compression_mode->aecs_comp_table_dma_addr; + desc->src2_size = sizeof(struct aecs_comp_table_record); + desc->completion_addr = idxd_desc->compl_dma; + + if (ctx->use_irq) { + desc->flags |= IDXD_OP_FLAG_RCI; + + idxd_desc->crypto.req = req; + idxd_desc->crypto.tfm = tfm; + idxd_desc->crypto.src_addr = src_addr; + idxd_desc->crypto.dst_addr = dst_addr; + idxd_desc->crypto.compress = true; + + dev_dbg(dev, "%s use_async_irq: compression mode %s," + " src_addr %llx, dst_addr %llx\n", __func__, + active_compression_mode->name, + src_addr, dst_addr); + } + + dev_dbg(dev, "%s: compression mode %s," + " desc->src1_addr %llx, desc->src1_size %d," + " desc->dst_addr %llx, desc->max_dst_size %d," + " desc->src2_addr %llx, desc->src2_size %d\n", __func__, + active_compression_mode->name, + desc->src1_addr, desc->src1_size, desc->dst_addr, + desc->max_dst_size, desc->src2_addr, desc->src2_size); + + ret = idxd_submit_desc(wq, idxd_desc); + if (ret) { + dev_dbg(dev, "submit_desc failed ret=%d\n", ret); + goto err; + } + + /* Update stats */ + update_total_comp_calls(); + update_wq_comp_calls(wq); + + if (ctx->async_mode) { + ret = -EINPROGRESS; + dev_dbg(dev, "%s: returning -EINPROGRESS\n", __func__); + goto out; + } + + ret = check_completion(dev, idxd_desc->iax_completion, true, false); + if (ret) { + dev_dbg(dev, "check_completion failed ret=%d\n", ret); + goto err; + } + + *dlen = idxd_desc->iax_completion->output_size; + + /* Update stats */ + update_total_comp_bytes_out(*dlen); + update_wq_comp_bytes(wq, *dlen); + + *compression_crc = idxd_desc->iax_completion->crc; + + if (!ctx->async_mode) + idxd_free_desc(wq, idxd_desc); +out: + return ret; +err: + idxd_free_desc(wq, idxd_desc); + dev_dbg(dev, "iaa compress failed: ret=%d\n", ret); + + goto out; +} + +static int iaa_remap_for_verify(struct device *dev, struct iaa_wq *iaa_wq, + struct acomp_req *req, + dma_addr_t *src_addr, dma_addr_t *dst_addr) +{ + int ret = 0; + int nr_sgs; + + dma_unmap_sg(dev, req->dst, sg_nents(req->dst), DMA_FROM_DEVICE); + dma_unmap_sg(dev, req->src, sg_nents(req->src), DMA_TO_DEVICE); + + nr_sgs = dma_map_sg(dev, req->src, sg_nents(req->src), DMA_FROM_DEVICE); + if (nr_sgs <= 0 || nr_sgs > 1) { + dev_dbg(dev, "verify: couldn't map src sg for iaa device %d," + " wq %d: ret=%d\n", iaa_wq->iaa_device->idxd->id, + iaa_wq->wq->id, ret); + ret = -EIO; + goto out; + } + *src_addr = sg_dma_address(req->src); + dev_dbg(dev, "verify: dma_map_sg, src_addr %llx, nr_sgs %d, req->src %p," + " req->slen %d, sg_dma_len(sg) %d\n", *src_addr, nr_sgs, + req->src, req->slen, sg_dma_len(req->src)); + + nr_sgs = dma_map_sg(dev, req->dst, sg_nents(req->dst), DMA_TO_DEVICE); + if (nr_sgs <= 0 || nr_sgs > 1) { + dev_dbg(dev, "verify: couldn't map dst sg for iaa device %d," + " wq %d: ret=%d\n", iaa_wq->iaa_device->idxd->id, + iaa_wq->wq->id, ret); + ret = -EIO; + dma_unmap_sg(dev, req->src, sg_nents(req->src), DMA_FROM_DEVICE); + goto out; + } + *dst_addr = sg_dma_address(req->dst); + dev_dbg(dev, "verify: dma_map_sg, dst_addr %llx, nr_sgs %d, req->dst %p," + " req->dlen %d, sg_dma_len(sg) %d\n", *dst_addr, nr_sgs, + req->dst, req->dlen, sg_dma_len(req->dst)); +out: + return ret; +} + +static int iaa_compress_verify(struct crypto_tfm *tfm, struct acomp_req *req, + struct idxd_wq *wq, + dma_addr_t src_addr, unsigned int slen, + dma_addr_t dst_addr, unsigned int *dlen) +{ + struct iaa_device_compression_mode *active_compression_mode; + struct iaa_compression_ctx *ctx = crypto_tfm_ctx(tfm); + u32 *compression_crc = acomp_request_ctx(req); + struct iaa_device *iaa_device; + struct idxd_desc *idxd_desc; + struct iax_hw_desc *desc; + struct idxd_device *idxd; + struct iaa_wq *iaa_wq; + struct pci_dev *pdev; + struct device *dev; + int ret = 0; + + iaa_wq = idxd_wq_get_private(wq); + iaa_device = iaa_wq->iaa_device; + idxd = iaa_device->idxd; + pdev = idxd->pdev; + dev = &pdev->dev; + + active_compression_mode = get_iaa_device_compression_mode(iaa_device, ctx->mode); + + idxd_desc = idxd_alloc_desc(wq, IDXD_OP_BLOCK); + if (IS_ERR(idxd_desc)) { + dev_dbg(dev, "idxd descriptor allocation failed\n"); + dev_dbg(dev, "iaa compress failed: ret=%ld\n", + PTR_ERR(idxd_desc)); + return PTR_ERR(idxd_desc); + } + desc = idxd_desc->iax_hw; + + /* Verify (optional) - decompress and check crc, suppress dest write */ + + desc->flags = IDXD_OP_FLAG_CRAV | IDXD_OP_FLAG_RCR | IDXD_OP_FLAG_CC; + desc->opcode = IAX_OPCODE_DECOMPRESS; + desc->decompr_flags = IAA_DECOMP_FLAGS | IAA_DECOMP_SUPPRESS_OUTPUT; + desc->priv = 0; + + desc->src1_addr = (u64)dst_addr; + desc->src1_size = *dlen; + desc->dst_addr = (u64)src_addr; + desc->max_dst_size = slen; + desc->completion_addr = idxd_desc->compl_dma; + + dev_dbg(dev, "(verify) compression mode %s," + " desc->src1_addr %llx, desc->src1_size %d," + " desc->dst_addr %llx, desc->max_dst_size %d," + " desc->src2_addr %llx, desc->src2_size %d\n", + active_compression_mode->name, + desc->src1_addr, desc->src1_size, desc->dst_addr, + desc->max_dst_size, desc->src2_addr, desc->src2_size); + + ret = idxd_submit_desc(wq, idxd_desc); + if (ret) { + dev_dbg(dev, "submit_desc (verify) failed ret=%d\n", ret); + goto err; + } + + ret = check_completion(dev, idxd_desc->iax_completion, false, false); + if (ret) { + dev_dbg(dev, "(verify) check_completion failed ret=%d\n", ret); + goto err; + } + + if (*compression_crc != idxd_desc->iax_completion->crc) { + ret = -EINVAL; + dev_dbg(dev, "(verify) iaa comp/decomp crc mismatch:" + " comp=0x%x, decomp=0x%x\n", *compression_crc, + idxd_desc->iax_completion->crc); + print_hex_dump(KERN_INFO, "cmp-rec: ", DUMP_PREFIX_OFFSET, + 8, 1, idxd_desc->iax_completion, 64, 0); + goto err; + } + + idxd_free_desc(wq, idxd_desc); +out: + return ret; +err: + idxd_free_desc(wq, idxd_desc); + dev_dbg(dev, "iaa compress failed: ret=%d\n", ret); + + goto out; +} + +static int iaa_decompress(struct crypto_tfm *tfm, struct acomp_req *req, + struct idxd_wq *wq, + dma_addr_t src_addr, unsigned int slen, + dma_addr_t dst_addr, unsigned int *dlen) +{ + struct iaa_device_compression_mode *active_compression_mode; + struct iaa_compression_ctx *ctx = crypto_tfm_ctx(tfm); + struct iaa_device *iaa_device; + struct idxd_desc *idxd_desc; + struct iax_hw_desc *desc; + struct idxd_device *idxd; + struct iaa_wq *iaa_wq; + struct pci_dev *pdev; + struct device *dev; + int ret = 0; + + iaa_wq = idxd_wq_get_private(wq); + iaa_device = iaa_wq->iaa_device; + idxd = iaa_device->idxd; + pdev = idxd->pdev; + dev = &pdev->dev; + + active_compression_mode = get_iaa_device_compression_mode(iaa_device, ctx->mode); + + idxd_desc = idxd_alloc_desc(wq, IDXD_OP_BLOCK); + if (IS_ERR(idxd_desc)) { + dev_dbg(dev, "idxd descriptor allocation failed\n"); + dev_dbg(dev, "iaa decompress failed: ret=%ld\n", + PTR_ERR(idxd_desc)); + return PTR_ERR(idxd_desc); + } + desc = idxd_desc->iax_hw; + + desc->flags = IDXD_OP_FLAG_CRAV | IDXD_OP_FLAG_RCR | IDXD_OP_FLAG_CC; + desc->opcode = IAX_OPCODE_DECOMPRESS; + desc->max_dst_size = PAGE_SIZE; + desc->decompr_flags = IAA_DECOMP_FLAGS; + desc->priv = 0; + + desc->src1_addr = (u64)src_addr; + desc->dst_addr = (u64)dst_addr; + desc->max_dst_size = *dlen; + desc->src1_size = slen; + desc->completion_addr = idxd_desc->compl_dma; + + if (ctx->use_irq) { + desc->flags |= IDXD_OP_FLAG_RCI; + + idxd_desc->crypto.req = req; + idxd_desc->crypto.tfm = tfm; + idxd_desc->crypto.src_addr = src_addr; + idxd_desc->crypto.dst_addr = dst_addr; + idxd_desc->crypto.compress = false; + + dev_dbg(dev, "%s: use_async_irq compression mode %s," + " src_addr %llx, dst_addr %llx\n", __func__, + active_compression_mode->name, + src_addr, dst_addr); + } + + dev_dbg(dev, "%s: decompression mode %s," + " desc->src1_addr %llx, desc->src1_size %d," + " desc->dst_addr %llx, desc->max_dst_size %d," + " desc->src2_addr %llx, desc->src2_size %d\n", __func__, + active_compression_mode->name, + desc->src1_addr, desc->src1_size, desc->dst_addr, + desc->max_dst_size, desc->src2_addr, desc->src2_size); + + ret = idxd_submit_desc(wq, idxd_desc); + if (ret) { + dev_dbg(dev, "submit_desc failed ret=%d\n", ret); + goto err; + } + + /* Update stats */ + update_total_decomp_calls(); + update_wq_decomp_calls(wq); + + if (ctx->async_mode) { + ret = -EINPROGRESS; + dev_dbg(dev, "%s: returning -EINPROGRESS\n", __func__); + goto out; + } + + ret = check_completion(dev, idxd_desc->iax_completion, false, false); + if (ret) { + dev_dbg(dev, "%s: check_completion failed ret=%d\n", __func__, ret); + if (idxd_desc->iax_completion->status == IAA_ANALYTICS_ERROR) { + pr_warn("%s: falling back to deflate-generic decompress, " + "analytics error code %x\n", __func__, + idxd_desc->iax_completion->error_code); + ret = deflate_generic_decompress(req); + if (ret) { + dev_dbg(dev, "%s: deflate-generic failed ret=%d\n", + __func__, ret); + goto err; + } + } else { + goto err; + } + } else { + req->dlen = idxd_desc->iax_completion->output_size; + } + + *dlen = req->dlen; + + if (!ctx->async_mode) + idxd_free_desc(wq, idxd_desc); + + /* Update stats */ + update_total_decomp_bytes_in(slen); + update_wq_decomp_bytes(wq, slen); +out: + return ret; +err: + idxd_free_desc(wq, idxd_desc); + dev_dbg(dev, "iaa decompress failed: ret=%d\n", ret); + + goto out; +} + +static int iaa_comp_acompress(struct acomp_req *req) +{ + struct iaa_compression_ctx *compression_ctx; + struct crypto_tfm *tfm = req->base.tfm; + dma_addr_t src_addr, dst_addr; + int nr_sgs, cpu, ret = 0; + struct iaa_wq *iaa_wq; + struct idxd_wq *wq; + struct device *dev; + + compression_ctx = crypto_tfm_ctx(tfm); + + if (!iaa_crypto_enabled) { + pr_debug("iaa_crypto disabled, not compressing\n"); + return -ENODEV; + } + + if (!req->src || !req->slen) { + pr_debug("invalid src, not compressing\n"); + return -EINVAL; + } + + cpu = get_cpu(); + wq = wq_table_next_wq(cpu); + put_cpu(); + if (!wq) { + pr_debug("no wq configured for cpu=%d\n", cpu); + return -ENODEV; + } + + ret = iaa_wq_get(wq); + if (ret) { + pr_debug("no wq available for cpu=%d\n", cpu); + return -ENODEV; + } + + iaa_wq = idxd_wq_get_private(wq); + + dev = &wq->idxd->pdev->dev; + + nr_sgs = dma_map_sg(dev, req->src, sg_nents(req->src), DMA_TO_DEVICE); + if (nr_sgs <= 0 || nr_sgs > 1) { + dev_dbg(dev, "couldn't map src sg for iaa device %d," + " wq %d: ret=%d\n", iaa_wq->iaa_device->idxd->id, + iaa_wq->wq->id, ret); + ret = -EIO; + goto out; + } + src_addr = sg_dma_address(req->src); + dev_dbg(dev, "dma_map_sg, src_addr %llx, nr_sgs %d, req->src %p," + " req->slen %d, sg_dma_len(sg) %d\n", src_addr, nr_sgs, + req->src, req->slen, sg_dma_len(req->src)); + + nr_sgs = dma_map_sg(dev, req->dst, sg_nents(req->dst), DMA_FROM_DEVICE); + if (nr_sgs <= 0 || nr_sgs > 1) { + dev_dbg(dev, "couldn't map dst sg for iaa device %d," + " wq %d: ret=%d\n", iaa_wq->iaa_device->idxd->id, + iaa_wq->wq->id, ret); + ret = -EIO; + goto err_map_dst; + } + dst_addr = sg_dma_address(req->dst); + dev_dbg(dev, "dma_map_sg, dst_addr %llx, nr_sgs %d, req->dst %p," + " req->dlen %d, sg_dma_len(sg) %d\n", dst_addr, nr_sgs, + req->dst, req->dlen, sg_dma_len(req->dst)); + + ret = iaa_compress(tfm, req, wq, src_addr, req->slen, dst_addr, + &req->dlen); + if (ret == -EINPROGRESS) + return ret; + + if (!ret && compression_ctx->verify_compress) { + ret = iaa_remap_for_verify(dev, iaa_wq, req, &src_addr, &dst_addr); + if (ret) { + dev_dbg(dev, "%s: compress verify remap failed ret=%d\n", __func__, ret); + goto out; + } + + ret = iaa_compress_verify(tfm, req, wq, src_addr, req->slen, + dst_addr, &req->dlen); + if (ret) + dev_dbg(dev, "asynchronous compress verification failed ret=%d\n", ret); + + dma_unmap_sg(dev, req->dst, sg_nents(req->dst), DMA_TO_DEVICE); + dma_unmap_sg(dev, req->src, sg_nents(req->src), DMA_FROM_DEVICE); + + goto out; + } + + if (ret) + dev_dbg(dev, "asynchronous compress failed ret=%d\n", ret); + + dma_unmap_sg(dev, req->dst, sg_nents(req->dst), DMA_FROM_DEVICE); +err_map_dst: + dma_unmap_sg(dev, req->src, sg_nents(req->src), DMA_TO_DEVICE); +out: + iaa_wq_put(wq); + + return ret; +} + +static int iaa_comp_adecompress(struct acomp_req *req) +{ + struct crypto_tfm *tfm = req->base.tfm; + dma_addr_t src_addr, dst_addr; + int nr_sgs, cpu, ret = 0; + struct iaa_wq *iaa_wq; + struct device *dev; + struct idxd_wq *wq; + + if (!iaa_crypto_enabled) { + pr_debug("iaa_crypto disabled, not decompressing\n"); + return -ENODEV; + } + + if (!req->src || !req->slen) { + pr_debug("invalid src, not decompressing\n"); + return -EINVAL; + } + + cpu = get_cpu(); + wq = wq_table_next_wq(cpu); + put_cpu(); + if (!wq) { + pr_debug("no wq configured for cpu=%d\n", cpu); + return -ENODEV; + } + + ret = iaa_wq_get(wq); + if (ret) { + pr_debug("no wq available for cpu=%d\n", cpu); + return -ENODEV; + } + + iaa_wq = idxd_wq_get_private(wq); + + dev = &wq->idxd->pdev->dev; + + nr_sgs = dma_map_sg(dev, req->src, sg_nents(req->src), DMA_TO_DEVICE); + if (nr_sgs <= 0 || nr_sgs > 1) { + dev_dbg(dev, "couldn't map src sg for iaa device %d," + " wq %d: ret=%d\n", iaa_wq->iaa_device->idxd->id, + iaa_wq->wq->id, ret); + ret = -EIO; + goto out; + } + src_addr = sg_dma_address(req->src); + dev_dbg(dev, "dma_map_sg, src_addr %llx, nr_sgs %d, req->src %p," + " req->slen %d, sg_dma_len(sg) %d\n", src_addr, nr_sgs, + req->src, req->slen, sg_dma_len(req->src)); + + nr_sgs = dma_map_sg(dev, req->dst, sg_nents(req->dst), DMA_FROM_DEVICE); + if (nr_sgs <= 0 || nr_sgs > 1) { + dev_dbg(dev, "couldn't map dst sg for iaa device %d," + " wq %d: ret=%d\n", iaa_wq->iaa_device->idxd->id, + iaa_wq->wq->id, ret); + ret = -EIO; + goto err_map_dst; + } + dst_addr = sg_dma_address(req->dst); + dev_dbg(dev, "dma_map_sg, dst_addr %llx, nr_sgs %d, req->dst %p," + " req->dlen %d, sg_dma_len(sg) %d\n", dst_addr, nr_sgs, + req->dst, req->dlen, sg_dma_len(req->dst)); + + ret = iaa_decompress(tfm, req, wq, src_addr, req->slen, + dst_addr, &req->dlen); + if (ret == -EINPROGRESS) + return ret; + + if (ret != 0) + dev_dbg(dev, "asynchronous decompress failed ret=%d\n", ret); + + dma_unmap_sg(dev, req->dst, sg_nents(req->dst), DMA_FROM_DEVICE); +err_map_dst: + dma_unmap_sg(dev, req->src, sg_nents(req->src), DMA_TO_DEVICE); +out: + iaa_wq_put(wq); + + return ret; +} + +static void compression_ctx_init(struct iaa_compression_ctx *ctx) +{ + ctx->verify_compress = iaa_verify_compress; + ctx->async_mode = async_mode; + ctx->use_irq = use_irq; +} + +static int iaa_comp_init_fixed(struct crypto_acomp *acomp_tfm) +{ + struct crypto_tfm *tfm = crypto_acomp_tfm(acomp_tfm); + struct iaa_compression_ctx *ctx = crypto_tfm_ctx(tfm); + + compression_ctx_init(ctx); + + ctx->mode = IAA_MODE_FIXED; + + return 0; +} + +static struct acomp_alg iaa_acomp_fixed_deflate = { + .init = iaa_comp_init_fixed, + .compress = iaa_comp_acompress, + .decompress = iaa_comp_adecompress, + .base = { + .cra_name = "deflate", + .cra_driver_name = "deflate-iaa", + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_ctxsize = sizeof(struct iaa_compression_ctx), + .cra_reqsize = sizeof(u32), + .cra_module = THIS_MODULE, + .cra_priority = IAA_ALG_PRIORITY, + } +}; + +static int iaa_register_compression_device(void) +{ + int ret; + + ret = crypto_register_acomp(&iaa_acomp_fixed_deflate); + if (ret) { + pr_err("deflate algorithm acomp fixed registration failed (%d)\n", ret); + goto out; + } + + iaa_crypto_registered = true; +out: + return ret; +} + +static int iaa_unregister_compression_device(void) +{ + if (iaa_crypto_registered) + crypto_unregister_acomp(&iaa_acomp_fixed_deflate); + + return 0; +} + +static int iaa_crypto_probe(struct idxd_dev *idxd_dev) +{ + struct idxd_wq *wq = idxd_dev_to_wq(idxd_dev); + struct idxd_device *idxd = wq->idxd; + struct idxd_driver_data *data = idxd->data; + struct device *dev = &idxd_dev->conf_dev; + bool first_wq = false; + int ret = 0; + + if (idxd->state != IDXD_DEV_ENABLED) + return -ENXIO; + + if (data->type != IDXD_TYPE_IAX) + return -ENODEV; + + mutex_lock(&wq->wq_lock); + + if (idxd_wq_get_private(wq)) { + mutex_unlock(&wq->wq_lock); + return -EBUSY; + } + + if (!idxd_wq_driver_name_match(wq, dev)) { + dev_dbg(dev, "wq %d.%d driver_name match failed: wq driver_name %s, dev driver name %s\n", + idxd->id, wq->id, wq->driver_name, dev->driver->name); + idxd->cmd_status = IDXD_SCMD_WQ_NO_DRV_NAME; + ret = -ENODEV; + goto err; + } + + wq->type = IDXD_WQT_KERNEL; + + ret = idxd_drv_enable_wq(wq); + if (ret < 0) { + dev_dbg(dev, "enable wq %d.%d failed: %d\n", + idxd->id, wq->id, ret); + ret = -ENXIO; + goto err; + } + + mutex_lock(&iaa_devices_lock); + + if (list_empty(&iaa_devices)) { + ret = alloc_wq_table(wq->idxd->max_wqs); + if (ret) + goto err_alloc; + first_wq = true; + } + + ret = save_iaa_wq(wq); + if (ret) + goto err_save; + + rebalance_wq_table(); + + if (first_wq) { + iaa_crypto_enabled = true; + ret = iaa_register_compression_device(); + if (ret != 0) { + iaa_crypto_enabled = false; + dev_dbg(dev, "IAA compression device registration failed\n"); + goto err_register; + } + try_module_get(THIS_MODULE); + + pr_info("iaa_crypto now ENABLED\n"); + } + + mutex_unlock(&iaa_devices_lock); +out: + mutex_unlock(&wq->wq_lock); + + return ret; + +err_register: + remove_iaa_wq(wq); + free_iaa_wq(idxd_wq_get_private(wq)); +err_save: + if (first_wq) + free_wq_table(); +err_alloc: + mutex_unlock(&iaa_devices_lock); + idxd_drv_disable_wq(wq); +err: + wq->type = IDXD_WQT_NONE; + + goto out; +} + +static void iaa_crypto_remove(struct idxd_dev *idxd_dev) +{ + struct idxd_wq *wq = idxd_dev_to_wq(idxd_dev); + struct idxd_device *idxd = wq->idxd; + struct iaa_wq *iaa_wq; + bool free = false; + + idxd_wq_quiesce(wq); + + mutex_lock(&wq->wq_lock); + mutex_lock(&iaa_devices_lock); + + remove_iaa_wq(wq); + + spin_lock(&idxd->dev_lock); + iaa_wq = idxd_wq_get_private(wq); + if (!iaa_wq) { + spin_unlock(&idxd->dev_lock); + pr_err("%s: no iaa_wq available to remove\n", __func__); + goto out; + } + + if (iaa_wq->ref) { + iaa_wq->remove = true; + } else { + wq = iaa_wq->wq; + idxd_wq_set_private(wq, NULL); + free = true; + } + spin_unlock(&idxd->dev_lock); + if (free) { + __free_iaa_wq(iaa_wq); + kfree(iaa_wq); + } + + idxd_drv_disable_wq(wq); + rebalance_wq_table(); + + if (nr_iaa == 0) { + iaa_crypto_enabled = false; + free_wq_table(); + module_put(THIS_MODULE); + + pr_info("iaa_crypto now DISABLED\n"); + } +out: + mutex_unlock(&iaa_devices_lock); + mutex_unlock(&wq->wq_lock); +} + +static enum idxd_dev_type dev_types[] = { + IDXD_DEV_WQ, + IDXD_DEV_NONE, +}; + +static struct idxd_device_driver iaa_crypto_driver = { + .probe = iaa_crypto_probe, + .remove = iaa_crypto_remove, + .name = IDXD_SUBDRIVER_NAME, + .type = dev_types, + .desc_complete = iaa_desc_complete, +}; + +static int __init iaa_crypto_init_module(void) +{ + int ret = 0; + int node; + + nr_cpus = num_possible_cpus(); + for_each_node_with_cpus(node) + nr_nodes++; + if (!nr_nodes) { + pr_err("IAA couldn't find any nodes with cpus\n"); + return -ENODEV; + } + nr_cpus_per_node = nr_cpus / nr_nodes; + + ret = iaa_aecs_init_fixed(); + if (ret < 0) { + pr_debug("IAA fixed compression mode init failed\n"); + goto err_aecs_init; + } + + ret = idxd_driver_register(&iaa_crypto_driver); + if (ret) { + pr_debug("IAA wq sub-driver registration failed\n"); + goto err_driver_reg; + } + + ret = driver_create_file(&iaa_crypto_driver.drv, + &driver_attr_verify_compress); + if (ret) { + pr_debug("IAA verify_compress attr creation failed\n"); + goto err_verify_attr_create; + } + + ret = driver_create_file(&iaa_crypto_driver.drv, + &driver_attr_sync_mode); + if (ret) { + pr_debug("IAA sync mode attr creation failed\n"); + goto err_sync_attr_create; + } + + if (iaa_crypto_debugfs_init()) + pr_warn("debugfs init failed, stats not available\n"); + + pr_debug("initialized\n"); +out: + return ret; + +err_sync_attr_create: + driver_remove_file(&iaa_crypto_driver.drv, + &driver_attr_verify_compress); +err_verify_attr_create: + idxd_driver_unregister(&iaa_crypto_driver); +err_driver_reg: + iaa_aecs_cleanup_fixed(); +err_aecs_init: + + goto out; +} + +static void __exit iaa_crypto_cleanup_module(void) +{ + if (iaa_unregister_compression_device()) + pr_debug("IAA compression device unregister failed\n"); + + iaa_crypto_debugfs_cleanup(); + driver_remove_file(&iaa_crypto_driver.drv, + &driver_attr_sync_mode); + driver_remove_file(&iaa_crypto_driver.drv, + &driver_attr_verify_compress); + idxd_driver_unregister(&iaa_crypto_driver); + iaa_aecs_cleanup_fixed(); + + pr_debug("cleaned up\n"); +} + +MODULE_IMPORT_NS("IDXD"); +MODULE_LICENSE("GPL"); +MODULE_ALIAS_IDXD_DEVICE(0); +MODULE_AUTHOR("Intel Corporation"); +MODULE_DESCRIPTION("IAA Compression Accelerator Crypto Driver"); + +module_init(iaa_crypto_init_module); +module_exit(iaa_crypto_cleanup_module); diff --git a/drivers/crypto/intel/iaa/iaa_crypto_stats.c b/drivers/crypto/intel/iaa/iaa_crypto_stats.c new file mode 100644 index 000000000000..f5cc3d29ca19 --- /dev/null +++ b/drivers/crypto/intel/iaa/iaa_crypto_stats.c @@ -0,0 +1,271 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright(c) 2021 Intel Corporation. All rights rsvd. */ + +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/highmem.h> +#include <linux/mm.h> +#include <linux/slab.h> +#include <linux/delay.h> +#include <linux/smp.h> +#include <uapi/linux/idxd.h> +#include <linux/idxd.h> +#include <linux/dmaengine.h> +#include "../../dma/idxd/idxd.h" +#include <linux/debugfs.h> +#include <crypto/internal/acompress.h> +#include "iaa_crypto.h" +#include "iaa_crypto_stats.h" + +static atomic64_t total_comp_calls; +static atomic64_t total_decomp_calls; +static atomic64_t total_sw_decomp_calls; +static atomic64_t total_comp_bytes_out; +static atomic64_t total_decomp_bytes_in; +static atomic64_t total_completion_einval_errors; +static atomic64_t total_completion_timeout_errors; +static atomic64_t total_completion_comp_buf_overflow_errors; + +static struct dentry *iaa_crypto_debugfs_root; + +void update_total_comp_calls(void) +{ + atomic64_inc(&total_comp_calls); +} + +void update_total_comp_bytes_out(int n) +{ + atomic64_add(n, &total_comp_bytes_out); +} + +void update_total_decomp_calls(void) +{ + atomic64_inc(&total_decomp_calls); +} + +void update_total_sw_decomp_calls(void) +{ + atomic64_inc(&total_sw_decomp_calls); +} + +void update_total_decomp_bytes_in(int n) +{ + atomic64_add(n, &total_decomp_bytes_in); +} + +void update_completion_einval_errs(void) +{ + atomic64_inc(&total_completion_einval_errors); +} + +void update_completion_timeout_errs(void) +{ + atomic64_inc(&total_completion_timeout_errors); +} + +void update_completion_comp_buf_overflow_errs(void) +{ + atomic64_inc(&total_completion_comp_buf_overflow_errors); +} + +void update_wq_comp_calls(struct idxd_wq *idxd_wq) +{ + struct iaa_wq *wq = idxd_wq_get_private(idxd_wq); + + atomic64_inc(&wq->comp_calls); + atomic64_inc(&wq->iaa_device->comp_calls); +} + +void update_wq_comp_bytes(struct idxd_wq *idxd_wq, int n) +{ + struct iaa_wq *wq = idxd_wq_get_private(idxd_wq); + + atomic64_add(n, &wq->comp_bytes); + atomic64_add(n, &wq->iaa_device->comp_bytes); +} + +void update_wq_decomp_calls(struct idxd_wq *idxd_wq) +{ + struct iaa_wq *wq = idxd_wq_get_private(idxd_wq); + + atomic64_inc(&wq->decomp_calls); + atomic64_inc(&wq->iaa_device->decomp_calls); +} + +void update_wq_decomp_bytes(struct idxd_wq *idxd_wq, int n) +{ + struct iaa_wq *wq = idxd_wq_get_private(idxd_wq); + + atomic64_add(n, &wq->decomp_bytes); + atomic64_add(n, &wq->iaa_device->decomp_bytes); +} + +static void reset_iaa_crypto_stats(void) +{ + atomic64_set(&total_comp_calls, 0); + atomic64_set(&total_decomp_calls, 0); + atomic64_set(&total_sw_decomp_calls, 0); + atomic64_set(&total_comp_bytes_out, 0); + atomic64_set(&total_decomp_bytes_in, 0); + atomic64_set(&total_completion_einval_errors, 0); + atomic64_set(&total_completion_timeout_errors, 0); + atomic64_set(&total_completion_comp_buf_overflow_errors, 0); +} + +static void reset_wq_stats(struct iaa_wq *wq) +{ + atomic64_set(&wq->comp_calls, 0); + atomic64_set(&wq->comp_bytes, 0); + atomic64_set(&wq->decomp_calls, 0); + atomic64_set(&wq->decomp_bytes, 0); +} + +static void reset_device_stats(struct iaa_device *iaa_device) +{ + struct iaa_wq *iaa_wq; + + atomic64_set(&iaa_device->comp_calls, 0); + atomic64_set(&iaa_device->comp_bytes, 0); + atomic64_set(&iaa_device->decomp_calls, 0); + atomic64_set(&iaa_device->decomp_bytes, 0); + + list_for_each_entry(iaa_wq, &iaa_device->wqs, list) + reset_wq_stats(iaa_wq); +} + +static void wq_show(struct seq_file *m, struct iaa_wq *iaa_wq) +{ + seq_printf(m, " name: %s\n", iaa_wq->wq->name); + seq_printf(m, " comp_calls: %llu\n", + atomic64_read(&iaa_wq->comp_calls)); + seq_printf(m, " comp_bytes: %llu\n", + atomic64_read(&iaa_wq->comp_bytes)); + seq_printf(m, " decomp_calls: %llu\n", + atomic64_read(&iaa_wq->decomp_calls)); + seq_printf(m, " decomp_bytes: %llu\n\n", + atomic64_read(&iaa_wq->decomp_bytes)); +} + +static void device_stats_show(struct seq_file *m, struct iaa_device *iaa_device) +{ + struct iaa_wq *iaa_wq; + + seq_puts(m, "iaa device:\n"); + seq_printf(m, " id: %d\n", iaa_device->idxd->id); + seq_printf(m, " n_wqs: %d\n", iaa_device->n_wq); + seq_printf(m, " comp_calls: %llu\n", + atomic64_read(&iaa_device->comp_calls)); + seq_printf(m, " comp_bytes: %llu\n", + atomic64_read(&iaa_device->comp_bytes)); + seq_printf(m, " decomp_calls: %llu\n", + atomic64_read(&iaa_device->decomp_calls)); + seq_printf(m, " decomp_bytes: %llu\n", + atomic64_read(&iaa_device->decomp_bytes)); + seq_puts(m, " wqs:\n"); + + list_for_each_entry(iaa_wq, &iaa_device->wqs, list) + wq_show(m, iaa_wq); +} + +static int global_stats_show(struct seq_file *m, void *v) +{ + seq_puts(m, "global stats:\n"); + seq_printf(m, " total_comp_calls: %llu\n", + atomic64_read(&total_comp_calls)); + seq_printf(m, " total_decomp_calls: %llu\n", + atomic64_read(&total_decomp_calls)); + seq_printf(m, " total_sw_decomp_calls: %llu\n", + atomic64_read(&total_sw_decomp_calls)); + seq_printf(m, " total_comp_bytes_out: %llu\n", + atomic64_read(&total_comp_bytes_out)); + seq_printf(m, " total_decomp_bytes_in: %llu\n", + atomic64_read(&total_decomp_bytes_in)); + seq_printf(m, " total_completion_einval_errors: %llu\n", + atomic64_read(&total_completion_einval_errors)); + seq_printf(m, " total_completion_timeout_errors: %llu\n", + atomic64_read(&total_completion_timeout_errors)); + seq_printf(m, " total_completion_comp_buf_overflow_errors: %llu\n\n", + atomic64_read(&total_completion_comp_buf_overflow_errors)); + + return 0; +} + +static int wq_stats_show(struct seq_file *m, void *v) +{ + struct iaa_device *iaa_device; + + mutex_lock(&iaa_devices_lock); + + list_for_each_entry(iaa_device, &iaa_devices, list) + device_stats_show(m, iaa_device); + + mutex_unlock(&iaa_devices_lock); + + return 0; +} + +static int iaa_crypto_stats_reset(void *data, u64 value) +{ + struct iaa_device *iaa_device; + + reset_iaa_crypto_stats(); + + mutex_lock(&iaa_devices_lock); + + list_for_each_entry(iaa_device, &iaa_devices, list) + reset_device_stats(iaa_device); + + mutex_unlock(&iaa_devices_lock); + + return 0; +} + +static int wq_stats_open(struct inode *inode, struct file *file) +{ + return single_open(file, wq_stats_show, file); +} + +static const struct file_operations wq_stats_fops = { + .open = wq_stats_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int global_stats_open(struct inode *inode, struct file *file) +{ + return single_open(file, global_stats_show, file); +} + +static const struct file_operations global_stats_fops = { + .open = global_stats_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +DEFINE_DEBUGFS_ATTRIBUTE(wq_stats_reset_fops, NULL, iaa_crypto_stats_reset, "%llu\n"); + +int __init iaa_crypto_debugfs_init(void) +{ + if (!debugfs_initialized()) + return -ENODEV; + + iaa_crypto_debugfs_root = debugfs_create_dir("iaa_crypto", NULL); + + debugfs_create_file("global_stats", 0644, iaa_crypto_debugfs_root, NULL, + &global_stats_fops); + debugfs_create_file("wq_stats", 0644, iaa_crypto_debugfs_root, NULL, + &wq_stats_fops); + debugfs_create_file("stats_reset", 0644, iaa_crypto_debugfs_root, NULL, + &wq_stats_reset_fops); + + return 0; +} + +void __exit iaa_crypto_debugfs_cleanup(void) +{ + debugfs_remove_recursive(iaa_crypto_debugfs_root); +} + +MODULE_LICENSE("GPL"); diff --git a/drivers/crypto/intel/iaa/iaa_crypto_stats.h b/drivers/crypto/intel/iaa/iaa_crypto_stats.h new file mode 100644 index 000000000000..3787a5f507eb --- /dev/null +++ b/drivers/crypto/intel/iaa/iaa_crypto_stats.h @@ -0,0 +1,45 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Copyright(c) 2021 Intel Corporation. All rights rsvd. */ + +#ifndef __CRYPTO_DEV_IAA_CRYPTO_STATS_H__ +#define __CRYPTO_DEV_IAA_CRYPTO_STATS_H__ + +#if defined(CONFIG_CRYPTO_DEV_IAA_CRYPTO_STATS) +int iaa_crypto_debugfs_init(void); +void iaa_crypto_debugfs_cleanup(void); + +void update_total_comp_calls(void); +void update_total_comp_bytes_out(int n); +void update_total_decomp_calls(void); +void update_total_sw_decomp_calls(void); +void update_total_decomp_bytes_in(int n); +void update_completion_einval_errs(void); +void update_completion_timeout_errs(void); +void update_completion_comp_buf_overflow_errs(void); + +void update_wq_comp_calls(struct idxd_wq *idxd_wq); +void update_wq_comp_bytes(struct idxd_wq *idxd_wq, int n); +void update_wq_decomp_calls(struct idxd_wq *idxd_wq); +void update_wq_decomp_bytes(struct idxd_wq *idxd_wq, int n); + +#else +static inline int iaa_crypto_debugfs_init(void) { return 0; } +static inline void iaa_crypto_debugfs_cleanup(void) {} + +static inline void update_total_comp_calls(void) {} +static inline void update_total_comp_bytes_out(int n) {} +static inline void update_total_decomp_calls(void) {} +static inline void update_total_sw_decomp_calls(void) {} +static inline void update_total_decomp_bytes_in(int n) {} +static inline void update_completion_einval_errs(void) {} +static inline void update_completion_timeout_errs(void) {} +static inline void update_completion_comp_buf_overflow_errs(void) {} + +static inline void update_wq_comp_calls(struct idxd_wq *idxd_wq) {} +static inline void update_wq_comp_bytes(struct idxd_wq *idxd_wq, int n) {} +static inline void update_wq_decomp_calls(struct idxd_wq *idxd_wq) {} +static inline void update_wq_decomp_bytes(struct idxd_wq *idxd_wq, int n) {} + +#endif // CONFIG_CRYPTO_DEV_IAA_CRYPTO_STATS + +#endif diff --git a/drivers/crypto/intel/ixp4xx/Kconfig b/drivers/crypto/intel/ixp4xx/Kconfig new file mode 100644 index 000000000000..af3cc5688328 --- /dev/null +++ b/drivers/crypto/intel/ixp4xx/Kconfig @@ -0,0 +1,14 @@ +config CRYPTO_DEV_IXP4XX + tristate "Driver for IXP4xx crypto hardware acceleration" + depends on (ARCH_IXP4XX || COMPILE_TEST) && IXP4XX_QMGR && IXP4XX_NPE + select CRYPTO_AES + select CRYPTO_DES + select CRYPTO_ECB + select CRYPTO_CBC + select CRYPTO_CTR + select CRYPTO_LIB_DES + select CRYPTO_AEAD + select CRYPTO_AUTHENC + select CRYPTO_SKCIPHER + help + Driver for the IXP4xx NPE crypto engine. diff --git a/drivers/crypto/intel/ixp4xx/Makefile b/drivers/crypto/intel/ixp4xx/Makefile new file mode 100644 index 000000000000..74ebefd93046 --- /dev/null +++ b/drivers/crypto/intel/ixp4xx/Makefile @@ -0,0 +1,2 @@ +# SPDX-License-Identifier: GPL-2.0 +obj-$(CONFIG_CRYPTO_DEV_IXP4XX) += ixp4xx_crypto.o diff --git a/drivers/crypto/intel/ixp4xx/ixp4xx_crypto.c b/drivers/crypto/intel/ixp4xx/ixp4xx_crypto.c new file mode 100644 index 000000000000..fcc0cf4df637 --- /dev/null +++ b/drivers/crypto/intel/ixp4xx/ixp4xx_crypto.c @@ -0,0 +1,1605 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Intel IXP4xx NPE-C crypto driver + * + * Copyright (C) 2008 Christian Hohnstaedt <chohnstaedt@innominate.com> + */ + +#include <linux/platform_device.h> +#include <linux/dma-mapping.h> +#include <linux/dmapool.h> +#include <linux/crypto.h> +#include <linux/kernel.h> +#include <linux/rtnetlink.h> +#include <linux/interrupt.h> +#include <linux/spinlock.h> +#include <linux/gfp.h> +#include <linux/module.h> +#include <linux/of.h> + +#include <crypto/ctr.h> +#include <crypto/internal/des.h> +#include <crypto/aes.h> +#include <crypto/hmac.h> +#include <crypto/sha1.h> +#include <crypto/algapi.h> +#include <crypto/internal/aead.h> +#include <crypto/internal/skcipher.h> +#include <crypto/authenc.h> +#include <crypto/scatterwalk.h> + +#include <linux/soc/ixp4xx/npe.h> +#include <linux/soc/ixp4xx/qmgr.h> + +/* Intermittent includes, delete this after v5.14-rc1 */ +#include <linux/soc/ixp4xx/cpu.h> + +#define MAX_KEYLEN 32 + +/* hash: cfgword + 2 * digestlen; crypt: keylen + cfgword */ +#define NPE_CTX_LEN 80 +#define AES_BLOCK128 16 + +#define NPE_OP_HASH_VERIFY 0x01 +#define NPE_OP_CCM_ENABLE 0x04 +#define NPE_OP_CRYPT_ENABLE 0x08 +#define NPE_OP_HASH_ENABLE 0x10 +#define NPE_OP_NOT_IN_PLACE 0x20 +#define NPE_OP_HMAC_DISABLE 0x40 +#define NPE_OP_CRYPT_ENCRYPT 0x80 + +#define NPE_OP_CCM_GEN_MIC 0xcc +#define NPE_OP_HASH_GEN_ICV 0x50 +#define NPE_OP_ENC_GEN_KEY 0xc9 + +#define MOD_ECB 0x0000 +#define MOD_CTR 0x1000 +#define MOD_CBC_ENC 0x2000 +#define MOD_CBC_DEC 0x3000 +#define MOD_CCM_ENC 0x4000 +#define MOD_CCM_DEC 0x5000 + +#define KEYLEN_128 4 +#define KEYLEN_192 6 +#define KEYLEN_256 8 + +#define CIPH_DECR 0x0000 +#define CIPH_ENCR 0x0400 + +#define MOD_DES 0x0000 +#define MOD_TDEA2 0x0100 +#define MOD_3DES 0x0200 +#define MOD_AES 0x0800 +#define MOD_AES128 (0x0800 | KEYLEN_128) +#define MOD_AES192 (0x0900 | KEYLEN_192) +#define MOD_AES256 (0x0a00 | KEYLEN_256) + +#define MAX_IVLEN 16 +#define NPE_QLEN 16 +/* Space for registering when the first + * NPE_QLEN crypt_ctl are busy */ +#define NPE_QLEN_TOTAL 64 + +#define CTL_FLAG_UNUSED 0x0000 +#define CTL_FLAG_USED 0x1000 +#define CTL_FLAG_PERFORM_ABLK 0x0001 +#define CTL_FLAG_GEN_ICV 0x0002 +#define CTL_FLAG_GEN_REVAES 0x0004 +#define CTL_FLAG_PERFORM_AEAD 0x0008 +#define CTL_FLAG_MASK 0x000f + +#define HMAC_PAD_BLOCKLEN SHA1_BLOCK_SIZE + +#define MD5_DIGEST_SIZE 16 + +struct buffer_desc { + u32 phys_next; +#ifdef __ARMEB__ + u16 buf_len; + u16 pkt_len; +#else + u16 pkt_len; + u16 buf_len; +#endif + dma_addr_t phys_addr; + u32 __reserved[4]; + struct buffer_desc *next; + enum dma_data_direction dir; +}; + +struct crypt_ctl { +#ifdef __ARMEB__ + u8 mode; /* NPE_OP_* operation mode */ + u8 init_len; + u16 reserved; +#else + u16 reserved; + u8 init_len; + u8 mode; /* NPE_OP_* operation mode */ +#endif + u8 iv[MAX_IVLEN]; /* IV for CBC mode or CTR IV for CTR mode */ + u32 icv_rev_aes; /* icv or rev aes */ + u32 src_buf; + u32 dst_buf; +#ifdef __ARMEB__ + u16 auth_offs; /* Authentication start offset */ + u16 auth_len; /* Authentication data length */ + u16 crypt_offs; /* Cryption start offset */ + u16 crypt_len; /* Cryption data length */ +#else + u16 auth_len; /* Authentication data length */ + u16 auth_offs; /* Authentication start offset */ + u16 crypt_len; /* Cryption data length */ + u16 crypt_offs; /* Cryption start offset */ +#endif + u32 aadAddr; /* Additional Auth Data Addr for CCM mode */ + u32 crypto_ctx; /* NPE Crypto Param structure address */ + + /* Used by Host: 4*4 bytes*/ + unsigned int ctl_flags; + union { + struct skcipher_request *ablk_req; + struct aead_request *aead_req; + struct crypto_tfm *tfm; + } data; + struct buffer_desc *regist_buf; + u8 *regist_ptr; +}; + +struct ablk_ctx { + struct buffer_desc *src; + struct buffer_desc *dst; + u8 iv[MAX_IVLEN]; + bool encrypt; + struct skcipher_request fallback_req; // keep at the end +}; + +struct aead_ctx { + struct buffer_desc *src; + struct buffer_desc *dst; + struct scatterlist ivlist; + /* used when the hmac is not on one sg entry */ + u8 *hmac_virt; + int encrypt; +}; + +struct ix_hash_algo { + u32 cfgword; + unsigned char *icv; +}; + +struct ix_sa_dir { + unsigned char *npe_ctx; + dma_addr_t npe_ctx_phys; + int npe_ctx_idx; + u8 npe_mode; +}; + +struct ixp_ctx { + struct ix_sa_dir encrypt; + struct ix_sa_dir decrypt; + int authkey_len; + u8 authkey[MAX_KEYLEN]; + int enckey_len; + u8 enckey[MAX_KEYLEN]; + u8 salt[MAX_IVLEN]; + u8 nonce[CTR_RFC3686_NONCE_SIZE]; + unsigned int salted; + atomic_t configuring; + struct completion completion; + struct crypto_skcipher *fallback_tfm; +}; + +struct ixp_alg { + struct skcipher_alg crypto; + const struct ix_hash_algo *hash; + u32 cfg_enc; + u32 cfg_dec; + + int registered; +}; + +struct ixp_aead_alg { + struct aead_alg crypto; + const struct ix_hash_algo *hash; + u32 cfg_enc; + u32 cfg_dec; + + int registered; +}; + +static const struct ix_hash_algo hash_alg_md5 = { + .cfgword = 0xAA010004, + .icv = "\x01\x23\x45\x67\x89\xAB\xCD\xEF" + "\xFE\xDC\xBA\x98\x76\x54\x32\x10", +}; + +static const struct ix_hash_algo hash_alg_sha1 = { + .cfgword = 0x00000005, + .icv = "\x67\x45\x23\x01\xEF\xCD\xAB\x89\x98\xBA" + "\xDC\xFE\x10\x32\x54\x76\xC3\xD2\xE1\xF0", +}; + +static struct npe *npe_c; + +static unsigned int send_qid; +static unsigned int recv_qid; +static struct dma_pool *buffer_pool; +static struct dma_pool *ctx_pool; + +static struct crypt_ctl *crypt_virt; +static dma_addr_t crypt_phys; + +static int support_aes = 1; + +static struct platform_device *pdev; + +static inline dma_addr_t crypt_virt2phys(struct crypt_ctl *virt) +{ + return crypt_phys + (virt - crypt_virt) * sizeof(struct crypt_ctl); +} + +static inline struct crypt_ctl *crypt_phys2virt(dma_addr_t phys) +{ + return crypt_virt + (phys - crypt_phys) / sizeof(struct crypt_ctl); +} + +static inline u32 cipher_cfg_enc(struct crypto_tfm *tfm) +{ + return container_of(tfm->__crt_alg, struct ixp_alg, crypto.base)->cfg_enc; +} + +static inline u32 cipher_cfg_dec(struct crypto_tfm *tfm) +{ + return container_of(tfm->__crt_alg, struct ixp_alg, crypto.base)->cfg_dec; +} + +static inline const struct ix_hash_algo *ix_hash(struct crypto_tfm *tfm) +{ + return container_of(tfm->__crt_alg, struct ixp_alg, crypto.base)->hash; +} + +static int setup_crypt_desc(void) +{ + struct device *dev = &pdev->dev; + + BUILD_BUG_ON(!(IS_ENABLED(CONFIG_COMPILE_TEST) && + IS_ENABLED(CONFIG_64BIT)) && + sizeof(struct crypt_ctl) != 64); + crypt_virt = dma_alloc_coherent(dev, + NPE_QLEN * sizeof(struct crypt_ctl), + &crypt_phys, GFP_ATOMIC); + if (!crypt_virt) + return -ENOMEM; + return 0; +} + +static DEFINE_SPINLOCK(desc_lock); +static struct crypt_ctl *get_crypt_desc(void) +{ + int i; + static int idx; + unsigned long flags; + + spin_lock_irqsave(&desc_lock, flags); + + if (unlikely(!crypt_virt)) + setup_crypt_desc(); + if (unlikely(!crypt_virt)) { + spin_unlock_irqrestore(&desc_lock, flags); + return NULL; + } + i = idx; + if (crypt_virt[i].ctl_flags == CTL_FLAG_UNUSED) { + if (++idx >= NPE_QLEN) + idx = 0; + crypt_virt[i].ctl_flags = CTL_FLAG_USED; + spin_unlock_irqrestore(&desc_lock, flags); + return crypt_virt + i; + } else { + spin_unlock_irqrestore(&desc_lock, flags); + return NULL; + } +} + +static DEFINE_SPINLOCK(emerg_lock); +static struct crypt_ctl *get_crypt_desc_emerg(void) +{ + int i; + static int idx = NPE_QLEN; + struct crypt_ctl *desc; + unsigned long flags; + + desc = get_crypt_desc(); + if (desc) + return desc; + if (unlikely(!crypt_virt)) + return NULL; + + spin_lock_irqsave(&emerg_lock, flags); + i = idx; + if (crypt_virt[i].ctl_flags == CTL_FLAG_UNUSED) { + if (++idx >= NPE_QLEN_TOTAL) + idx = NPE_QLEN; + crypt_virt[i].ctl_flags = CTL_FLAG_USED; + spin_unlock_irqrestore(&emerg_lock, flags); + return crypt_virt + i; + } else { + spin_unlock_irqrestore(&emerg_lock, flags); + return NULL; + } +} + +static void free_buf_chain(struct device *dev, struct buffer_desc *buf, + dma_addr_t phys) +{ + while (buf) { + struct buffer_desc *buf1; + u32 phys1; + + buf1 = buf->next; + phys1 = buf->phys_next; + dma_unmap_single(dev, buf->phys_addr, buf->buf_len, buf->dir); + dma_pool_free(buffer_pool, buf, phys); + buf = buf1; + phys = phys1; + } +} + +static struct tasklet_struct crypto_done_tasklet; + +static void finish_scattered_hmac(struct crypt_ctl *crypt) +{ + struct aead_request *req = crypt->data.aead_req; + struct aead_ctx *req_ctx = aead_request_ctx(req); + struct crypto_aead *tfm = crypto_aead_reqtfm(req); + int authsize = crypto_aead_authsize(tfm); + int decryptlen = req->assoclen + req->cryptlen - authsize; + + if (req_ctx->encrypt) { + scatterwalk_map_and_copy(req_ctx->hmac_virt, req->dst, + decryptlen, authsize, 1); + } + dma_pool_free(buffer_pool, req_ctx->hmac_virt, crypt->icv_rev_aes); +} + +static void one_packet(dma_addr_t phys) +{ + struct device *dev = &pdev->dev; + struct crypt_ctl *crypt; + struct ixp_ctx *ctx; + int failed; + + failed = phys & 0x1 ? -EBADMSG : 0; + phys &= ~0x3; + crypt = crypt_phys2virt(phys); + + switch (crypt->ctl_flags & CTL_FLAG_MASK) { + case CTL_FLAG_PERFORM_AEAD: { + struct aead_request *req = crypt->data.aead_req; + struct aead_ctx *req_ctx = aead_request_ctx(req); + + free_buf_chain(dev, req_ctx->src, crypt->src_buf); + free_buf_chain(dev, req_ctx->dst, crypt->dst_buf); + if (req_ctx->hmac_virt) + finish_scattered_hmac(crypt); + + aead_request_complete(req, failed); + break; + } + case CTL_FLAG_PERFORM_ABLK: { + struct skcipher_request *req = crypt->data.ablk_req; + struct ablk_ctx *req_ctx = skcipher_request_ctx(req); + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + unsigned int ivsize = crypto_skcipher_ivsize(tfm); + unsigned int offset; + + if (ivsize > 0) { + offset = req->cryptlen - ivsize; + if (req_ctx->encrypt) { + scatterwalk_map_and_copy(req->iv, req->dst, + offset, ivsize, 0); + } else { + memcpy(req->iv, req_ctx->iv, ivsize); + memzero_explicit(req_ctx->iv, ivsize); + } + } + + if (req_ctx->dst) + free_buf_chain(dev, req_ctx->dst, crypt->dst_buf); + + free_buf_chain(dev, req_ctx->src, crypt->src_buf); + skcipher_request_complete(req, failed); + break; + } + case CTL_FLAG_GEN_ICV: + ctx = crypto_tfm_ctx(crypt->data.tfm); + dma_pool_free(ctx_pool, crypt->regist_ptr, + crypt->regist_buf->phys_addr); + dma_pool_free(buffer_pool, crypt->regist_buf, crypt->src_buf); + if (atomic_dec_and_test(&ctx->configuring)) + complete(&ctx->completion); + break; + case CTL_FLAG_GEN_REVAES: + ctx = crypto_tfm_ctx(crypt->data.tfm); + *(__be32 *)ctx->decrypt.npe_ctx &= cpu_to_be32(~CIPH_ENCR); + if (atomic_dec_and_test(&ctx->configuring)) + complete(&ctx->completion); + break; + default: + BUG(); + } + crypt->ctl_flags = CTL_FLAG_UNUSED; +} + +static void irqhandler(void *_unused) +{ + tasklet_schedule(&crypto_done_tasklet); +} + +static void crypto_done_action(unsigned long arg) +{ + int i; + + for (i = 0; i < 4; i++) { + dma_addr_t phys = qmgr_get_entry(recv_qid); + if (!phys) + return; + one_packet(phys); + } + tasklet_schedule(&crypto_done_tasklet); +} + +static int init_ixp_crypto(struct device *dev) +{ + struct device_node *np = dev->of_node; + u32 msg[2] = { 0, 0 }; + int ret = -ENODEV; + u32 npe_id; + + dev_info(dev, "probing...\n"); + + /* Locate the NPE and queue manager to use from device tree */ + if (IS_ENABLED(CONFIG_OF) && np) { + struct of_phandle_args queue_spec; + struct of_phandle_args npe_spec; + + ret = of_parse_phandle_with_fixed_args(np, "intel,npe-handle", + 1, 0, &npe_spec); + if (ret) { + dev_err(dev, "no NPE engine specified\n"); + return -ENODEV; + } + npe_id = npe_spec.args[0]; + of_node_put(npe_spec.np); + + ret = of_parse_phandle_with_fixed_args(np, "queue-rx", 1, 0, + &queue_spec); + if (ret) { + dev_err(dev, "no rx queue phandle\n"); + return -ENODEV; + } + recv_qid = queue_spec.args[0]; + of_node_put(queue_spec.np); + + ret = of_parse_phandle_with_fixed_args(np, "queue-txready", 1, 0, + &queue_spec); + if (ret) { + dev_err(dev, "no txready queue phandle\n"); + return -ENODEV; + } + send_qid = queue_spec.args[0]; + of_node_put(queue_spec.np); + } else { + /* + * Hardcoded engine when using platform data, this goes away + * when we switch to using DT only. + */ + npe_id = 2; + send_qid = 29; + recv_qid = 30; + } + + npe_c = npe_request(npe_id); + if (!npe_c) + return ret; + + if (!npe_running(npe_c)) { + ret = npe_load_firmware(npe_c, npe_name(npe_c), dev); + if (ret) + goto npe_release; + if (npe_recv_message(npe_c, msg, "STATUS_MSG")) + goto npe_error; + } else { + if (npe_send_message(npe_c, msg, "STATUS_MSG")) + goto npe_error; + + if (npe_recv_message(npe_c, msg, "STATUS_MSG")) + goto npe_error; + } + + switch ((msg[1] >> 16) & 0xff) { + case 3: + dev_warn(dev, "Firmware of %s lacks AES support\n", npe_name(npe_c)); + support_aes = 0; + break; + case 4: + case 5: + support_aes = 1; + break; + default: + dev_err(dev, "Firmware of %s lacks crypto support\n", npe_name(npe_c)); + ret = -ENODEV; + goto npe_release; + } + /* buffer_pool will also be used to sometimes store the hmac, + * so assure it is large enough + */ + BUILD_BUG_ON(SHA1_DIGEST_SIZE > sizeof(struct buffer_desc)); + buffer_pool = dma_pool_create("buffer", dev, sizeof(struct buffer_desc), + 32, 0); + ret = -ENOMEM; + if (!buffer_pool) + goto err; + + ctx_pool = dma_pool_create("context", dev, NPE_CTX_LEN, 16, 0); + if (!ctx_pool) + goto err; + + ret = qmgr_request_queue(send_qid, NPE_QLEN_TOTAL, 0, 0, + "ixp_crypto:out", NULL); + if (ret) + goto err; + ret = qmgr_request_queue(recv_qid, NPE_QLEN, 0, 0, + "ixp_crypto:in", NULL); + if (ret) { + qmgr_release_queue(send_qid); + goto err; + } + qmgr_set_irq(recv_qid, QUEUE_IRQ_SRC_NOT_EMPTY, irqhandler, NULL); + tasklet_init(&crypto_done_tasklet, crypto_done_action, 0); + + qmgr_enable_irq(recv_qid); + return 0; + +npe_error: + dev_err(dev, "%s not responding\n", npe_name(npe_c)); + ret = -EIO; +err: + dma_pool_destroy(ctx_pool); + dma_pool_destroy(buffer_pool); +npe_release: + npe_release(npe_c); + return ret; +} + +static void release_ixp_crypto(struct device *dev) +{ + qmgr_disable_irq(recv_qid); + tasklet_kill(&crypto_done_tasklet); + + qmgr_release_queue(send_qid); + qmgr_release_queue(recv_qid); + + dma_pool_destroy(ctx_pool); + dma_pool_destroy(buffer_pool); + + npe_release(npe_c); + + if (crypt_virt) + dma_free_coherent(dev, NPE_QLEN * sizeof(struct crypt_ctl), + crypt_virt, crypt_phys); +} + +static void reset_sa_dir(struct ix_sa_dir *dir) +{ + memset(dir->npe_ctx, 0, NPE_CTX_LEN); + dir->npe_ctx_idx = 0; + dir->npe_mode = 0; +} + +static int init_sa_dir(struct ix_sa_dir *dir) +{ + dir->npe_ctx = dma_pool_alloc(ctx_pool, GFP_KERNEL, &dir->npe_ctx_phys); + if (!dir->npe_ctx) + return -ENOMEM; + + reset_sa_dir(dir); + return 0; +} + +static void free_sa_dir(struct ix_sa_dir *dir) +{ + memset(dir->npe_ctx, 0, NPE_CTX_LEN); + dma_pool_free(ctx_pool, dir->npe_ctx, dir->npe_ctx_phys); +} + +static int init_tfm(struct crypto_tfm *tfm) +{ + struct ixp_ctx *ctx = crypto_tfm_ctx(tfm); + int ret; + + atomic_set(&ctx->configuring, 0); + ret = init_sa_dir(&ctx->encrypt); + if (ret) + return ret; + ret = init_sa_dir(&ctx->decrypt); + if (ret) + free_sa_dir(&ctx->encrypt); + + return ret; +} + +static int init_tfm_ablk(struct crypto_skcipher *tfm) +{ + struct crypto_tfm *ctfm = crypto_skcipher_tfm(tfm); + struct ixp_ctx *ctx = crypto_tfm_ctx(ctfm); + const char *name = crypto_tfm_alg_name(ctfm); + + ctx->fallback_tfm = crypto_alloc_skcipher(name, 0, CRYPTO_ALG_NEED_FALLBACK); + if (IS_ERR(ctx->fallback_tfm)) { + pr_err("ERROR: Cannot allocate fallback for %s %ld\n", + name, PTR_ERR(ctx->fallback_tfm)); + return PTR_ERR(ctx->fallback_tfm); + } + + pr_info("Fallback for %s is %s\n", + crypto_tfm_alg_driver_name(&tfm->base), + crypto_tfm_alg_driver_name(crypto_skcipher_tfm(ctx->fallback_tfm)) + ); + + crypto_skcipher_set_reqsize(tfm, sizeof(struct ablk_ctx) + crypto_skcipher_reqsize(ctx->fallback_tfm)); + return init_tfm(crypto_skcipher_tfm(tfm)); +} + +static int init_tfm_aead(struct crypto_aead *tfm) +{ + crypto_aead_set_reqsize(tfm, sizeof(struct aead_ctx)); + return init_tfm(crypto_aead_tfm(tfm)); +} + +static void exit_tfm(struct crypto_tfm *tfm) +{ + struct ixp_ctx *ctx = crypto_tfm_ctx(tfm); + + free_sa_dir(&ctx->encrypt); + free_sa_dir(&ctx->decrypt); +} + +static void exit_tfm_ablk(struct crypto_skcipher *tfm) +{ + struct crypto_tfm *ctfm = crypto_skcipher_tfm(tfm); + struct ixp_ctx *ctx = crypto_tfm_ctx(ctfm); + + crypto_free_skcipher(ctx->fallback_tfm); + exit_tfm(crypto_skcipher_tfm(tfm)); +} + +static void exit_tfm_aead(struct crypto_aead *tfm) +{ + exit_tfm(crypto_aead_tfm(tfm)); +} + +static int register_chain_var(struct crypto_tfm *tfm, u8 xpad, u32 target, + int init_len, u32 ctx_addr, const u8 *key, + int key_len) +{ + struct ixp_ctx *ctx = crypto_tfm_ctx(tfm); + struct crypt_ctl *crypt; + struct buffer_desc *buf; + int i; + u8 *pad; + dma_addr_t pad_phys, buf_phys; + + BUILD_BUG_ON(NPE_CTX_LEN < HMAC_PAD_BLOCKLEN); + pad = dma_pool_alloc(ctx_pool, GFP_KERNEL, &pad_phys); + if (!pad) + return -ENOMEM; + buf = dma_pool_alloc(buffer_pool, GFP_KERNEL, &buf_phys); + if (!buf) { + dma_pool_free(ctx_pool, pad, pad_phys); + return -ENOMEM; + } + crypt = get_crypt_desc_emerg(); + if (!crypt) { + dma_pool_free(ctx_pool, pad, pad_phys); + dma_pool_free(buffer_pool, buf, buf_phys); + return -EAGAIN; + } + + memcpy(pad, key, key_len); + memset(pad + key_len, 0, HMAC_PAD_BLOCKLEN - key_len); + for (i = 0; i < HMAC_PAD_BLOCKLEN; i++) + pad[i] ^= xpad; + + crypt->data.tfm = tfm; + crypt->regist_ptr = pad; + crypt->regist_buf = buf; + + crypt->auth_offs = 0; + crypt->auth_len = HMAC_PAD_BLOCKLEN; + crypt->crypto_ctx = ctx_addr; + crypt->src_buf = buf_phys; + crypt->icv_rev_aes = target; + crypt->mode = NPE_OP_HASH_GEN_ICV; + crypt->init_len = init_len; + crypt->ctl_flags |= CTL_FLAG_GEN_ICV; + + buf->next = NULL; + buf->buf_len = HMAC_PAD_BLOCKLEN; + buf->pkt_len = 0; + buf->phys_addr = pad_phys; + + atomic_inc(&ctx->configuring); + qmgr_put_entry(send_qid, crypt_virt2phys(crypt)); + BUG_ON(qmgr_stat_overflow(send_qid)); + return 0; +} + +static int setup_auth(struct crypto_tfm *tfm, int encrypt, unsigned int authsize, + const u8 *key, int key_len, unsigned int digest_len) +{ + u32 itarget, otarget, npe_ctx_addr; + unsigned char *cinfo; + int init_len, ret = 0; + u32 cfgword; + struct ix_sa_dir *dir; + struct ixp_ctx *ctx = crypto_tfm_ctx(tfm); + const struct ix_hash_algo *algo; + + dir = encrypt ? &ctx->encrypt : &ctx->decrypt; + cinfo = dir->npe_ctx + dir->npe_ctx_idx; + algo = ix_hash(tfm); + + /* write cfg word to cryptinfo */ + cfgword = algo->cfgword | (authsize << 6); /* (authsize/4) << 8 */ +#ifndef __ARMEB__ + cfgword ^= 0xAA000000; /* change the "byte swap" flags */ +#endif + *(__be32 *)cinfo = cpu_to_be32(cfgword); + cinfo += sizeof(cfgword); + + /* write ICV to cryptinfo */ + memcpy(cinfo, algo->icv, digest_len); + cinfo += digest_len; + + itarget = dir->npe_ctx_phys + dir->npe_ctx_idx + + sizeof(algo->cfgword); + otarget = itarget + digest_len; + init_len = cinfo - (dir->npe_ctx + dir->npe_ctx_idx); + npe_ctx_addr = dir->npe_ctx_phys + dir->npe_ctx_idx; + + dir->npe_ctx_idx += init_len; + dir->npe_mode |= NPE_OP_HASH_ENABLE; + + if (!encrypt) + dir->npe_mode |= NPE_OP_HASH_VERIFY; + + ret = register_chain_var(tfm, HMAC_OPAD_VALUE, otarget, + init_len, npe_ctx_addr, key, key_len); + if (ret) + return ret; + return register_chain_var(tfm, HMAC_IPAD_VALUE, itarget, + init_len, npe_ctx_addr, key, key_len); +} + +static int gen_rev_aes_key(struct crypto_tfm *tfm) +{ + struct crypt_ctl *crypt; + struct ixp_ctx *ctx = crypto_tfm_ctx(tfm); + struct ix_sa_dir *dir = &ctx->decrypt; + + crypt = get_crypt_desc_emerg(); + if (!crypt) + return -EAGAIN; + + *(__be32 *)dir->npe_ctx |= cpu_to_be32(CIPH_ENCR); + + crypt->data.tfm = tfm; + crypt->crypt_offs = 0; + crypt->crypt_len = AES_BLOCK128; + crypt->src_buf = 0; + crypt->crypto_ctx = dir->npe_ctx_phys; + crypt->icv_rev_aes = dir->npe_ctx_phys + sizeof(u32); + crypt->mode = NPE_OP_ENC_GEN_KEY; + crypt->init_len = dir->npe_ctx_idx; + crypt->ctl_flags |= CTL_FLAG_GEN_REVAES; + + atomic_inc(&ctx->configuring); + qmgr_put_entry(send_qid, crypt_virt2phys(crypt)); + BUG_ON(qmgr_stat_overflow(send_qid)); + return 0; +} + +static int setup_cipher(struct crypto_tfm *tfm, int encrypt, const u8 *key, + int key_len) +{ + u8 *cinfo; + u32 cipher_cfg; + u32 keylen_cfg = 0; + struct ix_sa_dir *dir; + struct ixp_ctx *ctx = crypto_tfm_ctx(tfm); + int err; + + dir = encrypt ? &ctx->encrypt : &ctx->decrypt; + cinfo = dir->npe_ctx; + + if (encrypt) { + cipher_cfg = cipher_cfg_enc(tfm); + dir->npe_mode |= NPE_OP_CRYPT_ENCRYPT; + } else { + cipher_cfg = cipher_cfg_dec(tfm); + } + if (cipher_cfg & MOD_AES) { + switch (key_len) { + case 16: + keylen_cfg = MOD_AES128; + break; + case 24: + keylen_cfg = MOD_AES192; + break; + case 32: + keylen_cfg = MOD_AES256; + break; + default: + return -EINVAL; + } + cipher_cfg |= keylen_cfg; + } else { + err = crypto_des_verify_key(tfm, key); + if (err) + return err; + } + /* write cfg word to cryptinfo */ + *(__be32 *)cinfo = cpu_to_be32(cipher_cfg); + cinfo += sizeof(cipher_cfg); + + /* write cipher key to cryptinfo */ + memcpy(cinfo, key, key_len); + /* NPE wants keylen set to DES3_EDE_KEY_SIZE even for single DES */ + if (key_len < DES3_EDE_KEY_SIZE && !(cipher_cfg & MOD_AES)) { + memset(cinfo + key_len, 0, DES3_EDE_KEY_SIZE - key_len); + key_len = DES3_EDE_KEY_SIZE; + } + dir->npe_ctx_idx = sizeof(cipher_cfg) + key_len; + dir->npe_mode |= NPE_OP_CRYPT_ENABLE; + if ((cipher_cfg & MOD_AES) && !encrypt) + return gen_rev_aes_key(tfm); + + return 0; +} + +static struct buffer_desc *chainup_buffers(struct device *dev, + struct scatterlist *sg, unsigned int nbytes, + struct buffer_desc *buf, gfp_t flags, + enum dma_data_direction dir) +{ + for (; nbytes > 0; sg = sg_next(sg)) { + unsigned int len = min(nbytes, sg->length); + struct buffer_desc *next_buf; + dma_addr_t next_buf_phys; + void *ptr; + + nbytes -= len; + ptr = sg_virt(sg); + next_buf = dma_pool_alloc(buffer_pool, flags, &next_buf_phys); + if (!next_buf) { + buf = NULL; + break; + } + sg_dma_address(sg) = dma_map_single(dev, ptr, len, dir); + buf->next = next_buf; + buf->phys_next = next_buf_phys; + buf = next_buf; + + buf->phys_addr = sg_dma_address(sg); + buf->buf_len = len; + buf->dir = dir; + } + buf->next = NULL; + buf->phys_next = 0; + return buf; +} + +static int ablk_setkey(struct crypto_skcipher *tfm, const u8 *key, + unsigned int key_len) +{ + struct ixp_ctx *ctx = crypto_skcipher_ctx(tfm); + int ret; + + init_completion(&ctx->completion); + atomic_inc(&ctx->configuring); + + reset_sa_dir(&ctx->encrypt); + reset_sa_dir(&ctx->decrypt); + + ctx->encrypt.npe_mode = NPE_OP_HMAC_DISABLE; + ctx->decrypt.npe_mode = NPE_OP_HMAC_DISABLE; + + ret = setup_cipher(&tfm->base, 0, key, key_len); + if (ret) + goto out; + ret = setup_cipher(&tfm->base, 1, key, key_len); +out: + if (!atomic_dec_and_test(&ctx->configuring)) + wait_for_completion(&ctx->completion); + if (ret) + return ret; + crypto_skcipher_clear_flags(ctx->fallback_tfm, CRYPTO_TFM_REQ_MASK); + crypto_skcipher_set_flags(ctx->fallback_tfm, tfm->base.crt_flags & CRYPTO_TFM_REQ_MASK); + + return crypto_skcipher_setkey(ctx->fallback_tfm, key, key_len); +} + +static int ablk_des3_setkey(struct crypto_skcipher *tfm, const u8 *key, + unsigned int key_len) +{ + return verify_skcipher_des3_key(tfm, key) ?: + ablk_setkey(tfm, key, key_len); +} + +static int ablk_rfc3686_setkey(struct crypto_skcipher *tfm, const u8 *key, + unsigned int key_len) +{ + struct ixp_ctx *ctx = crypto_skcipher_ctx(tfm); + + /* the nonce is stored in bytes at end of key */ + if (key_len < CTR_RFC3686_NONCE_SIZE) + return -EINVAL; + + memcpy(ctx->nonce, key + (key_len - CTR_RFC3686_NONCE_SIZE), + CTR_RFC3686_NONCE_SIZE); + + key_len -= CTR_RFC3686_NONCE_SIZE; + return ablk_setkey(tfm, key, key_len); +} + +static int ixp4xx_cipher_fallback(struct skcipher_request *areq, int encrypt) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(areq); + struct ixp_ctx *op = crypto_skcipher_ctx(tfm); + struct ablk_ctx *rctx = skcipher_request_ctx(areq); + int err; + + skcipher_request_set_tfm(&rctx->fallback_req, op->fallback_tfm); + skcipher_request_set_callback(&rctx->fallback_req, areq->base.flags, + areq->base.complete, areq->base.data); + skcipher_request_set_crypt(&rctx->fallback_req, areq->src, areq->dst, + areq->cryptlen, areq->iv); + if (encrypt) + err = crypto_skcipher_encrypt(&rctx->fallback_req); + else + err = crypto_skcipher_decrypt(&rctx->fallback_req); + return err; +} + +static int ablk_perform(struct skcipher_request *req, int encrypt) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ixp_ctx *ctx = crypto_skcipher_ctx(tfm); + unsigned int ivsize = crypto_skcipher_ivsize(tfm); + struct ix_sa_dir *dir; + struct crypt_ctl *crypt; + unsigned int nbytes = req->cryptlen; + enum dma_data_direction src_direction = DMA_BIDIRECTIONAL; + struct ablk_ctx *req_ctx = skcipher_request_ctx(req); + struct buffer_desc src_hook; + struct device *dev = &pdev->dev; + unsigned int offset; + gfp_t flags = req->base.flags & CRYPTO_TFM_REQ_MAY_SLEEP ? + GFP_KERNEL : GFP_ATOMIC; + + if (sg_nents(req->src) > 1 || sg_nents(req->dst) > 1) + return ixp4xx_cipher_fallback(req, encrypt); + + if (qmgr_stat_full(send_qid)) + return -EAGAIN; + if (atomic_read(&ctx->configuring)) + return -EAGAIN; + + dir = encrypt ? &ctx->encrypt : &ctx->decrypt; + req_ctx->encrypt = encrypt; + + crypt = get_crypt_desc(); + if (!crypt) + return -ENOMEM; + + crypt->data.ablk_req = req; + crypt->crypto_ctx = dir->npe_ctx_phys; + crypt->mode = dir->npe_mode; + crypt->init_len = dir->npe_ctx_idx; + + crypt->crypt_offs = 0; + crypt->crypt_len = nbytes; + + BUG_ON(ivsize && !req->iv); + memcpy(crypt->iv, req->iv, ivsize); + if (ivsize > 0 && !encrypt) { + offset = req->cryptlen - ivsize; + scatterwalk_map_and_copy(req_ctx->iv, req->src, offset, ivsize, 0); + } + if (req->src != req->dst) { + struct buffer_desc dst_hook; + + crypt->mode |= NPE_OP_NOT_IN_PLACE; + /* This was never tested by Intel + * for more than one dst buffer, I think. */ + req_ctx->dst = NULL; + if (!chainup_buffers(dev, req->dst, nbytes, &dst_hook, + flags, DMA_FROM_DEVICE)) + goto free_buf_dest; + src_direction = DMA_TO_DEVICE; + req_ctx->dst = dst_hook.next; + crypt->dst_buf = dst_hook.phys_next; + } else { + req_ctx->dst = NULL; + } + req_ctx->src = NULL; + if (!chainup_buffers(dev, req->src, nbytes, &src_hook, flags, + src_direction)) + goto free_buf_src; + + req_ctx->src = src_hook.next; + crypt->src_buf = src_hook.phys_next; + crypt->ctl_flags |= CTL_FLAG_PERFORM_ABLK; + qmgr_put_entry(send_qid, crypt_virt2phys(crypt)); + BUG_ON(qmgr_stat_overflow(send_qid)); + return -EINPROGRESS; + +free_buf_src: + free_buf_chain(dev, req_ctx->src, crypt->src_buf); +free_buf_dest: + if (req->src != req->dst) + free_buf_chain(dev, req_ctx->dst, crypt->dst_buf); + + crypt->ctl_flags = CTL_FLAG_UNUSED; + return -ENOMEM; +} + +static int ablk_encrypt(struct skcipher_request *req) +{ + return ablk_perform(req, 1); +} + +static int ablk_decrypt(struct skcipher_request *req) +{ + return ablk_perform(req, 0); +} + +static int ablk_rfc3686_crypt(struct skcipher_request *req) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ixp_ctx *ctx = crypto_skcipher_ctx(tfm); + u8 iv[CTR_RFC3686_BLOCK_SIZE]; + u8 *info = req->iv; + int ret; + + /* set up counter block */ + memcpy(iv, ctx->nonce, CTR_RFC3686_NONCE_SIZE); + memcpy(iv + CTR_RFC3686_NONCE_SIZE, info, CTR_RFC3686_IV_SIZE); + + /* initialize counter portion of counter block */ + *(__be32 *)(iv + CTR_RFC3686_NONCE_SIZE + CTR_RFC3686_IV_SIZE) = + cpu_to_be32(1); + + req->iv = iv; + ret = ablk_perform(req, 1); + req->iv = info; + return ret; +} + +static int aead_perform(struct aead_request *req, int encrypt, + int cryptoffset, int eff_cryptlen, u8 *iv) +{ + struct crypto_aead *tfm = crypto_aead_reqtfm(req); + struct ixp_ctx *ctx = crypto_aead_ctx(tfm); + unsigned int ivsize = crypto_aead_ivsize(tfm); + unsigned int authsize = crypto_aead_authsize(tfm); + struct ix_sa_dir *dir; + struct crypt_ctl *crypt; + unsigned int cryptlen; + struct buffer_desc *buf, src_hook; + struct aead_ctx *req_ctx = aead_request_ctx(req); + struct device *dev = &pdev->dev; + gfp_t flags = req->base.flags & CRYPTO_TFM_REQ_MAY_SLEEP ? + GFP_KERNEL : GFP_ATOMIC; + enum dma_data_direction src_direction = DMA_BIDIRECTIONAL; + unsigned int lastlen; + + if (qmgr_stat_full(send_qid)) + return -EAGAIN; + if (atomic_read(&ctx->configuring)) + return -EAGAIN; + + if (encrypt) { + dir = &ctx->encrypt; + cryptlen = req->cryptlen; + } else { + dir = &ctx->decrypt; + /* req->cryptlen includes the authsize when decrypting */ + cryptlen = req->cryptlen - authsize; + eff_cryptlen -= authsize; + } + crypt = get_crypt_desc(); + if (!crypt) + return -ENOMEM; + + crypt->data.aead_req = req; + crypt->crypto_ctx = dir->npe_ctx_phys; + crypt->mode = dir->npe_mode; + crypt->init_len = dir->npe_ctx_idx; + + crypt->crypt_offs = cryptoffset; + crypt->crypt_len = eff_cryptlen; + + crypt->auth_offs = 0; + crypt->auth_len = req->assoclen + cryptlen; + BUG_ON(ivsize && !req->iv); + memcpy(crypt->iv, req->iv, ivsize); + + buf = chainup_buffers(dev, req->src, crypt->auth_len, + &src_hook, flags, src_direction); + req_ctx->src = src_hook.next; + crypt->src_buf = src_hook.phys_next; + if (!buf) + goto free_buf_src; + + lastlen = buf->buf_len; + if (lastlen >= authsize) + crypt->icv_rev_aes = buf->phys_addr + + buf->buf_len - authsize; + + req_ctx->dst = NULL; + + if (req->src != req->dst) { + struct buffer_desc dst_hook; + + crypt->mode |= NPE_OP_NOT_IN_PLACE; + src_direction = DMA_TO_DEVICE; + + buf = chainup_buffers(dev, req->dst, crypt->auth_len, + &dst_hook, flags, DMA_FROM_DEVICE); + req_ctx->dst = dst_hook.next; + crypt->dst_buf = dst_hook.phys_next; + + if (!buf) + goto free_buf_dst; + + if (encrypt) { + lastlen = buf->buf_len; + if (lastlen >= authsize) + crypt->icv_rev_aes = buf->phys_addr + + buf->buf_len - authsize; + } + } + + if (unlikely(lastlen < authsize)) { + dma_addr_t dma; + /* The 12 hmac bytes are scattered, + * we need to copy them into a safe buffer */ + req_ctx->hmac_virt = dma_pool_alloc(buffer_pool, flags, &dma); + if (unlikely(!req_ctx->hmac_virt)) + goto free_buf_dst; + crypt->icv_rev_aes = dma; + if (!encrypt) { + scatterwalk_map_and_copy(req_ctx->hmac_virt, + req->src, cryptlen, authsize, 0); + } + req_ctx->encrypt = encrypt; + } else { + req_ctx->hmac_virt = NULL; + } + + crypt->ctl_flags |= CTL_FLAG_PERFORM_AEAD; + qmgr_put_entry(send_qid, crypt_virt2phys(crypt)); + BUG_ON(qmgr_stat_overflow(send_qid)); + return -EINPROGRESS; + +free_buf_dst: + free_buf_chain(dev, req_ctx->dst, crypt->dst_buf); +free_buf_src: + free_buf_chain(dev, req_ctx->src, crypt->src_buf); + crypt->ctl_flags = CTL_FLAG_UNUSED; + return -ENOMEM; +} + +static int aead_setup(struct crypto_aead *tfm, unsigned int authsize) +{ + struct ixp_ctx *ctx = crypto_aead_ctx(tfm); + unsigned int digest_len = crypto_aead_maxauthsize(tfm); + int ret; + + if (!ctx->enckey_len && !ctx->authkey_len) + return 0; + init_completion(&ctx->completion); + atomic_inc(&ctx->configuring); + + reset_sa_dir(&ctx->encrypt); + reset_sa_dir(&ctx->decrypt); + + ret = setup_cipher(&tfm->base, 0, ctx->enckey, ctx->enckey_len); + if (ret) + goto out; + ret = setup_cipher(&tfm->base, 1, ctx->enckey, ctx->enckey_len); + if (ret) + goto out; + ret = setup_auth(&tfm->base, 0, authsize, ctx->authkey, + ctx->authkey_len, digest_len); + if (ret) + goto out; + ret = setup_auth(&tfm->base, 1, authsize, ctx->authkey, + ctx->authkey_len, digest_len); +out: + if (!atomic_dec_and_test(&ctx->configuring)) + wait_for_completion(&ctx->completion); + return ret; +} + +static int aead_setauthsize(struct crypto_aead *tfm, unsigned int authsize) +{ + int max = crypto_aead_maxauthsize(tfm) >> 2; + + if ((authsize >> 2) < 1 || (authsize >> 2) > max || (authsize & 3)) + return -EINVAL; + return aead_setup(tfm, authsize); +} + +static int aead_setkey(struct crypto_aead *tfm, const u8 *key, + unsigned int keylen) +{ + struct ixp_ctx *ctx = crypto_aead_ctx(tfm); + struct crypto_authenc_keys keys; + + if (crypto_authenc_extractkeys(&keys, key, keylen) != 0) + goto badkey; + + if (keys.authkeylen > sizeof(ctx->authkey)) + goto badkey; + + if (keys.enckeylen > sizeof(ctx->enckey)) + goto badkey; + + memcpy(ctx->authkey, keys.authkey, keys.authkeylen); + memcpy(ctx->enckey, keys.enckey, keys.enckeylen); + ctx->authkey_len = keys.authkeylen; + ctx->enckey_len = keys.enckeylen; + + memzero_explicit(&keys, sizeof(keys)); + return aead_setup(tfm, crypto_aead_authsize(tfm)); +badkey: + memzero_explicit(&keys, sizeof(keys)); + return -EINVAL; +} + +static int des3_aead_setkey(struct crypto_aead *tfm, const u8 *key, + unsigned int keylen) +{ + struct ixp_ctx *ctx = crypto_aead_ctx(tfm); + struct crypto_authenc_keys keys; + int err; + + err = crypto_authenc_extractkeys(&keys, key, keylen); + if (unlikely(err)) + goto badkey; + + err = -EINVAL; + if (keys.authkeylen > sizeof(ctx->authkey)) + goto badkey; + + err = verify_aead_des3_key(tfm, keys.enckey, keys.enckeylen); + if (err) + goto badkey; + + memcpy(ctx->authkey, keys.authkey, keys.authkeylen); + memcpy(ctx->enckey, keys.enckey, keys.enckeylen); + ctx->authkey_len = keys.authkeylen; + ctx->enckey_len = keys.enckeylen; + + memzero_explicit(&keys, sizeof(keys)); + return aead_setup(tfm, crypto_aead_authsize(tfm)); +badkey: + memzero_explicit(&keys, sizeof(keys)); + return err; +} + +static int aead_encrypt(struct aead_request *req) +{ + return aead_perform(req, 1, req->assoclen, req->cryptlen, req->iv); +} + +static int aead_decrypt(struct aead_request *req) +{ + return aead_perform(req, 0, req->assoclen, req->cryptlen, req->iv); +} + +static struct ixp_alg ixp4xx_algos[] = { +{ + .crypto = { + .base.cra_name = "cbc(des)", + .base.cra_blocksize = DES_BLOCK_SIZE, + + .min_keysize = DES_KEY_SIZE, + .max_keysize = DES_KEY_SIZE, + .ivsize = DES_BLOCK_SIZE, + }, + .cfg_enc = CIPH_ENCR | MOD_DES | MOD_CBC_ENC | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_DES | MOD_CBC_DEC | KEYLEN_192, + +}, { + .crypto = { + .base.cra_name = "ecb(des)", + .base.cra_blocksize = DES_BLOCK_SIZE, + .min_keysize = DES_KEY_SIZE, + .max_keysize = DES_KEY_SIZE, + }, + .cfg_enc = CIPH_ENCR | MOD_DES | MOD_ECB | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_DES | MOD_ECB | KEYLEN_192, +}, { + .crypto = { + .base.cra_name = "cbc(des3_ede)", + .base.cra_blocksize = DES3_EDE_BLOCK_SIZE, + + .min_keysize = DES3_EDE_KEY_SIZE, + .max_keysize = DES3_EDE_KEY_SIZE, + .ivsize = DES3_EDE_BLOCK_SIZE, + .setkey = ablk_des3_setkey, + }, + .cfg_enc = CIPH_ENCR | MOD_3DES | MOD_CBC_ENC | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_3DES | MOD_CBC_DEC | KEYLEN_192, +}, { + .crypto = { + .base.cra_name = "ecb(des3_ede)", + .base.cra_blocksize = DES3_EDE_BLOCK_SIZE, + + .min_keysize = DES3_EDE_KEY_SIZE, + .max_keysize = DES3_EDE_KEY_SIZE, + .setkey = ablk_des3_setkey, + }, + .cfg_enc = CIPH_ENCR | MOD_3DES | MOD_ECB | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_3DES | MOD_ECB | KEYLEN_192, +}, { + .crypto = { + .base.cra_name = "cbc(aes)", + .base.cra_blocksize = AES_BLOCK_SIZE, + + .min_keysize = AES_MIN_KEY_SIZE, + .max_keysize = AES_MAX_KEY_SIZE, + .ivsize = AES_BLOCK_SIZE, + }, + .cfg_enc = CIPH_ENCR | MOD_AES | MOD_CBC_ENC, + .cfg_dec = CIPH_DECR | MOD_AES | MOD_CBC_DEC, +}, { + .crypto = { + .base.cra_name = "ecb(aes)", + .base.cra_blocksize = AES_BLOCK_SIZE, + + .min_keysize = AES_MIN_KEY_SIZE, + .max_keysize = AES_MAX_KEY_SIZE, + }, + .cfg_enc = CIPH_ENCR | MOD_AES | MOD_ECB, + .cfg_dec = CIPH_DECR | MOD_AES | MOD_ECB, +}, { + .crypto = { + .base.cra_name = "ctr(aes)", + .base.cra_blocksize = 1, + + .min_keysize = AES_MIN_KEY_SIZE, + .max_keysize = AES_MAX_KEY_SIZE, + .ivsize = AES_BLOCK_SIZE, + }, + .cfg_enc = CIPH_ENCR | MOD_AES | MOD_CTR, + .cfg_dec = CIPH_ENCR | MOD_AES | MOD_CTR, +}, { + .crypto = { + .base.cra_name = "rfc3686(ctr(aes))", + .base.cra_blocksize = 1, + + .min_keysize = AES_MIN_KEY_SIZE, + .max_keysize = AES_MAX_KEY_SIZE, + .ivsize = AES_BLOCK_SIZE, + .setkey = ablk_rfc3686_setkey, + .encrypt = ablk_rfc3686_crypt, + .decrypt = ablk_rfc3686_crypt, + }, + .cfg_enc = CIPH_ENCR | MOD_AES | MOD_CTR, + .cfg_dec = CIPH_ENCR | MOD_AES | MOD_CTR, +} }; + +static struct ixp_aead_alg ixp4xx_aeads[] = { +{ + .crypto = { + .base = { + .cra_name = "authenc(hmac(md5),cbc(des))", + .cra_blocksize = DES_BLOCK_SIZE, + }, + .ivsize = DES_BLOCK_SIZE, + .maxauthsize = MD5_DIGEST_SIZE, + }, + .hash = &hash_alg_md5, + .cfg_enc = CIPH_ENCR | MOD_DES | MOD_CBC_ENC | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_DES | MOD_CBC_DEC | KEYLEN_192, +}, { + .crypto = { + .base = { + .cra_name = "authenc(hmac(md5),cbc(des3_ede))", + .cra_blocksize = DES3_EDE_BLOCK_SIZE, + }, + .ivsize = DES3_EDE_BLOCK_SIZE, + .maxauthsize = MD5_DIGEST_SIZE, + .setkey = des3_aead_setkey, + }, + .hash = &hash_alg_md5, + .cfg_enc = CIPH_ENCR | MOD_3DES | MOD_CBC_ENC | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_3DES | MOD_CBC_DEC | KEYLEN_192, +}, { + .crypto = { + .base = { + .cra_name = "authenc(hmac(sha1),cbc(des))", + .cra_blocksize = DES_BLOCK_SIZE, + }, + .ivsize = DES_BLOCK_SIZE, + .maxauthsize = SHA1_DIGEST_SIZE, + }, + .hash = &hash_alg_sha1, + .cfg_enc = CIPH_ENCR | MOD_DES | MOD_CBC_ENC | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_DES | MOD_CBC_DEC | KEYLEN_192, +}, { + .crypto = { + .base = { + .cra_name = "authenc(hmac(sha1),cbc(des3_ede))", + .cra_blocksize = DES3_EDE_BLOCK_SIZE, + }, + .ivsize = DES3_EDE_BLOCK_SIZE, + .maxauthsize = SHA1_DIGEST_SIZE, + .setkey = des3_aead_setkey, + }, + .hash = &hash_alg_sha1, + .cfg_enc = CIPH_ENCR | MOD_3DES | MOD_CBC_ENC | KEYLEN_192, + .cfg_dec = CIPH_DECR | MOD_3DES | MOD_CBC_DEC | KEYLEN_192, +}, { + .crypto = { + .base = { + .cra_name = "authenc(hmac(md5),cbc(aes))", + .cra_blocksize = AES_BLOCK_SIZE, + }, + .ivsize = AES_BLOCK_SIZE, + .maxauthsize = MD5_DIGEST_SIZE, + }, + .hash = &hash_alg_md5, + .cfg_enc = CIPH_ENCR | MOD_AES | MOD_CBC_ENC, + .cfg_dec = CIPH_DECR | MOD_AES | MOD_CBC_DEC, +}, { + .crypto = { + .base = { + .cra_name = "authenc(hmac(sha1),cbc(aes))", + .cra_blocksize = AES_BLOCK_SIZE, + }, + .ivsize = AES_BLOCK_SIZE, + .maxauthsize = SHA1_DIGEST_SIZE, + }, + .hash = &hash_alg_sha1, + .cfg_enc = CIPH_ENCR | MOD_AES | MOD_CBC_ENC, + .cfg_dec = CIPH_DECR | MOD_AES | MOD_CBC_DEC, +} }; + +#define IXP_POSTFIX "-ixp4xx" + +static int ixp_crypto_probe(struct platform_device *_pdev) +{ + struct device *dev = &_pdev->dev; + int num = ARRAY_SIZE(ixp4xx_algos); + int i, err; + + pdev = _pdev; + + err = init_ixp_crypto(dev); + if (err) + return err; + + for (i = 0; i < num; i++) { + struct skcipher_alg *cra = &ixp4xx_algos[i].crypto; + + if (snprintf(cra->base.cra_driver_name, CRYPTO_MAX_ALG_NAME, + "%s"IXP_POSTFIX, cra->base.cra_name) >= + CRYPTO_MAX_ALG_NAME) + continue; + if (!support_aes && (ixp4xx_algos[i].cfg_enc & MOD_AES)) + continue; + + /* block ciphers */ + cra->base.cra_flags = CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_ASYNC | + CRYPTO_ALG_ALLOCATES_MEMORY | + CRYPTO_ALG_NEED_FALLBACK; + if (!cra->setkey) + cra->setkey = ablk_setkey; + if (!cra->encrypt) + cra->encrypt = ablk_encrypt; + if (!cra->decrypt) + cra->decrypt = ablk_decrypt; + cra->init = init_tfm_ablk; + cra->exit = exit_tfm_ablk; + + cra->base.cra_ctxsize = sizeof(struct ixp_ctx); + cra->base.cra_module = THIS_MODULE; + cra->base.cra_alignmask = 3; + cra->base.cra_priority = 300; + if (crypto_register_skcipher(cra)) + dev_err(&pdev->dev, "Failed to register '%s'\n", + cra->base.cra_name); + else + ixp4xx_algos[i].registered = 1; + } + + for (i = 0; i < ARRAY_SIZE(ixp4xx_aeads); i++) { + struct aead_alg *cra = &ixp4xx_aeads[i].crypto; + + if (snprintf(cra->base.cra_driver_name, CRYPTO_MAX_ALG_NAME, + "%s"IXP_POSTFIX, cra->base.cra_name) >= + CRYPTO_MAX_ALG_NAME) + continue; + if (!support_aes && (ixp4xx_algos[i].cfg_enc & MOD_AES)) + continue; + + /* authenc */ + cra->base.cra_flags = CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_ASYNC | + CRYPTO_ALG_ALLOCATES_MEMORY; + cra->setkey = cra->setkey ?: aead_setkey; + cra->setauthsize = aead_setauthsize; + cra->encrypt = aead_encrypt; + cra->decrypt = aead_decrypt; + cra->init = init_tfm_aead; + cra->exit = exit_tfm_aead; + + cra->base.cra_ctxsize = sizeof(struct ixp_ctx); + cra->base.cra_module = THIS_MODULE; + cra->base.cra_alignmask = 3; + cra->base.cra_priority = 300; + + if (crypto_register_aead(cra)) + dev_err(&pdev->dev, "Failed to register '%s'\n", + cra->base.cra_driver_name); + else + ixp4xx_aeads[i].registered = 1; + } + return 0; +} + +static void ixp_crypto_remove(struct platform_device *pdev) +{ + int num = ARRAY_SIZE(ixp4xx_algos); + int i; + + for (i = 0; i < ARRAY_SIZE(ixp4xx_aeads); i++) { + if (ixp4xx_aeads[i].registered) + crypto_unregister_aead(&ixp4xx_aeads[i].crypto); + } + + for (i = 0; i < num; i++) { + if (ixp4xx_algos[i].registered) + crypto_unregister_skcipher(&ixp4xx_algos[i].crypto); + } + release_ixp_crypto(&pdev->dev); +} +static const struct of_device_id ixp4xx_crypto_of_match[] = { + { + .compatible = "intel,ixp4xx-crypto", + }, + {}, +}; + +static struct platform_driver ixp_crypto_driver = { + .probe = ixp_crypto_probe, + .remove = ixp_crypto_remove, + .driver = { + .name = "ixp4xx_crypto", + .of_match_table = ixp4xx_crypto_of_match, + }, +}; +module_platform_driver(ixp_crypto_driver); + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Christian Hohnstaedt <chohnstaedt@innominate.com>"); +MODULE_DESCRIPTION("IXP4xx hardware crypto"); + diff --git a/drivers/crypto/intel/keembay/Kconfig b/drivers/crypto/intel/keembay/Kconfig new file mode 100644 index 000000000000..1cd62f9c3e3a --- /dev/null +++ b/drivers/crypto/intel/keembay/Kconfig @@ -0,0 +1,90 @@ +config CRYPTO_DEV_KEEMBAY_OCS_AES_SM4 + tristate "Support for Intel Keem Bay OCS AES/SM4 HW acceleration" + depends on HAS_IOMEM + depends on ARCH_KEEMBAY || COMPILE_TEST + select CRYPTO_SKCIPHER + select CRYPTO_AEAD + select CRYPTO_ENGINE + help + Support for Intel Keem Bay Offload and Crypto Subsystem (OCS) AES and + SM4 cipher hardware acceleration for use with Crypto API. + + Provides HW acceleration for the following transformations: + cbc(aes), ctr(aes), ccm(aes), gcm(aes), cbc(sm4), ctr(sm4), ccm(sm4) + and gcm(sm4). + + Optionally, support for the following transformations can also be + enabled: ecb(aes), cts(cbc(aes)), ecb(sm4) and cts(cbc(sm4)). + +config CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB + bool "Support for Intel Keem Bay OCS AES/SM4 ECB HW acceleration" + depends on CRYPTO_DEV_KEEMBAY_OCS_AES_SM4 + help + Support for Intel Keem Bay Offload and Crypto Subsystem (OCS) + AES/SM4 ECB mode hardware acceleration for use with Crypto API. + + Provides OCS version of ecb(aes) and ecb(sm4) + + Intel does not recommend use of ECB mode with AES/SM4. + +config CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS + bool "Support for Intel Keem Bay OCS AES/SM4 CTS HW acceleration" + depends on CRYPTO_DEV_KEEMBAY_OCS_AES_SM4 + help + Support for Intel Keem Bay Offload and Crypto Subsystem (OCS) + AES/SM4 CBC with CTS mode hardware acceleration for use with + Crypto API. + + Provides OCS version of cts(cbc(aes)) and cts(cbc(sm4)). + + Intel does not recommend use of CTS mode with AES/SM4. + +config CRYPTO_DEV_KEEMBAY_OCS_ECC + tristate "Support for Intel Keem Bay OCS ECC HW acceleration" + depends on ARCH_KEEMBAY || COMPILE_TEST + depends on OF + depends on HAS_IOMEM + select CRYPTO_ECDH + select CRYPTO_ENGINE + help + Support for Intel Keem Bay Offload and Crypto Subsystem (OCS) + Elliptic Curve Cryptography (ECC) hardware acceleration for use with + Crypto API. + + Provides OCS acceleration for ECDH-256 and ECDH-384. + + Say Y or M if you are compiling for the Intel Keem Bay SoC. The + module will be called keembay-ocs-ecc. + + If unsure, say N. + +config CRYPTO_DEV_KEEMBAY_OCS_HCU + tristate "Support for Intel Keem Bay OCS HCU HW acceleration" + select CRYPTO_HASH + select CRYPTO_ENGINE + depends on HAS_IOMEM + depends on ARCH_KEEMBAY || COMPILE_TEST + depends on OF + help + Support for Intel Keem Bay Offload and Crypto Subsystem (OCS) Hash + Control Unit (HCU) hardware acceleration for use with Crypto API. + + Provides OCS HCU hardware acceleration of sha256, sha384, sha512, and + sm3, as well as the HMAC variant of these algorithms. + + Say Y or M if you're building for the Intel Keem Bay SoC. If compiled + as a module, the module will be called keembay-ocs-hcu. + + If unsure, say N. + +config CRYPTO_DEV_KEEMBAY_OCS_HCU_HMAC_SHA224 + bool "Enable sha224 and hmac(sha224) support in Intel Keem Bay OCS HCU" + depends on CRYPTO_DEV_KEEMBAY_OCS_HCU + help + Enables support for sha224 and hmac(sha224) algorithms in the Intel + Keem Bay OCS HCU driver. Intel recommends not to use these + algorithms. + + Provides OCS HCU hardware acceleration of sha224 and hmac(224). + + If unsure, say N. diff --git a/drivers/crypto/intel/keembay/Makefile b/drivers/crypto/intel/keembay/Makefile new file mode 100644 index 000000000000..7c12c3c138bd --- /dev/null +++ b/drivers/crypto/intel/keembay/Makefile @@ -0,0 +1,10 @@ +# +# Makefile for Intel Keem Bay OCS Crypto API Linux drivers +# +obj-$(CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4) += keembay-ocs-aes.o +keembay-ocs-aes-objs := keembay-ocs-aes-core.o ocs-aes.o + +obj-$(CONFIG_CRYPTO_DEV_KEEMBAY_OCS_ECC) += keembay-ocs-ecc.o + +obj-$(CONFIG_CRYPTO_DEV_KEEMBAY_OCS_HCU) += keembay-ocs-hcu.o +keembay-ocs-hcu-objs := keembay-ocs-hcu-core.o ocs-hcu.o diff --git a/drivers/crypto/intel/keembay/keembay-ocs-aes-core.c b/drivers/crypto/intel/keembay/keembay-ocs-aes-core.c new file mode 100644 index 000000000000..8a8f6c81e010 --- /dev/null +++ b/drivers/crypto/intel/keembay/keembay-ocs-aes-core.c @@ -0,0 +1,1689 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Intel Keem Bay OCS AES Crypto Driver. + * + * Copyright (C) 2018-2020 Intel Corporation + */ + +#include <crypto/aes.h> +#include <crypto/engine.h> +#include <crypto/gcm.h> +#include <crypto/internal/aead.h> +#include <crypto/internal/skcipher.h> +#include <crypto/scatterwalk.h> +#include <linux/clk.h> +#include <linux/completion.h> +#include <linux/dma-mapping.h> +#include <linux/err.h> +#include <linux/interrupt.h> +#include <linux/io.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/platform_device.h> +#include <linux/string.h> + +#include "ocs-aes.h" + +#define KMB_OCS_PRIORITY 350 +#define DRV_NAME "keembay-ocs-aes" + +#define OCS_AES_MIN_KEY_SIZE 16 +#define OCS_AES_MAX_KEY_SIZE 32 +#define OCS_AES_KEYSIZE_128 16 +#define OCS_AES_KEYSIZE_192 24 +#define OCS_AES_KEYSIZE_256 32 +#define OCS_SM4_KEY_SIZE 16 + +/** + * struct ocs_aes_tctx - OCS AES Transform context + * @aes_dev: The OCS AES device. + * @key: AES/SM4 key. + * @key_len: The length (in bytes) of @key. + * @cipher: OCS cipher to use (either AES or SM4). + * @sw_cipher: The cipher to use as fallback. + * @use_fallback: Whether or not fallback cipher should be used. + */ +struct ocs_aes_tctx { + struct ocs_aes_dev *aes_dev; + u8 key[OCS_AES_KEYSIZE_256]; + unsigned int key_len; + enum ocs_cipher cipher; + union { + struct crypto_sync_skcipher *sk; + struct crypto_aead *aead; + } sw_cipher; + bool use_fallback; +}; + +/** + * struct ocs_aes_rctx - OCS AES Request context. + * @instruction: Instruction to be executed (encrypt / decrypt). + * @mode: Mode to use (ECB, CBC, CTR, CCm, GCM, CTS) + * @src_nents: Number of source SG entries. + * @dst_nents: Number of destination SG entries. + * @src_dma_count: The number of DMA-mapped entries of the source SG. + * @dst_dma_count: The number of DMA-mapped entries of the destination SG. + * @in_place: Whether or not this is an in place request, i.e., + * src_sg == dst_sg. + * @src_dll: OCS DMA linked list for input data. + * @dst_dll: OCS DMA linked list for output data. + * @last_ct_blk: Buffer to hold last cipher text block (only used in CBC + * mode). + * @cts_swap: Whether or not CTS swap must be performed. + * @aad_src_dll: OCS DMA linked list for input AAD data. + * @aad_dst_dll: OCS DMA linked list for output AAD data. + * @in_tag: Buffer to hold input encrypted tag (only used for + * CCM/GCM decrypt). + * @out_tag: Buffer to hold output encrypted / decrypted tag (only + * used for GCM encrypt / decrypt). + */ +struct ocs_aes_rctx { + /* Fields common across all modes. */ + enum ocs_instruction instruction; + enum ocs_mode mode; + int src_nents; + int dst_nents; + int src_dma_count; + int dst_dma_count; + bool in_place; + struct ocs_dll_desc src_dll; + struct ocs_dll_desc dst_dll; + + /* CBC specific */ + u8 last_ct_blk[AES_BLOCK_SIZE]; + + /* CTS specific */ + int cts_swap; + + /* CCM/GCM specific */ + struct ocs_dll_desc aad_src_dll; + struct ocs_dll_desc aad_dst_dll; + u8 in_tag[AES_BLOCK_SIZE]; + + /* GCM specific */ + u8 out_tag[AES_BLOCK_SIZE]; +}; + +/* Driver data. */ +struct ocs_aes_drv { + struct list_head dev_list; + spinlock_t lock; /* Protects dev_list. */ +}; + +static struct ocs_aes_drv ocs_aes = { + .dev_list = LIST_HEAD_INIT(ocs_aes.dev_list), + .lock = __SPIN_LOCK_UNLOCKED(ocs_aes.lock), +}; + +static struct ocs_aes_dev *kmb_ocs_aes_find_dev(struct ocs_aes_tctx *tctx) +{ + struct ocs_aes_dev *aes_dev; + + spin_lock(&ocs_aes.lock); + + if (tctx->aes_dev) { + aes_dev = tctx->aes_dev; + goto exit; + } + + /* Only a single OCS device available */ + aes_dev = list_first_entry(&ocs_aes.dev_list, struct ocs_aes_dev, list); + tctx->aes_dev = aes_dev; + +exit: + spin_unlock(&ocs_aes.lock); + + return aes_dev; +} + +/* + * Ensure key is 128-bit or 256-bit for AES or 128-bit for SM4 and an actual + * key is being passed in. + * + * Return: 0 if key is valid, -EINVAL otherwise. + */ +static int check_key(const u8 *in_key, size_t key_len, enum ocs_cipher cipher) +{ + if (!in_key) + return -EINVAL; + + /* For AES, only 128-byte or 256-byte keys are supported. */ + if (cipher == OCS_AES && (key_len == OCS_AES_KEYSIZE_128 || + key_len == OCS_AES_KEYSIZE_256)) + return 0; + + /* For SM4, only 128-byte keys are supported. */ + if (cipher == OCS_SM4 && key_len == OCS_AES_KEYSIZE_128) + return 0; + + /* Everything else is unsupported. */ + return -EINVAL; +} + +/* Save key into transformation context. */ +static int save_key(struct ocs_aes_tctx *tctx, const u8 *in_key, size_t key_len, + enum ocs_cipher cipher) +{ + int ret; + + ret = check_key(in_key, key_len, cipher); + if (ret) + return ret; + + memcpy(tctx->key, in_key, key_len); + tctx->key_len = key_len; + tctx->cipher = cipher; + + return 0; +} + +/* Set key for symmetric cypher. */ +static int kmb_ocs_sk_set_key(struct crypto_skcipher *tfm, const u8 *in_key, + size_t key_len, enum ocs_cipher cipher) +{ + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + + /* Fallback is used for AES with 192-bit key. */ + tctx->use_fallback = (cipher == OCS_AES && + key_len == OCS_AES_KEYSIZE_192); + + if (!tctx->use_fallback) + return save_key(tctx, in_key, key_len, cipher); + + crypto_sync_skcipher_clear_flags(tctx->sw_cipher.sk, + CRYPTO_TFM_REQ_MASK); + crypto_sync_skcipher_set_flags(tctx->sw_cipher.sk, + tfm->base.crt_flags & + CRYPTO_TFM_REQ_MASK); + + return crypto_sync_skcipher_setkey(tctx->sw_cipher.sk, in_key, key_len); +} + +/* Set key for AEAD cipher. */ +static int kmb_ocs_aead_set_key(struct crypto_aead *tfm, const u8 *in_key, + size_t key_len, enum ocs_cipher cipher) +{ + struct ocs_aes_tctx *tctx = crypto_aead_ctx(tfm); + + /* Fallback is used for AES with 192-bit key. */ + tctx->use_fallback = (cipher == OCS_AES && + key_len == OCS_AES_KEYSIZE_192); + + if (!tctx->use_fallback) + return save_key(tctx, in_key, key_len, cipher); + + crypto_aead_clear_flags(tctx->sw_cipher.aead, CRYPTO_TFM_REQ_MASK); + crypto_aead_set_flags(tctx->sw_cipher.aead, + crypto_aead_get_flags(tfm) & CRYPTO_TFM_REQ_MASK); + + return crypto_aead_setkey(tctx->sw_cipher.aead, in_key, key_len); +} + +/* Swap two AES blocks in SG lists. */ +static void sg_swap_blocks(struct scatterlist *sgl, unsigned int nents, + off_t blk1_offset, off_t blk2_offset) +{ + u8 tmp_buf1[AES_BLOCK_SIZE], tmp_buf2[AES_BLOCK_SIZE]; + + /* + * No easy way to copy within sg list, so copy both blocks to temporary + * buffers first. + */ + sg_pcopy_to_buffer(sgl, nents, tmp_buf1, AES_BLOCK_SIZE, blk1_offset); + sg_pcopy_to_buffer(sgl, nents, tmp_buf2, AES_BLOCK_SIZE, blk2_offset); + sg_pcopy_from_buffer(sgl, nents, tmp_buf1, AES_BLOCK_SIZE, blk2_offset); + sg_pcopy_from_buffer(sgl, nents, tmp_buf2, AES_BLOCK_SIZE, blk1_offset); +} + +/* Initialize request context to default values. */ +static void ocs_aes_init_rctx(struct ocs_aes_rctx *rctx) +{ + /* Zero everything. */ + memset(rctx, 0, sizeof(*rctx)); + + /* Set initial value for DMA addresses. */ + rctx->src_dll.dma_addr = DMA_MAPPING_ERROR; + rctx->dst_dll.dma_addr = DMA_MAPPING_ERROR; + rctx->aad_src_dll.dma_addr = DMA_MAPPING_ERROR; + rctx->aad_dst_dll.dma_addr = DMA_MAPPING_ERROR; +} + +static int kmb_ocs_sk_validate_input(struct skcipher_request *req, + enum ocs_mode mode) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + int iv_size = crypto_skcipher_ivsize(tfm); + + switch (mode) { + case OCS_MODE_ECB: + /* Ensure input length is multiple of block size */ + if (req->cryptlen % AES_BLOCK_SIZE != 0) + return -EINVAL; + + return 0; + + case OCS_MODE_CBC: + /* Ensure input length is multiple of block size */ + if (req->cryptlen % AES_BLOCK_SIZE != 0) + return -EINVAL; + + /* Ensure IV is present and block size in length */ + if (!req->iv || iv_size != AES_BLOCK_SIZE) + return -EINVAL; + /* + * NOTE: Since req->cryptlen == 0 case was already handled in + * kmb_ocs_sk_common(), the above two conditions also guarantee + * that: cryptlen >= iv_size + */ + return 0; + + case OCS_MODE_CTR: + /* Ensure IV is present and block size in length */ + if (!req->iv || iv_size != AES_BLOCK_SIZE) + return -EINVAL; + return 0; + + case OCS_MODE_CTS: + /* Ensure input length >= block size */ + if (req->cryptlen < AES_BLOCK_SIZE) + return -EINVAL; + + /* Ensure IV is present and block size in length */ + if (!req->iv || iv_size != AES_BLOCK_SIZE) + return -EINVAL; + + return 0; + default: + return -EINVAL; + } +} + +/* + * Called by encrypt() / decrypt() skcipher functions. + * + * Use fallback if needed, otherwise initialize context and enqueue request + * into engine. + */ +static int kmb_ocs_sk_common(struct skcipher_request *req, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + enum ocs_mode mode) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ocs_aes_rctx *rctx = skcipher_request_ctx(req); + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + struct ocs_aes_dev *aes_dev; + int rc; + + if (tctx->use_fallback) { + SYNC_SKCIPHER_REQUEST_ON_STACK(subreq, tctx->sw_cipher.sk); + + skcipher_request_set_sync_tfm(subreq, tctx->sw_cipher.sk); + skcipher_request_set_callback(subreq, req->base.flags, NULL, + NULL); + skcipher_request_set_crypt(subreq, req->src, req->dst, + req->cryptlen, req->iv); + + if (instruction == OCS_ENCRYPT) + rc = crypto_skcipher_encrypt(subreq); + else + rc = crypto_skcipher_decrypt(subreq); + + skcipher_request_zero(subreq); + + return rc; + } + + /* + * If cryptlen == 0, no processing needed for ECB, CBC and CTR. + * + * For CTS continue: kmb_ocs_sk_validate_input() will return -EINVAL. + */ + if (!req->cryptlen && mode != OCS_MODE_CTS) + return 0; + + rc = kmb_ocs_sk_validate_input(req, mode); + if (rc) + return rc; + + aes_dev = kmb_ocs_aes_find_dev(tctx); + if (!aes_dev) + return -ENODEV; + + if (cipher != tctx->cipher) + return -EINVAL; + + ocs_aes_init_rctx(rctx); + rctx->instruction = instruction; + rctx->mode = mode; + + return crypto_transfer_skcipher_request_to_engine(aes_dev->engine, req); +} + +static void cleanup_ocs_dma_linked_list(struct device *dev, + struct ocs_dll_desc *dll) +{ + if (dll->vaddr) + dma_free_coherent(dev, dll->size, dll->vaddr, dll->dma_addr); + dll->vaddr = NULL; + dll->size = 0; + dll->dma_addr = DMA_MAPPING_ERROR; +} + +static void kmb_ocs_sk_dma_cleanup(struct skcipher_request *req) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ocs_aes_rctx *rctx = skcipher_request_ctx(req); + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + struct device *dev = tctx->aes_dev->dev; + + if (rctx->src_dma_count) { + dma_unmap_sg(dev, req->src, rctx->src_nents, DMA_TO_DEVICE); + rctx->src_dma_count = 0; + } + + if (rctx->dst_dma_count) { + dma_unmap_sg(dev, req->dst, rctx->dst_nents, rctx->in_place ? + DMA_BIDIRECTIONAL : + DMA_FROM_DEVICE); + rctx->dst_dma_count = 0; + } + + /* Clean up OCS DMA linked lists */ + cleanup_ocs_dma_linked_list(dev, &rctx->src_dll); + cleanup_ocs_dma_linked_list(dev, &rctx->dst_dll); +} + +static int kmb_ocs_sk_prepare_inplace(struct skcipher_request *req) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ocs_aes_rctx *rctx = skcipher_request_ctx(req); + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + int iv_size = crypto_skcipher_ivsize(tfm); + int rc; + + /* + * For CBC decrypt, save last block (iv) to last_ct_blk buffer. + * + * Note: if we are here, we already checked that cryptlen >= iv_size + * and iv_size == AES_BLOCK_SIZE (i.e., the size of last_ct_blk); see + * kmb_ocs_sk_validate_input(). + */ + if (rctx->mode == OCS_MODE_CBC && rctx->instruction == OCS_DECRYPT) + scatterwalk_map_and_copy(rctx->last_ct_blk, req->src, + req->cryptlen - iv_size, iv_size, 0); + + /* For CTS decrypt, swap last two blocks, if needed. */ + if (rctx->cts_swap && rctx->instruction == OCS_DECRYPT) + sg_swap_blocks(req->dst, rctx->dst_nents, + req->cryptlen - AES_BLOCK_SIZE, + req->cryptlen - (2 * AES_BLOCK_SIZE)); + + /* src and dst buffers are the same, use bidirectional DMA mapping. */ + rctx->dst_dma_count = dma_map_sg(tctx->aes_dev->dev, req->dst, + rctx->dst_nents, DMA_BIDIRECTIONAL); + if (rctx->dst_dma_count == 0) { + dev_err(tctx->aes_dev->dev, "Failed to map destination sg\n"); + return -ENOMEM; + } + + /* Create DST linked list */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->dst, + rctx->dst_dma_count, &rctx->dst_dll, + req->cryptlen, 0); + if (rc) + return rc; + /* + * If descriptor creation was successful, set the src_dll.dma_addr to + * the value of dst_dll.dma_addr, as we do in-place AES operation on + * the src. + */ + rctx->src_dll.dma_addr = rctx->dst_dll.dma_addr; + + return 0; +} + +static int kmb_ocs_sk_prepare_notinplace(struct skcipher_request *req) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ocs_aes_rctx *rctx = skcipher_request_ctx(req); + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + int rc; + + rctx->src_nents = sg_nents_for_len(req->src, req->cryptlen); + if (rctx->src_nents < 0) + return -EBADMSG; + + /* Map SRC SG. */ + rctx->src_dma_count = dma_map_sg(tctx->aes_dev->dev, req->src, + rctx->src_nents, DMA_TO_DEVICE); + if (rctx->src_dma_count == 0) { + dev_err(tctx->aes_dev->dev, "Failed to map source sg\n"); + return -ENOMEM; + } + + /* Create SRC linked list */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->src, + rctx->src_dma_count, &rctx->src_dll, + req->cryptlen, 0); + if (rc) + return rc; + + /* Map DST SG. */ + rctx->dst_dma_count = dma_map_sg(tctx->aes_dev->dev, req->dst, + rctx->dst_nents, DMA_FROM_DEVICE); + if (rctx->dst_dma_count == 0) { + dev_err(tctx->aes_dev->dev, "Failed to map destination sg\n"); + return -ENOMEM; + } + + /* Create DST linked list */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->dst, + rctx->dst_dma_count, &rctx->dst_dll, + req->cryptlen, 0); + if (rc) + return rc; + + /* If this is not a CTS decrypt operation with swapping, we are done. */ + if (!(rctx->cts_swap && rctx->instruction == OCS_DECRYPT)) + return 0; + + /* + * Otherwise, we have to copy src to dst (as we cannot modify src). + * Use OCS AES bypass mode to copy src to dst via DMA. + * + * NOTE: for anything other than small data sizes this is rather + * inefficient. + */ + rc = ocs_aes_bypass_op(tctx->aes_dev, rctx->dst_dll.dma_addr, + rctx->src_dll.dma_addr, req->cryptlen); + if (rc) + return rc; + + /* + * Now dst == src, so clean up what we did so far and use in_place + * logic. + */ + kmb_ocs_sk_dma_cleanup(req); + rctx->in_place = true; + + return kmb_ocs_sk_prepare_inplace(req); +} + +static int kmb_ocs_sk_run(struct skcipher_request *req) +{ + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ocs_aes_rctx *rctx = skcipher_request_ctx(req); + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + struct ocs_aes_dev *aes_dev = tctx->aes_dev; + int iv_size = crypto_skcipher_ivsize(tfm); + int rc; + + rctx->dst_nents = sg_nents_for_len(req->dst, req->cryptlen); + if (rctx->dst_nents < 0) + return -EBADMSG; + + /* + * If 2 blocks or greater, and multiple of block size swap last two + * blocks to be compatible with other crypto API CTS implementations: + * OCS mode uses CBC-CS2, whereas other crypto API implementations use + * CBC-CS3. + * CBC-CS2 and CBC-CS3 defined by: + * https://nvlpubs.nist.gov/nistpubs/Legacy/SP/nistspecialpublication800-38a-add.pdf + */ + rctx->cts_swap = (rctx->mode == OCS_MODE_CTS && + req->cryptlen > AES_BLOCK_SIZE && + req->cryptlen % AES_BLOCK_SIZE == 0); + + rctx->in_place = (req->src == req->dst); + + if (rctx->in_place) + rc = kmb_ocs_sk_prepare_inplace(req); + else + rc = kmb_ocs_sk_prepare_notinplace(req); + + if (rc) + goto error; + + rc = ocs_aes_op(aes_dev, rctx->mode, tctx->cipher, rctx->instruction, + rctx->dst_dll.dma_addr, rctx->src_dll.dma_addr, + req->cryptlen, req->iv, iv_size); + if (rc) + goto error; + + /* Clean-up DMA before further processing output. */ + kmb_ocs_sk_dma_cleanup(req); + + /* For CTS Encrypt, swap last 2 blocks, if needed. */ + if (rctx->cts_swap && rctx->instruction == OCS_ENCRYPT) { + sg_swap_blocks(req->dst, rctx->dst_nents, + req->cryptlen - AES_BLOCK_SIZE, + req->cryptlen - (2 * AES_BLOCK_SIZE)); + return 0; + } + + /* For CBC copy IV to req->IV. */ + if (rctx->mode == OCS_MODE_CBC) { + /* CBC encrypt case. */ + if (rctx->instruction == OCS_ENCRYPT) { + scatterwalk_map_and_copy(req->iv, req->dst, + req->cryptlen - iv_size, + iv_size, 0); + return 0; + } + /* CBC decrypt case. */ + if (rctx->in_place) + memcpy(req->iv, rctx->last_ct_blk, iv_size); + else + scatterwalk_map_and_copy(req->iv, req->src, + req->cryptlen - iv_size, + iv_size, 0); + return 0; + } + /* For all other modes there's nothing to do. */ + + return 0; + +error: + kmb_ocs_sk_dma_cleanup(req); + + return rc; +} + +static int kmb_ocs_aead_validate_input(struct aead_request *req, + enum ocs_instruction instruction, + enum ocs_mode mode) +{ + struct crypto_aead *tfm = crypto_aead_reqtfm(req); + int tag_size = crypto_aead_authsize(tfm); + int iv_size = crypto_aead_ivsize(tfm); + + /* For decrypt crytplen == len(PT) + len(tag). */ + if (instruction == OCS_DECRYPT && req->cryptlen < tag_size) + return -EINVAL; + + /* IV is mandatory. */ + if (!req->iv) + return -EINVAL; + + switch (mode) { + case OCS_MODE_GCM: + if (iv_size != GCM_AES_IV_SIZE) + return -EINVAL; + + return 0; + + case OCS_MODE_CCM: + /* Ensure IV is present and block size in length */ + if (iv_size != AES_BLOCK_SIZE) + return -EINVAL; + + return 0; + + default: + return -EINVAL; + } +} + +/* + * Called by encrypt() / decrypt() aead functions. + * + * Use fallback if needed, otherwise initialize context and enqueue request + * into engine. + */ +static int kmb_ocs_aead_common(struct aead_request *req, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + enum ocs_mode mode) +{ + struct ocs_aes_tctx *tctx = crypto_aead_ctx(crypto_aead_reqtfm(req)); + struct ocs_aes_rctx *rctx = aead_request_ctx(req); + struct ocs_aes_dev *dd; + int rc; + + if (tctx->use_fallback) { + struct aead_request *subreq = aead_request_ctx(req); + + aead_request_set_tfm(subreq, tctx->sw_cipher.aead); + aead_request_set_callback(subreq, req->base.flags, + req->base.complete, req->base.data); + aead_request_set_crypt(subreq, req->src, req->dst, + req->cryptlen, req->iv); + aead_request_set_ad(subreq, req->assoclen); + rc = crypto_aead_setauthsize(tctx->sw_cipher.aead, + crypto_aead_authsize(crypto_aead_reqtfm(req))); + if (rc) + return rc; + + return (instruction == OCS_ENCRYPT) ? + crypto_aead_encrypt(subreq) : + crypto_aead_decrypt(subreq); + } + + rc = kmb_ocs_aead_validate_input(req, instruction, mode); + if (rc) + return rc; + + dd = kmb_ocs_aes_find_dev(tctx); + if (!dd) + return -ENODEV; + + if (cipher != tctx->cipher) + return -EINVAL; + + ocs_aes_init_rctx(rctx); + rctx->instruction = instruction; + rctx->mode = mode; + + return crypto_transfer_aead_request_to_engine(dd->engine, req); +} + +static void kmb_ocs_aead_dma_cleanup(struct aead_request *req) +{ + struct ocs_aes_tctx *tctx = crypto_aead_ctx(crypto_aead_reqtfm(req)); + struct ocs_aes_rctx *rctx = aead_request_ctx(req); + struct device *dev = tctx->aes_dev->dev; + + if (rctx->src_dma_count) { + dma_unmap_sg(dev, req->src, rctx->src_nents, DMA_TO_DEVICE); + rctx->src_dma_count = 0; + } + + if (rctx->dst_dma_count) { + dma_unmap_sg(dev, req->dst, rctx->dst_nents, rctx->in_place ? + DMA_BIDIRECTIONAL : + DMA_FROM_DEVICE); + rctx->dst_dma_count = 0; + } + /* Clean up OCS DMA linked lists */ + cleanup_ocs_dma_linked_list(dev, &rctx->src_dll); + cleanup_ocs_dma_linked_list(dev, &rctx->dst_dll); + cleanup_ocs_dma_linked_list(dev, &rctx->aad_src_dll); + cleanup_ocs_dma_linked_list(dev, &rctx->aad_dst_dll); +} + +/** + * kmb_ocs_aead_dma_prepare() - Do DMA mapping for AEAD processing. + * @req: The AEAD request being processed. + * @src_dll_size: Where to store the length of the data mapped into the + * src_dll OCS DMA list. + * + * Do the following: + * - DMA map req->src and req->dst + * - Initialize the following OCS DMA linked lists: rctx->src_dll, + * rctx->dst_dll, rctx->aad_src_dll and rxtc->aad_dst_dll. + * + * Return: 0 on success, negative error code otherwise. + */ +static int kmb_ocs_aead_dma_prepare(struct aead_request *req, u32 *src_dll_size) +{ + struct ocs_aes_tctx *tctx = crypto_aead_ctx(crypto_aead_reqtfm(req)); + const int tag_size = crypto_aead_authsize(crypto_aead_reqtfm(req)); + struct ocs_aes_rctx *rctx = aead_request_ctx(req); + u32 in_size; /* The length of the data to be mapped by src_dll. */ + u32 out_size; /* The length of the data to be mapped by dst_dll. */ + u32 dst_size; /* The length of the data in dst_sg. */ + int rc; + + /* Get number of entries in input data SG list. */ + rctx->src_nents = sg_nents_for_len(req->src, + req->assoclen + req->cryptlen); + if (rctx->src_nents < 0) + return -EBADMSG; + + if (rctx->instruction == OCS_DECRYPT) { + /* + * For decrypt: + * - src sg list is: AAD|CT|tag + * - dst sg list expects: AAD|PT + * + * in_size == len(CT); out_size == len(PT) + */ + + /* req->cryptlen includes both CT and tag. */ + in_size = req->cryptlen - tag_size; + + /* out_size = PT size == CT size */ + out_size = in_size; + + /* len(dst_sg) == len(AAD) + len(PT) */ + dst_size = req->assoclen + out_size; + + /* + * Copy tag from source SG list to 'in_tag' buffer. + * + * Note: this needs to be done here, before DMA mapping src_sg. + */ + sg_pcopy_to_buffer(req->src, rctx->src_nents, rctx->in_tag, + tag_size, req->assoclen + in_size); + + } else { /* OCS_ENCRYPT */ + /* + * For encrypt: + * src sg list is: AAD|PT + * dst sg list expects: AAD|CT|tag + */ + /* in_size == len(PT) */ + in_size = req->cryptlen; + + /* + * In CCM mode the OCS engine appends the tag to the ciphertext, + * but in GCM mode the tag must be read from the tag registers + * and appended manually below + */ + out_size = (rctx->mode == OCS_MODE_CCM) ? in_size + tag_size : + in_size; + /* len(dst_sg) == len(AAD) + len(CT) + len(tag) */ + dst_size = req->assoclen + in_size + tag_size; + } + *src_dll_size = in_size; + + /* Get number of entries in output data SG list. */ + rctx->dst_nents = sg_nents_for_len(req->dst, dst_size); + if (rctx->dst_nents < 0) + return -EBADMSG; + + rctx->in_place = (req->src == req->dst) ? 1 : 0; + + /* Map destination; use bidirectional mapping for in-place case. */ + rctx->dst_dma_count = dma_map_sg(tctx->aes_dev->dev, req->dst, + rctx->dst_nents, + rctx->in_place ? DMA_BIDIRECTIONAL : + DMA_FROM_DEVICE); + if (rctx->dst_dma_count == 0 && rctx->dst_nents != 0) { + dev_err(tctx->aes_dev->dev, "Failed to map destination sg\n"); + return -ENOMEM; + } + + /* Create AAD DST list: maps dst[0:AAD_SIZE-1]. */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->dst, + rctx->dst_dma_count, + &rctx->aad_dst_dll, req->assoclen, + 0); + if (rc) + return rc; + + /* Create DST list: maps dst[AAD_SIZE:out_size] */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->dst, + rctx->dst_dma_count, &rctx->dst_dll, + out_size, req->assoclen); + if (rc) + return rc; + + if (rctx->in_place) { + /* If this is not CCM encrypt, we are done. */ + if (!(rctx->mode == OCS_MODE_CCM && + rctx->instruction == OCS_ENCRYPT)) { + /* + * SRC and DST are the same, so re-use the same DMA + * addresses (to avoid allocating new DMA lists + * identical to the dst ones). + */ + rctx->src_dll.dma_addr = rctx->dst_dll.dma_addr; + rctx->aad_src_dll.dma_addr = rctx->aad_dst_dll.dma_addr; + + return 0; + } + /* + * For CCM encrypt the input and output linked lists contain + * different amounts of data, so, we need to create different + * SRC and AAD SRC lists, even for the in-place case. + */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->dst, + rctx->dst_dma_count, + &rctx->aad_src_dll, + req->assoclen, 0); + if (rc) + return rc; + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->dst, + rctx->dst_dma_count, + &rctx->src_dll, in_size, + req->assoclen); + if (rc) + return rc; + + return 0; + } + /* Not in-place case. */ + + /* Map source SG. */ + rctx->src_dma_count = dma_map_sg(tctx->aes_dev->dev, req->src, + rctx->src_nents, DMA_TO_DEVICE); + if (rctx->src_dma_count == 0 && rctx->src_nents != 0) { + dev_err(tctx->aes_dev->dev, "Failed to map source sg\n"); + return -ENOMEM; + } + + /* Create AAD SRC list. */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->src, + rctx->src_dma_count, + &rctx->aad_src_dll, + req->assoclen, 0); + if (rc) + return rc; + + /* Create SRC list. */ + rc = ocs_create_linked_list_from_sg(tctx->aes_dev, req->src, + rctx->src_dma_count, + &rctx->src_dll, in_size, + req->assoclen); + if (rc) + return rc; + + if (req->assoclen == 0) + return 0; + + /* Copy AAD from src sg to dst sg using OCS DMA. */ + rc = ocs_aes_bypass_op(tctx->aes_dev, rctx->aad_dst_dll.dma_addr, + rctx->aad_src_dll.dma_addr, req->cryptlen); + if (rc) + dev_err(tctx->aes_dev->dev, + "Failed to copy source AAD to destination AAD\n"); + + return rc; +} + +static int kmb_ocs_aead_run(struct aead_request *req) +{ + struct ocs_aes_tctx *tctx = crypto_aead_ctx(crypto_aead_reqtfm(req)); + const int tag_size = crypto_aead_authsize(crypto_aead_reqtfm(req)); + struct ocs_aes_rctx *rctx = aead_request_ctx(req); + u32 in_size; /* The length of the data mapped by src_dll. */ + int rc; + + rc = kmb_ocs_aead_dma_prepare(req, &in_size); + if (rc) + goto exit; + + /* For CCM, we just call the OCS processing and we are done. */ + if (rctx->mode == OCS_MODE_CCM) { + rc = ocs_aes_ccm_op(tctx->aes_dev, tctx->cipher, + rctx->instruction, rctx->dst_dll.dma_addr, + rctx->src_dll.dma_addr, in_size, + req->iv, + rctx->aad_src_dll.dma_addr, req->assoclen, + rctx->in_tag, tag_size); + goto exit; + } + /* GCM case; invoke OCS processing. */ + rc = ocs_aes_gcm_op(tctx->aes_dev, tctx->cipher, + rctx->instruction, + rctx->dst_dll.dma_addr, + rctx->src_dll.dma_addr, in_size, + req->iv, + rctx->aad_src_dll.dma_addr, req->assoclen, + rctx->out_tag, tag_size); + if (rc) + goto exit; + + /* For GCM decrypt, we have to compare in_tag with out_tag. */ + if (rctx->instruction == OCS_DECRYPT) { + rc = memcmp(rctx->in_tag, rctx->out_tag, tag_size) ? + -EBADMSG : 0; + goto exit; + } + + /* For GCM encrypt, we must manually copy out_tag to DST sg. */ + + /* Clean-up must be called before the sg_pcopy_from_buffer() below. */ + kmb_ocs_aead_dma_cleanup(req); + + /* Copy tag to destination sg after AAD and CT. */ + sg_pcopy_from_buffer(req->dst, rctx->dst_nents, rctx->out_tag, + tag_size, req->assoclen + req->cryptlen); + + /* Return directly as DMA cleanup already done. */ + return 0; + +exit: + kmb_ocs_aead_dma_cleanup(req); + + return rc; +} + +static int kmb_ocs_aes_sk_do_one_request(struct crypto_engine *engine, + void *areq) +{ + struct skcipher_request *req = + container_of(areq, struct skcipher_request, base); + struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req); + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + int err; + + if (!tctx->aes_dev) { + err = -ENODEV; + goto exit; + } + + err = ocs_aes_set_key(tctx->aes_dev, tctx->key_len, tctx->key, + tctx->cipher); + if (err) + goto exit; + + err = kmb_ocs_sk_run(req); + +exit: + crypto_finalize_skcipher_request(engine, req, err); + + return 0; +} + +static int kmb_ocs_aes_aead_do_one_request(struct crypto_engine *engine, + void *areq) +{ + struct aead_request *req = container_of(areq, + struct aead_request, base); + struct ocs_aes_tctx *tctx = crypto_aead_ctx(crypto_aead_reqtfm(req)); + int err; + + if (!tctx->aes_dev) + return -ENODEV; + + err = ocs_aes_set_key(tctx->aes_dev, tctx->key_len, tctx->key, + tctx->cipher); + if (err) + goto exit; + + err = kmb_ocs_aead_run(req); + +exit: + crypto_finalize_aead_request(tctx->aes_dev->engine, req, err); + + return 0; +} + +static int kmb_ocs_aes_set_key(struct crypto_skcipher *tfm, const u8 *in_key, + unsigned int key_len) +{ + return kmb_ocs_sk_set_key(tfm, in_key, key_len, OCS_AES); +} + +static int kmb_ocs_aes_aead_set_key(struct crypto_aead *tfm, const u8 *in_key, + unsigned int key_len) +{ + return kmb_ocs_aead_set_key(tfm, in_key, key_len, OCS_AES); +} + +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB +static int kmb_ocs_aes_ecb_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_ENCRYPT, OCS_MODE_ECB); +} + +static int kmb_ocs_aes_ecb_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_DECRYPT, OCS_MODE_ECB); +} +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB */ + +static int kmb_ocs_aes_cbc_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_ENCRYPT, OCS_MODE_CBC); +} + +static int kmb_ocs_aes_cbc_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_DECRYPT, OCS_MODE_CBC); +} + +static int kmb_ocs_aes_ctr_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_ENCRYPT, OCS_MODE_CTR); +} + +static int kmb_ocs_aes_ctr_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_DECRYPT, OCS_MODE_CTR); +} + +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS +static int kmb_ocs_aes_cts_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_ENCRYPT, OCS_MODE_CTS); +} + +static int kmb_ocs_aes_cts_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_AES, OCS_DECRYPT, OCS_MODE_CTS); +} +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS */ + +static int kmb_ocs_aes_gcm_encrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_AES, OCS_ENCRYPT, OCS_MODE_GCM); +} + +static int kmb_ocs_aes_gcm_decrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_AES, OCS_DECRYPT, OCS_MODE_GCM); +} + +static int kmb_ocs_aes_ccm_encrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_AES, OCS_ENCRYPT, OCS_MODE_CCM); +} + +static int kmb_ocs_aes_ccm_decrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_AES, OCS_DECRYPT, OCS_MODE_CCM); +} + +static int kmb_ocs_sm4_set_key(struct crypto_skcipher *tfm, const u8 *in_key, + unsigned int key_len) +{ + return kmb_ocs_sk_set_key(tfm, in_key, key_len, OCS_SM4); +} + +static int kmb_ocs_sm4_aead_set_key(struct crypto_aead *tfm, const u8 *in_key, + unsigned int key_len) +{ + return kmb_ocs_aead_set_key(tfm, in_key, key_len, OCS_SM4); +} + +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB +static int kmb_ocs_sm4_ecb_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_ENCRYPT, OCS_MODE_ECB); +} + +static int kmb_ocs_sm4_ecb_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_DECRYPT, OCS_MODE_ECB); +} +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB */ + +static int kmb_ocs_sm4_cbc_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_ENCRYPT, OCS_MODE_CBC); +} + +static int kmb_ocs_sm4_cbc_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_DECRYPT, OCS_MODE_CBC); +} + +static int kmb_ocs_sm4_ctr_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_ENCRYPT, OCS_MODE_CTR); +} + +static int kmb_ocs_sm4_ctr_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_DECRYPT, OCS_MODE_CTR); +} + +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS +static int kmb_ocs_sm4_cts_encrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_ENCRYPT, OCS_MODE_CTS); +} + +static int kmb_ocs_sm4_cts_decrypt(struct skcipher_request *req) +{ + return kmb_ocs_sk_common(req, OCS_SM4, OCS_DECRYPT, OCS_MODE_CTS); +} +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS */ + +static int kmb_ocs_sm4_gcm_encrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_SM4, OCS_ENCRYPT, OCS_MODE_GCM); +} + +static int kmb_ocs_sm4_gcm_decrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_SM4, OCS_DECRYPT, OCS_MODE_GCM); +} + +static int kmb_ocs_sm4_ccm_encrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_SM4, OCS_ENCRYPT, OCS_MODE_CCM); +} + +static int kmb_ocs_sm4_ccm_decrypt(struct aead_request *req) +{ + return kmb_ocs_aead_common(req, OCS_SM4, OCS_DECRYPT, OCS_MODE_CCM); +} + +static int ocs_aes_init_tfm(struct crypto_skcipher *tfm) +{ + const char *alg_name = crypto_tfm_alg_name(&tfm->base); + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + struct crypto_sync_skcipher *blk; + + /* set fallback cipher in case it will be needed */ + blk = crypto_alloc_sync_skcipher(alg_name, 0, CRYPTO_ALG_NEED_FALLBACK); + if (IS_ERR(blk)) + return PTR_ERR(blk); + + tctx->sw_cipher.sk = blk; + + crypto_skcipher_set_reqsize(tfm, sizeof(struct ocs_aes_rctx)); + + return 0; +} + +static int ocs_sm4_init_tfm(struct crypto_skcipher *tfm) +{ + crypto_skcipher_set_reqsize(tfm, sizeof(struct ocs_aes_rctx)); + + return 0; +} + +static inline void clear_key(struct ocs_aes_tctx *tctx) +{ + memzero_explicit(tctx->key, OCS_AES_KEYSIZE_256); + + /* Zero key registers if set */ + if (tctx->aes_dev) + ocs_aes_set_key(tctx->aes_dev, OCS_AES_KEYSIZE_256, + tctx->key, OCS_AES); +} + +static void ocs_exit_tfm(struct crypto_skcipher *tfm) +{ + struct ocs_aes_tctx *tctx = crypto_skcipher_ctx(tfm); + + clear_key(tctx); + + if (tctx->sw_cipher.sk) { + crypto_free_sync_skcipher(tctx->sw_cipher.sk); + tctx->sw_cipher.sk = NULL; + } +} + +static int ocs_aes_aead_cra_init(struct crypto_aead *tfm) +{ + const char *alg_name = crypto_tfm_alg_name(&tfm->base); + struct ocs_aes_tctx *tctx = crypto_aead_ctx(tfm); + struct crypto_aead *blk; + + /* Set fallback cipher in case it will be needed */ + blk = crypto_alloc_aead(alg_name, 0, CRYPTO_ALG_NEED_FALLBACK); + if (IS_ERR(blk)) + return PTR_ERR(blk); + + tctx->sw_cipher.aead = blk; + + crypto_aead_set_reqsize(tfm, + max(sizeof(struct ocs_aes_rctx), + (sizeof(struct aead_request) + + crypto_aead_reqsize(tctx->sw_cipher.aead)))); + + return 0; +} + +static int kmb_ocs_aead_ccm_setauthsize(struct crypto_aead *tfm, + unsigned int authsize) +{ + switch (authsize) { + case 4: + case 6: + case 8: + case 10: + case 12: + case 14: + case 16: + return 0; + default: + return -EINVAL; + } +} + +static int kmb_ocs_aead_gcm_setauthsize(struct crypto_aead *tfm, + unsigned int authsize) +{ + return crypto_gcm_check_authsize(authsize); +} + +static int ocs_sm4_aead_cra_init(struct crypto_aead *tfm) +{ + crypto_aead_set_reqsize(tfm, sizeof(struct ocs_aes_rctx)); + + return 0; +} + +static void ocs_aead_cra_exit(struct crypto_aead *tfm) +{ + struct ocs_aes_tctx *tctx = crypto_aead_ctx(tfm); + + clear_key(tctx); + + if (tctx->sw_cipher.aead) { + crypto_free_aead(tctx->sw_cipher.aead); + tctx->sw_cipher.aead = NULL; + } +} + +static struct skcipher_engine_alg algs[] = { +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB + { + .base.base.cra_name = "ecb(aes)", + .base.base.cra_driver_name = "ecb-aes-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_NEED_FALLBACK, + .base.base.cra_blocksize = AES_BLOCK_SIZE, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_AES_MIN_KEY_SIZE, + .base.max_keysize = OCS_AES_MAX_KEY_SIZE, + .base.setkey = kmb_ocs_aes_set_key, + .base.encrypt = kmb_ocs_aes_ecb_encrypt, + .base.decrypt = kmb_ocs_aes_ecb_decrypt, + .base.init = ocs_aes_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + }, +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB */ + { + .base.base.cra_name = "cbc(aes)", + .base.base.cra_driver_name = "cbc-aes-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_NEED_FALLBACK, + .base.base.cra_blocksize = AES_BLOCK_SIZE, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_AES_MIN_KEY_SIZE, + .base.max_keysize = OCS_AES_MAX_KEY_SIZE, + .base.ivsize = AES_BLOCK_SIZE, + .base.setkey = kmb_ocs_aes_set_key, + .base.encrypt = kmb_ocs_aes_cbc_encrypt, + .base.decrypt = kmb_ocs_aes_cbc_decrypt, + .base.init = ocs_aes_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + }, + { + .base.base.cra_name = "ctr(aes)", + .base.base.cra_driver_name = "ctr-aes-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_NEED_FALLBACK, + .base.base.cra_blocksize = 1, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_AES_MIN_KEY_SIZE, + .base.max_keysize = OCS_AES_MAX_KEY_SIZE, + .base.ivsize = AES_BLOCK_SIZE, + .base.setkey = kmb_ocs_aes_set_key, + .base.encrypt = kmb_ocs_aes_ctr_encrypt, + .base.decrypt = kmb_ocs_aes_ctr_decrypt, + .base.init = ocs_aes_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + }, +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS + { + .base.base.cra_name = "cts(cbc(aes))", + .base.base.cra_driver_name = "cts-aes-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_NEED_FALLBACK, + .base.base.cra_blocksize = AES_BLOCK_SIZE, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_AES_MIN_KEY_SIZE, + .base.max_keysize = OCS_AES_MAX_KEY_SIZE, + .base.ivsize = AES_BLOCK_SIZE, + .base.setkey = kmb_ocs_aes_set_key, + .base.encrypt = kmb_ocs_aes_cts_encrypt, + .base.decrypt = kmb_ocs_aes_cts_decrypt, + .base.init = ocs_aes_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + }, +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS */ +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB + { + .base.base.cra_name = "ecb(sm4)", + .base.base.cra_driver_name = "ecb-sm4-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY, + .base.base.cra_blocksize = AES_BLOCK_SIZE, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_SM4_KEY_SIZE, + .base.max_keysize = OCS_SM4_KEY_SIZE, + .base.setkey = kmb_ocs_sm4_set_key, + .base.encrypt = kmb_ocs_sm4_ecb_encrypt, + .base.decrypt = kmb_ocs_sm4_ecb_decrypt, + .base.init = ocs_sm4_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + }, +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB */ + { + .base.base.cra_name = "cbc(sm4)", + .base.base.cra_driver_name = "cbc-sm4-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY, + .base.base.cra_blocksize = AES_BLOCK_SIZE, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_SM4_KEY_SIZE, + .base.max_keysize = OCS_SM4_KEY_SIZE, + .base.ivsize = AES_BLOCK_SIZE, + .base.setkey = kmb_ocs_sm4_set_key, + .base.encrypt = kmb_ocs_sm4_cbc_encrypt, + .base.decrypt = kmb_ocs_sm4_cbc_decrypt, + .base.init = ocs_sm4_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + }, + { + .base.base.cra_name = "ctr(sm4)", + .base.base.cra_driver_name = "ctr-sm4-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY, + .base.base.cra_blocksize = 1, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_SM4_KEY_SIZE, + .base.max_keysize = OCS_SM4_KEY_SIZE, + .base.ivsize = AES_BLOCK_SIZE, + .base.setkey = kmb_ocs_sm4_set_key, + .base.encrypt = kmb_ocs_sm4_ctr_encrypt, + .base.decrypt = kmb_ocs_sm4_ctr_decrypt, + .base.init = ocs_sm4_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + }, +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS + { + .base.base.cra_name = "cts(cbc(sm4))", + .base.base.cra_driver_name = "cts-sm4-keembay-ocs", + .base.base.cra_priority = KMB_OCS_PRIORITY, + .base.base.cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY, + .base.base.cra_blocksize = AES_BLOCK_SIZE, + .base.base.cra_ctxsize = sizeof(struct ocs_aes_tctx), + .base.base.cra_module = THIS_MODULE, + .base.base.cra_alignmask = 0, + + .base.min_keysize = OCS_SM4_KEY_SIZE, + .base.max_keysize = OCS_SM4_KEY_SIZE, + .base.ivsize = AES_BLOCK_SIZE, + .base.setkey = kmb_ocs_sm4_set_key, + .base.encrypt = kmb_ocs_sm4_cts_encrypt, + .base.decrypt = kmb_ocs_sm4_cts_decrypt, + .base.init = ocs_sm4_init_tfm, + .base.exit = ocs_exit_tfm, + .op.do_one_request = kmb_ocs_aes_sk_do_one_request, + } +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS */ +}; + +static struct aead_engine_alg algs_aead[] = { + { + .base.base = { + .cra_name = "gcm(aes)", + .cra_driver_name = "gcm-aes-keembay-ocs", + .cra_priority = KMB_OCS_PRIORITY, + .cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_NEED_FALLBACK, + .cra_blocksize = 1, + .cra_ctxsize = sizeof(struct ocs_aes_tctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + }, + .base.init = ocs_aes_aead_cra_init, + .base.exit = ocs_aead_cra_exit, + .base.ivsize = GCM_AES_IV_SIZE, + .base.maxauthsize = AES_BLOCK_SIZE, + .base.setauthsize = kmb_ocs_aead_gcm_setauthsize, + .base.setkey = kmb_ocs_aes_aead_set_key, + .base.encrypt = kmb_ocs_aes_gcm_encrypt, + .base.decrypt = kmb_ocs_aes_gcm_decrypt, + .op.do_one_request = kmb_ocs_aes_aead_do_one_request, + }, + { + .base.base = { + .cra_name = "ccm(aes)", + .cra_driver_name = "ccm-aes-keembay-ocs", + .cra_priority = KMB_OCS_PRIORITY, + .cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY | + CRYPTO_ALG_NEED_FALLBACK, + .cra_blocksize = 1, + .cra_ctxsize = sizeof(struct ocs_aes_tctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + }, + .base.init = ocs_aes_aead_cra_init, + .base.exit = ocs_aead_cra_exit, + .base.ivsize = AES_BLOCK_SIZE, + .base.maxauthsize = AES_BLOCK_SIZE, + .base.setauthsize = kmb_ocs_aead_ccm_setauthsize, + .base.setkey = kmb_ocs_aes_aead_set_key, + .base.encrypt = kmb_ocs_aes_ccm_encrypt, + .base.decrypt = kmb_ocs_aes_ccm_decrypt, + .op.do_one_request = kmb_ocs_aes_aead_do_one_request, + }, + { + .base.base = { + .cra_name = "gcm(sm4)", + .cra_driver_name = "gcm-sm4-keembay-ocs", + .cra_priority = KMB_OCS_PRIORITY, + .cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY, + .cra_blocksize = 1, + .cra_ctxsize = sizeof(struct ocs_aes_tctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + }, + .base.init = ocs_sm4_aead_cra_init, + .base.exit = ocs_aead_cra_exit, + .base.ivsize = GCM_AES_IV_SIZE, + .base.maxauthsize = AES_BLOCK_SIZE, + .base.setauthsize = kmb_ocs_aead_gcm_setauthsize, + .base.setkey = kmb_ocs_sm4_aead_set_key, + .base.encrypt = kmb_ocs_sm4_gcm_encrypt, + .base.decrypt = kmb_ocs_sm4_gcm_decrypt, + .op.do_one_request = kmb_ocs_aes_aead_do_one_request, + }, + { + .base.base = { + .cra_name = "ccm(sm4)", + .cra_driver_name = "ccm-sm4-keembay-ocs", + .cra_priority = KMB_OCS_PRIORITY, + .cra_flags = CRYPTO_ALG_ASYNC | + CRYPTO_ALG_KERN_DRIVER_ONLY, + .cra_blocksize = 1, + .cra_ctxsize = sizeof(struct ocs_aes_tctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + }, + .base.init = ocs_sm4_aead_cra_init, + .base.exit = ocs_aead_cra_exit, + .base.ivsize = AES_BLOCK_SIZE, + .base.maxauthsize = AES_BLOCK_SIZE, + .base.setauthsize = kmb_ocs_aead_ccm_setauthsize, + .base.setkey = kmb_ocs_sm4_aead_set_key, + .base.encrypt = kmb_ocs_sm4_ccm_encrypt, + .base.decrypt = kmb_ocs_sm4_ccm_decrypt, + .op.do_one_request = kmb_ocs_aes_aead_do_one_request, + } +}; + +static void unregister_aes_algs(struct ocs_aes_dev *aes_dev) +{ + crypto_engine_unregister_aeads(algs_aead, ARRAY_SIZE(algs_aead)); + crypto_engine_unregister_skciphers(algs, ARRAY_SIZE(algs)); +} + +static int register_aes_algs(struct ocs_aes_dev *aes_dev) +{ + int ret; + + /* + * If any algorithm fails to register, all preceding algorithms that + * were successfully registered will be automatically unregistered. + */ + ret = crypto_engine_register_aeads(algs_aead, ARRAY_SIZE(algs_aead)); + if (ret) + return ret; + + ret = crypto_engine_register_skciphers(algs, ARRAY_SIZE(algs)); + if (ret) + crypto_engine_unregister_aeads(algs_aead, ARRAY_SIZE(algs)); + + return ret; +} + +/* Device tree driver match. */ +static const struct of_device_id kmb_ocs_aes_of_match[] = { + { + .compatible = "intel,keembay-ocs-aes", + }, + {} +}; + +static void kmb_ocs_aes_remove(struct platform_device *pdev) +{ + struct ocs_aes_dev *aes_dev; + + aes_dev = platform_get_drvdata(pdev); + + unregister_aes_algs(aes_dev); + + spin_lock(&ocs_aes.lock); + list_del(&aes_dev->list); + spin_unlock(&ocs_aes.lock); + + crypto_engine_exit(aes_dev->engine); +} + +static int kmb_ocs_aes_probe(struct platform_device *pdev) +{ + struct device *dev = &pdev->dev; + struct ocs_aes_dev *aes_dev; + int rc; + + aes_dev = devm_kzalloc(dev, sizeof(*aes_dev), GFP_KERNEL); + if (!aes_dev) + return -ENOMEM; + + aes_dev->dev = dev; + + platform_set_drvdata(pdev, aes_dev); + + rc = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32)); + if (rc) { + dev_err(dev, "Failed to set 32 bit dma mask %d\n", rc); + return rc; + } + + /* Get base register address. */ + aes_dev->base_reg = devm_platform_ioremap_resource(pdev, 0); + if (IS_ERR(aes_dev->base_reg)) + return PTR_ERR(aes_dev->base_reg); + + /* Get and request IRQ */ + aes_dev->irq = platform_get_irq(pdev, 0); + if (aes_dev->irq < 0) + return aes_dev->irq; + + rc = devm_request_threaded_irq(dev, aes_dev->irq, ocs_aes_irq_handler, + NULL, 0, "keembay-ocs-aes", aes_dev); + if (rc < 0) { + dev_err(dev, "Could not request IRQ\n"); + return rc; + } + + INIT_LIST_HEAD(&aes_dev->list); + spin_lock(&ocs_aes.lock); + list_add_tail(&aes_dev->list, &ocs_aes.dev_list); + spin_unlock(&ocs_aes.lock); + + init_completion(&aes_dev->irq_completion); + + /* Initialize crypto engine */ + aes_dev->engine = crypto_engine_alloc_init(dev, true); + if (!aes_dev->engine) { + rc = -ENOMEM; + goto list_del; + } + + rc = crypto_engine_start(aes_dev->engine); + if (rc) { + dev_err(dev, "Could not start crypto engine\n"); + goto cleanup; + } + + rc = register_aes_algs(aes_dev); + if (rc) { + dev_err(dev, + "Could not register OCS algorithms with Crypto API\n"); + goto cleanup; + } + + return 0; + +cleanup: + crypto_engine_exit(aes_dev->engine); +list_del: + spin_lock(&ocs_aes.lock); + list_del(&aes_dev->list); + spin_unlock(&ocs_aes.lock); + + return rc; +} + +/* The OCS driver is a platform device. */ +static struct platform_driver kmb_ocs_aes_driver = { + .probe = kmb_ocs_aes_probe, + .remove = kmb_ocs_aes_remove, + .driver = { + .name = DRV_NAME, + .of_match_table = kmb_ocs_aes_of_match, + }, +}; + +module_platform_driver(kmb_ocs_aes_driver); + +MODULE_DESCRIPTION("Intel Keem Bay Offload and Crypto Subsystem (OCS) AES/SM4 Driver"); +MODULE_LICENSE("GPL"); + +MODULE_ALIAS_CRYPTO("cbc-aes-keembay-ocs"); +MODULE_ALIAS_CRYPTO("ctr-aes-keembay-ocs"); +MODULE_ALIAS_CRYPTO("gcm-aes-keembay-ocs"); +MODULE_ALIAS_CRYPTO("ccm-aes-keembay-ocs"); + +MODULE_ALIAS_CRYPTO("cbc-sm4-keembay-ocs"); +MODULE_ALIAS_CRYPTO("ctr-sm4-keembay-ocs"); +MODULE_ALIAS_CRYPTO("gcm-sm4-keembay-ocs"); +MODULE_ALIAS_CRYPTO("ccm-sm4-keembay-ocs"); + +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB +MODULE_ALIAS_CRYPTO("ecb-aes-keembay-ocs"); +MODULE_ALIAS_CRYPTO("ecb-sm4-keembay-ocs"); +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_ECB */ + +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS +MODULE_ALIAS_CRYPTO("cts-aes-keembay-ocs"); +MODULE_ALIAS_CRYPTO("cts-sm4-keembay-ocs"); +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_AES_SM4_CTS */ diff --git a/drivers/crypto/intel/keembay/keembay-ocs-ecc.c b/drivers/crypto/intel/keembay/keembay-ocs-ecc.c new file mode 100644 index 000000000000..59308926399d --- /dev/null +++ b/drivers/crypto/intel/keembay/keembay-ocs-ecc.c @@ -0,0 +1,1007 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Intel Keem Bay OCS ECC Crypto Driver. + * + * Copyright (C) 2019-2021 Intel Corporation + */ + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include <crypto/ecc_curve.h> +#include <crypto/ecdh.h> +#include <crypto/engine.h> +#include <crypto/internal/ecc.h> +#include <crypto/internal/kpp.h> +#include <crypto/kpp.h> +#include <crypto/rng.h> +#include <linux/clk.h> +#include <linux/completion.h> +#include <linux/err.h> +#include <linux/fips.h> +#include <linux/interrupt.h> +#include <linux/io.h> +#include <linux/iopoll.h> +#include <linux/irq.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/platform_device.h> +#include <linux/scatterlist.h> +#include <linux/string.h> + +#define DRV_NAME "keembay-ocs-ecc" + +#define KMB_OCS_ECC_PRIORITY 350 + +#define HW_OFFS_OCS_ECC_COMMAND 0x00000000 +#define HW_OFFS_OCS_ECC_STATUS 0x00000004 +#define HW_OFFS_OCS_ECC_DATA_IN 0x00000080 +#define HW_OFFS_OCS_ECC_CX_DATA_OUT 0x00000100 +#define HW_OFFS_OCS_ECC_CY_DATA_OUT 0x00000180 +#define HW_OFFS_OCS_ECC_ISR 0x00000400 +#define HW_OFFS_OCS_ECC_IER 0x00000404 + +#define HW_OCS_ECC_ISR_INT_STATUS_DONE BIT(0) +#define HW_OCS_ECC_COMMAND_INS_BP BIT(0) + +#define HW_OCS_ECC_COMMAND_START_VAL BIT(0) + +#define OCS_ECC_OP_SIZE_384 BIT(8) +#define OCS_ECC_OP_SIZE_256 0 + +/* ECC Instruction : for ECC_COMMAND */ +#define OCS_ECC_INST_WRITE_AX (0x1 << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_WRITE_AY (0x2 << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_WRITE_BX_D (0x3 << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_WRITE_BY_L (0x4 << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_WRITE_P (0x5 << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_WRITE_A (0x6 << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_CALC_D_IDX_A (0x8 << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_CALC_A_POW_B_MODP (0xB << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_CALC_A_MUL_B_MODP (0xC << HW_OCS_ECC_COMMAND_INS_BP) +#define OCS_ECC_INST_CALC_A_ADD_B_MODP (0xD << HW_OCS_ECC_COMMAND_INS_BP) + +#define ECC_ENABLE_INTR 1 + +#define POLL_USEC 100 +#define TIMEOUT_USEC 10000 + +#define KMB_ECC_VLI_MAX_DIGITS ECC_CURVE_NIST_P384_DIGITS +#define KMB_ECC_VLI_MAX_BYTES (KMB_ECC_VLI_MAX_DIGITS \ + << ECC_DIGITS_TO_BYTES_SHIFT) + +#define POW_CUBE 3 + +/** + * struct ocs_ecc_dev - ECC device context + * @list: List of device contexts + * @dev: OCS ECC device + * @base_reg: IO base address of OCS ECC + * @engine: Crypto engine for the device + * @irq_done: IRQ done completion. + * @irq: IRQ number + */ +struct ocs_ecc_dev { + struct list_head list; + struct device *dev; + void __iomem *base_reg; + struct crypto_engine *engine; + struct completion irq_done; + int irq; +}; + +/** + * struct ocs_ecc_ctx - Transformation context. + * @ecc_dev: The ECC driver associated with this context. + * @curve: The elliptic curve used by this transformation. + * @private_key: The private key. + */ +struct ocs_ecc_ctx { + struct ocs_ecc_dev *ecc_dev; + const struct ecc_curve *curve; + u64 private_key[KMB_ECC_VLI_MAX_DIGITS]; +}; + +/* Driver data. */ +struct ocs_ecc_drv { + struct list_head dev_list; + spinlock_t lock; /* Protects dev_list. */ +}; + +/* Global variable holding the list of OCS ECC devices (only one expected). */ +static struct ocs_ecc_drv ocs_ecc = { + .dev_list = LIST_HEAD_INIT(ocs_ecc.dev_list), + .lock = __SPIN_LOCK_UNLOCKED(ocs_ecc.lock), +}; + +/* Get OCS ECC tfm context from kpp_request. */ +static inline struct ocs_ecc_ctx *kmb_ocs_ecc_tctx(struct kpp_request *req) +{ + return kpp_tfm_ctx(crypto_kpp_reqtfm(req)); +} + +/* Converts number of digits to number of bytes. */ +static inline unsigned int digits_to_bytes(unsigned int n) +{ + return n << ECC_DIGITS_TO_BYTES_SHIFT; +} + +/* + * Wait for ECC idle i.e when an operation (other than write operations) + * is done. + */ +static inline int ocs_ecc_wait_idle(struct ocs_ecc_dev *dev) +{ + u32 value; + + return readl_poll_timeout((dev->base_reg + HW_OFFS_OCS_ECC_STATUS), + value, + !(value & HW_OCS_ECC_ISR_INT_STATUS_DONE), + POLL_USEC, TIMEOUT_USEC); +} + +static void ocs_ecc_cmd_start(struct ocs_ecc_dev *ecc_dev, u32 op_size) +{ + iowrite32(op_size | HW_OCS_ECC_COMMAND_START_VAL, + ecc_dev->base_reg + HW_OFFS_OCS_ECC_COMMAND); +} + +/* Direct write of u32 buffer to ECC engine with associated instruction. */ +static void ocs_ecc_write_cmd_and_data(struct ocs_ecc_dev *dev, + u32 op_size, + u32 inst, + const void *data_in, + size_t data_size) +{ + iowrite32(op_size | inst, dev->base_reg + HW_OFFS_OCS_ECC_COMMAND); + + /* MMIO Write src uint32 to dst. */ + memcpy_toio(dev->base_reg + HW_OFFS_OCS_ECC_DATA_IN, data_in, + data_size); +} + +/* Start OCS ECC operation and wait for its completion. */ +static int ocs_ecc_trigger_op(struct ocs_ecc_dev *ecc_dev, u32 op_size, + u32 inst) +{ + reinit_completion(&ecc_dev->irq_done); + + iowrite32(ECC_ENABLE_INTR, ecc_dev->base_reg + HW_OFFS_OCS_ECC_IER); + iowrite32(op_size | inst, ecc_dev->base_reg + HW_OFFS_OCS_ECC_COMMAND); + + return wait_for_completion_interruptible(&ecc_dev->irq_done); +} + +/** + * ocs_ecc_read_cx_out() - Read the CX data output buffer. + * @dev: The OCS ECC device to read from. + * @cx_out: The buffer where to store the CX value. Must be at least + * @byte_count byte long. + * @byte_count: The amount of data to read. + */ +static inline void ocs_ecc_read_cx_out(struct ocs_ecc_dev *dev, void *cx_out, + size_t byte_count) +{ + memcpy_fromio(cx_out, dev->base_reg + HW_OFFS_OCS_ECC_CX_DATA_OUT, + byte_count); +} + +/** + * ocs_ecc_read_cy_out() - Read the CX data output buffer. + * @dev: The OCS ECC device to read from. + * @cy_out: The buffer where to store the CY value. Must be at least + * @byte_count byte long. + * @byte_count: The amount of data to read. + */ +static inline void ocs_ecc_read_cy_out(struct ocs_ecc_dev *dev, void *cy_out, + size_t byte_count) +{ + memcpy_fromio(cy_out, dev->base_reg + HW_OFFS_OCS_ECC_CY_DATA_OUT, + byte_count); +} + +static struct ocs_ecc_dev *kmb_ocs_ecc_find_dev(struct ocs_ecc_ctx *tctx) +{ + if (tctx->ecc_dev) + return tctx->ecc_dev; + + spin_lock(&ocs_ecc.lock); + + /* Only a single OCS device available. */ + tctx->ecc_dev = list_first_entry(&ocs_ecc.dev_list, struct ocs_ecc_dev, + list); + + spin_unlock(&ocs_ecc.lock); + + return tctx->ecc_dev; +} + +/* Do point multiplication using OCS ECC HW. */ +static int kmb_ecc_point_mult(struct ocs_ecc_dev *ecc_dev, + struct ecc_point *result, + const struct ecc_point *point, + u64 *scalar, + const struct ecc_curve *curve) +{ + u8 sca[KMB_ECC_VLI_MAX_BYTES]; /* Use the maximum data size. */ + u32 op_size = (curve->g.ndigits > ECC_CURVE_NIST_P256_DIGITS) ? + OCS_ECC_OP_SIZE_384 : OCS_ECC_OP_SIZE_256; + size_t nbytes = digits_to_bytes(curve->g.ndigits); + int rc = 0; + + /* Generate random nbytes for Simple and Differential SCA protection. */ + rc = crypto_get_default_rng(); + if (rc) + return rc; + + rc = crypto_rng_get_bytes(crypto_default_rng, sca, nbytes); + crypto_put_default_rng(); + if (rc) + return rc; + + /* Wait engine to be idle before starting new operation. */ + rc = ocs_ecc_wait_idle(ecc_dev); + if (rc) + return rc; + + /* Send ecc_start pulse as well as indicating operation size. */ + ocs_ecc_cmd_start(ecc_dev, op_size); + + /* Write ax param; Base point (Gx). */ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_AX, + point->x, nbytes); + + /* Write ay param; Base point (Gy). */ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_AY, + point->y, nbytes); + + /* + * Write the private key into DATA_IN reg. + * + * Since DATA_IN register is used to write different values during the + * computation private Key value is overwritten with + * side-channel-resistance value. + */ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_BX_D, + scalar, nbytes); + + /* Write operand by/l. */ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_BY_L, + sca, nbytes); + memzero_explicit(sca, sizeof(sca)); + + /* Write p = curve prime(GF modulus). */ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_P, + curve->p, nbytes); + + /* Write a = curve coefficient. */ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_A, + curve->a, nbytes); + + /* Make hardware perform the multiplication. */ + rc = ocs_ecc_trigger_op(ecc_dev, op_size, OCS_ECC_INST_CALC_D_IDX_A); + if (rc) + return rc; + + /* Read result. */ + ocs_ecc_read_cx_out(ecc_dev, result->x, nbytes); + ocs_ecc_read_cy_out(ecc_dev, result->y, nbytes); + + return 0; +} + +/** + * kmb_ecc_do_scalar_op() - Perform Scalar operation using OCS ECC HW. + * @ecc_dev: The OCS ECC device to use. + * @scalar_out: Where to store the output scalar. + * @scalar_a: Input scalar operand 'a'. + * @scalar_b: Input scalar operand 'b' + * @curve: The curve on which the operation is performed. + * @ndigits: The size of the operands (in digits). + * @inst: The operation to perform (as an OCS ECC instruction). + * + * Return: 0 on success, negative error code otherwise. + */ +static int kmb_ecc_do_scalar_op(struct ocs_ecc_dev *ecc_dev, u64 *scalar_out, + const u64 *scalar_a, const u64 *scalar_b, + const struct ecc_curve *curve, + unsigned int ndigits, const u32 inst) +{ + u32 op_size = (ndigits > ECC_CURVE_NIST_P256_DIGITS) ? + OCS_ECC_OP_SIZE_384 : OCS_ECC_OP_SIZE_256; + size_t nbytes = digits_to_bytes(ndigits); + int rc; + + /* Wait engine to be idle before starting new operation. */ + rc = ocs_ecc_wait_idle(ecc_dev); + if (rc) + return rc; + + /* Send ecc_start pulse as well as indicating operation size. */ + ocs_ecc_cmd_start(ecc_dev, op_size); + + /* Write ax param (Base point (Gx).*/ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_AX, + scalar_a, nbytes); + + /* Write ay param Base point (Gy).*/ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_AY, + scalar_b, nbytes); + + /* Write p = curve prime(GF modulus).*/ + ocs_ecc_write_cmd_and_data(ecc_dev, op_size, OCS_ECC_INST_WRITE_P, + curve->p, nbytes); + + /* Give instruction A.B or A+B to ECC engine. */ + rc = ocs_ecc_trigger_op(ecc_dev, op_size, inst); + if (rc) + return rc; + + ocs_ecc_read_cx_out(ecc_dev, scalar_out, nbytes); + + if (vli_is_zero(scalar_out, ndigits)) + return -EINVAL; + + return 0; +} + +/* SP800-56A section 5.6.2.3.4 partial verification: ephemeral keys only */ +static int kmb_ocs_ecc_is_pubkey_valid_partial(struct ocs_ecc_dev *ecc_dev, + const struct ecc_curve *curve, + struct ecc_point *pk) +{ + u64 xxx[KMB_ECC_VLI_MAX_DIGITS] = { 0 }; + u64 yy[KMB_ECC_VLI_MAX_DIGITS] = { 0 }; + u64 w[KMB_ECC_VLI_MAX_DIGITS] = { 0 }; + int rc; + + if (WARN_ON(pk->ndigits != curve->g.ndigits)) + return -EINVAL; + + /* Check 1: Verify key is not the zero point. */ + if (ecc_point_is_zero(pk)) + return -EINVAL; + + /* Check 2: Verify key is in the range [0, p-1]. */ + if (vli_cmp(curve->p, pk->x, pk->ndigits) != 1) + return -EINVAL; + + if (vli_cmp(curve->p, pk->y, pk->ndigits) != 1) + return -EINVAL; + + /* Check 3: Verify that y^2 == (x^3 + a·x + b) mod p */ + + /* y^2 */ + /* Compute y^2 -> store in yy */ + rc = kmb_ecc_do_scalar_op(ecc_dev, yy, pk->y, pk->y, curve, pk->ndigits, + OCS_ECC_INST_CALC_A_MUL_B_MODP); + if (rc) + goto exit; + + /* x^3 */ + /* Assigning w = 3, used for calculating x^3. */ + w[0] = POW_CUBE; + /* Load the next stage.*/ + rc = kmb_ecc_do_scalar_op(ecc_dev, xxx, pk->x, w, curve, pk->ndigits, + OCS_ECC_INST_CALC_A_POW_B_MODP); + if (rc) + goto exit; + + /* Do a*x -> store in w. */ + rc = kmb_ecc_do_scalar_op(ecc_dev, w, curve->a, pk->x, curve, + pk->ndigits, + OCS_ECC_INST_CALC_A_MUL_B_MODP); + if (rc) + goto exit; + + /* Do ax + b == w + b; store in w. */ + rc = kmb_ecc_do_scalar_op(ecc_dev, w, w, curve->b, curve, + pk->ndigits, + OCS_ECC_INST_CALC_A_ADD_B_MODP); + if (rc) + goto exit; + + /* x^3 + ax + b == x^3 + w -> store in w. */ + rc = kmb_ecc_do_scalar_op(ecc_dev, w, xxx, w, curve, pk->ndigits, + OCS_ECC_INST_CALC_A_ADD_B_MODP); + if (rc) + goto exit; + + /* Compare y^2 == x^3 + a·x + b. */ + rc = vli_cmp(yy, w, pk->ndigits); + if (rc) + rc = -EINVAL; + +exit: + memzero_explicit(xxx, sizeof(xxx)); + memzero_explicit(yy, sizeof(yy)); + memzero_explicit(w, sizeof(w)); + + return rc; +} + +/* SP800-56A section 5.6.2.3.3 full verification */ +static int kmb_ocs_ecc_is_pubkey_valid_full(struct ocs_ecc_dev *ecc_dev, + const struct ecc_curve *curve, + struct ecc_point *pk) +{ + struct ecc_point *nQ; + int rc; + + /* Checks 1 through 3 */ + rc = kmb_ocs_ecc_is_pubkey_valid_partial(ecc_dev, curve, pk); + if (rc) + return rc; + + /* Check 4: Verify that nQ is the zero point. */ + nQ = ecc_alloc_point(pk->ndigits); + if (!nQ) + return -ENOMEM; + + rc = kmb_ecc_point_mult(ecc_dev, nQ, pk, curve->n, curve); + if (rc) + goto exit; + + if (!ecc_point_is_zero(nQ)) + rc = -EINVAL; + +exit: + ecc_free_point(nQ); + + return rc; +} + +static int kmb_ecc_is_key_valid(const struct ecc_curve *curve, + const u64 *private_key, size_t private_key_len) +{ + size_t ndigits = curve->g.ndigits; + u64 one[KMB_ECC_VLI_MAX_DIGITS] = {1}; + u64 res[KMB_ECC_VLI_MAX_DIGITS]; + + if (private_key_len != digits_to_bytes(ndigits)) + return -EINVAL; + + if (!private_key) + return -EINVAL; + + /* Make sure the private key is in the range [2, n-3]. */ + if (vli_cmp(one, private_key, ndigits) != -1) + return -EINVAL; + + vli_sub(res, curve->n, one, ndigits); + vli_sub(res, res, one, ndigits); + if (vli_cmp(res, private_key, ndigits) != 1) + return -EINVAL; + + return 0; +} + +/* + * ECC private keys are generated using the method of extra random bits, + * equivalent to that described in FIPS 186-4, Appendix B.4.1. + * + * d = (c mod(n–1)) + 1 where c is a string of random bits, 64 bits longer + * than requested + * 0 <= c mod(n-1) <= n-2 and implies that + * 1 <= d <= n-1 + * + * This method generates a private key uniformly distributed in the range + * [1, n-1]. + */ +static int kmb_ecc_gen_privkey(const struct ecc_curve *curve, u64 *privkey) +{ + size_t nbytes = digits_to_bytes(curve->g.ndigits); + u64 priv[KMB_ECC_VLI_MAX_DIGITS]; + size_t nbits; + int rc; + + nbits = vli_num_bits(curve->n, curve->g.ndigits); + + /* Check that N is included in Table 1 of FIPS 186-4, section 6.1.1 */ + if (nbits < 160 || curve->g.ndigits > ARRAY_SIZE(priv)) + return -EINVAL; + + /* + * FIPS 186-4 recommends that the private key should be obtained from a + * RBG with a security strength equal to or greater than the security + * strength associated with N. + * + * The maximum security strength identified by NIST SP800-57pt1r4 for + * ECC is 256 (N >= 512). + * + * This condition is met by the default RNG because it selects a favored + * DRBG with a security strength of 256. + */ + if (crypto_get_default_rng()) + return -EFAULT; + + rc = crypto_rng_get_bytes(crypto_default_rng, (u8 *)priv, nbytes); + crypto_put_default_rng(); + if (rc) + goto cleanup; + + rc = kmb_ecc_is_key_valid(curve, priv, nbytes); + if (rc) + goto cleanup; + + ecc_swap_digits(priv, privkey, curve->g.ndigits); + +cleanup: + memzero_explicit(&priv, sizeof(priv)); + + return rc; +} + +static int kmb_ocs_ecdh_set_secret(struct crypto_kpp *tfm, const void *buf, + unsigned int len) +{ + struct ocs_ecc_ctx *tctx = kpp_tfm_ctx(tfm); + struct ecdh params; + int rc = 0; + + rc = crypto_ecdh_decode_key(buf, len, ¶ms); + if (rc) + goto cleanup; + + /* Ensure key size is not bigger then expected. */ + if (params.key_size > digits_to_bytes(tctx->curve->g.ndigits)) { + rc = -EINVAL; + goto cleanup; + } + + /* Auto-generate private key is not provided. */ + if (!params.key || !params.key_size) { + rc = kmb_ecc_gen_privkey(tctx->curve, tctx->private_key); + goto cleanup; + } + + rc = kmb_ecc_is_key_valid(tctx->curve, (const u64 *)params.key, + params.key_size); + if (rc) + goto cleanup; + + ecc_swap_digits((const u64 *)params.key, tctx->private_key, + tctx->curve->g.ndigits); +cleanup: + memzero_explicit(¶ms, sizeof(params)); + + if (rc) + tctx->curve = NULL; + + return rc; +} + +/* Compute shared secret. */ +static int kmb_ecc_do_shared_secret(struct ocs_ecc_ctx *tctx, + struct kpp_request *req) +{ + struct ocs_ecc_dev *ecc_dev = tctx->ecc_dev; + const struct ecc_curve *curve = tctx->curve; + u64 shared_secret[KMB_ECC_VLI_MAX_DIGITS]; + u64 pubk_buf[KMB_ECC_VLI_MAX_DIGITS * 2]; + size_t copied, nbytes, pubk_len; + struct ecc_point *pk, *result; + int rc; + + nbytes = digits_to_bytes(curve->g.ndigits); + + /* Public key is a point, thus it has two coordinates */ + pubk_len = 2 * nbytes; + + /* Copy public key from SG list to pubk_buf. */ + copied = sg_copy_to_buffer(req->src, + sg_nents_for_len(req->src, pubk_len), + pubk_buf, pubk_len); + if (copied != pubk_len) + return -EINVAL; + + /* Allocate and initialize public key point. */ + pk = ecc_alloc_point(curve->g.ndigits); + if (!pk) + return -ENOMEM; + + ecc_swap_digits(pubk_buf, pk->x, curve->g.ndigits); + ecc_swap_digits(&pubk_buf[curve->g.ndigits], pk->y, curve->g.ndigits); + + /* + * Check the public key for following + * Check 1: Verify key is not the zero point. + * Check 2: Verify key is in the range [1, p-1]. + * Check 3: Verify that y^2 == (x^3 + a·x + b) mod p + */ + rc = kmb_ocs_ecc_is_pubkey_valid_partial(ecc_dev, curve, pk); + if (rc) + goto exit_free_pk; + + /* Allocate point for storing computed shared secret. */ + result = ecc_alloc_point(pk->ndigits); + if (!result) { + rc = -ENOMEM; + goto exit_free_pk; + } + + /* Calculate the shared secret.*/ + rc = kmb_ecc_point_mult(ecc_dev, result, pk, tctx->private_key, curve); + if (rc) + goto exit_free_result; + + if (ecc_point_is_zero(result)) { + rc = -EFAULT; + goto exit_free_result; + } + + /* Copy shared secret from point to buffer. */ + ecc_swap_digits(result->x, shared_secret, result->ndigits); + + /* Request might ask for less bytes than what we have. */ + nbytes = min_t(size_t, nbytes, req->dst_len); + + copied = sg_copy_from_buffer(req->dst, + sg_nents_for_len(req->dst, nbytes), + shared_secret, nbytes); + + if (copied != nbytes) + rc = -EINVAL; + + memzero_explicit(shared_secret, sizeof(shared_secret)); + +exit_free_result: + ecc_free_point(result); + +exit_free_pk: + ecc_free_point(pk); + + return rc; +} + +/* Compute public key. */ +static int kmb_ecc_do_public_key(struct ocs_ecc_ctx *tctx, + struct kpp_request *req) +{ + const struct ecc_curve *curve = tctx->curve; + u64 pubk_buf[KMB_ECC_VLI_MAX_DIGITS * 2]; + struct ecc_point *pk; + size_t pubk_len; + size_t copied; + int rc; + + /* Public key is a point, so it has double the digits. */ + pubk_len = 2 * digits_to_bytes(curve->g.ndigits); + + pk = ecc_alloc_point(curve->g.ndigits); + if (!pk) + return -ENOMEM; + + /* Public Key(pk) = priv * G. */ + rc = kmb_ecc_point_mult(tctx->ecc_dev, pk, &curve->g, tctx->private_key, + curve); + if (rc) + goto exit; + + /* SP800-56A rev 3 5.6.2.1.3 key check */ + if (kmb_ocs_ecc_is_pubkey_valid_full(tctx->ecc_dev, curve, pk)) { + rc = -EAGAIN; + goto exit; + } + + /* Copy public key from point to buffer. */ + ecc_swap_digits(pk->x, pubk_buf, pk->ndigits); + ecc_swap_digits(pk->y, &pubk_buf[pk->ndigits], pk->ndigits); + + /* Copy public key to req->dst. */ + copied = sg_copy_from_buffer(req->dst, + sg_nents_for_len(req->dst, pubk_len), + pubk_buf, pubk_len); + + if (copied != pubk_len) + rc = -EINVAL; + +exit: + ecc_free_point(pk); + + return rc; +} + +static int kmb_ocs_ecc_do_one_request(struct crypto_engine *engine, + void *areq) +{ + struct kpp_request *req = container_of(areq, struct kpp_request, base); + struct ocs_ecc_ctx *tctx = kmb_ocs_ecc_tctx(req); + struct ocs_ecc_dev *ecc_dev = tctx->ecc_dev; + int rc; + + if (req->src) + rc = kmb_ecc_do_shared_secret(tctx, req); + else + rc = kmb_ecc_do_public_key(tctx, req); + + crypto_finalize_kpp_request(ecc_dev->engine, req, rc); + + return 0; +} + +static int kmb_ocs_ecdh_generate_public_key(struct kpp_request *req) +{ + struct ocs_ecc_ctx *tctx = kmb_ocs_ecc_tctx(req); + const struct ecc_curve *curve = tctx->curve; + + /* Ensure kmb_ocs_ecdh_set_secret() has been successfully called. */ + if (!tctx->curve) + return -EINVAL; + + /* Ensure dst is present. */ + if (!req->dst) + return -EINVAL; + + /* Check the request dst is big enough to hold the public key. */ + if (req->dst_len < (2 * digits_to_bytes(curve->g.ndigits))) + return -EINVAL; + + /* 'src' is not supposed to be present when generate pubk is called. */ + if (req->src) + return -EINVAL; + + return crypto_transfer_kpp_request_to_engine(tctx->ecc_dev->engine, + req); +} + +static int kmb_ocs_ecdh_compute_shared_secret(struct kpp_request *req) +{ + struct ocs_ecc_ctx *tctx = kmb_ocs_ecc_tctx(req); + const struct ecc_curve *curve = tctx->curve; + + /* Ensure kmb_ocs_ecdh_set_secret() has been successfully called. */ + if (!tctx->curve) + return -EINVAL; + + /* Ensure dst is present. */ + if (!req->dst) + return -EINVAL; + + /* Ensure src is present. */ + if (!req->src) + return -EINVAL; + + /* + * req->src is expected to the (other-side) public key, so its length + * must be 2 * coordinate size (in bytes). + */ + if (req->src_len != 2 * digits_to_bytes(curve->g.ndigits)) + return -EINVAL; + + return crypto_transfer_kpp_request_to_engine(tctx->ecc_dev->engine, + req); +} + +static int kmb_ecc_tctx_init(struct ocs_ecc_ctx *tctx, unsigned int curve_id) +{ + memset(tctx, 0, sizeof(*tctx)); + + tctx->ecc_dev = kmb_ocs_ecc_find_dev(tctx); + + if (IS_ERR(tctx->ecc_dev)) { + pr_err("Failed to find the device : %ld\n", + PTR_ERR(tctx->ecc_dev)); + return PTR_ERR(tctx->ecc_dev); + } + + tctx->curve = ecc_get_curve(curve_id); + if (!tctx->curve) + return -EOPNOTSUPP; + + return 0; +} + +static int kmb_ocs_ecdh_nist_p256_init_tfm(struct crypto_kpp *tfm) +{ + struct ocs_ecc_ctx *tctx = kpp_tfm_ctx(tfm); + + return kmb_ecc_tctx_init(tctx, ECC_CURVE_NIST_P256); +} + +static int kmb_ocs_ecdh_nist_p384_init_tfm(struct crypto_kpp *tfm) +{ + struct ocs_ecc_ctx *tctx = kpp_tfm_ctx(tfm); + + return kmb_ecc_tctx_init(tctx, ECC_CURVE_NIST_P384); +} + +static void kmb_ocs_ecdh_exit_tfm(struct crypto_kpp *tfm) +{ + struct ocs_ecc_ctx *tctx = kpp_tfm_ctx(tfm); + + memzero_explicit(tctx->private_key, sizeof(*tctx->private_key)); +} + +static unsigned int kmb_ocs_ecdh_max_size(struct crypto_kpp *tfm) +{ + struct ocs_ecc_ctx *tctx = kpp_tfm_ctx(tfm); + + /* Public key is made of two coordinates, so double the digits. */ + return digits_to_bytes(tctx->curve->g.ndigits) * 2; +} + +static struct kpp_engine_alg ocs_ecdh_p256 = { + .base.set_secret = kmb_ocs_ecdh_set_secret, + .base.generate_public_key = kmb_ocs_ecdh_generate_public_key, + .base.compute_shared_secret = kmb_ocs_ecdh_compute_shared_secret, + .base.init = kmb_ocs_ecdh_nist_p256_init_tfm, + .base.exit = kmb_ocs_ecdh_exit_tfm, + .base.max_size = kmb_ocs_ecdh_max_size, + .base.base = { + .cra_name = "ecdh-nist-p256", + .cra_driver_name = "ecdh-nist-p256-keembay-ocs", + .cra_priority = KMB_OCS_ECC_PRIORITY, + .cra_module = THIS_MODULE, + .cra_ctxsize = sizeof(struct ocs_ecc_ctx), + }, + .op.do_one_request = kmb_ocs_ecc_do_one_request, +}; + +static struct kpp_engine_alg ocs_ecdh_p384 = { + .base.set_secret = kmb_ocs_ecdh_set_secret, + .base.generate_public_key = kmb_ocs_ecdh_generate_public_key, + .base.compute_shared_secret = kmb_ocs_ecdh_compute_shared_secret, + .base.init = kmb_ocs_ecdh_nist_p384_init_tfm, + .base.exit = kmb_ocs_ecdh_exit_tfm, + .base.max_size = kmb_ocs_ecdh_max_size, + .base.base = { + .cra_name = "ecdh-nist-p384", + .cra_driver_name = "ecdh-nist-p384-keembay-ocs", + .cra_priority = KMB_OCS_ECC_PRIORITY, + .cra_module = THIS_MODULE, + .cra_ctxsize = sizeof(struct ocs_ecc_ctx), + }, + .op.do_one_request = kmb_ocs_ecc_do_one_request, +}; + +static irqreturn_t ocs_ecc_irq_handler(int irq, void *dev_id) +{ + struct ocs_ecc_dev *ecc_dev = dev_id; + u32 status; + + /* + * Read the status register and write it back to clear the + * DONE_INT_STATUS bit. + */ + status = ioread32(ecc_dev->base_reg + HW_OFFS_OCS_ECC_ISR); + iowrite32(status, ecc_dev->base_reg + HW_OFFS_OCS_ECC_ISR); + + if (!(status & HW_OCS_ECC_ISR_INT_STATUS_DONE)) + return IRQ_NONE; + + complete(&ecc_dev->irq_done); + + return IRQ_HANDLED; +} + +static int kmb_ocs_ecc_probe(struct platform_device *pdev) +{ + struct device *dev = &pdev->dev; + struct ocs_ecc_dev *ecc_dev; + int rc; + + ecc_dev = devm_kzalloc(dev, sizeof(*ecc_dev), GFP_KERNEL); + if (!ecc_dev) + return -ENOMEM; + + ecc_dev->dev = dev; + + platform_set_drvdata(pdev, ecc_dev); + + INIT_LIST_HEAD(&ecc_dev->list); + init_completion(&ecc_dev->irq_done); + + /* Get base register address. */ + ecc_dev->base_reg = devm_platform_ioremap_resource(pdev, 0); + if (IS_ERR(ecc_dev->base_reg)) { + dev_err(dev, "Failed to get base address\n"); + rc = PTR_ERR(ecc_dev->base_reg); + goto list_del; + } + + /* Get and request IRQ */ + ecc_dev->irq = platform_get_irq(pdev, 0); + if (ecc_dev->irq < 0) { + rc = ecc_dev->irq; + goto list_del; + } + + rc = devm_request_threaded_irq(dev, ecc_dev->irq, ocs_ecc_irq_handler, + NULL, 0, "keembay-ocs-ecc", ecc_dev); + if (rc < 0) { + dev_err(dev, "Could not request IRQ\n"); + goto list_del; + } + + /* Add device to the list of OCS ECC devices. */ + spin_lock(&ocs_ecc.lock); + list_add_tail(&ecc_dev->list, &ocs_ecc.dev_list); + spin_unlock(&ocs_ecc.lock); + + /* Initialize crypto engine. */ + ecc_dev->engine = crypto_engine_alloc_init(dev, 1); + if (!ecc_dev->engine) { + dev_err(dev, "Could not allocate crypto engine\n"); + rc = -ENOMEM; + goto list_del; + } + + rc = crypto_engine_start(ecc_dev->engine); + if (rc) { + dev_err(dev, "Could not start crypto engine\n"); + goto cleanup; + } + + /* Register the KPP algo. */ + rc = crypto_engine_register_kpp(&ocs_ecdh_p256); + if (rc) { + dev_err(dev, + "Could not register OCS algorithms with Crypto API\n"); + goto cleanup; + } + + rc = crypto_engine_register_kpp(&ocs_ecdh_p384); + if (rc) { + dev_err(dev, + "Could not register OCS algorithms with Crypto API\n"); + goto ocs_ecdh_p384_error; + } + + return 0; + +ocs_ecdh_p384_error: + crypto_engine_unregister_kpp(&ocs_ecdh_p256); + +cleanup: + crypto_engine_exit(ecc_dev->engine); + +list_del: + spin_lock(&ocs_ecc.lock); + list_del(&ecc_dev->list); + spin_unlock(&ocs_ecc.lock); + + return rc; +} + +static void kmb_ocs_ecc_remove(struct platform_device *pdev) +{ + struct ocs_ecc_dev *ecc_dev; + + ecc_dev = platform_get_drvdata(pdev); + + crypto_engine_unregister_kpp(&ocs_ecdh_p384); + crypto_engine_unregister_kpp(&ocs_ecdh_p256); + + spin_lock(&ocs_ecc.lock); + list_del(&ecc_dev->list); + spin_unlock(&ocs_ecc.lock); + + crypto_engine_exit(ecc_dev->engine); +} + +/* Device tree driver match. */ +static const struct of_device_id kmb_ocs_ecc_of_match[] = { + { + .compatible = "intel,keembay-ocs-ecc", + }, + {} +}; + +/* The OCS driver is a platform device. */ +static struct platform_driver kmb_ocs_ecc_driver = { + .probe = kmb_ocs_ecc_probe, + .remove = kmb_ocs_ecc_remove, + .driver = { + .name = DRV_NAME, + .of_match_table = kmb_ocs_ecc_of_match, + }, +}; +module_platform_driver(kmb_ocs_ecc_driver); + +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("Intel Keem Bay OCS ECC Driver"); +MODULE_ALIAS_CRYPTO("ecdh-nist-p256"); +MODULE_ALIAS_CRYPTO("ecdh-nist-p384"); +MODULE_ALIAS_CRYPTO("ecdh-nist-p256-keembay-ocs"); +MODULE_ALIAS_CRYPTO("ecdh-nist-p384-keembay-ocs"); diff --git a/drivers/crypto/intel/keembay/keembay-ocs-hcu-core.c b/drivers/crypto/intel/keembay/keembay-ocs-hcu-core.c new file mode 100644 index 000000000000..48281d882260 --- /dev/null +++ b/drivers/crypto/intel/keembay/keembay-ocs-hcu-core.c @@ -0,0 +1,1264 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Intel Keem Bay OCS HCU Crypto Driver. + * + * Copyright (C) 2018-2020 Intel Corporation + */ + +#include <crypto/engine.h> +#include <crypto/hmac.h> +#include <crypto/internal/hash.h> +#include <crypto/scatterwalk.h> +#include <crypto/sha2.h> +#include <crypto/sm3.h> +#include <linux/completion.h> +#include <linux/dma-mapping.h> +#include <linux/err.h> +#include <linux/interrupt.h> +#include <linux/kernel.h> +#include <linux/mod_devicetable.h> +#include <linux/module.h> +#include <linux/platform_device.h> +#include <linux/string.h> + +#include "ocs-hcu.h" + +#define DRV_NAME "keembay-ocs-hcu" + +/* Flag marking a final request. */ +#define REQ_FINAL BIT(0) +/* Flag marking a HMAC request. */ +#define REQ_FLAGS_HMAC BIT(1) +/* Flag set when HW HMAC is being used. */ +#define REQ_FLAGS_HMAC_HW BIT(2) +/* Flag set when SW HMAC is being used. */ +#define REQ_FLAGS_HMAC_SW BIT(3) + +/** + * struct ocs_hcu_ctx: OCS HCU Transform context. + * @hcu_dev: The OCS HCU device used by the transformation. + * @key: The key (used only for HMAC transformations). + * @key_len: The length of the key. + * @is_sm3_tfm: Whether or not this is an SM3 transformation. + * @is_hmac_tfm: Whether or not this is a HMAC transformation. + */ +struct ocs_hcu_ctx { + struct ocs_hcu_dev *hcu_dev; + u8 key[SHA512_BLOCK_SIZE]; + size_t key_len; + bool is_sm3_tfm; + bool is_hmac_tfm; +}; + +/** + * struct ocs_hcu_rctx - Context for the request. + * @hcu_dev: OCS HCU device to be used to service the request. + * @flags: Flags tracking request status. + * @algo: Algorithm to use for the request. + * @blk_sz: Block size of the transformation / request. + * @dig_sz: Digest size of the transformation / request. + * @dma_list: OCS DMA linked list. + * @hash_ctx: OCS HCU hashing context. + * @buffer: Buffer to store: partial block of data and SW HMAC + * artifacts (ipad, opad, etc.). + * @buf_cnt: Number of bytes currently stored in the buffer. + * @buf_dma_addr: The DMA address of @buffer (when mapped). + * @buf_dma_count: The number of bytes in @buffer currently DMA-mapped. + * @sg: Head of the scatterlist entries containing data. + * @sg_data_total: Total data in the SG list at any time. + * @sg_data_offset: Offset into the data of the current individual SG node. + * @sg_dma_nents: Number of sg entries mapped in dma_list. + * @nents: Number of entries in the scatterlist. + */ +struct ocs_hcu_rctx { + struct ocs_hcu_dev *hcu_dev; + u32 flags; + enum ocs_hcu_algo algo; + size_t blk_sz; + size_t dig_sz; + struct ocs_hcu_dma_list *dma_list; + struct ocs_hcu_hash_ctx hash_ctx; + /* + * Buffer is double the block size because we need space for SW HMAC + * artifacts, i.e: + * - ipad (1 block) + a possible partial block of data. + * - opad (1 block) + digest of H(k ^ ipad || m) + */ + u8 buffer[2 * SHA512_BLOCK_SIZE]; + size_t buf_cnt; + dma_addr_t buf_dma_addr; + size_t buf_dma_count; + struct scatterlist *sg; + unsigned int sg_data_total; + unsigned int sg_data_offset; + unsigned int sg_dma_nents; + unsigned int nents; +}; + +/** + * struct ocs_hcu_drv - Driver data + * @dev_list: The list of HCU devices. + * @lock: The lock protecting dev_list. + */ +struct ocs_hcu_drv { + struct list_head dev_list; + spinlock_t lock; /* Protects dev_list. */ +}; + +static struct ocs_hcu_drv ocs_hcu = { + .dev_list = LIST_HEAD_INIT(ocs_hcu.dev_list), + .lock = __SPIN_LOCK_UNLOCKED(ocs_hcu.lock), +}; + +/* + * Return the total amount of data in the request; that is: the data in the + * request buffer + the data in the sg list. + */ +static inline unsigned int kmb_get_total_data(struct ocs_hcu_rctx *rctx) +{ + return rctx->sg_data_total + rctx->buf_cnt; +} + +/* Move remaining content of scatter-gather list to context buffer. */ +static int flush_sg_to_ocs_buffer(struct ocs_hcu_rctx *rctx) +{ + size_t count; + + if (rctx->sg_data_total > (sizeof(rctx->buffer) - rctx->buf_cnt)) { + WARN(1, "%s: sg data does not fit in buffer\n", __func__); + return -EINVAL; + } + + while (rctx->sg_data_total) { + if (!rctx->sg) { + WARN(1, "%s: unexpected NULL sg\n", __func__); + return -EINVAL; + } + /* + * If current sg has been fully processed, skip to the next + * one. + */ + if (rctx->sg_data_offset == rctx->sg->length) { + rctx->sg = sg_next(rctx->sg); + rctx->sg_data_offset = 0; + continue; + } + /* + * Determine the maximum data available to copy from the node. + * Minimum of the length left in the sg node, or the total data + * in the request. + */ + count = min(rctx->sg->length - rctx->sg_data_offset, + rctx->sg_data_total); + /* Copy from scatter-list entry to context buffer. */ + scatterwalk_map_and_copy(&rctx->buffer[rctx->buf_cnt], + rctx->sg, rctx->sg_data_offset, + count, 0); + + rctx->sg_data_offset += count; + rctx->sg_data_total -= count; + rctx->buf_cnt += count; + } + + return 0; +} + +static struct ocs_hcu_dev *kmb_ocs_hcu_find_dev(struct ahash_request *req) +{ + struct crypto_ahash *tfm = crypto_ahash_reqtfm(req); + struct ocs_hcu_ctx *tctx = crypto_ahash_ctx(tfm); + + /* If the HCU device for the request was previously set, return it. */ + if (tctx->hcu_dev) + return tctx->hcu_dev; + + /* + * Otherwise, get the first HCU device available (there should be one + * and only one device). + */ + spin_lock_bh(&ocs_hcu.lock); + tctx->hcu_dev = list_first_entry_or_null(&ocs_hcu.dev_list, + struct ocs_hcu_dev, + list); + spin_unlock_bh(&ocs_hcu.lock); + + return tctx->hcu_dev; +} + +/* Free OCS DMA linked list and DMA-able context buffer. */ +static void kmb_ocs_hcu_dma_cleanup(struct ahash_request *req, + struct ocs_hcu_rctx *rctx) +{ + struct ocs_hcu_dev *hcu_dev = rctx->hcu_dev; + struct device *dev = hcu_dev->dev; + + /* Unmap rctx->buffer (if mapped). */ + if (rctx->buf_dma_count) { + dma_unmap_single(dev, rctx->buf_dma_addr, rctx->buf_dma_count, + DMA_TO_DEVICE); + rctx->buf_dma_count = 0; + } + + /* Unmap req->src (if mapped). */ + if (rctx->sg_dma_nents) { + dma_unmap_sg(dev, req->src, rctx->nents, DMA_TO_DEVICE); + rctx->sg_dma_nents = 0; + } + + /* Free dma_list (if allocated). */ + if (rctx->dma_list) { + ocs_hcu_dma_list_free(hcu_dev, rctx->dma_list); + rctx->dma_list = NULL; + } +} + +/* + * Prepare for DMA operation: + * - DMA-map request context buffer (if needed) + * - DMA-map SG list (only the entries to be processed, see note below) + * - Allocate OCS HCU DMA linked list (number of elements = SG entries to + * process + context buffer (if not empty)). + * - Add DMA-mapped request context buffer to OCS HCU DMA list. + * - Add SG entries to DMA list. + * + * Note: if this is a final request, we process all the data in the SG list, + * otherwise we can only process up to the maximum amount of block-aligned data + * (the remainder will be put into the context buffer and processed in the next + * request). + */ +static int kmb_ocs_dma_prepare(struct ahash_request *req) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + struct device *dev = rctx->hcu_dev->dev; + unsigned int remainder = 0; + unsigned int total; + int nents; + size_t count; + int rc; + int i; + + /* This function should be called only when there is data to process. */ + total = kmb_get_total_data(rctx); + if (!total) + return -EINVAL; + + /* + * If this is not a final DMA (terminated DMA), the data passed to the + * HCU must be aligned to the block size; compute the remainder data to + * be processed in the next request. + */ + if (!(rctx->flags & REQ_FINAL)) + remainder = total % rctx->blk_sz; + + /* Determine the number of scatter gather list entries to process. */ + nents = sg_nents_for_len(req->src, rctx->sg_data_total - remainder); + + if (nents < 0) + return nents; + + /* If there are entries to process, map them. */ + if (nents) { + rctx->sg_dma_nents = dma_map_sg(dev, req->src, nents, + DMA_TO_DEVICE); + if (!rctx->sg_dma_nents) { + dev_err(dev, "Failed to MAP SG\n"); + rc = -ENOMEM; + goto cleanup; + } + + /* Save the value of nents to pass to dma_unmap_sg. */ + rctx->nents = nents; + + /* + * The value returned by dma_map_sg() can be < nents; so update + * nents accordingly. + */ + nents = rctx->sg_dma_nents; + } + + /* + * If context buffer is not empty, map it and add extra DMA entry for + * it. + */ + if (rctx->buf_cnt) { + rctx->buf_dma_addr = dma_map_single(dev, rctx->buffer, + rctx->buf_cnt, + DMA_TO_DEVICE); + if (dma_mapping_error(dev, rctx->buf_dma_addr)) { + dev_err(dev, "Failed to map request context buffer\n"); + rc = -ENOMEM; + goto cleanup; + } + rctx->buf_dma_count = rctx->buf_cnt; + /* Increase number of dma entries. */ + nents++; + } + + /* Allocate OCS HCU DMA list. */ + rctx->dma_list = ocs_hcu_dma_list_alloc(rctx->hcu_dev, nents); + if (!rctx->dma_list) { + rc = -ENOMEM; + goto cleanup; + } + + /* Add request context buffer (if previously DMA-mapped) */ + if (rctx->buf_dma_count) { + rc = ocs_hcu_dma_list_add_tail(rctx->hcu_dev, rctx->dma_list, + rctx->buf_dma_addr, + rctx->buf_dma_count); + if (rc) + goto cleanup; + } + + /* Add the SG nodes to be processed to the DMA linked list. */ + for_each_sg(req->src, rctx->sg, rctx->sg_dma_nents, i) { + /* + * The number of bytes to add to the list entry is the minimum + * between: + * - The DMA length of the SG entry. + * - The data left to be processed. + */ + count = min(rctx->sg_data_total - remainder, + sg_dma_len(rctx->sg) - rctx->sg_data_offset); + /* + * Do not create a zero length DMA descriptor. Check in case of + * zero length SG node. + */ + if (count == 0) + continue; + /* Add sg to HCU DMA list. */ + rc = ocs_hcu_dma_list_add_tail(rctx->hcu_dev, + rctx->dma_list, + rctx->sg->dma_address, + count); + if (rc) + goto cleanup; + + /* Update amount of data remaining in SG list. */ + rctx->sg_data_total -= count; + + /* + * If remaining data is equal to remainder (note: 'less than' + * case should never happen in practice), we are done: update + * offset and exit the loop. + */ + if (rctx->sg_data_total <= remainder) { + WARN_ON(rctx->sg_data_total < remainder); + rctx->sg_data_offset += count; + break; + } + + /* + * If we get here is because we need to process the next sg in + * the list; set offset within the sg to 0. + */ + rctx->sg_data_offset = 0; + } + + return 0; +cleanup: + dev_err(dev, "Failed to prepare DMA.\n"); + kmb_ocs_hcu_dma_cleanup(req, rctx); + + return rc; +} + +static void kmb_ocs_hcu_secure_cleanup(struct ahash_request *req) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + + /* Clear buffer of any data. */ + memzero_explicit(rctx->buffer, sizeof(rctx->buffer)); +} + +static int kmb_ocs_hcu_handle_queue(struct ahash_request *req) +{ + struct ocs_hcu_dev *hcu_dev = kmb_ocs_hcu_find_dev(req); + + if (!hcu_dev) + return -ENOENT; + + return crypto_transfer_hash_request_to_engine(hcu_dev->engine, req); +} + +static int prepare_ipad(struct ahash_request *req) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + struct crypto_ahash *tfm = crypto_ahash_reqtfm(req); + struct ocs_hcu_ctx *ctx = crypto_ahash_ctx(tfm); + int i; + + WARN(rctx->buf_cnt, "%s: Context buffer is not empty\n", __func__); + WARN(!(rctx->flags & REQ_FLAGS_HMAC_SW), + "%s: HMAC_SW flag is not set\n", __func__); + /* + * Key length must be equal to block size. If key is shorter, + * we pad it with zero (note: key cannot be longer, since + * longer keys are hashed by kmb_ocs_hcu_setkey()). + */ + if (ctx->key_len > rctx->blk_sz) { + WARN(1, "%s: Invalid key length in tfm context\n", __func__); + return -EINVAL; + } + memzero_explicit(&ctx->key[ctx->key_len], + rctx->blk_sz - ctx->key_len); + ctx->key_len = rctx->blk_sz; + /* + * Prepare IPAD for HMAC. Only done for first block. + * HMAC(k,m) = H(k ^ opad || H(k ^ ipad || m)) + * k ^ ipad will be first hashed block. + * k ^ opad will be calculated in the final request. + * Only needed if not using HW HMAC. + */ + for (i = 0; i < rctx->blk_sz; i++) + rctx->buffer[i] = ctx->key[i] ^ HMAC_IPAD_VALUE; + rctx->buf_cnt = rctx->blk_sz; + + return 0; +} + +static int kmb_ocs_hcu_do_one_request(struct crypto_engine *engine, void *areq) +{ + struct ahash_request *req = container_of(areq, struct ahash_request, + base); + struct ocs_hcu_dev *hcu_dev = kmb_ocs_hcu_find_dev(req); + struct crypto_ahash *tfm = crypto_ahash_reqtfm(req); + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + struct ocs_hcu_ctx *tctx = crypto_ahash_ctx(tfm); + int rc; + int i; + + if (!hcu_dev) { + rc = -ENOENT; + goto error; + } + + /* + * If hardware HMAC flag is set, perform HMAC in hardware. + * + * NOTE: this flag implies REQ_FINAL && kmb_get_total_data(rctx) + */ + if (rctx->flags & REQ_FLAGS_HMAC_HW) { + /* Map input data into the HCU DMA linked list. */ + rc = kmb_ocs_dma_prepare(req); + if (rc) + goto error; + + rc = ocs_hcu_hmac(hcu_dev, rctx->algo, tctx->key, tctx->key_len, + rctx->dma_list, req->result, rctx->dig_sz); + + /* Unmap data and free DMA list regardless of return code. */ + kmb_ocs_hcu_dma_cleanup(req, rctx); + + /* Process previous return code. */ + if (rc) + goto error; + + goto done; + } + + /* Handle update request case. */ + if (!(rctx->flags & REQ_FINAL)) { + /* Update should always have input data. */ + if (!kmb_get_total_data(rctx)) + return -EINVAL; + + /* Map input data into the HCU DMA linked list. */ + rc = kmb_ocs_dma_prepare(req); + if (rc) + goto error; + + /* Do hashing step. */ + rc = ocs_hcu_hash_update(hcu_dev, &rctx->hash_ctx, + rctx->dma_list); + + /* Unmap data and free DMA list regardless of return code. */ + kmb_ocs_hcu_dma_cleanup(req, rctx); + + /* Process previous return code. */ + if (rc) + goto error; + + /* + * Reset request buffer count (data in the buffer was just + * processed). + */ + rctx->buf_cnt = 0; + /* + * Move remaining sg data into the request buffer, so that it + * will be processed during the next request. + * + * NOTE: we have remaining data if kmb_get_total_data() was not + * a multiple of block size. + */ + rc = flush_sg_to_ocs_buffer(rctx); + if (rc) + goto error; + + goto done; + } + + /* If we get here, this is a final request. */ + + /* If there is data to process, use finup. */ + if (kmb_get_total_data(rctx)) { + /* Map input data into the HCU DMA linked list. */ + rc = kmb_ocs_dma_prepare(req); + if (rc) + goto error; + + /* Do hashing step. */ + rc = ocs_hcu_hash_finup(hcu_dev, &rctx->hash_ctx, + rctx->dma_list, + req->result, rctx->dig_sz); + /* Free DMA list regardless of return code. */ + kmb_ocs_hcu_dma_cleanup(req, rctx); + + /* Process previous return code. */ + if (rc) + goto error; + + } else { /* Otherwise (if we have no data), use final. */ + rc = ocs_hcu_hash_final(hcu_dev, &rctx->hash_ctx, req->result, + rctx->dig_sz); + if (rc) + goto error; + } + + /* + * If we are finalizing a SW HMAC request, we just computed the result + * of: H(k ^ ipad || m). + * + * We now need to complete the HMAC calculation with the OPAD step, + * that is, we need to compute H(k ^ opad || digest), where digest is + * the digest we just obtained, i.e., H(k ^ ipad || m). + */ + if (rctx->flags & REQ_FLAGS_HMAC_SW) { + /* + * Compute k ^ opad and store it in the request buffer (which + * is not used anymore at this point). + * Note: key has been padded / hashed already (so keylen == + * blksz) . + */ + WARN_ON(tctx->key_len != rctx->blk_sz); + for (i = 0; i < rctx->blk_sz; i++) + rctx->buffer[i] = tctx->key[i] ^ HMAC_OPAD_VALUE; + /* Now append the digest to the rest of the buffer. */ + for (i = 0; (i < rctx->dig_sz); i++) + rctx->buffer[rctx->blk_sz + i] = req->result[i]; + + /* Now hash the buffer to obtain the final HMAC. */ + rc = ocs_hcu_digest(hcu_dev, rctx->algo, rctx->buffer, + rctx->blk_sz + rctx->dig_sz, req->result, + rctx->dig_sz); + if (rc) + goto error; + } + + /* Perform secure clean-up. */ + kmb_ocs_hcu_secure_cleanup(req); +done: + crypto_finalize_hash_request(hcu_dev->engine, req, 0); + + return 0; + +error: + kmb_ocs_hcu_secure_cleanup(req); + return rc; +} + +static int kmb_ocs_hcu_init(struct ahash_request *req) +{ + struct ocs_hcu_dev *hcu_dev = kmb_ocs_hcu_find_dev(req); + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + struct crypto_ahash *tfm = crypto_ahash_reqtfm(req); + struct ocs_hcu_ctx *ctx = crypto_ahash_ctx(tfm); + + if (!hcu_dev) + return -ENOENT; + + /* Initialize entire request context to zero. */ + memset(rctx, 0, sizeof(*rctx)); + + rctx->hcu_dev = hcu_dev; + rctx->dig_sz = crypto_ahash_digestsize(tfm); + + switch (rctx->dig_sz) { +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_HCU_HMAC_SHA224 + case SHA224_DIGEST_SIZE: + rctx->blk_sz = SHA224_BLOCK_SIZE; + rctx->algo = OCS_HCU_ALGO_SHA224; + break; +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_HCU_HMAC_SHA224 */ + case SHA256_DIGEST_SIZE: + rctx->blk_sz = SHA256_BLOCK_SIZE; + /* + * SHA256 and SM3 have the same digest size: use info from tfm + * context to find out which one we should use. + */ + rctx->algo = ctx->is_sm3_tfm ? OCS_HCU_ALGO_SM3 : + OCS_HCU_ALGO_SHA256; + break; + case SHA384_DIGEST_SIZE: + rctx->blk_sz = SHA384_BLOCK_SIZE; + rctx->algo = OCS_HCU_ALGO_SHA384; + break; + case SHA512_DIGEST_SIZE: + rctx->blk_sz = SHA512_BLOCK_SIZE; + rctx->algo = OCS_HCU_ALGO_SHA512; + break; + default: + return -EINVAL; + } + + /* Initialize intermediate data. */ + ocs_hcu_hash_init(&rctx->hash_ctx, rctx->algo); + + /* If this a HMAC request, set HMAC flag. */ + if (ctx->is_hmac_tfm) + rctx->flags |= REQ_FLAGS_HMAC; + + return 0; +} + +static int kmb_ocs_hcu_update(struct ahash_request *req) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + int rc; + + if (!req->nbytes) + return 0; + + rctx->sg_data_total = req->nbytes; + rctx->sg_data_offset = 0; + rctx->sg = req->src; + + /* + * If we are doing HMAC, then we must use SW-assisted HMAC, since HW + * HMAC does not support context switching (there it can only be used + * with finup() or digest()). + */ + if (rctx->flags & REQ_FLAGS_HMAC && + !(rctx->flags & REQ_FLAGS_HMAC_SW)) { + rctx->flags |= REQ_FLAGS_HMAC_SW; + rc = prepare_ipad(req); + if (rc) + return rc; + } + + /* + * If remaining sg_data fits into ctx buffer, just copy it there; we'll + * process it at the next update() or final(). + */ + if (rctx->sg_data_total <= (sizeof(rctx->buffer) - rctx->buf_cnt)) + return flush_sg_to_ocs_buffer(rctx); + + return kmb_ocs_hcu_handle_queue(req); +} + +/* Common logic for kmb_ocs_hcu_final() and kmb_ocs_hcu_finup(). */ +static int kmb_ocs_hcu_fin_common(struct ahash_request *req) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + struct crypto_ahash *tfm = crypto_ahash_reqtfm(req); + struct ocs_hcu_ctx *ctx = crypto_ahash_ctx(tfm); + int rc; + + rctx->flags |= REQ_FINAL; + + /* + * If this is a HMAC request and, so far, we didn't have to switch to + * SW HMAC, check if we can use HW HMAC. + */ + if (rctx->flags & REQ_FLAGS_HMAC && + !(rctx->flags & REQ_FLAGS_HMAC_SW)) { + /* + * If we are here, it means we never processed any data so far, + * so we can use HW HMAC, but only if there is some data to + * process (since OCS HW MAC does not support zero-length + * messages) and the key length is supported by the hardware + * (OCS HCU HW only supports length <= 64); if HW HMAC cannot + * be used, fall back to SW-assisted HMAC. + */ + if (kmb_get_total_data(rctx) && + ctx->key_len <= OCS_HCU_HW_KEY_LEN) { + rctx->flags |= REQ_FLAGS_HMAC_HW; + } else { + rctx->flags |= REQ_FLAGS_HMAC_SW; + rc = prepare_ipad(req); + if (rc) + return rc; + } + } + + return kmb_ocs_hcu_handle_queue(req); +} + +static int kmb_ocs_hcu_final(struct ahash_request *req) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + + rctx->sg_data_total = 0; + rctx->sg_data_offset = 0; + rctx->sg = NULL; + + return kmb_ocs_hcu_fin_common(req); +} + +static int kmb_ocs_hcu_finup(struct ahash_request *req) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + + rctx->sg_data_total = req->nbytes; + rctx->sg_data_offset = 0; + rctx->sg = req->src; + + return kmb_ocs_hcu_fin_common(req); +} + +static int kmb_ocs_hcu_digest(struct ahash_request *req) +{ + int rc = 0; + struct ocs_hcu_dev *hcu_dev = kmb_ocs_hcu_find_dev(req); + + if (!hcu_dev) + return -ENOENT; + + rc = kmb_ocs_hcu_init(req); + if (rc) + return rc; + + rc = kmb_ocs_hcu_finup(req); + + return rc; +} + +static int kmb_ocs_hcu_export(struct ahash_request *req, void *out) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + + /* Intermediate data is always stored and applied per request. */ + memcpy(out, rctx, sizeof(*rctx)); + + return 0; +} + +static int kmb_ocs_hcu_import(struct ahash_request *req, const void *in) +{ + struct ocs_hcu_rctx *rctx = ahash_request_ctx_dma(req); + + /* Intermediate data is always stored and applied per request. */ + memcpy(rctx, in, sizeof(*rctx)); + + return 0; +} + +static int kmb_ocs_hcu_setkey(struct crypto_ahash *tfm, const u8 *key, + unsigned int keylen) +{ + unsigned int digestsize = crypto_ahash_digestsize(tfm); + struct ocs_hcu_ctx *ctx = crypto_ahash_ctx(tfm); + size_t blk_sz = crypto_ahash_blocksize(tfm); + struct crypto_ahash *ahash_tfm; + struct ahash_request *req; + struct crypto_wait wait; + struct scatterlist sg; + const char *alg_name; + int rc; + + /* + * Key length must be equal to block size: + * - If key is shorter, we are done for now (the key will be padded + * later on); this is to maximize the use of HW HMAC (which works + * only for keys <= 64 bytes). + * - If key is longer, we hash it. + */ + if (keylen <= blk_sz) { + memcpy(ctx->key, key, keylen); + ctx->key_len = keylen; + return 0; + } + + switch (digestsize) { +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_HCU_HMAC_SHA224 + case SHA224_DIGEST_SIZE: + alg_name = "sha224-keembay-ocs"; + break; +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_HCU_HMAC_SHA224 */ + case SHA256_DIGEST_SIZE: + alg_name = ctx->is_sm3_tfm ? "sm3-keembay-ocs" : + "sha256-keembay-ocs"; + break; + case SHA384_DIGEST_SIZE: + alg_name = "sha384-keembay-ocs"; + break; + case SHA512_DIGEST_SIZE: + alg_name = "sha512-keembay-ocs"; + break; + default: + return -EINVAL; + } + + ahash_tfm = crypto_alloc_ahash(alg_name, 0, 0); + if (IS_ERR(ahash_tfm)) + return PTR_ERR(ahash_tfm); + + req = ahash_request_alloc(ahash_tfm, GFP_KERNEL); + if (!req) { + rc = -ENOMEM; + goto err_free_ahash; + } + + crypto_init_wait(&wait); + ahash_request_set_callback(req, CRYPTO_TFM_REQ_MAY_BACKLOG, + crypto_req_done, &wait); + crypto_ahash_clear_flags(ahash_tfm, ~0); + + sg_init_one(&sg, key, keylen); + ahash_request_set_crypt(req, &sg, ctx->key, keylen); + + rc = crypto_wait_req(crypto_ahash_digest(req), &wait); + if (rc == 0) + ctx->key_len = digestsize; + + ahash_request_free(req); +err_free_ahash: + crypto_free_ahash(ahash_tfm); + + return rc; +} + +/* Set request size and initialize tfm context. */ +static void __cra_init(struct crypto_tfm *tfm, struct ocs_hcu_ctx *ctx) +{ + crypto_ahash_set_reqsize_dma(__crypto_ahash_cast(tfm), + sizeof(struct ocs_hcu_rctx)); +} + +static int kmb_ocs_hcu_sha_cra_init(struct crypto_tfm *tfm) +{ + struct ocs_hcu_ctx *ctx = crypto_tfm_ctx(tfm); + + __cra_init(tfm, ctx); + + return 0; +} + +static int kmb_ocs_hcu_sm3_cra_init(struct crypto_tfm *tfm) +{ + struct ocs_hcu_ctx *ctx = crypto_tfm_ctx(tfm); + + __cra_init(tfm, ctx); + + ctx->is_sm3_tfm = true; + + return 0; +} + +static int kmb_ocs_hcu_hmac_sm3_cra_init(struct crypto_tfm *tfm) +{ + struct ocs_hcu_ctx *ctx = crypto_tfm_ctx(tfm); + + __cra_init(tfm, ctx); + + ctx->is_sm3_tfm = true; + ctx->is_hmac_tfm = true; + + return 0; +} + +static int kmb_ocs_hcu_hmac_cra_init(struct crypto_tfm *tfm) +{ + struct ocs_hcu_ctx *ctx = crypto_tfm_ctx(tfm); + + __cra_init(tfm, ctx); + + ctx->is_hmac_tfm = true; + + return 0; +} + +/* Function called when 'tfm' is de-initialized. */ +static void kmb_ocs_hcu_hmac_cra_exit(struct crypto_tfm *tfm) +{ + struct ocs_hcu_ctx *ctx = crypto_tfm_ctx(tfm); + + /* Clear the key. */ + memzero_explicit(ctx->key, sizeof(ctx->key)); +} + +static struct ahash_engine_alg ocs_hcu_algs[] = { +#ifdef CONFIG_CRYPTO_DEV_KEEMBAY_OCS_HCU_HMAC_SHA224 +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.halg = { + .digestsize = SHA224_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "sha224", + .cra_driver_name = "sha224-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA224_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_sha_cra_init, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.setkey = kmb_ocs_hcu_setkey, + .base.halg = { + .digestsize = SHA224_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "hmac(sha224)", + .cra_driver_name = "hmac-sha224-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA224_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_hmac_cra_init, + .cra_exit = kmb_ocs_hcu_hmac_cra_exit, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +#endif /* CONFIG_CRYPTO_DEV_KEEMBAY_OCS_HCU_HMAC_SHA224 */ +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.halg = { + .digestsize = SHA256_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "sha256", + .cra_driver_name = "sha256-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA256_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_sha_cra_init, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.setkey = kmb_ocs_hcu_setkey, + .base.halg = { + .digestsize = SHA256_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "hmac(sha256)", + .cra_driver_name = "hmac-sha256-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA256_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_hmac_cra_init, + .cra_exit = kmb_ocs_hcu_hmac_cra_exit, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.halg = { + .digestsize = SM3_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "sm3", + .cra_driver_name = "sm3-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SM3_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_sm3_cra_init, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.setkey = kmb_ocs_hcu_setkey, + .base.halg = { + .digestsize = SM3_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "hmac(sm3)", + .cra_driver_name = "hmac-sm3-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SM3_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_hmac_sm3_cra_init, + .cra_exit = kmb_ocs_hcu_hmac_cra_exit, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.halg = { + .digestsize = SHA384_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "sha384", + .cra_driver_name = "sha384-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA384_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_sha_cra_init, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.setkey = kmb_ocs_hcu_setkey, + .base.halg = { + .digestsize = SHA384_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "hmac(sha384)", + .cra_driver_name = "hmac-sha384-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA384_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_hmac_cra_init, + .cra_exit = kmb_ocs_hcu_hmac_cra_exit, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.halg = { + .digestsize = SHA512_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "sha512", + .cra_driver_name = "sha512-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA512_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_sha_cra_init, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +{ + .base.init = kmb_ocs_hcu_init, + .base.update = kmb_ocs_hcu_update, + .base.final = kmb_ocs_hcu_final, + .base.finup = kmb_ocs_hcu_finup, + .base.digest = kmb_ocs_hcu_digest, + .base.export = kmb_ocs_hcu_export, + .base.import = kmb_ocs_hcu_import, + .base.setkey = kmb_ocs_hcu_setkey, + .base.halg = { + .digestsize = SHA512_DIGEST_SIZE, + .statesize = sizeof(struct ocs_hcu_rctx), + .base = { + .cra_name = "hmac(sha512)", + .cra_driver_name = "hmac-sha512-keembay-ocs", + .cra_priority = 255, + .cra_flags = CRYPTO_ALG_ASYNC, + .cra_blocksize = SHA512_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct ocs_hcu_ctx), + .cra_alignmask = 0, + .cra_module = THIS_MODULE, + .cra_init = kmb_ocs_hcu_hmac_cra_init, + .cra_exit = kmb_ocs_hcu_hmac_cra_exit, + } + }, + .op.do_one_request = kmb_ocs_hcu_do_one_request, +}, +}; + +/* Device tree driver match. */ +static const struct of_device_id kmb_ocs_hcu_of_match[] = { + { + .compatible = "intel,keembay-ocs-hcu", + }, + {} +}; +MODULE_DEVICE_TABLE(of, kmb_ocs_hcu_of_match); + +static void kmb_ocs_hcu_remove(struct platform_device *pdev) +{ + struct ocs_hcu_dev *hcu_dev = platform_get_drvdata(pdev); + + crypto_engine_unregister_ahashes(ocs_hcu_algs, ARRAY_SIZE(ocs_hcu_algs)); + + crypto_engine_exit(hcu_dev->engine); + + spin_lock_bh(&ocs_hcu.lock); + list_del(&hcu_dev->list); + spin_unlock_bh(&ocs_hcu.lock); +} + +static int kmb_ocs_hcu_probe(struct platform_device *pdev) +{ + struct device *dev = &pdev->dev; + struct ocs_hcu_dev *hcu_dev; + int rc; + + hcu_dev = devm_kzalloc(dev, sizeof(*hcu_dev), GFP_KERNEL); + if (!hcu_dev) + return -ENOMEM; + + hcu_dev->dev = dev; + + platform_set_drvdata(pdev, hcu_dev); + rc = dma_set_mask_and_coherent(&pdev->dev, OCS_HCU_DMA_BIT_MASK); + if (rc) + return rc; + + hcu_dev->io_base = devm_platform_ioremap_resource(pdev, 0); + if (IS_ERR(hcu_dev->io_base)) + return PTR_ERR(hcu_dev->io_base); + + init_completion(&hcu_dev->irq_done); + + /* Get and request IRQ. */ + hcu_dev->irq = platform_get_irq(pdev, 0); + if (hcu_dev->irq < 0) + return hcu_dev->irq; + + rc = devm_request_threaded_irq(&pdev->dev, hcu_dev->irq, + ocs_hcu_irq_handler, NULL, 0, + "keembay-ocs-hcu", hcu_dev); + if (rc < 0) { + dev_err(dev, "Could not request IRQ.\n"); + return rc; + } + + INIT_LIST_HEAD(&hcu_dev->list); + + spin_lock_bh(&ocs_hcu.lock); + list_add_tail(&hcu_dev->list, &ocs_hcu.dev_list); + spin_unlock_bh(&ocs_hcu.lock); + + /* Initialize crypto engine */ + hcu_dev->engine = crypto_engine_alloc_init(dev, 1); + if (!hcu_dev->engine) { + rc = -ENOMEM; + goto list_del; + } + + rc = crypto_engine_start(hcu_dev->engine); + if (rc) { + dev_err(dev, "Could not start engine.\n"); + goto cleanup; + } + + /* Security infrastructure guarantees OCS clock is enabled. */ + + rc = crypto_engine_register_ahashes(ocs_hcu_algs, ARRAY_SIZE(ocs_hcu_algs)); + if (rc) { + dev_err(dev, "Could not register algorithms.\n"); + goto cleanup; + } + + return 0; + +cleanup: + crypto_engine_exit(hcu_dev->engine); +list_del: + spin_lock_bh(&ocs_hcu.lock); + list_del(&hcu_dev->list); + spin_unlock_bh(&ocs_hcu.lock); + + return rc; +} + +/* The OCS driver is a platform device. */ +static struct platform_driver kmb_ocs_hcu_driver = { + .probe = kmb_ocs_hcu_probe, + .remove = kmb_ocs_hcu_remove, + .driver = { + .name = DRV_NAME, + .of_match_table = kmb_ocs_hcu_of_match, + }, +}; + +module_platform_driver(kmb_ocs_hcu_driver); + +MODULE_LICENSE("GPL"); diff --git a/drivers/crypto/intel/keembay/ocs-aes.c b/drivers/crypto/intel/keembay/ocs-aes.c new file mode 100644 index 000000000000..bb6f33f6b4d3 --- /dev/null +++ b/drivers/crypto/intel/keembay/ocs-aes.c @@ -0,0 +1,1489 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Intel Keem Bay OCS AES Crypto Driver. + * + * Copyright (C) 2018-2020 Intel Corporation + */ + +#include <linux/dma-mapping.h> +#include <linux/interrupt.h> +#include <linux/kernel.h> +#include <linux/platform_device.h> +#include <linux/slab.h> +#include <linux/swab.h> + +#include <asm/byteorder.h> +#include <asm/errno.h> + +#include <crypto/aes.h> +#include <crypto/gcm.h> + +#include "ocs-aes.h" + +#define AES_COMMAND_OFFSET 0x0000 +#define AES_KEY_0_OFFSET 0x0004 +#define AES_KEY_1_OFFSET 0x0008 +#define AES_KEY_2_OFFSET 0x000C +#define AES_KEY_3_OFFSET 0x0010 +#define AES_KEY_4_OFFSET 0x0014 +#define AES_KEY_5_OFFSET 0x0018 +#define AES_KEY_6_OFFSET 0x001C +#define AES_KEY_7_OFFSET 0x0020 +#define AES_IV_0_OFFSET 0x0024 +#define AES_IV_1_OFFSET 0x0028 +#define AES_IV_2_OFFSET 0x002C +#define AES_IV_3_OFFSET 0x0030 +#define AES_ACTIVE_OFFSET 0x0034 +#define AES_STATUS_OFFSET 0x0038 +#define AES_KEY_SIZE_OFFSET 0x0044 +#define AES_IER_OFFSET 0x0048 +#define AES_ISR_OFFSET 0x005C +#define AES_MULTIPURPOSE1_0_OFFSET 0x0200 +#define AES_MULTIPURPOSE1_1_OFFSET 0x0204 +#define AES_MULTIPURPOSE1_2_OFFSET 0x0208 +#define AES_MULTIPURPOSE1_3_OFFSET 0x020C +#define AES_MULTIPURPOSE2_0_OFFSET 0x0220 +#define AES_MULTIPURPOSE2_1_OFFSET 0x0224 +#define AES_MULTIPURPOSE2_2_OFFSET 0x0228 +#define AES_MULTIPURPOSE2_3_OFFSET 0x022C +#define AES_BYTE_ORDER_CFG_OFFSET 0x02C0 +#define AES_TLEN_OFFSET 0x0300 +#define AES_T_MAC_0_OFFSET 0x0304 +#define AES_T_MAC_1_OFFSET 0x0308 +#define AES_T_MAC_2_OFFSET 0x030C +#define AES_T_MAC_3_OFFSET 0x0310 +#define AES_PLEN_OFFSET 0x0314 +#define AES_A_DMA_SRC_ADDR_OFFSET 0x0400 +#define AES_A_DMA_DST_ADDR_OFFSET 0x0404 +#define AES_A_DMA_SRC_SIZE_OFFSET 0x0408 +#define AES_A_DMA_DST_SIZE_OFFSET 0x040C +#define AES_A_DMA_DMA_MODE_OFFSET 0x0410 +#define AES_A_DMA_NEXT_SRC_DESCR_OFFSET 0x0418 +#define AES_A_DMA_NEXT_DST_DESCR_OFFSET 0x041C +#define AES_A_DMA_WHILE_ACTIVE_MODE_OFFSET 0x0420 +#define AES_A_DMA_LOG_OFFSET 0x0424 +#define AES_A_DMA_STATUS_OFFSET 0x0428 +#define AES_A_DMA_PERF_CNTR_OFFSET 0x042C +#define AES_A_DMA_MSI_ISR_OFFSET 0x0480 +#define AES_A_DMA_MSI_IER_OFFSET 0x0484 +#define AES_A_DMA_MSI_MASK_OFFSET 0x0488 +#define AES_A_DMA_INBUFFER_WRITE_FIFO_OFFSET 0x0600 +#define AES_A_DMA_OUTBUFFER_READ_FIFO_OFFSET 0x0700 + +/* + * AES_A_DMA_DMA_MODE register. + * Default: 0x00000000. + * bit[31] ACTIVE + * This bit activates the DMA. When the DMA finishes, it resets + * this bit to zero. + * bit[30:26] Unused by this driver. + * bit[25] SRC_LINK_LIST_EN + * Source link list enable bit. When the linked list is terminated + * this bit is reset by the DMA. + * bit[24] DST_LINK_LIST_EN + * Destination link list enable bit. When the linked list is + * terminated this bit is reset by the DMA. + * bit[23:0] Unused by this driver. + */ +#define AES_A_DMA_DMA_MODE_ACTIVE BIT(31) +#define AES_A_DMA_DMA_MODE_SRC_LINK_LIST_EN BIT(25) +#define AES_A_DMA_DMA_MODE_DST_LINK_LIST_EN BIT(24) + +/* + * AES_ACTIVE register + * default 0x00000000 + * bit[31:10] Reserved + * bit[9] LAST_ADATA + * bit[8] LAST_GCX + * bit[7:2] Reserved + * bit[1] TERMINATION + * bit[0] TRIGGER + */ +#define AES_ACTIVE_LAST_ADATA BIT(9) +#define AES_ACTIVE_LAST_CCM_GCM BIT(8) +#define AES_ACTIVE_TERMINATION BIT(1) +#define AES_ACTIVE_TRIGGER BIT(0) + +#define AES_DISABLE_INT 0x00000000 +#define AES_DMA_CPD_ERR_INT BIT(8) +#define AES_DMA_OUTBUF_RD_ERR_INT BIT(7) +#define AES_DMA_OUTBUF_WR_ERR_INT BIT(6) +#define AES_DMA_INBUF_RD_ERR_INT BIT(5) +#define AES_DMA_INBUF_WR_ERR_INT BIT(4) +#define AES_DMA_BAD_COMP_INT BIT(3) +#define AES_DMA_SAI_INT BIT(2) +#define AES_DMA_SRC_DONE_INT BIT(0) +#define AES_COMPLETE_INT BIT(1) + +#define AES_DMA_MSI_MASK_CLEAR BIT(0) + +#define AES_128_BIT_KEY 0x00000000 +#define AES_256_BIT_KEY BIT(0) + +#define AES_DEACTIVATE_PERF_CNTR 0x00000000 +#define AES_ACTIVATE_PERF_CNTR BIT(0) + +#define AES_MAX_TAG_SIZE_U32 4 + +#define OCS_LL_DMA_FLAG_TERMINATE BIT(31) + +/* + * There is an inconsistency in the documentation. This is documented as a + * 11-bit value, but it is actually 10-bits. + */ +#define AES_DMA_STATUS_INPUT_BUFFER_OCCUPANCY_MASK 0x3FF + +/* + * During CCM decrypt, the OCS block needs to finish processing the ciphertext + * before the tag is written. For 128-bit mode this required delay is 28 OCS + * clock cycles. For 256-bit mode it is 36 OCS clock cycles. + */ +#define CCM_DECRYPT_DELAY_TAG_CLK_COUNT 36UL + +/* + * During CCM decrypt there must be a delay of at least 42 OCS clock cycles + * between setting the TRIGGER bit in AES_ACTIVE and setting the LAST_CCM_GCM + * bit in the same register (as stated in the OCS databook) + */ +#define CCM_DECRYPT_DELAY_LAST_GCX_CLK_COUNT 42UL + +/* See RFC3610 section 2.2 */ +#define L_PRIME_MIN (1) +#define L_PRIME_MAX (7) +/* + * CCM IV format from RFC 3610 section 2.3 + * + * Octet Number Contents + * ------------ --------- + * 0 Flags + * 1 ... 15-L Nonce N + * 16-L ... 15 Counter i + * + * Flags = L' = L - 1 + */ +#define L_PRIME_IDX 0 +#define COUNTER_START(lprime) (16 - ((lprime) + 1)) +#define COUNTER_LEN(lprime) ((lprime) + 1) + +enum aes_counter_mode { + AES_CTR_M_NO_INC = 0, + AES_CTR_M_32_INC = 1, + AES_CTR_M_64_INC = 2, + AES_CTR_M_128_INC = 3, +}; + +/** + * struct ocs_dma_linked_list - OCS DMA linked list entry. + * @src_addr: Source address of the data. + * @src_len: Length of data to be fetched. + * @next: Next dma_list to fetch. + * @ll_flags: Flags (Freeze @ terminate) for the DMA engine. + */ +struct ocs_dma_linked_list { + u32 src_addr; + u32 src_len; + u32 next; + u32 ll_flags; +} __packed; + +/* + * Set endianness of inputs and outputs + * AES_BYTE_ORDER_CFG + * default 0x00000000 + * bit [10] - KEY_HI_LO_SWAP + * bit [9] - KEY_HI_SWAP_DWORDS_IN_OCTWORD + * bit [8] - KEY_HI_SWAP_BYTES_IN_DWORD + * bit [7] - KEY_LO_SWAP_DWORDS_IN_OCTWORD + * bit [6] - KEY_LO_SWAP_BYTES_IN_DWORD + * bit [5] - IV_SWAP_DWORDS_IN_OCTWORD + * bit [4] - IV_SWAP_BYTES_IN_DWORD + * bit [3] - DOUT_SWAP_DWORDS_IN_OCTWORD + * bit [2] - DOUT_SWAP_BYTES_IN_DWORD + * bit [1] - DOUT_SWAP_DWORDS_IN_OCTWORD + * bit [0] - DOUT_SWAP_BYTES_IN_DWORD + */ +static inline void aes_a_set_endianness(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(0x7FF, aes_dev->base_reg + AES_BYTE_ORDER_CFG_OFFSET); +} + +/* Trigger AES process start. */ +static inline void aes_a_op_trigger(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_ACTIVE_TRIGGER, aes_dev->base_reg + AES_ACTIVE_OFFSET); +} + +/* Indicate last bulk of data. */ +static inline void aes_a_op_termination(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_ACTIVE_TERMINATION, + aes_dev->base_reg + AES_ACTIVE_OFFSET); +} + +/* + * Set LAST_CCM_GCM in AES_ACTIVE register and clear all other bits. + * + * Called when DMA is programmed to fetch the last batch of data. + * - For AES-CCM it is called for the last batch of Payload data and Ciphertext + * data. + * - For AES-GCM, it is called for the last batch of Plaintext data and + * Ciphertext data. + */ +static inline void aes_a_set_last_gcx(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_ACTIVE_LAST_CCM_GCM, + aes_dev->base_reg + AES_ACTIVE_OFFSET); +} + +/* Wait for LAST_CCM_GCM bit to be unset. */ +static inline void aes_a_wait_last_gcx(const struct ocs_aes_dev *aes_dev) +{ + u32 aes_active_reg; + + do { + aes_active_reg = ioread32(aes_dev->base_reg + + AES_ACTIVE_OFFSET); + } while (aes_active_reg & AES_ACTIVE_LAST_CCM_GCM); +} + +/* Wait for 10 bits of input occupancy. */ +static void aes_a_dma_wait_input_buffer_occupancy(const struct ocs_aes_dev *aes_dev) +{ + u32 reg; + + do { + reg = ioread32(aes_dev->base_reg + AES_A_DMA_STATUS_OFFSET); + } while (reg & AES_DMA_STATUS_INPUT_BUFFER_OCCUPANCY_MASK); +} + + /* + * Set LAST_CCM_GCM and LAST_ADATA bits in AES_ACTIVE register (and clear all + * other bits). + * + * Called when DMA is programmed to fetch the last batch of Associated Data + * (CCM case) or Additional Authenticated Data (GCM case). + */ +static inline void aes_a_set_last_gcx_and_adata(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_ACTIVE_LAST_ADATA | AES_ACTIVE_LAST_CCM_GCM, + aes_dev->base_reg + AES_ACTIVE_OFFSET); +} + +/* Set DMA src and dst transfer size to 0 */ +static inline void aes_a_dma_set_xfer_size_zero(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(0, aes_dev->base_reg + AES_A_DMA_SRC_SIZE_OFFSET); + iowrite32(0, aes_dev->base_reg + AES_A_DMA_DST_SIZE_OFFSET); +} + +/* Activate DMA for zero-byte transfer case. */ +static inline void aes_a_dma_active(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_A_DMA_DMA_MODE_ACTIVE, + aes_dev->base_reg + AES_A_DMA_DMA_MODE_OFFSET); +} + +/* Activate DMA and enable src linked list */ +static inline void aes_a_dma_active_src_ll_en(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_A_DMA_DMA_MODE_ACTIVE | + AES_A_DMA_DMA_MODE_SRC_LINK_LIST_EN, + aes_dev->base_reg + AES_A_DMA_DMA_MODE_OFFSET); +} + +/* Activate DMA and enable dst linked list */ +static inline void aes_a_dma_active_dst_ll_en(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_A_DMA_DMA_MODE_ACTIVE | + AES_A_DMA_DMA_MODE_DST_LINK_LIST_EN, + aes_dev->base_reg + AES_A_DMA_DMA_MODE_OFFSET); +} + +/* Activate DMA and enable src and dst linked lists */ +static inline void aes_a_dma_active_src_dst_ll_en(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(AES_A_DMA_DMA_MODE_ACTIVE | + AES_A_DMA_DMA_MODE_SRC_LINK_LIST_EN | + AES_A_DMA_DMA_MODE_DST_LINK_LIST_EN, + aes_dev->base_reg + AES_A_DMA_DMA_MODE_OFFSET); +} + +/* Reset PERF_CNTR to 0 and activate it */ +static inline void aes_a_dma_reset_and_activate_perf_cntr(const struct ocs_aes_dev *aes_dev) +{ + iowrite32(0x00000000, aes_dev->base_reg + AES_A_DMA_PERF_CNTR_OFFSET); + iowrite32(AES_ACTIVATE_PERF_CNTR, + aes_dev->base_reg + AES_A_DMA_WHILE_ACTIVE_MODE_OFFSET); +} + +/* Wait until PERF_CNTR is > delay, then deactivate it */ +static inline void aes_a_dma_wait_and_deactivate_perf_cntr(const struct ocs_aes_dev *aes_dev, + int delay) +{ + while (ioread32(aes_dev->base_reg + AES_A_DMA_PERF_CNTR_OFFSET) < delay) + ; + iowrite32(AES_DEACTIVATE_PERF_CNTR, + aes_dev->base_reg + AES_A_DMA_WHILE_ACTIVE_MODE_OFFSET); +} + +/* Disable AES and DMA IRQ. */ +static void aes_irq_disable(struct ocs_aes_dev *aes_dev) +{ + u32 isr_val = 0; + + /* Disable interrupts */ + iowrite32(AES_DISABLE_INT, + aes_dev->base_reg + AES_A_DMA_MSI_IER_OFFSET); + iowrite32(AES_DISABLE_INT, aes_dev->base_reg + AES_IER_OFFSET); + + /* Clear any pending interrupt */ + isr_val = ioread32(aes_dev->base_reg + AES_A_DMA_MSI_ISR_OFFSET); + if (isr_val) + iowrite32(isr_val, + aes_dev->base_reg + AES_A_DMA_MSI_ISR_OFFSET); + + isr_val = ioread32(aes_dev->base_reg + AES_A_DMA_MSI_MASK_OFFSET); + if (isr_val) + iowrite32(isr_val, + aes_dev->base_reg + AES_A_DMA_MSI_MASK_OFFSET); + + isr_val = ioread32(aes_dev->base_reg + AES_ISR_OFFSET); + if (isr_val) + iowrite32(isr_val, aes_dev->base_reg + AES_ISR_OFFSET); +} + +/* Enable AES or DMA IRQ. IRQ is disabled once fired. */ +static void aes_irq_enable(struct ocs_aes_dev *aes_dev, u8 irq) +{ + if (irq == AES_COMPLETE_INT) { + /* Ensure DMA error interrupts are enabled */ + iowrite32(AES_DMA_CPD_ERR_INT | + AES_DMA_OUTBUF_RD_ERR_INT | + AES_DMA_OUTBUF_WR_ERR_INT | + AES_DMA_INBUF_RD_ERR_INT | + AES_DMA_INBUF_WR_ERR_INT | + AES_DMA_BAD_COMP_INT | + AES_DMA_SAI_INT, + aes_dev->base_reg + AES_A_DMA_MSI_IER_OFFSET); + /* + * AES_IER + * default 0x00000000 + * bits [31:3] - reserved + * bit [2] - EN_SKS_ERR + * bit [1] - EN_AES_COMPLETE + * bit [0] - reserved + */ + iowrite32(AES_COMPLETE_INT, aes_dev->base_reg + AES_IER_OFFSET); + return; + } + if (irq == AES_DMA_SRC_DONE_INT) { + /* Ensure AES interrupts are disabled */ + iowrite32(AES_DISABLE_INT, aes_dev->base_reg + AES_IER_OFFSET); + /* + * DMA_MSI_IER + * default 0x00000000 + * bits [31:9] - reserved + * bit [8] - CPD_ERR_INT_EN + * bit [7] - OUTBUF_RD_ERR_INT_EN + * bit [6] - OUTBUF_WR_ERR_INT_EN + * bit [5] - INBUF_RD_ERR_INT_EN + * bit [4] - INBUF_WR_ERR_INT_EN + * bit [3] - BAD_COMP_INT_EN + * bit [2] - SAI_INT_EN + * bit [1] - DST_DONE_INT_EN + * bit [0] - SRC_DONE_INT_EN + */ + iowrite32(AES_DMA_CPD_ERR_INT | + AES_DMA_OUTBUF_RD_ERR_INT | + AES_DMA_OUTBUF_WR_ERR_INT | + AES_DMA_INBUF_RD_ERR_INT | + AES_DMA_INBUF_WR_ERR_INT | + AES_DMA_BAD_COMP_INT | + AES_DMA_SAI_INT | + AES_DMA_SRC_DONE_INT, + aes_dev->base_reg + AES_A_DMA_MSI_IER_OFFSET); + } +} + +/* Enable and wait for IRQ (either from OCS AES engine or DMA) */ +static int ocs_aes_irq_enable_and_wait(struct ocs_aes_dev *aes_dev, u8 irq) +{ + int rc; + + reinit_completion(&aes_dev->irq_completion); + aes_irq_enable(aes_dev, irq); + rc = wait_for_completion_interruptible(&aes_dev->irq_completion); + if (rc) + return rc; + + return aes_dev->dma_err_mask ? -EIO : 0; +} + +/* Configure DMA to OCS, linked list mode */ +static inline void dma_to_ocs_aes_ll(struct ocs_aes_dev *aes_dev, + dma_addr_t dma_list) +{ + iowrite32(0, aes_dev->base_reg + AES_A_DMA_SRC_SIZE_OFFSET); + iowrite32(dma_list, + aes_dev->base_reg + AES_A_DMA_NEXT_SRC_DESCR_OFFSET); +} + +/* Configure DMA from OCS, linked list mode */ +static inline void dma_from_ocs_aes_ll(struct ocs_aes_dev *aes_dev, + dma_addr_t dma_list) +{ + iowrite32(0, aes_dev->base_reg + AES_A_DMA_DST_SIZE_OFFSET); + iowrite32(dma_list, + aes_dev->base_reg + AES_A_DMA_NEXT_DST_DESCR_OFFSET); +} + +irqreturn_t ocs_aes_irq_handler(int irq, void *dev_id) +{ + struct ocs_aes_dev *aes_dev = dev_id; + u32 aes_dma_isr; + + /* Read DMA ISR status. */ + aes_dma_isr = ioread32(aes_dev->base_reg + AES_A_DMA_MSI_ISR_OFFSET); + + /* Disable and clear interrupts. */ + aes_irq_disable(aes_dev); + + /* Save DMA error status. */ + aes_dev->dma_err_mask = aes_dma_isr & + (AES_DMA_CPD_ERR_INT | + AES_DMA_OUTBUF_RD_ERR_INT | + AES_DMA_OUTBUF_WR_ERR_INT | + AES_DMA_INBUF_RD_ERR_INT | + AES_DMA_INBUF_WR_ERR_INT | + AES_DMA_BAD_COMP_INT | + AES_DMA_SAI_INT); + + /* Signal IRQ completion. */ + complete(&aes_dev->irq_completion); + + return IRQ_HANDLED; +} + +/** + * ocs_aes_set_key() - Write key into OCS AES hardware. + * @aes_dev: The OCS AES device to write the key to. + * @key_size: The size of the key (in bytes). + * @key: The key to write. + * @cipher: The cipher the key is for. + * + * For AES @key_size must be either 16 or 32. For SM4 @key_size must be 16. + * + * Return: 0 on success, negative error code otherwise. + */ +int ocs_aes_set_key(struct ocs_aes_dev *aes_dev, u32 key_size, const u8 *key, + enum ocs_cipher cipher) +{ + const u32 *key_u32; + u32 val; + int i; + + /* OCS AES supports 128-bit and 256-bit keys only. */ + if (cipher == OCS_AES && !(key_size == 32 || key_size == 16)) { + dev_err(aes_dev->dev, + "%d-bit keys not supported by AES cipher\n", + key_size * 8); + return -EINVAL; + } + /* OCS SM4 supports 128-bit keys only. */ + if (cipher == OCS_SM4 && key_size != 16) { + dev_err(aes_dev->dev, + "%d-bit keys not supported for SM4 cipher\n", + key_size * 8); + return -EINVAL; + } + + if (!key) + return -EINVAL; + + key_u32 = (const u32 *)key; + + /* Write key to AES_KEY[0-7] registers */ + for (i = 0; i < (key_size / sizeof(u32)); i++) { + iowrite32(key_u32[i], + aes_dev->base_reg + AES_KEY_0_OFFSET + + (i * sizeof(u32))); + } + /* + * Write key size + * bits [31:1] - reserved + * bit [0] - AES_KEY_SIZE + * 0 - 128 bit key + * 1 - 256 bit key + */ + val = (key_size == 16) ? AES_128_BIT_KEY : AES_256_BIT_KEY; + iowrite32(val, aes_dev->base_reg + AES_KEY_SIZE_OFFSET); + + return 0; +} + +/* Write AES_COMMAND */ +static inline void set_ocs_aes_command(struct ocs_aes_dev *aes_dev, + enum ocs_cipher cipher, + enum ocs_mode mode, + enum ocs_instruction instruction) +{ + u32 val; + + /* AES_COMMAND + * default 0x000000CC + * bit [14] - CIPHER_SELECT + * 0 - AES + * 1 - SM4 + * bits [11:8] - OCS_AES_MODE + * 0000 - ECB + * 0001 - CBC + * 0010 - CTR + * 0110 - CCM + * 0111 - GCM + * 1001 - CTS + * bits [7:6] - AES_INSTRUCTION + * 00 - ENCRYPT + * 01 - DECRYPT + * 10 - EXPAND + * 11 - BYPASS + * bits [3:2] - CTR_M_BITS + * 00 - No increment + * 01 - Least significant 32 bits are incremented + * 10 - Least significant 64 bits are incremented + * 11 - Full 128 bits are incremented + */ + val = (cipher << 14) | (mode << 8) | (instruction << 6) | + (AES_CTR_M_128_INC << 2); + iowrite32(val, aes_dev->base_reg + AES_COMMAND_OFFSET); +} + +static void ocs_aes_init(struct ocs_aes_dev *aes_dev, + enum ocs_mode mode, + enum ocs_cipher cipher, + enum ocs_instruction instruction) +{ + /* Ensure interrupts are disabled and pending interrupts cleared. */ + aes_irq_disable(aes_dev); + + /* Set endianness recommended by data-sheet. */ + aes_a_set_endianness(aes_dev); + + /* Set AES_COMMAND register. */ + set_ocs_aes_command(aes_dev, cipher, mode, instruction); +} + +/* + * Write the byte length of the last AES/SM4 block of Payload data (without + * zero padding and without the length of the MAC) in register AES_PLEN. + */ +static inline void ocs_aes_write_last_data_blk_len(struct ocs_aes_dev *aes_dev, + u32 size) +{ + u32 val; + + if (size == 0) { + val = 0; + goto exit; + } + + val = size % AES_BLOCK_SIZE; + if (val == 0) + val = AES_BLOCK_SIZE; + +exit: + iowrite32(val, aes_dev->base_reg + AES_PLEN_OFFSET); +} + +/* + * Validate inputs according to mode. + * If OK return 0; else return -EINVAL. + */ +static int ocs_aes_validate_inputs(dma_addr_t src_dma_list, u32 src_size, + const u8 *iv, u32 iv_size, + dma_addr_t aad_dma_list, u32 aad_size, + const u8 *tag, u32 tag_size, + enum ocs_cipher cipher, enum ocs_mode mode, + enum ocs_instruction instruction, + dma_addr_t dst_dma_list) +{ + /* Ensure cipher, mode and instruction are valid. */ + if (!(cipher == OCS_AES || cipher == OCS_SM4)) + return -EINVAL; + + if (mode != OCS_MODE_ECB && mode != OCS_MODE_CBC && + mode != OCS_MODE_CTR && mode != OCS_MODE_CCM && + mode != OCS_MODE_GCM && mode != OCS_MODE_CTS) + return -EINVAL; + + if (instruction != OCS_ENCRYPT && instruction != OCS_DECRYPT && + instruction != OCS_EXPAND && instruction != OCS_BYPASS) + return -EINVAL; + + /* + * When instruction is OCS_BYPASS, OCS simply copies data from source + * to destination using DMA. + * + * AES mode is irrelevant, but both source and destination DMA + * linked-list must be defined. + */ + if (instruction == OCS_BYPASS) { + if (src_dma_list == DMA_MAPPING_ERROR || + dst_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + return 0; + } + + /* + * For performance reasons switch based on mode to limit unnecessary + * conditionals for each mode + */ + switch (mode) { + case OCS_MODE_ECB: + /* Ensure input length is multiple of block size */ + if (src_size % AES_BLOCK_SIZE != 0) + return -EINVAL; + + /* Ensure source and destination linked lists are created */ + if (src_dma_list == DMA_MAPPING_ERROR || + dst_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + return 0; + + case OCS_MODE_CBC: + /* Ensure input length is multiple of block size */ + if (src_size % AES_BLOCK_SIZE != 0) + return -EINVAL; + + /* Ensure source and destination linked lists are created */ + if (src_dma_list == DMA_MAPPING_ERROR || + dst_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + /* Ensure IV is present and block size in length */ + if (!iv || iv_size != AES_BLOCK_SIZE) + return -EINVAL; + + return 0; + + case OCS_MODE_CTR: + /* Ensure input length of 1 byte or greater */ + if (src_size == 0) + return -EINVAL; + + /* Ensure source and destination linked lists are created */ + if (src_dma_list == DMA_MAPPING_ERROR || + dst_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + /* Ensure IV is present and block size in length */ + if (!iv || iv_size != AES_BLOCK_SIZE) + return -EINVAL; + + return 0; + + case OCS_MODE_CTS: + /* Ensure input length >= block size */ + if (src_size < AES_BLOCK_SIZE) + return -EINVAL; + + /* Ensure source and destination linked lists are created */ + if (src_dma_list == DMA_MAPPING_ERROR || + dst_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + /* Ensure IV is present and block size in length */ + if (!iv || iv_size != AES_BLOCK_SIZE) + return -EINVAL; + + return 0; + + case OCS_MODE_GCM: + /* Ensure IV is present and GCM_AES_IV_SIZE in length */ + if (!iv || iv_size != GCM_AES_IV_SIZE) + return -EINVAL; + + /* + * If input data present ensure source and destination linked + * lists are created + */ + if (src_size && (src_dma_list == DMA_MAPPING_ERROR || + dst_dma_list == DMA_MAPPING_ERROR)) + return -EINVAL; + + /* If aad present ensure aad linked list is created */ + if (aad_size && aad_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + /* Ensure tag destination is set */ + if (!tag) + return -EINVAL; + + /* Just ensure that tag_size doesn't cause overflows. */ + if (tag_size > (AES_MAX_TAG_SIZE_U32 * sizeof(u32))) + return -EINVAL; + + return 0; + + case OCS_MODE_CCM: + /* Ensure IV is present and block size in length */ + if (!iv || iv_size != AES_BLOCK_SIZE) + return -EINVAL; + + /* 2 <= L <= 8, so 1 <= L' <= 7 */ + if (iv[L_PRIME_IDX] < L_PRIME_MIN || + iv[L_PRIME_IDX] > L_PRIME_MAX) + return -EINVAL; + + /* If aad present ensure aad linked list is created */ + if (aad_size && aad_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + /* Just ensure that tag_size doesn't cause overflows. */ + if (tag_size > (AES_MAX_TAG_SIZE_U32 * sizeof(u32))) + return -EINVAL; + + if (instruction == OCS_DECRYPT) { + /* + * If input data present ensure source and destination + * linked lists are created + */ + if (src_size && (src_dma_list == DMA_MAPPING_ERROR || + dst_dma_list == DMA_MAPPING_ERROR)) + return -EINVAL; + + /* Ensure input tag is present */ + if (!tag) + return -EINVAL; + + return 0; + } + + /* Instruction == OCS_ENCRYPT */ + + /* + * Destination linked list always required (for tag even if no + * input data) + */ + if (dst_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + /* If input data present ensure src linked list is created */ + if (src_size && src_dma_list == DMA_MAPPING_ERROR) + return -EINVAL; + + return 0; + + default: + return -EINVAL; + } +} + +/** + * ocs_aes_op() - Perform AES/SM4 operation. + * @aes_dev: The OCS AES device to use. + * @mode: The mode to use (ECB, CBC, CTR, or CTS). + * @cipher: The cipher to use (AES or SM4). + * @instruction: The instruction to perform (encrypt or decrypt). + * @dst_dma_list: The OCS DMA list mapping output memory. + * @src_dma_list: The OCS DMA list mapping input payload data. + * @src_size: The amount of data mapped by @src_dma_list. + * @iv: The IV vector. + * @iv_size: The size (in bytes) of @iv. + * + * Return: 0 on success, negative error code otherwise. + */ +int ocs_aes_op(struct ocs_aes_dev *aes_dev, + enum ocs_mode mode, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size, + u8 *iv, + u32 iv_size) +{ + u32 *iv32; + int rc; + + rc = ocs_aes_validate_inputs(src_dma_list, src_size, iv, iv_size, 0, 0, + NULL, 0, cipher, mode, instruction, + dst_dma_list); + if (rc) + return rc; + /* + * ocs_aes_validate_inputs() is a generic check, now ensure mode is not + * GCM or CCM. + */ + if (mode == OCS_MODE_GCM || mode == OCS_MODE_CCM) + return -EINVAL; + + /* Cast IV to u32 array. */ + iv32 = (u32 *)iv; + + ocs_aes_init(aes_dev, mode, cipher, instruction); + + if (mode == OCS_MODE_CTS) { + /* Write the byte length of the last data block to engine. */ + ocs_aes_write_last_data_blk_len(aes_dev, src_size); + } + + /* ECB is the only mode that doesn't use IV. */ + if (mode != OCS_MODE_ECB) { + iowrite32(iv32[0], aes_dev->base_reg + AES_IV_0_OFFSET); + iowrite32(iv32[1], aes_dev->base_reg + AES_IV_1_OFFSET); + iowrite32(iv32[2], aes_dev->base_reg + AES_IV_2_OFFSET); + iowrite32(iv32[3], aes_dev->base_reg + AES_IV_3_OFFSET); + } + + /* Set AES_ACTIVE.TRIGGER to start the operation. */ + aes_a_op_trigger(aes_dev); + + /* Configure and activate input / output DMA. */ + dma_to_ocs_aes_ll(aes_dev, src_dma_list); + dma_from_ocs_aes_ll(aes_dev, dst_dma_list); + aes_a_dma_active_src_dst_ll_en(aes_dev); + + if (mode == OCS_MODE_CTS) { + /* + * For CTS mode, instruct engine to activate ciphertext + * stealing if last block of data is incomplete. + */ + aes_a_set_last_gcx(aes_dev); + } else { + /* For all other modes, just write the 'termination' bit. */ + aes_a_op_termination(aes_dev); + } + + /* Wait for engine to complete processing. */ + rc = ocs_aes_irq_enable_and_wait(aes_dev, AES_COMPLETE_INT); + if (rc) + return rc; + + if (mode == OCS_MODE_CTR) { + /* Read back IV for streaming mode */ + iv32[0] = ioread32(aes_dev->base_reg + AES_IV_0_OFFSET); + iv32[1] = ioread32(aes_dev->base_reg + AES_IV_1_OFFSET); + iv32[2] = ioread32(aes_dev->base_reg + AES_IV_2_OFFSET); + iv32[3] = ioread32(aes_dev->base_reg + AES_IV_3_OFFSET); + } + + return 0; +} + +/* Compute and write J0 to engine registers. */ +static void ocs_aes_gcm_write_j0(const struct ocs_aes_dev *aes_dev, + const u8 *iv) +{ + const u32 *j0 = (u32 *)iv; + + /* + * IV must be 12 bytes; Other sizes not supported as Linux crypto API + * does only expects/allows 12 byte IV for GCM + */ + iowrite32(0x00000001, aes_dev->base_reg + AES_IV_0_OFFSET); + iowrite32(__swab32(j0[2]), aes_dev->base_reg + AES_IV_1_OFFSET); + iowrite32(__swab32(j0[1]), aes_dev->base_reg + AES_IV_2_OFFSET); + iowrite32(__swab32(j0[0]), aes_dev->base_reg + AES_IV_3_OFFSET); +} + +/* Read GCM tag from engine registers. */ +static inline void ocs_aes_gcm_read_tag(struct ocs_aes_dev *aes_dev, + u8 *tag, u32 tag_size) +{ + u32 tag_u32[AES_MAX_TAG_SIZE_U32]; + + /* + * The Authentication Tag T is stored in Little Endian order in the + * registers with the most significant bytes stored from AES_T_MAC[3] + * downward. + */ + tag_u32[0] = __swab32(ioread32(aes_dev->base_reg + AES_T_MAC_3_OFFSET)); + tag_u32[1] = __swab32(ioread32(aes_dev->base_reg + AES_T_MAC_2_OFFSET)); + tag_u32[2] = __swab32(ioread32(aes_dev->base_reg + AES_T_MAC_1_OFFSET)); + tag_u32[3] = __swab32(ioread32(aes_dev->base_reg + AES_T_MAC_0_OFFSET)); + + memcpy(tag, tag_u32, tag_size); +} + +/** + * ocs_aes_gcm_op() - Perform GCM operation. + * @aes_dev: The OCS AES device to use. + * @cipher: The Cipher to use (AES or SM4). + * @instruction: The instruction to perform (encrypt or decrypt). + * @dst_dma_list: The OCS DMA list mapping output memory. + * @src_dma_list: The OCS DMA list mapping input payload data. + * @src_size: The amount of data mapped by @src_dma_list. + * @iv: The input IV vector. + * @aad_dma_list: The OCS DMA list mapping input AAD data. + * @aad_size: The amount of data mapped by @aad_dma_list. + * @out_tag: Where to store computed tag. + * @tag_size: The size (in bytes) of @out_tag. + * + * Return: 0 on success, negative error code otherwise. + */ +int ocs_aes_gcm_op(struct ocs_aes_dev *aes_dev, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size, + const u8 *iv, + dma_addr_t aad_dma_list, + u32 aad_size, + u8 *out_tag, + u32 tag_size) +{ + u64 bit_len; + u32 val; + int rc; + + rc = ocs_aes_validate_inputs(src_dma_list, src_size, iv, + GCM_AES_IV_SIZE, aad_dma_list, + aad_size, out_tag, tag_size, cipher, + OCS_MODE_GCM, instruction, + dst_dma_list); + if (rc) + return rc; + + ocs_aes_init(aes_dev, OCS_MODE_GCM, cipher, instruction); + + /* Compute and write J0 to OCS HW. */ + ocs_aes_gcm_write_j0(aes_dev, iv); + + /* Write out_tag byte length */ + iowrite32(tag_size, aes_dev->base_reg + AES_TLEN_OFFSET); + + /* Write the byte length of the last plaintext / ciphertext block. */ + ocs_aes_write_last_data_blk_len(aes_dev, src_size); + + /* Write ciphertext bit length */ + bit_len = (u64)src_size * 8; + val = bit_len & 0xFFFFFFFF; + iowrite32(val, aes_dev->base_reg + AES_MULTIPURPOSE2_0_OFFSET); + val = bit_len >> 32; + iowrite32(val, aes_dev->base_reg + AES_MULTIPURPOSE2_1_OFFSET); + + /* Write aad bit length */ + bit_len = (u64)aad_size * 8; + val = bit_len & 0xFFFFFFFF; + iowrite32(val, aes_dev->base_reg + AES_MULTIPURPOSE2_2_OFFSET); + val = bit_len >> 32; + iowrite32(val, aes_dev->base_reg + AES_MULTIPURPOSE2_3_OFFSET); + + /* Set AES_ACTIVE.TRIGGER to start the operation. */ + aes_a_op_trigger(aes_dev); + + /* Process AAD. */ + if (aad_size) { + /* If aad present, configure DMA to feed it to the engine. */ + dma_to_ocs_aes_ll(aes_dev, aad_dma_list); + aes_a_dma_active_src_ll_en(aes_dev); + + /* Instructs engine to pad last block of aad, if needed. */ + aes_a_set_last_gcx_and_adata(aes_dev); + + /* Wait for DMA transfer to complete. */ + rc = ocs_aes_irq_enable_and_wait(aes_dev, AES_DMA_SRC_DONE_INT); + if (rc) + return rc; + } else { + aes_a_set_last_gcx_and_adata(aes_dev); + } + + /* Wait until adata (if present) has been processed. */ + aes_a_wait_last_gcx(aes_dev); + aes_a_dma_wait_input_buffer_occupancy(aes_dev); + + /* Now process payload. */ + if (src_size) { + /* Configure and activate DMA for both input and output data. */ + dma_to_ocs_aes_ll(aes_dev, src_dma_list); + dma_from_ocs_aes_ll(aes_dev, dst_dma_list); + aes_a_dma_active_src_dst_ll_en(aes_dev); + } else { + aes_a_dma_set_xfer_size_zero(aes_dev); + aes_a_dma_active(aes_dev); + } + + /* Instruct AES/SMA4 engine payload processing is over. */ + aes_a_set_last_gcx(aes_dev); + + /* Wait for OCS AES engine to complete processing. */ + rc = ocs_aes_irq_enable_and_wait(aes_dev, AES_COMPLETE_INT); + if (rc) + return rc; + + ocs_aes_gcm_read_tag(aes_dev, out_tag, tag_size); + + return 0; +} + +/* Write encrypted tag to AES/SM4 engine. */ +static void ocs_aes_ccm_write_encrypted_tag(struct ocs_aes_dev *aes_dev, + const u8 *in_tag, u32 tag_size) +{ + int i; + + /* Ensure DMA input buffer is empty */ + aes_a_dma_wait_input_buffer_occupancy(aes_dev); + + /* + * During CCM decrypt, the OCS block needs to finish processing the + * ciphertext before the tag is written. So delay needed after DMA has + * completed writing the ciphertext + */ + aes_a_dma_reset_and_activate_perf_cntr(aes_dev); + aes_a_dma_wait_and_deactivate_perf_cntr(aes_dev, + CCM_DECRYPT_DELAY_TAG_CLK_COUNT); + + /* Write encrypted tag to AES/SM4 engine. */ + for (i = 0; i < tag_size; i++) { + iowrite8(in_tag[i], aes_dev->base_reg + + AES_A_DMA_INBUFFER_WRITE_FIFO_OFFSET); + } +} + +/* + * Write B0 CCM block to OCS AES HW. + * + * Note: B0 format is documented in NIST Special Publication 800-38C + * https://nvlpubs.nist.gov/nistpubs/Legacy/SP/nistspecialpublication800-38c.pdf + * (see Section A.2.1) + */ +static int ocs_aes_ccm_write_b0(const struct ocs_aes_dev *aes_dev, + const u8 *iv, u32 adata_size, u32 tag_size, + u32 cryptlen) +{ + u8 b0[16]; /* CCM B0 block is 16 bytes long. */ + int i, q; + + /* Initialize B0 to 0. */ + memset(b0, 0, sizeof(b0)); + + /* + * B0[0] is the 'Flags Octet' and has the following structure: + * bit 7: Reserved + * bit 6: Adata flag + * bit 5-3: t value encoded as (t-2)/2 + * bit 2-0: q value encoded as q - 1 + */ + /* If there is AAD data, set the Adata flag. */ + if (adata_size) + b0[0] |= BIT(6); + /* + * t denotes the octet length of T. + * t can only be an element of { 4, 6, 8, 10, 12, 14, 16} and is + * encoded as (t - 2) / 2 + */ + b0[0] |= (((tag_size - 2) / 2) & 0x7) << 3; + /* + * q is the octet length of Q. + * q can only be an element of {2, 3, 4, 5, 6, 7, 8} and is encoded as + * q - 1 == iv[0] & 0x7; + */ + b0[0] |= iv[0] & 0x7; + /* + * Copy the Nonce N from IV to B0; N is located in iv[1]..iv[15 - q] + * and must be copied to b0[1]..b0[15-q]. + * q == (iv[0] & 0x7) + 1 + */ + q = (iv[0] & 0x7) + 1; + for (i = 1; i <= 15 - q; i++) + b0[i] = iv[i]; + /* + * The rest of B0 must contain Q, i.e., the message length. + * Q is encoded in q octets, in big-endian order, so to write it, we + * start from the end of B0 and we move backward. + */ + i = sizeof(b0) - 1; + while (q) { + b0[i] = cryptlen & 0xff; + cryptlen >>= 8; + i--; + q--; + } + /* + * If cryptlen is not zero at this point, it means that its original + * value was too big. + */ + if (cryptlen) + return -EOVERFLOW; + /* Now write B0 to OCS AES input buffer. */ + for (i = 0; i < sizeof(b0); i++) + iowrite8(b0[i], aes_dev->base_reg + + AES_A_DMA_INBUFFER_WRITE_FIFO_OFFSET); + return 0; +} + +/* + * Write adata length to OCS AES HW. + * + * Note: adata len encoding is documented in NIST Special Publication 800-38C + * https://nvlpubs.nist.gov/nistpubs/Legacy/SP/nistspecialpublication800-38c.pdf + * (see Section A.2.2) + */ +static void ocs_aes_ccm_write_adata_len(const struct ocs_aes_dev *aes_dev, + u64 adata_len) +{ + u8 enc_a[10]; /* Maximum encoded size: 10 octets. */ + int i, len; + + /* + * adata_len ('a') is encoded as follows: + * If 0 < a < 2^16 - 2^8 ==> 'a' encoded as [a]16, i.e., two octets + * (big endian). + * If 2^16 - 2^8 ≤ a < 2^32 ==> 'a' encoded as 0xff || 0xfe || [a]32, + * i.e., six octets (big endian). + * If 2^32 ≤ a < 2^64 ==> 'a' encoded as 0xff || 0xff || [a]64, + * i.e., ten octets (big endian). + */ + if (adata_len < 65280) { + len = 2; + *(__be16 *)enc_a = cpu_to_be16(adata_len); + } else if (adata_len <= 0xFFFFFFFF) { + len = 6; + *(__be16 *)enc_a = cpu_to_be16(0xfffe); + *(__be32 *)&enc_a[2] = cpu_to_be32(adata_len); + } else { /* adata_len >= 2^32 */ + len = 10; + *(__be16 *)enc_a = cpu_to_be16(0xffff); + *(__be64 *)&enc_a[2] = cpu_to_be64(adata_len); + } + for (i = 0; i < len; i++) + iowrite8(enc_a[i], + aes_dev->base_reg + + AES_A_DMA_INBUFFER_WRITE_FIFO_OFFSET); +} + +static int ocs_aes_ccm_do_adata(struct ocs_aes_dev *aes_dev, + dma_addr_t adata_dma_list, u32 adata_size) +{ + int rc; + + if (!adata_size) { + /* Since no aad the LAST_GCX bit can be set now */ + aes_a_set_last_gcx_and_adata(aes_dev); + goto exit; + } + + /* Adata case. */ + + /* + * Form the encoding of the Associated data length and write it + * to the AES/SM4 input buffer. + */ + ocs_aes_ccm_write_adata_len(aes_dev, adata_size); + + /* Configure the AES/SM4 DMA to fetch the Associated Data */ + dma_to_ocs_aes_ll(aes_dev, adata_dma_list); + + /* Activate DMA to fetch Associated data. */ + aes_a_dma_active_src_ll_en(aes_dev); + + /* Set LAST_GCX and LAST_ADATA in AES ACTIVE register. */ + aes_a_set_last_gcx_and_adata(aes_dev); + + /* Wait for DMA transfer to complete. */ + rc = ocs_aes_irq_enable_and_wait(aes_dev, AES_DMA_SRC_DONE_INT); + if (rc) + return rc; + +exit: + /* Wait until adata (if present) has been processed. */ + aes_a_wait_last_gcx(aes_dev); + aes_a_dma_wait_input_buffer_occupancy(aes_dev); + + return 0; +} + +static int ocs_aes_ccm_encrypt_do_payload(struct ocs_aes_dev *aes_dev, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size) +{ + if (src_size) { + /* + * Configure and activate DMA for both input and output + * data. + */ + dma_to_ocs_aes_ll(aes_dev, src_dma_list); + dma_from_ocs_aes_ll(aes_dev, dst_dma_list); + aes_a_dma_active_src_dst_ll_en(aes_dev); + } else { + /* Configure and activate DMA for output data only. */ + dma_from_ocs_aes_ll(aes_dev, dst_dma_list); + aes_a_dma_active_dst_ll_en(aes_dev); + } + + /* + * Set the LAST GCX bit in AES_ACTIVE Register to instruct + * AES/SM4 engine to pad the last block of data. + */ + aes_a_set_last_gcx(aes_dev); + + /* We are done, wait for IRQ and return. */ + return ocs_aes_irq_enable_and_wait(aes_dev, AES_COMPLETE_INT); +} + +static int ocs_aes_ccm_decrypt_do_payload(struct ocs_aes_dev *aes_dev, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size) +{ + if (!src_size) { + /* Let engine process 0-length input. */ + aes_a_dma_set_xfer_size_zero(aes_dev); + aes_a_dma_active(aes_dev); + aes_a_set_last_gcx(aes_dev); + + return 0; + } + + /* + * Configure and activate DMA for both input and output + * data. + */ + dma_to_ocs_aes_ll(aes_dev, src_dma_list); + dma_from_ocs_aes_ll(aes_dev, dst_dma_list); + aes_a_dma_active_src_dst_ll_en(aes_dev); + /* + * Set the LAST GCX bit in AES_ACTIVE Register; this allows the + * AES/SM4 engine to differentiate between encrypted data and + * encrypted MAC. + */ + aes_a_set_last_gcx(aes_dev); + /* + * Enable DMA DONE interrupt; once DMA transfer is over, + * interrupt handler will process the MAC/tag. + */ + return ocs_aes_irq_enable_and_wait(aes_dev, AES_DMA_SRC_DONE_INT); +} + +/* + * Compare Tag to Yr. + * + * Only used at the end of CCM decrypt. If tag == yr, message authentication + * has succeeded. + */ +static inline int ccm_compare_tag_to_yr(struct ocs_aes_dev *aes_dev, + u8 tag_size_bytes) +{ + u32 tag[AES_MAX_TAG_SIZE_U32]; + u32 yr[AES_MAX_TAG_SIZE_U32]; + u8 i; + + /* Read Tag and Yr from AES registers. */ + for (i = 0; i < AES_MAX_TAG_SIZE_U32; i++) { + tag[i] = ioread32(aes_dev->base_reg + + AES_T_MAC_0_OFFSET + (i * sizeof(u32))); + yr[i] = ioread32(aes_dev->base_reg + + AES_MULTIPURPOSE2_0_OFFSET + + (i * sizeof(u32))); + } + + return memcmp(tag, yr, tag_size_bytes) ? -EBADMSG : 0; +} + +/** + * ocs_aes_ccm_op() - Perform CCM operation. + * @aes_dev: The OCS AES device to use. + * @cipher: The Cipher to use (AES or SM4). + * @instruction: The instruction to perform (encrypt or decrypt). + * @dst_dma_list: The OCS DMA list mapping output memory. + * @src_dma_list: The OCS DMA list mapping input payload data. + * @src_size: The amount of data mapped by @src_dma_list. + * @iv: The input IV vector. + * @adata_dma_list: The OCS DMA list mapping input A-data. + * @adata_size: The amount of data mapped by @adata_dma_list. + * @in_tag: Input tag. + * @tag_size: The size (in bytes) of @in_tag. + * + * Note: for encrypt the tag is appended to the ciphertext (in the memory + * mapped by @dst_dma_list). + * + * Return: 0 on success, negative error code otherwise. + */ +int ocs_aes_ccm_op(struct ocs_aes_dev *aes_dev, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size, + u8 *iv, + dma_addr_t adata_dma_list, + u32 adata_size, + u8 *in_tag, + u32 tag_size) +{ + u32 *iv_32; + u8 lprime; + int rc; + + rc = ocs_aes_validate_inputs(src_dma_list, src_size, iv, + AES_BLOCK_SIZE, adata_dma_list, adata_size, + in_tag, tag_size, cipher, OCS_MODE_CCM, + instruction, dst_dma_list); + if (rc) + return rc; + + ocs_aes_init(aes_dev, OCS_MODE_CCM, cipher, instruction); + + /* + * Note: rfc 3610 and NIST 800-38C require counter of zero to encrypt + * auth tag so ensure this is the case + */ + lprime = iv[L_PRIME_IDX]; + memset(&iv[COUNTER_START(lprime)], 0, COUNTER_LEN(lprime)); + + /* + * Nonce is already converted to ctr0 before being passed into this + * function as iv. + */ + iv_32 = (u32 *)iv; + iowrite32(__swab32(iv_32[0]), + aes_dev->base_reg + AES_MULTIPURPOSE1_3_OFFSET); + iowrite32(__swab32(iv_32[1]), + aes_dev->base_reg + AES_MULTIPURPOSE1_2_OFFSET); + iowrite32(__swab32(iv_32[2]), + aes_dev->base_reg + AES_MULTIPURPOSE1_1_OFFSET); + iowrite32(__swab32(iv_32[3]), + aes_dev->base_reg + AES_MULTIPURPOSE1_0_OFFSET); + + /* Write MAC/tag length in register AES_TLEN */ + iowrite32(tag_size, aes_dev->base_reg + AES_TLEN_OFFSET); + /* + * Write the byte length of the last AES/SM4 block of Payload data + * (without zero padding and without the length of the MAC) in register + * AES_PLEN. + */ + ocs_aes_write_last_data_blk_len(aes_dev, src_size); + + /* Set AES_ACTIVE.TRIGGER to start the operation. */ + aes_a_op_trigger(aes_dev); + + aes_a_dma_reset_and_activate_perf_cntr(aes_dev); + + /* Form block B0 and write it to the AES/SM4 input buffer. */ + rc = ocs_aes_ccm_write_b0(aes_dev, iv, adata_size, tag_size, src_size); + if (rc) + return rc; + /* + * Ensure there has been at least CCM_DECRYPT_DELAY_LAST_GCX_CLK_COUNT + * clock cycles since TRIGGER bit was set + */ + aes_a_dma_wait_and_deactivate_perf_cntr(aes_dev, + CCM_DECRYPT_DELAY_LAST_GCX_CLK_COUNT); + + /* Process Adata. */ + ocs_aes_ccm_do_adata(aes_dev, adata_dma_list, adata_size); + + /* For Encrypt case we just process the payload and return. */ + if (instruction == OCS_ENCRYPT) { + return ocs_aes_ccm_encrypt_do_payload(aes_dev, dst_dma_list, + src_dma_list, src_size); + } + /* For Decypt we need to process the payload and then the tag. */ + rc = ocs_aes_ccm_decrypt_do_payload(aes_dev, dst_dma_list, + src_dma_list, src_size); + if (rc) + return rc; + + /* Process MAC/tag directly: feed tag to engine and wait for IRQ. */ + ocs_aes_ccm_write_encrypted_tag(aes_dev, in_tag, tag_size); + rc = ocs_aes_irq_enable_and_wait(aes_dev, AES_COMPLETE_INT); + if (rc) + return rc; + + return ccm_compare_tag_to_yr(aes_dev, tag_size); +} + +/** + * ocs_create_linked_list_from_sg() - Create OCS DMA linked list from SG list. + * @aes_dev: The OCS AES device the list will be created for. + * @sg: The SG list OCS DMA linked list will be created from. When + * passed to this function, @sg must have been already mapped + * with dma_map_sg(). + * @sg_dma_count: The number of DMA-mapped entries in @sg. This must be the + * value returned by dma_map_sg() when @sg was mapped. + * @dll_desc: The OCS DMA dma_list to use to store information about the + * created linked list. + * @data_size: The size of the data (from the SG list) to be mapped into the + * OCS DMA linked list. + * @data_offset: The offset (within the SG list) of the data to be mapped. + * + * Return: 0 on success, negative error code otherwise. + */ +int ocs_create_linked_list_from_sg(const struct ocs_aes_dev *aes_dev, + struct scatterlist *sg, + int sg_dma_count, + struct ocs_dll_desc *dll_desc, + size_t data_size, size_t data_offset) +{ + struct ocs_dma_linked_list *ll = NULL; + struct scatterlist *sg_tmp; + unsigned int tmp; + int dma_nents; + int i; + + if (!dll_desc || !sg || !aes_dev) + return -EINVAL; + + /* Default values for when no ddl_desc is created. */ + dll_desc->vaddr = NULL; + dll_desc->dma_addr = DMA_MAPPING_ERROR; + dll_desc->size = 0; + + if (data_size == 0) + return 0; + + /* Loop over sg_list until we reach entry at specified offset. */ + while (data_offset >= sg_dma_len(sg)) { + data_offset -= sg_dma_len(sg); + sg_dma_count--; + sg = sg_next(sg); + /* If we reach the end of the list, offset was invalid. */ + if (!sg || sg_dma_count == 0) + return -EINVAL; + } + + /* Compute number of DMA-mapped SG entries to add into OCS DMA list. */ + dma_nents = 0; + tmp = 0; + sg_tmp = sg; + while (tmp < data_offset + data_size) { + /* If we reach the end of the list, data_size was invalid. */ + if (!sg_tmp) + return -EINVAL; + tmp += sg_dma_len(sg_tmp); + dma_nents++; + sg_tmp = sg_next(sg_tmp); + } + if (dma_nents > sg_dma_count) + return -EINVAL; + + /* Allocate the DMA list, one entry for each SG entry. */ + dll_desc->size = sizeof(struct ocs_dma_linked_list) * dma_nents; + dll_desc->vaddr = dma_alloc_coherent(aes_dev->dev, dll_desc->size, + &dll_desc->dma_addr, GFP_KERNEL); + if (!dll_desc->vaddr) + return -ENOMEM; + + /* Populate DMA linked list entries. */ + ll = dll_desc->vaddr; + for (i = 0; i < dma_nents; i++, sg = sg_next(sg)) { + ll[i].src_addr = sg_dma_address(sg) + data_offset; + ll[i].src_len = min(sg_dma_len(sg) - data_offset, data_size); + data_offset = 0; + data_size -= ll[i].src_len; + /* Current element points to the DMA address of the next one. */ + ll[i].next = dll_desc->dma_addr + (sizeof(*ll) * (i + 1)); + ll[i].ll_flags = 0; + } + /* Terminate last element. */ + ll[i - 1].next = 0; + ll[i - 1].ll_flags = OCS_LL_DMA_FLAG_TERMINATE; + + return 0; +} diff --git a/drivers/crypto/intel/keembay/ocs-aes.h b/drivers/crypto/intel/keembay/ocs-aes.h new file mode 100644 index 000000000000..c035fc48b7ed --- /dev/null +++ b/drivers/crypto/intel/keembay/ocs-aes.h @@ -0,0 +1,129 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * Intel Keem Bay OCS AES Crypto Driver. + * + * Copyright (C) 2018-2020 Intel Corporation + */ + +#ifndef _CRYPTO_OCS_AES_H +#define _CRYPTO_OCS_AES_H + +#include <linux/dma-mapping.h> + +enum ocs_cipher { + OCS_AES = 0, + OCS_SM4 = 1, +}; + +enum ocs_mode { + OCS_MODE_ECB = 0, + OCS_MODE_CBC = 1, + OCS_MODE_CTR = 2, + OCS_MODE_CCM = 6, + OCS_MODE_GCM = 7, + OCS_MODE_CTS = 9, +}; + +enum ocs_instruction { + OCS_ENCRYPT = 0, + OCS_DECRYPT = 1, + OCS_EXPAND = 2, + OCS_BYPASS = 3, +}; + +/** + * struct ocs_aes_dev - AES device context. + * @list: List head for insertion into device list hold + * by driver. + * @dev: OCS AES device. + * @irq: IRQ number. + * @base_reg: IO base address of OCS AES. + * @irq_copy_completion: Completion to indicate IRQ has been triggered. + * @dma_err_mask: Error reported by OCS DMA interrupts. + * @engine: Crypto engine for the device. + */ +struct ocs_aes_dev { + struct list_head list; + struct device *dev; + int irq; + void __iomem *base_reg; + struct completion irq_completion; + u32 dma_err_mask; + struct crypto_engine *engine; +}; + +/** + * struct ocs_dll_desc - Descriptor of an OCS DMA Linked List. + * @vaddr: Virtual address of the linked list head. + * @dma_addr: DMA address of the linked list head. + * @size: Size (in bytes) of the linked list. + */ +struct ocs_dll_desc { + void *vaddr; + dma_addr_t dma_addr; + size_t size; +}; + +int ocs_aes_set_key(struct ocs_aes_dev *aes_dev, const u32 key_size, + const u8 *key, const enum ocs_cipher cipher); + +int ocs_aes_op(struct ocs_aes_dev *aes_dev, + enum ocs_mode mode, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size, + u8 *iv, + u32 iv_size); + +/** + * ocs_aes_bypass_op() - Use OCS DMA to copy data. + * @aes_dev: The OCS AES device to use. + * @dst_dma_list: The OCS DMA list mapping the memory where input data + * will be copied to. + * @src_dma_list: The OCS DMA list mapping input data. + * @src_size: The amount of data to copy. + */ +static inline int ocs_aes_bypass_op(struct ocs_aes_dev *aes_dev, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, u32 src_size) +{ + return ocs_aes_op(aes_dev, OCS_MODE_ECB, OCS_AES, OCS_BYPASS, + dst_dma_list, src_dma_list, src_size, NULL, 0); +} + +int ocs_aes_gcm_op(struct ocs_aes_dev *aes_dev, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size, + const u8 *iv, + dma_addr_t aad_dma_list, + u32 aad_size, + u8 *out_tag, + u32 tag_size); + +int ocs_aes_ccm_op(struct ocs_aes_dev *aes_dev, + enum ocs_cipher cipher, + enum ocs_instruction instruction, + dma_addr_t dst_dma_list, + dma_addr_t src_dma_list, + u32 src_size, + u8 *iv, + dma_addr_t adata_dma_list, + u32 adata_size, + u8 *in_tag, + u32 tag_size); + +int ocs_create_linked_list_from_sg(const struct ocs_aes_dev *aes_dev, + struct scatterlist *sg, + int sg_dma_count, + struct ocs_dll_desc *dll_desc, + size_t data_size, + size_t data_offset); + +irqreturn_t ocs_aes_irq_handler(int irq, void *dev_id); + +#endif diff --git a/drivers/crypto/intel/keembay/ocs-hcu.c b/drivers/crypto/intel/keembay/ocs-hcu.c new file mode 100644 index 000000000000..55a41e6ab103 --- /dev/null +++ b/drivers/crypto/intel/keembay/ocs-hcu.c @@ -0,0 +1,841 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Intel Keem Bay OCS HCU Crypto Driver. + * + * Copyright (C) 2018-2020 Intel Corporation + */ + +#include <linux/delay.h> +#include <linux/device.h> +#include <linux/iopoll.h> +#include <linux/irq.h> +#include <linux/module.h> + +#include <crypto/sha2.h> + +#include "ocs-hcu.h" + +/* Registers. */ +#define OCS_HCU_MODE 0x00 +#define OCS_HCU_CHAIN 0x04 +#define OCS_HCU_OPERATION 0x08 +#define OCS_HCU_KEY_0 0x0C +#define OCS_HCU_ISR 0x50 +#define OCS_HCU_IER 0x54 +#define OCS_HCU_STATUS 0x58 +#define OCS_HCU_MSG_LEN_LO 0x60 +#define OCS_HCU_MSG_LEN_HI 0x64 +#define OCS_HCU_KEY_BYTE_ORDER_CFG 0x80 +#define OCS_HCU_DMA_SRC_ADDR 0x400 +#define OCS_HCU_DMA_SRC_SIZE 0x408 +#define OCS_HCU_DMA_DST_SIZE 0x40C +#define OCS_HCU_DMA_DMA_MODE 0x410 +#define OCS_HCU_DMA_NEXT_SRC_DESCR 0x418 +#define OCS_HCU_DMA_MSI_ISR 0x480 +#define OCS_HCU_DMA_MSI_IER 0x484 +#define OCS_HCU_DMA_MSI_MASK 0x488 + +/* Register bit definitions. */ +#define HCU_MODE_ALGO_SHIFT 16 +#define HCU_MODE_HMAC_SHIFT 22 + +#define HCU_STATUS_BUSY BIT(0) + +#define HCU_BYTE_ORDER_SWAP BIT(0) + +#define HCU_IRQ_HASH_DONE BIT(2) +#define HCU_IRQ_HASH_ERR_MASK (BIT(3) | BIT(1) | BIT(0)) + +#define HCU_DMA_IRQ_SRC_DONE BIT(0) +#define HCU_DMA_IRQ_SAI_ERR BIT(2) +#define HCU_DMA_IRQ_BAD_COMP_ERR BIT(3) +#define HCU_DMA_IRQ_INBUF_RD_ERR BIT(4) +#define HCU_DMA_IRQ_INBUF_WD_ERR BIT(5) +#define HCU_DMA_IRQ_OUTBUF_WR_ERR BIT(6) +#define HCU_DMA_IRQ_OUTBUF_RD_ERR BIT(7) +#define HCU_DMA_IRQ_CRD_ERR BIT(8) +#define HCU_DMA_IRQ_ERR_MASK (HCU_DMA_IRQ_SAI_ERR | \ + HCU_DMA_IRQ_BAD_COMP_ERR | \ + HCU_DMA_IRQ_INBUF_RD_ERR | \ + HCU_DMA_IRQ_INBUF_WD_ERR | \ + HCU_DMA_IRQ_OUTBUF_WR_ERR | \ + HCU_DMA_IRQ_OUTBUF_RD_ERR | \ + HCU_DMA_IRQ_CRD_ERR) + +#define HCU_DMA_SNOOP_MASK (0x7 << 28) +#define HCU_DMA_SRC_LL_EN BIT(25) +#define HCU_DMA_EN BIT(31) + +#define OCS_HCU_ENDIANNESS_VALUE 0x2A + +#define HCU_DMA_MSI_UNMASK BIT(0) +#define HCU_DMA_MSI_DISABLE 0 +#define HCU_IRQ_DISABLE 0 + +#define OCS_HCU_START BIT(0) +#define OCS_HCU_TERMINATE BIT(1) + +#define OCS_LL_DMA_FLAG_TERMINATE BIT(31) + +#define OCS_HCU_HW_KEY_LEN_U32 (OCS_HCU_HW_KEY_LEN / sizeof(u32)) + +#define HCU_DATA_WRITE_ENDIANNESS_OFFSET 26 + +#define OCS_HCU_NUM_CHAINS_SHA256_224_SM3 (SHA256_DIGEST_SIZE / sizeof(u32)) +#define OCS_HCU_NUM_CHAINS_SHA384_512 (SHA512_DIGEST_SIZE / sizeof(u32)) + +/* + * While polling on a busy HCU, wait maximum 200us between one check and the + * other. + */ +#define OCS_HCU_WAIT_BUSY_RETRY_DELAY_US 200 +/* Wait on a busy HCU for maximum 1 second. */ +#define OCS_HCU_WAIT_BUSY_TIMEOUT_US 1000000 + +/** + * struct ocs_hcu_dma_entry - An entry in an OCS DMA linked list. + * @src_addr: Source address of the data. + * @src_len: Length of data to be fetched. + * @nxt_desc: Next descriptor to fetch. + * @ll_flags: Flags (Freeze @ terminate) for the DMA engine. + */ +struct ocs_hcu_dma_entry { + u32 src_addr; + u32 src_len; + u32 nxt_desc; + u32 ll_flags; +}; + +/** + * struct ocs_hcu_dma_list - OCS-specific DMA linked list. + * @head: The head of the list (points to the array backing the list). + * @tail: The current tail of the list; NULL if the list is empty. + * @dma_addr: The DMA address of @head (i.e., the DMA address of the backing + * array). + * @max_nents: Maximum number of entries in the list (i.e., number of elements + * in the backing array). + * + * The OCS DMA list is an array-backed list of OCS DMA descriptors. The array + * backing the list is allocated with dma_alloc_coherent() and pointed by + * @head. + */ +struct ocs_hcu_dma_list { + struct ocs_hcu_dma_entry *head; + struct ocs_hcu_dma_entry *tail; + dma_addr_t dma_addr; + size_t max_nents; +}; + +static inline u32 ocs_hcu_num_chains(enum ocs_hcu_algo algo) +{ + switch (algo) { + case OCS_HCU_ALGO_SHA224: + case OCS_HCU_ALGO_SHA256: + case OCS_HCU_ALGO_SM3: + return OCS_HCU_NUM_CHAINS_SHA256_224_SM3; + case OCS_HCU_ALGO_SHA384: + case OCS_HCU_ALGO_SHA512: + return OCS_HCU_NUM_CHAINS_SHA384_512; + default: + return 0; + }; +} + +static inline u32 ocs_hcu_digest_size(enum ocs_hcu_algo algo) +{ + switch (algo) { + case OCS_HCU_ALGO_SHA224: + return SHA224_DIGEST_SIZE; + case OCS_HCU_ALGO_SHA256: + case OCS_HCU_ALGO_SM3: + /* SM3 shares the same block size. */ + return SHA256_DIGEST_SIZE; + case OCS_HCU_ALGO_SHA384: + return SHA384_DIGEST_SIZE; + case OCS_HCU_ALGO_SHA512: + return SHA512_DIGEST_SIZE; + default: + return 0; + } +} + +/** + * ocs_hcu_wait_busy() - Wait for HCU OCS hardware to became usable. + * @hcu_dev: OCS HCU device to wait for. + * + * Return: 0 if device free, -ETIMEOUT if device busy and internal timeout has + * expired. + */ +static int ocs_hcu_wait_busy(struct ocs_hcu_dev *hcu_dev) +{ + long val; + + return readl_poll_timeout(hcu_dev->io_base + OCS_HCU_STATUS, val, + !(val & HCU_STATUS_BUSY), + OCS_HCU_WAIT_BUSY_RETRY_DELAY_US, + OCS_HCU_WAIT_BUSY_TIMEOUT_US); +} + +static void ocs_hcu_done_irq_en(struct ocs_hcu_dev *hcu_dev) +{ + /* Clear any pending interrupts. */ + writel(0xFFFFFFFF, hcu_dev->io_base + OCS_HCU_ISR); + hcu_dev->irq_err = false; + /* Enable error and HCU done interrupts. */ + writel(HCU_IRQ_HASH_DONE | HCU_IRQ_HASH_ERR_MASK, + hcu_dev->io_base + OCS_HCU_IER); +} + +static void ocs_hcu_dma_irq_en(struct ocs_hcu_dev *hcu_dev) +{ + /* Clear any pending interrupts. */ + writel(0xFFFFFFFF, hcu_dev->io_base + OCS_HCU_DMA_MSI_ISR); + hcu_dev->irq_err = false; + /* Only operating on DMA source completion and error interrupts. */ + writel(HCU_DMA_IRQ_ERR_MASK | HCU_DMA_IRQ_SRC_DONE, + hcu_dev->io_base + OCS_HCU_DMA_MSI_IER); + /* Unmask */ + writel(HCU_DMA_MSI_UNMASK, hcu_dev->io_base + OCS_HCU_DMA_MSI_MASK); +} + +static void ocs_hcu_irq_dis(struct ocs_hcu_dev *hcu_dev) +{ + writel(HCU_IRQ_DISABLE, hcu_dev->io_base + OCS_HCU_IER); + writel(HCU_DMA_MSI_DISABLE, hcu_dev->io_base + OCS_HCU_DMA_MSI_IER); +} + +static int ocs_hcu_wait_and_disable_irq(struct ocs_hcu_dev *hcu_dev) +{ + int rc; + + rc = wait_for_completion_interruptible(&hcu_dev->irq_done); + if (rc) + goto exit; + + if (hcu_dev->irq_err) { + /* Unset flag and return error. */ + hcu_dev->irq_err = false; + rc = -EIO; + goto exit; + } + +exit: + ocs_hcu_irq_dis(hcu_dev); + + return rc; +} + +/** + * ocs_hcu_get_intermediate_data() - Get intermediate data. + * @hcu_dev: The target HCU device. + * @data: Where to store the intermediate. + * @algo: The algorithm being used. + * + * This function is used to save the current hashing process state in order to + * continue it in the future. + * + * Note: once all data has been processed, the intermediate data actually + * contains the hashing result. So this function is also used to retrieve the + * final result of a hashing process. + * + * Return: 0 on success, negative error code otherwise. + */ +static int ocs_hcu_get_intermediate_data(struct ocs_hcu_dev *hcu_dev, + struct ocs_hcu_idata *data, + enum ocs_hcu_algo algo) +{ + const int n = ocs_hcu_num_chains(algo); + u32 *chain; + int rc; + int i; + + /* Data not requested. */ + if (!data) + return -EINVAL; + + chain = (u32 *)data->digest; + + /* Ensure that the OCS is no longer busy before reading the chains. */ + rc = ocs_hcu_wait_busy(hcu_dev); + if (rc) + return rc; + + /* + * This loops is safe because data->digest is an array of + * SHA512_DIGEST_SIZE bytes and the maximum value returned by + * ocs_hcu_num_chains() is OCS_HCU_NUM_CHAINS_SHA384_512 which is equal + * to SHA512_DIGEST_SIZE / sizeof(u32). + */ + for (i = 0; i < n; i++) + chain[i] = readl(hcu_dev->io_base + OCS_HCU_CHAIN); + + data->msg_len_lo = readl(hcu_dev->io_base + OCS_HCU_MSG_LEN_LO); + data->msg_len_hi = readl(hcu_dev->io_base + OCS_HCU_MSG_LEN_HI); + + return 0; +} + +/** + * ocs_hcu_set_intermediate_data() - Set intermediate data. + * @hcu_dev: The target HCU device. + * @data: The intermediate data to be set. + * @algo: The algorithm being used. + * + * This function is used to continue a previous hashing process. + */ +static void ocs_hcu_set_intermediate_data(struct ocs_hcu_dev *hcu_dev, + const struct ocs_hcu_idata *data, + enum ocs_hcu_algo algo) +{ + const int n = ocs_hcu_num_chains(algo); + u32 *chain = (u32 *)data->digest; + int i; + + /* + * This loops is safe because data->digest is an array of + * SHA512_DIGEST_SIZE bytes and the maximum value returned by + * ocs_hcu_num_chains() is OCS_HCU_NUM_CHAINS_SHA384_512 which is equal + * to SHA512_DIGEST_SIZE / sizeof(u32). + */ + for (i = 0; i < n; i++) + writel(chain[i], hcu_dev->io_base + OCS_HCU_CHAIN); + + writel(data->msg_len_lo, hcu_dev->io_base + OCS_HCU_MSG_LEN_LO); + writel(data->msg_len_hi, hcu_dev->io_base + OCS_HCU_MSG_LEN_HI); +} + +static int ocs_hcu_get_digest(struct ocs_hcu_dev *hcu_dev, + enum ocs_hcu_algo algo, u8 *dgst, size_t dgst_len) +{ + u32 *chain; + int rc; + int i; + + if (!dgst) + return -EINVAL; + + /* Length of the output buffer must match the algo digest size. */ + if (dgst_len != ocs_hcu_digest_size(algo)) + return -EINVAL; + + /* Ensure that the OCS is no longer busy before reading the chains. */ + rc = ocs_hcu_wait_busy(hcu_dev); + if (rc) + return rc; + + chain = (u32 *)dgst; + for (i = 0; i < dgst_len / sizeof(u32); i++) + chain[i] = readl(hcu_dev->io_base + OCS_HCU_CHAIN); + + return 0; +} + +/** + * ocs_hcu_hw_cfg() - Configure the HCU hardware. + * @hcu_dev: The HCU device to configure. + * @algo: The algorithm to be used by the HCU device. + * @use_hmac: Whether or not HW HMAC should be used. + * + * Return: 0 on success, negative error code otherwise. + */ +static int ocs_hcu_hw_cfg(struct ocs_hcu_dev *hcu_dev, enum ocs_hcu_algo algo, + bool use_hmac) +{ + u32 cfg; + int rc; + + if (algo != OCS_HCU_ALGO_SHA256 && algo != OCS_HCU_ALGO_SHA224 && + algo != OCS_HCU_ALGO_SHA384 && algo != OCS_HCU_ALGO_SHA512 && + algo != OCS_HCU_ALGO_SM3) + return -EINVAL; + + rc = ocs_hcu_wait_busy(hcu_dev); + if (rc) + return rc; + + /* Ensure interrupts are disabled. */ + ocs_hcu_irq_dis(hcu_dev); + + /* Configure endianness, hashing algorithm and HW HMAC (if needed) */ + cfg = OCS_HCU_ENDIANNESS_VALUE << HCU_DATA_WRITE_ENDIANNESS_OFFSET; + cfg |= algo << HCU_MODE_ALGO_SHIFT; + if (use_hmac) + cfg |= BIT(HCU_MODE_HMAC_SHIFT); + + writel(cfg, hcu_dev->io_base + OCS_HCU_MODE); + + return 0; +} + +/** + * ocs_hcu_clear_key() - Clear key stored in OCS HMAC KEY registers. + * @hcu_dev: The OCS HCU device whose key registers should be cleared. + */ +static void ocs_hcu_clear_key(struct ocs_hcu_dev *hcu_dev) +{ + int reg_off; + + /* Clear OCS_HCU_KEY_[0..15] */ + for (reg_off = 0; reg_off < OCS_HCU_HW_KEY_LEN; reg_off += sizeof(u32)) + writel(0, hcu_dev->io_base + OCS_HCU_KEY_0 + reg_off); +} + +/** + * ocs_hcu_write_key() - Write key to OCS HMAC KEY registers. + * @hcu_dev: The OCS HCU device the key should be written to. + * @key: The key to be written. + * @len: The size of the key to write. It must be OCS_HCU_HW_KEY_LEN. + * + * Return: 0 on success, negative error code otherwise. + */ +static int ocs_hcu_write_key(struct ocs_hcu_dev *hcu_dev, const u8 *key, size_t len) +{ + u32 key_u32[OCS_HCU_HW_KEY_LEN_U32]; + int i; + + if (len > OCS_HCU_HW_KEY_LEN) + return -EINVAL; + + /* Copy key into temporary u32 array. */ + memcpy(key_u32, key, len); + + /* + * Hardware requires all the bytes of the HW Key vector to be + * written. So pad with zero until we reach OCS_HCU_HW_KEY_LEN. + */ + memzero_explicit((u8 *)key_u32 + len, OCS_HCU_HW_KEY_LEN - len); + + /* + * OCS hardware expects the MSB of the key to be written at the highest + * address of the HCU Key vector; in other word, the key must be + * written in reverse order. + * + * Therefore, we first enable byte swapping for the HCU key vector; + * so that bytes of 32-bit word written to OCS_HCU_KEY_[0..15] will be + * swapped: + * 3 <---> 0, 2 <---> 1. + */ + writel(HCU_BYTE_ORDER_SWAP, + hcu_dev->io_base + OCS_HCU_KEY_BYTE_ORDER_CFG); + /* + * And then we write the 32-bit words composing the key starting from + * the end of the key. + */ + for (i = 0; i < OCS_HCU_HW_KEY_LEN_U32; i++) + writel(key_u32[OCS_HCU_HW_KEY_LEN_U32 - 1 - i], + hcu_dev->io_base + OCS_HCU_KEY_0 + (sizeof(u32) * i)); + + memzero_explicit(key_u32, OCS_HCU_HW_KEY_LEN); + + return 0; +} + +/** + * ocs_hcu_ll_dma_start() - Start OCS HCU hashing via DMA + * @hcu_dev: The OCS HCU device to use. + * @dma_list: The OCS DMA list mapping the data to hash. + * @finalize: Whether or not this is the last hashing operation and therefore + * the final hash should be compute even if data is not + * block-aligned. + * + * Return: 0 on success, negative error code otherwise. + */ +static int ocs_hcu_ll_dma_start(struct ocs_hcu_dev *hcu_dev, + const struct ocs_hcu_dma_list *dma_list, + bool finalize) +{ + u32 cfg = HCU_DMA_SNOOP_MASK | HCU_DMA_SRC_LL_EN | HCU_DMA_EN; + int rc; + + if (!dma_list) + return -EINVAL; + + /* + * For final requests we use HCU_DONE IRQ to be notified when all input + * data has been processed by the HCU; however, we cannot do so for + * non-final requests, because we don't get a HCU_DONE IRQ when we + * don't terminate the operation. + * + * Therefore, for non-final requests, we use the DMA IRQ, which + * triggers when DMA has finishing feeding all the input data to the + * HCU, but the HCU may still be processing it. This is fine, since we + * will wait for the HCU processing to be completed when we try to read + * intermediate results, in ocs_hcu_get_intermediate_data(). + */ + if (finalize) + ocs_hcu_done_irq_en(hcu_dev); + else + ocs_hcu_dma_irq_en(hcu_dev); + + reinit_completion(&hcu_dev->irq_done); + writel(dma_list->dma_addr, hcu_dev->io_base + OCS_HCU_DMA_NEXT_SRC_DESCR); + writel(0, hcu_dev->io_base + OCS_HCU_DMA_SRC_SIZE); + writel(0, hcu_dev->io_base + OCS_HCU_DMA_DST_SIZE); + + writel(OCS_HCU_START, hcu_dev->io_base + OCS_HCU_OPERATION); + + writel(cfg, hcu_dev->io_base + OCS_HCU_DMA_DMA_MODE); + + if (finalize) + writel(OCS_HCU_TERMINATE, hcu_dev->io_base + OCS_HCU_OPERATION); + + rc = ocs_hcu_wait_and_disable_irq(hcu_dev); + if (rc) + return rc; + + return 0; +} + +struct ocs_hcu_dma_list *ocs_hcu_dma_list_alloc(struct ocs_hcu_dev *hcu_dev, + int max_nents) +{ + struct ocs_hcu_dma_list *dma_list; + + dma_list = kmalloc(sizeof(*dma_list), GFP_KERNEL); + if (!dma_list) + return NULL; + + /* Total size of the DMA list to allocate. */ + dma_list->head = dma_alloc_coherent(hcu_dev->dev, + sizeof(*dma_list->head) * max_nents, + &dma_list->dma_addr, GFP_KERNEL); + if (!dma_list->head) { + kfree(dma_list); + return NULL; + } + dma_list->max_nents = max_nents; + dma_list->tail = NULL; + + return dma_list; +} + +void ocs_hcu_dma_list_free(struct ocs_hcu_dev *hcu_dev, + struct ocs_hcu_dma_list *dma_list) +{ + if (!dma_list) + return; + + dma_free_coherent(hcu_dev->dev, + sizeof(*dma_list->head) * dma_list->max_nents, + dma_list->head, dma_list->dma_addr); + + kfree(dma_list); +} + +/* Add a new DMA entry at the end of the OCS DMA list. */ +int ocs_hcu_dma_list_add_tail(struct ocs_hcu_dev *hcu_dev, + struct ocs_hcu_dma_list *dma_list, + dma_addr_t addr, u32 len) +{ + struct device *dev = hcu_dev->dev; + struct ocs_hcu_dma_entry *old_tail; + struct ocs_hcu_dma_entry *new_tail; + + if (!len) + return 0; + + if (!dma_list) + return -EINVAL; + + if (addr & ~OCS_HCU_DMA_BIT_MASK) { + dev_err(dev, + "Unexpected error: Invalid DMA address for OCS HCU\n"); + return -EINVAL; + } + + old_tail = dma_list->tail; + new_tail = old_tail ? old_tail + 1 : dma_list->head; + + /* Check if list is full. */ + if (new_tail - dma_list->head >= dma_list->max_nents) + return -ENOMEM; + + /* + * If there was an old tail (i.e., this is not the first element we are + * adding), un-terminate the old tail and make it point to the new one. + */ + if (old_tail) { + old_tail->ll_flags &= ~OCS_LL_DMA_FLAG_TERMINATE; + /* + * The old tail 'nxt_desc' must point to the DMA address of the + * new tail. + */ + old_tail->nxt_desc = dma_list->dma_addr + + sizeof(*dma_list->tail) * (new_tail - + dma_list->head); + } + + new_tail->src_addr = (u32)addr; + new_tail->src_len = (u32)len; + new_tail->ll_flags = OCS_LL_DMA_FLAG_TERMINATE; + new_tail->nxt_desc = 0; + + /* Update list tail with new tail. */ + dma_list->tail = new_tail; + + return 0; +} + +/** + * ocs_hcu_hash_init() - Initialize hash operation context. + * @ctx: The context to initialize. + * @algo: The hashing algorithm to use. + * + * Return: 0 on success, negative error code otherwise. + */ +int ocs_hcu_hash_init(struct ocs_hcu_hash_ctx *ctx, enum ocs_hcu_algo algo) +{ + if (!ctx) + return -EINVAL; + + ctx->algo = algo; + ctx->idata.msg_len_lo = 0; + ctx->idata.msg_len_hi = 0; + /* No need to set idata.digest to 0. */ + + return 0; +} + +/** + * ocs_hcu_hash_update() - Perform a hashing iteration. + * @hcu_dev: The OCS HCU device to use. + * @ctx: The OCS HCU hashing context. + * @dma_list: The OCS DMA list mapping the input data to process. + * + * Return: 0 on success; negative error code otherwise. + */ +int ocs_hcu_hash_update(struct ocs_hcu_dev *hcu_dev, + struct ocs_hcu_hash_ctx *ctx, + const struct ocs_hcu_dma_list *dma_list) +{ + int rc; + + if (!hcu_dev || !ctx) + return -EINVAL; + + /* Configure the hardware for the current request. */ + rc = ocs_hcu_hw_cfg(hcu_dev, ctx->algo, false); + if (rc) + return rc; + + /* If we already processed some data, idata needs to be set. */ + if (ctx->idata.msg_len_lo || ctx->idata.msg_len_hi) + ocs_hcu_set_intermediate_data(hcu_dev, &ctx->idata, ctx->algo); + + /* Start linked-list DMA hashing. */ + rc = ocs_hcu_ll_dma_start(hcu_dev, dma_list, false); + if (rc) + return rc; + + /* Update idata and return. */ + return ocs_hcu_get_intermediate_data(hcu_dev, &ctx->idata, ctx->algo); +} + +/** + * ocs_hcu_hash_finup() - Update and finalize hash computation. + * @hcu_dev: The OCS HCU device to use. + * @ctx: The OCS HCU hashing context. + * @dma_list: The OCS DMA list mapping the input data to process. + * @dgst: The buffer where to save the computed digest. + * @dgst_len: The length of @dgst. + * + * Return: 0 on success; negative error code otherwise. + */ +int ocs_hcu_hash_finup(struct ocs_hcu_dev *hcu_dev, + const struct ocs_hcu_hash_ctx *ctx, + const struct ocs_hcu_dma_list *dma_list, + u8 *dgst, size_t dgst_len) +{ + int rc; + + if (!hcu_dev || !ctx) + return -EINVAL; + + /* Configure the hardware for the current request. */ + rc = ocs_hcu_hw_cfg(hcu_dev, ctx->algo, false); + if (rc) + return rc; + + /* If we already processed some data, idata needs to be set. */ + if (ctx->idata.msg_len_lo || ctx->idata.msg_len_hi) + ocs_hcu_set_intermediate_data(hcu_dev, &ctx->idata, ctx->algo); + + /* Start linked-list DMA hashing. */ + rc = ocs_hcu_ll_dma_start(hcu_dev, dma_list, true); + if (rc) + return rc; + + /* Get digest and return. */ + return ocs_hcu_get_digest(hcu_dev, ctx->algo, dgst, dgst_len); +} + +/** + * ocs_hcu_hash_final() - Finalize hash computation. + * @hcu_dev: The OCS HCU device to use. + * @ctx: The OCS HCU hashing context. + * @dgst: The buffer where to save the computed digest. + * @dgst_len: The length of @dgst. + * + * Return: 0 on success; negative error code otherwise. + */ +int ocs_hcu_hash_final(struct ocs_hcu_dev *hcu_dev, + const struct ocs_hcu_hash_ctx *ctx, u8 *dgst, + size_t dgst_len) +{ + int rc; + + if (!hcu_dev || !ctx) + return -EINVAL; + + /* Configure the hardware for the current request. */ + rc = ocs_hcu_hw_cfg(hcu_dev, ctx->algo, false); + if (rc) + return rc; + + /* If we already processed some data, idata needs to be set. */ + if (ctx->idata.msg_len_lo || ctx->idata.msg_len_hi) + ocs_hcu_set_intermediate_data(hcu_dev, &ctx->idata, ctx->algo); + + /* + * Enable HCU interrupts, so that HCU_DONE will be triggered once the + * final hash is computed. + */ + ocs_hcu_done_irq_en(hcu_dev); + reinit_completion(&hcu_dev->irq_done); + writel(OCS_HCU_TERMINATE, hcu_dev->io_base + OCS_HCU_OPERATION); + + rc = ocs_hcu_wait_and_disable_irq(hcu_dev); + if (rc) + return rc; + + /* Get digest and return. */ + return ocs_hcu_get_digest(hcu_dev, ctx->algo, dgst, dgst_len); +} + +/** + * ocs_hcu_digest() - Compute hash digest. + * @hcu_dev: The OCS HCU device to use. + * @algo: The hash algorithm to use. + * @data: The input data to process. + * @data_len: The length of @data. + * @dgst: The buffer where to save the computed digest. + * @dgst_len: The length of @dgst. + * + * Return: 0 on success; negative error code otherwise. + */ +int ocs_hcu_digest(struct ocs_hcu_dev *hcu_dev, enum ocs_hcu_algo algo, + void *data, size_t data_len, u8 *dgst, size_t dgst_len) +{ + struct device *dev = hcu_dev->dev; + dma_addr_t dma_handle; + u32 reg; + int rc; + + /* Configure the hardware for the current request. */ + rc = ocs_hcu_hw_cfg(hcu_dev, algo, false); + if (rc) + return rc; + + dma_handle = dma_map_single(dev, data, data_len, DMA_TO_DEVICE); + if (dma_mapping_error(dev, dma_handle)) + return -EIO; + + reg = HCU_DMA_SNOOP_MASK | HCU_DMA_EN; + + ocs_hcu_done_irq_en(hcu_dev); + + reinit_completion(&hcu_dev->irq_done); + + writel(dma_handle, hcu_dev->io_base + OCS_HCU_DMA_SRC_ADDR); + writel(data_len, hcu_dev->io_base + OCS_HCU_DMA_SRC_SIZE); + writel(OCS_HCU_START, hcu_dev->io_base + OCS_HCU_OPERATION); + writel(reg, hcu_dev->io_base + OCS_HCU_DMA_DMA_MODE); + + writel(OCS_HCU_TERMINATE, hcu_dev->io_base + OCS_HCU_OPERATION); + + rc = ocs_hcu_wait_and_disable_irq(hcu_dev); + if (rc) + return rc; + + dma_unmap_single(dev, dma_handle, data_len, DMA_TO_DEVICE); + + return ocs_hcu_get_digest(hcu_dev, algo, dgst, dgst_len); +} + +/** + * ocs_hcu_hmac() - Compute HMAC. + * @hcu_dev: The OCS HCU device to use. + * @algo: The hash algorithm to use with HMAC. + * @key: The key to use. + * @dma_list: The OCS DMA list mapping the input data to process. + * @key_len: The length of @key. + * @dgst: The buffer where to save the computed HMAC. + * @dgst_len: The length of @dgst. + * + * Return: 0 on success; negative error code otherwise. + */ +int ocs_hcu_hmac(struct ocs_hcu_dev *hcu_dev, enum ocs_hcu_algo algo, + const u8 *key, size_t key_len, + const struct ocs_hcu_dma_list *dma_list, + u8 *dgst, size_t dgst_len) +{ + int rc; + + /* Ensure 'key' is not NULL. */ + if (!key || key_len == 0) + return -EINVAL; + + /* Configure the hardware for the current request. */ + rc = ocs_hcu_hw_cfg(hcu_dev, algo, true); + if (rc) + return rc; + + rc = ocs_hcu_write_key(hcu_dev, key, key_len); + if (rc) + return rc; + + rc = ocs_hcu_ll_dma_start(hcu_dev, dma_list, true); + + /* Clear HW key before processing return code. */ + ocs_hcu_clear_key(hcu_dev); + + if (rc) + return rc; + + return ocs_hcu_get_digest(hcu_dev, algo, dgst, dgst_len); +} + +irqreturn_t ocs_hcu_irq_handler(int irq, void *dev_id) +{ + struct ocs_hcu_dev *hcu_dev = dev_id; + u32 hcu_irq; + u32 dma_irq; + + /* Read and clear the HCU interrupt. */ + hcu_irq = readl(hcu_dev->io_base + OCS_HCU_ISR); + writel(hcu_irq, hcu_dev->io_base + OCS_HCU_ISR); + + /* Read and clear the HCU DMA interrupt. */ + dma_irq = readl(hcu_dev->io_base + OCS_HCU_DMA_MSI_ISR); + writel(dma_irq, hcu_dev->io_base + OCS_HCU_DMA_MSI_ISR); + + /* Check for errors. */ + if (hcu_irq & HCU_IRQ_HASH_ERR_MASK || dma_irq & HCU_DMA_IRQ_ERR_MASK) { + hcu_dev->irq_err = true; + goto complete; + } + + /* Check for DONE IRQs. */ + if (hcu_irq & HCU_IRQ_HASH_DONE || dma_irq & HCU_DMA_IRQ_SRC_DONE) + goto complete; + + return IRQ_NONE; + +complete: + complete(&hcu_dev->irq_done); + + return IRQ_HANDLED; +} + +MODULE_DESCRIPTION("Intel Keem Bay OCS HCU Crypto Driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/crypto/intel/keembay/ocs-hcu.h b/drivers/crypto/intel/keembay/ocs-hcu.h new file mode 100644 index 000000000000..fbbbb92a0592 --- /dev/null +++ b/drivers/crypto/intel/keembay/ocs-hcu.h @@ -0,0 +1,106 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * Intel Keem Bay OCS HCU Crypto Driver. + * + * Copyright (C) 2018-2020 Intel Corporation + */ + +#include <linux/dma-mapping.h> + +#ifndef _CRYPTO_OCS_HCU_H +#define _CRYPTO_OCS_HCU_H + +#define OCS_HCU_DMA_BIT_MASK DMA_BIT_MASK(32) + +#define OCS_HCU_HW_KEY_LEN 64 + +struct ocs_hcu_dma_list; + +enum ocs_hcu_algo { + OCS_HCU_ALGO_SHA256 = 2, + OCS_HCU_ALGO_SHA224 = 3, + OCS_HCU_ALGO_SHA384 = 4, + OCS_HCU_ALGO_SHA512 = 5, + OCS_HCU_ALGO_SM3 = 6, +}; + +/** + * struct ocs_hcu_dev - OCS HCU device context. + * @list: List of device contexts. + * @dev: OCS HCU device. + * @io_base: Base address of OCS HCU registers. + * @engine: Crypto engine for the device. + * @irq: IRQ number. + * @irq_done: Completion for IRQ. + * @irq_err: Flag indicating an IRQ error has happened. + */ +struct ocs_hcu_dev { + struct list_head list; + struct device *dev; + void __iomem *io_base; + struct crypto_engine *engine; + int irq; + struct completion irq_done; + bool irq_err; +}; + +/** + * struct ocs_hcu_idata - Intermediate data generated by the HCU. + * @msg_len_lo: Length of data the HCU has operated on in bits, low 32b. + * @msg_len_hi: Length of data the HCU has operated on in bits, high 32b. + * @digest: The digest read from the HCU. If the HCU is terminated, it will + * contain the actual hash digest. Otherwise it is the intermediate + * state. + */ +struct ocs_hcu_idata { + u32 msg_len_lo; + u32 msg_len_hi; + u8 digest[SHA512_DIGEST_SIZE]; +}; + +/** + * struct ocs_hcu_hash_ctx - Context for OCS HCU hashing operation. + * @algo: The hashing algorithm being used. + * @idata: The current intermediate data. + */ +struct ocs_hcu_hash_ctx { + enum ocs_hcu_algo algo; + struct ocs_hcu_idata idata; +}; + +irqreturn_t ocs_hcu_irq_handler(int irq, void *dev_id); + +struct ocs_hcu_dma_list *ocs_hcu_dma_list_alloc(struct ocs_hcu_dev *hcu_dev, + int max_nents); + +void ocs_hcu_dma_list_free(struct ocs_hcu_dev *hcu_dev, + struct ocs_hcu_dma_list *dma_list); + +int ocs_hcu_dma_list_add_tail(struct ocs_hcu_dev *hcu_dev, + struct ocs_hcu_dma_list *dma_list, + dma_addr_t addr, u32 len); + +int ocs_hcu_hash_init(struct ocs_hcu_hash_ctx *ctx, enum ocs_hcu_algo algo); + +int ocs_hcu_hash_update(struct ocs_hcu_dev *hcu_dev, + struct ocs_hcu_hash_ctx *ctx, + const struct ocs_hcu_dma_list *dma_list); + +int ocs_hcu_hash_finup(struct ocs_hcu_dev *hcu_dev, + const struct ocs_hcu_hash_ctx *ctx, + const struct ocs_hcu_dma_list *dma_list, + u8 *dgst, size_t dgst_len); + +int ocs_hcu_hash_final(struct ocs_hcu_dev *hcu_dev, + const struct ocs_hcu_hash_ctx *ctx, u8 *dgst, + size_t dgst_len); + +int ocs_hcu_digest(struct ocs_hcu_dev *hcu_dev, enum ocs_hcu_algo algo, + void *data, size_t data_len, u8 *dgst, size_t dgst_len); + +int ocs_hcu_hmac(struct ocs_hcu_dev *hcu_dev, enum ocs_hcu_algo algo, + const u8 *key, size_t key_len, + const struct ocs_hcu_dma_list *dma_list, + u8 *dgst, size_t dgst_len); + +#endif /* _CRYPTO_OCS_HCU_H */ diff --git a/drivers/crypto/intel/qat/Kconfig b/drivers/crypto/intel/qat/Kconfig new file mode 100644 index 000000000000..4b4861460dd4 --- /dev/null +++ b/drivers/crypto/intel/qat/Kconfig @@ -0,0 +1,133 @@ +# SPDX-License-Identifier: GPL-2.0-only +config CRYPTO_DEV_QAT + tristate + select CRYPTO_AEAD + select CRYPTO_AUTHENC + select CRYPTO_SKCIPHER + select CRYPTO_AKCIPHER + select CRYPTO_DH + select CRYPTO_RSA + select CRYPTO_LIB_AES + select CRYPTO_LIB_SHA1 + select CRYPTO_LIB_SHA256 + select CRYPTO_LIB_SHA512 + select FW_LOADER + select CRC8 + +config CRYPTO_DEV_QAT_DH895xCC + tristate "Support for Intel(R) DH895xCC" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select CRYPTO_DEV_QAT + help + Support for Intel(R) DH895xcc with Intel(R) QuickAssist Technology + for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_dh895xcc. + +config CRYPTO_DEV_QAT_C3XXX + tristate "Support for Intel(R) C3XXX" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select CRYPTO_DEV_QAT + help + Support for Intel(R) C3xxx with Intel(R) QuickAssist Technology + for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_c3xxx. + +config CRYPTO_DEV_QAT_C62X + tristate "Support for Intel(R) C62X" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select CRYPTO_DEV_QAT + help + Support for Intel(R) C62x with Intel(R) QuickAssist Technology + for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_c62x. + +config CRYPTO_DEV_QAT_4XXX + tristate "Support for Intel(R) QAT_4XXX" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select CRYPTO_DEV_QAT + help + Support for Intel(R) QuickAssist Technology QAT_4xxx + for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_4xxx. + +config CRYPTO_DEV_QAT_420XX + tristate "Support for Intel(R) QAT_420XX" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select CRYPTO_DEV_QAT + help + Support for Intel(R) QuickAssist Technology QAT_420xx + for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_420xx. + +config CRYPTO_DEV_QAT_6XXX + tristate "Support for Intel(R) QuickAssist Technology QAT_6XXX" + depends on (X86 || COMPILE_TEST) + depends on PCI + select CRYPTO_DEV_QAT + help + Support for Intel(R) QuickAssist Technology QAT_6xxx + for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_6xxx. + +config CRYPTO_DEV_QAT_DH895xCCVF + tristate "Support for Intel(R) DH895xCC Virtual Function" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select PCI_IOV + select CRYPTO_DEV_QAT + + help + Support for Intel(R) DH895xcc with Intel(R) QuickAssist Technology + Virtual Function for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_dh895xccvf. + +config CRYPTO_DEV_QAT_C3XXXVF + tristate "Support for Intel(R) C3XXX Virtual Function" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select PCI_IOV + select CRYPTO_DEV_QAT + help + Support for Intel(R) C3xxx with Intel(R) QuickAssist Technology + Virtual Function for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_c3xxxvf. + +config CRYPTO_DEV_QAT_C62XVF + tristate "Support for Intel(R) C62X Virtual Function" + depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST) + select PCI_IOV + select CRYPTO_DEV_QAT + help + Support for Intel(R) C62x with Intel(R) QuickAssist Technology + Virtual Function for accelerating crypto and compression workloads. + + To compile this as a module, choose M here: the module + will be called qat_c62xvf. + +config CRYPTO_DEV_QAT_ERROR_INJECTION + bool "Support for Intel(R) QAT Devices Heartbeat Error Injection" + depends on CRYPTO_DEV_QAT + depends on DEBUG_FS + help + Enables a mechanism that allows to inject a heartbeat error on + Intel(R) QuickAssist devices for testing purposes. + + This is intended for developer use only. + If unsure, say N. + + This functionality is available via debugfs entry of the Intel(R) + QuickAssist device diff --git a/drivers/crypto/intel/qat/Makefile b/drivers/crypto/intel/qat/Makefile new file mode 100644 index 000000000000..abef14207afa --- /dev/null +++ b/drivers/crypto/intel/qat/Makefile @@ -0,0 +1,12 @@ +# SPDX-License-Identifier: GPL-2.0 +subdir-ccflags-y := -I$(src)/qat_common +obj-$(CONFIG_CRYPTO_DEV_QAT) += qat_common/ +obj-$(CONFIG_CRYPTO_DEV_QAT_DH895xCC) += qat_dh895xcc/ +obj-$(CONFIG_CRYPTO_DEV_QAT_C3XXX) += qat_c3xxx/ +obj-$(CONFIG_CRYPTO_DEV_QAT_C62X) += qat_c62x/ +obj-$(CONFIG_CRYPTO_DEV_QAT_4XXX) += qat_4xxx/ +obj-$(CONFIG_CRYPTO_DEV_QAT_420XX) += qat_420xx/ +obj-$(CONFIG_CRYPTO_DEV_QAT_6XXX) += qat_6xxx/ +obj-$(CONFIG_CRYPTO_DEV_QAT_DH895xCCVF) += qat_dh895xccvf/ +obj-$(CONFIG_CRYPTO_DEV_QAT_C3XXXVF) += qat_c3xxxvf/ +obj-$(CONFIG_CRYPTO_DEV_QAT_C62XVF) += qat_c62xvf/ diff --git a/drivers/crypto/intel/qat/qat_420xx/Makefile b/drivers/crypto/intel/qat/qat_420xx/Makefile new file mode 100644 index 000000000000..f6df54d2993e --- /dev/null +++ b/drivers/crypto/intel/qat/qat_420xx/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_420XX) += qat_420xx.o +qat_420xx-y := adf_drv.o adf_420xx_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_420xx/adf_420xx_hw_data.c b/drivers/crypto/intel/qat/qat_420xx/adf_420xx_hw_data.c new file mode 100644 index 000000000000..53fa91d577ed --- /dev/null +++ b/drivers/crypto/intel/qat/qat_420xx/adf_420xx_hw_data.c @@ -0,0 +1,486 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include <linux/iopoll.h> +#include <adf_accel_devices.h> +#include <adf_admin.h> +#include <adf_cfg.h> +#include <adf_cfg_services.h> +#include <adf_clock.h> +#include <adf_common_drv.h> +#include <adf_fw_config.h> +#include <adf_gen4_config.h> +#include <adf_gen4_hw_csr_data.h> +#include <adf_gen4_hw_data.h> +#include <adf_gen4_pfvf.h> +#include <adf_gen4_pm.h> +#include <adf_gen4_ras.h> +#include <adf_gen4_tl.h> +#include <adf_gen4_vf_mig.h> +#include <adf_timer.h> +#include "adf_420xx_hw_data.h" +#include "icp_qat_hw.h" + +#define ADF_AE_GROUP_0 GENMASK(3, 0) +#define ADF_AE_GROUP_1 GENMASK(7, 4) +#define ADF_AE_GROUP_2 GENMASK(11, 8) +#define ADF_AE_GROUP_3 GENMASK(15, 12) +#define ADF_AE_GROUP_4 BIT(16) + +#define ENA_THD_MASK_ASYM GENMASK(1, 0) +#define ENA_THD_MASK_SYM GENMASK(3, 0) +#define ENA_THD_MASK_DC GENMASK(1, 0) + +static const char * const adf_420xx_fw_objs[] = { + [ADF_FW_SYM_OBJ] = ADF_420XX_SYM_OBJ, + [ADF_FW_ASYM_OBJ] = ADF_420XX_ASYM_OBJ, + [ADF_FW_DC_OBJ] = ADF_420XX_DC_OBJ, + [ADF_FW_ADMIN_OBJ] = ADF_420XX_ADMIN_OBJ, +}; + +static const struct adf_fw_config adf_fw_cy_config[] = { + {ADF_AE_GROUP_3, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_1, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_4, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_dc_config[] = { + {ADF_AE_GROUP_1, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_4, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_sym_config[] = { + {ADF_AE_GROUP_3, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_1, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_4, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_asym_config[] = { + {ADF_AE_GROUP_3, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_1, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_4, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_asym_dc_config[] = { + {ADF_AE_GROUP_3, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_1, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_4, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_sym_dc_config[] = { + {ADF_AE_GROUP_2, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_1, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_4, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_dcc_config[] = { + {ADF_AE_GROUP_1, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_4, ADF_FW_ADMIN_OBJ}, +}; + + +static struct adf_hw_device_class adf_420xx_class = { + .name = ADF_420XX_DEVICE_NAME, + .type = DEV_420XX, +}; + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + u32 me_disable = self->fuses[ADF_FUSECTL4]; + + return ~me_disable & ADF_420XX_ACCELENGINES_MASK; +} + +static u32 uof_get_num_objs(struct adf_accel_dev *accel_dev) +{ + switch (adf_get_service_enabled(accel_dev)) { + case SVC_SYM_ASYM: + return ARRAY_SIZE(adf_fw_cy_config); + case SVC_DC: + return ARRAY_SIZE(adf_fw_dc_config); + case SVC_DCC: + return ARRAY_SIZE(adf_fw_dcc_config); + case SVC_SYM: + return ARRAY_SIZE(adf_fw_sym_config); + case SVC_ASYM: + return ARRAY_SIZE(adf_fw_asym_config); + case SVC_ASYM_DC: + return ARRAY_SIZE(adf_fw_asym_dc_config); + case SVC_SYM_DC: + return ARRAY_SIZE(adf_fw_sym_dc_config); + default: + return 0; + } +} + +static const struct adf_fw_config *get_fw_config(struct adf_accel_dev *accel_dev) +{ + switch (adf_get_service_enabled(accel_dev)) { + case SVC_SYM_ASYM: + return adf_fw_cy_config; + case SVC_DC: + return adf_fw_dc_config; + case SVC_DCC: + return adf_fw_dcc_config; + case SVC_SYM: + return adf_fw_sym_config; + case SVC_ASYM: + return adf_fw_asym_config; + case SVC_ASYM_DC: + return adf_fw_asym_dc_config; + case SVC_SYM_DC: + return adf_fw_sym_dc_config; + default: + return NULL; + } +} + +static void update_ae_mask(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + const struct adf_fw_config *fw_config; + u32 config_ae_mask = 0; + u32 ae_mask, num_objs; + int i; + + ae_mask = get_ae_mask(hw_data); + + /* Modify the AE mask based on the firmware configuration loaded */ + fw_config = get_fw_config(accel_dev); + num_objs = uof_get_num_objs(accel_dev); + + config_ae_mask |= ADF_420XX_ADMIN_AE_MASK; + for (i = 0; i < num_objs; i++) + config_ae_mask |= fw_config[i].ae_mask; + + hw_data->ae_mask = ae_mask & config_ae_mask; +} + +static u32 get_accel_cap(struct adf_accel_dev *accel_dev) +{ + u32 capabilities_sym, capabilities_asym, capabilities_dc; + struct pci_dev *pdev = accel_dev->accel_pci_dev.pci_dev; + u32 capabilities_dcc; + u32 fusectl1; + + /* As a side effect, update ae_mask based on configuration */ + update_ae_mask(accel_dev); + + /* Read accelerator capabilities mask */ + pci_read_config_dword(pdev, ADF_GEN4_FUSECTL1_OFFSET, &fusectl1); + + capabilities_sym = ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC | + ICP_ACCEL_CAPABILITIES_CIPHER | + ICP_ACCEL_CAPABILITIES_AUTHENTICATION | + ICP_ACCEL_CAPABILITIES_SHA3 | + ICP_ACCEL_CAPABILITIES_SHA3_EXT | + ICP_ACCEL_CAPABILITIES_HKDF | + ICP_ACCEL_CAPABILITIES_CHACHA_POLY | + ICP_ACCEL_CAPABILITIES_AESGCM_SPC | + ICP_ACCEL_CAPABILITIES_SM3 | + ICP_ACCEL_CAPABILITIES_SM4 | + ICP_ACCEL_CAPABILITIES_AES_V2 | + ICP_ACCEL_CAPABILITIES_ZUC | + ICP_ACCEL_CAPABILITIES_WIRELESS_CRYPTO_EXT | + ICP_ACCEL_CAPABILITIES_EXT_ALGCHAIN; + + /* A set bit in fusectl1 means the feature is OFF in this SKU */ + if (fusectl1 & ICP_ACCEL_GEN4_MASK_CIPHER_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_HKDF; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_UCS_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CHACHA_POLY; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AESGCM_SPC; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AES_V2; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_AUTH_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AUTHENTICATION; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SHA3; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SHA3_EXT; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_SMX_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SM3; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SM4; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_WCP_WAT_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_ZUC; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_WIRELESS_CRYPTO_EXT; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_EIA3_SLICE) + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_ZUC; + + capabilities_asym = ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC | + ICP_ACCEL_CAPABILITIES_SM2 | + ICP_ACCEL_CAPABILITIES_ECEDMONT; + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_PKE_SLICE) { + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC; + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_SM2; + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_ECEDMONT; + } + + capabilities_dc = ICP_ACCEL_CAPABILITIES_COMPRESSION | + ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION | + ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION | + ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64; + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_COMPRESS_SLICE) { + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64; + } + + switch (adf_get_service_enabled(accel_dev)) { + case SVC_SYM_ASYM: + return capabilities_sym | capabilities_asym; + case SVC_DC: + return capabilities_dc; + case SVC_DCC: + /* + * Sym capabilities are available for chaining operations, + * but sym crypto instances cannot be supported + */ + capabilities_dcc = capabilities_dc | capabilities_sym; + capabilities_dcc &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + return capabilities_dcc; + case SVC_SYM: + return capabilities_sym; + case SVC_ASYM: + return capabilities_asym; + case SVC_ASYM_DC: + return capabilities_asym | capabilities_dc; + case SVC_SYM_DC: + return capabilities_sym | capabilities_dc; + default: + return 0; + } +} + +static const u32 *adf_get_arbiter_mapping(struct adf_accel_dev *accel_dev) +{ + if (adf_gen4_init_thd2arb_map(accel_dev)) + dev_warn(&GET_DEV(accel_dev), + "Failed to generate thread to arbiter mapping"); + + return GET_HW_DATA(accel_dev)->thd_to_arb_map; +} + +static void adf_init_rl_data(struct adf_rl_hw_data *rl_data) +{ + rl_data->pciout_tb_offset = ADF_GEN4_RL_TOKEN_PCIEOUT_BUCKET_OFFSET; + rl_data->pciin_tb_offset = ADF_GEN4_RL_TOKEN_PCIEIN_BUCKET_OFFSET; + rl_data->r2l_offset = ADF_GEN4_RL_R2L_OFFSET; + rl_data->l2c_offset = ADF_GEN4_RL_L2C_OFFSET; + rl_data->c2s_offset = ADF_GEN4_RL_C2S_OFFSET; + + rl_data->pcie_scale_div = ADF_420XX_RL_PCIE_SCALE_FACTOR_DIV; + rl_data->pcie_scale_mul = ADF_420XX_RL_PCIE_SCALE_FACTOR_MUL; + rl_data->dcpr_correction = ADF_420XX_RL_DCPR_CORRECTION; + rl_data->max_tp[SVC_ASYM] = ADF_420XX_RL_MAX_TP_ASYM; + rl_data->max_tp[SVC_SYM] = ADF_420XX_RL_MAX_TP_SYM; + rl_data->max_tp[SVC_DC] = ADF_420XX_RL_MAX_TP_DC; + rl_data->scan_interval = ADF_420XX_RL_SCANS_PER_SEC; + rl_data->scale_ref = ADF_420XX_RL_SLICE_REF; + + adf_gen4_init_num_svc_aes(rl_data); +} + +static int get_rp_group(struct adf_accel_dev *accel_dev, u32 ae_mask) +{ + switch (ae_mask) { + case ADF_AE_GROUP_0: + return RP_GROUP_0; + case ADF_AE_GROUP_1: + case ADF_AE_GROUP_3: + return RP_GROUP_1; + case ADF_AE_GROUP_2: + if (get_fw_config(accel_dev) == adf_fw_cy_config) + return RP_GROUP_0; + else + return RP_GROUP_1; + default: + dev_dbg(&GET_DEV(accel_dev), "ae_mask not recognized"); + return -EINVAL; + } +} + +static u32 get_ena_thd_mask(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + const struct adf_fw_config *fw_config; + + if (obj_num >= uof_get_num_objs(accel_dev)) + return ADF_GEN4_ENA_THD_MASK_ERROR; + + fw_config = get_fw_config(accel_dev); + if (!fw_config) + return ADF_GEN4_ENA_THD_MASK_ERROR; + + switch (fw_config[obj_num].obj) { + case ADF_FW_ASYM_OBJ: + return ENA_THD_MASK_ASYM; + case ADF_FW_SYM_OBJ: + return ENA_THD_MASK_SYM; + case ADF_FW_DC_OBJ: + return ENA_THD_MASK_DC; + default: + return ADF_GEN4_ENA_THD_MASK_ERROR; + } +} + +static const char *uof_get_name(struct adf_accel_dev *accel_dev, u32 obj_num, + const char * const fw_objs[], int num_objs) +{ + const struct adf_fw_config *fw_config; + int id; + + fw_config = get_fw_config(accel_dev); + if (fw_config) + id = fw_config[obj_num].obj; + else + id = -EINVAL; + + if (id < 0 || id >= num_objs) + return NULL; + + return fw_objs[id]; +} + +static const char *uof_get_name_420xx(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + int num_fw_objs = ARRAY_SIZE(adf_420xx_fw_objs); + + return uof_get_name(accel_dev, obj_num, adf_420xx_fw_objs, num_fw_objs); +} + +static int uof_get_obj_type(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + const struct adf_fw_config *fw_config; + + if (obj_num >= uof_get_num_objs(accel_dev)) + return -EINVAL; + + fw_config = get_fw_config(accel_dev); + if (!fw_config) + return -EINVAL; + + return fw_config[obj_num].obj; +} + +static u32 uof_get_ae_mask(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + const struct adf_fw_config *fw_config; + + fw_config = get_fw_config(accel_dev); + if (!fw_config) + return 0; + + return fw_config[obj_num].ae_mask; +} + +static void adf_gen4_set_err_mask(struct adf_dev_err_mask *dev_err_mask) +{ + dev_err_mask->cppagentcmdpar_mask = ADF_420XX_HICPPAGENTCMDPARERRLOG_MASK; + dev_err_mask->parerr_ath_cph_mask = ADF_420XX_PARITYERRORMASK_ATH_CPH_MASK; + dev_err_mask->parerr_cpr_xlt_mask = ADF_420XX_PARITYERRORMASK_CPR_XLT_MASK; + dev_err_mask->parerr_dcpr_ucs_mask = ADF_420XX_PARITYERRORMASK_DCPR_UCS_MASK; + dev_err_mask->parerr_pke_mask = ADF_420XX_PARITYERRORMASK_PKE_MASK; + dev_err_mask->parerr_wat_wcp_mask = ADF_420XX_PARITYERRORMASK_WAT_WCP_MASK; + dev_err_mask->ssmfeatren_mask = ADF_420XX_SSMFEATREN_MASK; +} + +void adf_init_hw_data_420xx(struct adf_hw_device_data *hw_data, u32 dev_id) +{ + hw_data->dev_class = &adf_420xx_class; + hw_data->instance_id = adf_420xx_class.instances++; + hw_data->num_banks = ADF_GEN4_ETR_MAX_BANKS; + hw_data->num_banks_per_vf = ADF_GEN4_NUM_BANKS_PER_VF; + hw_data->num_rings_per_bank = ADF_GEN4_NUM_RINGS_PER_BANK; + hw_data->num_accel = ADF_GEN4_MAX_ACCELERATORS; + hw_data->num_engines = ADF_420XX_MAX_ACCELENGINES; + hw_data->num_logical_accel = 1; + hw_data->tx_rx_gap = ADF_GEN4_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_GEN4_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN4_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_isr_resource_alloc; + hw_data->free_irq = adf_isr_resource_free; + hw_data->enable_error_correction = adf_gen4_enable_error_correction; + hw_data->get_accel_mask = adf_gen4_get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_num_accels = adf_gen4_get_num_accels; + hw_data->get_num_aes = adf_gen4_get_num_aes; + hw_data->get_sram_bar_id = adf_gen4_get_sram_bar_id; + hw_data->get_etr_bar_id = adf_gen4_get_etr_bar_id; + hw_data->get_misc_bar_id = adf_gen4_get_misc_bar_id; + hw_data->get_arb_info = adf_gen4_get_arb_info; + hw_data->get_admin_info = adf_gen4_get_admin_info; + hw_data->get_accel_cap = get_accel_cap; + hw_data->get_sku = adf_gen4_get_sku; + hw_data->init_admin_comms = adf_init_admin_comms; + hw_data->exit_admin_comms = adf_exit_admin_comms; + hw_data->send_admin_init = adf_send_admin_init; + hw_data->init_arb = adf_init_arb; + hw_data->exit_arb = adf_exit_arb; + hw_data->get_arb_mapping = adf_get_arbiter_mapping; + hw_data->enable_ints = adf_gen4_enable_ints; + hw_data->init_device = adf_gen4_init_device; + hw_data->reset_device = adf_reset_flr; + hw_data->admin_ae_mask = ADF_420XX_ADMIN_AE_MASK; + hw_data->num_rps = ADF_GEN4_MAX_RPS; + hw_data->fw_name = ADF_420XX_FW; + hw_data->fw_mmp_name = ADF_420XX_MMP; + hw_data->uof_get_name = uof_get_name_420xx; + hw_data->uof_get_num_objs = uof_get_num_objs; + hw_data->uof_get_obj_type = uof_get_obj_type; + hw_data->uof_get_ae_mask = uof_get_ae_mask; + hw_data->get_rp_group = get_rp_group; + hw_data->get_ena_thd_mask = get_ena_thd_mask; + hw_data->set_msix_rttable = adf_gen4_set_msix_default_rttable; + hw_data->set_ssm_wdtimer = adf_gen4_set_ssm_wdtimer; + hw_data->get_ring_to_svc_map = adf_gen4_get_ring_to_svc_map; + hw_data->disable_iov = adf_disable_sriov; + hw_data->ring_pair_reset = adf_gen4_ring_pair_reset; + hw_data->enable_pm = adf_gen4_enable_pm; + hw_data->handle_pm_interrupt = adf_gen4_handle_pm_interrupt; + hw_data->dev_config = adf_gen4_dev_config; + hw_data->start_timer = adf_timer_start; + hw_data->stop_timer = adf_timer_stop; + hw_data->get_hb_clock = adf_gen4_get_heartbeat_clock; + hw_data->num_hb_ctrs = ADF_NUM_HB_CNT_PER_AE; + hw_data->clock_frequency = ADF_420XX_AE_FREQ; + hw_data->services_supported = adf_gen4_services_supported; + hw_data->get_svc_slice_cnt = adf_gen4_get_svc_slice_cnt; + + adf_gen4_set_err_mask(&hw_data->dev_err_mask); + adf_gen4_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen4_init_pf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen4_init_dc_ops(&hw_data->dc_ops); + adf_gen4_init_ras_ops(&hw_data->ras_ops); + adf_gen4_init_tl_data(&hw_data->tl_data); + adf_gen4_init_vf_mig_ops(&hw_data->vfmig_ops); + adf_init_rl_data(&hw_data->rl_data); +} + +void adf_clean_hw_data_420xx(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; +} diff --git a/drivers/crypto/intel/qat/qat_420xx/adf_420xx_hw_data.h b/drivers/crypto/intel/qat/qat_420xx/adf_420xx_hw_data.h new file mode 100644 index 000000000000..99abbfc14820 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_420xx/adf_420xx_hw_data.h @@ -0,0 +1,55 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_420XX_HW_DATA_H_ +#define ADF_420XX_HW_DATA_H_ + +#include <adf_accel_devices.h> + +#define ADF_420XX_MAX_ACCELENGINES 17 + +#define ADF_420XX_ACCELENGINES_MASK 0x1FFFF +#define ADF_420XX_ADMIN_AE_MASK 0x10000 + +#define ADF_420XX_HICPPAGENTCMDPARERRLOG_MASK (0xFF) +#define ADF_420XX_PARITYERRORMASK_ATH_CPH_MASK (0xFF00FF) +#define ADF_420XX_PARITYERRORMASK_CPR_XLT_MASK (0x10001) +#define ADF_420XX_PARITYERRORMASK_DCPR_UCS_MASK (0xF0007) +#define ADF_420XX_PARITYERRORMASK_PKE_MASK (0xFFF) +#define ADF_420XX_PARITYERRORMASK_WAT_WCP_MASK (0x3FF03FF) + +/* + * SSMFEATREN bit mask + * BIT(4) - enables parity detection on CPP + * BIT(12) - enables the logging of push/pull data errors + * in pperr register + * BIT(16) - BIT(27) - enable parity detection on SPPs + */ +#define ADF_420XX_SSMFEATREN_MASK \ + (BIT(4) | BIT(12) | BIT(16) | BIT(17) | BIT(18) | BIT(19) | BIT(20) | \ + BIT(21) | BIT(22) | BIT(23) | BIT(24) | BIT(25) | BIT(26) | BIT(27)) + +/* Firmware Binaries */ +#define ADF_420XX_FW "qat_420xx.bin" +#define ADF_420XX_MMP "qat_420xx_mmp.bin" +#define ADF_420XX_SYM_OBJ "qat_420xx_sym.bin" +#define ADF_420XX_DC_OBJ "qat_420xx_dc.bin" +#define ADF_420XX_ASYM_OBJ "qat_420xx_asym.bin" +#define ADF_420XX_ADMIN_OBJ "qat_420xx_admin.bin" + +/* RL constants */ +#define ADF_420XX_RL_PCIE_SCALE_FACTOR_DIV 100 +#define ADF_420XX_RL_PCIE_SCALE_FACTOR_MUL 102 +#define ADF_420XX_RL_DCPR_CORRECTION 1 +#define ADF_420XX_RL_SCANS_PER_SEC 954 +#define ADF_420XX_RL_MAX_TP_ASYM 173750UL +#define ADF_420XX_RL_MAX_TP_SYM 95000UL +#define ADF_420XX_RL_MAX_TP_DC 40000UL +#define ADF_420XX_RL_SLICE_REF 1000UL + +/* Clocks frequency */ +#define ADF_420XX_AE_FREQ (1000 * HZ_PER_MHZ) + +void adf_init_hw_data_420xx(struct adf_hw_device_data *hw_data, u32 dev_id); +void adf_clean_hw_data_420xx(struct adf_hw_device_data *hw_data); + +#endif diff --git a/drivers/crypto/intel/qat/qat_420xx/adf_drv.c b/drivers/crypto/intel/qat/qat_420xx/adf_drv.c new file mode 100644 index 000000000000..cfa00daeb4fb --- /dev/null +++ b/drivers/crypto/intel/qat/qat_420xx/adf_drv.c @@ -0,0 +1,215 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include <linux/device.h> +#include <linux/module.h> +#include <linux/pci.h> + +#include <adf_accel_devices.h> +#include <adf_gen4_hw_data.h> +#include <adf_gen4_config.h> +#include <adf_cfg.h> +#include <adf_common_drv.h> +#include <adf_dbgfs.h> + +#include "adf_420xx_hw_data.h" + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_420XX) }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + if (accel_dev->hw_device) { + adf_clean_hw_data_420xx(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + adf_devmgr_rm_dev(accel_dev, NULL); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + struct adf_bar *bar; + int ret; + + if (num_possible_nodes() > 1 && dev_to_node(&pdev->dev) < 0) { + /* + * If the accelerator is connected to a node with no memory + * there is no point in using the accelerator since the remote + * memory transaction will be very slow. + */ + dev_err(&pdev->dev, "Invalid NUMA configuration.\n"); + return -EINVAL; + } + + accel_dev = devm_kzalloc(&pdev->dev, sizeof(*accel_dev), GFP_KERNEL); + if (!accel_dev) + return -ENOMEM; + + INIT_LIST_HEAD(&accel_dev->crypto_list); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* + * Add accel device to accel table + * This should be called before adf_cleanup_accel is called + */ + if (adf_devmgr_add_dev(accel_dev, NULL)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + return -EFAULT; + } + + accel_dev->owner = THIS_MODULE; + /* Allocate and initialise device hardware meta-data structure */ + hw_data = devm_kzalloc(&pdev->dev, sizeof(*hw_data), GFP_KERNEL); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + + accel_dev->hw_device = hw_data; + adf_init_hw_data_420xx(accel_dev->hw_device, ent->device); + + pci_read_config_byte(pdev, PCI_REVISION_ID, &accel_pci_dev->revid); + pci_read_config_dword(pdev, ADF_GEN4_FUSECTL4_OFFSET, &hw_data->fuses[ADF_FUSECTL4]); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + /* If the device has no acceleration engines then ignore it */ + if (!hw_data->accel_mask || !hw_data->ae_mask || + (~hw_data->ae_mask & 0x01)) { + dev_err(&pdev->dev, "No acceleration units found.\n"); + ret = -EFAULT; + goto out_err; + } + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + /* Enable PCI device */ + ret = pcim_enable_device(pdev); + if (ret) { + dev_err(&pdev->dev, "Can't enable PCI device.\n"); + goto out_err; + } + + /* Set DMA identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration.\n"); + goto out_err; + } + + ret = adf_gen4_cfg_dev_init(accel_dev); + if (ret) { + dev_err(&pdev->dev, "Failed to initialize configuration.\n"); + goto out_err; + } + + /* Get accelerator capabilities mask */ + hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev); + if (!hw_data->accel_capabilities_mask) { + dev_err(&pdev->dev, "Failed to get capabilities mask.\n"); + ret = -EINVAL; + goto out_err; + } + + /* Find and map all the device's BARS */ + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM) & ADF_GEN4_BAR_MASK; + + ret = pcim_request_all_regions(pdev, pci_name(pdev)); + if (ret) { + dev_err(&pdev->dev, "Failed to request PCI regions.\n"); + goto out_err; + } + + i = 0; + for_each_set_bit(bar_nr, &bar_mask, PCI_STD_NUM_BARS) { + bar = &accel_pci_dev->pci_bars[i++]; + bar->virt_addr = pcim_iomap(pdev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to ioremap PCI region.\n"); + ret = -ENOMEM; + goto out_err; + } + } + + pci_set_master(pdev); + + if (pci_save_state(pdev)) { + dev_err(&pdev->dev, "Failed to save pci state.\n"); + ret = -ENOMEM; + goto out_err; + } + + accel_dev->ras_errors.enabled = true; + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, true); + if (ret) + goto out_err_dev_stop; + + ret = adf_sysfs_init(accel_dev); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err: + adf_cleanup_accel(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); +} + +static void adf_shutdown(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + adf_dev_down(accel_dev); +} + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_420XX_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, + .shutdown = adf_shutdown, + .sriov_configure = adf_sriov_configure, + .err_handler = &adf_err_handler, +}; + +module_pci_driver(adf_driver); + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Intel"); +MODULE_FIRMWARE(ADF_420XX_FW); +MODULE_FIRMWARE(ADF_420XX_MMP); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_SOFTDEP("pre: crypto-intel_qat"); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_4xxx/Makefile b/drivers/crypto/intel/qat/qat_4xxx/Makefile new file mode 100644 index 000000000000..188b611445e6 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_4xxx/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +obj-$(CONFIG_CRYPTO_DEV_QAT_4XXX) += qat_4xxx.o +qat_4xxx-y := adf_drv.o adf_4xxx_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_4xxx/adf_4xxx_hw_data.c b/drivers/crypto/intel/qat/qat_4xxx/adf_4xxx_hw_data.c new file mode 100644 index 000000000000..740f68a36ac5 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_4xxx/adf_4xxx_hw_data.c @@ -0,0 +1,480 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2020 - 2021 Intel Corporation */ +#include <linux/iopoll.h> +#include <adf_accel_devices.h> +#include <adf_admin.h> +#include <adf_bank_state.h> +#include <adf_cfg.h> +#include <adf_cfg_services.h> +#include <adf_clock.h> +#include <adf_common_drv.h> +#include <adf_fw_config.h> +#include <adf_gen4_config.h> +#include <adf_gen4_hw_csr_data.h> +#include <adf_gen4_hw_data.h> +#include <adf_gen4_pfvf.h> +#include <adf_gen4_pm.h> +#include "adf_gen4_ras.h" +#include <adf_gen4_tl.h> +#include <adf_gen4_vf_mig.h> +#include <adf_timer.h> +#include "adf_4xxx_hw_data.h" +#include "icp_qat_hw.h" + +#define ADF_AE_GROUP_0 GENMASK(3, 0) +#define ADF_AE_GROUP_1 GENMASK(7, 4) +#define ADF_AE_GROUP_2 BIT(8) + +#define ENA_THD_MASK_ASYM GENMASK(1, 0) +#define ENA_THD_MASK_ASYM_401XX GENMASK(5, 0) +#define ENA_THD_MASK_SYM GENMASK(6, 0) +#define ENA_THD_MASK_DC GENMASK(1, 0) + +static const char * const adf_4xxx_fw_objs[] = { + [ADF_FW_SYM_OBJ] = ADF_4XXX_SYM_OBJ, + [ADF_FW_ASYM_OBJ] = ADF_4XXX_ASYM_OBJ, + [ADF_FW_DC_OBJ] = ADF_4XXX_DC_OBJ, + [ADF_FW_ADMIN_OBJ] = ADF_4XXX_ADMIN_OBJ, +}; + +static const char * const adf_402xx_fw_objs[] = { + [ADF_FW_SYM_OBJ] = ADF_402XX_SYM_OBJ, + [ADF_FW_ASYM_OBJ] = ADF_402XX_ASYM_OBJ, + [ADF_FW_DC_OBJ] = ADF_402XX_DC_OBJ, + [ADF_FW_ADMIN_OBJ] = ADF_402XX_ADMIN_OBJ, +}; + +static const struct adf_fw_config adf_fw_cy_config[] = { + {ADF_AE_GROUP_1, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_dc_config[] = { + {ADF_AE_GROUP_1, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_sym_config[] = { + {ADF_AE_GROUP_1, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_asym_config[] = { + {ADF_AE_GROUP_1, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_asym_dc_config[] = { + {ADF_AE_GROUP_1, ADF_FW_ASYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_sym_dc_config[] = { + {ADF_AE_GROUP_1, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ}, +}; + +static const struct adf_fw_config adf_fw_dcc_config[] = { + {ADF_AE_GROUP_1, ADF_FW_DC_OBJ}, + {ADF_AE_GROUP_0, ADF_FW_SYM_OBJ}, + {ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ}, +}; + +static_assert(ARRAY_SIZE(adf_fw_cy_config) == ARRAY_SIZE(adf_fw_dc_config)); +static_assert(ARRAY_SIZE(adf_fw_cy_config) == ARRAY_SIZE(adf_fw_sym_config)); +static_assert(ARRAY_SIZE(adf_fw_cy_config) == ARRAY_SIZE(adf_fw_asym_config)); +static_assert(ARRAY_SIZE(adf_fw_cy_config) == ARRAY_SIZE(adf_fw_asym_dc_config)); +static_assert(ARRAY_SIZE(adf_fw_cy_config) == ARRAY_SIZE(adf_fw_sym_dc_config)); +static_assert(ARRAY_SIZE(adf_fw_cy_config) == ARRAY_SIZE(adf_fw_dcc_config)); + +static struct adf_hw_device_class adf_4xxx_class = { + .name = ADF_4XXX_DEVICE_NAME, + .type = DEV_4XXX, +}; + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + u32 me_disable = self->fuses[ADF_FUSECTL4]; + + return ~me_disable & ADF_4XXX_ACCELENGINES_MASK; +} + +static u32 get_accel_cap(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_dev->accel_pci_dev.pci_dev; + u32 capabilities_sym, capabilities_asym, capabilities_dc; + u32 capabilities_dcc; + u32 fusectl1; + + /* Read accelerator capabilities mask */ + pci_read_config_dword(pdev, ADF_GEN4_FUSECTL1_OFFSET, &fusectl1); + + capabilities_sym = ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC | + ICP_ACCEL_CAPABILITIES_CIPHER | + ICP_ACCEL_CAPABILITIES_AUTHENTICATION | + ICP_ACCEL_CAPABILITIES_SHA3 | + ICP_ACCEL_CAPABILITIES_SHA3_EXT | + ICP_ACCEL_CAPABILITIES_HKDF | + ICP_ACCEL_CAPABILITIES_CHACHA_POLY | + ICP_ACCEL_CAPABILITIES_AESGCM_SPC | + ICP_ACCEL_CAPABILITIES_SM3 | + ICP_ACCEL_CAPABILITIES_SM4 | + ICP_ACCEL_CAPABILITIES_AES_V2; + + /* A set bit in fusectl1 means the feature is OFF in this SKU */ + if (fusectl1 & ICP_ACCEL_GEN4_MASK_CIPHER_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_HKDF; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_UCS_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CHACHA_POLY; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AESGCM_SPC; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AES_V2; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_AUTH_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AUTHENTICATION; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SHA3; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SHA3_EXT; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_SMX_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SM3; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SM4; + } + + capabilities_asym = ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC | + ICP_ACCEL_CAPABILITIES_CIPHER | + ICP_ACCEL_CAPABILITIES_SM2 | + ICP_ACCEL_CAPABILITIES_ECEDMONT; + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_PKE_SLICE) { + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC; + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_SM2; + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_ECEDMONT; + } + + capabilities_dc = ICP_ACCEL_CAPABILITIES_COMPRESSION | + ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION | + ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION | + ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64; + + if (fusectl1 & ICP_ACCEL_GEN4_MASK_COMPRESS_SLICE) { + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64; + } + + switch (adf_get_service_enabled(accel_dev)) { + case SVC_SYM_ASYM: + return capabilities_sym | capabilities_asym; + case SVC_DC: + return capabilities_dc; + case SVC_DCC: + /* + * Sym capabilities are available for chaining operations, + * but sym crypto instances cannot be supported + */ + capabilities_dcc = capabilities_dc | capabilities_sym; + capabilities_dcc &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + return capabilities_dcc; + case SVC_SYM: + return capabilities_sym; + case SVC_ASYM: + return capabilities_asym; + case SVC_ASYM_DC: + return capabilities_asym | capabilities_dc; + case SVC_SYM_DC: + return capabilities_sym | capabilities_dc; + default: + return 0; + } +} + +static const u32 *adf_get_arbiter_mapping(struct adf_accel_dev *accel_dev) +{ + if (adf_gen4_init_thd2arb_map(accel_dev)) + dev_warn(&GET_DEV(accel_dev), + "Failed to generate thread to arbiter mapping"); + + return GET_HW_DATA(accel_dev)->thd_to_arb_map; +} + +static void adf_init_rl_data(struct adf_rl_hw_data *rl_data) +{ + rl_data->pciout_tb_offset = ADF_GEN4_RL_TOKEN_PCIEOUT_BUCKET_OFFSET; + rl_data->pciin_tb_offset = ADF_GEN4_RL_TOKEN_PCIEIN_BUCKET_OFFSET; + rl_data->r2l_offset = ADF_GEN4_RL_R2L_OFFSET; + rl_data->l2c_offset = ADF_GEN4_RL_L2C_OFFSET; + rl_data->c2s_offset = ADF_GEN4_RL_C2S_OFFSET; + + rl_data->pcie_scale_div = ADF_4XXX_RL_PCIE_SCALE_FACTOR_DIV; + rl_data->pcie_scale_mul = ADF_4XXX_RL_PCIE_SCALE_FACTOR_MUL; + rl_data->dcpr_correction = ADF_4XXX_RL_DCPR_CORRECTION; + rl_data->max_tp[SVC_ASYM] = ADF_4XXX_RL_MAX_TP_ASYM; + rl_data->max_tp[SVC_SYM] = ADF_4XXX_RL_MAX_TP_SYM; + rl_data->max_tp[SVC_DC] = ADF_4XXX_RL_MAX_TP_DC; + rl_data->scan_interval = ADF_4XXX_RL_SCANS_PER_SEC; + rl_data->scale_ref = ADF_4XXX_RL_SLICE_REF; + + adf_gen4_init_num_svc_aes(rl_data); +} + +static u32 uof_get_num_objs(struct adf_accel_dev *accel_dev) +{ + return ARRAY_SIZE(adf_fw_cy_config); +} + +static const struct adf_fw_config *get_fw_config(struct adf_accel_dev *accel_dev) +{ + switch (adf_get_service_enabled(accel_dev)) { + case SVC_SYM_ASYM: + return adf_fw_cy_config; + case SVC_DC: + return adf_fw_dc_config; + case SVC_DCC: + return adf_fw_dcc_config; + case SVC_SYM: + return adf_fw_sym_config; + case SVC_ASYM: + return adf_fw_asym_config; + case SVC_ASYM_DC: + return adf_fw_asym_dc_config; + case SVC_SYM_DC: + return adf_fw_sym_dc_config; + default: + return NULL; + } +} + +static int get_rp_group(struct adf_accel_dev *accel_dev, u32 ae_mask) +{ + switch (ae_mask) { + case ADF_AE_GROUP_0: + return RP_GROUP_0; + case ADF_AE_GROUP_1: + return RP_GROUP_1; + default: + dev_dbg(&GET_DEV(accel_dev), "ae_mask not recognized"); + return -EINVAL; + } +} + +static u32 get_ena_thd_mask(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + const struct adf_fw_config *fw_config; + + if (obj_num >= uof_get_num_objs(accel_dev)) + return ADF_GEN4_ENA_THD_MASK_ERROR; + + fw_config = get_fw_config(accel_dev); + if (!fw_config) + return ADF_GEN4_ENA_THD_MASK_ERROR; + + switch (fw_config[obj_num].obj) { + case ADF_FW_ASYM_OBJ: + return ENA_THD_MASK_ASYM; + case ADF_FW_SYM_OBJ: + return ENA_THD_MASK_SYM; + case ADF_FW_DC_OBJ: + return ENA_THD_MASK_DC; + default: + return ADF_GEN4_ENA_THD_MASK_ERROR; + } +} + +static u32 get_ena_thd_mask_401xx(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + const struct adf_fw_config *fw_config; + + if (obj_num >= uof_get_num_objs(accel_dev)) + return ADF_GEN4_ENA_THD_MASK_ERROR; + + fw_config = get_fw_config(accel_dev); + if (!fw_config) + return ADF_GEN4_ENA_THD_MASK_ERROR; + + switch (fw_config[obj_num].obj) { + case ADF_FW_ASYM_OBJ: + return ENA_THD_MASK_ASYM_401XX; + case ADF_FW_SYM_OBJ: + return ENA_THD_MASK_SYM; + case ADF_FW_DC_OBJ: + return ENA_THD_MASK_DC; + default: + return ADF_GEN4_ENA_THD_MASK_ERROR; + } +} + +static const char *uof_get_name(struct adf_accel_dev *accel_dev, u32 obj_num, + const char * const fw_objs[], int num_objs) +{ + const struct adf_fw_config *fw_config; + int id; + + fw_config = get_fw_config(accel_dev); + if (fw_config) + id = fw_config[obj_num].obj; + else + id = -EINVAL; + + if (id < 0 || id >= num_objs) + return NULL; + + return fw_objs[id]; +} + +static const char *uof_get_name_4xxx(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + int num_fw_objs = ARRAY_SIZE(adf_4xxx_fw_objs); + + return uof_get_name(accel_dev, obj_num, adf_4xxx_fw_objs, num_fw_objs); +} + +static const char *uof_get_name_402xx(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + int num_fw_objs = ARRAY_SIZE(adf_402xx_fw_objs); + + return uof_get_name(accel_dev, obj_num, adf_402xx_fw_objs, num_fw_objs); +} + +static int uof_get_obj_type(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + const struct adf_fw_config *fw_config; + + if (obj_num >= uof_get_num_objs(accel_dev)) + return -EINVAL; + + fw_config = get_fw_config(accel_dev); + if (!fw_config) + return -EINVAL; + + return fw_config[obj_num].obj; +} + +static u32 uof_get_ae_mask(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + const struct adf_fw_config *fw_config; + + fw_config = get_fw_config(accel_dev); + if (!fw_config) + return 0; + + return fw_config[obj_num].ae_mask; +} + +static void adf_gen4_set_err_mask(struct adf_dev_err_mask *dev_err_mask) +{ + dev_err_mask->cppagentcmdpar_mask = ADF_4XXX_HICPPAGENTCMDPARERRLOG_MASK; + dev_err_mask->parerr_ath_cph_mask = ADF_4XXX_PARITYERRORMASK_ATH_CPH_MASK; + dev_err_mask->parerr_cpr_xlt_mask = ADF_4XXX_PARITYERRORMASK_CPR_XLT_MASK; + dev_err_mask->parerr_dcpr_ucs_mask = ADF_4XXX_PARITYERRORMASK_DCPR_UCS_MASK; + dev_err_mask->parerr_pke_mask = ADF_4XXX_PARITYERRORMASK_PKE_MASK; + dev_err_mask->ssmfeatren_mask = ADF_4XXX_SSMFEATREN_MASK; +} + +void adf_init_hw_data_4xxx(struct adf_hw_device_data *hw_data, u32 dev_id) +{ + hw_data->dev_class = &adf_4xxx_class; + hw_data->instance_id = adf_4xxx_class.instances++; + hw_data->num_banks = ADF_GEN4_ETR_MAX_BANKS; + hw_data->num_banks_per_vf = ADF_GEN4_NUM_BANKS_PER_VF; + hw_data->num_rings_per_bank = ADF_GEN4_NUM_RINGS_PER_BANK; + hw_data->num_accel = ADF_GEN4_MAX_ACCELERATORS; + hw_data->num_engines = ADF_4XXX_MAX_ACCELENGINES; + hw_data->num_logical_accel = 1; + hw_data->tx_rx_gap = ADF_GEN4_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_GEN4_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN4_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_isr_resource_alloc; + hw_data->free_irq = adf_isr_resource_free; + hw_data->enable_error_correction = adf_gen4_enable_error_correction; + hw_data->get_accel_mask = adf_gen4_get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_num_accels = adf_gen4_get_num_accels; + hw_data->get_num_aes = adf_gen4_get_num_aes; + hw_data->get_sram_bar_id = adf_gen4_get_sram_bar_id; + hw_data->get_etr_bar_id = adf_gen4_get_etr_bar_id; + hw_data->get_misc_bar_id = adf_gen4_get_misc_bar_id; + hw_data->get_arb_info = adf_gen4_get_arb_info; + hw_data->get_admin_info = adf_gen4_get_admin_info; + hw_data->get_accel_cap = get_accel_cap; + hw_data->get_sku = adf_gen4_get_sku; + hw_data->init_admin_comms = adf_init_admin_comms; + hw_data->exit_admin_comms = adf_exit_admin_comms; + hw_data->send_admin_init = adf_send_admin_init; + hw_data->init_arb = adf_init_arb; + hw_data->exit_arb = adf_exit_arb; + hw_data->get_arb_mapping = adf_get_arbiter_mapping; + hw_data->enable_ints = adf_gen4_enable_ints; + hw_data->init_device = adf_gen4_init_device; + hw_data->reset_device = adf_reset_flr; + hw_data->admin_ae_mask = ADF_4XXX_ADMIN_AE_MASK; + hw_data->num_rps = ADF_GEN4_MAX_RPS; + switch (dev_id) { + case PCI_DEVICE_ID_INTEL_QAT_402XX: + hw_data->fw_name = ADF_402XX_FW; + hw_data->fw_mmp_name = ADF_402XX_MMP; + hw_data->uof_get_name = uof_get_name_402xx; + hw_data->get_ena_thd_mask = get_ena_thd_mask; + break; + case PCI_DEVICE_ID_INTEL_QAT_401XX: + hw_data->fw_name = ADF_4XXX_FW; + hw_data->fw_mmp_name = ADF_4XXX_MMP; + hw_data->uof_get_name = uof_get_name_4xxx; + hw_data->get_ena_thd_mask = get_ena_thd_mask_401xx; + break; + default: + hw_data->fw_name = ADF_4XXX_FW; + hw_data->fw_mmp_name = ADF_4XXX_MMP; + hw_data->uof_get_name = uof_get_name_4xxx; + hw_data->get_ena_thd_mask = get_ena_thd_mask; + break; + } + hw_data->uof_get_num_objs = uof_get_num_objs; + hw_data->uof_get_obj_type = uof_get_obj_type; + hw_data->uof_get_ae_mask = uof_get_ae_mask; + hw_data->get_rp_group = get_rp_group; + hw_data->set_msix_rttable = adf_gen4_set_msix_default_rttable; + hw_data->set_ssm_wdtimer = adf_gen4_set_ssm_wdtimer; + hw_data->get_ring_to_svc_map = adf_gen4_get_ring_to_svc_map; + hw_data->disable_iov = adf_disable_sriov; + hw_data->ring_pair_reset = adf_gen4_ring_pair_reset; + hw_data->bank_state_save = adf_bank_state_save; + hw_data->bank_state_restore = adf_bank_state_restore; + hw_data->enable_pm = adf_gen4_enable_pm; + hw_data->handle_pm_interrupt = adf_gen4_handle_pm_interrupt; + hw_data->dev_config = adf_gen4_dev_config; + hw_data->start_timer = adf_timer_start; + hw_data->stop_timer = adf_timer_stop; + hw_data->get_hb_clock = adf_gen4_get_heartbeat_clock; + hw_data->num_hb_ctrs = ADF_NUM_HB_CNT_PER_AE; + hw_data->clock_frequency = ADF_4XXX_AE_FREQ; + hw_data->services_supported = adf_gen4_services_supported; + hw_data->get_svc_slice_cnt = adf_gen4_get_svc_slice_cnt; + + adf_gen4_set_err_mask(&hw_data->dev_err_mask); + adf_gen4_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen4_init_pf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen4_init_dc_ops(&hw_data->dc_ops); + adf_gen4_init_ras_ops(&hw_data->ras_ops); + adf_gen4_init_tl_data(&hw_data->tl_data); + adf_gen4_init_vf_mig_ops(&hw_data->vfmig_ops); + adf_init_rl_data(&hw_data->rl_data); +} + +void adf_clean_hw_data_4xxx(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; +} diff --git a/drivers/crypto/intel/qat/qat_4xxx/adf_4xxx_hw_data.h b/drivers/crypto/intel/qat/qat_4xxx/adf_4xxx_hw_data.h new file mode 100644 index 000000000000..76388363ea87 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_4xxx/adf_4xxx_hw_data.h @@ -0,0 +1,62 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_4XXX_HW_DATA_H_ +#define ADF_4XXX_HW_DATA_H_ + +#include <linux/units.h> +#include <adf_accel_devices.h> + +#define ADF_4XXX_MAX_ACCELENGINES 9 + +#define ADF_4XXX_ACCELENGINES_MASK (0x1FF) +#define ADF_4XXX_ADMIN_AE_MASK (0x100) + +#define ADF_4XXX_HICPPAGENTCMDPARERRLOG_MASK 0x1F +#define ADF_4XXX_PARITYERRORMASK_ATH_CPH_MASK 0xF000F +#define ADF_4XXX_PARITYERRORMASK_CPR_XLT_MASK 0x10001 +#define ADF_4XXX_PARITYERRORMASK_DCPR_UCS_MASK 0x30007 +#define ADF_4XXX_PARITYERRORMASK_PKE_MASK 0x3F + +/* + * SSMFEATREN bit mask + * BIT(4) - enables parity detection on CPP + * BIT(12) - enables the logging of push/pull data errors + * in pperr register + * BIT(16) - BIT(23) - enable parity detection on SPPs + */ +#define ADF_4XXX_SSMFEATREN_MASK \ + (BIT(4) | BIT(12) | BIT(16) | BIT(17) | BIT(18) | \ + BIT(19) | BIT(20) | BIT(21) | BIT(22) | BIT(23)) + +/* Firmware Binaries */ +#define ADF_4XXX_FW "qat_4xxx.bin" +#define ADF_4XXX_MMP "qat_4xxx_mmp.bin" +#define ADF_4XXX_SYM_OBJ "qat_4xxx_sym.bin" +#define ADF_4XXX_DC_OBJ "qat_4xxx_dc.bin" +#define ADF_4XXX_ASYM_OBJ "qat_4xxx_asym.bin" +#define ADF_4XXX_ADMIN_OBJ "qat_4xxx_admin.bin" +/* Firmware for 402XXX */ +#define ADF_402XX_FW "qat_402xx.bin" +#define ADF_402XX_MMP "qat_402xx_mmp.bin" +#define ADF_402XX_SYM_OBJ "qat_402xx_sym.bin" +#define ADF_402XX_DC_OBJ "qat_402xx_dc.bin" +#define ADF_402XX_ASYM_OBJ "qat_402xx_asym.bin" +#define ADF_402XX_ADMIN_OBJ "qat_402xx_admin.bin" + +/* RL constants */ +#define ADF_4XXX_RL_PCIE_SCALE_FACTOR_DIV 100 +#define ADF_4XXX_RL_PCIE_SCALE_FACTOR_MUL 102 +#define ADF_4XXX_RL_DCPR_CORRECTION 1 +#define ADF_4XXX_RL_SCANS_PER_SEC 954 +#define ADF_4XXX_RL_MAX_TP_ASYM 173750UL +#define ADF_4XXX_RL_MAX_TP_SYM 95000UL +#define ADF_4XXX_RL_MAX_TP_DC 45000UL +#define ADF_4XXX_RL_SLICE_REF 1000UL + +/* Clocks frequency */ +#define ADF_4XXX_AE_FREQ (1000 * HZ_PER_MHZ) + +void adf_init_hw_data_4xxx(struct adf_hw_device_data *hw_data, u32 dev_id); +void adf_clean_hw_data_4xxx(struct adf_hw_device_data *hw_data); + +#endif diff --git a/drivers/crypto/intel/qat/qat_4xxx/adf_drv.c b/drivers/crypto/intel/qat/qat_4xxx/adf_drv.c new file mode 100644 index 000000000000..c9be5dcddb27 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_4xxx/adf_drv.c @@ -0,0 +1,219 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2020 Intel Corporation */ +#include <linux/device.h> +#include <linux/module.h> +#include <linux/pci.h> + +#include <adf_accel_devices.h> +#include <adf_cfg.h> +#include <adf_common_drv.h> +#include <adf_dbgfs.h> +#include <adf_gen4_config.h> +#include <adf_gen4_hw_data.h> + +#include "adf_4xxx_hw_data.h" + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_4XXX) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_401XX) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_402XX) }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + if (accel_dev->hw_device) { + adf_clean_hw_data_4xxx(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + adf_devmgr_rm_dev(accel_dev, NULL); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + struct adf_bar *bar; + int ret; + + if (num_possible_nodes() > 1 && dev_to_node(&pdev->dev) < 0) { + /* + * If the accelerator is connected to a node with no memory + * there is no point in using the accelerator since the remote + * memory transaction will be very slow. + */ + dev_err(&pdev->dev, "Invalid NUMA configuration.\n"); + return -EINVAL; + } + + accel_dev = devm_kzalloc(&pdev->dev, sizeof(*accel_dev), GFP_KERNEL); + if (!accel_dev) + return -ENOMEM; + + INIT_LIST_HEAD(&accel_dev->crypto_list); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* + * Add accel device to accel table + * This should be called before adf_cleanup_accel is called + */ + if (adf_devmgr_add_dev(accel_dev, NULL)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + return -EFAULT; + } + + accel_dev->owner = THIS_MODULE; + /* Allocate and initialise device hardware meta-data structure */ + hw_data = devm_kzalloc(&pdev->dev, sizeof(*hw_data), GFP_KERNEL); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + + accel_dev->hw_device = hw_data; + adf_init_hw_data_4xxx(accel_dev->hw_device, ent->device); + + pci_read_config_byte(pdev, PCI_REVISION_ID, &accel_pci_dev->revid); + pci_read_config_dword(pdev, ADF_GEN4_FUSECTL4_OFFSET, &hw_data->fuses[ADF_FUSECTL4]); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + /* If the device has no acceleration engines then ignore it */ + if (!hw_data->accel_mask || !hw_data->ae_mask || + (~hw_data->ae_mask & 0x01)) { + dev_err(&pdev->dev, "No acceleration units found.\n"); + ret = -EFAULT; + goto out_err; + } + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + /* Enable PCI device */ + ret = pcim_enable_device(pdev); + if (ret) { + dev_err(&pdev->dev, "Can't enable PCI device.\n"); + goto out_err; + } + + /* Set DMA identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration.\n"); + goto out_err; + } + + ret = adf_gen4_cfg_dev_init(accel_dev); + if (ret) { + dev_err(&pdev->dev, "Failed to initialize configuration.\n"); + goto out_err; + } + + /* Get accelerator capabilities mask */ + hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev); + if (!hw_data->accel_capabilities_mask) { + dev_err(&pdev->dev, "Failed to get capabilities mask.\n"); + ret = -EINVAL; + goto out_err; + } + + /* Find and map all the device's BARS */ + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM) & ADF_GEN4_BAR_MASK; + + ret = pcim_request_all_regions(pdev, pci_name(pdev)); + if (ret) { + dev_err(&pdev->dev, "Failed to request PCI regions.\n"); + goto out_err; + } + + i = 0; + for_each_set_bit(bar_nr, &bar_mask, PCI_STD_NUM_BARS) { + bar = &accel_pci_dev->pci_bars[i++]; + bar->virt_addr = pcim_iomap(pdev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to ioremap PCI region.\n"); + ret = -ENOMEM; + goto out_err; + } + } + + pci_set_master(pdev); + + if (pci_save_state(pdev)) { + dev_err(&pdev->dev, "Failed to save pci state.\n"); + ret = -ENOMEM; + goto out_err; + } + + accel_dev->ras_errors.enabled = true; + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, true); + if (ret) + goto out_err_dev_stop; + + ret = adf_sysfs_init(accel_dev); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err: + adf_cleanup_accel(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); +} + +static void adf_shutdown(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + adf_dev_down(accel_dev); +} + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_4XXX_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, + .shutdown = adf_shutdown, + .sriov_configure = adf_sriov_configure, + .err_handler = &adf_err_handler, +}; + +module_pci_driver(adf_driver); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_FIRMWARE(ADF_4XXX_FW); +MODULE_FIRMWARE(ADF_402XX_FW); +MODULE_FIRMWARE(ADF_4XXX_MMP); +MODULE_FIRMWARE(ADF_402XX_MMP); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_SOFTDEP("pre: crypto-intel_qat"); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_6xxx/Makefile b/drivers/crypto/intel/qat/qat_6xxx/Makefile new file mode 100644 index 000000000000..4b4de67cb0c2 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_6xxx/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_6XXX) += qat_6xxx.o +qat_6xxx-y := adf_drv.o adf_6xxx_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_6xxx/adf_6xxx_hw_data.c b/drivers/crypto/intel/qat/qat_6xxx/adf_6xxx_hw_data.c new file mode 100644 index 000000000000..bed88d3ce8ca --- /dev/null +++ b/drivers/crypto/intel/qat/qat_6xxx/adf_6xxx_hw_data.c @@ -0,0 +1,950 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2025 Intel Corporation */ +#include <linux/array_size.h> +#include <linux/bitfield.h> +#include <linux/bitops.h> +#include <linux/bits.h> +#include <linux/iopoll.h> +#include <linux/pci.h> +#include <linux/types.h> + +#include <adf_accel_devices.h> +#include <adf_admin.h> +#include <adf_bank_state.h> +#include <adf_cfg.h> +#include <adf_cfg_services.h> +#include <adf_clock.h> +#include <adf_common_drv.h> +#include <adf_fw_config.h> +#include <adf_gen6_pm.h> +#include <adf_gen6_ras.h> +#include <adf_gen6_shared.h> +#include <adf_gen6_tl.h> +#include <adf_timer.h> +#include "adf_6xxx_hw_data.h" +#include "icp_qat_fw_comp.h" +#include "icp_qat_hw_51_comp.h" + +#define RP_GROUP_0_MASK (BIT(0) | BIT(2)) +#define RP_GROUP_1_MASK (BIT(1) | BIT(3)) +#define RP_GROUP_ALL_MASK (RP_GROUP_0_MASK | RP_GROUP_1_MASK) + +#define ADF_AE_GROUP_0 GENMASK(3, 0) +#define ADF_AE_GROUP_1 GENMASK(7, 4) +#define ADF_AE_GROUP_2 BIT(8) + +struct adf_ring_config { + u32 ring_mask; + enum adf_cfg_service_type ring_type; + const unsigned long *thrd_mask; +}; + +static u32 rmask_two_services[] = { + RP_GROUP_0_MASK, + RP_GROUP_1_MASK, +}; + +enum adf_gen6_rps { + RP0 = 0, + RP1 = 1, + RP2 = 2, + RP3 = 3, + RP_MAX = RP3 +}; + +/* + * thrd_mask_[sym|asym|cpr|dcc]: these static arrays define the thread + * configuration for handling requests of specific services across the + * accelerator engines. Each element in an array corresponds to an + * accelerator engine, with the value being a bitmask that specifies which + * threads within that engine are capable of processing the particular service. + * + * For example, a value of 0x0C means that threads 2 and 3 are enabled for the + * service in the respective accelerator engine. + */ +static const unsigned long thrd_mask_sym[ADF_6XXX_MAX_ACCELENGINES] = { + 0x0C, 0x0C, 0x0C, 0x0C, 0x1C, 0x1C, 0x1C, 0x1C, 0x00 +}; + +static const unsigned long thrd_mask_asym[ADF_6XXX_MAX_ACCELENGINES] = { + 0x70, 0x70, 0x70, 0x70, 0x60, 0x60, 0x60, 0x60, 0x00 +}; + +static const unsigned long thrd_mask_cpr[ADF_6XXX_MAX_ACCELENGINES] = { + 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x00 +}; + +static const unsigned long thrd_mask_dcc[ADF_6XXX_MAX_ACCELENGINES] = { + 0x00, 0x00, 0x00, 0x00, 0x07, 0x07, 0x03, 0x03, 0x00 +}; + +static const unsigned long thrd_mask_dcpr[ADF_6XXX_MAX_ACCELENGINES] = { + 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x02, 0x00 +}; + +static const char *const adf_6xxx_fw_objs[] = { + [ADF_FW_CY_OBJ] = ADF_6XXX_CY_OBJ, + [ADF_FW_DC_OBJ] = ADF_6XXX_DC_OBJ, + [ADF_FW_ADMIN_OBJ] = ADF_6XXX_ADMIN_OBJ, +}; + +static const struct adf_fw_config adf_default_fw_config[] = { + { ADF_AE_GROUP_1, ADF_FW_DC_OBJ }, + { ADF_AE_GROUP_0, ADF_FW_CY_OBJ }, + { ADF_AE_GROUP_2, ADF_FW_ADMIN_OBJ }, +}; + +static struct adf_hw_device_class adf_6xxx_class = { + .name = ADF_6XXX_DEVICE_NAME, + .type = DEV_6XXX, +}; + +static bool services_supported(unsigned long mask) +{ + int num_svc; + + if (mask >= BIT(SVC_COUNT)) + return false; + + num_svc = hweight_long(mask); + switch (num_svc) { + case ADF_ONE_SERVICE: + return true; + case ADF_TWO_SERVICES: + case ADF_THREE_SERVICES: + return !test_bit(SVC_DCC, &mask); + default: + return false; + } +} + +static int get_service(unsigned long *mask) +{ + if (test_and_clear_bit(SVC_ASYM, mask)) + return SVC_ASYM; + + if (test_and_clear_bit(SVC_SYM, mask)) + return SVC_SYM; + + if (test_and_clear_bit(SVC_DC, mask)) + return SVC_DC; + + if (test_and_clear_bit(SVC_DCC, mask)) + return SVC_DCC; + + if (test_and_clear_bit(SVC_DECOMP, mask)) + return SVC_DECOMP; + + return -EINVAL; +} + +static enum adf_cfg_service_type get_ring_type(unsigned int service) +{ + switch (service) { + case SVC_SYM: + return SYM; + case SVC_ASYM: + return ASYM; + case SVC_DC: + case SVC_DCC: + return COMP; + case SVC_DECOMP: + return DECOMP; + default: + return UNUSED; + } +} + +static const unsigned long *get_thrd_mask(unsigned int service) +{ + switch (service) { + case SVC_SYM: + return thrd_mask_sym; + case SVC_ASYM: + return thrd_mask_asym; + case SVC_DC: + return thrd_mask_cpr; + case SVC_DCC: + return thrd_mask_dcc; + case SVC_DECOMP: + return thrd_mask_dcpr; + default: + return NULL; + } +} + +static int get_rp_config(struct adf_accel_dev *accel_dev, struct adf_ring_config *rp_config, + unsigned int *num_services) +{ + unsigned int i, nservices; + unsigned long mask; + int ret, service; + + ret = adf_get_service_mask(accel_dev, &mask); + if (ret) + return ret; + + nservices = hweight_long(mask); + if (nservices > MAX_NUM_CONCURR_SVC) + return -EINVAL; + + for (i = 0; i < nservices; i++) { + service = get_service(&mask); + if (service < 0) + return service; + + rp_config[i].ring_type = get_ring_type(service); + rp_config[i].thrd_mask = get_thrd_mask(service); + + /* + * If there is only one service enabled, use all ring pairs for + * that service. + * If there are two services enabled, use ring pairs 0 and 2 for + * one service and ring pairs 1 and 3 for the other service. + */ + switch (nservices) { + case ADF_ONE_SERVICE: + rp_config[i].ring_mask = RP_GROUP_ALL_MASK; + break; + case ADF_TWO_SERVICES: + rp_config[i].ring_mask = rmask_two_services[i]; + break; + case ADF_THREE_SERVICES: + rp_config[i].ring_mask = BIT(i); + + /* If ASYM is enabled, use additional ring pair */ + if (service == SVC_ASYM) + rp_config[i].ring_mask |= BIT(RP3); + + break; + default: + return -EINVAL; + } + } + + *num_services = nservices; + + return 0; +} + +static u32 adf_gen6_get_arb_mask(struct adf_accel_dev *accel_dev, unsigned int ae) +{ + struct adf_ring_config rp_config[MAX_NUM_CONCURR_SVC]; + unsigned int num_services, i, thrd; + u32 ring_mask, thd2arb_mask = 0; + const unsigned long *p_mask; + + if (get_rp_config(accel_dev, rp_config, &num_services)) + return 0; + + /* + * The thd2arb_mask maps ring pairs to threads within an accelerator engine. + * It ensures that jobs submitted to ring pairs are scheduled on threads capable + * of handling the specified service type. + * + * Each group of 4 bits in the mask corresponds to a thread, with each bit + * indicating whether a job from a ring pair can be scheduled on that thread. + * The use of 4 bits is due to the organization of ring pairs into groups of + * four, where each group shares the same configuration. + */ + for (i = 0; i < num_services; i++) { + p_mask = &rp_config[i].thrd_mask[ae]; + ring_mask = rp_config[i].ring_mask; + + for_each_set_bit(thrd, p_mask, ADF_NUM_THREADS_PER_AE) + thd2arb_mask |= ring_mask << (thrd * 4); + } + + return thd2arb_mask; +} + +static u16 get_ring_to_svc_map(struct adf_accel_dev *accel_dev) +{ + enum adf_cfg_service_type rps[ADF_GEN6_NUM_BANKS_PER_VF] = { }; + struct adf_ring_config rp_config[MAX_NUM_CONCURR_SVC]; + unsigned int num_services, rp_num, i; + unsigned long cfg_mask; + u16 ring_to_svc_map; + + if (get_rp_config(accel_dev, rp_config, &num_services)) + return 0; + + /* + * Loop through the configured services and populate the `rps` array that + * contains what service that particular ring pair can handle (i.e. symmetric + * crypto, asymmetric crypto, data compression or compression chaining). + */ + for (i = 0; i < num_services; i++) { + cfg_mask = rp_config[i].ring_mask; + for_each_set_bit(rp_num, &cfg_mask, ADF_GEN6_NUM_BANKS_PER_VF) + rps[rp_num] = rp_config[i].ring_type; + } + + /* + * The ring_mask is structured into segments of 3 bits, with each + * segment representing the service configuration for a specific ring pair. + * Since ring pairs are organized into groups of 4, the ring_mask contains 4 + * such 3-bit segments, each corresponding to one ring pair. + * + * The device has 64 ring pairs, which are organized in groups of 4, namely + * 16 groups. Each group has the same configuration, represented here by + * `ring_to_svc_map`. + */ + ring_to_svc_map = rps[RP0] << ADF_CFG_SERV_RING_PAIR_0_SHIFT | + rps[RP1] << ADF_CFG_SERV_RING_PAIR_1_SHIFT | + rps[RP2] << ADF_CFG_SERV_RING_PAIR_2_SHIFT | + rps[RP3] << ADF_CFG_SERV_RING_PAIR_3_SHIFT; + + return ring_to_svc_map; +} + +static u32 get_accel_mask(struct adf_hw_device_data *self) +{ + return ADF_GEN6_ACCELERATORS_MASK; +} + +static u32 get_num_accels(struct adf_hw_device_data *self) +{ + return ADF_GEN6_MAX_ACCELERATORS; +} + +static u32 get_num_aes(struct adf_hw_device_data *self) +{ + return self ? hweight32(self->ae_mask) : 0; +} + +static u32 get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_GEN6_PMISC_BAR; +} + +static u32 get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_GEN6_ETR_BAR; +} + +static u32 get_sram_bar_id(struct adf_hw_device_data *self) +{ + return ADF_GEN6_SRAM_BAR; +} + +static enum dev_sku_info get_sku(struct adf_hw_device_data *self) +{ + return DEV_SKU_1; +} + +static void get_arb_info(struct arb_info *arb_info) +{ + arb_info->arb_cfg = ADF_GEN6_ARB_CONFIG; + arb_info->arb_offset = ADF_GEN6_ARB_OFFSET; + arb_info->wt2sam_offset = ADF_GEN6_ARB_WRK_2_SER_MAP_OFFSET; +} + +static void get_admin_info(struct admin_info *admin_csrs_info) +{ + admin_csrs_info->mailbox_offset = ADF_GEN6_MAILBOX_BASE_OFFSET; + admin_csrs_info->admin_msg_ur = ADF_GEN6_ADMINMSGUR_OFFSET; + admin_csrs_info->admin_msg_lr = ADF_GEN6_ADMINMSGLR_OFFSET; +} + +static u32 get_heartbeat_clock(struct adf_hw_device_data *self) +{ + return ADF_GEN6_COUNTER_FREQ; +} + +static void enable_error_correction(struct adf_accel_dev *accel_dev) +{ + void __iomem *csr = adf_get_pmisc_base(accel_dev); + + /* + * Enable all error notification bits in errsou3 except VFLR + * notification on host. + */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK3, ADF_GEN6_VFLNOTIFY); +} + +static void enable_ints(struct adf_accel_dev *accel_dev) +{ + void __iomem *addr = adf_get_pmisc_base(accel_dev); + + /* Enable bundle interrupts */ + ADF_CSR_WR(addr, ADF_GEN6_SMIAPF_RP_X0_MASK_OFFSET, 0); + ADF_CSR_WR(addr, ADF_GEN6_SMIAPF_RP_X1_MASK_OFFSET, 0); + + /* Enable misc interrupts */ + ADF_CSR_WR(addr, ADF_GEN6_SMIAPF_MASK_OFFSET, 0); +} + +static void set_ssm_wdtimer(struct adf_accel_dev *accel_dev) +{ + void __iomem *addr = adf_get_pmisc_base(accel_dev); + u64 val_pke = ADF_SSM_WDT_PKE_DEFAULT_VALUE; + u64 val = ADF_SSM_WDT_DEFAULT_VALUE; + + /* Enable watchdog timer for sym and dc */ + ADF_CSR_WR64_LO_HI(addr, ADF_SSMWDTATHL_OFFSET, ADF_SSMWDTATHH_OFFSET, val); + ADF_CSR_WR64_LO_HI(addr, ADF_SSMWDTCNVL_OFFSET, ADF_SSMWDTCNVH_OFFSET, val); + ADF_CSR_WR64_LO_HI(addr, ADF_SSMWDTUCSL_OFFSET, ADF_SSMWDTUCSH_OFFSET, val); + ADF_CSR_WR64_LO_HI(addr, ADF_SSMWDTDCPRL_OFFSET, ADF_SSMWDTDCPRH_OFFSET, val); + + /* Enable watchdog timer for pke */ + ADF_CSR_WR64_LO_HI(addr, ADF_SSMWDTPKEL_OFFSET, ADF_SSMWDTPKEH_OFFSET, val_pke); +} + +/* + * The vector routing table is used to select the MSI-X entry to use for each + * interrupt source. + * The first ADF_GEN6_ETR_MAX_BANKS entries correspond to ring interrupts. + * The final entry corresponds to VF2PF or error interrupts. + * This vector table could be used to configure one MSI-X entry to be shared + * between multiple interrupt sources. + * + * The default routing is set to have a one to one correspondence between the + * interrupt source and the MSI-X entry used. + */ +static void set_msix_default_rttable(struct adf_accel_dev *accel_dev) +{ + void __iomem *csr = adf_get_pmisc_base(accel_dev); + unsigned int i; + + for (i = 0; i <= ADF_GEN6_ETR_MAX_BANKS; i++) + ADF_CSR_WR(csr, ADF_GEN6_MSIX_RTTABLE_OFFSET(i), i); +} + +static int reset_ring_pair(void __iomem *csr, u32 bank_number) +{ + u32 status; + int ret; + + /* + * Write rpresetctl register BIT(0) as 1. + * Since rpresetctl registers have no RW fields, no need to preserve + * values for other bits. Just write directly. + */ + ADF_CSR_WR(csr, ADF_WQM_CSR_RPRESETCTL(bank_number), + ADF_WQM_CSR_RPRESETCTL_RESET); + + /* Read rpresetsts register and wait for rp reset to complete */ + ret = read_poll_timeout(ADF_CSR_RD, status, + status & ADF_WQM_CSR_RPRESETSTS_STATUS, + ADF_RPRESET_POLL_DELAY_US, + ADF_RPRESET_POLL_TIMEOUT_US, true, + csr, ADF_WQM_CSR_RPRESETSTS(bank_number)); + if (ret) + return ret; + + /* When ring pair reset is done, clear rpresetsts */ + ADF_CSR_WR(csr, ADF_WQM_CSR_RPRESETSTS(bank_number), ADF_WQM_CSR_RPRESETSTS_STATUS); + + return 0; +} + +static int ring_pair_reset(struct adf_accel_dev *accel_dev, u32 bank_number) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + void __iomem *csr = adf_get_etr_base(accel_dev); + int ret; + + if (bank_number >= hw_data->num_banks) + return -EINVAL; + + dev_dbg(&GET_DEV(accel_dev), "ring pair reset for bank:%d\n", bank_number); + + ret = reset_ring_pair(csr, bank_number); + if (ret) + dev_err(&GET_DEV(accel_dev), "ring pair reset failed (timeout)\n"); + else + dev_dbg(&GET_DEV(accel_dev), "ring pair reset successful\n"); + + return ret; +} + +static int build_comp_block(void *ctx, enum adf_dc_algo algo) +{ + struct icp_qat_fw_comp_req *req_tmpl = ctx; + struct icp_qat_fw_comp_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_hw_comp_51_config_csr_lower hw_comp_lower_csr = { }; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + u32 lower_val; + + switch (algo) { + case QAT_DEFLATE: + header->service_cmd_id = ICP_QAT_FW_COMP_CMD_DYNAMIC; + break; + default: + return -EINVAL; + } + + hw_comp_lower_csr.lllbd = ICP_QAT_HW_COMP_51_LLLBD_CTRL_LLLBD_DISABLED; + hw_comp_lower_csr.sd = ICP_QAT_HW_COMP_51_SEARCH_DEPTH_LEVEL_1; + lower_val = ICP_QAT_FW_COMP_51_BUILD_CONFIG_LOWER(hw_comp_lower_csr); + cd_pars->u.sl.comp_slice_cfg_word[0] = lower_val; + cd_pars->u.sl.comp_slice_cfg_word[1] = 0; + + return 0; +} + +static int build_decomp_block(void *ctx, enum adf_dc_algo algo) +{ + struct icp_qat_fw_comp_req *req_tmpl = ctx; + struct icp_qat_fw_comp_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + + switch (algo) { + case QAT_DEFLATE: + header->service_cmd_id = ICP_QAT_FW_COMP_CMD_DECOMPRESS; + break; + default: + return -EINVAL; + } + + cd_pars->u.sl.comp_slice_cfg_word[0] = 0; + cd_pars->u.sl.comp_slice_cfg_word[1] = 0; + + return 0; +} + +static void adf_gen6_init_dc_ops(struct adf_dc_ops *dc_ops) +{ + dc_ops->build_comp_block = build_comp_block; + dc_ops->build_decomp_block = build_decomp_block; +} + +static int adf_gen6_init_thd2arb_map(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + u32 *thd2arb_map = hw_data->thd_to_arb_map; + unsigned int i; + + for (i = 0; i < hw_data->num_engines; i++) { + thd2arb_map[i] = adf_gen6_get_arb_mask(accel_dev, i); + dev_dbg(&GET_DEV(accel_dev), "ME:%d arb_mask:%#x\n", i, thd2arb_map[i]); + } + + return 0; +} + +static void init_num_svc_aes(struct adf_rl_hw_data *device_data) +{ + enum adf_fw_objs obj_type, obj_iter; + unsigned int svc, i, num_grp; + u32 ae_mask; + + for (svc = 0; svc < SVC_BASE_COUNT; svc++) { + switch (svc) { + case SVC_SYM: + case SVC_ASYM: + obj_type = ADF_FW_CY_OBJ; + break; + case SVC_DC: + case SVC_DECOMP: + obj_type = ADF_FW_DC_OBJ; + break; + } + + num_grp = ARRAY_SIZE(adf_default_fw_config); + for (i = 0; i < num_grp; i++) { + obj_iter = adf_default_fw_config[i].obj; + if (obj_iter == obj_type) { + ae_mask = adf_default_fw_config[i].ae_mask; + device_data->svc_ae_mask[svc] = hweight32(ae_mask); + break; + } + } + } +} + +static u32 adf_gen6_get_svc_slice_cnt(struct adf_accel_dev *accel_dev, + enum adf_base_services svc) +{ + struct adf_rl_hw_data *device_data = &accel_dev->hw_device->rl_data; + + switch (svc) { + case SVC_SYM: + return device_data->slices.cph_cnt; + case SVC_ASYM: + return device_data->slices.pke_cnt; + case SVC_DC: + return device_data->slices.cpr_cnt + device_data->slices.dcpr_cnt; + case SVC_DECOMP: + return device_data->slices.dcpr_cnt; + default: + return 0; + } +} + +static void set_vc_csr_for_bank(void __iomem *csr, u32 bank_number) +{ + u32 value; + + /* + * After each PF FLR, for each of the 64 ring pairs in the PF, the + * driver must program the ringmodectl CSRs. + */ + value = ADF_CSR_RD(csr, ADF_GEN6_CSR_RINGMODECTL(bank_number)); + FIELD_MODIFY(ADF_GEN6_RINGMODECTL_TC_MASK, &value, ADF_GEN6_RINGMODECTL_TC_DEFAULT); + FIELD_MODIFY(ADF_GEN6_RINGMODECTL_TC_EN_MASK, &value, ADF_GEN6_RINGMODECTL_TC_EN_OP1); + ADF_CSR_WR(csr, ADF_GEN6_CSR_RINGMODECTL(bank_number), value); +} + +static int set_vc_config(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + u32 value; + int err; + + /* + * After each PF FLR, the driver must program the Port Virtual Channel (VC) + * Control Registers. + * Read PVC0CTL then write the masked values. + */ + pci_read_config_dword(pdev, ADF_GEN6_PVC0CTL_OFFSET, &value); + FIELD_MODIFY(ADF_GEN6_PVC0CTL_TCVCMAP_MASK, &value, ADF_GEN6_PVC0CTL_TCVCMAP_DEFAULT); + err = pci_write_config_dword(pdev, ADF_GEN6_PVC0CTL_OFFSET, value); + if (err) { + dev_err(&GET_DEV(accel_dev), "pci write to PVC0CTL failed\n"); + return pcibios_err_to_errno(err); + } + + /* Read PVC1CTL then write masked values */ + pci_read_config_dword(pdev, ADF_GEN6_PVC1CTL_OFFSET, &value); + FIELD_MODIFY(ADF_GEN6_PVC1CTL_TCVCMAP_MASK, &value, ADF_GEN6_PVC1CTL_TCVCMAP_DEFAULT); + FIELD_MODIFY(ADF_GEN6_PVC1CTL_VCEN_MASK, &value, ADF_GEN6_PVC1CTL_VCEN_ON); + err = pci_write_config_dword(pdev, ADF_GEN6_PVC1CTL_OFFSET, value); + if (err) + dev_err(&GET_DEV(accel_dev), "pci write to PVC1CTL failed\n"); + + return pcibios_err_to_errno(err); +} + +static int adf_gen6_set_vc(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + void __iomem *csr = adf_get_etr_base(accel_dev); + u32 i; + + for (i = 0; i < hw_data->num_banks; i++) { + dev_dbg(&GET_DEV(accel_dev), "set virtual channels for bank:%d\n", i); + set_vc_csr_for_bank(csr, i); + } + + return set_vc_config(accel_dev); +} + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + unsigned long fuses = self->fuses[ADF_FUSECTL4]; + u32 mask = ADF_6XXX_ACCELENGINES_MASK; + + /* + * If bit 0 is set in the fuses, the first 4 engines are disabled. + * If bit 4 is set, the second group of 4 engines are disabled. + * If bit 8 is set, the admin engine (bit 8) is disabled. + */ + if (test_bit(0, &fuses)) + mask &= ~ADF_AE_GROUP_0; + + if (test_bit(4, &fuses)) + mask &= ~ADF_AE_GROUP_1; + + if (test_bit(8, &fuses)) + mask &= ~ADF_AE_GROUP_2; + + return mask; +} + +static u32 get_accel_cap(struct adf_accel_dev *accel_dev) +{ + u32 capabilities_sym, capabilities_asym; + u32 capabilities_dc; + unsigned long mask; + u32 caps = 0; + u32 fusectl1; + + fusectl1 = GET_HW_DATA(accel_dev)->fuses[ADF_FUSECTL1]; + + /* Read accelerator capabilities mask */ + capabilities_sym = ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC | + ICP_ACCEL_CAPABILITIES_CIPHER | + ICP_ACCEL_CAPABILITIES_AUTHENTICATION | + ICP_ACCEL_CAPABILITIES_SHA3 | + ICP_ACCEL_CAPABILITIES_SHA3_EXT | + ICP_ACCEL_CAPABILITIES_CHACHA_POLY | + ICP_ACCEL_CAPABILITIES_AESGCM_SPC | + ICP_ACCEL_CAPABILITIES_AES_V2; + + /* A set bit in fusectl1 means the corresponding feature is OFF in this SKU */ + if (fusectl1 & ICP_ACCEL_GEN6_MASK_UCS_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CHACHA_POLY; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AESGCM_SPC; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AES_V2; + } + if (fusectl1 & ICP_ACCEL_GEN6_MASK_AUTH_SLICE) { + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_AUTHENTICATION; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SHA3; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_SHA3_EXT; + capabilities_sym &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + + capabilities_asym = ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC | + ICP_ACCEL_CAPABILITIES_SM2 | + ICP_ACCEL_CAPABILITIES_ECEDMONT; + + if (fusectl1 & ICP_ACCEL_GEN6_MASK_PKE_SLICE) { + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC; + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_SM2; + capabilities_asym &= ~ICP_ACCEL_CAPABILITIES_ECEDMONT; + } + + capabilities_dc = ICP_ACCEL_CAPABILITIES_COMPRESSION | + ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION | + ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION | + ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64; + + if (fusectl1 & ICP_ACCEL_GEN6_MASK_CPR_SLICE) { + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION; + capabilities_dc &= ~ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64; + } + + if (adf_get_service_mask(accel_dev, &mask)) + return 0; + + if (test_bit(SVC_ASYM, &mask)) + caps |= capabilities_asym; + if (test_bit(SVC_SYM, &mask)) + caps |= capabilities_sym; + if (test_bit(SVC_DC, &mask) || test_bit(SVC_DECOMP, &mask)) + caps |= capabilities_dc; + if (test_bit(SVC_DCC, &mask)) { + /* + * Sym capabilities are available for chaining operations, + * but sym crypto instances cannot be supported + */ + caps = capabilities_dc | capabilities_sym; + caps &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + } + + return caps; +} + +static u32 uof_get_num_objs(struct adf_accel_dev *accel_dev) +{ + return ARRAY_SIZE(adf_default_fw_config); +} + +static const char *uof_get_name(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + int num_fw_objs = ARRAY_SIZE(adf_6xxx_fw_objs); + int id; + + id = adf_default_fw_config[obj_num].obj; + if (id >= num_fw_objs) + return NULL; + + return adf_6xxx_fw_objs[id]; +} + +static const char *uof_get_name_6xxx(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + return uof_get_name(accel_dev, obj_num); +} + +static int uof_get_obj_type(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + if (obj_num >= uof_get_num_objs(accel_dev)) + return -EINVAL; + + return adf_default_fw_config[obj_num].obj; +} + +static u32 uof_get_ae_mask(struct adf_accel_dev *accel_dev, u32 obj_num) +{ + return adf_default_fw_config[obj_num].ae_mask; +} + +static const u32 *adf_get_arbiter_mapping(struct adf_accel_dev *accel_dev) +{ + if (adf_gen6_init_thd2arb_map(accel_dev)) + dev_warn(&GET_DEV(accel_dev), + "Failed to generate thread to arbiter mapping"); + + return GET_HW_DATA(accel_dev)->thd_to_arb_map; +} + +static int adf_init_device(struct adf_accel_dev *accel_dev) +{ + void __iomem *addr = adf_get_pmisc_base(accel_dev); + u32 status; + u32 csr; + int ret; + + /* Temporarily mask PM interrupt */ + csr = ADF_CSR_RD(addr, ADF_GEN6_ERRMSK2); + csr |= ADF_GEN6_PM_SOU; + ADF_CSR_WR(addr, ADF_GEN6_ERRMSK2, csr); + + /* Set DRV_ACTIVE bit to power up the device */ + ADF_CSR_WR(addr, ADF_GEN6_PM_INTERRUPT, ADF_GEN6_PM_DRV_ACTIVE); + + /* Poll status register to make sure the device is powered up */ + ret = read_poll_timeout(ADF_CSR_RD, status, + status & ADF_GEN6_PM_INIT_STATE, + ADF_GEN6_PM_POLL_DELAY_US, + ADF_GEN6_PM_POLL_TIMEOUT_US, true, addr, + ADF_GEN6_PM_STATUS); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Failed to power up the device\n"); + return ret; + } + + dev_dbg(&GET_DEV(accel_dev), "Setting virtual channels for device qat_dev%d\n", + accel_dev->accel_id); + + ret = adf_gen6_set_vc(accel_dev); + if (ret) + dev_err(&GET_DEV(accel_dev), "Failed to set virtual channels\n"); + + return ret; +} + +static int enable_pm(struct adf_accel_dev *accel_dev) +{ + int ret; + + ret = adf_init_admin_pm(accel_dev, ADF_GEN6_PM_DEFAULT_IDLE_FILTER); + if (ret) + return ret; + + /* Initialize PM internal data */ + adf_gen6_init_dev_pm_data(accel_dev); + + return 0; +} + +static int dev_config(struct adf_accel_dev *accel_dev) +{ + int ret; + + ret = adf_cfg_section_add(accel_dev, ADF_KERNEL_SEC); + if (ret) + return ret; + + ret = adf_cfg_section_add(accel_dev, "Accelerator0"); + if (ret) + return ret; + + switch (adf_get_service_enabled(accel_dev)) { + case SVC_DC: + case SVC_DCC: + ret = adf_gen6_comp_dev_config(accel_dev); + break; + default: + ret = adf_gen6_no_dev_config(accel_dev); + break; + } + if (ret) + return ret; + + __set_bit(ADF_STATUS_CONFIGURED, &accel_dev->status); + + return ret; +} + +static void adf_gen6_init_rl_data(struct adf_rl_hw_data *rl_data) +{ + rl_data->pciout_tb_offset = ADF_GEN6_RL_TOKEN_PCIEOUT_BUCKET_OFFSET; + rl_data->pciin_tb_offset = ADF_GEN6_RL_TOKEN_PCIEIN_BUCKET_OFFSET; + rl_data->r2l_offset = ADF_GEN6_RL_R2L_OFFSET; + rl_data->l2c_offset = ADF_GEN6_RL_L2C_OFFSET; + rl_data->c2s_offset = ADF_GEN6_RL_C2S_OFFSET; + rl_data->pcie_scale_div = ADF_6XXX_RL_PCIE_SCALE_FACTOR_DIV; + rl_data->pcie_scale_mul = ADF_6XXX_RL_PCIE_SCALE_FACTOR_MUL; + rl_data->max_tp[SVC_ASYM] = ADF_6XXX_RL_MAX_TP_ASYM; + rl_data->max_tp[SVC_SYM] = ADF_6XXX_RL_MAX_TP_SYM; + rl_data->max_tp[SVC_DC] = ADF_6XXX_RL_MAX_TP_DC; + rl_data->max_tp[SVC_DECOMP] = ADF_6XXX_RL_MAX_TP_DECOMP; + rl_data->scan_interval = ADF_6XXX_RL_SCANS_PER_SEC; + rl_data->scale_ref = ADF_6XXX_RL_SLICE_REF; + + init_num_svc_aes(rl_data); +} + +void adf_init_hw_data_6xxx(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class = &adf_6xxx_class; + hw_data->instance_id = adf_6xxx_class.instances++; + hw_data->num_banks = ADF_GEN6_ETR_MAX_BANKS; + hw_data->num_banks_per_vf = ADF_GEN6_NUM_BANKS_PER_VF; + hw_data->num_rings_per_bank = ADF_GEN6_NUM_RINGS_PER_BANK; + hw_data->num_accel = ADF_GEN6_MAX_ACCELERATORS; + hw_data->num_engines = ADF_6XXX_MAX_ACCELENGINES; + hw_data->num_logical_accel = 1; + hw_data->tx_rx_gap = ADF_GEN6_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_GEN6_TX_RINGS_MASK; + hw_data->ring_to_svc_map = 0; + hw_data->alloc_irq = adf_isr_resource_alloc; + hw_data->free_irq = adf_isr_resource_free; + hw_data->enable_error_correction = enable_error_correction; + hw_data->get_accel_mask = get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_num_accels = get_num_accels; + hw_data->get_num_aes = get_num_aes; + hw_data->get_sram_bar_id = get_sram_bar_id; + hw_data->get_etr_bar_id = get_etr_bar_id; + hw_data->get_misc_bar_id = get_misc_bar_id; + hw_data->get_arb_info = get_arb_info; + hw_data->get_admin_info = get_admin_info; + hw_data->get_accel_cap = get_accel_cap; + hw_data->get_sku = get_sku; + hw_data->init_admin_comms = adf_init_admin_comms; + hw_data->exit_admin_comms = adf_exit_admin_comms; + hw_data->send_admin_init = adf_send_admin_init; + hw_data->init_arb = adf_init_arb; + hw_data->exit_arb = adf_exit_arb; + hw_data->get_arb_mapping = adf_get_arbiter_mapping; + hw_data->enable_ints = enable_ints; + hw_data->reset_device = adf_reset_flr; + hw_data->admin_ae_mask = ADF_6XXX_ADMIN_AE_MASK; + hw_data->fw_name = ADF_6XXX_FW; + hw_data->fw_mmp_name = ADF_6XXX_MMP; + hw_data->uof_get_name = uof_get_name_6xxx; + hw_data->uof_get_num_objs = uof_get_num_objs; + hw_data->uof_get_obj_type = uof_get_obj_type; + hw_data->uof_get_ae_mask = uof_get_ae_mask; + hw_data->set_msix_rttable = set_msix_default_rttable; + hw_data->set_ssm_wdtimer = set_ssm_wdtimer; + hw_data->get_ring_to_svc_map = get_ring_to_svc_map; + hw_data->disable_iov = adf_disable_sriov; + hw_data->ring_pair_reset = ring_pair_reset; + hw_data->dev_config = dev_config; + hw_data->bank_state_save = adf_bank_state_save; + hw_data->bank_state_restore = adf_bank_state_restore; + hw_data->get_hb_clock = get_heartbeat_clock; + hw_data->num_hb_ctrs = ADF_NUM_HB_CNT_PER_AE; + hw_data->start_timer = adf_timer_start; + hw_data->stop_timer = adf_timer_stop; + hw_data->init_device = adf_init_device; + hw_data->enable_pm = enable_pm; + hw_data->services_supported = services_supported; + hw_data->num_rps = ADF_GEN6_ETR_MAX_BANKS; + hw_data->clock_frequency = ADF_6XXX_AE_FREQ; + hw_data->get_svc_slice_cnt = adf_gen6_get_svc_slice_cnt; + + adf_gen6_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen6_init_pf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen6_init_dc_ops(&hw_data->dc_ops); + adf_gen6_init_vf_mig_ops(&hw_data->vfmig_ops); + adf_gen6_init_ras_ops(&hw_data->ras_ops); + adf_gen6_init_tl_data(&hw_data->tl_data); + adf_gen6_init_rl_data(&hw_data->rl_data); +} + +void adf_clean_hw_data_6xxx(struct adf_hw_device_data *hw_data) +{ + if (hw_data->dev_class->instances) + hw_data->dev_class->instances--; +} diff --git a/drivers/crypto/intel/qat/qat_6xxx/adf_6xxx_hw_data.h b/drivers/crypto/intel/qat/qat_6xxx/adf_6xxx_hw_data.h new file mode 100644 index 000000000000..d822911fe68c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_6xxx/adf_6xxx_hw_data.h @@ -0,0 +1,168 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ADF_6XXX_HW_DATA_H_ +#define ADF_6XXX_HW_DATA_H_ + +#include <linux/bits.h> +#include <linux/time.h> +#include <linux/units.h> + +#include "adf_accel_devices.h" +#include "adf_cfg_common.h" +#include "adf_dc.h" + +/* PCIe configuration space */ +#define ADF_GEN6_BAR_MASK (BIT(0) | BIT(2) | BIT(4)) +#define ADF_GEN6_SRAM_BAR 0 +#define ADF_GEN6_PMISC_BAR 1 +#define ADF_GEN6_ETR_BAR 2 +#define ADF_6XXX_MAX_ACCELENGINES 9 + +/* Clocks frequency */ +#define ADF_GEN6_COUNTER_FREQ (100 * HZ_PER_MHZ) + +/* Physical function fuses */ +#define ADF_GEN6_FUSECTL0_OFFSET 0x2C8 +#define ADF_GEN6_FUSECTL1_OFFSET 0x2CC +#define ADF_GEN6_FUSECTL4_OFFSET 0x2D8 + +/* Accelerators */ +#define ADF_GEN6_ACCELERATORS_MASK 0x1 +#define ADF_GEN6_MAX_ACCELERATORS 1 + +/* MSI-X interrupt */ +#define ADF_GEN6_SMIAPF_RP_X0_MASK_OFFSET 0x41A040 +#define ADF_GEN6_SMIAPF_RP_X1_MASK_OFFSET 0x41A044 +#define ADF_GEN6_SMIAPF_MASK_OFFSET 0x41A084 +#define ADF_GEN6_MSIX_RTTABLE_OFFSET(i) (0x409000 + ((i) * 4)) + +/* Bank and ring configuration */ +#define ADF_GEN6_NUM_RINGS_PER_BANK 2 +#define ADF_GEN6_NUM_BANKS_PER_VF 4 +#define ADF_GEN6_ETR_MAX_BANKS 64 +#define ADF_GEN6_RX_RINGS_OFFSET 1 +#define ADF_GEN6_TX_RINGS_MASK 0x1 + +/* Arbiter configuration */ +#define ADF_GEN6_ARB_CONFIG (BIT(31) | BIT(6) | BIT(0)) +#define ADF_GEN6_ARB_OFFSET 0x000 +#define ADF_GEN6_ARB_WRK_2_SER_MAP_OFFSET 0x400 + +/* Admin interface configuration */ +#define ADF_GEN6_ADMINMSGUR_OFFSET 0x500574 +#define ADF_GEN6_ADMINMSGLR_OFFSET 0x500578 +#define ADF_GEN6_MAILBOX_BASE_OFFSET 0x600970 + +/* + * Watchdog timers + * Timeout is in cycles. Clock speed may vary across products but this + * value should be a few milli-seconds. + */ +#define ADF_SSM_WDT_DEFAULT_VALUE 0x7000000ULL +#define ADF_SSM_WDT_PKE_DEFAULT_VALUE 0x8000000ULL +#define ADF_SSMWDTATHL_OFFSET 0x5208 +#define ADF_SSMWDTATHH_OFFSET 0x520C +#define ADF_SSMWDTCNVL_OFFSET 0x5408 +#define ADF_SSMWDTCNVH_OFFSET 0x540C +#define ADF_SSMWDTUCSL_OFFSET 0x5808 +#define ADF_SSMWDTUCSH_OFFSET 0x580C +#define ADF_SSMWDTDCPRL_OFFSET 0x5A08 +#define ADF_SSMWDTDCPRH_OFFSET 0x5A0C +#define ADF_SSMWDTPKEL_OFFSET 0x5E08 +#define ADF_SSMWDTPKEH_OFFSET 0x5E0C + +/* Ring reset */ +#define ADF_RPRESET_POLL_TIMEOUT_US (5 * USEC_PER_SEC) +#define ADF_RPRESET_POLL_DELAY_US 20 +#define ADF_WQM_CSR_RPRESETCTL_RESET BIT(0) +#define ADF_WQM_CSR_RPRESETCTL(bank) (0x6000 + (bank) * 8) +#define ADF_WQM_CSR_RPRESETSTS_STATUS BIT(0) +#define ADF_WQM_CSR_RPRESETSTS(bank) (ADF_WQM_CSR_RPRESETCTL(bank) + 4) + +/* Controls and sets up the corresponding ring mode of operation */ +#define ADF_GEN6_CSR_RINGMODECTL(bank) (0x9000 + (bank) * 4) + +/* Specifies the traffic class to use for the transactions to/from the ring */ +#define ADF_GEN6_RINGMODECTL_TC_MASK GENMASK(18, 16) +#define ADF_GEN6_RINGMODECTL_TC_DEFAULT 0x7 + +/* Specifies usage of tc for the transactions to/from this ring */ +#define ADF_GEN6_RINGMODECTL_TC_EN_MASK GENMASK(20, 19) + +/* + * Use the value programmed in the tc field for request descriptor + * and metadata read transactions + */ +#define ADF_GEN6_RINGMODECTL_TC_EN_OP1 0x1 + +/* VC0 Resource Control Register */ +#define ADF_GEN6_PVC0CTL_OFFSET 0x204 +#define ADF_GEN6_PVC0CTL_TCVCMAP_OFFSET 1 +#define ADF_GEN6_PVC0CTL_TCVCMAP_MASK GENMASK(7, 1) +#define ADF_GEN6_PVC0CTL_TCVCMAP_DEFAULT 0x3F + +/* VC1 Resource Control Register */ +#define ADF_GEN6_PVC1CTL_OFFSET 0x210 +#define ADF_GEN6_PVC1CTL_TCVCMAP_OFFSET 1 +#define ADF_GEN6_PVC1CTL_TCVCMAP_MASK GENMASK(7, 1) +#define ADF_GEN6_PVC1CTL_TCVCMAP_DEFAULT 0x40 +#define ADF_GEN6_PVC1CTL_VCEN_OFFSET 31 +#define ADF_GEN6_PVC1CTL_VCEN_MASK BIT(31) +/* RW bit: 0x1 - enables a Virtual Channel, 0x0 - disables */ +#define ADF_GEN6_PVC1CTL_VCEN_ON 0x1 + +/* Error source mask registers */ +#define ADF_GEN6_ERRMSK0 0x41A210 +#define ADF_GEN6_ERRMSK1 0x41A214 +#define ADF_GEN6_ERRMSK2 0x41A218 +#define ADF_GEN6_ERRMSK3 0x41A21C + +#define ADF_GEN6_VFLNOTIFY BIT(7) + +/* Number of heartbeat counter pairs */ +#define ADF_NUM_HB_CNT_PER_AE ADF_NUM_THREADS_PER_AE + +/* Rate Limiting */ +#define ADF_GEN6_RL_R2L_OFFSET 0x508000 +#define ADF_GEN6_RL_L2C_OFFSET 0x509000 +#define ADF_GEN6_RL_C2S_OFFSET 0x508818 +#define ADF_GEN6_RL_TOKEN_PCIEIN_BUCKET_OFFSET 0x508800 +#define ADF_GEN6_RL_TOKEN_PCIEOUT_BUCKET_OFFSET 0x508804 + +/* Physical function fuses */ +#define ADF_6XXX_ACCELENGINES_MASK GENMASK(8, 0) +#define ADF_6XXX_ADMIN_AE_MASK GENMASK(8, 8) + +/* Firmware binaries */ +#define ADF_6XXX_FW "qat_6xxx.bin" +#define ADF_6XXX_MMP "qat_6xxx_mmp.bin" +#define ADF_6XXX_CY_OBJ "qat_6xxx_cy.bin" +#define ADF_6XXX_DC_OBJ "qat_6xxx_dc.bin" +#define ADF_6XXX_ADMIN_OBJ "qat_6xxx_admin.bin" + +/* RL constants */ +#define ADF_6XXX_RL_PCIE_SCALE_FACTOR_DIV 100 +#define ADF_6XXX_RL_PCIE_SCALE_FACTOR_MUL 102 +#define ADF_6XXX_RL_SCANS_PER_SEC 954 +#define ADF_6XXX_RL_MAX_TP_ASYM 173750UL +#define ADF_6XXX_RL_MAX_TP_SYM 95000UL +#define ADF_6XXX_RL_MAX_TP_DC 40000UL +#define ADF_6XXX_RL_MAX_TP_DECOMP 40000UL +#define ADF_6XXX_RL_SLICE_REF 1000UL + +/* Clock frequency */ +#define ADF_6XXX_AE_FREQ (1000 * HZ_PER_MHZ) + +enum icp_qat_gen6_slice_mask { + ICP_ACCEL_GEN6_MASK_UCS_SLICE = BIT(0), + ICP_ACCEL_GEN6_MASK_AUTH_SLICE = BIT(1), + ICP_ACCEL_GEN6_MASK_PKE_SLICE = BIT(2), + ICP_ACCEL_GEN6_MASK_CPR_SLICE = BIT(3), + ICP_ACCEL_GEN6_MASK_DCPRZ_SLICE = BIT(4), + ICP_ACCEL_GEN6_MASK_WCP_WAT_SLICE = BIT(6), +}; + +void adf_init_hw_data_6xxx(struct adf_hw_device_data *hw_data); +void adf_clean_hw_data_6xxx(struct adf_hw_device_data *hw_data); + +#endif /* ADF_6XXX_HW_DATA_H_ */ diff --git a/drivers/crypto/intel/qat/qat_6xxx/adf_drv.c b/drivers/crypto/intel/qat/qat_6xxx/adf_drv.c new file mode 100644 index 000000000000..c1dc9c56fdf5 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_6xxx/adf_drv.c @@ -0,0 +1,226 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2025 Intel Corporation */ +#include <linux/array_size.h> +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/errno.h> +#include <linux/list.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/types.h> + +#include <adf_accel_devices.h> +#include <adf_cfg.h> +#include <adf_common_drv.h> +#include <adf_dbgfs.h> + +#include "adf_gen6_shared.h" +#include "adf_6xxx_hw_data.h" + +static int bar_map[] = { + 0, /* SRAM */ + 2, /* PMISC */ + 4, /* ETR */ +}; + +static void adf_device_down(void *accel_dev) +{ + adf_dev_down(accel_dev); +} + +static void adf_dbgfs_cleanup(void *accel_dev) +{ + adf_dbgfs_exit(accel_dev); +} + +static void adf_cfg_device_remove(void *accel_dev) +{ + adf_cfg_dev_remove(accel_dev); +} + +static void adf_cleanup_hw_data(void *accel_dev) +{ + struct adf_accel_dev *accel_device = accel_dev; + + if (accel_device->hw_device) { + adf_clean_hw_data_6xxx(accel_device->hw_device); + accel_device->hw_device = NULL; + } +} + +static void adf_devmgr_remove(void *accel_dev) +{ + adf_devmgr_rm_dev(accel_dev, NULL); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + struct device *dev = &pdev->dev; + struct adf_accel_dev *accel_dev; + struct adf_bar *bar; + unsigned int i; + int ret; + + if (num_possible_nodes() > 1 && dev_to_node(dev) < 0) { + /* + * If the accelerator is connected to a node with no memory + * there is no point in using the accelerator since the remote + * memory transaction will be very slow. + */ + return dev_err_probe(dev, -EINVAL, "Invalid NUMA configuration.\n"); + } + + accel_dev = devm_kzalloc(dev, sizeof(*accel_dev), GFP_KERNEL); + if (!accel_dev) + return -ENOMEM; + + INIT_LIST_HEAD(&accel_dev->crypto_list); + INIT_LIST_HEAD(&accel_dev->list); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + accel_dev->owner = THIS_MODULE; + + hw_data = devm_kzalloc(dev, sizeof(*hw_data), GFP_KERNEL); + if (!hw_data) + return -ENOMEM; + + pci_read_config_byte(pdev, PCI_REVISION_ID, &accel_pci_dev->revid); + pci_read_config_dword(pdev, ADF_GEN6_FUSECTL4_OFFSET, &hw_data->fuses[ADF_FUSECTL4]); + pci_read_config_dword(pdev, ADF_GEN6_FUSECTL0_OFFSET, &hw_data->fuses[ADF_FUSECTL0]); + pci_read_config_dword(pdev, ADF_GEN6_FUSECTL1_OFFSET, &hw_data->fuses[ADF_FUSECTL1]); + + if (!(hw_data->fuses[ADF_FUSECTL1] & ICP_ACCEL_GEN6_MASK_WCP_WAT_SLICE)) + return dev_err_probe(dev, -EFAULT, "Wireless mode is not supported.\n"); + + /* Enable PCI device */ + ret = pcim_enable_device(pdev); + if (ret) + return dev_err_probe(dev, ret, "Cannot enable PCI device.\n"); + + ret = adf_devmgr_add_dev(accel_dev, NULL); + if (ret) + return dev_err_probe(dev, ret, "Failed to add new accelerator device.\n"); + + ret = devm_add_action_or_reset(dev, adf_devmgr_remove, accel_dev); + if (ret) + return ret; + + accel_dev->hw_device = hw_data; + adf_init_hw_data_6xxx(accel_dev->hw_device); + + ret = devm_add_action_or_reset(dev, adf_cleanup_hw_data, accel_dev); + if (ret) + return ret; + + /* Get Accelerators and Accelerator Engine masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + + /* If the device has no acceleration engines then ignore it */ + if (!hw_data->accel_mask || !hw_data->ae_mask || + (~hw_data->ae_mask & ADF_GEN6_ACCELERATORS_MASK)) { + ret = -EFAULT; + return dev_err_probe(dev, ret, "No acceleration units were found.\n"); + } + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + return ret; + + ret = devm_add_action_or_reset(dev, adf_cfg_device_remove, accel_dev); + if (ret) + return ret; + + /* Set DMA identifier */ + ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(64)); + if (ret) + return dev_err_probe(dev, ret, "No usable DMA configuration.\n"); + + ret = adf_gen6_cfg_dev_init(accel_dev); + if (ret) + return dev_err_probe(dev, ret, "Failed to initialize configuration.\n"); + + /* Get accelerator capability mask */ + hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev); + if (!hw_data->accel_capabilities_mask) { + ret = -EINVAL; + return dev_err_probe(dev, ret, "Failed to get capabilities mask.\n"); + } + + for (i = 0; i < ARRAY_SIZE(bar_map); i++) { + bar = &accel_pci_dev->pci_bars[i]; + + /* Map 64-bit PCIe BAR */ + bar->virt_addr = pcim_iomap_region(pdev, bar_map[i], pci_name(pdev)); + if (IS_ERR(bar->virt_addr)) { + ret = PTR_ERR(bar->virt_addr); + return dev_err_probe(dev, ret, "Failed to ioremap PCI region.\n"); + } + } + + pci_set_master(pdev); + + /* + * The PCI config space is saved at this point and will be restored + * after a Function Level Reset (FLR) as the FLR does not completely + * restore it. + */ + ret = pci_save_state(pdev); + if (ret) + return dev_err_probe(dev, ret, "Failed to save pci state.\n"); + + accel_dev->ras_errors.enabled = true; + + adf_dbgfs_init(accel_dev); + + ret = devm_add_action_or_reset(dev, adf_dbgfs_cleanup, accel_dev); + if (ret) + return ret; + + ret = adf_dev_up(accel_dev, true); + if (ret) + return ret; + + ret = devm_add_action_or_reset(dev, adf_device_down, accel_dev); + if (ret) + return ret; + + ret = adf_sysfs_init(accel_dev); + + return ret; +} + +static void adf_shutdown(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + adf_dev_down(accel_dev); +} + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_6XXX) }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_6XXX_DEVICE_NAME, + .probe = adf_probe, + .shutdown = adf_shutdown, + .sriov_configure = adf_sriov_configure, + .err_handler = &adf_err_handler, +}; +module_pci_driver(adf_driver); + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Intel"); +MODULE_FIRMWARE(ADF_6XXX_FW); +MODULE_FIRMWARE(ADF_6XXX_MMP); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology for GEN6 Devices"); +MODULE_SOFTDEP("pre: crypto-intel_qat"); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_c3xxx/Makefile b/drivers/crypto/intel/qat/qat_c3xxx/Makefile new file mode 100644 index 000000000000..43604c025f0c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxx/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_C3XXX) += qat_c3xxx.o +qat_c3xxx-y := adf_drv.o adf_c3xxx_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_c3xxx/adf_c3xxx_hw_data.c b/drivers/crypto/intel/qat/qat_c3xxx/adf_c3xxx_hw_data.c new file mode 100644 index 000000000000..07f2c42a68f5 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxx/adf_c3xxx_hw_data.c @@ -0,0 +1,167 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2021 Intel Corporation */ +#include <adf_accel_devices.h> +#include <adf_admin.h> +#include <adf_clock.h> +#include <adf_common_drv.h> +#include <adf_gen2_config.h> +#include <adf_gen2_hw_csr_data.h> +#include <adf_gen2_hw_data.h> +#include <adf_gen2_pfvf.h> +#include "adf_c3xxx_hw_data.h" +#include "adf_heartbeat.h" +#include "icp_qat_hw.h" + +/* Worker thread to service arbiter mappings */ +static const u32 thrd_to_arb_map[ADF_C3XXX_MAX_ACCELENGINES] = { + 0x12222AAA, 0x11222AAA, 0x12222AAA, + 0x11222AAA, 0x12222AAA, 0x11222AAA +}; + +static struct adf_hw_device_class c3xxx_class = { + .name = ADF_C3XXX_DEVICE_NAME, + .type = DEV_C3XXX, +}; + +static u32 get_accel_mask(struct adf_hw_device_data *self) +{ + u32 fuses = self->fuses[ADF_FUSECTL0]; + u32 straps = self->straps; + u32 accel; + + accel = ~(fuses | straps) >> ADF_C3XXX_ACCELERATORS_REG_OFFSET; + accel &= ADF_C3XXX_ACCELERATORS_MASK; + + return accel; +} + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + u32 fuses = self->fuses[ADF_FUSECTL0]; + u32 straps = self->straps; + unsigned long disabled; + u32 ae_disable; + int accel; + + /* If an accel is disabled, then disable the corresponding two AEs */ + disabled = ~get_accel_mask(self) & ADF_C3XXX_ACCELERATORS_MASK; + ae_disable = BIT(1) | BIT(0); + for_each_set_bit(accel, &disabled, ADF_C3XXX_MAX_ACCELERATORS) + straps |= ae_disable << (accel << 1); + + return ~(fuses | straps) & ADF_C3XXX_ACCELENGINES_MASK; +} + +static u32 get_ts_clock(struct adf_hw_device_data *self) +{ + /* + * Timestamp update interval is 16 AE clock ticks for c3xxx. + */ + return self->clock_frequency / 16; +} + +static int measure_clock(struct adf_accel_dev *accel_dev) +{ + u32 frequency; + int ret; + + ret = adf_dev_measure_clock(accel_dev, &frequency, ADF_C3XXX_MIN_AE_FREQ, + ADF_C3XXX_MAX_AE_FREQ); + if (ret) + return ret; + + accel_dev->hw_device->clock_frequency = frequency; + return 0; +} + +static u32 get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C3XXX_PMISC_BAR; +} + +static u32 get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C3XXX_ETR_BAR; +} + +static u32 get_sram_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C3XXX_SRAM_BAR; +} + +static enum dev_sku_info get_sku(struct adf_hw_device_data *self) +{ + int aes = self->get_num_aes(self); + + if (aes == 6) + return DEV_SKU_4; + + return DEV_SKU_UNKNOWN; +} + +static const u32 *adf_get_arbiter_mapping(struct adf_accel_dev *accel_dev) +{ + return thrd_to_arb_map; +} + +static void configure_iov_threads(struct adf_accel_dev *accel_dev, bool enable) +{ + adf_gen2_cfg_iov_thds(accel_dev, enable, + ADF_C3XXX_AE2FUNC_MAP_GRP_A_NUM_REGS, + ADF_C3XXX_AE2FUNC_MAP_GRP_B_NUM_REGS); +} + +void adf_init_hw_data_c3xxx(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class = &c3xxx_class; + hw_data->instance_id = c3xxx_class.instances++; + hw_data->num_banks = ADF_C3XXX_ETR_MAX_BANKS; + hw_data->num_rings_per_bank = ADF_ETR_MAX_RINGS_PER_BANK; + hw_data->num_accel = ADF_C3XXX_MAX_ACCELERATORS; + hw_data->num_logical_accel = 1; + hw_data->num_engines = ADF_C3XXX_MAX_ACCELENGINES; + hw_data->tx_rx_gap = ADF_GEN2_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_GEN2_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN2_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_isr_resource_alloc; + hw_data->free_irq = adf_isr_resource_free; + hw_data->enable_error_correction = adf_gen2_enable_error_correction; + hw_data->get_accel_mask = get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_accel_cap = adf_gen2_get_accel_cap; + hw_data->get_num_accels = adf_gen2_get_num_accels; + hw_data->get_num_aes = adf_gen2_get_num_aes; + hw_data->get_sram_bar_id = get_sram_bar_id; + hw_data->get_etr_bar_id = get_etr_bar_id; + hw_data->get_misc_bar_id = get_misc_bar_id; + hw_data->get_admin_info = adf_gen2_get_admin_info; + hw_data->get_arb_info = adf_gen2_get_arb_info; + hw_data->get_sku = get_sku; + hw_data->fw_name = ADF_C3XXX_FW; + hw_data->fw_mmp_name = ADF_C3XXX_MMP; + hw_data->init_admin_comms = adf_init_admin_comms; + hw_data->exit_admin_comms = adf_exit_admin_comms; + hw_data->configure_iov_threads = configure_iov_threads; + hw_data->send_admin_init = adf_send_admin_init; + hw_data->init_arb = adf_init_arb; + hw_data->exit_arb = adf_exit_arb; + hw_data->get_arb_mapping = adf_get_arbiter_mapping; + hw_data->enable_ints = adf_gen2_enable_ints; + hw_data->reset_device = adf_reset_flr; + hw_data->set_ssm_wdtimer = adf_gen2_set_ssm_wdtimer; + hw_data->disable_iov = adf_disable_sriov; + hw_data->dev_config = adf_gen2_dev_config; + hw_data->measure_clock = measure_clock; + hw_data->get_hb_clock = get_ts_clock; + hw_data->num_hb_ctrs = ADF_NUM_HB_CNT_PER_AE; + hw_data->check_hb_ctrs = adf_heartbeat_check_ctrs; + + adf_gen2_init_pf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen2_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen2_init_dc_ops(&hw_data->dc_ops); +} + +void adf_clean_hw_data_c3xxx(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; +} diff --git a/drivers/crypto/intel/qat/qat_c3xxx/adf_c3xxx_hw_data.h b/drivers/crypto/intel/qat/qat_c3xxx/adf_c3xxx_hw_data.h new file mode 100644 index 000000000000..690c6a1aa172 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxx/adf_c3xxx_hw_data.h @@ -0,0 +1,35 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_C3XXX_HW_DATA_H_ +#define ADF_C3XXX_HW_DATA_H_ + +#include <linux/units.h> + +/* PCIe configuration space */ +#define ADF_C3XXX_PMISC_BAR 0 +#define ADF_C3XXX_ETR_BAR 1 +#define ADF_C3XXX_SRAM_BAR 0 +#define ADF_C3XXX_MAX_ACCELERATORS 3 +#define ADF_C3XXX_MAX_ACCELENGINES 6 +#define ADF_C3XXX_ACCELERATORS_REG_OFFSET 16 +#define ADF_C3XXX_ACCELERATORS_MASK 0x7 +#define ADF_C3XXX_ACCELENGINES_MASK 0x3F +#define ADF_C3XXX_ETR_MAX_BANKS 16 +#define ADF_C3XXX_SOFTSTRAP_CSR_OFFSET 0x2EC + +/* AE to function mapping */ +#define ADF_C3XXX_AE2FUNC_MAP_GRP_A_NUM_REGS 48 +#define ADF_C3XXX_AE2FUNC_MAP_GRP_B_NUM_REGS 6 + +/* Clocks frequency */ +#define ADF_C3XXX_AE_FREQ (685 * HZ_PER_MHZ) +#define ADF_C3XXX_MIN_AE_FREQ (533 * HZ_PER_MHZ) +#define ADF_C3XXX_MAX_AE_FREQ (685 * HZ_PER_MHZ) + +/* Firmware Binary */ +#define ADF_C3XXX_FW "qat_c3xxx.bin" +#define ADF_C3XXX_MMP "qat_c3xxx_mmp.bin" + +void adf_init_hw_data_c3xxx(struct adf_hw_device_data *hw_data); +void adf_clean_hw_data_c3xxx(struct adf_hw_device_data *hw_data); +#endif diff --git a/drivers/crypto/intel/qat/qat_c3xxx/adf_drv.c b/drivers/crypto/intel/qat/qat_c3xxx/adf_drv.c new file mode 100644 index 000000000000..bceb5dd8b148 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxx/adf_drv.c @@ -0,0 +1,260 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/fs.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/platform_device.h> +#include <linux/workqueue.h> +#include <linux/io.h> +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_cfg.h> +#include <adf_dbgfs.h> +#include "adf_c3xxx_hw_data.h" + +static void adf_cleanup_pci_dev(struct adf_accel_dev *accel_dev) +{ + pci_release_regions(accel_dev->accel_pci_dev.pci_dev); + pci_disable_device(accel_dev->accel_pci_dev.pci_dev); +} + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *accel_pci_dev = &accel_dev->accel_pci_dev; + int i; + + for (i = 0; i < ADF_PCI_MAX_BARS; i++) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i]; + + if (bar->virt_addr) + pci_iounmap(accel_pci_dev->pci_dev, bar->virt_addr); + } + + if (accel_dev->hw_device) { + switch (accel_pci_dev->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_C3XXX: + adf_clean_hw_data_c3xxx(accel_dev->hw_device); + break; + default: + break; + } + kfree(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + adf_devmgr_rm_dev(accel_dev, NULL); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + int ret; + + switch (ent->device) { + case PCI_DEVICE_ID_INTEL_QAT_C3XXX: + break; + default: + dev_err(&pdev->dev, "Invalid device 0x%x.\n", ent->device); + return -ENODEV; + } + + if (num_possible_nodes() > 1 && dev_to_node(&pdev->dev) < 0) { + /* If the accelerator is connected to a node with no memory + * there is no point in using the accelerator since the remote + * memory transaction will be very slow. */ + dev_err(&pdev->dev, "Invalid NUMA configuration.\n"); + return -EINVAL; + } + + accel_dev = kzalloc_node(sizeof(*accel_dev), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!accel_dev) + return -ENOMEM; + + INIT_LIST_HEAD(&accel_dev->crypto_list); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* Add accel device to accel table. + * This should be called before adf_cleanup_accel is called */ + if (adf_devmgr_add_dev(accel_dev, NULL)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + kfree(accel_dev); + return -EFAULT; + } + + accel_dev->owner = THIS_MODULE; + /* Allocate and configure device configuration structure */ + hw_data = kzalloc_node(sizeof(*hw_data), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + + accel_dev->hw_device = hw_data; + adf_init_hw_data_c3xxx(accel_dev->hw_device); + pci_read_config_byte(pdev, PCI_REVISION_ID, &accel_pci_dev->revid); + pci_read_config_dword(pdev, ADF_DEVICE_FUSECTL_OFFSET, + &hw_data->fuses[ADF_FUSECTL0]); + pci_read_config_dword(pdev, ADF_C3XXX_SOFTSTRAP_CSR_OFFSET, + &hw_data->straps); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + /* If the device has no acceleration engines then ignore it. */ + if (!hw_data->accel_mask || !hw_data->ae_mask || + ((~hw_data->ae_mask) & 0x01)) { + dev_err(&pdev->dev, "No acceleration units found"); + ret = -EFAULT; + goto out_err; + } + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + /* enable PCI device */ + if (pci_enable_device(pdev)) { + ret = -EFAULT; + goto out_err; + } + + /* set dma identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(48)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration\n"); + goto out_err_disable; + } + + if (pci_request_regions(pdev, ADF_C3XXX_DEVICE_NAME)) { + ret = -EFAULT; + goto out_err_disable; + } + + /* Get accelerator capabilities mask */ + hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev); + + /* Find and map all the device's BARS */ + i = 0; + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM); + for_each_set_bit(bar_nr, &bar_mask, ADF_PCI_MAX_BARS * 2) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i++]; + + bar->base_addr = pci_resource_start(pdev, bar_nr); + if (!bar->base_addr) + break; + bar->size = pci_resource_len(pdev, bar_nr); + bar->virt_addr = pci_iomap(accel_pci_dev->pci_dev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to map BAR %d\n", bar_nr); + ret = -EFAULT; + goto out_err_free_reg; + } + } + pci_set_master(pdev); + + if (pci_save_state(pdev)) { + dev_err(&pdev->dev, "Failed to save pci state\n"); + ret = -ENOMEM; + goto out_err_free_reg; + } + + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, true); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err_free_reg: + pci_release_regions(accel_pci_dev->pci_dev); +out_err_disable: + pci_disable_device(accel_pci_dev->pci_dev); +out_err: + adf_cleanup_accel(accel_dev); + kfree(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); + adf_cleanup_pci_dev(accel_dev); + kfree(accel_dev); +} + +static void adf_shutdown(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + adf_dev_down(accel_dev); +} + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_C3XXX) }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_C3XXX_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, + .shutdown = adf_shutdown, + .sriov_configure = adf_sriov_configure, + .err_handler = &adf_err_handler, +}; + +static int __init adfdrv_init(void) +{ + request_module("intel_qat"); + + if (pci_register_driver(&adf_driver)) { + pr_err("QAT: Driver initialization failed\n"); + return -EFAULT; + } + return 0; +} + +static void __exit adfdrv_release(void) +{ + pci_unregister_driver(&adf_driver); +} + +module_init(adfdrv_init); +module_exit(adfdrv_release); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_FIRMWARE(ADF_C3XXX_FW); +MODULE_FIRMWARE(ADF_C3XXX_MMP); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_c3xxxvf/Makefile b/drivers/crypto/intel/qat/qat_c3xxxvf/Makefile new file mode 100644 index 000000000000..03f6745b4aa2 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxxvf/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_C3XXXVF) += qat_c3xxxvf.o +qat_c3xxxvf-y := adf_drv.o adf_c3xxxvf_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_c3xxxvf/adf_c3xxxvf_hw_data.c b/drivers/crypto/intel/qat/qat_c3xxxvf/adf_c3xxxvf_hw_data.c new file mode 100644 index 000000000000..db3c33fa1881 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxxvf/adf_c3xxxvf_hw_data.c @@ -0,0 +1,101 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_gen2_config.h> +#include <adf_gen2_hw_csr_data.h> +#include <adf_gen2_hw_data.h> +#include <adf_gen2_pfvf.h> +#include <adf_pfvf_vf_msg.h> +#include "adf_c3xxxvf_hw_data.h" + +static struct adf_hw_device_class c3xxxiov_class = { + .name = ADF_C3XXXVF_DEVICE_NAME, + .type = DEV_C3XXXVF, +}; + +static u32 get_accel_mask(struct adf_hw_device_data *self) +{ + return ADF_C3XXXIOV_ACCELERATORS_MASK; +} + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + return ADF_C3XXXIOV_ACCELENGINES_MASK; +} + +static u32 get_num_accels(struct adf_hw_device_data *self) +{ + return ADF_C3XXXIOV_MAX_ACCELERATORS; +} + +static u32 get_num_aes(struct adf_hw_device_data *self) +{ + return ADF_C3XXXIOV_MAX_ACCELENGINES; +} + +static u32 get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C3XXXIOV_PMISC_BAR; +} + +static u32 get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C3XXXIOV_ETR_BAR; +} + +static enum dev_sku_info get_sku(struct adf_hw_device_data *self) +{ + return DEV_SKU_VF; +} + +static int adf_vf_int_noop(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static void adf_vf_void_noop(struct adf_accel_dev *accel_dev) +{ +} + +void adf_init_hw_data_c3xxxiov(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class = &c3xxxiov_class; + hw_data->num_banks = ADF_C3XXXIOV_ETR_MAX_BANKS; + hw_data->num_rings_per_bank = ADF_ETR_MAX_RINGS_PER_BANK; + hw_data->num_accel = ADF_C3XXXIOV_MAX_ACCELERATORS; + hw_data->num_logical_accel = 1; + hw_data->num_engines = ADF_C3XXXIOV_MAX_ACCELENGINES; + hw_data->tx_rx_gap = ADF_C3XXXIOV_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_C3XXXIOV_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN2_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_vf_isr_resource_alloc; + hw_data->free_irq = adf_vf_isr_resource_free; + hw_data->enable_error_correction = adf_vf_void_noop; + hw_data->init_admin_comms = adf_vf_int_noop; + hw_data->exit_admin_comms = adf_vf_void_noop; + hw_data->send_admin_init = adf_vf2pf_notify_init; + hw_data->init_arb = adf_vf_int_noop; + hw_data->exit_arb = adf_vf_void_noop; + hw_data->disable_iov = adf_vf2pf_notify_shutdown; + hw_data->get_accel_mask = get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_num_accels = get_num_accels; + hw_data->get_num_aes = get_num_aes; + hw_data->get_etr_bar_id = get_etr_bar_id; + hw_data->get_misc_bar_id = get_misc_bar_id; + hw_data->get_sku = get_sku; + hw_data->enable_ints = adf_vf_void_noop; + hw_data->dev_class->instances++; + hw_data->dev_config = adf_gen2_dev_config; + adf_devmgr_update_class_index(hw_data); + adf_gen2_init_vf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen2_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen2_init_dc_ops(&hw_data->dc_ops); +} + +void adf_clean_hw_data_c3xxxiov(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; + adf_devmgr_update_class_index(hw_data); +} diff --git a/drivers/crypto/intel/qat/qat_c3xxxvf/adf_c3xxxvf_hw_data.h b/drivers/crypto/intel/qat/qat_c3xxxvf/adf_c3xxxvf_hw_data.h new file mode 100644 index 000000000000..6b4bf181d15b --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxxvf/adf_c3xxxvf_hw_data.h @@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2015 - 2020 Intel Corporation */ +#ifndef ADF_C3XXXVF_HW_DATA_H_ +#define ADF_C3XXXVF_HW_DATA_H_ + +#define ADF_C3XXXIOV_PMISC_BAR 1 +#define ADF_C3XXXIOV_ACCELERATORS_MASK 0x1 +#define ADF_C3XXXIOV_ACCELENGINES_MASK 0x1 +#define ADF_C3XXXIOV_MAX_ACCELERATORS 1 +#define ADF_C3XXXIOV_MAX_ACCELENGINES 1 +#define ADF_C3XXXIOV_RX_RINGS_OFFSET 8 +#define ADF_C3XXXIOV_TX_RINGS_MASK 0xFF +#define ADF_C3XXXIOV_ETR_BAR 0 +#define ADF_C3XXXIOV_ETR_MAX_BANKS 1 + +void adf_init_hw_data_c3xxxiov(struct adf_hw_device_data *hw_data); +void adf_clean_hw_data_c3xxxiov(struct adf_hw_device_data *hw_data); +#endif diff --git a/drivers/crypto/intel/qat/qat_c3xxxvf/adf_drv.c b/drivers/crypto/intel/qat/qat_c3xxxvf/adf_drv.c new file mode 100644 index 000000000000..c622793e94a8 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c3xxxvf/adf_drv.c @@ -0,0 +1,229 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/fs.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/platform_device.h> +#include <linux/workqueue.h> +#include <linux/io.h> +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_cfg.h> +#include <adf_dbgfs.h> +#include "adf_c3xxxvf_hw_data.h" + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_C3XXX_VF), }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static int adf_probe(struct pci_dev *dev, const struct pci_device_id *ent); +static void adf_remove(struct pci_dev *dev); + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_C3XXXVF_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, +}; + +static void adf_cleanup_pci_dev(struct adf_accel_dev *accel_dev) +{ + pci_release_regions(accel_dev->accel_pci_dev.pci_dev); + pci_disable_device(accel_dev->accel_pci_dev.pci_dev); +} + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *accel_pci_dev = &accel_dev->accel_pci_dev; + struct adf_accel_dev *pf; + int i; + + for (i = 0; i < ADF_PCI_MAX_BARS; i++) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i]; + + if (bar->virt_addr) + pci_iounmap(accel_pci_dev->pci_dev, bar->virt_addr); + } + + if (accel_dev->hw_device) { + switch (accel_pci_dev->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_C3XXX_VF: + adf_clean_hw_data_c3xxxiov(accel_dev->hw_device); + break; + default: + break; + } + kfree(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + pf = adf_devmgr_pci_to_accel_dev(accel_pci_dev->pci_dev->physfn); + adf_devmgr_rm_dev(accel_dev, pf); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_dev *pf; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + int ret; + + switch (ent->device) { + case PCI_DEVICE_ID_INTEL_QAT_C3XXX_VF: + break; + default: + dev_err(&pdev->dev, "Invalid device 0x%x.\n", ent->device); + return -ENODEV; + } + + accel_dev = kzalloc_node(sizeof(*accel_dev), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!accel_dev) + return -ENOMEM; + + accel_dev->is_vf = true; + pf = adf_devmgr_pci_to_accel_dev(pdev->physfn); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* Add accel device to accel table */ + if (adf_devmgr_add_dev(accel_dev, pf)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + kfree(accel_dev); + return -EFAULT; + } + INIT_LIST_HEAD(&accel_dev->crypto_list); + + accel_dev->owner = THIS_MODULE; + /* Allocate and configure device configuration structure */ + hw_data = kzalloc_node(sizeof(*hw_data), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + accel_dev->hw_device = hw_data; + adf_init_hw_data_c3xxxiov(accel_dev->hw_device); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + /* enable PCI device */ + if (pci_enable_device(pdev)) { + ret = -EFAULT; + goto out_err; + } + + /* set dma identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(48)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration\n"); + goto out_err_disable; + } + + if (pci_request_regions(pdev, ADF_C3XXXVF_DEVICE_NAME)) { + ret = -EFAULT; + goto out_err_disable; + } + + /* Find and map all the device's BARS */ + i = 0; + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM); + for_each_set_bit(bar_nr, &bar_mask, ADF_PCI_MAX_BARS * 2) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i++]; + + bar->base_addr = pci_resource_start(pdev, bar_nr); + if (!bar->base_addr) + break; + bar->size = pci_resource_len(pdev, bar_nr); + bar->virt_addr = pci_iomap(accel_pci_dev->pci_dev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to map BAR %d\n", bar_nr); + ret = -EFAULT; + goto out_err_free_reg; + } + } + pci_set_master(pdev); + /* Completion for VF2PF request/response message exchange */ + init_completion(&accel_dev->vf.msg_received); + + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, false); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err_free_reg: + pci_release_regions(accel_pci_dev->pci_dev); +out_err_disable: + pci_disable_device(accel_pci_dev->pci_dev); +out_err: + adf_cleanup_accel(accel_dev); + kfree(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_flush_vf_wq(accel_dev); + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); + adf_cleanup_pci_dev(accel_dev); + kfree(accel_dev); +} + +static int __init adfdrv_init(void) +{ + request_module("intel_qat"); + + if (pci_register_driver(&adf_driver)) { + pr_err("QAT: Driver initialization failed\n"); + return -EFAULT; + } + return 0; +} + +static void __exit adfdrv_release(void) +{ + pci_unregister_driver(&adf_driver); + adf_clean_vf_map(true); +} + +module_init(adfdrv_init); +module_exit(adfdrv_release); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_c62x/Makefile b/drivers/crypto/intel/qat/qat_c62x/Makefile new file mode 100644 index 000000000000..f3d722bef088 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62x/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_C62X) += qat_c62x.o +qat_c62x-y := adf_drv.o adf_c62x_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_c62x/adf_c62x_hw_data.c b/drivers/crypto/intel/qat/qat_c62x/adf_c62x_hw_data.c new file mode 100644 index 000000000000..0b410b41474d --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62x/adf_c62x_hw_data.c @@ -0,0 +1,169 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2021 Intel Corporation */ +#include <adf_accel_devices.h> +#include <adf_admin.h> +#include <adf_clock.h> +#include <adf_common_drv.h> +#include <adf_gen2_config.h> +#include <adf_gen2_hw_csr_data.h> +#include <adf_gen2_hw_data.h> +#include <adf_gen2_pfvf.h> +#include "adf_c62x_hw_data.h" +#include "adf_heartbeat.h" +#include "icp_qat_hw.h" + +/* Worker thread to service arbiter mappings */ +static const u32 thrd_to_arb_map[ADF_C62X_MAX_ACCELENGINES] = { + 0x12222AAA, 0x11222AAA, 0x12222AAA, 0x11222AAA, 0x12222AAA, + 0x11222AAA, 0x12222AAA, 0x11222AAA, 0x12222AAA, 0x11222AAA +}; + +static struct adf_hw_device_class c62x_class = { + .name = ADF_C62X_DEVICE_NAME, + .type = DEV_C62X, +}; + +static u32 get_accel_mask(struct adf_hw_device_data *self) +{ + u32 fuses = self->fuses[ADF_FUSECTL0]; + u32 straps = self->straps; + u32 accel; + + accel = ~(fuses | straps) >> ADF_C62X_ACCELERATORS_REG_OFFSET; + accel &= ADF_C62X_ACCELERATORS_MASK; + + return accel; +} + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + u32 fuses = self->fuses[ADF_FUSECTL0]; + u32 straps = self->straps; + unsigned long disabled; + u32 ae_disable; + int accel; + + /* If an accel is disabled, then disable the corresponding two AEs */ + disabled = ~get_accel_mask(self) & ADF_C62X_ACCELERATORS_MASK; + ae_disable = BIT(1) | BIT(0); + for_each_set_bit(accel, &disabled, ADF_C62X_MAX_ACCELERATORS) + straps |= ae_disable << (accel << 1); + + return ~(fuses | straps) & ADF_C62X_ACCELENGINES_MASK; +} + +static u32 get_ts_clock(struct adf_hw_device_data *self) +{ + /* + * Timestamp update interval is 16 AE clock ticks for c62x. + */ + return self->clock_frequency / 16; +} + +static int measure_clock(struct adf_accel_dev *accel_dev) +{ + u32 frequency; + int ret; + + ret = adf_dev_measure_clock(accel_dev, &frequency, ADF_C62X_MIN_AE_FREQ, + ADF_C62X_MAX_AE_FREQ); + if (ret) + return ret; + + accel_dev->hw_device->clock_frequency = frequency; + return 0; +} + +static u32 get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C62X_PMISC_BAR; +} + +static u32 get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C62X_ETR_BAR; +} + +static u32 get_sram_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C62X_SRAM_BAR; +} + +static enum dev_sku_info get_sku(struct adf_hw_device_data *self) +{ + int aes = self->get_num_aes(self); + + if (aes == 8) + return DEV_SKU_2; + else if (aes == 10) + return DEV_SKU_4; + + return DEV_SKU_UNKNOWN; +} + +static const u32 *adf_get_arbiter_mapping(struct adf_accel_dev *accel_dev) +{ + return thrd_to_arb_map; +} + +static void configure_iov_threads(struct adf_accel_dev *accel_dev, bool enable) +{ + adf_gen2_cfg_iov_thds(accel_dev, enable, + ADF_C62X_AE2FUNC_MAP_GRP_A_NUM_REGS, + ADF_C62X_AE2FUNC_MAP_GRP_B_NUM_REGS); +} + +void adf_init_hw_data_c62x(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class = &c62x_class; + hw_data->instance_id = c62x_class.instances++; + hw_data->num_banks = ADF_C62X_ETR_MAX_BANKS; + hw_data->num_rings_per_bank = ADF_ETR_MAX_RINGS_PER_BANK; + hw_data->num_accel = ADF_C62X_MAX_ACCELERATORS; + hw_data->num_logical_accel = 1; + hw_data->num_engines = ADF_C62X_MAX_ACCELENGINES; + hw_data->tx_rx_gap = ADF_GEN2_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_GEN2_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN2_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_isr_resource_alloc; + hw_data->free_irq = adf_isr_resource_free; + hw_data->enable_error_correction = adf_gen2_enable_error_correction; + hw_data->get_accel_mask = get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_accel_cap = adf_gen2_get_accel_cap; + hw_data->get_num_accels = adf_gen2_get_num_accels; + hw_data->get_num_aes = adf_gen2_get_num_aes; + hw_data->get_sram_bar_id = get_sram_bar_id; + hw_data->get_etr_bar_id = get_etr_bar_id; + hw_data->get_misc_bar_id = get_misc_bar_id; + hw_data->get_admin_info = adf_gen2_get_admin_info; + hw_data->get_arb_info = adf_gen2_get_arb_info; + hw_data->get_sku = get_sku; + hw_data->fw_name = ADF_C62X_FW; + hw_data->fw_mmp_name = ADF_C62X_MMP; + hw_data->init_admin_comms = adf_init_admin_comms; + hw_data->exit_admin_comms = adf_exit_admin_comms; + hw_data->configure_iov_threads = configure_iov_threads; + hw_data->send_admin_init = adf_send_admin_init; + hw_data->init_arb = adf_init_arb; + hw_data->exit_arb = adf_exit_arb; + hw_data->get_arb_mapping = adf_get_arbiter_mapping; + hw_data->enable_ints = adf_gen2_enable_ints; + hw_data->reset_device = adf_reset_flr; + hw_data->set_ssm_wdtimer = adf_gen2_set_ssm_wdtimer; + hw_data->disable_iov = adf_disable_sriov; + hw_data->dev_config = adf_gen2_dev_config; + hw_data->measure_clock = measure_clock; + hw_data->get_hb_clock = get_ts_clock; + hw_data->num_hb_ctrs = ADF_NUM_HB_CNT_PER_AE; + hw_data->check_hb_ctrs = adf_heartbeat_check_ctrs; + + adf_gen2_init_pf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen2_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen2_init_dc_ops(&hw_data->dc_ops); +} + +void adf_clean_hw_data_c62x(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; +} diff --git a/drivers/crypto/intel/qat/qat_c62x/adf_c62x_hw_data.h b/drivers/crypto/intel/qat/qat_c62x/adf_c62x_hw_data.h new file mode 100644 index 000000000000..13e6ebf6fd91 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62x/adf_c62x_hw_data.h @@ -0,0 +1,35 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_C62X_HW_DATA_H_ +#define ADF_C62X_HW_DATA_H_ + +#include <linux/units.h> + +/* PCIe configuration space */ +#define ADF_C62X_SRAM_BAR 0 +#define ADF_C62X_PMISC_BAR 1 +#define ADF_C62X_ETR_BAR 2 +#define ADF_C62X_MAX_ACCELERATORS 5 +#define ADF_C62X_MAX_ACCELENGINES 10 +#define ADF_C62X_ACCELERATORS_REG_OFFSET 16 +#define ADF_C62X_ACCELERATORS_MASK 0x1F +#define ADF_C62X_ACCELENGINES_MASK 0x3FF +#define ADF_C62X_ETR_MAX_BANKS 16 +#define ADF_C62X_SOFTSTRAP_CSR_OFFSET 0x2EC + +/* AE to function mapping */ +#define ADF_C62X_AE2FUNC_MAP_GRP_A_NUM_REGS 80 +#define ADF_C62X_AE2FUNC_MAP_GRP_B_NUM_REGS 10 + +/* Clocks frequency */ +#define ADF_C62X_AE_FREQ (685 * HZ_PER_MHZ) +#define ADF_C62X_MIN_AE_FREQ (533 * HZ_PER_MHZ) +#define ADF_C62X_MAX_AE_FREQ (800 * HZ_PER_MHZ) + +/* Firmware Binary */ +#define ADF_C62X_FW "qat_c62x.bin" +#define ADF_C62X_MMP "qat_c62x_mmp.bin" + +void adf_init_hw_data_c62x(struct adf_hw_device_data *hw_data); +void adf_clean_hw_data_c62x(struct adf_hw_device_data *hw_data); +#endif diff --git a/drivers/crypto/intel/qat/qat_c62x/adf_drv.c b/drivers/crypto/intel/qat/qat_c62x/adf_drv.c new file mode 100644 index 000000000000..23ccb72b6ea2 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62x/adf_drv.c @@ -0,0 +1,260 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/fs.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/platform_device.h> +#include <linux/workqueue.h> +#include <linux/io.h> +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_cfg.h> +#include <adf_dbgfs.h> +#include "adf_c62x_hw_data.h" + +static void adf_cleanup_pci_dev(struct adf_accel_dev *accel_dev) +{ + pci_release_regions(accel_dev->accel_pci_dev.pci_dev); + pci_disable_device(accel_dev->accel_pci_dev.pci_dev); +} + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *accel_pci_dev = &accel_dev->accel_pci_dev; + int i; + + for (i = 0; i < ADF_PCI_MAX_BARS; i++) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i]; + + if (bar->virt_addr) + pci_iounmap(accel_pci_dev->pci_dev, bar->virt_addr); + } + + if (accel_dev->hw_device) { + switch (accel_pci_dev->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_C62X: + adf_clean_hw_data_c62x(accel_dev->hw_device); + break; + default: + break; + } + kfree(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + adf_devmgr_rm_dev(accel_dev, NULL); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + int ret; + + switch (ent->device) { + case PCI_DEVICE_ID_INTEL_QAT_C62X: + break; + default: + dev_err(&pdev->dev, "Invalid device 0x%x.\n", ent->device); + return -ENODEV; + } + + if (num_possible_nodes() > 1 && dev_to_node(&pdev->dev) < 0) { + /* If the accelerator is connected to a node with no memory + * there is no point in using the accelerator since the remote + * memory transaction will be very slow. */ + dev_err(&pdev->dev, "Invalid NUMA configuration.\n"); + return -EINVAL; + } + + accel_dev = kzalloc_node(sizeof(*accel_dev), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!accel_dev) + return -ENOMEM; + + INIT_LIST_HEAD(&accel_dev->crypto_list); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* Add accel device to accel table. + * This should be called before adf_cleanup_accel is called */ + if (adf_devmgr_add_dev(accel_dev, NULL)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + kfree(accel_dev); + return -EFAULT; + } + + accel_dev->owner = THIS_MODULE; + /* Allocate and configure device configuration structure */ + hw_data = kzalloc_node(sizeof(*hw_data), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + + accel_dev->hw_device = hw_data; + adf_init_hw_data_c62x(accel_dev->hw_device); + pci_read_config_byte(pdev, PCI_REVISION_ID, &accel_pci_dev->revid); + pci_read_config_dword(pdev, ADF_DEVICE_FUSECTL_OFFSET, + &hw_data->fuses[ADF_FUSECTL0]); + pci_read_config_dword(pdev, ADF_C62X_SOFTSTRAP_CSR_OFFSET, + &hw_data->straps); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + /* If the device has no acceleration engines then ignore it. */ + if (!hw_data->accel_mask || !hw_data->ae_mask || + ((~hw_data->ae_mask) & 0x01)) { + dev_err(&pdev->dev, "No acceleration units found"); + ret = -EFAULT; + goto out_err; + } + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + /* enable PCI device */ + if (pci_enable_device(pdev)) { + ret = -EFAULT; + goto out_err; + } + + /* set dma identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(48)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration\n"); + goto out_err_disable; + } + + if (pci_request_regions(pdev, ADF_C62X_DEVICE_NAME)) { + ret = -EFAULT; + goto out_err_disable; + } + + /* Get accelerator capabilities mask */ + hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev); + + /* Find and map all the device's BARS */ + i = (hw_data->fuses[ADF_FUSECTL0] & ADF_DEVICE_FUSECTL_MASK) ? 1 : 0; + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM); + for_each_set_bit(bar_nr, &bar_mask, ADF_PCI_MAX_BARS * 2) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i++]; + + bar->base_addr = pci_resource_start(pdev, bar_nr); + if (!bar->base_addr) + break; + bar->size = pci_resource_len(pdev, bar_nr); + bar->virt_addr = pci_iomap(accel_pci_dev->pci_dev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to map BAR %d\n", bar_nr); + ret = -EFAULT; + goto out_err_free_reg; + } + } + pci_set_master(pdev); + + if (pci_save_state(pdev)) { + dev_err(&pdev->dev, "Failed to save pci state\n"); + ret = -ENOMEM; + goto out_err_free_reg; + } + + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, true); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err_free_reg: + pci_release_regions(accel_pci_dev->pci_dev); +out_err_disable: + pci_disable_device(accel_pci_dev->pci_dev); +out_err: + adf_cleanup_accel(accel_dev); + kfree(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); + adf_cleanup_pci_dev(accel_dev); + kfree(accel_dev); +} + +static void adf_shutdown(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + adf_dev_down(accel_dev); +} + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_C62X) }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_C62X_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, + .shutdown = adf_shutdown, + .sriov_configure = adf_sriov_configure, + .err_handler = &adf_err_handler, +}; + +static int __init adfdrv_init(void) +{ + request_module("intel_qat"); + + if (pci_register_driver(&adf_driver)) { + pr_err("QAT: Driver initialization failed\n"); + return -EFAULT; + } + return 0; +} + +static void __exit adfdrv_release(void) +{ + pci_unregister_driver(&adf_driver); +} + +module_init(adfdrv_init); +module_exit(adfdrv_release); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_FIRMWARE(ADF_C62X_FW); +MODULE_FIRMWARE(ADF_C62X_MMP); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_c62xvf/Makefile b/drivers/crypto/intel/qat/qat_c62xvf/Makefile new file mode 100644 index 000000000000..ed7f3f722d99 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62xvf/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_C62XVF) += qat_c62xvf.o +qat_c62xvf-y := adf_drv.o adf_c62xvf_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_c62xvf/adf_c62xvf_hw_data.c b/drivers/crypto/intel/qat/qat_c62xvf/adf_c62xvf_hw_data.c new file mode 100644 index 000000000000..7f00035d3661 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62xvf/adf_c62xvf_hw_data.c @@ -0,0 +1,101 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_gen2_config.h> +#include <adf_gen2_hw_csr_data.h> +#include <adf_gen2_hw_data.h> +#include <adf_gen2_pfvf.h> +#include <adf_pfvf_vf_msg.h> +#include "adf_c62xvf_hw_data.h" + +static struct adf_hw_device_class c62xiov_class = { + .name = ADF_C62XVF_DEVICE_NAME, + .type = DEV_C62XVF, +}; + +static u32 get_accel_mask(struct adf_hw_device_data *self) +{ + return ADF_C62XIOV_ACCELERATORS_MASK; +} + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + return ADF_C62XIOV_ACCELENGINES_MASK; +} + +static u32 get_num_accels(struct adf_hw_device_data *self) +{ + return ADF_C62XIOV_MAX_ACCELERATORS; +} + +static u32 get_num_aes(struct adf_hw_device_data *self) +{ + return ADF_C62XIOV_MAX_ACCELENGINES; +} + +static u32 get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C62XIOV_PMISC_BAR; +} + +static u32 get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_C62XIOV_ETR_BAR; +} + +static enum dev_sku_info get_sku(struct adf_hw_device_data *self) +{ + return DEV_SKU_VF; +} + +static int adf_vf_int_noop(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static void adf_vf_void_noop(struct adf_accel_dev *accel_dev) +{ +} + +void adf_init_hw_data_c62xiov(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class = &c62xiov_class; + hw_data->num_banks = ADF_C62XIOV_ETR_MAX_BANKS; + hw_data->num_rings_per_bank = ADF_ETR_MAX_RINGS_PER_BANK; + hw_data->num_accel = ADF_C62XIOV_MAX_ACCELERATORS; + hw_data->num_logical_accel = 1; + hw_data->num_engines = ADF_C62XIOV_MAX_ACCELENGINES; + hw_data->tx_rx_gap = ADF_C62XIOV_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_C62XIOV_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN2_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_vf_isr_resource_alloc; + hw_data->free_irq = adf_vf_isr_resource_free; + hw_data->enable_error_correction = adf_vf_void_noop; + hw_data->init_admin_comms = adf_vf_int_noop; + hw_data->exit_admin_comms = adf_vf_void_noop; + hw_data->send_admin_init = adf_vf2pf_notify_init; + hw_data->init_arb = adf_vf_int_noop; + hw_data->exit_arb = adf_vf_void_noop; + hw_data->disable_iov = adf_vf2pf_notify_shutdown; + hw_data->get_accel_mask = get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_num_accels = get_num_accels; + hw_data->get_num_aes = get_num_aes; + hw_data->get_etr_bar_id = get_etr_bar_id; + hw_data->get_misc_bar_id = get_misc_bar_id; + hw_data->get_sku = get_sku; + hw_data->enable_ints = adf_vf_void_noop; + hw_data->dev_class->instances++; + hw_data->dev_config = adf_gen2_dev_config; + adf_devmgr_update_class_index(hw_data); + adf_gen2_init_vf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen2_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen2_init_dc_ops(&hw_data->dc_ops); +} + +void adf_clean_hw_data_c62xiov(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; + adf_devmgr_update_class_index(hw_data); +} diff --git a/drivers/crypto/intel/qat/qat_c62xvf/adf_c62xvf_hw_data.h b/drivers/crypto/intel/qat/qat_c62xvf/adf_c62xvf_hw_data.h new file mode 100644 index 000000000000..a1a62c003ebf --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62xvf/adf_c62xvf_hw_data.h @@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2015 - 2020 Intel Corporation */ +#ifndef ADF_C62XVF_HW_DATA_H_ +#define ADF_C62XVF_HW_DATA_H_ + +#define ADF_C62XIOV_PMISC_BAR 1 +#define ADF_C62XIOV_ACCELERATORS_MASK 0x1 +#define ADF_C62XIOV_ACCELENGINES_MASK 0x1 +#define ADF_C62XIOV_MAX_ACCELERATORS 1 +#define ADF_C62XIOV_MAX_ACCELENGINES 1 +#define ADF_C62XIOV_RX_RINGS_OFFSET 8 +#define ADF_C62XIOV_TX_RINGS_MASK 0xFF +#define ADF_C62XIOV_ETR_BAR 0 +#define ADF_C62XIOV_ETR_MAX_BANKS 1 + +void adf_init_hw_data_c62xiov(struct adf_hw_device_data *hw_data); +void adf_clean_hw_data_c62xiov(struct adf_hw_device_data *hw_data); +#endif diff --git a/drivers/crypto/intel/qat/qat_c62xvf/adf_drv.c b/drivers/crypto/intel/qat/qat_c62xvf/adf_drv.c new file mode 100644 index 000000000000..4840d44bbd5b --- /dev/null +++ b/drivers/crypto/intel/qat/qat_c62xvf/adf_drv.c @@ -0,0 +1,229 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/fs.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/platform_device.h> +#include <linux/workqueue.h> +#include <linux/io.h> +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_cfg.h> +#include <adf_dbgfs.h> +#include "adf_c62xvf_hw_data.h" + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_C62X_VF), }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static int adf_probe(struct pci_dev *dev, const struct pci_device_id *ent); +static void adf_remove(struct pci_dev *dev); + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_C62XVF_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, +}; + +static void adf_cleanup_pci_dev(struct adf_accel_dev *accel_dev) +{ + pci_release_regions(accel_dev->accel_pci_dev.pci_dev); + pci_disable_device(accel_dev->accel_pci_dev.pci_dev); +} + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *accel_pci_dev = &accel_dev->accel_pci_dev; + struct adf_accel_dev *pf; + int i; + + for (i = 0; i < ADF_PCI_MAX_BARS; i++) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i]; + + if (bar->virt_addr) + pci_iounmap(accel_pci_dev->pci_dev, bar->virt_addr); + } + + if (accel_dev->hw_device) { + switch (accel_pci_dev->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_C62X_VF: + adf_clean_hw_data_c62xiov(accel_dev->hw_device); + break; + default: + break; + } + kfree(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + pf = adf_devmgr_pci_to_accel_dev(accel_pci_dev->pci_dev->physfn); + adf_devmgr_rm_dev(accel_dev, pf); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_dev *pf; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + int ret; + + switch (ent->device) { + case PCI_DEVICE_ID_INTEL_QAT_C62X_VF: + break; + default: + dev_err(&pdev->dev, "Invalid device 0x%x.\n", ent->device); + return -ENODEV; + } + + accel_dev = kzalloc_node(sizeof(*accel_dev), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!accel_dev) + return -ENOMEM; + + accel_dev->is_vf = true; + pf = adf_devmgr_pci_to_accel_dev(pdev->physfn); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* Add accel device to accel table */ + if (adf_devmgr_add_dev(accel_dev, pf)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + kfree(accel_dev); + return -EFAULT; + } + INIT_LIST_HEAD(&accel_dev->crypto_list); + + accel_dev->owner = THIS_MODULE; + /* Allocate and configure device configuration structure */ + hw_data = kzalloc_node(sizeof(*hw_data), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + accel_dev->hw_device = hw_data; + adf_init_hw_data_c62xiov(accel_dev->hw_device); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + /* enable PCI device */ + if (pci_enable_device(pdev)) { + ret = -EFAULT; + goto out_err; + } + + /* set dma identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(48)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration\n"); + goto out_err_disable; + } + + if (pci_request_regions(pdev, ADF_C62XVF_DEVICE_NAME)) { + ret = -EFAULT; + goto out_err_disable; + } + + /* Find and map all the device's BARS */ + i = 0; + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM); + for_each_set_bit(bar_nr, &bar_mask, ADF_PCI_MAX_BARS * 2) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i++]; + + bar->base_addr = pci_resource_start(pdev, bar_nr); + if (!bar->base_addr) + break; + bar->size = pci_resource_len(pdev, bar_nr); + bar->virt_addr = pci_iomap(accel_pci_dev->pci_dev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to map BAR %d\n", bar_nr); + ret = -EFAULT; + goto out_err_free_reg; + } + } + pci_set_master(pdev); + /* Completion for VF2PF request/response message exchange */ + init_completion(&accel_dev->vf.msg_received); + + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, false); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err_free_reg: + pci_release_regions(accel_pci_dev->pci_dev); +out_err_disable: + pci_disable_device(accel_pci_dev->pci_dev); +out_err: + adf_cleanup_accel(accel_dev); + kfree(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_flush_vf_wq(accel_dev); + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); + adf_cleanup_pci_dev(accel_dev); + kfree(accel_dev); +} + +static int __init adfdrv_init(void) +{ + request_module("intel_qat"); + + if (pci_register_driver(&adf_driver)) { + pr_err("QAT: Driver initialization failed\n"); + return -EFAULT; + } + return 0; +} + +static void __exit adfdrv_release(void) +{ + pci_unregister_driver(&adf_driver); + adf_clean_vf_map(true); +} + +module_init(adfdrv_init); +module_exit(adfdrv_release); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_common/Makefile b/drivers/crypto/intel/qat/qat_common/Makefile new file mode 100644 index 000000000000..89845754841b --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/Makefile @@ -0,0 +1,67 @@ +# SPDX-License-Identifier: GPL-2.0 +obj-$(CONFIG_CRYPTO_DEV_QAT) += intel_qat.o +ccflags-y += -DDEFAULT_SYMBOL_NAMESPACE='"CRYPTO_QAT"' +intel_qat-y := adf_accel_engine.o \ + adf_admin.o \ + adf_aer.o \ + adf_bank_state.o \ + adf_cfg.o \ + adf_cfg_services.o \ + adf_clock.o \ + adf_ctl_drv.o \ + adf_dc.o \ + adf_dev_mgr.o \ + adf_gen2_config.o \ + adf_gen2_hw_csr_data.o \ + adf_gen2_hw_data.o \ + adf_gen4_config.o \ + adf_gen4_hw_csr_data.o \ + adf_gen4_hw_data.o \ + adf_gen4_pm.o \ + adf_gen4_ras.o \ + adf_gen4_vf_mig.o \ + adf_gen6_ras.o \ + adf_gen6_shared.o \ + adf_hw_arbiter.o \ + adf_init.o \ + adf_isr.o \ + adf_mstate_mgr.o \ + adf_rl_admin.o \ + adf_rl.o \ + adf_sysfs.o \ + adf_sysfs_ras_counters.o \ + adf_sysfs_rl.o \ + adf_timer.o \ + adf_transport.o \ + qat_algs.o \ + qat_algs_send.o \ + qat_asym_algs.o \ + qat_bl.o \ + qat_comp_algs.o \ + qat_compression.o \ + qat_crypto.o \ + qat_hal.o \ + qat_mig_dev.o \ + qat_uclo.o + +intel_qat-$(CONFIG_DEBUG_FS) += adf_cnv_dbgfs.o \ + adf_dbgfs.o \ + adf_fw_counters.o \ + adf_gen4_pm_debugfs.o \ + adf_gen4_tl.o \ + adf_gen6_pm_dbgfs.o \ + adf_gen6_tl.o \ + adf_heartbeat_dbgfs.o \ + adf_heartbeat.o \ + adf_pm_dbgfs.o \ + adf_pm_dbgfs_utils.o \ + adf_telemetry.o \ + adf_tl_debugfs.o \ + adf_transport_debug.o + +intel_qat-$(CONFIG_PCI_IOV) += adf_gen2_pfvf.o adf_gen4_pfvf.o \ + adf_pfvf_pf_msg.o adf_pfvf_pf_proto.o \ + adf_pfvf_utils.o adf_pfvf_vf_msg.o \ + adf_pfvf_vf_proto.o adf_sriov.o adf_vf_isr.o + +intel_qat-$(CONFIG_CRYPTO_DEV_QAT_ERROR_INJECTION) += adf_heartbeat_inject.o diff --git a/drivers/crypto/intel/qat/qat_common/adf_accel_devices.h b/drivers/crypto/intel/qat/qat_common/adf_accel_devices.h new file mode 100644 index 000000000000..9fe3239f0114 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_accel_devices.h @@ -0,0 +1,494 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_ACCEL_DEVICES_H_ +#define ADF_ACCEL_DEVICES_H_ +#include <linux/interrupt.h> +#include <linux/module.h> +#include <linux/list.h> +#include <linux/io.h> +#include <linux/pci.h> +#include <linux/ratelimit.h> +#include <linux/types.h> +#include <linux/qat/qat_mig_dev.h> +#include <linux/wordpart.h> +#include "adf_cfg_common.h" +#include "adf_dc.h" +#include "adf_rl.h" +#include "adf_telemetry.h" +#include "adf_pfvf_msg.h" +#include "icp_qat_hw.h" + +#define ADF_DH895XCC_DEVICE_NAME "dh895xcc" +#define ADF_DH895XCCVF_DEVICE_NAME "dh895xccvf" +#define ADF_C62X_DEVICE_NAME "c6xx" +#define ADF_C62XVF_DEVICE_NAME "c6xxvf" +#define ADF_C3XXX_DEVICE_NAME "c3xxx" +#define ADF_C3XXXVF_DEVICE_NAME "c3xxxvf" +#define ADF_4XXX_DEVICE_NAME "4xxx" +#define ADF_420XX_DEVICE_NAME "420xx" +#define ADF_6XXX_DEVICE_NAME "6xxx" +#define PCI_DEVICE_ID_INTEL_QAT_4XXX 0x4940 +#define PCI_DEVICE_ID_INTEL_QAT_4XXXIOV 0x4941 +#define PCI_DEVICE_ID_INTEL_QAT_401XX 0x4942 +#define PCI_DEVICE_ID_INTEL_QAT_401XXIOV 0x4943 +#define PCI_DEVICE_ID_INTEL_QAT_402XX 0x4944 +#define PCI_DEVICE_ID_INTEL_QAT_402XXIOV 0x4945 +#define PCI_DEVICE_ID_INTEL_QAT_420XX 0x4946 +#define PCI_DEVICE_ID_INTEL_QAT_420XXIOV 0x4947 +#define PCI_DEVICE_ID_INTEL_QAT_6XXX 0x4948 +#define PCI_DEVICE_ID_INTEL_QAT_6XXX_IOV 0x4949 + +#define ADF_DEVICE_FUSECTL_OFFSET 0x40 +#define ADF_DEVICE_LEGFUSE_OFFSET 0x4C +#define ADF_DEVICE_FUSECTL_MASK 0x80000000 +#define ADF_PCI_MAX_BARS 3 +#define ADF_DEVICE_NAME_LENGTH 32 +#define ADF_ETR_MAX_RINGS_PER_BANK 16 +#define ADF_MAX_MSIX_VECTOR_NAME 48 +#define ADF_DEVICE_NAME_PREFIX "qat_" + +enum adf_accel_capabilities { + ADF_ACCEL_CAPABILITIES_NULL = 0, + ADF_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC = 1, + ADF_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC = 2, + ADF_ACCEL_CAPABILITIES_CIPHER = 4, + ADF_ACCEL_CAPABILITIES_AUTHENTICATION = 8, + ADF_ACCEL_CAPABILITIES_COMPRESSION = 32, + ADF_ACCEL_CAPABILITIES_LZS_COMPRESSION = 64, + ADF_ACCEL_CAPABILITIES_RANDOM_NUMBER = 128 +}; + +enum adf_fuses { + ADF_FUSECTL0, + ADF_FUSECTL1, + ADF_FUSECTL2, + ADF_FUSECTL3, + ADF_FUSECTL4, + ADF_FUSECTL5, + ADF_MAX_FUSES +}; + +struct adf_bar { + resource_size_t base_addr; + void __iomem *virt_addr; + resource_size_t size; +}; + +struct adf_irq { + bool enabled; + char name[ADF_MAX_MSIX_VECTOR_NAME]; +}; + +struct adf_accel_msix { + struct adf_irq *irqs; + u32 num_entries; +}; + +struct adf_accel_pci { + struct pci_dev *pci_dev; + struct adf_accel_msix msix_entries; + struct adf_bar pci_bars[ADF_PCI_MAX_BARS]; + u8 revid; + u8 sku; +}; + +enum dev_state { + DEV_DOWN = 0, + DEV_UP +}; + +enum dev_sku_info { + DEV_SKU_1 = 0, + DEV_SKU_2, + DEV_SKU_3, + DEV_SKU_4, + DEV_SKU_VF, + DEV_SKU_UNKNOWN, +}; + +enum ras_errors { + ADF_RAS_CORR, + ADF_RAS_UNCORR, + ADF_RAS_FATAL, + ADF_RAS_ERRORS, +}; + +struct adf_error_counters { + atomic_t counter[ADF_RAS_ERRORS]; + bool sysfs_added; + bool enabled; +}; + +static inline const char *get_sku_info(enum dev_sku_info info) +{ + switch (info) { + case DEV_SKU_1: + return "SKU1"; + case DEV_SKU_2: + return "SKU2"; + case DEV_SKU_3: + return "SKU3"; + case DEV_SKU_4: + return "SKU4"; + case DEV_SKU_VF: + return "SKUVF"; + case DEV_SKU_UNKNOWN: + default: + break; + } + return "Unknown SKU"; +} + +struct adf_hw_device_class { + const char *name; + const enum adf_device_type type; + u32 instances; +}; + +struct arb_info { + u32 arb_cfg; + u32 arb_offset; + u32 wt2sam_offset; +}; + +struct admin_info { + u32 admin_msg_ur; + u32 admin_msg_lr; + u32 mailbox_offset; +}; + +struct adf_bank_state; + +struct adf_hw_csr_ops { + u64 (*build_csr_ring_base_addr)(dma_addr_t addr, u32 size); + u32 (*read_csr_ring_head)(void __iomem *csr_base_addr, u32 bank, + u32 ring); + void (*write_csr_ring_head)(void __iomem *csr_base_addr, u32 bank, + u32 ring, u32 value); + u32 (*read_csr_ring_tail)(void __iomem *csr_base_addr, u32 bank, + u32 ring); + void (*write_csr_ring_tail)(void __iomem *csr_base_addr, u32 bank, + u32 ring, u32 value); + u32 (*read_csr_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_uo_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_e_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_ne_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_nf_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_f_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_c_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_exp_stat)(void __iomem *csr_base_addr, u32 bank); + u32 (*read_csr_exp_int_en)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_exp_int_en)(void __iomem *csr_base_addr, u32 bank, + u32 value); + u32 (*read_csr_ring_config)(void __iomem *csr_base_addr, u32 bank, + u32 ring); + void (*write_csr_ring_config)(void __iomem *csr_base_addr, u32 bank, + u32 ring, u32 value); + dma_addr_t (*read_csr_ring_base)(void __iomem *csr_base_addr, u32 bank, + u32 ring); + void (*write_csr_ring_base)(void __iomem *csr_base_addr, u32 bank, + u32 ring, dma_addr_t addr); + u32 (*read_csr_int_en)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_int_en)(void __iomem *csr_base_addr, u32 bank, + u32 value); + u32 (*read_csr_int_flag)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_int_flag)(void __iomem *csr_base_addr, u32 bank, + u32 value); + u32 (*read_csr_int_srcsel)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_int_srcsel)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_int_srcsel_w_val)(void __iomem *csr_base_addr, + u32 bank, u32 value); + u32 (*read_csr_int_col_en)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_int_col_en)(void __iomem *csr_base_addr, u32 bank, + u32 value); + u32 (*read_csr_int_col_ctl)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_int_col_ctl)(void __iomem *csr_base_addr, u32 bank, + u32 value); + u32 (*read_csr_int_flag_and_col)(void __iomem *csr_base_addr, + u32 bank); + void (*write_csr_int_flag_and_col)(void __iomem *csr_base_addr, + u32 bank, u32 value); + u32 (*read_csr_ring_srv_arb_en)(void __iomem *csr_base_addr, u32 bank); + void (*write_csr_ring_srv_arb_en)(void __iomem *csr_base_addr, u32 bank, + u32 value); + u32 (*get_int_col_ctl_enable_mask)(void); +}; + +struct adf_cfg_device_data; +struct adf_accel_dev; +struct adf_etr_data; +struct adf_etr_ring_data; + +struct adf_ras_ops { + void (*enable_ras_errors)(struct adf_accel_dev *accel_dev); + void (*disable_ras_errors)(struct adf_accel_dev *accel_dev); + bool (*handle_interrupt)(struct adf_accel_dev *accel_dev, + bool *reset_required); +}; + +struct adf_pfvf_ops { + int (*enable_comms)(struct adf_accel_dev *accel_dev); + u32 (*get_pf2vf_offset)(u32 i); + u32 (*get_vf2pf_offset)(u32 i); + void (*enable_vf2pf_interrupts)(void __iomem *pmisc_addr, u32 vf_mask); + void (*disable_all_vf2pf_interrupts)(void __iomem *pmisc_addr); + u32 (*disable_pending_vf2pf_interrupts)(void __iomem *pmisc_addr); + int (*send_msg)(struct adf_accel_dev *accel_dev, struct pfvf_message msg, + u32 pfvf_offset, struct mutex *csr_lock); + struct pfvf_message (*recv_msg)(struct adf_accel_dev *accel_dev, + u32 pfvf_offset, u8 compat_ver); +}; + +struct adf_dc_ops { + int (*build_comp_block)(void *ctx, enum adf_dc_algo algo); + int (*build_decomp_block)(void *ctx, enum adf_dc_algo algo); +}; + +struct qat_migdev_ops { + int (*init)(struct qat_mig_dev *mdev); + void (*cleanup)(struct qat_mig_dev *mdev); + void (*reset)(struct qat_mig_dev *mdev); + int (*open)(struct qat_mig_dev *mdev); + void (*close)(struct qat_mig_dev *mdev); + int (*suspend)(struct qat_mig_dev *mdev); + int (*resume)(struct qat_mig_dev *mdev); + int (*save_state)(struct qat_mig_dev *mdev); + int (*save_setup)(struct qat_mig_dev *mdev); + int (*load_state)(struct qat_mig_dev *mdev); + int (*load_setup)(struct qat_mig_dev *mdev, int size); +}; + +struct adf_dev_err_mask { + u32 cppagentcmdpar_mask; + u32 parerr_ath_cph_mask; + u32 parerr_cpr_xlt_mask; + u32 parerr_dcpr_ucs_mask; + u32 parerr_pke_mask; + u32 parerr_wat_wcp_mask; + u32 ssmfeatren_mask; +}; + +struct adf_hw_device_data { + struct adf_hw_device_class *dev_class; + u32 (*get_accel_mask)(struct adf_hw_device_data *self); + u32 (*get_ae_mask)(struct adf_hw_device_data *self); + u32 (*get_accel_cap)(struct adf_accel_dev *accel_dev); + u32 (*get_sram_bar_id)(struct adf_hw_device_data *self); + u32 (*get_misc_bar_id)(struct adf_hw_device_data *self); + u32 (*get_etr_bar_id)(struct adf_hw_device_data *self); + u32 (*get_num_aes)(struct adf_hw_device_data *self); + u32 (*get_num_accels)(struct adf_hw_device_data *self); + void (*get_arb_info)(struct arb_info *arb_csrs_info); + void (*get_admin_info)(struct admin_info *admin_csrs_info); + enum dev_sku_info (*get_sku)(struct adf_hw_device_data *self); + u16 (*get_ring_to_svc_map)(struct adf_accel_dev *accel_dev); + int (*alloc_irq)(struct adf_accel_dev *accel_dev); + void (*free_irq)(struct adf_accel_dev *accel_dev); + void (*enable_error_correction)(struct adf_accel_dev *accel_dev); + int (*init_admin_comms)(struct adf_accel_dev *accel_dev); + void (*exit_admin_comms)(struct adf_accel_dev *accel_dev); + int (*send_admin_init)(struct adf_accel_dev *accel_dev); + int (*start_timer)(struct adf_accel_dev *accel_dev); + void (*stop_timer)(struct adf_accel_dev *accel_dev); + void (*check_hb_ctrs)(struct adf_accel_dev *accel_dev); + uint32_t (*get_hb_clock)(struct adf_hw_device_data *self); + int (*measure_clock)(struct adf_accel_dev *accel_dev); + int (*init_arb)(struct adf_accel_dev *accel_dev); + void (*exit_arb)(struct adf_accel_dev *accel_dev); + const u32 *(*get_arb_mapping)(struct adf_accel_dev *accel_dev); + int (*init_device)(struct adf_accel_dev *accel_dev); + int (*enable_pm)(struct adf_accel_dev *accel_dev); + bool (*handle_pm_interrupt)(struct adf_accel_dev *accel_dev); + void (*disable_iov)(struct adf_accel_dev *accel_dev); + void (*configure_iov_threads)(struct adf_accel_dev *accel_dev, + bool enable); + void (*enable_ints)(struct adf_accel_dev *accel_dev); + void (*set_ssm_wdtimer)(struct adf_accel_dev *accel_dev); + int (*ring_pair_reset)(struct adf_accel_dev *accel_dev, u32 bank_nr); + int (*bank_state_save)(struct adf_accel_dev *accel_dev, u32 bank_number, + struct adf_bank_state *state); + int (*bank_state_restore)(struct adf_accel_dev *accel_dev, + u32 bank_number, struct adf_bank_state *state); + void (*reset_device)(struct adf_accel_dev *accel_dev); + void (*set_msix_rttable)(struct adf_accel_dev *accel_dev); + const char *(*uof_get_name)(struct adf_accel_dev *accel_dev, u32 obj_num); + u32 (*uof_get_num_objs)(struct adf_accel_dev *accel_dev); + int (*uof_get_obj_type)(struct adf_accel_dev *accel_dev, u32 obj_num); + u32 (*uof_get_ae_mask)(struct adf_accel_dev *accel_dev, u32 obj_num); + int (*get_rp_group)(struct adf_accel_dev *accel_dev, u32 ae_mask); + u32 (*get_ena_thd_mask)(struct adf_accel_dev *accel_dev, u32 obj_num); + int (*dev_config)(struct adf_accel_dev *accel_dev); + bool (*services_supported)(unsigned long mask); + u32 (*get_svc_slice_cnt)(struct adf_accel_dev *accel_dev, + enum adf_base_services svc); + struct adf_pfvf_ops pfvf_ops; + struct adf_hw_csr_ops csr_ops; + struct adf_dc_ops dc_ops; + struct adf_ras_ops ras_ops; + struct adf_dev_err_mask dev_err_mask; + struct adf_rl_hw_data rl_data; + struct adf_tl_hw_data tl_data; + struct qat_migdev_ops vfmig_ops; + const char *fw_name; + const char *fw_mmp_name; + u32 fuses[ADF_MAX_FUSES]; + u32 straps; + u32 accel_capabilities_mask; + u32 extended_dc_capabilities; + u16 fw_capabilities; + u32 clock_frequency; + u32 instance_id; + u16 accel_mask; + u32 ae_mask; + u32 admin_ae_mask; + u16 tx_rings_mask; + u16 ring_to_svc_map; + u32 thd_to_arb_map[ICP_QAT_HW_AE_DELIMITER]; + u8 tx_rx_gap; + u8 num_banks; + u16 num_banks_per_vf; + u8 num_rings_per_bank; + u8 num_accel; + u8 num_logical_accel; + u8 num_engines; + u32 num_hb_ctrs; + u8 num_rps; +}; + +/* CSR write macro */ +#define ADF_CSR_WR(csr_base, csr_offset, val) \ + __raw_writel(val, csr_base + csr_offset) +/* + * CSR write macro to handle cases where the high and low + * offsets are sparsely located. + */ +#define ADF_CSR_WR64_LO_HI(csr_base, csr_low_offset, csr_high_offset, val) \ +do { \ + ADF_CSR_WR(csr_base, csr_low_offset, lower_32_bits(val)); \ + ADF_CSR_WR(csr_base, csr_high_offset, upper_32_bits(val)); \ +} while (0) + +/* CSR read macro */ +#define ADF_CSR_RD(csr_base, csr_offset) __raw_readl(csr_base + csr_offset) + +#define ADF_CFG_NUM_SERVICES 4 +#define ADF_SRV_TYPE_BIT_LEN 3 +#define ADF_SRV_TYPE_MASK 0x7 +#define ADF_AE_ADMIN_THREAD 7 +#define ADF_NUM_THREADS_PER_AE 8 +#define ADF_NUM_PKE_STRAND 2 +#define ADF_AE_STRAND0_THREAD 8 +#define ADF_AE_STRAND1_THREAD 9 + +#define GET_DEV(accel_dev) ((accel_dev)->accel_pci_dev.pci_dev->dev) +#define GET_BARS(accel_dev) ((accel_dev)->accel_pci_dev.pci_bars) +#define GET_HW_DATA(accel_dev) (accel_dev->hw_device) +#define GET_MAX_BANKS(accel_dev) (GET_HW_DATA(accel_dev)->num_banks) +#define GET_NUM_RINGS_PER_BANK(accel_dev) \ + GET_HW_DATA(accel_dev)->num_rings_per_bank +#define GET_SRV_TYPE(accel_dev, idx) \ + (((GET_HW_DATA(accel_dev)->ring_to_svc_map) >> (ADF_SRV_TYPE_BIT_LEN * (idx))) \ + & ADF_SRV_TYPE_MASK) +#define GET_ERR_MASK(accel_dev) (&GET_HW_DATA(accel_dev)->dev_err_mask) +#define GET_MAX_ACCELENGINES(accel_dev) (GET_HW_DATA(accel_dev)->num_engines) +#define GET_CSR_OPS(accel_dev) (&(accel_dev)->hw_device->csr_ops) +#define GET_PFVF_OPS(accel_dev) (&(accel_dev)->hw_device->pfvf_ops) +#define GET_DC_OPS(accel_dev) (&(accel_dev)->hw_device->dc_ops) +#define GET_VFMIG_OPS(accel_dev) (&(accel_dev)->hw_device->vfmig_ops) +#define GET_TL_DATA(accel_dev) GET_HW_DATA(accel_dev)->tl_data +#define accel_to_pci_dev(accel_ptr) accel_ptr->accel_pci_dev.pci_dev + +struct adf_admin_comms; +struct icp_qat_fw_loader_handle; +struct adf_fw_loader_data { + struct icp_qat_fw_loader_handle *fw_loader; + const struct firmware *uof_fw; + const struct firmware *mmp_fw; +}; + +struct adf_accel_vf_info { + struct adf_accel_dev *accel_dev; + struct mutex pf2vf_lock; /* protect CSR access for PF2VF messages */ + struct mutex pfvf_mig_lock; /* protects PFVF state for migration */ + struct ratelimit_state vf2pf_ratelimit; + u32 vf_nr; + bool init; + bool restarting; + u8 vf_compat_ver; + /* + * Private area used for device migration. + * Memory allocation and free is managed by migration driver. + */ + void *mig_priv; +}; + +struct adf_dc_data { + u8 *ovf_buff; + size_t ovf_buff_sz; + dma_addr_t ovf_buff_p; +}; + +struct adf_pm { + struct dentry *debugfs_pm_status; + bool present; + int idle_irq_counters; + int throttle_irq_counters; + int fw_irq_counters; + int host_ack_counter; + int host_nack_counter; + ssize_t (*print_pm_status)(struct adf_accel_dev *accel_dev, + char __user *buf, size_t count, loff_t *pos); +}; + +struct adf_sysfs { + int ring_num; + struct rw_semaphore lock; /* protects access to the fields in this struct */ +}; + +struct adf_accel_dev { + struct adf_etr_data *transport; + struct adf_hw_device_data *hw_device; + struct adf_cfg_device_data *cfg; + struct adf_fw_loader_data *fw_loader; + struct adf_admin_comms *admin; + struct adf_telemetry *telemetry; + struct adf_dc_data *dc_data; + struct adf_pm power_management; + struct list_head crypto_list; + struct list_head compression_list; + unsigned long status; + atomic_t ref_count; + struct dentry *debugfs_dir; + struct dentry *fw_cntr_dbgfile; + struct dentry *cnv_dbgfile; + struct list_head list; + struct module *owner; + struct adf_accel_pci accel_pci_dev; + struct adf_timer *timer; + struct adf_heartbeat *heartbeat; + struct adf_rl *rate_limiting; + struct adf_sysfs sysfs; + union { + struct { + /* protects VF2PF interrupts access */ + spinlock_t vf2pf_ints_lock; + /* vf_info is non-zero when SR-IOV is init'ed */ + struct adf_accel_vf_info *vf_info; + } pf; + struct { + bool irq_enabled; + char irq_name[ADF_MAX_MSIX_VECTOR_NAME]; + struct tasklet_struct pf2vf_bh_tasklet; + struct mutex vf2pf_lock; /* protect CSR access */ + struct completion msg_received; + struct pfvf_message response; /* temp field holding pf2vf response */ + u8 pf_compat_ver; + } vf; + }; + struct adf_error_counters ras_errors; + struct mutex state_lock; /* protect state of the device */ + bool is_vf; + bool autoreset_on_error; + u32 accel_id; +}; +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_accel_engine.c b/drivers/crypto/intel/qat/qat_common/adf_accel_engine.c new file mode 100644 index 000000000000..4b5d0350fc2e --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_accel_engine.c @@ -0,0 +1,212 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/firmware.h> +#include <linux/pci.h> +#include "adf_cfg.h" +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "icp_qat_uclo.h" + +static int adf_ae_fw_load_images(struct adf_accel_dev *accel_dev, void *fw_addr, + u32 fw_size) +{ + struct adf_fw_loader_data *loader_data = accel_dev->fw_loader; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + struct icp_qat_fw_loader_handle *loader; + const char *obj_name; + u32 num_objs; + u32 ae_mask; + int i; + + loader = loader_data->fw_loader; + num_objs = hw_device->uof_get_num_objs(accel_dev); + + for (i = 0; i < num_objs; i++) { + obj_name = hw_device->uof_get_name(accel_dev, i); + ae_mask = hw_device->uof_get_ae_mask(accel_dev, i); + if (!obj_name || !ae_mask) { + dev_err(&GET_DEV(accel_dev), "Invalid UOF image\n"); + goto out_err; + } + + if (qat_uclo_set_cfg_ae_mask(loader, ae_mask)) { + dev_err(&GET_DEV(accel_dev), + "Invalid mask for UOF image\n"); + goto out_err; + } + if (qat_uclo_map_obj(loader, fw_addr, fw_size, obj_name)) { + dev_err(&GET_DEV(accel_dev), + "Failed to map UOF firmware\n"); + goto out_err; + } + if (qat_uclo_wr_all_uimage(loader)) { + dev_err(&GET_DEV(accel_dev), + "Failed to load UOF firmware\n"); + goto out_err; + } + qat_uclo_del_obj(loader); + } + + return 0; + +out_err: + adf_ae_fw_release(accel_dev); + return -EFAULT; +} + +int adf_ae_fw_load(struct adf_accel_dev *accel_dev) +{ + struct adf_fw_loader_data *loader_data = accel_dev->fw_loader; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + void *fw_addr, *mmp_addr; + u32 fw_size, mmp_size; + + if (!hw_device->fw_name) + return 0; + + if (request_firmware(&loader_data->mmp_fw, hw_device->fw_mmp_name, + &accel_dev->accel_pci_dev.pci_dev->dev)) { + dev_err(&GET_DEV(accel_dev), "Failed to load MMP firmware %s\n", + hw_device->fw_mmp_name); + return -EFAULT; + } + if (request_firmware(&loader_data->uof_fw, hw_device->fw_name, + &accel_dev->accel_pci_dev.pci_dev->dev)) { + dev_err(&GET_DEV(accel_dev), "Failed to load UOF firmware %s\n", + hw_device->fw_name); + goto out_err; + } + + fw_size = loader_data->uof_fw->size; + fw_addr = (void *)loader_data->uof_fw->data; + mmp_size = loader_data->mmp_fw->size; + mmp_addr = (void *)loader_data->mmp_fw->data; + + if (qat_uclo_wr_mimage(loader_data->fw_loader, mmp_addr, mmp_size)) { + dev_err(&GET_DEV(accel_dev), "Failed to load MMP\n"); + goto out_err; + } + + if (hw_device->uof_get_num_objs) + return adf_ae_fw_load_images(accel_dev, fw_addr, fw_size); + + if (qat_uclo_map_obj(loader_data->fw_loader, fw_addr, fw_size, NULL)) { + dev_err(&GET_DEV(accel_dev), "Failed to map FW\n"); + goto out_err; + } + if (qat_uclo_wr_all_uimage(loader_data->fw_loader)) { + dev_err(&GET_DEV(accel_dev), "Failed to load UOF\n"); + goto out_err; + } + return 0; + +out_err: + adf_ae_fw_release(accel_dev); + return -EFAULT; +} + +void adf_ae_fw_release(struct adf_accel_dev *accel_dev) +{ + struct adf_fw_loader_data *loader_data = accel_dev->fw_loader; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + + if (!hw_device->fw_name) + return; + + qat_uclo_del_obj(loader_data->fw_loader); + qat_hal_deinit(loader_data->fw_loader); + release_firmware(loader_data->uof_fw); + release_firmware(loader_data->mmp_fw); + loader_data->uof_fw = NULL; + loader_data->mmp_fw = NULL; + loader_data->fw_loader = NULL; +} + +int adf_ae_start(struct adf_accel_dev *accel_dev) +{ + struct adf_fw_loader_data *loader_data = accel_dev->fw_loader; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u32 ae_ctr; + + if (!hw_data->fw_name) + return 0; + + ae_ctr = qat_hal_start(loader_data->fw_loader); + dev_info(&GET_DEV(accel_dev), + "qat_dev%d started %d acceleration engines\n", + accel_dev->accel_id, ae_ctr); + return 0; +} + +int adf_ae_stop(struct adf_accel_dev *accel_dev) +{ + struct adf_fw_loader_data *loader_data = accel_dev->fw_loader; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u32 ae_ctr, ae, max_aes = GET_MAX_ACCELENGINES(accel_dev); + + if (!hw_data->fw_name) + return 0; + + for (ae = 0, ae_ctr = 0; ae < max_aes; ae++) { + if (hw_data->ae_mask & (1 << ae)) { + qat_hal_stop(loader_data->fw_loader, ae, 0xFF); + ae_ctr++; + } + } + dev_info(&GET_DEV(accel_dev), + "qat_dev%d stopped %d acceleration engines\n", + accel_dev->accel_id, ae_ctr); + return 0; +} + +static int adf_ae_reset(struct adf_accel_dev *accel_dev, int ae) +{ + struct adf_fw_loader_data *loader_data = accel_dev->fw_loader; + + qat_hal_reset(loader_data->fw_loader); + if (qat_hal_clr_reset(loader_data->fw_loader)) + return -EFAULT; + + return 0; +} + +int adf_ae_init(struct adf_accel_dev *accel_dev) +{ + struct adf_fw_loader_data *loader_data; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + + if (!hw_device->fw_name) + return 0; + + loader_data = kzalloc(sizeof(*loader_data), GFP_KERNEL); + if (!loader_data) + return -ENOMEM; + + accel_dev->fw_loader = loader_data; + if (qat_hal_init(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed to init the AEs\n"); + kfree(loader_data); + return -EFAULT; + } + if (adf_ae_reset(accel_dev, 0)) { + dev_err(&GET_DEV(accel_dev), "Failed to reset the AEs\n"); + qat_hal_deinit(loader_data->fw_loader); + kfree(loader_data); + return -EFAULT; + } + return 0; +} + +int adf_ae_shutdown(struct adf_accel_dev *accel_dev) +{ + struct adf_fw_loader_data *loader_data = accel_dev->fw_loader; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + + if (!hw_device->fw_name) + return 0; + + qat_hal_deinit(loader_data->fw_loader); + kfree(accel_dev->fw_loader); + accel_dev->fw_loader = NULL; + return 0; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_admin.c b/drivers/crypto/intel/qat/qat_common/adf_admin.c new file mode 100644 index 000000000000..573388c37100 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_admin.c @@ -0,0 +1,610 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/types.h> +#include <linux/mutex.h> +#include <linux/slab.h> +#include <linux/iopoll.h> +#include <linux/pci.h> +#include <linux/dma-mapping.h> +#include "adf_accel_devices.h" +#include "adf_admin.h" +#include "adf_common_drv.h" +#include "adf_cfg.h" +#include "adf_heartbeat.h" +#include "icp_qat_fw_init_admin.h" + +#define ADF_ADMIN_MAILBOX_STRIDE 0x1000 +#define ADF_ADMINMSG_LEN 32 +#define ADF_CONST_TABLE_SIZE 1024 +#define ADF_ADMIN_POLL_DELAY_US 20 +#define ADF_ADMIN_POLL_TIMEOUT_US (5 * USEC_PER_SEC) +#define ADF_ONE_AE 1 + +static const u8 const_tab[1024] __aligned(1024) = { +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x01, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x11, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x21, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x03, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x01, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x03, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x13, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x13, 0x02, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x13, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x13, +0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x23, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x33, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x06, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x01, 0x23, 0x45, 0x67, 0x89, 0xab, 0xcd, 0xef, 0xfe, 0xdc, 0xba, 0x98, 0x76, +0x54, 0x32, 0x10, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x67, 0x45, 0x23, 0x01, 0xef, 0xcd, 0xab, +0x89, 0x98, 0xba, 0xdc, 0xfe, 0x10, 0x32, 0x54, 0x76, 0xc3, 0xd2, 0xe1, 0xf0, +0x00, 0x00, 0x00, 0x00, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x40, 0x00, 0x00, 0x00, 0x00, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc1, 0x05, 0x9e, +0xd8, 0x36, 0x7c, 0xd5, 0x07, 0x30, 0x70, 0xdd, 0x17, 0xf7, 0x0e, 0x59, 0x39, +0xff, 0xc0, 0x0b, 0x31, 0x68, 0x58, 0x15, 0x11, 0x64, 0xf9, 0x8f, 0xa7, 0xbe, +0xfa, 0x4f, 0xa4, 0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x6a, 0x09, 0xe6, 0x67, 0xbb, 0x67, 0xae, +0x85, 0x3c, 0x6e, 0xf3, 0x72, 0xa5, 0x4f, 0xf5, 0x3a, 0x51, 0x0e, 0x52, 0x7f, +0x9b, 0x05, 0x68, 0x8c, 0x1f, 0x83, 0xd9, 0xab, 0x5b, 0xe0, 0xcd, 0x19, 0x05, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0xcb, 0xbb, 0x9d, 0x5d, 0xc1, 0x05, 0x9e, 0xd8, 0x62, 0x9a, 0x29, +0x2a, 0x36, 0x7c, 0xd5, 0x07, 0x91, 0x59, 0x01, 0x5a, 0x30, 0x70, 0xdd, 0x17, +0x15, 0x2f, 0xec, 0xd8, 0xf7, 0x0e, 0x59, 0x39, 0x67, 0x33, 0x26, 0x67, 0xff, +0xc0, 0x0b, 0x31, 0x8e, 0xb4, 0x4a, 0x87, 0x68, 0x58, 0x15, 0x11, 0xdb, 0x0c, +0x2e, 0x0d, 0x64, 0xf9, 0x8f, 0xa7, 0x47, 0xb5, 0x48, 0x1d, 0xbe, 0xfa, 0x4f, +0xa4, 0x06, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x6a, 0x09, 0xe6, 0x67, 0xf3, 0xbc, 0xc9, 0x08, 0xbb, +0x67, 0xae, 0x85, 0x84, 0xca, 0xa7, 0x3b, 0x3c, 0x6e, 0xf3, 0x72, 0xfe, 0x94, +0xf8, 0x2b, 0xa5, 0x4f, 0xf5, 0x3a, 0x5f, 0x1d, 0x36, 0xf1, 0x51, 0x0e, 0x52, +0x7f, 0xad, 0xe6, 0x82, 0xd1, 0x9b, 0x05, 0x68, 0x8c, 0x2b, 0x3e, 0x6c, 0x1f, +0x1f, 0x83, 0xd9, 0xab, 0xfb, 0x41, 0xbd, 0x6b, 0x5b, 0xe0, 0xcd, 0x19, 0x13, +0x7e, 0x21, 0x79, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x40, 0x00, 0x00, 0x00, 0x00, 0x16, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x80, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x40, 0x00, 0x00, 0x00, 0x00, 0x18, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x14, 0x01, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x15, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x14, 0x02, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x14, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x15, 0x02, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x15, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x24, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x25, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x24, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x25, +0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x12, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x43, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x43, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x45, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x45, 0x01, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x44, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x44, 0x01, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x2B, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x2B, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x15, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x80, 0x00, 0x00, 0x00, 0x00, 0x17, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, +0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00}; + +struct adf_admin_comms { + dma_addr_t phy_addr; + dma_addr_t const_tbl_addr; + void *virt_addr; + void *virt_tbl_addr; + void __iomem *mailbox_addr; + struct mutex lock; /* protects adf_admin_comms struct */ +}; + +static int adf_put_admin_msg_sync(struct adf_accel_dev *accel_dev, u32 ae, + void *in, void *out) +{ + int ret; + u32 status; + struct adf_admin_comms *admin = accel_dev->admin; + int offset = ae * ADF_ADMINMSG_LEN * 2; + void __iomem *mailbox = admin->mailbox_addr; + int mb_offset = ae * ADF_ADMIN_MAILBOX_STRIDE; + struct icp_qat_fw_init_admin_req *request = in; + + mutex_lock(&admin->lock); + + if (ADF_CSR_RD(mailbox, mb_offset) == 1) { + mutex_unlock(&admin->lock); + return -EAGAIN; + } + + memcpy(admin->virt_addr + offset, in, ADF_ADMINMSG_LEN); + ADF_CSR_WR(mailbox, mb_offset, 1); + + ret = read_poll_timeout(ADF_CSR_RD, status, status == 0, + ADF_ADMIN_POLL_DELAY_US, + ADF_ADMIN_POLL_TIMEOUT_US, true, + mailbox, mb_offset); + if (ret < 0) { + /* Response timeout */ + dev_err(&GET_DEV(accel_dev), + "Failed to send admin msg %d to accelerator %d\n", + request->cmd_id, ae); + } else { + /* Response received from admin message, we can now + * make response data available in "out" parameter. + */ + memcpy(out, admin->virt_addr + offset + + ADF_ADMINMSG_LEN, ADF_ADMINMSG_LEN); + } + + mutex_unlock(&admin->lock); + return ret; +} + +static int adf_send_admin(struct adf_accel_dev *accel_dev, + struct icp_qat_fw_init_admin_req *req, + struct icp_qat_fw_init_admin_resp *resp, + const unsigned long ae_mask) +{ + u32 ae; + + for_each_set_bit(ae, &ae_mask, ICP_QAT_HW_AE_DELIMITER) + if (adf_put_admin_msg_sync(accel_dev, ae, req, resp) || + resp->status) + return -EFAULT; + + return 0; +} + +static int adf_init_ae(struct adf_accel_dev *accel_dev) +{ + struct icp_qat_fw_init_admin_req req; + struct icp_qat_fw_init_admin_resp resp; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + u32 ae_mask = hw_device->ae_mask; + + memset(&req, 0, sizeof(req)); + memset(&resp, 0, sizeof(resp)); + req.cmd_id = ICP_QAT_FW_INIT_AE; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} + +static int adf_set_fw_constants(struct adf_accel_dev *accel_dev) +{ + struct icp_qat_fw_init_admin_req req; + struct icp_qat_fw_init_admin_resp resp; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + u32 ae_mask = hw_device->admin_ae_mask ?: hw_device->ae_mask; + + memset(&req, 0, sizeof(req)); + memset(&resp, 0, sizeof(resp)); + req.cmd_id = ICP_QAT_FW_CONSTANTS_CFG; + + req.init_cfg_sz = ADF_CONST_TABLE_SIZE; + req.init_cfg_ptr = accel_dev->admin->const_tbl_addr; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} + +int adf_get_fw_timestamp(struct adf_accel_dev *accel_dev, u64 *timestamp) +{ + struct icp_qat_fw_init_admin_req req = { }; + struct icp_qat_fw_init_admin_resp resp; + unsigned int ae_mask = ADF_ONE_AE; + int ret; + + req.cmd_id = ICP_QAT_FW_TIMER_GET; + ret = adf_send_admin(accel_dev, &req, &resp, ae_mask); + if (ret) + return ret; + + *timestamp = resp.timestamp; + return 0; +} + +static int adf_set_chaining(struct adf_accel_dev *accel_dev) +{ + u32 ae_mask = GET_HW_DATA(accel_dev)->ae_mask; + struct icp_qat_fw_init_admin_resp resp = { }; + struct icp_qat_fw_init_admin_req req = { }; + + req.cmd_id = ICP_QAT_FW_DC_CHAIN_INIT; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} + +static int adf_get_dc_capabilities(struct adf_accel_dev *accel_dev, + u32 *capabilities) +{ + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + struct icp_qat_fw_init_admin_resp resp; + struct icp_qat_fw_init_admin_req req; + unsigned long ae_mask; + unsigned long ae; + int ret; + + /* Target only service accelerator engines */ + ae_mask = hw_device->ae_mask & ~hw_device->admin_ae_mask; + + memset(&req, 0, sizeof(req)); + memset(&resp, 0, sizeof(resp)); + req.cmd_id = ICP_QAT_FW_COMP_CAPABILITY_GET; + + *capabilities = 0; + for_each_set_bit(ae, &ae_mask, GET_MAX_ACCELENGINES(accel_dev)) { + ret = adf_send_admin(accel_dev, &req, &resp, 1ULL << ae); + if (ret) + return ret; + + *capabilities |= resp.extended_features; + } + + return 0; +} + +int adf_get_ae_fw_counters(struct adf_accel_dev *accel_dev, u16 ae, u64 *reqs, u64 *resps) +{ + struct icp_qat_fw_init_admin_resp resp = { }; + struct icp_qat_fw_init_admin_req req = { }; + int ret; + + req.cmd_id = ICP_QAT_FW_COUNTERS_GET; + + ret = adf_put_admin_msg_sync(accel_dev, ae, &req, &resp); + if (ret || resp.status) + return -EFAULT; + + *reqs = resp.req_rec_count; + *resps = resp.resp_sent_count; + + return 0; +} + +int adf_send_admin_tim_sync(struct adf_accel_dev *accel_dev, u32 cnt) +{ + u32 ae_mask = accel_dev->hw_device->ae_mask; + struct icp_qat_fw_init_admin_req req = { }; + struct icp_qat_fw_init_admin_resp resp = { }; + + req.cmd_id = ICP_QAT_FW_SYNC; + req.int_timer_ticks = cnt; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} + +int adf_send_admin_hb_timer(struct adf_accel_dev *accel_dev, uint32_t ticks) +{ + u32 ae_mask = accel_dev->hw_device->ae_mask; + struct icp_qat_fw_init_admin_req req = { }; + struct icp_qat_fw_init_admin_resp resp; + + req.cmd_id = ICP_QAT_FW_HEARTBEAT_TIMER_SET; + req.init_cfg_ptr = accel_dev->heartbeat->dma.phy_addr; + req.heartbeat_ticks = ticks; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} + +static bool is_dcc_enabled(struct adf_accel_dev *accel_dev) +{ + char services[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = {0}; + int ret; + + ret = adf_cfg_get_param_value(accel_dev, ADF_GENERAL_SEC, + ADF_SERVICES_ENABLED, services); + if (ret) + return false; + + return !strcmp(services, "dcc"); +} + +static int adf_get_fw_capabilities(struct adf_accel_dev *accel_dev, u16 *caps) +{ + u32 ae_mask = accel_dev->hw_device->admin_ae_mask; + struct icp_qat_fw_init_admin_resp resp = { }; + struct icp_qat_fw_init_admin_req req = { }; + int ret; + + if (!ae_mask) + return 0; + + req.cmd_id = ICP_QAT_FW_CAPABILITIES_GET; + ret = adf_send_admin(accel_dev, &req, &resp, ae_mask); + if (ret) + return ret; + + *caps = resp.fw_capabilities; + + return 0; +} + +int adf_send_admin_rl_init(struct adf_accel_dev *accel_dev, + struct icp_qat_fw_init_admin_slice_cnt *slices) +{ + u32 ae_mask = accel_dev->hw_device->admin_ae_mask; + struct icp_qat_fw_init_admin_resp resp = { }; + struct icp_qat_fw_init_admin_req req = { }; + int ret; + + req.cmd_id = ICP_QAT_FW_RL_INIT; + + ret = adf_send_admin(accel_dev, &req, &resp, ae_mask); + if (ret) + return ret; + + memcpy(slices, &resp.slices, sizeof(*slices)); + + return 0; +} + +int adf_send_admin_rl_add_update(struct adf_accel_dev *accel_dev, + struct icp_qat_fw_init_admin_req *req) +{ + u32 ae_mask = accel_dev->hw_device->admin_ae_mask; + struct icp_qat_fw_init_admin_resp resp = { }; + + /* + * req struct filled in rl implementation. Used commands + * ICP_QAT_FW_RL_ADD for a new SLA + * ICP_QAT_FW_RL_UPDATE for update SLA + */ + return adf_send_admin(accel_dev, req, &resp, ae_mask); +} + +int adf_send_admin_rl_delete(struct adf_accel_dev *accel_dev, u16 node_id, + u8 node_type) +{ + u32 ae_mask = accel_dev->hw_device->admin_ae_mask; + struct icp_qat_fw_init_admin_resp resp = { }; + struct icp_qat_fw_init_admin_req req = { }; + + req.cmd_id = ICP_QAT_FW_RL_REMOVE; + req.node_id = node_id; + req.node_type = node_type; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} + +/** + * adf_send_admin_init() - Function sends init message to FW + * @accel_dev: Pointer to acceleration device. + * + * Function sends admin init message to the FW + * + * Return: 0 on success, error code otherwise. + */ +int adf_send_admin_init(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + u32 dc_capabilities = 0; + int ret; + + ret = adf_set_fw_constants(accel_dev); + if (ret) + return ret; + + if (is_dcc_enabled(accel_dev)) { + ret = adf_set_chaining(accel_dev); + if (ret) + return ret; + } + + ret = adf_get_dc_capabilities(accel_dev, &dc_capabilities); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Cannot get dc capabilities\n"); + return ret; + } + accel_dev->hw_device->extended_dc_capabilities = dc_capabilities; + + adf_get_fw_capabilities(accel_dev, &hw_data->fw_capabilities); + + return adf_init_ae(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_send_admin_init); + +/** + * adf_init_admin_pm() - Function sends PM init message to FW + * @accel_dev: Pointer to acceleration device. + * @idle_delay: QAT HW idle time before power gating is initiated. + * 000 - 64us + * 001 - 128us + * 010 - 256us + * 011 - 512us + * 100 - 1ms + * 101 - 2ms + * 110 - 4ms + * 111 - 8ms + * + * Function sends to the FW the admin init message for the PM state + * configuration. + * + * Return: 0 on success, error code otherwise. + */ +int adf_init_admin_pm(struct adf_accel_dev *accel_dev, u32 idle_delay) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct icp_qat_fw_init_admin_resp resp = {0}; + struct icp_qat_fw_init_admin_req req = {0}; + u32 ae_mask = hw_data->admin_ae_mask; + + if (!accel_dev->admin) { + dev_err(&GET_DEV(accel_dev), "adf_admin is not available\n"); + return -EFAULT; + } + + req.cmd_id = ICP_QAT_FW_PM_STATE_CONFIG; + req.idle_filter = idle_delay; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} +EXPORT_SYMBOL_GPL(adf_init_admin_pm); + +int adf_get_pm_info(struct adf_accel_dev *accel_dev, dma_addr_t p_state_addr, + size_t buff_size) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct icp_qat_fw_init_admin_req req = { }; + struct icp_qat_fw_init_admin_resp resp; + u32 ae_mask = hw_data->admin_ae_mask; + int ret; + + /* Query pm info via init/admin cmd */ + if (!accel_dev->admin) { + dev_err(&GET_DEV(accel_dev), "adf_admin is not available\n"); + return -EFAULT; + } + + req.cmd_id = ICP_QAT_FW_PM_INFO; + req.init_cfg_sz = buff_size; + req.init_cfg_ptr = p_state_addr; + + ret = adf_send_admin(accel_dev, &req, &resp, ae_mask); + if (ret) + dev_err(&GET_DEV(accel_dev), + "Failed to query power-management info\n"); + + return ret; +} + +int adf_get_cnv_stats(struct adf_accel_dev *accel_dev, u16 ae, u16 *err_cnt, + u16 *latest_err) +{ + struct icp_qat_fw_init_admin_req req = { }; + struct icp_qat_fw_init_admin_resp resp; + int ret; + + req.cmd_id = ICP_QAT_FW_CNV_STATS_GET; + + ret = adf_put_admin_msg_sync(accel_dev, ae, &req, &resp); + if (ret) + return ret; + if (resp.status) + return -EPROTONOSUPPORT; + + *err_cnt = resp.error_count; + *latest_err = resp.latest_error; + + return ret; +} + +int adf_send_admin_tl_start(struct adf_accel_dev *accel_dev, + dma_addr_t tl_dma_addr, size_t layout_sz, u8 *rp_indexes, + struct icp_qat_fw_init_admin_slice_cnt *slice_count) +{ + u32 ae_mask = GET_HW_DATA(accel_dev)->admin_ae_mask; + struct icp_qat_fw_init_admin_resp resp = { }; + struct icp_qat_fw_init_admin_req req = { }; + int ret; + + req.cmd_id = ICP_QAT_FW_TL_START; + req.init_cfg_ptr = tl_dma_addr; + req.init_cfg_sz = layout_sz; + + if (rp_indexes) + memcpy(&req.rp_indexes, rp_indexes, sizeof(req.rp_indexes)); + + ret = adf_send_admin(accel_dev, &req, &resp, ae_mask); + if (ret) + return ret; + + memcpy(slice_count, &resp.slices, sizeof(*slice_count)); + + return 0; +} + +int adf_send_admin_tl_stop(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct icp_qat_fw_init_admin_resp resp = { }; + struct icp_qat_fw_init_admin_req req = { }; + u32 ae_mask = hw_data->admin_ae_mask; + + req.cmd_id = ICP_QAT_FW_TL_STOP; + + return adf_send_admin(accel_dev, &req, &resp, ae_mask); +} + +int adf_init_admin_comms(struct adf_accel_dev *accel_dev) +{ + struct adf_admin_comms *admin; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + struct admin_info admin_csrs_info; + u32 mailbox_offset, adminmsg_u, adminmsg_l; + void __iomem *mailbox; + u64 reg_val; + + admin = kzalloc_node(sizeof(*accel_dev->admin), GFP_KERNEL, + dev_to_node(&GET_DEV(accel_dev))); + if (!admin) + return -ENOMEM; + admin->virt_addr = dma_alloc_coherent(&GET_DEV(accel_dev), PAGE_SIZE, + &admin->phy_addr, GFP_KERNEL); + if (!admin->virt_addr) { + dev_err(&GET_DEV(accel_dev), "Failed to allocate dma buff\n"); + kfree(admin); + return -ENOMEM; + } + + admin->virt_tbl_addr = dma_alloc_coherent(&GET_DEV(accel_dev), + PAGE_SIZE, + &admin->const_tbl_addr, + GFP_KERNEL); + if (!admin->virt_tbl_addr) { + dev_err(&GET_DEV(accel_dev), "Failed to allocate const_tbl\n"); + dma_free_coherent(&GET_DEV(accel_dev), PAGE_SIZE, + admin->virt_addr, admin->phy_addr); + kfree(admin); + return -ENOMEM; + } + + memcpy(admin->virt_tbl_addr, const_tab, sizeof(const_tab)); + hw_data->get_admin_info(&admin_csrs_info); + + mailbox_offset = admin_csrs_info.mailbox_offset; + mailbox = pmisc_addr + mailbox_offset; + adminmsg_u = admin_csrs_info.admin_msg_ur; + adminmsg_l = admin_csrs_info.admin_msg_lr; + + reg_val = (u64)admin->phy_addr; + ADF_CSR_WR(pmisc_addr, adminmsg_u, upper_32_bits(reg_val)); + ADF_CSR_WR(pmisc_addr, adminmsg_l, lower_32_bits(reg_val)); + + mutex_init(&admin->lock); + admin->mailbox_addr = mailbox; + accel_dev->admin = admin; + return 0; +} +EXPORT_SYMBOL_GPL(adf_init_admin_comms); + +void adf_exit_admin_comms(struct adf_accel_dev *accel_dev) +{ + struct adf_admin_comms *admin = accel_dev->admin; + + if (!admin) + return; + + if (admin->virt_addr) + dma_free_coherent(&GET_DEV(accel_dev), PAGE_SIZE, + admin->virt_addr, admin->phy_addr); + if (admin->virt_tbl_addr) + dma_free_coherent(&GET_DEV(accel_dev), PAGE_SIZE, + admin->virt_tbl_addr, admin->const_tbl_addr); + + mutex_destroy(&admin->lock); + kfree(admin); + accel_dev->admin = NULL; +} +EXPORT_SYMBOL_GPL(adf_exit_admin_comms); diff --git a/drivers/crypto/intel/qat/qat_common/adf_admin.h b/drivers/crypto/intel/qat/qat_common/adf_admin.h new file mode 100644 index 000000000000..647c8e196752 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_admin.h @@ -0,0 +1,31 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_ADMIN +#define ADF_ADMIN + +#include "icp_qat_fw_init_admin.h" + +struct adf_accel_dev; + +int adf_init_admin_comms(struct adf_accel_dev *accel_dev); +void adf_exit_admin_comms(struct adf_accel_dev *accel_dev); +int adf_send_admin_init(struct adf_accel_dev *accel_dev); +int adf_get_ae_fw_counters(struct adf_accel_dev *accel_dev, u16 ae, u64 *reqs, u64 *resps); +int adf_init_admin_pm(struct adf_accel_dev *accel_dev, u32 idle_delay); +int adf_send_admin_tim_sync(struct adf_accel_dev *accel_dev, u32 cnt); +int adf_send_admin_hb_timer(struct adf_accel_dev *accel_dev, uint32_t ticks); +int adf_send_admin_rl_init(struct adf_accel_dev *accel_dev, + struct icp_qat_fw_init_admin_slice_cnt *slices); +int adf_send_admin_rl_add_update(struct adf_accel_dev *accel_dev, + struct icp_qat_fw_init_admin_req *req); +int adf_send_admin_rl_delete(struct adf_accel_dev *accel_dev, u16 node_id, + u8 node_type); +int adf_get_fw_timestamp(struct adf_accel_dev *accel_dev, u64 *timestamp); +int adf_get_pm_info(struct adf_accel_dev *accel_dev, dma_addr_t p_state_addr, size_t buff_size); +int adf_get_cnv_stats(struct adf_accel_dev *accel_dev, u16 ae, u16 *err_cnt, u16 *latest_err); +int adf_send_admin_tl_start(struct adf_accel_dev *accel_dev, + dma_addr_t tl_dma_addr, size_t layout_sz, u8 *rp_indexes, + struct icp_qat_fw_init_admin_slice_cnt *slice_count); +int adf_send_admin_tl_stop(struct adf_accel_dev *accel_dev); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_aer.c b/drivers/crypto/intel/qat/qat_common/adf_aer.c new file mode 100644 index 000000000000..11728cf32653 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_aer.c @@ -0,0 +1,300 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/pci.h> +#include <linux/completion.h> +#include <linux/workqueue.h> +#include <linux/delay.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_pfvf_pf_msg.h" + +struct adf_fatal_error_data { + struct adf_accel_dev *accel_dev; + struct work_struct work; +}; + +static struct workqueue_struct *device_reset_wq; +static struct workqueue_struct *device_sriov_wq; + +static pci_ers_result_t adf_error_detected(struct pci_dev *pdev, + pci_channel_state_t state) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + dev_info(&pdev->dev, "Acceleration driver hardware error detected.\n"); + if (!accel_dev) { + dev_err(&pdev->dev, "Can't find acceleration device\n"); + return PCI_ERS_RESULT_DISCONNECT; + } + + if (state == pci_channel_io_perm_failure) { + dev_err(&pdev->dev, "Can't recover from device error\n"); + return PCI_ERS_RESULT_DISCONNECT; + } + + set_bit(ADF_STATUS_RESTARTING, &accel_dev->status); + if (accel_dev->hw_device->exit_arb) { + dev_dbg(&pdev->dev, "Disabling arbitration\n"); + accel_dev->hw_device->exit_arb(accel_dev); + } + adf_error_notifier(accel_dev); + adf_pf2vf_notify_fatal_error(accel_dev); + adf_dev_restarting_notify(accel_dev); + adf_pf2vf_notify_restarting(accel_dev); + adf_pf2vf_wait_for_restarting_complete(accel_dev); + pci_clear_master(pdev); + adf_dev_down(accel_dev); + + return PCI_ERS_RESULT_NEED_RESET; +} + +/* reset dev data */ +struct adf_reset_dev_data { + int mode; + struct adf_accel_dev *accel_dev; + struct completion compl; + struct work_struct reset_work; +}; + +/* sriov dev data */ +struct adf_sriov_dev_data { + struct adf_accel_dev *accel_dev; + struct completion compl; + struct work_struct sriov_work; +}; + +void adf_reset_sbr(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + struct pci_dev *parent = pdev->bus->self; + u16 bridge_ctl = 0; + + if (!parent) + parent = pdev; + + if (!pci_wait_for_pending_transaction(pdev)) + dev_info(&GET_DEV(accel_dev), + "Transaction still in progress. Proceeding\n"); + + dev_info(&GET_DEV(accel_dev), "Secondary bus reset\n"); + + pci_read_config_word(parent, PCI_BRIDGE_CONTROL, &bridge_ctl); + bridge_ctl |= PCI_BRIDGE_CTL_BUS_RESET; + pci_write_config_word(parent, PCI_BRIDGE_CONTROL, bridge_ctl); + msleep(100); + bridge_ctl &= ~PCI_BRIDGE_CTL_BUS_RESET; + pci_write_config_word(parent, PCI_BRIDGE_CONTROL, bridge_ctl); + msleep(100); +} +EXPORT_SYMBOL_GPL(adf_reset_sbr); + +void adf_reset_flr(struct adf_accel_dev *accel_dev) +{ + pcie_flr(accel_to_pci_dev(accel_dev)); +} +EXPORT_SYMBOL_GPL(adf_reset_flr); + +void adf_dev_restore(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + + if (hw_device->reset_device) { + dev_info(&GET_DEV(accel_dev), "Resetting device qat_dev%d\n", + accel_dev->accel_id); + hw_device->reset_device(accel_dev); + pci_restore_state(pdev); + } +} + +static void adf_device_sriov_worker(struct work_struct *work) +{ + struct adf_sriov_dev_data *sriov_data = + container_of(work, struct adf_sriov_dev_data, sriov_work); + + adf_reenable_sriov(sriov_data->accel_dev); + complete(&sriov_data->compl); +} + +static void adf_device_reset_worker(struct work_struct *work) +{ + struct adf_reset_dev_data *reset_data = + container_of(work, struct adf_reset_dev_data, reset_work); + struct adf_accel_dev *accel_dev = reset_data->accel_dev; + unsigned long wait_jiffies = msecs_to_jiffies(10000); + struct adf_sriov_dev_data sriov_data; + + adf_dev_restarting_notify(accel_dev); + if (adf_dev_restart(accel_dev)) { + /* The device hanged and we can't restart it so stop here */ + dev_err(&GET_DEV(accel_dev), "Restart device failed\n"); + if (reset_data->mode == ADF_DEV_RESET_ASYNC) + kfree(reset_data); + WARN(1, "QAT: device restart failed. Device is unusable\n"); + return; + } + + sriov_data.accel_dev = accel_dev; + init_completion(&sriov_data.compl); + INIT_WORK(&sriov_data.sriov_work, adf_device_sriov_worker); + queue_work(device_sriov_wq, &sriov_data.sriov_work); + if (wait_for_completion_timeout(&sriov_data.compl, wait_jiffies)) + adf_pf2vf_notify_restarted(accel_dev); + + adf_dev_restarted_notify(accel_dev); + clear_bit(ADF_STATUS_RESTARTING, &accel_dev->status); + + /* The dev is back alive. Notify the caller if in sync mode */ + if (reset_data->mode == ADF_DEV_RESET_ASYNC) + kfree(reset_data); + else + complete(&reset_data->compl); +} + +static int adf_dev_aer_schedule_reset(struct adf_accel_dev *accel_dev, + enum adf_dev_reset_mode mode) +{ + struct adf_reset_dev_data *reset_data; + + if (!adf_dev_started(accel_dev) || + test_bit(ADF_STATUS_RESTARTING, &accel_dev->status)) + return 0; + + set_bit(ADF_STATUS_RESTARTING, &accel_dev->status); + reset_data = kzalloc(sizeof(*reset_data), GFP_KERNEL); + if (!reset_data) + return -ENOMEM; + reset_data->accel_dev = accel_dev; + init_completion(&reset_data->compl); + reset_data->mode = mode; + INIT_WORK(&reset_data->reset_work, adf_device_reset_worker); + queue_work(device_reset_wq, &reset_data->reset_work); + + /* If in sync mode wait for the result */ + if (mode == ADF_DEV_RESET_SYNC) { + int ret = 0; + /* Maximum device reset time is 10 seconds */ + unsigned long wait_jiffies = msecs_to_jiffies(10000); + unsigned long timeout = wait_for_completion_timeout( + &reset_data->compl, wait_jiffies); + if (!timeout) { + dev_err(&GET_DEV(accel_dev), + "Reset device timeout expired\n"); + cancel_work_sync(&reset_data->reset_work); + ret = -EFAULT; + } + kfree(reset_data); + return ret; + } + return 0; +} + +static pci_ers_result_t adf_slot_reset(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + int res = 0; + + if (!accel_dev) { + pr_err("QAT: Can't find acceleration device\n"); + return PCI_ERS_RESULT_DISCONNECT; + } + + if (!pdev->is_busmaster) + pci_set_master(pdev); + pci_restore_state(pdev); + res = adf_dev_up(accel_dev, false); + if (res && res != -EALREADY) + return PCI_ERS_RESULT_DISCONNECT; + + adf_reenable_sriov(accel_dev); + adf_pf2vf_notify_restarted(accel_dev); + adf_dev_restarted_notify(accel_dev); + clear_bit(ADF_STATUS_RESTARTING, &accel_dev->status); + return PCI_ERS_RESULT_RECOVERED; +} + +static void adf_resume(struct pci_dev *pdev) +{ + dev_info(&pdev->dev, "Acceleration driver reset completed\n"); + dev_info(&pdev->dev, "Device is up and running\n"); +} + +const struct pci_error_handlers adf_err_handler = { + .error_detected = adf_error_detected, + .slot_reset = adf_slot_reset, + .resume = adf_resume, +}; +EXPORT_SYMBOL_GPL(adf_err_handler); + +static int adf_dev_autoreset(struct adf_accel_dev *accel_dev) +{ + if (accel_dev->autoreset_on_error) + return adf_dev_aer_schedule_reset(accel_dev, ADF_DEV_RESET_ASYNC); + + return 0; +} + +static void adf_notify_fatal_error_worker(struct work_struct *work) +{ + struct adf_fatal_error_data *wq_data = + container_of(work, struct adf_fatal_error_data, work); + struct adf_accel_dev *accel_dev = wq_data->accel_dev; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + + adf_error_notifier(accel_dev); + + if (!accel_dev->is_vf) { + /* Disable arbitration to stop processing of new requests */ + if (accel_dev->autoreset_on_error && hw_device->exit_arb) + hw_device->exit_arb(accel_dev); + if (accel_dev->pf.vf_info) + adf_pf2vf_notify_fatal_error(accel_dev); + adf_dev_autoreset(accel_dev); + } + + kfree(wq_data); +} + +int adf_notify_fatal_error(struct adf_accel_dev *accel_dev) +{ + struct adf_fatal_error_data *wq_data; + + wq_data = kzalloc(sizeof(*wq_data), GFP_ATOMIC); + if (!wq_data) + return -ENOMEM; + + wq_data->accel_dev = accel_dev; + INIT_WORK(&wq_data->work, adf_notify_fatal_error_worker); + adf_misc_wq_queue_work(&wq_data->work); + + return 0; +} + +int adf_init_aer(void) +{ + device_reset_wq = alloc_workqueue("qat_device_reset_wq", + WQ_MEM_RECLAIM | WQ_PERCPU, 0); + if (!device_reset_wq) + return -EFAULT; + + device_sriov_wq = alloc_workqueue("qat_device_sriov_wq", WQ_PERCPU, 0); + if (!device_sriov_wq) { + destroy_workqueue(device_reset_wq); + device_reset_wq = NULL; + return -EFAULT; + } + + return 0; +} + +void adf_exit_aer(void) +{ + if (device_reset_wq) + destroy_workqueue(device_reset_wq); + device_reset_wq = NULL; + + if (device_sriov_wq) + destroy_workqueue(device_sriov_wq); + device_sriov_wq = NULL; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_bank_state.c b/drivers/crypto/intel/qat/qat_common/adf_bank_state.c new file mode 100644 index 000000000000..225d55d56a4b --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_bank_state.c @@ -0,0 +1,238 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2025 Intel Corporation */ + +#define pr_fmt(fmt) "QAT: " fmt + +#include <linux/bits.h> +#include <linux/dev_printk.h> +#include <linux/printk.h> +#include "adf_accel_devices.h" +#include "adf_bank_state.h" +#include "adf_common_drv.h" + +/* Ring interrupt masks */ +#define ADF_RP_INT_SRC_SEL_F_RISE_MASK GENMASK(1, 0) +#define ADF_RP_INT_SRC_SEL_F_FALL_MASK GENMASK(2, 0) +#define ADF_RP_INT_SRC_SEL_RANGE_WIDTH 4 + +static inline int check_stat(u32 (*op)(void __iomem *, u32), u32 expect_val, + const char *name, void __iomem *base, u32 bank) +{ + u32 actual_val = op(base, bank); + + if (expect_val == actual_val) + return 0; + + pr_err("Fail to restore %s register. Expected %#x, actual %#x\n", + name, expect_val, actual_val); + + return -EINVAL; +} + +static void bank_state_save(struct adf_hw_csr_ops *ops, void __iomem *base, + u32 bank, struct adf_bank_state *state, u32 num_rings) +{ + u32 i; + + state->ringstat0 = ops->read_csr_stat(base, bank); + state->ringuostat = ops->read_csr_uo_stat(base, bank); + state->ringestat = ops->read_csr_e_stat(base, bank); + state->ringnestat = ops->read_csr_ne_stat(base, bank); + state->ringnfstat = ops->read_csr_nf_stat(base, bank); + state->ringfstat = ops->read_csr_f_stat(base, bank); + state->ringcstat0 = ops->read_csr_c_stat(base, bank); + state->iaintflagen = ops->read_csr_int_en(base, bank); + state->iaintflagreg = ops->read_csr_int_flag(base, bank); + state->iaintflagsrcsel0 = ops->read_csr_int_srcsel(base, bank); + state->iaintcolen = ops->read_csr_int_col_en(base, bank); + state->iaintcolctl = ops->read_csr_int_col_ctl(base, bank); + state->iaintflagandcolen = ops->read_csr_int_flag_and_col(base, bank); + state->ringexpstat = ops->read_csr_exp_stat(base, bank); + state->ringexpintenable = ops->read_csr_exp_int_en(base, bank); + state->ringsrvarben = ops->read_csr_ring_srv_arb_en(base, bank); + + for (i = 0; i < num_rings; i++) { + state->rings[i].head = ops->read_csr_ring_head(base, bank, i); + state->rings[i].tail = ops->read_csr_ring_tail(base, bank, i); + state->rings[i].config = ops->read_csr_ring_config(base, bank, i); + state->rings[i].base = ops->read_csr_ring_base(base, bank, i); + } +} + +static int bank_state_restore(struct adf_hw_csr_ops *ops, void __iomem *base, + u32 bank, struct adf_bank_state *state, u32 num_rings, + int tx_rx_gap) +{ + u32 val, tmp_val, i; + int ret; + + for (i = 0; i < num_rings; i++) + ops->write_csr_ring_base(base, bank, i, state->rings[i].base); + + for (i = 0; i < num_rings; i++) + ops->write_csr_ring_config(base, bank, i, state->rings[i].config); + + for (i = 0; i < num_rings / 2; i++) { + int tx = i * (tx_rx_gap + 1); + int rx = tx + tx_rx_gap; + + ops->write_csr_ring_head(base, bank, tx, state->rings[tx].head); + ops->write_csr_ring_tail(base, bank, tx, state->rings[tx].tail); + + /* + * The TX ring head needs to be updated again to make sure that + * the HW will not consider the ring as full when it is empty + * and the correct state flags are set to match the recovered state. + */ + if (state->ringestat & BIT(tx)) { + val = ops->read_csr_int_srcsel(base, bank); + val |= ADF_RP_INT_SRC_SEL_F_RISE_MASK; + ops->write_csr_int_srcsel_w_val(base, bank, val); + ops->write_csr_ring_head(base, bank, tx, state->rings[tx].head); + } + + ops->write_csr_ring_tail(base, bank, rx, state->rings[rx].tail); + val = ops->read_csr_int_srcsel(base, bank); + val |= ADF_RP_INT_SRC_SEL_F_RISE_MASK << ADF_RP_INT_SRC_SEL_RANGE_WIDTH; + ops->write_csr_int_srcsel_w_val(base, bank, val); + + ops->write_csr_ring_head(base, bank, rx, state->rings[rx].head); + val = ops->read_csr_int_srcsel(base, bank); + val |= ADF_RP_INT_SRC_SEL_F_FALL_MASK << ADF_RP_INT_SRC_SEL_RANGE_WIDTH; + ops->write_csr_int_srcsel_w_val(base, bank, val); + + /* + * The RX ring tail needs to be updated again to make sure that + * the HW will not consider the ring as empty when it is full + * and the correct state flags are set to match the recovered state. + */ + if (state->ringfstat & BIT(rx)) + ops->write_csr_ring_tail(base, bank, rx, state->rings[rx].tail); + } + + ops->write_csr_int_flag_and_col(base, bank, state->iaintflagandcolen); + ops->write_csr_int_en(base, bank, state->iaintflagen); + ops->write_csr_int_col_en(base, bank, state->iaintcolen); + ops->write_csr_int_srcsel_w_val(base, bank, state->iaintflagsrcsel0); + ops->write_csr_exp_int_en(base, bank, state->ringexpintenable); + ops->write_csr_int_col_ctl(base, bank, state->iaintcolctl); + + /* + * Verify whether any exceptions were raised during the bank save process. + * If exceptions occurred, the status and exception registers cannot + * be directly restored. Consequently, further restoration is not + * feasible, and the current state of the ring should be maintained. + */ + val = state->ringexpstat; + if (val) { + pr_info("Bank %u state not fully restored due to exception in saved state (%#x)\n", + bank, val); + return 0; + } + + /* Ensure that the restoration process completed without exceptions */ + tmp_val = ops->read_csr_exp_stat(base, bank); + if (tmp_val) { + pr_err("Bank %u restored with exception: %#x\n", bank, tmp_val); + return -EFAULT; + } + + ops->write_csr_ring_srv_arb_en(base, bank, state->ringsrvarben); + + /* Check that all ring statuses match the saved state. */ + ret = check_stat(ops->read_csr_stat, state->ringstat0, "ringstat", + base, bank); + if (ret) + return ret; + + ret = check_stat(ops->read_csr_e_stat, state->ringestat, "ringestat", + base, bank); + if (ret) + return ret; + + ret = check_stat(ops->read_csr_ne_stat, state->ringnestat, "ringnestat", + base, bank); + if (ret) + return ret; + + ret = check_stat(ops->read_csr_nf_stat, state->ringnfstat, "ringnfstat", + base, bank); + if (ret) + return ret; + + ret = check_stat(ops->read_csr_f_stat, state->ringfstat, "ringfstat", + base, bank); + if (ret) + return ret; + + ret = check_stat(ops->read_csr_c_stat, state->ringcstat0, "ringcstat", + base, bank); + if (ret) + return ret; + + return 0; +} + +/** + * adf_bank_state_save() - save state of bank-related registers + * @accel_dev: Pointer to the device structure + * @bank_number: Bank number + * @state: Pointer to bank state structure + * + * This function saves the state of a bank by reading the bank CSRs and + * writing them in the @state structure. + * + * Returns 0 on success, error code otherwise + */ +int adf_bank_state_save(struct adf_accel_dev *accel_dev, u32 bank_number, + struct adf_bank_state *state) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(accel_dev); + void __iomem *csr_base = adf_get_etr_base(accel_dev); + + if (bank_number >= hw_data->num_banks || !state) + return -EINVAL; + + dev_dbg(&GET_DEV(accel_dev), "Saving state of bank %d\n", bank_number); + + bank_state_save(csr_ops, csr_base, bank_number, state, + hw_data->num_rings_per_bank); + + return 0; +} +EXPORT_SYMBOL_GPL(adf_bank_state_save); + +/** + * adf_bank_state_restore() - restore state of bank-related registers + * @accel_dev: Pointer to the device structure + * @bank_number: Bank number + * @state: Pointer to bank state structure + * + * This function attempts to restore the state of a bank by writing the + * bank CSRs to the values in the state structure. + * + * Returns 0 on success, error code otherwise + */ +int adf_bank_state_restore(struct adf_accel_dev *accel_dev, u32 bank_number, + struct adf_bank_state *state) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(accel_dev); + void __iomem *csr_base = adf_get_etr_base(accel_dev); + int ret; + + if (bank_number >= hw_data->num_banks || !state) + return -EINVAL; + + dev_dbg(&GET_DEV(accel_dev), "Restoring state of bank %d\n", bank_number); + + ret = bank_state_restore(csr_ops, csr_base, bank_number, state, + hw_data->num_rings_per_bank, hw_data->tx_rx_gap); + if (ret) + dev_err(&GET_DEV(accel_dev), + "Unable to restore state of bank %d\n", bank_number); + + return ret; +} +EXPORT_SYMBOL_GPL(adf_bank_state_restore); diff --git a/drivers/crypto/intel/qat/qat_common/adf_bank_state.h b/drivers/crypto/intel/qat/qat_common/adf_bank_state.h new file mode 100644 index 000000000000..48b573d692dd --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_bank_state.h @@ -0,0 +1,49 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ADF_BANK_STATE_H_ +#define ADF_BANK_STATE_H_ + +#include <linux/types.h> + +struct adf_accel_dev; + +struct ring_config { + u64 base; + u32 config; + u32 head; + u32 tail; + u32 reserved0; +}; + +struct adf_bank_state { + u32 ringstat0; + u32 ringstat1; + u32 ringuostat; + u32 ringestat; + u32 ringnestat; + u32 ringnfstat; + u32 ringfstat; + u32 ringcstat0; + u32 ringcstat1; + u32 ringcstat2; + u32 ringcstat3; + u32 iaintflagen; + u32 iaintflagreg; + u32 iaintflagsrcsel0; + u32 iaintflagsrcsel1; + u32 iaintcolen; + u32 iaintcolctl; + u32 iaintflagandcolen; + u32 ringexpstat; + u32 ringexpintenable; + u32 ringsrvarben; + u32 reserved0; + struct ring_config rings[ADF_ETR_MAX_RINGS_PER_BANK]; +}; + +int adf_bank_state_restore(struct adf_accel_dev *accel_dev, u32 bank_number, + struct adf_bank_state *state); +int adf_bank_state_save(struct adf_accel_dev *accel_dev, u32 bank_number, + struct adf_bank_state *state); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_cfg.c b/drivers/crypto/intel/qat/qat_common/adf_cfg.c new file mode 100644 index 000000000000..b0fc453fa3fb --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cfg.c @@ -0,0 +1,384 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/mutex.h> +#include <linux/slab.h> +#include <linux/list.h> +#include <linux/seq_file.h> +#include "adf_accel_devices.h" +#include "adf_cfg.h" +#include "adf_common_drv.h" + +static DEFINE_MUTEX(qat_cfg_read_lock); + +static void *qat_dev_cfg_start(struct seq_file *sfile, loff_t *pos) +{ + struct adf_cfg_device_data *dev_cfg = sfile->private; + + mutex_lock(&qat_cfg_read_lock); + return seq_list_start(&dev_cfg->sec_list, *pos); +} + +static int qat_dev_cfg_show(struct seq_file *sfile, void *v) +{ + struct list_head *list; + struct adf_cfg_section *sec = + list_entry(v, struct adf_cfg_section, list); + + seq_printf(sfile, "[%s]\n", sec->name); + list_for_each(list, &sec->param_head) { + struct adf_cfg_key_val *ptr = + list_entry(list, struct adf_cfg_key_val, list); + seq_printf(sfile, "%s = %s\n", ptr->key, ptr->val); + } + return 0; +} + +static void *qat_dev_cfg_next(struct seq_file *sfile, void *v, loff_t *pos) +{ + struct adf_cfg_device_data *dev_cfg = sfile->private; + + return seq_list_next(v, &dev_cfg->sec_list, pos); +} + +static void qat_dev_cfg_stop(struct seq_file *sfile, void *v) +{ + mutex_unlock(&qat_cfg_read_lock); +} + +static const struct seq_operations qat_dev_cfg_sops = { + .start = qat_dev_cfg_start, + .next = qat_dev_cfg_next, + .stop = qat_dev_cfg_stop, + .show = qat_dev_cfg_show +}; + +DEFINE_SEQ_ATTRIBUTE(qat_dev_cfg); + +/** + * adf_cfg_dev_add() - Create an acceleration device configuration table. + * @accel_dev: Pointer to acceleration device. + * + * Function creates a configuration table for the given acceleration device. + * The table stores device specific config values. + * To be used by QAT device specific drivers. + * + * Return: 0 on success, error code otherwise. + */ +int adf_cfg_dev_add(struct adf_accel_dev *accel_dev) +{ + struct adf_cfg_device_data *dev_cfg_data; + + dev_cfg_data = kzalloc(sizeof(*dev_cfg_data), GFP_KERNEL); + if (!dev_cfg_data) + return -ENOMEM; + INIT_LIST_HEAD(&dev_cfg_data->sec_list); + init_rwsem(&dev_cfg_data->lock); + accel_dev->cfg = dev_cfg_data; + return 0; +} +EXPORT_SYMBOL_GPL(adf_cfg_dev_add); + +void adf_cfg_dev_dbgfs_add(struct adf_accel_dev *accel_dev) +{ + struct adf_cfg_device_data *dev_cfg_data = accel_dev->cfg; + + dev_cfg_data->debug = debugfs_create_file("dev_cfg", 0400, + accel_dev->debugfs_dir, + dev_cfg_data, + &qat_dev_cfg_fops); +} + +void adf_cfg_dev_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ + struct adf_cfg_device_data *dev_cfg_data = accel_dev->cfg; + + if (!dev_cfg_data) + return; + + debugfs_remove(dev_cfg_data->debug); + dev_cfg_data->debug = NULL; +} + +static void adf_cfg_section_del_all(struct list_head *head); +static void adf_cfg_section_del_all_except(struct list_head *head, + const char *section_name); + +void adf_cfg_del_all(struct adf_accel_dev *accel_dev) +{ + struct adf_cfg_device_data *dev_cfg_data = accel_dev->cfg; + + down_write(&dev_cfg_data->lock); + adf_cfg_section_del_all(&dev_cfg_data->sec_list); + up_write(&dev_cfg_data->lock); + clear_bit(ADF_STATUS_CONFIGURED, &accel_dev->status); +} + +void adf_cfg_del_all_except(struct adf_accel_dev *accel_dev, + const char *section_name) +{ + struct adf_cfg_device_data *dev_cfg_data = accel_dev->cfg; + + down_write(&dev_cfg_data->lock); + adf_cfg_section_del_all_except(&dev_cfg_data->sec_list, section_name); + up_write(&dev_cfg_data->lock); + clear_bit(ADF_STATUS_CONFIGURED, &accel_dev->status); +} + +/** + * adf_cfg_dev_remove() - Clears acceleration device configuration table. + * @accel_dev: Pointer to acceleration device. + * + * Function removes configuration table from the given acceleration device + * and frees all allocated memory. + * To be used by QAT device specific drivers. + * + * Return: void + */ +void adf_cfg_dev_remove(struct adf_accel_dev *accel_dev) +{ + struct adf_cfg_device_data *dev_cfg_data = accel_dev->cfg; + + if (!dev_cfg_data) + return; + + down_write(&dev_cfg_data->lock); + adf_cfg_section_del_all(&dev_cfg_data->sec_list); + up_write(&dev_cfg_data->lock); + kfree(dev_cfg_data); + accel_dev->cfg = NULL; +} +EXPORT_SYMBOL_GPL(adf_cfg_dev_remove); + +static void adf_cfg_keyval_add(struct adf_cfg_key_val *new, + struct adf_cfg_section *sec) +{ + list_add_tail(&new->list, &sec->param_head); +} + +static void adf_cfg_keyval_remove(const char *key, struct adf_cfg_section *sec) +{ + struct list_head *head = &sec->param_head; + struct list_head *list_ptr, *tmp; + + list_for_each_prev_safe(list_ptr, tmp, head) { + struct adf_cfg_key_val *ptr = + list_entry(list_ptr, struct adf_cfg_key_val, list); + + if (strncmp(ptr->key, key, sizeof(ptr->key))) + continue; + + list_del(list_ptr); + kfree(ptr); + break; + } +} + +static void adf_cfg_keyval_del_all(struct list_head *head) +{ + struct list_head *list_ptr, *tmp; + + list_for_each_prev_safe(list_ptr, tmp, head) { + struct adf_cfg_key_val *ptr = + list_entry(list_ptr, struct adf_cfg_key_val, list); + list_del(list_ptr); + kfree(ptr); + } +} + +static void adf_cfg_section_del_all(struct list_head *head) +{ + struct adf_cfg_section *ptr; + struct list_head *list, *tmp; + + list_for_each_prev_safe(list, tmp, head) { + ptr = list_entry(list, struct adf_cfg_section, list); + adf_cfg_keyval_del_all(&ptr->param_head); + list_del(list); + kfree(ptr); + } +} + +static void adf_cfg_section_del_all_except(struct list_head *head, + const char *section_name) +{ + struct list_head *list, *tmp; + struct adf_cfg_section *ptr; + + list_for_each_prev_safe(list, tmp, head) { + ptr = list_entry(list, struct adf_cfg_section, list); + if (!strcmp(ptr->name, section_name)) + continue; + adf_cfg_keyval_del_all(&ptr->param_head); + list_del(list); + kfree(ptr); + } +} + +static struct adf_cfg_key_val *adf_cfg_key_value_find(struct adf_cfg_section *s, + const char *key) +{ + struct list_head *list; + + list_for_each(list, &s->param_head) { + struct adf_cfg_key_val *ptr = + list_entry(list, struct adf_cfg_key_val, list); + if (!strcmp(ptr->key, key)) + return ptr; + } + return NULL; +} + +static struct adf_cfg_section *adf_cfg_sec_find(struct adf_accel_dev *accel_dev, + const char *sec_name) +{ + struct adf_cfg_device_data *cfg = accel_dev->cfg; + struct list_head *list; + + list_for_each(list, &cfg->sec_list) { + struct adf_cfg_section *ptr = + list_entry(list, struct adf_cfg_section, list); + if (!strcmp(ptr->name, sec_name)) + return ptr; + } + return NULL; +} + +static int adf_cfg_key_val_get(struct adf_accel_dev *accel_dev, + const char *sec_name, + const char *key_name, + char *val) +{ + struct adf_cfg_section *sec = adf_cfg_sec_find(accel_dev, sec_name); + struct adf_cfg_key_val *keyval = NULL; + + if (sec) + keyval = adf_cfg_key_value_find(sec, key_name); + if (keyval) { + memcpy(val, keyval->val, ADF_CFG_MAX_VAL_LEN_IN_BYTES); + return 0; + } + return -ENODATA; +} + +/** + * adf_cfg_add_key_value_param() - Add key-value config entry to config table. + * @accel_dev: Pointer to acceleration device. + * @section_name: Name of the section where the param will be added + * @key: The key string + * @val: Value pain for the given @key + * @type: Type - string, int or address + * + * Function adds configuration key - value entry in the appropriate section + * in the given acceleration device. If the key exists already, the value + * is updated. + * To be used by QAT device specific drivers. + * + * Return: 0 on success, error code otherwise. + */ +int adf_cfg_add_key_value_param(struct adf_accel_dev *accel_dev, + const char *section_name, + const char *key, const void *val, + enum adf_cfg_val_type type) +{ + struct adf_cfg_device_data *cfg = accel_dev->cfg; + struct adf_cfg_key_val *key_val; + struct adf_cfg_section *section = adf_cfg_sec_find(accel_dev, + section_name); + char temp_val[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + + if (!section) + return -EFAULT; + + key_val = kzalloc(sizeof(*key_val), GFP_KERNEL); + if (!key_val) + return -ENOMEM; + + INIT_LIST_HEAD(&key_val->list); + strscpy(key_val->key, key, sizeof(key_val->key)); + + if (type == ADF_DEC) { + snprintf(key_val->val, ADF_CFG_MAX_VAL_LEN_IN_BYTES, + "%ld", (*((long *)val))); + } else if (type == ADF_STR) { + strscpy(key_val->val, (char *)val, sizeof(key_val->val)); + } else if (type == ADF_HEX) { + snprintf(key_val->val, ADF_CFG_MAX_VAL_LEN_IN_BYTES, + "0x%lx", (unsigned long)val); + } else { + dev_err(&GET_DEV(accel_dev), "Unknown type given.\n"); + kfree(key_val); + return -EINVAL; + } + key_val->type = type; + + /* Add the key-value pair as below policy: + * 1. if the key doesn't exist, add it; + * 2. if the key already exists with a different value then update it + * to the new value (the key is deleted and the newly created + * key_val containing the new value is added to the database); + * 3. if the key exists with the same value, then return without doing + * anything (the newly created key_val is freed). + */ + down_write(&cfg->lock); + if (!adf_cfg_key_val_get(accel_dev, section_name, key, temp_val)) { + if (strncmp(temp_val, key_val->val, sizeof(temp_val))) { + adf_cfg_keyval_remove(key, section); + } else { + kfree(key_val); + goto out; + } + } + + adf_cfg_keyval_add(key_val, section); + +out: + up_write(&cfg->lock); + return 0; +} +EXPORT_SYMBOL_GPL(adf_cfg_add_key_value_param); + +/** + * adf_cfg_section_add() - Add config section entry to config table. + * @accel_dev: Pointer to acceleration device. + * @name: Name of the section + * + * Function adds configuration section where key - value entries + * will be stored. + * To be used by QAT device specific drivers. + * + * Return: 0 on success, error code otherwise. + */ +int adf_cfg_section_add(struct adf_accel_dev *accel_dev, const char *name) +{ + struct adf_cfg_device_data *cfg = accel_dev->cfg; + struct adf_cfg_section *sec = adf_cfg_sec_find(accel_dev, name); + + if (sec) + return 0; + + sec = kzalloc(sizeof(*sec), GFP_KERNEL); + if (!sec) + return -ENOMEM; + + strscpy(sec->name, name, sizeof(sec->name)); + INIT_LIST_HEAD(&sec->param_head); + down_write(&cfg->lock); + list_add_tail(&sec->list, &cfg->sec_list); + up_write(&cfg->lock); + return 0; +} +EXPORT_SYMBOL_GPL(adf_cfg_section_add); + +int adf_cfg_get_param_value(struct adf_accel_dev *accel_dev, + const char *section, const char *name, + char *value) +{ + struct adf_cfg_device_data *cfg = accel_dev->cfg; + int ret; + + down_read(&cfg->lock); + ret = adf_cfg_key_val_get(accel_dev, section, name, value); + up_read(&cfg->lock); + return ret; +} +EXPORT_SYMBOL_GPL(adf_cfg_get_param_value); diff --git a/drivers/crypto/intel/qat/qat_common/adf_cfg.h b/drivers/crypto/intel/qat/qat_common/adf_cfg.h new file mode 100644 index 000000000000..2afa6f0d15c5 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cfg.h @@ -0,0 +1,47 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_CFG_H_ +#define ADF_CFG_H_ + +#include <linux/list.h> +#include <linux/rwsem.h> +#include <linux/debugfs.h> +#include "adf_accel_devices.h" +#include "adf_cfg_common.h" +#include "adf_cfg_strings.h" + +struct adf_cfg_key_val { + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + char val[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + enum adf_cfg_val_type type; + struct list_head list; +}; + +struct adf_cfg_section { + char name[ADF_CFG_MAX_SECTION_LEN_IN_BYTES]; + struct list_head list; + struct list_head param_head; +}; + +struct adf_cfg_device_data { + struct list_head sec_list; + struct dentry *debug; + struct rw_semaphore lock; +}; + +int adf_cfg_dev_add(struct adf_accel_dev *accel_dev); +void adf_cfg_dev_remove(struct adf_accel_dev *accel_dev); +void adf_cfg_dev_dbgfs_add(struct adf_accel_dev *accel_dev); +void adf_cfg_dev_dbgfs_rm(struct adf_accel_dev *accel_dev); +int adf_cfg_section_add(struct adf_accel_dev *accel_dev, const char *name); +void adf_cfg_del_all(struct adf_accel_dev *accel_dev); +void adf_cfg_del_all_except(struct adf_accel_dev *accel_dev, + const char *section_name); +int adf_cfg_add_key_value_param(struct adf_accel_dev *accel_dev, + const char *section_name, + const char *key, const void *val, + enum adf_cfg_val_type type); +int adf_cfg_get_param_value(struct adf_accel_dev *accel_dev, + const char *section, const char *name, char *value); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_cfg_common.h b/drivers/crypto/intel/qat/qat_common/adf_cfg_common.h new file mode 100644 index 000000000000..81e9e9d7eccd --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cfg_common.h @@ -0,0 +1,79 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_CFG_COMMON_H_ +#define ADF_CFG_COMMON_H_ + +#include <linux/types.h> +#include <linux/ioctl.h> + +#define ADF_CFG_MAX_STR_LEN 64 +#define ADF_CFG_MAX_KEY_LEN_IN_BYTES ADF_CFG_MAX_STR_LEN +#define ADF_CFG_MAX_VAL_LEN_IN_BYTES ADF_CFG_MAX_STR_LEN +#define ADF_CFG_MAX_SECTION_LEN_IN_BYTES ADF_CFG_MAX_STR_LEN +#define ADF_CFG_BASE_DEC 10 +#define ADF_CFG_BASE_HEX 16 +#define ADF_CFG_ALL_DEVICES 0xFE +#define ADF_CFG_NO_DEVICE 0xFF +#define ADF_CFG_AFFINITY_WHATEVER 0xFF +#define MAX_DEVICE_NAME_SIZE 32 +#define ADF_MAX_DEVICES (32 * 32) +#define ADF_DEVS_ARRAY_SIZE BITS_TO_LONGS(ADF_MAX_DEVICES) + +#define ADF_CFG_SERV_RING_PAIR_0_SHIFT 0 +#define ADF_CFG_SERV_RING_PAIR_1_SHIFT 3 +#define ADF_CFG_SERV_RING_PAIR_2_SHIFT 6 +#define ADF_CFG_SERV_RING_PAIR_3_SHIFT 9 +enum adf_cfg_service_type { + UNUSED = 0, + CRYPTO, + COMP, + SYM, + ASYM, + DECOMP, + USED +}; + +enum adf_cfg_val_type { + ADF_DEC, + ADF_HEX, + ADF_STR +}; + +enum adf_device_type { + DEV_UNKNOWN = 0, + DEV_DH895XCC, + DEV_DH895XCCVF, + DEV_C62X, + DEV_C62XVF, + DEV_C3XXX, + DEV_C3XXXVF, + DEV_4XXX, + DEV_420XX, + DEV_6XXX, +}; + +struct adf_dev_status_info { + enum adf_device_type type; + __u32 accel_id; + __u32 instance_id; + __u8 num_ae; + __u8 num_accel; + __u8 num_logical_accel; + __u8 banks_per_accel; + __u8 state; + __u8 bus; + __u8 dev; + __u8 fun; + char name[MAX_DEVICE_NAME_SIZE]; +}; + +#define ADF_CTL_IOC_MAGIC 'a' +#define IOCTL_CONFIG_SYS_RESOURCE_PARAMETERS _IOW(ADF_CTL_IOC_MAGIC, 0, \ + struct adf_user_cfg_ctl_data) +#define IOCTL_STOP_ACCEL_DEV _IOW(ADF_CTL_IOC_MAGIC, 1, \ + struct adf_user_cfg_ctl_data) +#define IOCTL_START_ACCEL_DEV _IOW(ADF_CTL_IOC_MAGIC, 2, \ + struct adf_user_cfg_ctl_data) +#define IOCTL_STATUS_ACCEL_DEV _IOW(ADF_CTL_IOC_MAGIC, 3, __u32) +#define IOCTL_GET_NUM_DEVICES _IOW(ADF_CTL_IOC_MAGIC, 4, __s32) +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_cfg_services.c b/drivers/crypto/intel/qat/qat_common/adf_cfg_services.c new file mode 100644 index 000000000000..7d00bcb41ce7 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cfg_services.c @@ -0,0 +1,212 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/array_size.h> +#include <linux/bitops.h> +#include <linux/export.h> +#include <linux/pci.h> +#include <linux/string.h> +#include "adf_cfg.h" +#include "adf_cfg_common.h" +#include "adf_cfg_services.h" +#include "adf_cfg_strings.h" + +static const char *const adf_cfg_services[] = { + [SVC_ASYM] = ADF_CFG_ASYM, + [SVC_SYM] = ADF_CFG_SYM, + [SVC_DC] = ADF_CFG_DC, + [SVC_DCC] = ADF_CFG_DCC, + [SVC_DECOMP] = ADF_CFG_DECOMP, +}; + +/* + * Ensure that the size of the array matches the number of services, + * SVC_COUNT, that is used to size the bitmap. + */ +static_assert(ARRAY_SIZE(adf_cfg_services) == SVC_COUNT); + +/* + * Ensure that the maximum number of concurrent services that can be + * enabled on a device is less than or equal to the number of total + * supported services. + */ +static_assert(ARRAY_SIZE(adf_cfg_services) >= MAX_NUM_CONCURR_SVC); + +/* + * Ensure that the number of services fit a single unsigned long, as each + * service is represented by a bit in the mask. + */ +static_assert(BITS_PER_LONG >= SVC_COUNT); + +/* + * Ensure that size of the concatenation of all service strings is smaller + * than the size of the buffer that will contain them. + */ +static_assert(sizeof(ADF_CFG_SYM ADF_SERVICES_DELIMITER + ADF_CFG_ASYM ADF_SERVICES_DELIMITER + ADF_CFG_DC ADF_SERVICES_DELIMITER + ADF_CFG_DECOMP ADF_SERVICES_DELIMITER + ADF_CFG_DCC) < ADF_CFG_MAX_VAL_LEN_IN_BYTES); + +static int adf_service_string_to_mask(struct adf_accel_dev *accel_dev, const char *buf, + size_t len, unsigned long *out_mask) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + char services[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = { }; + unsigned long mask = 0; + char *substr, *token; + int id, num_svc = 0; + + if (len > ADF_CFG_MAX_VAL_LEN_IN_BYTES - 1) + return -EINVAL; + + strscpy(services, buf, ADF_CFG_MAX_VAL_LEN_IN_BYTES); + substr = services; + + while ((token = strsep(&substr, ADF_SERVICES_DELIMITER))) { + id = sysfs_match_string(adf_cfg_services, token); + if (id < 0) + return id; + + if (test_and_set_bit(id, &mask)) + return -EINVAL; + + if (num_svc++ == MAX_NUM_CONCURR_SVC) + return -EINVAL; + } + + if (hw_data->services_supported && !hw_data->services_supported(mask)) + return -EINVAL; + + *out_mask = mask; + + return 0; +} + +static int adf_service_mask_to_string(unsigned long mask, char *buf, size_t len) +{ + int offset = 0; + int bit; + + if (len < ADF_CFG_MAX_VAL_LEN_IN_BYTES) + return -ENOSPC; + + for_each_set_bit(bit, &mask, SVC_COUNT) { + if (offset) + offset += scnprintf(buf + offset, len - offset, + ADF_SERVICES_DELIMITER); + + offset += scnprintf(buf + offset, len - offset, "%s", + adf_cfg_services[bit]); + } + + return 0; +} + +int adf_parse_service_string(struct adf_accel_dev *accel_dev, const char *in, + size_t in_len, char *out, size_t out_len) +{ + unsigned long mask; + int ret; + + ret = adf_service_string_to_mask(accel_dev, in, in_len, &mask); + if (ret) + return ret; + + if (!mask) + return -EINVAL; + + return adf_service_mask_to_string(mask, out, out_len); +} + +int adf_get_service_mask(struct adf_accel_dev *accel_dev, unsigned long *mask) +{ + char services[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = { }; + size_t len; + int ret; + + ret = adf_cfg_get_param_value(accel_dev, ADF_GENERAL_SEC, + ADF_SERVICES_ENABLED, services); + if (ret) { + dev_err(&GET_DEV(accel_dev), "%s param not found\n", + ADF_SERVICES_ENABLED); + return ret; + } + + len = strnlen(services, ADF_CFG_MAX_VAL_LEN_IN_BYTES); + ret = adf_service_string_to_mask(accel_dev, services, len, mask); + if (ret) + dev_err(&GET_DEV(accel_dev), "Invalid value of %s param: %s\n", + ADF_SERVICES_ENABLED, services); + + return ret; +} +EXPORT_SYMBOL_GPL(adf_get_service_mask); + +int adf_get_service_enabled(struct adf_accel_dev *accel_dev) +{ + unsigned long mask; + int ret; + + ret = adf_get_service_mask(accel_dev, &mask); + if (ret) + return ret; + + if (test_bit(SVC_SYM, &mask) && test_bit(SVC_ASYM, &mask)) + return SVC_SYM_ASYM; + + if (test_bit(SVC_SYM, &mask) && test_bit(SVC_DC, &mask)) + return SVC_SYM_DC; + + if (test_bit(SVC_ASYM, &mask) && test_bit(SVC_DC, &mask)) + return SVC_ASYM_DC; + + if (test_bit(SVC_SYM, &mask)) + return SVC_SYM; + + if (test_bit(SVC_ASYM, &mask)) + return SVC_ASYM; + + if (test_bit(SVC_DC, &mask)) + return SVC_DC; + + if (test_bit(SVC_DECOMP, &mask)) + return SVC_DECOMP; + + if (test_bit(SVC_DCC, &mask)) + return SVC_DCC; + + return -EINVAL; +} +EXPORT_SYMBOL_GPL(adf_get_service_enabled); + +enum adf_cfg_service_type adf_srv_to_cfg_svc_type(enum adf_base_services svc) +{ + switch (svc) { + case SVC_ASYM: + return ASYM; + case SVC_SYM: + return SYM; + case SVC_DC: + return COMP; + case SVC_DECOMP: + return DECOMP; + default: + return UNUSED; + } +} + +bool adf_is_service_enabled(struct adf_accel_dev *accel_dev, enum adf_base_services svc) +{ + enum adf_cfg_service_type arb_srv = adf_srv_to_cfg_svc_type(svc); + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + u8 rps_per_bundle = hw_data->num_banks_per_vf; + int i; + + for (i = 0; i < rps_per_bundle; i++) { + if (GET_SRV_TYPE(accel_dev, i) == arb_srv) + return true; + } + + return false; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_cfg_services.h b/drivers/crypto/intel/qat/qat_common/adf_cfg_services.h new file mode 100644 index 000000000000..913d717280af --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cfg_services.h @@ -0,0 +1,44 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef _ADF_CFG_SERVICES_H_ +#define _ADF_CFG_SERVICES_H_ + +#include "adf_cfg_strings.h" + +struct adf_accel_dev; + +enum adf_base_services { + SVC_ASYM = 0, + SVC_SYM, + SVC_DC, + SVC_DECOMP, + SVC_BASE_COUNT +}; + +enum adf_extended_services { + SVC_DCC = SVC_BASE_COUNT, + SVC_COUNT +}; + +enum adf_composed_services { + SVC_SYM_ASYM = SVC_COUNT, + SVC_SYM_DC, + SVC_ASYM_DC, +}; + +enum { + ADF_ONE_SERVICE = 1, + ADF_TWO_SERVICES, + ADF_THREE_SERVICES, +}; + +#define MAX_NUM_CONCURR_SVC ADF_THREE_SERVICES + +int adf_parse_service_string(struct adf_accel_dev *accel_dev, const char *in, + size_t in_len, char *out, size_t out_len); +int adf_get_service_enabled(struct adf_accel_dev *accel_dev); +int adf_get_service_mask(struct adf_accel_dev *accel_dev, unsigned long *mask); +enum adf_cfg_service_type adf_srv_to_cfg_svc_type(enum adf_base_services svc); +bool adf_is_service_enabled(struct adf_accel_dev *accel_dev, enum adf_base_services svc); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_cfg_strings.h b/drivers/crypto/intel/qat/qat_common/adf_cfg_strings.h new file mode 100644 index 000000000000..30107a02ee7f --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cfg_strings.h @@ -0,0 +1,51 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_CFG_STRINGS_H_ +#define ADF_CFG_STRINGS_H_ + +#define ADF_GENERAL_SEC "GENERAL" +#define ADF_KERNEL_SEC "KERNEL" +#define ADF_ACCEL_SEC "Accelerator" +#define ADF_NUM_CY "NumberCyInstances" +#define ADF_NUM_DC "NumberDcInstances" +#define ADF_RING_SYM_SIZE "NumConcurrentSymRequests" +#define ADF_RING_ASYM_SIZE "NumConcurrentAsymRequests" +#define ADF_RING_DC_SIZE "NumConcurrentRequests" +#define ADF_RING_ASYM_TX "RingAsymTx" +#define ADF_RING_SYM_TX "RingSymTx" +#define ADF_RING_ASYM_RX "RingAsymRx" +#define ADF_RING_SYM_RX "RingSymRx" +#define ADF_RING_DC_TX "RingTx" +#define ADF_RING_DC_RX "RingRx" +#define ADF_ETRMGR_BANK "Bank" +#define ADF_RING_SYM_BANK_NUM "BankSymNumber" +#define ADF_RING_ASYM_BANK_NUM "BankAsymNumber" +#define ADF_RING_DC_BANK_NUM "BankDcNumber" +#define ADF_CY "Cy" +#define ADF_DC "Dc" +#define ADF_CFG_DC "dc" +#define ADF_CFG_DECOMP "decomp" +#define ADF_CFG_CY "sym;asym" +#define ADF_CFG_SYM "sym" +#define ADF_CFG_ASYM "asym" +#define ADF_CFG_DCC "dcc" +#define ADF_SERVICES_ENABLED "ServicesEnabled" +#define ADF_SERVICES_DELIMITER ";" +#define ADF_PM_IDLE_SUPPORT "PmIdleSupport" +#define ADF_ETRMGR_COALESCING_ENABLED "InterruptCoalescingEnabled" +#define ADF_ETRMGR_COALESCING_ENABLED_FORMAT \ + ADF_ETRMGR_BANK "%d" ADF_ETRMGR_COALESCING_ENABLED +#define ADF_ETRMGR_COALESCE_TIMER "InterruptCoalescingTimerNs" +#define ADF_ETRMGR_COALESCE_TIMER_FORMAT \ + ADF_ETRMGR_BANK "%d" ADF_ETRMGR_COALESCE_TIMER +#define ADF_ETRMGR_COALESCING_MSG_ENABLED "InterruptCoalescingNumResponses" +#define ADF_ETRMGR_COALESCING_MSG_ENABLED_FORMAT \ + ADF_ETRMGR_BANK "%d" ADF_ETRMGR_COALESCING_MSG_ENABLED +#define ADF_ETRMGR_CORE_AFFINITY "CoreAffinity" +#define ADF_ETRMGR_CORE_AFFINITY_FORMAT \ + ADF_ETRMGR_BANK "%d" ADF_ETRMGR_CORE_AFFINITY +#define ADF_ACCEL_STR "Accelerator%d" +#define ADF_HEARTBEAT_TIMER "HeartbeatTimer" +#define ADF_SRIOV_ENABLED "SriovEnabled" + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_cfg_user.h b/drivers/crypto/intel/qat/qat_common/adf_cfg_user.h new file mode 100644 index 000000000000..421f4fb8b4dd --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cfg_user.h @@ -0,0 +1,38 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_CFG_USER_H_ +#define ADF_CFG_USER_H_ + +#include "adf_cfg_common.h" +#include "adf_cfg_strings.h" + +struct adf_user_cfg_key_val { + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + char val[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + union { + struct adf_user_cfg_key_val *next; + __u64 padding3; + }; + enum adf_cfg_val_type type; +} __packed; + +struct adf_user_cfg_section { + char name[ADF_CFG_MAX_SECTION_LEN_IN_BYTES]; + union { + struct adf_user_cfg_key_val *params; + __u64 padding1; + }; + union { + struct adf_user_cfg_section *next; + __u64 padding3; + }; +} __packed; + +struct adf_user_cfg_ctl_data { + union { + struct adf_user_cfg_section *config_section; + __u64 padding; + }; + __u8 device_id; +} __packed; +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_clock.c b/drivers/crypto/intel/qat/qat_common/adf_clock.c new file mode 100644 index 000000000000..cf89f57de2a7 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_clock.c @@ -0,0 +1,135 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/delay.h> +#include <linux/dev_printk.h> +#include <linux/export.h> +#include <linux/math.h> +#include <linux/minmax.h> +#include <linux/time64.h> +#include <linux/types.h> +#include <linux/units.h> +#include <asm/errno.h> +#include "adf_admin.h" +#include "adf_accel_devices.h" +#include "adf_clock.h" +#include "adf_common_drv.h" + +#define MEASURE_CLOCK_RETRIES 10 +#define MEASURE_CLOCK_DELAY_US 10000 +#define ME_CLK_DIVIDER 16 +#define MEASURE_CLOCK_DELTA_THRESHOLD_US 100 + +static inline u64 timespec_to_us(const struct timespec64 *ts) +{ + return (u64)DIV_ROUND_CLOSEST_ULL(timespec64_to_ns(ts), NSEC_PER_USEC); +} + +static inline u64 timespec_to_ms(const struct timespec64 *ts) +{ + return (u64)DIV_ROUND_CLOSEST_ULL(timespec64_to_ns(ts), NSEC_PER_MSEC); +} + +u64 adf_clock_get_current_time(void) +{ + struct timespec64 ts; + + ktime_get_real_ts64(&ts); + return timespec_to_ms(&ts); +} + +static int measure_clock(struct adf_accel_dev *accel_dev, u32 *frequency) +{ + struct timespec64 ts1, ts2, ts3, ts4; + u64 timestamp1, timestamp2, temp; + u32 delta_us, tries; + int ret; + + tries = MEASURE_CLOCK_RETRIES; + do { + ktime_get_real_ts64(&ts1); + ret = adf_get_fw_timestamp(accel_dev, ×tamp1); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to get fw timestamp\n"); + return ret; + } + ktime_get_real_ts64(&ts2); + delta_us = timespec_to_us(&ts2) - timespec_to_us(&ts1); + } while (delta_us > MEASURE_CLOCK_DELTA_THRESHOLD_US && --tries); + + if (!tries) { + dev_err(&GET_DEV(accel_dev), "Excessive clock measure delay\n"); + return -ETIMEDOUT; + } + + fsleep(MEASURE_CLOCK_DELAY_US); + + tries = MEASURE_CLOCK_RETRIES; + do { + ktime_get_real_ts64(&ts3); + if (adf_get_fw_timestamp(accel_dev, ×tamp2)) { + dev_err(&GET_DEV(accel_dev), + "Failed to get fw timestamp\n"); + return -EIO; + } + ktime_get_real_ts64(&ts4); + delta_us = timespec_to_us(&ts4) - timespec_to_us(&ts3); + } while (delta_us > MEASURE_CLOCK_DELTA_THRESHOLD_US && --tries); + + if (!tries) { + dev_err(&GET_DEV(accel_dev), "Excessive clock measure delay\n"); + return -ETIMEDOUT; + } + + delta_us = timespec_to_us(&ts3) - timespec_to_us(&ts1); + if (!delta_us) + return -EINVAL; + + temp = (timestamp2 - timestamp1) * ME_CLK_DIVIDER * 10; + temp = DIV_ROUND_CLOSEST_ULL(temp, delta_us); + /* + * Enclose the division to allow the preprocessor to precalculate it, + * and avoid promoting r-value to 64-bit before division. + */ + *frequency = temp * (HZ_PER_MHZ / 10); + + return 0; +} + +/** + * adf_dev_measure_clock() - measures device clock frequency + * @accel_dev: Pointer to acceleration device. + * @frequency: Pointer to variable where result will be stored + * @min: Minimal allowed frequency value + * @max: Maximal allowed frequency value + * + * If the measurement result will go beyond the min/max thresholds the value + * will take the value of the crossed threshold. + * + * This algorithm compares the device firmware timestamp with the kernel + * timestamp. So we can't expect too high accuracy from this measurement. + * + * Return: + * * 0 - measurement succeed + * * -ETIMEDOUT - measurement failed + */ +int adf_dev_measure_clock(struct adf_accel_dev *accel_dev, + u32 *frequency, u32 min, u32 max) +{ + int ret; + u32 freq; + + ret = measure_clock(accel_dev, &freq); + if (ret) + return ret; + + *frequency = clamp(freq, min, max); + + if (*frequency != freq) + dev_warn(&GET_DEV(accel_dev), + "Measured clock %d Hz is out of range, assuming %d\n", + freq, *frequency); + return 0; +} +EXPORT_SYMBOL_GPL(adf_dev_measure_clock); diff --git a/drivers/crypto/intel/qat/qat_common/adf_clock.h b/drivers/crypto/intel/qat/qat_common/adf_clock.h new file mode 100644 index 000000000000..e309bc0dc35c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_clock.h @@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_CLOCK_H +#define ADF_CLOCK_H + +#include <linux/types.h> + +struct adf_accel_dev; + +int adf_dev_measure_clock(struct adf_accel_dev *accel_dev, u32 *frequency, + u32 min, u32 max); +u64 adf_clock_get_current_time(void); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_cnv_dbgfs.c b/drivers/crypto/intel/qat/qat_common/adf_cnv_dbgfs.c new file mode 100644 index 000000000000..627953a72d47 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cnv_dbgfs.c @@ -0,0 +1,299 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/bitfield.h> +#include <linux/debugfs.h> +#include <linux/kernel.h> + +#include "adf_accel_devices.h" +#include "adf_admin.h" +#include "adf_common_drv.h" +#include "adf_cnv_dbgfs.h" +#include "qat_compression.h" + +#define CNV_DEBUGFS_FILENAME "cnv_errors" +#define CNV_MIN_PADDING 16 + +#define CNV_ERR_INFO_MASK GENMASK(11, 0) +#define CNV_ERR_TYPE_MASK GENMASK(15, 12) +#define CNV_SLICE_ERR_SIGN_BIT_INDEX 7 +#define CNV_DELTA_ERR_SIGN_BIT_INDEX 11 + +enum cnv_error_type { + CNV_ERR_TYPE_NONE, + CNV_ERR_TYPE_CHECKSUM, + CNV_ERR_TYPE_DECOMP_PRODUCED_LENGTH, + CNV_ERR_TYPE_DECOMPRESSION, + CNV_ERR_TYPE_TRANSLATION, + CNV_ERR_TYPE_DECOMP_CONSUMED_LENGTH, + CNV_ERR_TYPE_UNKNOWN, + CNV_ERR_TYPES_COUNT +}; + +#define CNV_ERROR_TYPE_GET(latest_err) \ + min_t(u16, u16_get_bits(latest_err, CNV_ERR_TYPE_MASK), CNV_ERR_TYPE_UNKNOWN) + +#define CNV_GET_DELTA_ERR_INFO(latest_error) \ + sign_extend32(latest_error, CNV_DELTA_ERR_SIGN_BIT_INDEX) + +#define CNV_GET_SLICE_ERR_INFO(latest_error) \ + sign_extend32(latest_error, CNV_SLICE_ERR_SIGN_BIT_INDEX) + +#define CNV_GET_DEFAULT_ERR_INFO(latest_error) \ + u16_get_bits(latest_error, CNV_ERR_INFO_MASK) + +enum cnv_fields { + CNV_ERR_COUNT, + CNV_LATEST_ERR, + CNV_FIELDS_COUNT +}; + +static const char * const cnv_field_names[CNV_FIELDS_COUNT] = { + [CNV_ERR_COUNT] = "Total Errors", + [CNV_LATEST_ERR] = "Last Error", +}; + +static const char * const cnv_error_names[CNV_ERR_TYPES_COUNT] = { + [CNV_ERR_TYPE_NONE] = "No Error", + [CNV_ERR_TYPE_CHECKSUM] = "Checksum Error", + [CNV_ERR_TYPE_DECOMP_PRODUCED_LENGTH] = "Length Error-P", + [CNV_ERR_TYPE_DECOMPRESSION] = "Decomp Error", + [CNV_ERR_TYPE_TRANSLATION] = "Xlat Error", + [CNV_ERR_TYPE_DECOMP_CONSUMED_LENGTH] = "Length Error-C", + [CNV_ERR_TYPE_UNKNOWN] = "Unknown Error", +}; + +struct ae_cnv_errors { + u16 ae; + u16 err_cnt; + u16 latest_err; + bool is_comp_ae; +}; + +struct cnv_err_stats { + u16 ae_count; + struct ae_cnv_errors ae_cnv_errors[]; +}; + +static s16 get_err_info(u8 error_type, u16 latest) +{ + switch (error_type) { + case CNV_ERR_TYPE_DECOMP_PRODUCED_LENGTH: + case CNV_ERR_TYPE_DECOMP_CONSUMED_LENGTH: + return CNV_GET_DELTA_ERR_INFO(latest); + case CNV_ERR_TYPE_DECOMPRESSION: + case CNV_ERR_TYPE_TRANSLATION: + return CNV_GET_SLICE_ERR_INFO(latest); + default: + return CNV_GET_DEFAULT_ERR_INFO(latest); + } +} + +static void *qat_cnv_errors_seq_start(struct seq_file *sfile, loff_t *pos) +{ + struct cnv_err_stats *err_stats = sfile->private; + + if (*pos == 0) + return SEQ_START_TOKEN; + + if (*pos > err_stats->ae_count) + return NULL; + + return &err_stats->ae_cnv_errors[*pos - 1]; +} + +static void *qat_cnv_errors_seq_next(struct seq_file *sfile, void *v, + loff_t *pos) +{ + struct cnv_err_stats *err_stats = sfile->private; + + (*pos)++; + + if (*pos > err_stats->ae_count) + return NULL; + + return &err_stats->ae_cnv_errors[*pos - 1]; +} + +static void qat_cnv_errors_seq_stop(struct seq_file *sfile, void *v) +{ +} + +static int qat_cnv_errors_seq_show(struct seq_file *sfile, void *v) +{ + struct ae_cnv_errors *ae_errors; + unsigned int i; + s16 err_info; + u8 err_type; + + if (v == SEQ_START_TOKEN) { + seq_puts(sfile, "AE "); + for (i = 0; i < CNV_FIELDS_COUNT; ++i) + seq_printf(sfile, " %*s", CNV_MIN_PADDING, + cnv_field_names[i]); + } else { + ae_errors = v; + + if (!ae_errors->is_comp_ae) + return 0; + + err_type = CNV_ERROR_TYPE_GET(ae_errors->latest_err); + err_info = get_err_info(err_type, ae_errors->latest_err); + + seq_printf(sfile, "%d:", ae_errors->ae); + seq_printf(sfile, " %*d", CNV_MIN_PADDING, ae_errors->err_cnt); + seq_printf(sfile, "%*s [%d]", CNV_MIN_PADDING, + cnv_error_names[err_type], err_info); + } + seq_putc(sfile, '\n'); + + return 0; +} + +static const struct seq_operations qat_cnv_errors_sops = { + .start = qat_cnv_errors_seq_start, + .next = qat_cnv_errors_seq_next, + .stop = qat_cnv_errors_seq_stop, + .show = qat_cnv_errors_seq_show, +}; + +/** + * cnv_err_stats_alloc() - Get CNV stats for the provided device. + * @accel_dev: Pointer to a QAT acceleration device + * + * Allocates and populates table of CNV errors statistics for each non-admin AE + * available through the supplied acceleration device. The caller becomes the + * owner of such memory and is responsible for the deallocation through a call + * to kfree(). + * + * Returns: a pointer to a dynamically allocated struct cnv_err_stats on success + * or a negative value on error. + */ +static struct cnv_err_stats *cnv_err_stats_alloc(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct cnv_err_stats *err_stats; + unsigned long ae_count; + unsigned long ae_mask; + size_t err_stats_size; + unsigned long ae; + unsigned int i; + u16 latest_err; + u16 err_cnt; + int ret; + + if (!adf_dev_started(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "QAT Device not started\n"); + return ERR_PTR(-EBUSY); + } + + /* Ignore the admin AEs */ + ae_mask = hw_data->ae_mask & ~hw_data->admin_ae_mask; + ae_count = hweight_long(ae_mask); + if (unlikely(!ae_count)) + return ERR_PTR(-EINVAL); + + err_stats_size = struct_size(err_stats, ae_cnv_errors, ae_count); + err_stats = kmalloc(err_stats_size, GFP_KERNEL); + if (!err_stats) + return ERR_PTR(-ENOMEM); + + err_stats->ae_count = ae_count; + + i = 0; + for_each_set_bit(ae, &ae_mask, GET_MAX_ACCELENGINES(accel_dev)) { + ret = adf_get_cnv_stats(accel_dev, ae, &err_cnt, &latest_err); + if (ret) { + dev_dbg(&GET_DEV(accel_dev), + "Failed to get CNV stats for ae %ld, [%d].\n", + ae, ret); + err_stats->ae_cnv_errors[i++].is_comp_ae = false; + continue; + } + err_stats->ae_cnv_errors[i].is_comp_ae = true; + err_stats->ae_cnv_errors[i].latest_err = latest_err; + err_stats->ae_cnv_errors[i].err_cnt = err_cnt; + err_stats->ae_cnv_errors[i].ae = ae; + i++; + } + + return err_stats; +} + +static int qat_cnv_errors_file_open(struct inode *inode, struct file *file) +{ + struct adf_accel_dev *accel_dev = inode->i_private; + struct seq_file *cnv_errors_seq_file; + struct cnv_err_stats *cnv_err_stats; + int ret; + + cnv_err_stats = cnv_err_stats_alloc(accel_dev); + if (IS_ERR(cnv_err_stats)) + return PTR_ERR(cnv_err_stats); + + ret = seq_open(file, &qat_cnv_errors_sops); + if (unlikely(ret)) { + kfree(cnv_err_stats); + return ret; + } + + cnv_errors_seq_file = file->private_data; + cnv_errors_seq_file->private = cnv_err_stats; + return ret; +} + +static int qat_cnv_errors_file_release(struct inode *inode, struct file *file) +{ + struct seq_file *cnv_errors_seq_file = file->private_data; + + kfree(cnv_errors_seq_file->private); + cnv_errors_seq_file->private = NULL; + + return seq_release(inode, file); +} + +static const struct file_operations qat_cnv_fops = { + .owner = THIS_MODULE, + .open = qat_cnv_errors_file_open, + .read = seq_read, + .llseek = seq_lseek, + .release = qat_cnv_errors_file_release, +}; + +static ssize_t no_comp_file_read(struct file *f, char __user *buf, size_t count, + loff_t *pos) +{ + char *file_msg = "No engine configured for comp\n"; + + return simple_read_from_buffer(buf, count, pos, file_msg, + strlen(file_msg)); +} + +static const struct file_operations qat_cnv_no_comp_fops = { + .owner = THIS_MODULE, + .read = no_comp_file_read, +}; + +void adf_cnv_dbgfs_add(struct adf_accel_dev *accel_dev) +{ + const struct file_operations *fops; + void *data; + + if (adf_hw_dev_has_compression(accel_dev)) { + fops = &qat_cnv_fops; + data = accel_dev; + } else { + fops = &qat_cnv_no_comp_fops; + data = NULL; + } + + accel_dev->cnv_dbgfile = debugfs_create_file(CNV_DEBUGFS_FILENAME, 0400, + accel_dev->debugfs_dir, + data, fops); +} + +void adf_cnv_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ + debugfs_remove(accel_dev->cnv_dbgfile); + accel_dev->cnv_dbgfile = NULL; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_cnv_dbgfs.h b/drivers/crypto/intel/qat/qat_common/adf_cnv_dbgfs.h new file mode 100644 index 000000000000..b02b0961c433 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_cnv_dbgfs.h @@ -0,0 +1,11 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_CNV_DBG_H +#define ADF_CNV_DBG_H + +struct adf_accel_dev; + +void adf_cnv_dbgfs_add(struct adf_accel_dev *accel_dev); +void adf_cnv_dbgfs_rm(struct adf_accel_dev *accel_dev); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_common_drv.h b/drivers/crypto/intel/qat/qat_common/adf_common_drv.h new file mode 100644 index 000000000000..6cf3a95489e8 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_common_drv.h @@ -0,0 +1,270 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2021 Intel Corporation */ +#ifndef ADF_DRV_H +#define ADF_DRV_H + +#include <linux/list.h> +#include <linux/pci.h> +#include "adf_accel_devices.h" +#include "icp_qat_fw_loader_handle.h" +#include "icp_qat_hal.h" + +#define ADF_MAJOR_VERSION 0 +#define ADF_MINOR_VERSION 6 +#define ADF_BUILD_VERSION 0 +#define ADF_DRV_VERSION __stringify(ADF_MAJOR_VERSION) "." \ + __stringify(ADF_MINOR_VERSION) "." \ + __stringify(ADF_BUILD_VERSION) + +#define ADF_STATUS_RESTARTING 0 +#define ADF_STATUS_STARTING 1 +#define ADF_STATUS_CONFIGURED 2 +#define ADF_STATUS_STARTED 3 +#define ADF_STATUS_AE_INITIALISED 4 +#define ADF_STATUS_AE_UCODE_LOADED 5 +#define ADF_STATUS_AE_STARTED 6 +#define ADF_STATUS_PF_RUNNING 7 +#define ADF_STATUS_IRQ_ALLOCATED 8 +#define ADF_STATUS_CRYPTO_ALGS_REGISTERED 9 +#define ADF_STATUS_COMP_ALGS_REGISTERED 10 + +enum adf_dev_reset_mode { + ADF_DEV_RESET_ASYNC = 0, + ADF_DEV_RESET_SYNC +}; + +enum adf_event { + ADF_EVENT_INIT = 0, + ADF_EVENT_START, + ADF_EVENT_STOP, + ADF_EVENT_SHUTDOWN, + ADF_EVENT_RESTARTING, + ADF_EVENT_RESTARTED, + ADF_EVENT_FATAL_ERROR, +}; + +struct service_hndl { + int (*event_hld)(struct adf_accel_dev *accel_dev, + enum adf_event event); + unsigned long init_status[ADF_DEVS_ARRAY_SIZE]; + unsigned long start_status[ADF_DEVS_ARRAY_SIZE]; + char *name; + struct list_head list; +}; + +int adf_service_register(struct service_hndl *service); +int adf_service_unregister(struct service_hndl *service); + +int adf_dev_up(struct adf_accel_dev *accel_dev, bool init_config); +int adf_dev_down(struct adf_accel_dev *accel_dev); +int adf_dev_restart(struct adf_accel_dev *accel_dev); + +void adf_devmgr_update_class_index(struct adf_hw_device_data *hw_data); +void adf_clean_vf_map(bool); +int adf_notify_fatal_error(struct adf_accel_dev *accel_dev); +void adf_error_notifier(struct adf_accel_dev *accel_dev); +int adf_devmgr_add_dev(struct adf_accel_dev *accel_dev, + struct adf_accel_dev *pf); +void adf_devmgr_rm_dev(struct adf_accel_dev *accel_dev, + struct adf_accel_dev *pf); +struct list_head *adf_devmgr_get_head(void); +struct adf_accel_dev *adf_devmgr_get_dev_by_id(u32 id); +struct adf_accel_dev *adf_devmgr_pci_to_accel_dev(struct pci_dev *pci_dev); +int adf_devmgr_verify_id(u32 id); +void adf_devmgr_get_num_dev(u32 *num); +int adf_devmgr_in_reset(struct adf_accel_dev *accel_dev); +int adf_dev_started(struct adf_accel_dev *accel_dev); +int adf_dev_restarting_notify(struct adf_accel_dev *accel_dev); +int adf_dev_restarted_notify(struct adf_accel_dev *accel_dev); +int adf_ae_init(struct adf_accel_dev *accel_dev); +int adf_ae_shutdown(struct adf_accel_dev *accel_dev); +int adf_ae_fw_load(struct adf_accel_dev *accel_dev); +void adf_ae_fw_release(struct adf_accel_dev *accel_dev); +int adf_ae_start(struct adf_accel_dev *accel_dev); +int adf_ae_stop(struct adf_accel_dev *accel_dev); + +extern const struct pci_error_handlers adf_err_handler; +void adf_reset_sbr(struct adf_accel_dev *accel_dev); +void adf_reset_flr(struct adf_accel_dev *accel_dev); +void adf_dev_restore(struct adf_accel_dev *accel_dev); +int adf_init_aer(void); +void adf_exit_aer(void); +int adf_init_arb(struct adf_accel_dev *accel_dev); +void adf_exit_arb(struct adf_accel_dev *accel_dev); +void adf_update_ring_arb(struct adf_etr_ring_data *ring); +int adf_disable_arb_thd(struct adf_accel_dev *accel_dev, u32 ae, u32 thr); + +int adf_dev_get(struct adf_accel_dev *accel_dev); +void adf_dev_put(struct adf_accel_dev *accel_dev); +int adf_dev_in_use(struct adf_accel_dev *accel_dev); +int adf_init_etr_data(struct adf_accel_dev *accel_dev); +void adf_cleanup_etr_data(struct adf_accel_dev *accel_dev); +int qat_crypto_register(void); +int qat_crypto_unregister(void); +int qat_crypto_vf_dev_config(struct adf_accel_dev *accel_dev); +struct qat_crypto_instance *qat_crypto_get_instance_node(int node); +void qat_crypto_put_instance(struct qat_crypto_instance *inst); +void qat_alg_callback(void *resp); +void qat_alg_asym_callback(void *resp); +int qat_algs_register(void); +void qat_algs_unregister(void); +int qat_asym_algs_register(void); +void qat_asym_algs_unregister(void); + +struct qat_compression_instance *qat_compression_get_instance_node(int node); +void qat_compression_put_instance(struct qat_compression_instance *inst); +int qat_compression_register(void); +int qat_compression_unregister(void); +int qat_comp_algs_register(void); +void qat_comp_algs_unregister(void); +void qat_comp_alg_callback(void *resp); + +int adf_isr_resource_alloc(struct adf_accel_dev *accel_dev); +void adf_isr_resource_free(struct adf_accel_dev *accel_dev); +int adf_vf_isr_resource_alloc(struct adf_accel_dev *accel_dev); +void adf_vf_isr_resource_free(struct adf_accel_dev *accel_dev); + +int adf_pfvf_comms_disabled(struct adf_accel_dev *accel_dev); + +int adf_sysfs_init(struct adf_accel_dev *accel_dev); + +int qat_hal_init(struct adf_accel_dev *accel_dev); +void qat_hal_deinit(struct icp_qat_fw_loader_handle *handle); +int qat_hal_start(struct icp_qat_fw_loader_handle *handle); +void qat_hal_stop(struct icp_qat_fw_loader_handle *handle, unsigned char ae, + unsigned int ctx_mask); +void qat_hal_reset(struct icp_qat_fw_loader_handle *handle); +int qat_hal_clr_reset(struct icp_qat_fw_loader_handle *handle); +void qat_hal_set_live_ctx(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask); +int qat_hal_check_ae_active(struct icp_qat_fw_loader_handle *handle, + unsigned int ae); +int qat_hal_set_ae_lm_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, enum icp_qat_uof_regtype lm_type, + unsigned char mode); +int qat_hal_set_ae_ctx_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char mode); +int qat_hal_set_ae_nn_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char mode); +void qat_hal_set_pc(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask, unsigned int upc); +void qat_hal_wr_uwords(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int uaddr, + unsigned int words_num, u64 *uword); +void qat_hal_wr_umem(struct icp_qat_fw_loader_handle *handle, unsigned char ae, + unsigned int uword_addr, unsigned int words_num, + unsigned int *data); +int qat_hal_get_ins_num(void); +int qat_hal_batch_wr_lm(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, + struct icp_qat_uof_batch_init *lm_init_header); +int qat_hal_init_gpr(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int regdata); +int qat_hal_init_wr_xfer(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int regdata); +int qat_hal_init_rd_xfer(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int regdata); +int qat_hal_init_nn(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + unsigned short reg_num, unsigned int regdata); +void qat_hal_set_ae_tindex_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char mode); +int qat_uclo_wr_all_uimage(struct icp_qat_fw_loader_handle *handle); +void qat_uclo_del_obj(struct icp_qat_fw_loader_handle *handle); +int qat_uclo_wr_mimage(struct icp_qat_fw_loader_handle *handle, void *addr_ptr, + int mem_size); +int qat_uclo_map_obj(struct icp_qat_fw_loader_handle *handle, + void *addr_ptr, u32 mem_size, const char *obj_name); +int qat_uclo_set_cfg_ae_mask(struct icp_qat_fw_loader_handle *handle, + unsigned int cfg_ae_mask); +int adf_init_misc_wq(void); +void adf_exit_misc_wq(void); +bool adf_misc_wq_queue_work(struct work_struct *work); +bool adf_misc_wq_queue_delayed_work(struct delayed_work *work, + unsigned long delay); +void adf_misc_wq_flush(void); +#if defined(CONFIG_PCI_IOV) +int adf_sriov_configure(struct pci_dev *pdev, int numvfs); +void adf_disable_sriov(struct adf_accel_dev *accel_dev); +void adf_reenable_sriov(struct adf_accel_dev *accel_dev); +void adf_enable_vf2pf_interrupts(struct adf_accel_dev *accel_dev, u32 vf_mask); +void adf_disable_all_vf2pf_interrupts(struct adf_accel_dev *accel_dev); +bool adf_recv_and_handle_pf2vf_msg(struct adf_accel_dev *accel_dev); +bool adf_recv_and_handle_vf2pf_msg(struct adf_accel_dev *accel_dev, u32 vf_nr); +int adf_pf2vf_handle_pf_restarting(struct adf_accel_dev *accel_dev); +void adf_enable_pf2vf_interrupts(struct adf_accel_dev *accel_dev); +void adf_disable_pf2vf_interrupts(struct adf_accel_dev *accel_dev); +void adf_schedule_vf2pf_handler(struct adf_accel_vf_info *vf_info); +int adf_init_pf_wq(void); +void adf_exit_pf_wq(void); +int adf_init_vf_wq(void); +void adf_exit_vf_wq(void); +void adf_flush_vf_wq(struct adf_accel_dev *accel_dev); +#else +#define adf_sriov_configure NULL + +static inline void adf_disable_sriov(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_reenable_sriov(struct adf_accel_dev *accel_dev) +{ +} + +static inline int adf_init_pf_wq(void) +{ + return 0; +} + +static inline void adf_exit_pf_wq(void) +{ +} + +static inline int adf_init_vf_wq(void) +{ + return 0; +} + +static inline void adf_exit_vf_wq(void) +{ +} + +#endif + +static inline void __iomem *adf_get_pmisc_base(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_bar *pmisc; + + pmisc = &GET_BARS(accel_dev)[hw_data->get_misc_bar_id(hw_data)]; + + return pmisc->virt_addr; +} + +static inline void __iomem *adf_get_etr_base(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_bar *etr; + + etr = &GET_BARS(accel_dev)[hw_data->get_etr_bar_id(hw_data)]; + + return etr->virt_addr; +} + +static inline void __iomem *adf_get_aram_base(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_bar *param; + + param = &GET_BARS(accel_dev)[hw_data->get_sram_bar_id(hw_data)]; + + return param->virt_addr; +} + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_ctl_drv.c b/drivers/crypto/intel/qat/qat_common/adf_ctl_drv.c new file mode 100644 index 000000000000..c2e6f0cb7480 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_ctl_drv.c @@ -0,0 +1,466 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ + +#include <crypto/algapi.h> +#include <linux/module.h> +#include <linux/mutex.h> +#include <linux/slab.h> +#include <linux/fs.h> +#include <linux/bitops.h> +#include <linux/pci.h> +#include <linux/cdev.h> +#include <linux/uaccess.h> + +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_cfg.h" +#include "adf_cfg_common.h" +#include "adf_cfg_user.h" + +#define ADF_CFG_MAX_SECTION 512 +#define ADF_CFG_MAX_KEY_VAL 256 + +#define DEVICE_NAME "qat_adf_ctl" + +static DEFINE_MUTEX(adf_ctl_lock); +static long adf_ctl_ioctl(struct file *fp, unsigned int cmd, unsigned long arg); + +static const struct file_operations adf_ctl_ops = { + .owner = THIS_MODULE, + .unlocked_ioctl = adf_ctl_ioctl, + .compat_ioctl = compat_ptr_ioctl, +}; + +static const struct class adf_ctl_class = { + .name = DEVICE_NAME, +}; + +struct adf_ctl_drv_info { + unsigned int major; + struct cdev drv_cdev; +}; + +static struct adf_ctl_drv_info adf_ctl_drv; + +static void adf_chr_drv_destroy(void) +{ + device_destroy(&adf_ctl_class, MKDEV(adf_ctl_drv.major, 0)); + cdev_del(&adf_ctl_drv.drv_cdev); + class_unregister(&adf_ctl_class); + unregister_chrdev_region(MKDEV(adf_ctl_drv.major, 0), 1); +} + +static int adf_chr_drv_create(void) +{ + dev_t dev_id; + struct device *drv_device; + int ret; + + if (alloc_chrdev_region(&dev_id, 0, 1, DEVICE_NAME)) { + pr_err("QAT: unable to allocate chrdev region\n"); + return -EFAULT; + } + + ret = class_register(&adf_ctl_class); + if (ret) + goto err_chrdev_unreg; + + adf_ctl_drv.major = MAJOR(dev_id); + cdev_init(&adf_ctl_drv.drv_cdev, &adf_ctl_ops); + if (cdev_add(&adf_ctl_drv.drv_cdev, dev_id, 1)) { + pr_err("QAT: cdev add failed\n"); + goto err_class_destr; + } + + drv_device = device_create(&adf_ctl_class, NULL, + MKDEV(adf_ctl_drv.major, 0), + NULL, DEVICE_NAME); + if (IS_ERR(drv_device)) { + pr_err("QAT: failed to create device\n"); + goto err_cdev_del; + } + return 0; +err_cdev_del: + cdev_del(&adf_ctl_drv.drv_cdev); +err_class_destr: + class_unregister(&adf_ctl_class); +err_chrdev_unreg: + unregister_chrdev_region(dev_id, 1); + return -EFAULT; +} + +static struct adf_user_cfg_ctl_data *adf_ctl_alloc_resources(unsigned long arg) +{ + struct adf_user_cfg_ctl_data *cfg_data; + + cfg_data = memdup_user((void __user *)arg, sizeof(*cfg_data)); + if (IS_ERR(cfg_data)) + pr_err("QAT: failed to copy from user cfg_data.\n"); + return cfg_data; +} + +static int adf_add_key_value_data(struct adf_accel_dev *accel_dev, + const char *section, + const struct adf_user_cfg_key_val *key_val) +{ + if (key_val->type == ADF_HEX) { + long *ptr = (long *)key_val->val; + long val = *ptr; + + if (adf_cfg_add_key_value_param(accel_dev, section, + key_val->key, (void *)val, + key_val->type)) { + dev_err(&GET_DEV(accel_dev), + "failed to add hex keyvalue.\n"); + return -EFAULT; + } + } else { + if (adf_cfg_add_key_value_param(accel_dev, section, + key_val->key, key_val->val, + key_val->type)) { + dev_err(&GET_DEV(accel_dev), + "failed to add keyvalue.\n"); + return -EFAULT; + } + } + return 0; +} + +static int adf_copy_key_value_data(struct adf_accel_dev *accel_dev, + struct adf_user_cfg_ctl_data *ctl_data) +{ + struct adf_user_cfg_key_val key_val; + struct adf_user_cfg_key_val *params_head; + struct adf_user_cfg_section section, *section_head; + int i, j; + + section_head = ctl_data->config_section; + + for (i = 0; section_head && i < ADF_CFG_MAX_SECTION; i++) { + if (copy_from_user(§ion, (void __user *)section_head, + sizeof(*section_head))) { + dev_err(&GET_DEV(accel_dev), + "failed to copy section info\n"); + goto out_err; + } + + if (adf_cfg_section_add(accel_dev, section.name)) { + dev_err(&GET_DEV(accel_dev), + "failed to add section.\n"); + goto out_err; + } + + params_head = section.params; + + for (j = 0; params_head && j < ADF_CFG_MAX_KEY_VAL; j++) { + if (copy_from_user(&key_val, (void __user *)params_head, + sizeof(key_val))) { + dev_err(&GET_DEV(accel_dev), + "Failed to copy keyvalue.\n"); + goto out_err; + } + if (adf_add_key_value_data(accel_dev, section.name, + &key_val)) { + goto out_err; + } + params_head = key_val.next; + } + section_head = section.next; + } + return 0; +out_err: + adf_cfg_del_all(accel_dev); + return -EFAULT; +} + +static int adf_ctl_ioctl_dev_config(struct file *fp, unsigned int cmd, + unsigned long arg) +{ + struct adf_user_cfg_ctl_data *ctl_data; + struct adf_accel_dev *accel_dev; + int ret = 0; + + ctl_data = adf_ctl_alloc_resources(arg); + if (IS_ERR(ctl_data)) + return PTR_ERR(ctl_data); + + accel_dev = adf_devmgr_get_dev_by_id(ctl_data->device_id); + if (!accel_dev) { + ret = -EFAULT; + goto out; + } + + if (adf_dev_started(accel_dev)) { + ret = -EFAULT; + goto out; + } + + if (adf_copy_key_value_data(accel_dev, ctl_data)) { + ret = -EFAULT; + goto out; + } + set_bit(ADF_STATUS_CONFIGURED, &accel_dev->status); +out: + kfree(ctl_data); + return ret; +} + +static int adf_ctl_is_device_in_use(int id) +{ + struct adf_accel_dev *dev; + + list_for_each_entry(dev, adf_devmgr_get_head(), list) { + if (id == dev->accel_id || id == ADF_CFG_ALL_DEVICES) { + if (adf_devmgr_in_reset(dev) || adf_dev_in_use(dev)) { + dev_info(&GET_DEV(dev), + "device qat_dev%d is busy\n", + dev->accel_id); + return -EBUSY; + } + } + } + return 0; +} + +static void adf_ctl_stop_devices(u32 id) +{ + struct adf_accel_dev *accel_dev; + + list_for_each_entry(accel_dev, adf_devmgr_get_head(), list) { + if (id == accel_dev->accel_id || id == ADF_CFG_ALL_DEVICES) { + if (!adf_dev_started(accel_dev)) + continue; + + /* First stop all VFs */ + if (!accel_dev->is_vf) + continue; + + adf_dev_down(accel_dev); + } + } + + list_for_each_entry(accel_dev, adf_devmgr_get_head(), list) { + if (id == accel_dev->accel_id || id == ADF_CFG_ALL_DEVICES) { + if (!adf_dev_started(accel_dev)) + continue; + + adf_dev_down(accel_dev); + } + } +} + +static int adf_ctl_ioctl_dev_stop(struct file *fp, unsigned int cmd, + unsigned long arg) +{ + int ret; + struct adf_user_cfg_ctl_data *ctl_data; + + ctl_data = adf_ctl_alloc_resources(arg); + if (IS_ERR(ctl_data)) + return PTR_ERR(ctl_data); + + if (adf_devmgr_verify_id(ctl_data->device_id)) { + pr_err("QAT: Device %d not found\n", ctl_data->device_id); + ret = -ENODEV; + goto out; + } + + ret = adf_ctl_is_device_in_use(ctl_data->device_id); + if (ret) + goto out; + + if (ctl_data->device_id == ADF_CFG_ALL_DEVICES) + pr_info("QAT: Stopping all acceleration devices.\n"); + else + pr_info("QAT: Stopping acceleration device qat_dev%d.\n", + ctl_data->device_id); + + adf_ctl_stop_devices(ctl_data->device_id); + +out: + kfree(ctl_data); + return ret; +} + +static int adf_ctl_ioctl_dev_start(struct file *fp, unsigned int cmd, + unsigned long arg) +{ + int ret; + struct adf_user_cfg_ctl_data *ctl_data; + struct adf_accel_dev *accel_dev; + + ctl_data = adf_ctl_alloc_resources(arg); + if (IS_ERR(ctl_data)) + return PTR_ERR(ctl_data); + + ret = -ENODEV; + accel_dev = adf_devmgr_get_dev_by_id(ctl_data->device_id); + if (!accel_dev) + goto out; + + dev_info(&GET_DEV(accel_dev), + "Starting acceleration device qat_dev%d.\n", + ctl_data->device_id); + + ret = adf_dev_up(accel_dev, false); + + if (ret) { + dev_err(&GET_DEV(accel_dev), "Failed to start qat_dev%d\n", + ctl_data->device_id); + adf_dev_down(accel_dev); + } +out: + kfree(ctl_data); + return ret; +} + +static int adf_ctl_ioctl_get_num_devices(struct file *fp, unsigned int cmd, + unsigned long arg) +{ + u32 num_devices = 0; + + adf_devmgr_get_num_dev(&num_devices); + if (copy_to_user((void __user *)arg, &num_devices, sizeof(num_devices))) + return -EFAULT; + + return 0; +} + +static int adf_ctl_ioctl_get_status(struct file *fp, unsigned int cmd, + unsigned long arg) +{ + struct adf_hw_device_data *hw_data; + struct adf_dev_status_info dev_info; + struct adf_accel_dev *accel_dev; + + if (copy_from_user(&dev_info, (void __user *)arg, + sizeof(struct adf_dev_status_info))) { + pr_err("QAT: failed to copy from user.\n"); + return -EFAULT; + } + + accel_dev = adf_devmgr_get_dev_by_id(dev_info.accel_id); + if (!accel_dev) + return -ENODEV; + + hw_data = accel_dev->hw_device; + dev_info.state = adf_dev_started(accel_dev) ? DEV_UP : DEV_DOWN; + dev_info.num_ae = hw_data->get_num_aes(hw_data); + dev_info.num_accel = hw_data->get_num_accels(hw_data); + dev_info.num_logical_accel = hw_data->num_logical_accel; + dev_info.banks_per_accel = hw_data->num_banks + / hw_data->num_logical_accel; + strscpy(dev_info.name, hw_data->dev_class->name, sizeof(dev_info.name)); + dev_info.instance_id = hw_data->instance_id; + dev_info.type = hw_data->dev_class->type; + dev_info.bus = accel_to_pci_dev(accel_dev)->bus->number; + dev_info.dev = PCI_SLOT(accel_to_pci_dev(accel_dev)->devfn); + dev_info.fun = PCI_FUNC(accel_to_pci_dev(accel_dev)->devfn); + + if (copy_to_user((void __user *)arg, &dev_info, + sizeof(struct adf_dev_status_info))) { + dev_err(&GET_DEV(accel_dev), "failed to copy status.\n"); + return -EFAULT; + } + return 0; +} + +static long adf_ctl_ioctl(struct file *fp, unsigned int cmd, unsigned long arg) +{ + int ret; + + if (mutex_lock_interruptible(&adf_ctl_lock)) + return -EFAULT; + + switch (cmd) { + case IOCTL_CONFIG_SYS_RESOURCE_PARAMETERS: + ret = adf_ctl_ioctl_dev_config(fp, cmd, arg); + break; + + case IOCTL_STOP_ACCEL_DEV: + ret = adf_ctl_ioctl_dev_stop(fp, cmd, arg); + break; + + case IOCTL_START_ACCEL_DEV: + ret = adf_ctl_ioctl_dev_start(fp, cmd, arg); + break; + + case IOCTL_GET_NUM_DEVICES: + ret = adf_ctl_ioctl_get_num_devices(fp, cmd, arg); + break; + + case IOCTL_STATUS_ACCEL_DEV: + ret = adf_ctl_ioctl_get_status(fp, cmd, arg); + break; + default: + pr_err_ratelimited("QAT: Invalid ioctl %d\n", cmd); + ret = -EFAULT; + break; + } + mutex_unlock(&adf_ctl_lock); + return ret; +} + +static int __init adf_register_ctl_device_driver(void) +{ + if (adf_chr_drv_create()) + goto err_chr_dev; + + if (adf_init_misc_wq()) + goto err_misc_wq; + + if (adf_init_aer()) + goto err_aer; + + if (adf_init_pf_wq()) + goto err_pf_wq; + + if (adf_init_vf_wq()) + goto err_vf_wq; + + if (qat_crypto_register()) + goto err_crypto_register; + + if (qat_compression_register()) + goto err_compression_register; + + return 0; + +err_compression_register: + qat_crypto_unregister(); +err_crypto_register: + adf_exit_vf_wq(); +err_vf_wq: + adf_exit_pf_wq(); +err_pf_wq: + adf_exit_aer(); +err_aer: + adf_exit_misc_wq(); +err_misc_wq: + adf_chr_drv_destroy(); +err_chr_dev: + mutex_destroy(&adf_ctl_lock); + return -EFAULT; +} + +static void __exit adf_unregister_ctl_device_driver(void) +{ + adf_chr_drv_destroy(); + adf_exit_misc_wq(); + adf_exit_aer(); + adf_exit_vf_wq(); + adf_exit_pf_wq(); + qat_crypto_unregister(); + qat_compression_unregister(); + adf_clean_vf_map(false); + mutex_destroy(&adf_ctl_lock); +} + +module_init(adf_register_ctl_device_driver); +module_exit(adf_unregister_ctl_device_driver); +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_ALIAS_CRYPTO("intel_qat"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_IMPORT_NS("CRYPTO_INTERNAL"); diff --git a/drivers/crypto/intel/qat/qat_common/adf_dbgfs.c b/drivers/crypto/intel/qat/qat_common/adf_dbgfs.c new file mode 100644 index 000000000000..4c11ad1ebcf0 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_dbgfs.c @@ -0,0 +1,79 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/debugfs.h> +#include "adf_accel_devices.h" +#include "adf_cfg.h" +#include "adf_common_drv.h" +#include "adf_cnv_dbgfs.h" +#include "adf_dbgfs.h" +#include "adf_fw_counters.h" +#include "adf_heartbeat_dbgfs.h" +#include "adf_pm_dbgfs.h" +#include "adf_tl_debugfs.h" + +/** + * adf_dbgfs_init() - add persistent debugfs entries + * @accel_dev: Pointer to acceleration device. + * + * This function creates debugfs entries that are persistent through a device + * state change (from up to down or vice versa). + */ +void adf_dbgfs_init(struct adf_accel_dev *accel_dev) +{ + char name[ADF_DEVICE_NAME_LENGTH]; + + /* Create dev top level debugfs entry */ + snprintf(name, sizeof(name), "%s%s_%s", ADF_DEVICE_NAME_PREFIX, + accel_dev->hw_device->dev_class->name, + pci_name(accel_dev->accel_pci_dev.pci_dev)); + + accel_dev->debugfs_dir = debugfs_create_dir(name, NULL); + + adf_cfg_dev_dbgfs_add(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_dbgfs_init); + +/** + * adf_dbgfs_exit() - remove persistent debugfs entries + * @accel_dev: Pointer to acceleration device. + */ +void adf_dbgfs_exit(struct adf_accel_dev *accel_dev) +{ + adf_cfg_dev_dbgfs_rm(accel_dev); + debugfs_remove(accel_dev->debugfs_dir); +} +EXPORT_SYMBOL_GPL(adf_dbgfs_exit); + +/** + * adf_dbgfs_add() - add non-persistent debugfs entries + * @accel_dev: Pointer to acceleration device. + * + * This function creates debugfs entries that are not persistent through + * a device state change (from up to down or vice versa). + */ +void adf_dbgfs_add(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->is_vf) { + adf_fw_counters_dbgfs_add(accel_dev); + adf_heartbeat_dbgfs_add(accel_dev); + adf_pm_dbgfs_add(accel_dev); + adf_cnv_dbgfs_add(accel_dev); + adf_tl_dbgfs_add(accel_dev); + } +} + +/** + * adf_dbgfs_rm() - remove non-persistent debugfs entries + * @accel_dev: Pointer to acceleration device. + */ +void adf_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->is_vf) { + adf_tl_dbgfs_rm(accel_dev); + adf_cnv_dbgfs_rm(accel_dev); + adf_pm_dbgfs_rm(accel_dev); + adf_heartbeat_dbgfs_rm(accel_dev); + adf_fw_counters_dbgfs_rm(accel_dev); + } +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_dbgfs.h b/drivers/crypto/intel/qat/qat_common/adf_dbgfs.h new file mode 100644 index 000000000000..e0cb2c2a2ed0 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_dbgfs.h @@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_DBGFS_H +#define ADF_DBGFS_H + +#ifdef CONFIG_DEBUG_FS +void adf_dbgfs_init(struct adf_accel_dev *accel_dev); +void adf_dbgfs_add(struct adf_accel_dev *accel_dev); +void adf_dbgfs_rm(struct adf_accel_dev *accel_dev); +void adf_dbgfs_exit(struct adf_accel_dev *accel_dev); +#else +static inline void adf_dbgfs_init(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_dbgfs_add(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_dbgfs_exit(struct adf_accel_dev *accel_dev) +{ +} +#endif +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_dc.c b/drivers/crypto/intel/qat/qat_common/adf_dc.c new file mode 100644 index 000000000000..3e8fb4e3ed97 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_dc.c @@ -0,0 +1,64 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2022 Intel Corporation */ +#include "adf_accel_devices.h" +#include "adf_dc.h" +#include "icp_qat_fw_comp.h" + +int qat_comp_build_ctx(struct adf_accel_dev *accel_dev, void *ctx, enum adf_dc_algo algo) +{ + struct icp_qat_fw_comp_req *req_tmpl = ctx; + struct icp_qat_fw_comp_cd_hdr *comp_cd_ctrl = &req_tmpl->comp_cd_ctrl; + struct icp_qat_fw_comp_req_params *req_pars = &req_tmpl->comp_pars; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + int ret; + + memset(req_tmpl, 0, sizeof(*req_tmpl)); + header->hdr_flags = + ICP_QAT_FW_COMN_HDR_FLAGS_BUILD(ICP_QAT_FW_COMN_REQ_FLAG_SET); + header->service_type = ICP_QAT_FW_COMN_REQ_CPM_FW_COMP; + header->comn_req_flags = + ICP_QAT_FW_COMN_FLAGS_BUILD(QAT_COMN_CD_FLD_TYPE_16BYTE_DATA, + QAT_COMN_PTR_TYPE_SGL); + header->serv_specif_flags = + ICP_QAT_FW_COMP_FLAGS_BUILD(ICP_QAT_FW_COMP_STATELESS_SESSION, + ICP_QAT_FW_COMP_NOT_AUTO_SELECT_BEST, + ICP_QAT_FW_COMP_NOT_ENH_AUTO_SELECT_BEST, + ICP_QAT_FW_COMP_NOT_DISABLE_TYPE0_ENH_AUTO_SELECT_BEST, + ICP_QAT_FW_COMP_ENABLE_SECURE_RAM_USED_AS_INTMD_BUF); + + /* Build HW config block for compression */ + ret = GET_DC_OPS(accel_dev)->build_comp_block(ctx, algo); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Failed to build compression block\n"); + return ret; + } + + req_pars->crc.legacy.initial_adler = COMP_CPR_INITIAL_ADLER; + req_pars->crc.legacy.initial_crc32 = COMP_CPR_INITIAL_CRC; + req_pars->req_par_flags = + ICP_QAT_FW_COMP_REQ_PARAM_FLAGS_BUILD(ICP_QAT_FW_COMP_SOP, + ICP_QAT_FW_COMP_EOP, + ICP_QAT_FW_COMP_BFINAL, + ICP_QAT_FW_COMP_CNV, + ICP_QAT_FW_COMP_CNV_RECOVERY, + ICP_QAT_FW_COMP_NO_CNV_DFX, + ICP_QAT_FW_COMP_CRC_MODE_LEGACY, + ICP_QAT_FW_COMP_NO_XXHASH_ACC, + ICP_QAT_FW_COMP_CNV_ERROR_NONE, + ICP_QAT_FW_COMP_NO_APPEND_CRC, + ICP_QAT_FW_COMP_NO_DROP_DATA, + ICP_QAT_FW_COMP_NO_PARTIAL_DECOMPRESS); + ICP_QAT_FW_COMN_NEXT_ID_SET(comp_cd_ctrl, ICP_QAT_FW_SLICE_DRAM_WR); + ICP_QAT_FW_COMN_CURR_ID_SET(comp_cd_ctrl, ICP_QAT_FW_SLICE_COMP); + + /* Fill second half of the template for decompression */ + memcpy(req_tmpl + 1, req_tmpl, sizeof(*req_tmpl)); + req_tmpl++; + + /* Build HW config block for decompression */ + ret = GET_DC_OPS(accel_dev)->build_decomp_block(req_tmpl, algo); + if (ret) + dev_err(&GET_DEV(accel_dev), "Failed to build decompression block\n"); + + return ret; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_dc.h b/drivers/crypto/intel/qat/qat_common/adf_dc.h new file mode 100644 index 000000000000..6cb5e09054a6 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_dc.h @@ -0,0 +1,17 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ADF_DC_H +#define ADF_DC_H + +struct adf_accel_dev; + +enum adf_dc_algo { + QAT_DEFLATE, + QAT_LZ4, + QAT_LZ4S, + QAT_ZSTD, +}; + +int qat_comp_build_ctx(struct adf_accel_dev *accel_dev, void *ctx, enum adf_dc_algo algo); + +#endif /* ADF_DC_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_dev_mgr.c b/drivers/crypto/intel/qat/qat_common/adf_dev_mgr.c new file mode 100644 index 000000000000..34b9f7731c78 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_dev_mgr.c @@ -0,0 +1,442 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/mutex.h> +#include <linux/list.h> +#include "adf_cfg.h" +#include "adf_common_drv.h" + +static LIST_HEAD(accel_table); +static LIST_HEAD(vfs_table); +static DEFINE_MUTEX(table_lock); +static u32 num_devices; +static u8 id_map[ADF_MAX_DEVICES]; + +struct vf_id_map { + u32 bdf; + u32 id; + u32 fake_id; + bool attached; + struct list_head list; +}; + +static int adf_get_vf_id(struct adf_accel_dev *vf) +{ + return ((7 * (PCI_SLOT(accel_to_pci_dev(vf)->devfn) - 1)) + + PCI_FUNC(accel_to_pci_dev(vf)->devfn) + + (PCI_SLOT(accel_to_pci_dev(vf)->devfn) - 1)); +} + +static int adf_get_vf_num(struct adf_accel_dev *vf) +{ + return (accel_to_pci_dev(vf)->bus->number << 8) | adf_get_vf_id(vf); +} + +static struct vf_id_map *adf_find_vf(u32 bdf) +{ + struct list_head *itr; + + list_for_each(itr, &vfs_table) { + struct vf_id_map *ptr = + list_entry(itr, struct vf_id_map, list); + + if (ptr->bdf == bdf) + return ptr; + } + return NULL; +} + +static int adf_get_vf_real_id(u32 fake) +{ + struct list_head *itr; + + list_for_each(itr, &vfs_table) { + struct vf_id_map *ptr = + list_entry(itr, struct vf_id_map, list); + if (ptr->fake_id == fake) + return ptr->id; + } + return -1; +} + +/** + * adf_clean_vf_map() - Cleans VF id mappings + * @vf: flag indicating whether mappings is cleaned + * for vfs only or for vfs and pfs + * + * Function cleans internal ids for virtual functions. + */ +void adf_clean_vf_map(bool vf) +{ + struct vf_id_map *map; + struct list_head *ptr, *tmp; + + mutex_lock(&table_lock); + list_for_each_safe(ptr, tmp, &vfs_table) { + map = list_entry(ptr, struct vf_id_map, list); + if (map->bdf != -1) { + id_map[map->id] = 0; + num_devices--; + } + + if (vf && map->bdf == -1) + continue; + + list_del(ptr); + kfree(map); + } + mutex_unlock(&table_lock); +} +EXPORT_SYMBOL_GPL(adf_clean_vf_map); + +/** + * adf_devmgr_update_class_index() - Update internal index + * @hw_data: Pointer to internal device data. + * + * Function updates internal dev index for VFs + */ +void adf_devmgr_update_class_index(struct adf_hw_device_data *hw_data) +{ + struct adf_hw_device_class *class = hw_data->dev_class; + struct list_head *itr; + int i = 0; + + list_for_each(itr, &accel_table) { + struct adf_accel_dev *ptr = + list_entry(itr, struct adf_accel_dev, list); + + if (ptr->hw_device->dev_class == class) + ptr->hw_device->instance_id = i++; + + if (i == class->instances) + break; + } +} +EXPORT_SYMBOL_GPL(adf_devmgr_update_class_index); + +static unsigned int adf_find_free_id(void) +{ + unsigned int i; + + for (i = 0; i < ADF_MAX_DEVICES; i++) { + if (!id_map[i]) { + id_map[i] = 1; + return i; + } + } + return ADF_MAX_DEVICES + 1; +} + +/** + * adf_devmgr_add_dev() - Add accel_dev to the acceleration framework + * @accel_dev: Pointer to acceleration device. + * @pf: Corresponding PF if the accel_dev is a VF + * + * Function adds acceleration device to the acceleration framework. + * To be used by QAT device specific drivers. + * + * Return: 0 on success, error code otherwise. + */ +int adf_devmgr_add_dev(struct adf_accel_dev *accel_dev, + struct adf_accel_dev *pf) +{ + struct list_head *itr; + int ret = 0; + + if (num_devices == ADF_MAX_DEVICES) { + dev_err(&GET_DEV(accel_dev), "Only support up to %d devices\n", + ADF_MAX_DEVICES); + return -EFAULT; + } + + mutex_lock(&table_lock); + atomic_set(&accel_dev->ref_count, 0); + + /* PF on host or VF on guest - optimized to remove redundant is_vf */ + if (!accel_dev->is_vf || !pf) { + struct vf_id_map *map; + + list_for_each(itr, &accel_table) { + struct adf_accel_dev *ptr = + list_entry(itr, struct adf_accel_dev, list); + + if (ptr == accel_dev) { + ret = -EEXIST; + goto unlock; + } + } + + list_add_tail(&accel_dev->list, &accel_table); + accel_dev->accel_id = adf_find_free_id(); + if (accel_dev->accel_id > ADF_MAX_DEVICES) { + ret = -EFAULT; + goto unlock; + } + num_devices++; + map = kzalloc(sizeof(*map), GFP_KERNEL); + if (!map) { + ret = -ENOMEM; + goto unlock; + } + map->bdf = ~0; + map->id = accel_dev->accel_id; + map->fake_id = map->id; + map->attached = true; + list_add_tail(&map->list, &vfs_table); + } else if (accel_dev->is_vf && pf) { + /* VF on host */ + struct vf_id_map *map; + + map = adf_find_vf(adf_get_vf_num(accel_dev)); + if (map) { + struct vf_id_map *next; + + accel_dev->accel_id = map->id; + list_add_tail(&accel_dev->list, &accel_table); + map->fake_id++; + map->attached = true; + next = list_next_entry(map, list); + while (next && &next->list != &vfs_table) { + next->fake_id++; + next = list_next_entry(next, list); + } + + ret = 0; + goto unlock; + } + + map = kzalloc(sizeof(*map), GFP_KERNEL); + if (!map) { + ret = -ENOMEM; + goto unlock; + } + accel_dev->accel_id = adf_find_free_id(); + if (accel_dev->accel_id > ADF_MAX_DEVICES) { + kfree(map); + ret = -EFAULT; + goto unlock; + } + num_devices++; + list_add_tail(&accel_dev->list, &accel_table); + map->bdf = adf_get_vf_num(accel_dev); + map->id = accel_dev->accel_id; + map->fake_id = map->id; + map->attached = true; + list_add_tail(&map->list, &vfs_table); + } + mutex_init(&accel_dev->state_lock); +unlock: + mutex_unlock(&table_lock); + return ret; +} +EXPORT_SYMBOL_GPL(adf_devmgr_add_dev); + +struct list_head *adf_devmgr_get_head(void) +{ + return &accel_table; +} + +/** + * adf_devmgr_rm_dev() - Remove accel_dev from the acceleration framework. + * @accel_dev: Pointer to acceleration device. + * @pf: Corresponding PF if the accel_dev is a VF + * + * Function removes acceleration device from the acceleration framework. + * To be used by QAT device specific drivers. + * + * Return: void + */ +void adf_devmgr_rm_dev(struct adf_accel_dev *accel_dev, + struct adf_accel_dev *pf) +{ + mutex_lock(&table_lock); + /* PF on host or VF on guest - optimized to remove redundant is_vf */ + if (!accel_dev->is_vf || !pf) { + id_map[accel_dev->accel_id] = 0; + num_devices--; + } else if (accel_dev->is_vf && pf) { + struct vf_id_map *map, *next; + + map = adf_find_vf(adf_get_vf_num(accel_dev)); + if (!map) { + dev_err(&GET_DEV(accel_dev), "Failed to find VF map\n"); + goto unlock; + } + map->fake_id--; + map->attached = false; + next = list_next_entry(map, list); + while (next && &next->list != &vfs_table) { + next->fake_id--; + next = list_next_entry(next, list); + } + } +unlock: + mutex_destroy(&accel_dev->state_lock); + list_del(&accel_dev->list); + mutex_unlock(&table_lock); +} +EXPORT_SYMBOL_GPL(adf_devmgr_rm_dev); + +/** + * adf_devmgr_pci_to_accel_dev() - Get accel_dev associated with the pci_dev. + * @pci_dev: Pointer to PCI device. + * + * Function returns acceleration device associated with the given PCI device. + * To be used by QAT device specific drivers. + * + * Return: pointer to accel_dev or NULL if not found. + */ +struct adf_accel_dev *adf_devmgr_pci_to_accel_dev(struct pci_dev *pci_dev) +{ + struct list_head *itr; + + mutex_lock(&table_lock); + list_for_each(itr, &accel_table) { + struct adf_accel_dev *ptr = + list_entry(itr, struct adf_accel_dev, list); + + if (ptr->accel_pci_dev.pci_dev == pci_dev) { + mutex_unlock(&table_lock); + return ptr; + } + } + mutex_unlock(&table_lock); + return NULL; +} +EXPORT_SYMBOL_GPL(adf_devmgr_pci_to_accel_dev); + +struct adf_accel_dev *adf_devmgr_get_dev_by_id(u32 id) +{ + struct list_head *itr; + int real_id; + + mutex_lock(&table_lock); + real_id = adf_get_vf_real_id(id); + if (real_id < 0) + goto unlock; + + id = real_id; + + list_for_each(itr, &accel_table) { + struct adf_accel_dev *ptr = + list_entry(itr, struct adf_accel_dev, list); + if (ptr->accel_id == id) { + mutex_unlock(&table_lock); + return ptr; + } + } +unlock: + mutex_unlock(&table_lock); + return NULL; +} + +int adf_devmgr_verify_id(u32 id) +{ + if (id == ADF_CFG_ALL_DEVICES) + return 0; + + if (adf_devmgr_get_dev_by_id(id)) + return 0; + + return -ENODEV; +} + +static int adf_get_num_dettached_vfs(void) +{ + struct list_head *itr; + int vfs = 0; + + mutex_lock(&table_lock); + list_for_each(itr, &vfs_table) { + struct vf_id_map *ptr = + list_entry(itr, struct vf_id_map, list); + if (ptr->bdf != ~0 && !ptr->attached) + vfs++; + } + mutex_unlock(&table_lock); + return vfs; +} + +void adf_devmgr_get_num_dev(u32 *num) +{ + *num = num_devices - adf_get_num_dettached_vfs(); +} + +/** + * adf_dev_in_use() - Check whether accel_dev is currently in use + * @accel_dev: Pointer to acceleration device. + * + * To be used by QAT device specific drivers. + * + * Return: 1 when device is in use, 0 otherwise. + */ +int adf_dev_in_use(struct adf_accel_dev *accel_dev) +{ + return atomic_read(&accel_dev->ref_count) != 0; +} +EXPORT_SYMBOL_GPL(adf_dev_in_use); + +/** + * adf_dev_get() - Increment accel_dev reference count + * @accel_dev: Pointer to acceleration device. + * + * Increment the accel_dev refcount and if this is the first time + * incrementing it during this period the accel_dev is in use, + * increment the module refcount too. + * To be used by QAT device specific drivers. + * + * Return: 0 when successful, EFAULT when fail to bump module refcount + */ +int adf_dev_get(struct adf_accel_dev *accel_dev) +{ + if (atomic_add_return(1, &accel_dev->ref_count) == 1) + if (!try_module_get(accel_dev->owner)) + return -EFAULT; + return 0; +} +EXPORT_SYMBOL_GPL(adf_dev_get); + +/** + * adf_dev_put() - Decrement accel_dev reference count + * @accel_dev: Pointer to acceleration device. + * + * Decrement the accel_dev refcount and if this is the last time + * decrementing it during this period the accel_dev is in use, + * decrement the module refcount too. + * To be used by QAT device specific drivers. + * + * Return: void + */ +void adf_dev_put(struct adf_accel_dev *accel_dev) +{ + if (atomic_sub_return(1, &accel_dev->ref_count) == 0) + module_put(accel_dev->owner); +} +EXPORT_SYMBOL_GPL(adf_dev_put); + +/** + * adf_devmgr_in_reset() - Check whether device is in reset + * @accel_dev: Pointer to acceleration device. + * + * To be used by QAT device specific drivers. + * + * Return: 1 when the device is being reset, 0 otherwise. + */ +int adf_devmgr_in_reset(struct adf_accel_dev *accel_dev) +{ + return test_bit(ADF_STATUS_RESTARTING, &accel_dev->status); +} +EXPORT_SYMBOL_GPL(adf_devmgr_in_reset); + +/** + * adf_dev_started() - Check whether device has started + * @accel_dev: Pointer to acceleration device. + * + * To be used by QAT device specific drivers. + * + * Return: 1 when the device has started, 0 otherwise + */ +int adf_dev_started(struct adf_accel_dev *accel_dev) +{ + return test_bit(ADF_STATUS_STARTED, &accel_dev->status); +} +EXPORT_SYMBOL_GPL(adf_dev_started); diff --git a/drivers/crypto/intel/qat/qat_common/adf_fw_config.h b/drivers/crypto/intel/qat/qat_common/adf_fw_config.h new file mode 100644 index 000000000000..78957fa900b7 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_fw_config.h @@ -0,0 +1,19 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_FW_CONFIG_H_ +#define ADF_FW_CONFIG_H_ + +enum adf_fw_objs { + ADF_FW_SYM_OBJ, + ADF_FW_ASYM_OBJ, + ADF_FW_DC_OBJ, + ADF_FW_ADMIN_OBJ, + ADF_FW_CY_OBJ, +}; + +struct adf_fw_config { + u32 ae_mask; + enum adf_fw_objs obj; +}; + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_fw_counters.c b/drivers/crypto/intel/qat/qat_common/adf_fw_counters.c new file mode 100644 index 000000000000..98fb7ccfed9f --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_fw_counters.c @@ -0,0 +1,265 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include <linux/bitops.h> +#include <linux/debugfs.h> +#include <linux/err.h> +#include <linux/fs.h> +#include <linux/kernel.h> +#include <linux/seq_file.h> +#include <linux/types.h> + +#include "adf_accel_devices.h" +#include "adf_admin.h" +#include "adf_common_drv.h" +#include "adf_fw_counters.h" + +#define ADF_FW_COUNTERS_MAX_PADDING 16 + +enum adf_fw_counters_types { + ADF_FW_REQUESTS, + ADF_FW_RESPONSES, + ADF_FW_COUNTERS_COUNT +}; + +static const char * const adf_fw_counter_names[] = { + [ADF_FW_REQUESTS] = "Requests", + [ADF_FW_RESPONSES] = "Responses", +}; + +static_assert(ARRAY_SIZE(adf_fw_counter_names) == ADF_FW_COUNTERS_COUNT); + +struct adf_ae_counters { + u16 ae; + u64 values[ADF_FW_COUNTERS_COUNT]; +}; + +struct adf_fw_counters { + u16 ae_count; + struct adf_ae_counters ae_counters[] __counted_by(ae_count); +}; + +static void adf_fw_counters_parse_ae_values(struct adf_ae_counters *ae_counters, u32 ae, + u64 req_count, u64 resp_count) +{ + ae_counters->ae = ae; + ae_counters->values[ADF_FW_REQUESTS] = req_count; + ae_counters->values[ADF_FW_RESPONSES] = resp_count; +} + +static int adf_fw_counters_load_from_device(struct adf_accel_dev *accel_dev, + struct adf_fw_counters *fw_counters) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + unsigned long ae_mask; + unsigned int i; + unsigned long ae; + + /* Ignore the admin AEs */ + ae_mask = hw_data->ae_mask & ~hw_data->admin_ae_mask; + + if (hweight_long(ae_mask) > fw_counters->ae_count) + return -EINVAL; + + i = 0; + for_each_set_bit(ae, &ae_mask, GET_MAX_ACCELENGINES(accel_dev)) { + u64 req_count, resp_count; + int ret; + + ret = adf_get_ae_fw_counters(accel_dev, ae, &req_count, &resp_count); + if (ret) + return ret; + + adf_fw_counters_parse_ae_values(&fw_counters->ae_counters[i++], ae, + req_count, resp_count); + } + + return 0; +} + +static struct adf_fw_counters *adf_fw_counters_allocate(unsigned long ae_count) +{ + struct adf_fw_counters *fw_counters; + + if (unlikely(!ae_count)) + return ERR_PTR(-EINVAL); + + fw_counters = kmalloc(struct_size(fw_counters, ae_counters, ae_count), GFP_KERNEL); + if (!fw_counters) + return ERR_PTR(-ENOMEM); + + fw_counters->ae_count = ae_count; + + return fw_counters; +} + +/** + * adf_fw_counters_get() - Return FW counters for the provided device. + * @accel_dev: Pointer to a QAT acceleration device + * + * Allocates and returns a table of counters containing execution statistics + * for each non-admin AE available through the supplied acceleration device. + * The caller becomes the owner of such memory and is responsible for + * the deallocation through a call to kfree(). + * + * Returns: a pointer to a dynamically allocated struct adf_fw_counters + * on success, or a negative value on error. + */ +static struct adf_fw_counters *adf_fw_counters_get(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct adf_fw_counters *fw_counters; + unsigned long ae_count; + int ret; + + if (!adf_dev_started(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "QAT Device not started\n"); + return ERR_PTR(-EFAULT); + } + + /* Ignore the admin AEs */ + ae_count = hweight_long(hw_data->ae_mask & ~hw_data->admin_ae_mask); + + fw_counters = adf_fw_counters_allocate(ae_count); + if (IS_ERR(fw_counters)) + return fw_counters; + + ret = adf_fw_counters_load_from_device(accel_dev, fw_counters); + if (ret) { + kfree(fw_counters); + dev_err(&GET_DEV(accel_dev), + "Failed to create QAT fw_counters file table [%d].\n", ret); + return ERR_PTR(ret); + } + + return fw_counters; +} + +static void *qat_fw_counters_seq_start(struct seq_file *sfile, loff_t *pos) +{ + struct adf_fw_counters *fw_counters = sfile->private; + + if (*pos == 0) + return SEQ_START_TOKEN; + + if (*pos > fw_counters->ae_count) + return NULL; + + return &fw_counters->ae_counters[*pos - 1]; +} + +static void *qat_fw_counters_seq_next(struct seq_file *sfile, void *v, loff_t *pos) +{ + struct adf_fw_counters *fw_counters = sfile->private; + + (*pos)++; + + if (*pos > fw_counters->ae_count) + return NULL; + + return &fw_counters->ae_counters[*pos - 1]; +} + +static void qat_fw_counters_seq_stop(struct seq_file *sfile, void *v) {} + +static int qat_fw_counters_seq_show(struct seq_file *sfile, void *v) +{ + int i; + + if (v == SEQ_START_TOKEN) { + seq_puts(sfile, "AE "); + for (i = 0; i < ADF_FW_COUNTERS_COUNT; ++i) + seq_printf(sfile, " %*s", ADF_FW_COUNTERS_MAX_PADDING, + adf_fw_counter_names[i]); + } else { + struct adf_ae_counters *ae_counters = (struct adf_ae_counters *)v; + + seq_printf(sfile, "%2d:", ae_counters->ae); + for (i = 0; i < ADF_FW_COUNTERS_COUNT; ++i) + seq_printf(sfile, " %*llu", ADF_FW_COUNTERS_MAX_PADDING, + ae_counters->values[i]); + } + seq_putc(sfile, '\n'); + + return 0; +} + +static const struct seq_operations qat_fw_counters_sops = { + .start = qat_fw_counters_seq_start, + .next = qat_fw_counters_seq_next, + .stop = qat_fw_counters_seq_stop, + .show = qat_fw_counters_seq_show, +}; + +static int qat_fw_counters_file_open(struct inode *inode, struct file *file) +{ + struct adf_accel_dev *accel_dev = inode->i_private; + struct seq_file *fw_counters_seq_file; + struct adf_fw_counters *fw_counters; + int ret; + + fw_counters = adf_fw_counters_get(accel_dev); + if (IS_ERR(fw_counters)) + return PTR_ERR(fw_counters); + + ret = seq_open(file, &qat_fw_counters_sops); + if (unlikely(ret)) { + kfree(fw_counters); + return ret; + } + + fw_counters_seq_file = file->private_data; + fw_counters_seq_file->private = fw_counters; + return ret; +} + +static int qat_fw_counters_file_release(struct inode *inode, struct file *file) +{ + struct seq_file *seq = file->private_data; + + kfree(seq->private); + seq->private = NULL; + + return seq_release(inode, file); } + +static const struct file_operations qat_fw_counters_fops = { + .owner = THIS_MODULE, + .open = qat_fw_counters_file_open, + .read = seq_read, + .llseek = seq_lseek, + .release = qat_fw_counters_file_release, +}; + +/** + * adf_fw_counters_dbgfs_add() - Create a debugfs file containing FW + * execution counters. + * @accel_dev: Pointer to a QAT acceleration device + * + * Function creates a file to display a table with statistics for the given + * QAT acceleration device. The table stores device specific execution values + * for each AE, such as the number of requests sent to the FW and responses + * received from the FW. + * + * Return: void + */ +void adf_fw_counters_dbgfs_add(struct adf_accel_dev *accel_dev) +{ + accel_dev->fw_cntr_dbgfile = debugfs_create_file("fw_counters", 0400, + accel_dev->debugfs_dir, + accel_dev, + &qat_fw_counters_fops); +} + +/** + * adf_fw_counters_dbgfs_rm() - Remove the debugfs file containing FW counters. + * @accel_dev: Pointer to a QAT acceleration device. + * + * Function removes the file providing the table of statistics for the given + * QAT acceleration device. + * + * Return: void + */ +void adf_fw_counters_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ + debugfs_remove(accel_dev->fw_cntr_dbgfile); + accel_dev->fw_cntr_dbgfile = NULL; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_fw_counters.h b/drivers/crypto/intel/qat/qat_common/adf_fw_counters.h new file mode 100644 index 000000000000..91b3b6a95f1f --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_fw_counters.h @@ -0,0 +1,11 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_FW_COUNTERS_H +#define ADF_FW_COUNTERS_H + +struct adf_accel_dev; + +void adf_fw_counters_dbgfs_add(struct adf_accel_dev *accel_dev); +void adf_fw_counters_dbgfs_rm(struct adf_accel_dev *accel_dev); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_config.c b/drivers/crypto/intel/qat/qat_common/adf_gen2_config.c new file mode 100644 index 000000000000..c27ff6d18e11 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_config.c @@ -0,0 +1,213 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2022 Intel Corporation */ +#include "adf_accel_devices.h" +#include "adf_cfg.h" +#include "adf_cfg_strings.h" +#include "adf_gen2_config.h" +#include "adf_common_drv.h" +#include "qat_crypto.h" +#include "qat_compression.h" +#include "adf_heartbeat.h" +#include "adf_transport_access_macros.h" + +static int adf_gen2_crypto_dev_config(struct adf_accel_dev *accel_dev) +{ + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + int banks = GET_MAX_BANKS(accel_dev); + int cpus = num_online_cpus(); + unsigned long val; + int instances; + int ret; + int i; + + if (adf_hw_dev_has_crypto(accel_dev)) + instances = min(cpus, banks); + else + instances = 0; + + for (i = 0; i < instances; i++) { + val = i; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_BANK_NUM, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_BANK_NUM, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_ETRMGR_CORE_AFFINITY, + i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_SIZE, i); + val = 128; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 512; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_SIZE, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 0; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_TX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 2; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_TX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 8; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_RX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 10; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_RX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = ADF_COALESCING_DEF_TIME; + snprintf(key, sizeof(key), ADF_ETRMGR_COALESCE_TIMER_FORMAT, i); + ret = adf_cfg_add_key_value_param(accel_dev, "Accelerator0", + key, &val, ADF_DEC); + if (ret) + goto err; + } + + val = i; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_CY, + &val, ADF_DEC); + if (ret) + goto err; + + return ret; + +err: + dev_err(&GET_DEV(accel_dev), "Failed to add configuration for crypto\n"); + return ret; +} + +static int adf_gen2_comp_dev_config(struct adf_accel_dev *accel_dev) +{ + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + int banks = GET_MAX_BANKS(accel_dev); + int cpus = num_online_cpus(); + unsigned long val; + int instances; + int ret; + int i; + + if (adf_hw_dev_has_compression(accel_dev)) + instances = min(cpus, banks); + else + instances = 0; + + for (i = 0; i < instances; i++) { + val = i; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_BANK_NUM, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 512; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_SIZE, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 6; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_TX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 14; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_RX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + } + + val = i; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_DC, + &val, ADF_DEC); + if (ret) + return ret; + + return ret; + +err: + dev_err(&GET_DEV(accel_dev), "Failed to add configuration for compression\n"); + return ret; +} + +/** + * adf_gen2_dev_config() - create dev config required to create instances + * + * @accel_dev: Pointer to acceleration device. + * + * Function creates device configuration required to create instances + * + * Return: 0 on success, error code otherwise. + */ +int adf_gen2_dev_config(struct adf_accel_dev *accel_dev) +{ + int ret; + + ret = adf_cfg_section_add(accel_dev, ADF_KERNEL_SEC); + if (ret) + goto err; + + ret = adf_cfg_section_add(accel_dev, "Accelerator0"); + if (ret) + goto err; + + ret = adf_gen2_crypto_dev_config(accel_dev); + if (ret) + goto err; + + ret = adf_gen2_comp_dev_config(accel_dev); + if (ret) + goto err; + + ret = adf_cfg_section_add(accel_dev, ADF_GENERAL_SEC); + if (ret) + goto err; + + adf_heartbeat_save_cfg_param(accel_dev, ADF_CFG_HB_TIMER_DEFAULT_MS); + + set_bit(ADF_STATUS_CONFIGURED, &accel_dev->status); + + return ret; + +err: + dev_err(&GET_DEV(accel_dev), "Failed to configure QAT driver\n"); + return ret; +} +EXPORT_SYMBOL_GPL(adf_gen2_dev_config); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_config.h b/drivers/crypto/intel/qat/qat_common/adf_gen2_config.h new file mode 100644 index 000000000000..4bf9da2de68a --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_config.h @@ -0,0 +1,10 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef ADF_GEN2_CONFIG_H_ +#define ADF_GEN2_CONFIG_H_ + +#include "adf_accel_devices.h" + +int adf_gen2_dev_config(struct adf_accel_dev *accel_dev); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_csr_data.c b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_csr_data.c new file mode 100644 index 000000000000..650c9edd8a66 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_csr_data.c @@ -0,0 +1,101 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2024 Intel Corporation */ +#include <linux/types.h> +#include "adf_gen2_hw_csr_data.h" + +static u64 build_csr_ring_base_addr(dma_addr_t addr, u32 size) +{ + return BUILD_RING_BASE_ADDR(addr, size); +} + +static u32 read_csr_ring_head(void __iomem *csr_base_addr, u32 bank, u32 ring) +{ + return READ_CSR_RING_HEAD(csr_base_addr, bank, ring); +} + +static void write_csr_ring_head(void __iomem *csr_base_addr, u32 bank, u32 ring, + u32 value) +{ + WRITE_CSR_RING_HEAD(csr_base_addr, bank, ring, value); +} + +static u32 read_csr_ring_tail(void __iomem *csr_base_addr, u32 bank, u32 ring) +{ + return READ_CSR_RING_TAIL(csr_base_addr, bank, ring); +} + +static void write_csr_ring_tail(void __iomem *csr_base_addr, u32 bank, u32 ring, + u32 value) +{ + WRITE_CSR_RING_TAIL(csr_base_addr, bank, ring, value); +} + +static u32 read_csr_e_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_E_STAT(csr_base_addr, bank); +} + +static void write_csr_ring_config(void __iomem *csr_base_addr, u32 bank, + u32 ring, u32 value) +{ + WRITE_CSR_RING_CONFIG(csr_base_addr, bank, ring, value); +} + +static void write_csr_ring_base(void __iomem *csr_base_addr, u32 bank, u32 ring, + dma_addr_t addr) +{ + WRITE_CSR_RING_BASE(csr_base_addr, bank, ring, addr); +} + +static void write_csr_int_flag(void __iomem *csr_base_addr, u32 bank, u32 value) +{ + WRITE_CSR_INT_FLAG(csr_base_addr, bank, value); +} + +static void write_csr_int_srcsel(void __iomem *csr_base_addr, u32 bank) +{ + WRITE_CSR_INT_SRCSEL(csr_base_addr, bank); +} + +static void write_csr_int_col_en(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_INT_COL_EN(csr_base_addr, bank, value); +} + +static void write_csr_int_col_ctl(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_INT_COL_CTL(csr_base_addr, bank, value); +} + +static void write_csr_int_flag_and_col(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_INT_FLAG_AND_COL(csr_base_addr, bank, value); +} + +static void write_csr_ring_srv_arb_en(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_RING_SRV_ARB_EN(csr_base_addr, bank, value); +} + +void adf_gen2_init_hw_csr_ops(struct adf_hw_csr_ops *csr_ops) +{ + csr_ops->build_csr_ring_base_addr = build_csr_ring_base_addr; + csr_ops->read_csr_ring_head = read_csr_ring_head; + csr_ops->write_csr_ring_head = write_csr_ring_head; + csr_ops->read_csr_ring_tail = read_csr_ring_tail; + csr_ops->write_csr_ring_tail = write_csr_ring_tail; + csr_ops->read_csr_e_stat = read_csr_e_stat; + csr_ops->write_csr_ring_config = write_csr_ring_config; + csr_ops->write_csr_ring_base = write_csr_ring_base; + csr_ops->write_csr_int_flag = write_csr_int_flag; + csr_ops->write_csr_int_srcsel = write_csr_int_srcsel; + csr_ops->write_csr_int_col_en = write_csr_int_col_en; + csr_ops->write_csr_int_col_ctl = write_csr_int_col_ctl; + csr_ops->write_csr_int_flag_and_col = write_csr_int_flag_and_col; + csr_ops->write_csr_ring_srv_arb_en = write_csr_ring_srv_arb_en; +} +EXPORT_SYMBOL_GPL(adf_gen2_init_hw_csr_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_csr_data.h b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_csr_data.h new file mode 100644 index 000000000000..55058b0f9e52 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_csr_data.h @@ -0,0 +1,86 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2024 Intel Corporation */ +#ifndef ADF_GEN2_HW_CSR_DATA_H_ +#define ADF_GEN2_HW_CSR_DATA_H_ + +#include <linux/bitops.h> +#include "adf_accel_devices.h" + +#define ADF_BANK_INT_SRC_SEL_MASK_0 0x4444444CUL +#define ADF_BANK_INT_SRC_SEL_MASK_X 0x44444444UL +#define ADF_RING_CSR_RING_CONFIG 0x000 +#define ADF_RING_CSR_RING_LBASE 0x040 +#define ADF_RING_CSR_RING_UBASE 0x080 +#define ADF_RING_CSR_RING_HEAD 0x0C0 +#define ADF_RING_CSR_RING_TAIL 0x100 +#define ADF_RING_CSR_E_STAT 0x14C +#define ADF_RING_CSR_INT_FLAG 0x170 +#define ADF_RING_CSR_INT_SRCSEL 0x174 +#define ADF_RING_CSR_INT_SRCSEL_2 0x178 +#define ADF_RING_CSR_INT_COL_EN 0x17C +#define ADF_RING_CSR_INT_COL_CTL 0x180 +#define ADF_RING_CSR_INT_FLAG_AND_COL 0x184 +#define ADF_RING_CSR_INT_COL_CTL_ENABLE 0x80000000 +#define ADF_RING_BUNDLE_SIZE 0x1000 +#define ADF_ARB_REG_SLOT 0x1000 +#define ADF_ARB_RINGSRVARBEN_OFFSET 0x19C + +#define BUILD_RING_BASE_ADDR(addr, size) \ + (((addr) >> 6) & (GENMASK_ULL(63, 0) << (size))) +#define READ_CSR_RING_HEAD(csr_base_addr, bank, ring) \ + ADF_CSR_RD(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_RING_HEAD + ((ring) << 2)) +#define READ_CSR_RING_TAIL(csr_base_addr, bank, ring) \ + ADF_CSR_RD(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_RING_TAIL + ((ring) << 2)) +#define READ_CSR_E_STAT(csr_base_addr, bank) \ + ADF_CSR_RD(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_E_STAT) +#define WRITE_CSR_RING_CONFIG(csr_base_addr, bank, ring, value) \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_RING_CONFIG + ((ring) << 2), value) +#define WRITE_CSR_RING_BASE(csr_base_addr, bank, ring, value) \ +do { \ + u32 l_base = 0, u_base = 0; \ + l_base = (u32)((value) & 0xFFFFFFFF); \ + u_base = (u32)(((value) & 0xFFFFFFFF00000000ULL) >> 32); \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_RING_LBASE + ((ring) << 2), l_base); \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_RING_UBASE + ((ring) << 2), u_base); \ +} while (0) + +#define WRITE_CSR_RING_HEAD(csr_base_addr, bank, ring, value) \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_RING_HEAD + ((ring) << 2), value) +#define WRITE_CSR_RING_TAIL(csr_base_addr, bank, ring, value) \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_RING_TAIL + ((ring) << 2), value) +#define WRITE_CSR_INT_FLAG(csr_base_addr, bank, value) \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_INT_FLAG, value) +#define WRITE_CSR_INT_SRCSEL(csr_base_addr, bank) \ +do { \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_INT_SRCSEL, ADF_BANK_INT_SRC_SEL_MASK_0); \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_INT_SRCSEL_2, ADF_BANK_INT_SRC_SEL_MASK_X); \ +} while (0) +#define WRITE_CSR_INT_COL_EN(csr_base_addr, bank, value) \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_INT_COL_EN, value) +#define WRITE_CSR_INT_COL_CTL(csr_base_addr, bank, value) \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_INT_COL_CTL, \ + ADF_RING_CSR_INT_COL_CTL_ENABLE | (value)) +#define WRITE_CSR_INT_FLAG_AND_COL(csr_base_addr, bank, value) \ + ADF_CSR_WR(csr_base_addr, (ADF_RING_BUNDLE_SIZE * (bank)) + \ + ADF_RING_CSR_INT_FLAG_AND_COL, value) + +#define WRITE_CSR_RING_SRV_ARB_EN(csr_addr, index, value) \ + ADF_CSR_WR(csr_addr, ADF_ARB_RINGSRVARBEN_OFFSET + \ + (ADF_ARB_REG_SLOT * (index)), value) + +void adf_gen2_init_hw_csr_ops(struct adf_hw_csr_ops *csr_ops); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_data.c b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_data.c new file mode 100644 index 000000000000..6a505e9a5cf9 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_data.c @@ -0,0 +1,228 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2020 Intel Corporation */ +#include "adf_common_drv.h" +#include "adf_dc.h" +#include "adf_gen2_hw_data.h" +#include "icp_qat_fw_comp.h" +#include "icp_qat_hw.h" +#include <linux/pci.h> + +u32 adf_gen2_get_num_accels(struct adf_hw_device_data *self) +{ + if (!self || !self->accel_mask) + return 0; + + return hweight16(self->accel_mask); +} +EXPORT_SYMBOL_GPL(adf_gen2_get_num_accels); + +u32 adf_gen2_get_num_aes(struct adf_hw_device_data *self) +{ + if (!self || !self->ae_mask) + return 0; + + return hweight32(self->ae_mask); +} +EXPORT_SYMBOL_GPL(adf_gen2_get_num_aes); + +void adf_gen2_enable_error_correction(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + unsigned long accel_mask = hw_data->accel_mask; + unsigned long ae_mask = hw_data->ae_mask; + unsigned int val, i; + + /* Enable Accel Engine error detection & correction */ + for_each_set_bit(i, &ae_mask, hw_data->num_engines) { + val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_AE_CTX_ENABLES(i)); + val |= ADF_GEN2_ENABLE_AE_ECC_ERR; + ADF_CSR_WR(pmisc_addr, ADF_GEN2_AE_CTX_ENABLES(i), val); + val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_AE_MISC_CONTROL(i)); + val |= ADF_GEN2_ENABLE_AE_ECC_PARITY_CORR; + ADF_CSR_WR(pmisc_addr, ADF_GEN2_AE_MISC_CONTROL(i), val); + } + + /* Enable shared memory error detection & correction */ + for_each_set_bit(i, &accel_mask, hw_data->num_accel) { + val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_UERRSSMSH(i)); + val |= ADF_GEN2_ERRSSMSH_EN; + ADF_CSR_WR(pmisc_addr, ADF_GEN2_UERRSSMSH(i), val); + val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_CERRSSMSH(i)); + val |= ADF_GEN2_ERRSSMSH_EN; + ADF_CSR_WR(pmisc_addr, ADF_GEN2_CERRSSMSH(i), val); + } +} +EXPORT_SYMBOL_GPL(adf_gen2_enable_error_correction); + +void adf_gen2_cfg_iov_thds(struct adf_accel_dev *accel_dev, bool enable, + int num_a_regs, int num_b_regs) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u32 reg; + int i; + + /* Set/Unset Valid bit in AE Thread to PCIe Function Mapping Group A */ + for (i = 0; i < num_a_regs; i++) { + reg = READ_CSR_AE2FUNCTION_MAP_A(pmisc_addr, i); + if (enable) + reg |= AE2FUNCTION_MAP_VALID; + else + reg &= ~AE2FUNCTION_MAP_VALID; + WRITE_CSR_AE2FUNCTION_MAP_A(pmisc_addr, i, reg); + } + + /* Set/Unset Valid bit in AE Thread to PCIe Function Mapping Group B */ + for (i = 0; i < num_b_regs; i++) { + reg = READ_CSR_AE2FUNCTION_MAP_B(pmisc_addr, i); + if (enable) + reg |= AE2FUNCTION_MAP_VALID; + else + reg &= ~AE2FUNCTION_MAP_VALID; + WRITE_CSR_AE2FUNCTION_MAP_B(pmisc_addr, i, reg); + } +} +EXPORT_SYMBOL_GPL(adf_gen2_cfg_iov_thds); + +void adf_gen2_get_admin_info(struct admin_info *admin_csrs_info) +{ + admin_csrs_info->mailbox_offset = ADF_MAILBOX_BASE_OFFSET; + admin_csrs_info->admin_msg_ur = ADF_ADMINMSGUR_OFFSET; + admin_csrs_info->admin_msg_lr = ADF_ADMINMSGLR_OFFSET; +} +EXPORT_SYMBOL_GPL(adf_gen2_get_admin_info); + +void adf_gen2_get_arb_info(struct arb_info *arb_info) +{ + arb_info->arb_cfg = ADF_ARB_CONFIG; + arb_info->arb_offset = ADF_ARB_OFFSET; + arb_info->wt2sam_offset = ADF_ARB_WRK_2_SER_MAP_OFFSET; +} +EXPORT_SYMBOL_GPL(adf_gen2_get_arb_info); + +void adf_gen2_enable_ints(struct adf_accel_dev *accel_dev) +{ + void __iomem *addr = adf_get_pmisc_base(accel_dev); + u32 val; + + val = accel_dev->pf.vf_info ? 0 : BIT_ULL(GET_MAX_BANKS(accel_dev)) - 1; + + /* Enable bundle and misc interrupts */ + ADF_CSR_WR(addr, ADF_GEN2_SMIAPF0_MASK_OFFSET, val); + ADF_CSR_WR(addr, ADF_GEN2_SMIAPF1_MASK_OFFSET, ADF_GEN2_SMIA1_MASK); +} +EXPORT_SYMBOL_GPL(adf_gen2_enable_ints); + +u32 adf_gen2_get_accel_cap(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct pci_dev *pdev = accel_dev->accel_pci_dev.pci_dev; + u32 fuses = hw_data->fuses[ADF_FUSECTL0]; + u32 straps = hw_data->straps; + u32 legfuses; + u32 capabilities = ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC | + ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC | + ICP_ACCEL_CAPABILITIES_AUTHENTICATION | + ICP_ACCEL_CAPABILITIES_CIPHER | + ICP_ACCEL_CAPABILITIES_COMPRESSION; + + /* Read accelerator capabilities mask */ + pci_read_config_dword(pdev, ADF_DEVICE_LEGFUSE_OFFSET, &legfuses); + + /* A set bit in legfuses means the feature is OFF in this SKU */ + if (legfuses & ICP_ACCEL_MASK_CIPHER_SLICE) { + capabilities &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + capabilities &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + if (legfuses & ICP_ACCEL_MASK_PKE_SLICE) + capabilities &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC; + if (legfuses & ICP_ACCEL_MASK_AUTH_SLICE) { + capabilities &= ~ICP_ACCEL_CAPABILITIES_AUTHENTICATION; + capabilities &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + if (legfuses & ICP_ACCEL_MASK_COMPRESS_SLICE) + capabilities &= ~ICP_ACCEL_CAPABILITIES_COMPRESSION; + + if ((straps | fuses) & ADF_POWERGATE_PKE) + capabilities &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC; + + if ((straps | fuses) & ADF_POWERGATE_DC) + capabilities &= ~ICP_ACCEL_CAPABILITIES_COMPRESSION; + + return capabilities; +} +EXPORT_SYMBOL_GPL(adf_gen2_get_accel_cap); + +void adf_gen2_set_ssm_wdtimer(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u32 timer_val_pke = ADF_SSM_WDT_PKE_DEFAULT_VALUE; + u32 timer_val = ADF_SSM_WDT_DEFAULT_VALUE; + unsigned long accel_mask = hw_data->accel_mask; + u32 i = 0; + + /* Configures WDT timers */ + for_each_set_bit(i, &accel_mask, hw_data->num_accel) { + /* Enable WDT for sym and dc */ + ADF_CSR_WR(pmisc_addr, ADF_SSMWDT(i), timer_val); + /* Enable WDT for pke */ + ADF_CSR_WR(pmisc_addr, ADF_SSMWDTPKE(i), timer_val_pke); + } +} +EXPORT_SYMBOL_GPL(adf_gen2_set_ssm_wdtimer); + +static int adf_gen2_build_comp_block(void *ctx, enum adf_dc_algo algo) +{ + struct icp_qat_fw_comp_req *req_tmpl = ctx; + struct icp_qat_fw_comp_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + + switch (algo) { + case QAT_DEFLATE: + header->service_cmd_id = ICP_QAT_FW_COMP_CMD_STATIC; + break; + default: + return -EINVAL; + } + + cd_pars->u.sl.comp_slice_cfg_word[0] = + ICP_QAT_HW_COMPRESSION_CONFIG_BUILD(ICP_QAT_HW_COMPRESSION_DIR_COMPRESS, + ICP_QAT_HW_COMPRESSION_DELAYED_MATCH_DISABLED, + ICP_QAT_HW_COMPRESSION_ALGO_DEFLATE, + ICP_QAT_HW_COMPRESSION_DEPTH_1, + ICP_QAT_HW_COMPRESSION_FILE_TYPE_0); + + return 0; +} + +static int adf_gen2_build_decomp_block(void *ctx, enum adf_dc_algo algo) +{ + struct icp_qat_fw_comp_req *req_tmpl = ctx; + struct icp_qat_fw_comp_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + + switch (algo) { + case QAT_DEFLATE: + header->service_cmd_id = ICP_QAT_FW_COMP_CMD_DECOMPRESS; + break; + default: + return -EINVAL; + } + + cd_pars->u.sl.comp_slice_cfg_word[0] = + ICP_QAT_HW_COMPRESSION_CONFIG_BUILD(ICP_QAT_HW_COMPRESSION_DIR_DECOMPRESS, + ICP_QAT_HW_COMPRESSION_DELAYED_MATCH_DISABLED, + ICP_QAT_HW_COMPRESSION_ALGO_DEFLATE, + ICP_QAT_HW_COMPRESSION_DEPTH_1, + ICP_QAT_HW_COMPRESSION_FILE_TYPE_0); + + return 0; +} + +void adf_gen2_init_dc_ops(struct adf_dc_ops *dc_ops) +{ + dc_ops->build_comp_block = adf_gen2_build_comp_block; + dc_ops->build_decomp_block = adf_gen2_build_decomp_block; +} +EXPORT_SYMBOL_GPL(adf_gen2_init_dc_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_data.h b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_data.h new file mode 100644 index 000000000000..59bad368a921 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_hw_data.h @@ -0,0 +1,93 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2020 Intel Corporation */ +#ifndef ADF_GEN2_HW_DATA_H_ +#define ADF_GEN2_HW_DATA_H_ + +#include "adf_accel_devices.h" +#include "adf_cfg_common.h" + +#define ADF_GEN2_RX_RINGS_OFFSET 8 +#define ADF_GEN2_TX_RINGS_MASK 0xFF + +/* AE to function map */ +#define AE2FUNCTION_MAP_A_OFFSET (0x3A400 + 0x190) +#define AE2FUNCTION_MAP_B_OFFSET (0x3A400 + 0x310) +#define AE2FUNCTION_MAP_REG_SIZE 4 +#define AE2FUNCTION_MAP_VALID BIT(7) + +#define READ_CSR_AE2FUNCTION_MAP_A(pmisc_bar_addr, index) \ + ADF_CSR_RD(pmisc_bar_addr, AE2FUNCTION_MAP_A_OFFSET + \ + AE2FUNCTION_MAP_REG_SIZE * (index)) +#define WRITE_CSR_AE2FUNCTION_MAP_A(pmisc_bar_addr, index, value) \ + ADF_CSR_WR(pmisc_bar_addr, AE2FUNCTION_MAP_A_OFFSET + \ + AE2FUNCTION_MAP_REG_SIZE * (index), value) +#define READ_CSR_AE2FUNCTION_MAP_B(pmisc_bar_addr, index) \ + ADF_CSR_RD(pmisc_bar_addr, AE2FUNCTION_MAP_B_OFFSET + \ + AE2FUNCTION_MAP_REG_SIZE * (index)) +#define WRITE_CSR_AE2FUNCTION_MAP_B(pmisc_bar_addr, index, value) \ + ADF_CSR_WR(pmisc_bar_addr, AE2FUNCTION_MAP_B_OFFSET + \ + AE2FUNCTION_MAP_REG_SIZE * (index), value) + +/* Admin Interface Offsets */ +#define ADF_ADMINMSGUR_OFFSET (0x3A000 + 0x574) +#define ADF_ADMINMSGLR_OFFSET (0x3A000 + 0x578) +#define ADF_MAILBOX_BASE_OFFSET 0x20970 + +/* Arbiter configuration */ +#define ADF_ARB_OFFSET 0x30000 +#define ADF_ARB_WRK_2_SER_MAP_OFFSET 0x180 +#define ADF_ARB_CONFIG (BIT(31) | BIT(6) | BIT(0)) + +/* Power gating */ +#define ADF_POWERGATE_DC BIT(23) +#define ADF_POWERGATE_PKE BIT(24) + +/* Default ring mapping */ +#define ADF_GEN2_DEFAULT_RING_TO_SRV_MAP \ + (CRYPTO << ADF_CFG_SERV_RING_PAIR_0_SHIFT | \ + CRYPTO << ADF_CFG_SERV_RING_PAIR_1_SHIFT | \ + UNUSED << ADF_CFG_SERV_RING_PAIR_2_SHIFT | \ + COMP << ADF_CFG_SERV_RING_PAIR_3_SHIFT) + +/* WDT timers + * + * Timeout is in cycles. Clock speed may vary across products but this + * value should be a few milli-seconds. + */ +#define ADF_SSM_WDT_DEFAULT_VALUE 0x200000 +#define ADF_SSM_WDT_PKE_DEFAULT_VALUE 0x2000000 +#define ADF_SSMWDT_OFFSET 0x54 +#define ADF_SSMWDTPKE_OFFSET 0x58 +#define ADF_SSMWDT(i) (ADF_SSMWDT_OFFSET + ((i) * 0x4000)) +#define ADF_SSMWDTPKE(i) (ADF_SSMWDTPKE_OFFSET + ((i) * 0x4000)) + +/* Error detection and correction */ +#define ADF_GEN2_AE_CTX_ENABLES(i) ((i) * 0x1000 + 0x20818) +#define ADF_GEN2_AE_MISC_CONTROL(i) ((i) * 0x1000 + 0x20960) +#define ADF_GEN2_ENABLE_AE_ECC_ERR BIT(28) +#define ADF_GEN2_ENABLE_AE_ECC_PARITY_CORR (BIT(24) | BIT(12)) +#define ADF_GEN2_UERRSSMSH(i) ((i) * 0x4000 + 0x18) +#define ADF_GEN2_CERRSSMSH(i) ((i) * 0x4000 + 0x10) +#define ADF_GEN2_ERRSSMSH_EN BIT(3) + +/* Number of heartbeat counter pairs */ +#define ADF_NUM_HB_CNT_PER_AE ADF_NUM_THREADS_PER_AE + +/* Interrupts */ +#define ADF_GEN2_SMIAPF0_MASK_OFFSET (0x3A000 + 0x28) +#define ADF_GEN2_SMIAPF1_MASK_OFFSET (0x3A000 + 0x30) +#define ADF_GEN2_SMIA1_MASK 0x1 + +u32 adf_gen2_get_num_accels(struct adf_hw_device_data *self); +u32 adf_gen2_get_num_aes(struct adf_hw_device_data *self); +void adf_gen2_enable_error_correction(struct adf_accel_dev *accel_dev); +void adf_gen2_cfg_iov_thds(struct adf_accel_dev *accel_dev, bool enable, + int num_a_regs, int num_b_regs); +void adf_gen2_get_admin_info(struct admin_info *admin_csrs_info); +void adf_gen2_get_arb_info(struct arb_info *arb_info); +void adf_gen2_enable_ints(struct adf_accel_dev *accel_dev); +u32 adf_gen2_get_accel_cap(struct adf_accel_dev *accel_dev); +void adf_gen2_set_ssm_wdtimer(struct adf_accel_dev *accel_dev); +void adf_gen2_init_dc_ops(struct adf_dc_ops *dc_ops); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_pfvf.c b/drivers/crypto/intel/qat/qat_common/adf_gen2_pfvf.c new file mode 100644 index 000000000000..43af81fcab86 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_pfvf.c @@ -0,0 +1,401 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2021 Intel Corporation */ +#include <linux/delay.h> +#include <linux/iopoll.h> +#include <linux/mutex.h> +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_gen2_pfvf.h" +#include "adf_pfvf_msg.h" +#include "adf_pfvf_pf_proto.h" +#include "adf_pfvf_vf_proto.h" +#include "adf_pfvf_utils.h" + + /* VF2PF interrupts */ +#define ADF_GEN2_VF_MSK 0xFFFF +#define ADF_GEN2_ERR_REG_VF2PF(vf_src) (((vf_src) & 0x01FFFE00) >> 9) +#define ADF_GEN2_ERR_MSK_VF2PF(vf_mask) (((vf_mask) & ADF_GEN2_VF_MSK) << 9) + +#define ADF_GEN2_PF_PF2VF_OFFSET(i) (0x3A000 + 0x280 + ((i) * 0x04)) +#define ADF_GEN2_VF_PF2VF_OFFSET 0x200 + +#define ADF_GEN2_CSR_IN_USE 0x6AC2 +#define ADF_GEN2_CSR_IN_USE_MASK 0xFFFE + +enum gen2_csr_pos { + ADF_GEN2_CSR_PF2VF_OFFSET = 0, + ADF_GEN2_CSR_VF2PF_OFFSET = 16, +}; + +#define ADF_PFVF_GEN2_MSGTYPE_SHIFT 2 +#define ADF_PFVF_GEN2_MSGTYPE_MASK 0x0F +#define ADF_PFVF_GEN2_MSGDATA_SHIFT 6 +#define ADF_PFVF_GEN2_MSGDATA_MASK 0x3FF + +static const struct pfvf_csr_format csr_gen2_fmt = { + { ADF_PFVF_GEN2_MSGTYPE_SHIFT, ADF_PFVF_GEN2_MSGTYPE_MASK }, + { ADF_PFVF_GEN2_MSGDATA_SHIFT, ADF_PFVF_GEN2_MSGDATA_MASK }, +}; + +#define ADF_PFVF_MSG_RETRY_DELAY 5 +#define ADF_PFVF_MSG_MAX_RETRIES 3 + +static u32 adf_gen2_pf_get_pfvf_offset(u32 i) +{ + return ADF_GEN2_PF_PF2VF_OFFSET(i); +} + +static u32 adf_gen2_vf_get_pfvf_offset(u32 i) +{ + return ADF_GEN2_VF_PF2VF_OFFSET; +} + +static void adf_gen2_enable_vf2pf_interrupts(void __iomem *pmisc_addr, u32 vf_mask) +{ + /* Enable VF2PF Messaging Ints - VFs 0 through 15 per vf_mask[15:0] */ + if (vf_mask & ADF_GEN2_VF_MSK) { + u32 val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK3) + & ~ADF_GEN2_ERR_MSK_VF2PF(vf_mask); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, val); + } +} + +static void adf_gen2_disable_all_vf2pf_interrupts(void __iomem *pmisc_addr) +{ + /* Disable VF2PF interrupts for VFs 0 through 15 per vf_mask[15:0] */ + u32 val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK3) + | ADF_GEN2_ERR_MSK_VF2PF(ADF_GEN2_VF_MSK); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, val); +} + +static u32 adf_gen2_disable_pending_vf2pf_interrupts(void __iomem *pmisc_addr) +{ + u32 sources, disabled, pending; + u32 errsou3, errmsk3; + + /* Get the interrupt sources triggered by VFs */ + errsou3 = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRSOU3); + sources = ADF_GEN2_ERR_REG_VF2PF(errsou3); + + if (!sources) + return 0; + + /* Get the already disabled interrupts */ + errmsk3 = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK3); + disabled = ADF_GEN2_ERR_REG_VF2PF(errmsk3); + + pending = sources & ~disabled; + if (!pending) + return 0; + + /* Due to HW limitations, when disabling the interrupts, we can't + * just disable the requested sources, as this would lead to missed + * interrupts if ERRSOU3 changes just before writing to ERRMSK3. + * To work around it, disable all and re-enable only the sources that + * are not in vf_mask and were not already disabled. Re-enabling will + * trigger a new interrupt for the sources that have changed in the + * meantime, if any. + */ + errmsk3 |= ADF_GEN2_ERR_MSK_VF2PF(ADF_GEN2_VF_MSK); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, errmsk3); + + /* Update only section of errmsk3 related to VF2PF */ + errmsk3 &= ~ADF_GEN2_ERR_MSK_VF2PF(ADF_GEN2_VF_MSK); + errmsk3 |= ADF_GEN2_ERR_MSK_VF2PF(sources | disabled); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, errmsk3); + + /* Return the sources of the (new) interrupt(s) */ + return pending; +} + +static u32 gen2_csr_get_int_bit(enum gen2_csr_pos offset) +{ + return ADF_PFVF_INT << offset; +} + +static u32 gen2_csr_msg_to_position(u32 csr_msg, enum gen2_csr_pos offset) +{ + return (csr_msg & 0xFFFF) << offset; +} + +static u32 gen2_csr_msg_from_position(u32 csr_val, enum gen2_csr_pos offset) +{ + return (csr_val >> offset) & 0xFFFF; +} + +static bool gen2_csr_is_in_use(u32 msg, enum gen2_csr_pos offset) +{ + return ((msg >> offset) & ADF_GEN2_CSR_IN_USE_MASK) == ADF_GEN2_CSR_IN_USE; +} + +static void gen2_csr_clear_in_use(u32 *msg, enum gen2_csr_pos offset) +{ + *msg &= ~(ADF_GEN2_CSR_IN_USE_MASK << offset); +} + +static void gen2_csr_set_in_use(u32 *msg, enum gen2_csr_pos offset) +{ + *msg |= (ADF_GEN2_CSR_IN_USE << offset); +} + +static bool is_legacy_user_pfvf_message(u32 msg) +{ + return !(msg & ADF_PFVF_MSGORIGIN_SYSTEM); +} + +static bool is_pf2vf_notification(u8 msg_type) +{ + switch (msg_type) { + case ADF_PF2VF_MSGTYPE_RESTARTING: + return true; + default: + return false; + } +} + +static bool is_vf2pf_notification(u8 msg_type) +{ + switch (msg_type) { + case ADF_VF2PF_MSGTYPE_INIT: + case ADF_VF2PF_MSGTYPE_SHUTDOWN: + return true; + default: + return false; + } +} + +struct pfvf_gen2_params { + u32 pfvf_offset; + struct mutex *csr_lock; /* lock preventing concurrent access of CSR */ + enum gen2_csr_pos local_offset; + enum gen2_csr_pos remote_offset; + bool (*is_notification_message)(u8 msg_type); + u8 compat_ver; +}; + +static int adf_gen2_pfvf_send(struct adf_accel_dev *accel_dev, + struct pfvf_message msg, + struct pfvf_gen2_params *params) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + enum gen2_csr_pos remote_offset = params->remote_offset; + enum gen2_csr_pos local_offset = params->local_offset; + unsigned int retries = ADF_PFVF_MSG_MAX_RETRIES; + struct mutex *lock = params->csr_lock; + u32 pfvf_offset = params->pfvf_offset; + u32 int_bit; + u32 csr_val; + u32 csr_msg; + int ret; + + /* Gen2 messages, both PF->VF and VF->PF, are all 16 bits long. This + * allows us to build and read messages as if they where all 0 based. + * However, send and receive are in a single shared 32 bits register, + * so we need to shift and/or mask the message half before decoding + * it and after encoding it. Which one to shift depends on the + * direction. + */ + + int_bit = gen2_csr_get_int_bit(local_offset); + + csr_msg = adf_pfvf_csr_msg_of(accel_dev, msg, &csr_gen2_fmt); + if (unlikely(!csr_msg)) + return -EINVAL; + + /* Prepare for CSR format, shifting the wire message in place and + * setting the in use pattern + */ + csr_msg = gen2_csr_msg_to_position(csr_msg, local_offset); + gen2_csr_set_in_use(&csr_msg, remote_offset); + + mutex_lock(lock); + +start: + /* Check if the PFVF CSR is in use by remote function */ + csr_val = ADF_CSR_RD(pmisc_addr, pfvf_offset); + if (gen2_csr_is_in_use(csr_val, local_offset)) { + dev_dbg(&GET_DEV(accel_dev), + "PFVF CSR in use by remote function\n"); + goto retry; + } + + /* Attempt to get ownership of the PFVF CSR */ + ADF_CSR_WR(pmisc_addr, pfvf_offset, csr_msg | int_bit); + + /* Wait for confirmation from remote func it received the message */ + ret = read_poll_timeout(ADF_CSR_RD, csr_val, !(csr_val & int_bit), + ADF_PFVF_MSG_ACK_DELAY_US, + ADF_PFVF_MSG_ACK_MAX_DELAY_US, + true, pmisc_addr, pfvf_offset); + if (unlikely(ret < 0)) { + dev_dbg(&GET_DEV(accel_dev), "ACK not received from remote\n"); + csr_val &= ~int_bit; + } + + /* For fire-and-forget notifications, the receiver does not clear + * the in-use pattern. This is used to detect collisions. + */ + if (params->is_notification_message(msg.type) && csr_val != csr_msg) { + /* Collision must have overwritten the message */ + dev_err(&GET_DEV(accel_dev), + "Collision on notification - PFVF CSR overwritten by remote function\n"); + goto retry; + } + + /* If the far side did not clear the in-use pattern it is either + * 1) Notification - message left intact to detect collision + * 2) Older protocol (compatibility version < 3) on the far side + * where the sender is responsible for clearing the in-use + * pattern after the received has acknowledged receipt. + * In either case, clear the in-use pattern now. + */ + if (gen2_csr_is_in_use(csr_val, remote_offset)) { + gen2_csr_clear_in_use(&csr_val, remote_offset); + ADF_CSR_WR(pmisc_addr, pfvf_offset, csr_val); + } + +out: + mutex_unlock(lock); + return ret; + +retry: + if (--retries) { + msleep(ADF_PFVF_MSG_RETRY_DELAY); + goto start; + } else { + ret = -EBUSY; + goto out; + } +} + +static struct pfvf_message adf_gen2_pfvf_recv(struct adf_accel_dev *accel_dev, + struct pfvf_gen2_params *params) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + enum gen2_csr_pos remote_offset = params->remote_offset; + enum gen2_csr_pos local_offset = params->local_offset; + u32 pfvf_offset = params->pfvf_offset; + struct pfvf_message msg = { 0 }; + u32 int_bit; + u32 csr_val; + u16 csr_msg; + + int_bit = gen2_csr_get_int_bit(local_offset); + + /* Read message */ + csr_val = ADF_CSR_RD(pmisc_addr, pfvf_offset); + if (!(csr_val & int_bit)) { + dev_info(&GET_DEV(accel_dev), + "Spurious PFVF interrupt, msg 0x%.8x. Ignored\n", csr_val); + return msg; + } + + /* Extract the message from the CSR */ + csr_msg = gen2_csr_msg_from_position(csr_val, local_offset); + + /* Ignore legacy non-system (non-kernel) messages */ + if (unlikely(is_legacy_user_pfvf_message(csr_msg))) { + dev_dbg(&GET_DEV(accel_dev), + "Ignored non-system message (0x%.8x);\n", csr_val); + /* Because this must be a legacy message, the far side + * must clear the in-use pattern, so don't do it. + */ + return msg; + } + + /* Return the pfvf_message format */ + msg = adf_pfvf_message_of(accel_dev, csr_msg, &csr_gen2_fmt); + + /* The in-use pattern is not cleared for notifications (so that + * it can be used for collision detection) or older implementations + */ + if (params->compat_ver >= ADF_PFVF_COMPAT_FAST_ACK && + !params->is_notification_message(msg.type)) + gen2_csr_clear_in_use(&csr_val, remote_offset); + + /* To ACK, clear the INT bit */ + csr_val &= ~int_bit; + ADF_CSR_WR(pmisc_addr, pfvf_offset, csr_val); + + return msg; +} + +static int adf_gen2_pf2vf_send(struct adf_accel_dev *accel_dev, struct pfvf_message msg, + u32 pfvf_offset, struct mutex *csr_lock) +{ + struct pfvf_gen2_params params = { + .csr_lock = csr_lock, + .pfvf_offset = pfvf_offset, + .local_offset = ADF_GEN2_CSR_PF2VF_OFFSET, + .remote_offset = ADF_GEN2_CSR_VF2PF_OFFSET, + .is_notification_message = is_pf2vf_notification, + }; + + return adf_gen2_pfvf_send(accel_dev, msg, ¶ms); +} + +static int adf_gen2_vf2pf_send(struct adf_accel_dev *accel_dev, struct pfvf_message msg, + u32 pfvf_offset, struct mutex *csr_lock) +{ + struct pfvf_gen2_params params = { + .csr_lock = csr_lock, + .pfvf_offset = pfvf_offset, + .local_offset = ADF_GEN2_CSR_VF2PF_OFFSET, + .remote_offset = ADF_GEN2_CSR_PF2VF_OFFSET, + .is_notification_message = is_vf2pf_notification, + }; + + return adf_gen2_pfvf_send(accel_dev, msg, ¶ms); +} + +static struct pfvf_message adf_gen2_pf2vf_recv(struct adf_accel_dev *accel_dev, + u32 pfvf_offset, u8 compat_ver) +{ + struct pfvf_gen2_params params = { + .pfvf_offset = pfvf_offset, + .local_offset = ADF_GEN2_CSR_PF2VF_OFFSET, + .remote_offset = ADF_GEN2_CSR_VF2PF_OFFSET, + .is_notification_message = is_pf2vf_notification, + .compat_ver = compat_ver, + }; + + return adf_gen2_pfvf_recv(accel_dev, ¶ms); +} + +static struct pfvf_message adf_gen2_vf2pf_recv(struct adf_accel_dev *accel_dev, + u32 pfvf_offset, u8 compat_ver) +{ + struct pfvf_gen2_params params = { + .pfvf_offset = pfvf_offset, + .local_offset = ADF_GEN2_CSR_VF2PF_OFFSET, + .remote_offset = ADF_GEN2_CSR_PF2VF_OFFSET, + .is_notification_message = is_vf2pf_notification, + .compat_ver = compat_ver, + }; + + return adf_gen2_pfvf_recv(accel_dev, ¶ms); +} + +void adf_gen2_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops) +{ + pfvf_ops->enable_comms = adf_enable_pf2vf_comms; + pfvf_ops->get_pf2vf_offset = adf_gen2_pf_get_pfvf_offset; + pfvf_ops->get_vf2pf_offset = adf_gen2_pf_get_pfvf_offset; + pfvf_ops->enable_vf2pf_interrupts = adf_gen2_enable_vf2pf_interrupts; + pfvf_ops->disable_all_vf2pf_interrupts = adf_gen2_disable_all_vf2pf_interrupts; + pfvf_ops->disable_pending_vf2pf_interrupts = adf_gen2_disable_pending_vf2pf_interrupts; + pfvf_ops->send_msg = adf_gen2_pf2vf_send; + pfvf_ops->recv_msg = adf_gen2_vf2pf_recv; +} +EXPORT_SYMBOL_GPL(adf_gen2_init_pf_pfvf_ops); + +void adf_gen2_init_vf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops) +{ + pfvf_ops->enable_comms = adf_enable_vf2pf_comms; + pfvf_ops->get_pf2vf_offset = adf_gen2_vf_get_pfvf_offset; + pfvf_ops->get_vf2pf_offset = adf_gen2_vf_get_pfvf_offset; + pfvf_ops->send_msg = adf_gen2_vf2pf_send; + pfvf_ops->recv_msg = adf_gen2_pf2vf_recv; +} +EXPORT_SYMBOL_GPL(adf_gen2_init_vf_pfvf_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen2_pfvf.h b/drivers/crypto/intel/qat/qat_common/adf_gen2_pfvf.h new file mode 100644 index 000000000000..34a63cf40db2 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen2_pfvf.h @@ -0,0 +1,30 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2021 Intel Corporation */ +#ifndef ADF_GEN2_PFVF_H +#define ADF_GEN2_PFVF_H + +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" + +#define ADF_GEN2_ERRSOU3 (0x3A000 + 0x0C) +#define ADF_GEN2_ERRSOU5 (0x3A000 + 0xD8) +#define ADF_GEN2_ERRMSK3 (0x3A000 + 0x1C) +#define ADF_GEN2_ERRMSK5 (0x3A000 + 0xDC) + +#if defined(CONFIG_PCI_IOV) +void adf_gen2_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops); +void adf_gen2_init_vf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops); +#else +static inline void adf_gen2_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops) +{ + pfvf_ops->enable_comms = adf_pfvf_comms_disabled; +} + +static inline void adf_gen2_init_vf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops) +{ + pfvf_ops->enable_comms = adf_pfvf_comms_disabled; +} +#endif + +#endif /* ADF_GEN2_PFVF_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_config.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_config.c new file mode 100644 index 000000000000..afcdfdd0a37a --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_config.c @@ -0,0 +1,276 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include "adf_accel_devices.h" +#include "adf_cfg.h" +#include "adf_cfg_services.h" +#include "adf_cfg_strings.h" +#include "adf_common_drv.h" +#include "adf_gen4_config.h" +#include "adf_heartbeat.h" +#include "adf_transport_access_macros.h" +#include "qat_compression.h" +#include "qat_crypto.h" + +int adf_crypto_dev_config(struct adf_accel_dev *accel_dev) +{ + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + int banks = GET_MAX_BANKS(accel_dev); + int cpus = num_online_cpus(); + unsigned long bank, val; + int instances; + int ret; + int i; + + if (adf_hw_dev_has_crypto(accel_dev)) + instances = min(cpus, banks / 2); + else + instances = 0; + + for (i = 0; i < instances; i++) { + val = i; + bank = i * 2; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_BANK_NUM, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &bank, ADF_DEC); + if (ret) + goto err; + + bank += 1; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_BANK_NUM, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &bank, ADF_DEC); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_ETRMGR_CORE_AFFINITY, + i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_SIZE, i); + val = 128; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 512; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_SIZE, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 0; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_TX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 0; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_TX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 1; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_RX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 1; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_RX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = ADF_COALESCING_DEF_TIME; + snprintf(key, sizeof(key), ADF_ETRMGR_COALESCE_TIMER_FORMAT, i); + ret = adf_cfg_add_key_value_param(accel_dev, "Accelerator0", + key, &val, ADF_DEC); + if (ret) + goto err; + } + + val = i; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_CY, + &val, ADF_DEC); + if (ret) + goto err; + + val = 0; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_DC, + &val, ADF_DEC); + if (ret) + goto err; + + return 0; +err: + dev_err(&GET_DEV(accel_dev), "Failed to add configuration for crypto\n"); + return ret; +} + +int adf_comp_dev_config(struct adf_accel_dev *accel_dev) +{ + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + int banks = GET_MAX_BANKS(accel_dev); + int cpus = num_online_cpus(); + unsigned long val; + int instances; + int ret; + int i; + + if (adf_hw_dev_has_compression(accel_dev)) + instances = min(cpus, banks); + else + instances = 0; + + for (i = 0; i < instances; i++) { + val = i; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_BANK_NUM, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 512; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_SIZE, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 0; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_TX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = 1; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_RX, i); + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, + key, &val, ADF_DEC); + if (ret) + goto err; + + val = ADF_COALESCING_DEF_TIME; + snprintf(key, sizeof(key), ADF_ETRMGR_COALESCE_TIMER_FORMAT, i); + ret = adf_cfg_add_key_value_param(accel_dev, "Accelerator0", + key, &val, ADF_DEC); + if (ret) + goto err; + } + + val = i; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_DC, + &val, ADF_DEC); + if (ret) + goto err; + + val = 0; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_CY, + &val, ADF_DEC); + if (ret) + goto err; + + return 0; +err: + dev_err(&GET_DEV(accel_dev), "Failed to add configuration for compression\n"); + return ret; +} + +int adf_no_dev_config(struct adf_accel_dev *accel_dev) +{ + unsigned long val; + int ret; + + val = 0; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_DC, + &val, ADF_DEC); + if (ret) + return ret; + + return adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_CY, + &val, ADF_DEC); +} + +/** + * adf_gen4_dev_config() - create dev config required to create instances + * + * @accel_dev: Pointer to acceleration device. + * + * Function creates device configuration required to create instances + * + * Return: 0 on success, error code otherwise. + */ +int adf_gen4_dev_config(struct adf_accel_dev *accel_dev) +{ + int ret; + + ret = adf_cfg_section_add(accel_dev, ADF_KERNEL_SEC); + if (ret) + goto err; + + ret = adf_cfg_section_add(accel_dev, "Accelerator0"); + if (ret) + goto err; + + switch (adf_get_service_enabled(accel_dev)) { + case SVC_SYM_ASYM: + ret = adf_crypto_dev_config(accel_dev); + break; + case SVC_DC: + case SVC_DCC: + ret = adf_comp_dev_config(accel_dev); + break; + default: + ret = adf_no_dev_config(accel_dev); + break; + } + + if (ret) + goto err; + + set_bit(ADF_STATUS_CONFIGURED, &accel_dev->status); + + return ret; + +err: + dev_err(&GET_DEV(accel_dev), "Failed to configure QAT driver\n"); + return ret; +} +EXPORT_SYMBOL_GPL(adf_gen4_dev_config); + +int adf_gen4_cfg_dev_init(struct adf_accel_dev *accel_dev) +{ + const char *config; + int ret; + + config = accel_dev->accel_id % 2 ? ADF_CFG_DC : ADF_CFG_CY; + + ret = adf_cfg_section_add(accel_dev, ADF_GENERAL_SEC); + if (ret) + return ret; + + /* Default configuration is crypto only for even devices + * and compression for odd devices + */ + ret = adf_cfg_add_key_value_param(accel_dev, ADF_GENERAL_SEC, + ADF_SERVICES_ENABLED, config, + ADF_STR); + if (ret) + return ret; + + adf_heartbeat_save_cfg_param(accel_dev, ADF_CFG_HB_TIMER_MIN_MS); + + return 0; +} +EXPORT_SYMBOL_GPL(adf_gen4_cfg_dev_init); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_config.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_config.h new file mode 100644 index 000000000000..38a674c27e40 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_config.h @@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_GEN4_CONFIG_H_ +#define ADF_GEN4_CONFIG_H_ + +#include "adf_accel_devices.h" + +int adf_gen4_dev_config(struct adf_accel_dev *accel_dev); +int adf_gen4_cfg_dev_init(struct adf_accel_dev *accel_dev); +int adf_crypto_dev_config(struct adf_accel_dev *accel_dev); +int adf_comp_dev_config(struct adf_accel_dev *accel_dev); +int adf_no_dev_config(struct adf_accel_dev *accel_dev); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_csr_data.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_csr_data.c new file mode 100644 index 000000000000..6609c248aaba --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_csr_data.c @@ -0,0 +1,231 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2024 Intel Corporation */ +#include <linux/types.h> +#include "adf_gen4_hw_csr_data.h" + +static u64 build_csr_ring_base_addr(dma_addr_t addr, u32 size) +{ + return BUILD_RING_BASE_ADDR(addr, size); +} + +static u32 read_csr_ring_head(void __iomem *csr_base_addr, u32 bank, u32 ring) +{ + return READ_CSR_RING_HEAD(csr_base_addr, bank, ring); +} + +static void write_csr_ring_head(void __iomem *csr_base_addr, u32 bank, u32 ring, + u32 value) +{ + WRITE_CSR_RING_HEAD(csr_base_addr, bank, ring, value); +} + +static u32 read_csr_ring_tail(void __iomem *csr_base_addr, u32 bank, u32 ring) +{ + return READ_CSR_RING_TAIL(csr_base_addr, bank, ring); +} + +static void write_csr_ring_tail(void __iomem *csr_base_addr, u32 bank, u32 ring, + u32 value) +{ + WRITE_CSR_RING_TAIL(csr_base_addr, bank, ring, value); +} + +static u32 read_csr_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_STAT(csr_base_addr, bank); +} + +static u32 read_csr_uo_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_UO_STAT(csr_base_addr, bank); +} + +static u32 read_csr_e_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_E_STAT(csr_base_addr, bank); +} + +static u32 read_csr_ne_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_NE_STAT(csr_base_addr, bank); +} + +static u32 read_csr_nf_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_NF_STAT(csr_base_addr, bank); +} + +static u32 read_csr_f_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_F_STAT(csr_base_addr, bank); +} + +static u32 read_csr_c_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_C_STAT(csr_base_addr, bank); +} + +static u32 read_csr_exp_stat(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_EXP_STAT(csr_base_addr, bank); +} + +static u32 read_csr_exp_int_en(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_EXP_INT_EN(csr_base_addr, bank); +} + +static void write_csr_exp_int_en(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_EXP_INT_EN(csr_base_addr, bank, value); +} + +static u32 read_csr_ring_config(void __iomem *csr_base_addr, u32 bank, + u32 ring) +{ + return READ_CSR_RING_CONFIG(csr_base_addr, bank, ring); +} + +static void write_csr_ring_config(void __iomem *csr_base_addr, u32 bank, u32 ring, + u32 value) +{ + WRITE_CSR_RING_CONFIG(csr_base_addr, bank, ring, value); +} + +static dma_addr_t read_csr_ring_base(void __iomem *csr_base_addr, u32 bank, + u32 ring) +{ + return READ_CSR_RING_BASE(csr_base_addr, bank, ring); +} + +static void write_csr_ring_base(void __iomem *csr_base_addr, u32 bank, u32 ring, + dma_addr_t addr) +{ + WRITE_CSR_RING_BASE(csr_base_addr, bank, ring, addr); +} + +static u32 read_csr_int_en(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_INT_EN(csr_base_addr, bank); +} + +static void write_csr_int_en(void __iomem *csr_base_addr, u32 bank, u32 value) +{ + WRITE_CSR_INT_EN(csr_base_addr, bank, value); +} + +static u32 read_csr_int_flag(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_INT_FLAG(csr_base_addr, bank); +} + +static void write_csr_int_flag(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_INT_FLAG(csr_base_addr, bank, value); +} + +static u32 read_csr_int_srcsel(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_INT_SRCSEL(csr_base_addr, bank); +} + +static void write_csr_int_srcsel(void __iomem *csr_base_addr, u32 bank) +{ + WRITE_CSR_INT_SRCSEL(csr_base_addr, bank); +} + +static void write_csr_int_srcsel_w_val(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_INT_SRCSEL_W_VAL(csr_base_addr, bank, value); +} + +static u32 read_csr_int_col_en(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_INT_COL_EN(csr_base_addr, bank); +} + +static void write_csr_int_col_en(void __iomem *csr_base_addr, u32 bank, u32 value) +{ + WRITE_CSR_INT_COL_EN(csr_base_addr, bank, value); +} + +static u32 read_csr_int_col_ctl(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_INT_COL_CTL(csr_base_addr, bank); +} + +static void write_csr_int_col_ctl(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_INT_COL_CTL(csr_base_addr, bank, value); +} + +static u32 read_csr_int_flag_and_col(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_INT_FLAG_AND_COL(csr_base_addr, bank); +} + +static void write_csr_int_flag_and_col(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_INT_FLAG_AND_COL(csr_base_addr, bank, value); +} + +static u32 read_csr_ring_srv_arb_en(void __iomem *csr_base_addr, u32 bank) +{ + return READ_CSR_RING_SRV_ARB_EN(csr_base_addr, bank); +} + +static void write_csr_ring_srv_arb_en(void __iomem *csr_base_addr, u32 bank, + u32 value) +{ + WRITE_CSR_RING_SRV_ARB_EN(csr_base_addr, bank, value); +} + +static u32 get_int_col_ctl_enable_mask(void) +{ + return ADF_RING_CSR_INT_COL_CTL_ENABLE; +} + +void adf_gen4_init_hw_csr_ops(struct adf_hw_csr_ops *csr_ops) +{ + csr_ops->build_csr_ring_base_addr = build_csr_ring_base_addr; + csr_ops->read_csr_ring_head = read_csr_ring_head; + csr_ops->write_csr_ring_head = write_csr_ring_head; + csr_ops->read_csr_ring_tail = read_csr_ring_tail; + csr_ops->write_csr_ring_tail = write_csr_ring_tail; + csr_ops->read_csr_stat = read_csr_stat; + csr_ops->read_csr_uo_stat = read_csr_uo_stat; + csr_ops->read_csr_e_stat = read_csr_e_stat; + csr_ops->read_csr_ne_stat = read_csr_ne_stat; + csr_ops->read_csr_nf_stat = read_csr_nf_stat; + csr_ops->read_csr_f_stat = read_csr_f_stat; + csr_ops->read_csr_c_stat = read_csr_c_stat; + csr_ops->read_csr_exp_stat = read_csr_exp_stat; + csr_ops->read_csr_exp_int_en = read_csr_exp_int_en; + csr_ops->write_csr_exp_int_en = write_csr_exp_int_en; + csr_ops->read_csr_ring_config = read_csr_ring_config; + csr_ops->write_csr_ring_config = write_csr_ring_config; + csr_ops->read_csr_ring_base = read_csr_ring_base; + csr_ops->write_csr_ring_base = write_csr_ring_base; + csr_ops->read_csr_int_en = read_csr_int_en; + csr_ops->write_csr_int_en = write_csr_int_en; + csr_ops->read_csr_int_flag = read_csr_int_flag; + csr_ops->write_csr_int_flag = write_csr_int_flag; + csr_ops->read_csr_int_srcsel = read_csr_int_srcsel; + csr_ops->write_csr_int_srcsel = write_csr_int_srcsel; + csr_ops->write_csr_int_srcsel_w_val = write_csr_int_srcsel_w_val; + csr_ops->read_csr_int_col_en = read_csr_int_col_en; + csr_ops->write_csr_int_col_en = write_csr_int_col_en; + csr_ops->read_csr_int_col_ctl = read_csr_int_col_ctl; + csr_ops->write_csr_int_col_ctl = write_csr_int_col_ctl; + csr_ops->read_csr_int_flag_and_col = read_csr_int_flag_and_col; + csr_ops->write_csr_int_flag_and_col = write_csr_int_flag_and_col; + csr_ops->read_csr_ring_srv_arb_en = read_csr_ring_srv_arb_en; + csr_ops->write_csr_ring_srv_arb_en = write_csr_ring_srv_arb_en; + csr_ops->get_int_col_ctl_enable_mask = get_int_col_ctl_enable_mask; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_hw_csr_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_csr_data.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_csr_data.h new file mode 100644 index 000000000000..6f33e7c87c2c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_csr_data.h @@ -0,0 +1,188 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2024 Intel Corporation */ +#ifndef ADF_GEN4_HW_CSR_DATA_H_ +#define ADF_GEN4_HW_CSR_DATA_H_ + +#include <linux/bitops.h> +#include "adf_accel_devices.h" + +#define ADF_BANK_INT_SRC_SEL_MASK 0x44UL +#define ADF_RING_CSR_RING_CONFIG 0x1000 +#define ADF_RING_CSR_RING_LBASE 0x1040 +#define ADF_RING_CSR_RING_UBASE 0x1080 +#define ADF_RING_CSR_RING_HEAD 0x0C0 +#define ADF_RING_CSR_RING_TAIL 0x100 +#define ADF_RING_CSR_STAT 0x140 +#define ADF_RING_CSR_UO_STAT 0x148 +#define ADF_RING_CSR_E_STAT 0x14C +#define ADF_RING_CSR_NE_STAT 0x150 +#define ADF_RING_CSR_NF_STAT 0x154 +#define ADF_RING_CSR_F_STAT 0x158 +#define ADF_RING_CSR_C_STAT 0x15C +#define ADF_RING_CSR_INT_FLAG_EN 0x16C +#define ADF_RING_CSR_INT_FLAG 0x170 +#define ADF_RING_CSR_INT_SRCSEL 0x174 +#define ADF_RING_CSR_INT_COL_EN 0x17C +#define ADF_RING_CSR_INT_COL_CTL 0x180 +#define ADF_RING_CSR_INT_FLAG_AND_COL 0x184 +#define ADF_RING_CSR_EXP_STAT 0x188 +#define ADF_RING_CSR_EXP_INT_EN 0x18C +#define ADF_RING_CSR_INT_COL_CTL_ENABLE 0x80000000 +#define ADF_RING_CSR_ADDR_OFFSET 0x100000 +#define ADF_RING_BUNDLE_SIZE 0x2000 +#define ADF_RING_CSR_RING_SRV_ARB_EN 0x19C + +#define BUILD_RING_BASE_ADDR(addr, size) \ + ((((addr) >> 6) & (GENMASK_ULL(63, 0) << (size))) << 6) +#define READ_CSR_RING_HEAD(csr_base_addr, bank, ring) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_HEAD + ((ring) << 2)) +#define READ_CSR_RING_TAIL(csr_base_addr, bank, ring) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_TAIL + ((ring) << 2)) +#define READ_CSR_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_STAT) +#define READ_CSR_UO_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_UO_STAT) +#define READ_CSR_E_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_E_STAT) +#define READ_CSR_NE_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_NE_STAT) +#define READ_CSR_NF_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_NF_STAT) +#define READ_CSR_F_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_F_STAT) +#define READ_CSR_C_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_C_STAT) +#define READ_CSR_EXP_STAT(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_EXP_STAT) +#define READ_CSR_EXP_INT_EN(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_EXP_INT_EN) +#define WRITE_CSR_EXP_INT_EN(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_EXP_INT_EN, value) +#define READ_CSR_RING_CONFIG(csr_base_addr, bank, ring) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_CONFIG + ((ring) << 2)) +#define WRITE_CSR_RING_CONFIG(csr_base_addr, bank, ring, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_CONFIG + ((ring) << 2), value) +#define WRITE_CSR_RING_BASE(csr_base_addr, bank, ring, value) \ +do { \ + void __iomem *_csr_base_addr = csr_base_addr; \ + u32 _bank = bank; \ + u32 _ring = ring; \ + dma_addr_t _value = value; \ + u32 l_base = 0, u_base = 0; \ + l_base = lower_32_bits(_value); \ + u_base = upper_32_bits(_value); \ + ADF_CSR_WR((_csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (_bank) + \ + ADF_RING_CSR_RING_LBASE + ((_ring) << 2), l_base); \ + ADF_CSR_WR((_csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (_bank) + \ + ADF_RING_CSR_RING_UBASE + ((_ring) << 2), u_base); \ +} while (0) + +static inline u64 read_base(void __iomem *csr_base_addr, u32 bank, u32 ring) +{ + u32 l_base, u_base; + + /* + * Use special IO wrapper for ring base as LBASE and UBASE are + * not physically contigious + */ + l_base = ADF_CSR_RD(csr_base_addr, (ADF_RING_BUNDLE_SIZE * bank) + + ADF_RING_CSR_RING_LBASE + (ring << 2)); + u_base = ADF_CSR_RD(csr_base_addr, (ADF_RING_BUNDLE_SIZE * bank) + + ADF_RING_CSR_RING_UBASE + (ring << 2)); + + return (u64)u_base << 32 | (u64)l_base; +} + +#define READ_CSR_RING_BASE(csr_base_addr, bank, ring) \ + read_base((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, (bank), (ring)) + +#define WRITE_CSR_RING_HEAD(csr_base_addr, bank, ring, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_HEAD + ((ring) << 2), value) +#define WRITE_CSR_RING_TAIL(csr_base_addr, bank, ring, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_TAIL + ((ring) << 2), value) +#define READ_CSR_INT_EN(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_INT_FLAG_EN) +#define WRITE_CSR_INT_EN(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_FLAG_EN, (value)) +#define READ_CSR_INT_FLAG(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_INT_FLAG) +#define WRITE_CSR_INT_FLAG(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_FLAG, (value)) +#define READ_CSR_INT_SRCSEL(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_INT_SRCSEL) +#define WRITE_CSR_INT_SRCSEL(csr_base_addr, bank) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_SRCSEL, ADF_BANK_INT_SRC_SEL_MASK) +#define WRITE_CSR_INT_SRCSEL_W_VAL(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_SRCSEL, (value)) +#define READ_CSR_INT_COL_EN(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_INT_COL_EN) +#define WRITE_CSR_INT_COL_EN(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_COL_EN, (value)) +#define READ_CSR_INT_COL_CTL(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + ADF_RING_CSR_INT_COL_CTL) +#define WRITE_CSR_INT_COL_CTL(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_COL_CTL, \ + ADF_RING_CSR_INT_COL_CTL_ENABLE | (value)) +#define READ_CSR_INT_FLAG_AND_COL(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_FLAG_AND_COL) +#define WRITE_CSR_INT_FLAG_AND_COL(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_INT_FLAG_AND_COL, (value)) + +#define READ_CSR_RING_SRV_ARB_EN(csr_base_addr, bank) \ + ADF_CSR_RD((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_SRV_ARB_EN) +#define WRITE_CSR_RING_SRV_ARB_EN(csr_base_addr, bank, value) \ + ADF_CSR_WR((csr_base_addr) + ADF_RING_CSR_ADDR_OFFSET, \ + ADF_RING_BUNDLE_SIZE * (bank) + \ + ADF_RING_CSR_RING_SRV_ARB_EN, (value)) + +void adf_gen4_init_hw_csr_ops(struct adf_hw_csr_ops *csr_ops); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_data.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_data.c new file mode 100644 index 000000000000..349fdb323763 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_data.c @@ -0,0 +1,600 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2020 Intel Corporation */ + +#define pr_fmt(fmt) "QAT: " fmt + +#include <linux/bitops.h> +#include <linux/iopoll.h> +#include <asm/div64.h> +#include "adf_accel_devices.h" +#include "adf_cfg_services.h" +#include "adf_common_drv.h" +#include "adf_fw_config.h" +#include "adf_gen4_hw_data.h" +#include "adf_gen4_pm.h" +#include "icp_qat_fw_comp.h" +#include "icp_qat_hw_20_comp.h" + +u32 adf_gen4_get_accel_mask(struct adf_hw_device_data *self) +{ + return ADF_GEN4_ACCELERATORS_MASK; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_accel_mask); + +u32 adf_gen4_get_num_accels(struct adf_hw_device_data *self) +{ + return ADF_GEN4_MAX_ACCELERATORS; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_num_accels); + +u32 adf_gen4_get_num_aes(struct adf_hw_device_data *self) +{ + if (!self || !self->ae_mask) + return 0; + + return hweight32(self->ae_mask); +} +EXPORT_SYMBOL_GPL(adf_gen4_get_num_aes); + +u32 adf_gen4_get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_GEN4_PMISC_BAR; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_misc_bar_id); + +u32 adf_gen4_get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_GEN4_ETR_BAR; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_etr_bar_id); + +u32 adf_gen4_get_sram_bar_id(struct adf_hw_device_data *self) +{ + return ADF_GEN4_SRAM_BAR; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_sram_bar_id); + +enum dev_sku_info adf_gen4_get_sku(struct adf_hw_device_data *self) +{ + return DEV_SKU_1; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_sku); + +void adf_gen4_get_arb_info(struct arb_info *arb_info) +{ + arb_info->arb_cfg = ADF_GEN4_ARB_CONFIG; + arb_info->arb_offset = ADF_GEN4_ARB_OFFSET; + arb_info->wt2sam_offset = ADF_GEN4_ARB_WRK_2_SER_MAP_OFFSET; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_arb_info); + +void adf_gen4_get_admin_info(struct admin_info *admin_csrs_info) +{ + admin_csrs_info->mailbox_offset = ADF_GEN4_MAILBOX_BASE_OFFSET; + admin_csrs_info->admin_msg_ur = ADF_GEN4_ADMINMSGUR_OFFSET; + admin_csrs_info->admin_msg_lr = ADF_GEN4_ADMINMSGLR_OFFSET; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_admin_info); + +u32 adf_gen4_get_heartbeat_clock(struct adf_hw_device_data *self) +{ + /* + * GEN4 uses KPT counter for HB + */ + return ADF_GEN4_KPT_COUNTER_FREQ; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_heartbeat_clock); + +void adf_gen4_enable_error_correction(struct adf_accel_dev *accel_dev) +{ + struct adf_bar *misc_bar = &GET_BARS(accel_dev)[ADF_GEN4_PMISC_BAR]; + void __iomem *csr = misc_bar->virt_addr; + + /* Enable all in errsou3 except VFLR notification on host */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK3, ADF_GEN4_VFLNOTIFY); +} +EXPORT_SYMBOL_GPL(adf_gen4_enable_error_correction); + +void adf_gen4_enable_ints(struct adf_accel_dev *accel_dev) +{ + void __iomem *addr; + + addr = (&GET_BARS(accel_dev)[ADF_GEN4_PMISC_BAR])->virt_addr; + + /* Enable bundle interrupts */ + ADF_CSR_WR(addr, ADF_GEN4_SMIAPF_RP_X0_MASK_OFFSET, 0); + ADF_CSR_WR(addr, ADF_GEN4_SMIAPF_RP_X1_MASK_OFFSET, 0); + + /* Enable misc interrupts */ + ADF_CSR_WR(addr, ADF_GEN4_SMIAPF_MASK_OFFSET, 0); +} +EXPORT_SYMBOL_GPL(adf_gen4_enable_ints); + +int adf_gen4_init_device(struct adf_accel_dev *accel_dev) +{ + void __iomem *addr; + u32 status; + u32 csr; + int ret; + + addr = (&GET_BARS(accel_dev)[ADF_GEN4_PMISC_BAR])->virt_addr; + + /* Temporarily mask PM interrupt */ + csr = ADF_CSR_RD(addr, ADF_GEN4_ERRMSK2); + csr |= ADF_GEN4_PM_SOU; + ADF_CSR_WR(addr, ADF_GEN4_ERRMSK2, csr); + + /* Set DRV_ACTIVE bit to power up the device */ + ADF_CSR_WR(addr, ADF_GEN4_PM_INTERRUPT, ADF_GEN4_PM_DRV_ACTIVE); + + /* Poll status register to make sure the device is powered up */ + ret = read_poll_timeout(ADF_CSR_RD, status, + status & ADF_GEN4_PM_INIT_STATE, + ADF_GEN4_PM_POLL_DELAY_US, + ADF_GEN4_PM_POLL_TIMEOUT_US, true, addr, + ADF_GEN4_PM_STATUS); + if (ret) + dev_err(&GET_DEV(accel_dev), "Failed to power up the device\n"); + + return ret; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_device); + +void adf_gen4_set_ssm_wdtimer(struct adf_accel_dev *accel_dev) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u64 timer_val_pke = ADF_SSM_WDT_PKE_DEFAULT_VALUE; + u64 timer_val = ADF_SSM_WDT_DEFAULT_VALUE; + + /* Enable watchdog timer for sym and dc */ + ADF_CSR_WR64_LO_HI(pmisc_addr, ADF_SSMWDTL_OFFSET, ADF_SSMWDTH_OFFSET, timer_val); + + /* Enable watchdog timer for pke */ + ADF_CSR_WR64_LO_HI(pmisc_addr, ADF_SSMWDTPKEL_OFFSET, ADF_SSMWDTPKEH_OFFSET, + timer_val_pke); +} +EXPORT_SYMBOL_GPL(adf_gen4_set_ssm_wdtimer); + +/* + * The vector routing table is used to select the MSI-X entry to use for each + * interrupt source. + * The first ADF_GEN4_ETR_MAX_BANKS entries correspond to ring interrupts. + * The final entry corresponds to VF2PF or error interrupts. + * This vector table could be used to configure one MSI-X entry to be shared + * between multiple interrupt sources. + * + * The default routing is set to have a one to one correspondence between the + * interrupt source and the MSI-X entry used. + */ +void adf_gen4_set_msix_default_rttable(struct adf_accel_dev *accel_dev) +{ + void __iomem *csr; + int i; + + csr = (&GET_BARS(accel_dev)[ADF_GEN4_PMISC_BAR])->virt_addr; + for (i = 0; i <= ADF_GEN4_ETR_MAX_BANKS; i++) + ADF_CSR_WR(csr, ADF_GEN4_MSIX_RTTABLE_OFFSET(i), i); +} +EXPORT_SYMBOL_GPL(adf_gen4_set_msix_default_rttable); + +int adf_pfvf_comms_disabled(struct adf_accel_dev *accel_dev) +{ + return 0; +} +EXPORT_SYMBOL_GPL(adf_pfvf_comms_disabled); + +static int reset_ring_pair(void __iomem *csr, u32 bank_number) +{ + u32 status; + int ret; + + /* Write rpresetctl register BIT(0) as 1 + * Since rpresetctl registers have no RW fields, no need to preserve + * values for other bits. Just write directly. + */ + ADF_CSR_WR(csr, ADF_WQM_CSR_RPRESETCTL(bank_number), + ADF_WQM_CSR_RPRESETCTL_RESET); + + /* Read rpresetsts register and wait for rp reset to complete */ + ret = read_poll_timeout(ADF_CSR_RD, status, + status & ADF_WQM_CSR_RPRESETSTS_STATUS, + ADF_RPRESET_POLL_DELAY_US, + ADF_RPRESET_POLL_TIMEOUT_US, true, + csr, ADF_WQM_CSR_RPRESETSTS(bank_number)); + if (!ret) { + /* When rp reset is done, clear rpresetsts */ + ADF_CSR_WR(csr, ADF_WQM_CSR_RPRESETSTS(bank_number), + ADF_WQM_CSR_RPRESETSTS_STATUS); + } + + return ret; +} + +int adf_gen4_ring_pair_reset(struct adf_accel_dev *accel_dev, u32 bank_number) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + void __iomem *csr = adf_get_etr_base(accel_dev); + int ret; + + if (bank_number >= hw_data->num_banks) + return -EINVAL; + + dev_dbg(&GET_DEV(accel_dev), + "ring pair reset for bank:%d\n", bank_number); + + ret = reset_ring_pair(csr, bank_number); + if (ret) + dev_err(&GET_DEV(accel_dev), + "ring pair reset failed (timeout)\n"); + else + dev_dbg(&GET_DEV(accel_dev), "ring pair reset successful\n"); + + return ret; +} +EXPORT_SYMBOL_GPL(adf_gen4_ring_pair_reset); + +static const u32 thrd_to_arb_map_dcc[] = { + 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x0000FFFF, 0x0000FFFF, 0x0000FFFF, 0x0000FFFF, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x0 +}; + +static const u16 rp_group_to_arb_mask[] = { + [RP_GROUP_0] = 0x5, + [RP_GROUP_1] = 0xA, +}; + +static bool is_single_service(int service_id) +{ + switch (service_id) { + case SVC_DC: + case SVC_SYM: + case SVC_ASYM: + return true; + default: + return false; + } +} + +bool adf_gen4_services_supported(unsigned long mask) +{ + unsigned long num_svc = hweight_long(mask); + + if (mask >= BIT(SVC_COUNT)) + return false; + + if (test_bit(SVC_DECOMP, &mask)) + return false; + + switch (num_svc) { + case ADF_ONE_SERVICE: + return true; + case ADF_TWO_SERVICES: + return !test_bit(SVC_DCC, &mask); + default: + return false; + } +} +EXPORT_SYMBOL_GPL(adf_gen4_services_supported); + +int adf_gen4_init_thd2arb_map(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + u32 *thd2arb_map = hw_data->thd_to_arb_map; + unsigned int ae_cnt, worker_obj_cnt, i, j; + unsigned long ae_mask, thds_mask; + int srv_id, rp_group; + u32 thd2arb_map_base; + u16 arb_mask; + + if (!hw_data->get_rp_group || !hw_data->get_ena_thd_mask || + !hw_data->get_num_aes || !hw_data->uof_get_num_objs || + !hw_data->uof_get_ae_mask) + return -EFAULT; + + srv_id = adf_get_service_enabled(accel_dev); + if (srv_id < 0) + return srv_id; + + ae_cnt = hw_data->get_num_aes(hw_data); + worker_obj_cnt = hw_data->uof_get_num_objs(accel_dev) - + ADF_GEN4_ADMIN_ACCELENGINES; + + if (srv_id == SVC_DCC) { + if (ae_cnt > ICP_QAT_HW_AE_DELIMITER) + return -EINVAL; + + memcpy(thd2arb_map, thrd_to_arb_map_dcc, + array_size(sizeof(*thd2arb_map), ae_cnt)); + return 0; + } + + for (i = 0; i < worker_obj_cnt; i++) { + ae_mask = hw_data->uof_get_ae_mask(accel_dev, i); + rp_group = hw_data->get_rp_group(accel_dev, ae_mask); + thds_mask = hw_data->get_ena_thd_mask(accel_dev, i); + thd2arb_map_base = 0; + + if (rp_group >= RP_GROUP_COUNT || rp_group < RP_GROUP_0) + return -EINVAL; + + if (thds_mask == ADF_GEN4_ENA_THD_MASK_ERROR) + return -EINVAL; + + if (is_single_service(srv_id)) + arb_mask = rp_group_to_arb_mask[RP_GROUP_0] | + rp_group_to_arb_mask[RP_GROUP_1]; + else + arb_mask = rp_group_to_arb_mask[rp_group]; + + for_each_set_bit(j, &thds_mask, ADF_NUM_THREADS_PER_AE) + thd2arb_map_base |= arb_mask << (j * 4); + + for_each_set_bit(j, &ae_mask, ae_cnt) + thd2arb_map[j] = thd2arb_map_base; + } + return 0; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_thd2arb_map); + +u16 adf_gen4_get_ring_to_svc_map(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + enum adf_cfg_service_type rps[RP_GROUP_COUNT] = { }; + unsigned int ae_mask, start_id, worker_obj_cnt, i; + u16 ring_to_svc_map; + int rp_group; + + if (!hw_data->get_rp_group || !hw_data->uof_get_ae_mask || + !hw_data->uof_get_obj_type || !hw_data->uof_get_num_objs) + return 0; + + /* If dcc, all rings handle compression requests */ + if (adf_get_service_enabled(accel_dev) == SVC_DCC) { + for (i = 0; i < RP_GROUP_COUNT; i++) + rps[i] = COMP; + goto set_mask; + } + + worker_obj_cnt = hw_data->uof_get_num_objs(accel_dev) - + ADF_GEN4_ADMIN_ACCELENGINES; + start_id = worker_obj_cnt - RP_GROUP_COUNT; + + for (i = start_id; i < worker_obj_cnt; i++) { + ae_mask = hw_data->uof_get_ae_mask(accel_dev, i); + rp_group = hw_data->get_rp_group(accel_dev, ae_mask); + if (rp_group >= RP_GROUP_COUNT || rp_group < RP_GROUP_0) + return 0; + + switch (hw_data->uof_get_obj_type(accel_dev, i)) { + case ADF_FW_SYM_OBJ: + rps[rp_group] = SYM; + break; + case ADF_FW_ASYM_OBJ: + rps[rp_group] = ASYM; + break; + case ADF_FW_DC_OBJ: + rps[rp_group] = COMP; + break; + default: + rps[rp_group] = 0; + break; + } + } + +set_mask: + ring_to_svc_map = rps[RP_GROUP_0] << ADF_CFG_SERV_RING_PAIR_0_SHIFT | + rps[RP_GROUP_1] << ADF_CFG_SERV_RING_PAIR_1_SHIFT | + rps[RP_GROUP_0] << ADF_CFG_SERV_RING_PAIR_2_SHIFT | + rps[RP_GROUP_1] << ADF_CFG_SERV_RING_PAIR_3_SHIFT; + + return ring_to_svc_map; +} +EXPORT_SYMBOL_GPL(adf_gen4_get_ring_to_svc_map); + +/* + * adf_gen4_bank_quiesce_coal_timer() - quiesce bank coalesced interrupt timer + * @accel_dev: Pointer to the device structure + * @bank_idx: Offset to the bank within this device + * @timeout_ms: Timeout in milliseconds for the operation + * + * This function tries to quiesce the coalesced interrupt timer of a bank if + * it has been enabled and triggered. + * + * Returns 0 on success, error code otherwise + * + */ +int adf_gen4_bank_quiesce_coal_timer(struct adf_accel_dev *accel_dev, + u32 bank_idx, int timeout_ms) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(accel_dev); + void __iomem *csr_misc = adf_get_pmisc_base(accel_dev); + void __iomem *csr_etr = adf_get_etr_base(accel_dev); + u32 int_col_ctl, int_col_mask, int_col_en; + u32 e_stat, intsrc; + u64 wait_us; + int ret; + + if (timeout_ms < 0) + return -EINVAL; + + int_col_ctl = csr_ops->read_csr_int_col_ctl(csr_etr, bank_idx); + int_col_mask = csr_ops->get_int_col_ctl_enable_mask(); + if (!(int_col_ctl & int_col_mask)) + return 0; + + int_col_en = csr_ops->read_csr_int_col_en(csr_etr, bank_idx); + int_col_en &= BIT(ADF_WQM_CSR_RP_IDX_RX); + + e_stat = csr_ops->read_csr_e_stat(csr_etr, bank_idx); + if (!(~e_stat & int_col_en)) + return 0; + + wait_us = 2 * ((int_col_ctl & ~int_col_mask) << 8) * USEC_PER_SEC; + do_div(wait_us, hw_data->clock_frequency); + wait_us = min(wait_us, (u64)timeout_ms * USEC_PER_MSEC); + dev_dbg(&GET_DEV(accel_dev), + "wait for bank %d - coalesced timer expires in %llu us (max=%u ms estat=0x%x intcolen=0x%x)\n", + bank_idx, wait_us, timeout_ms, e_stat, int_col_en); + + ret = read_poll_timeout(ADF_CSR_RD, intsrc, intsrc, + ADF_COALESCED_POLL_DELAY_US, wait_us, true, + csr_misc, ADF_WQM_CSR_RPINTSOU(bank_idx)); + if (ret) + dev_warn(&GET_DEV(accel_dev), + "coalesced timer for bank %d expired (%llu us)\n", + bank_idx, wait_us); + + return ret; +} +EXPORT_SYMBOL_GPL(adf_gen4_bank_quiesce_coal_timer); + +static int drain_bank(void __iomem *csr, u32 bank_number, int timeout_us) +{ + u32 status; + + ADF_CSR_WR(csr, ADF_WQM_CSR_RPRESETCTL(bank_number), + ADF_WQM_CSR_RPRESETCTL_DRAIN); + + return read_poll_timeout(ADF_CSR_RD, status, + status & ADF_WQM_CSR_RPRESETSTS_STATUS, + ADF_RPRESET_POLL_DELAY_US, timeout_us, true, + csr, ADF_WQM_CSR_RPRESETSTS(bank_number)); +} + +void adf_gen4_bank_drain_finish(struct adf_accel_dev *accel_dev, + u32 bank_number) +{ + void __iomem *csr = adf_get_etr_base(accel_dev); + + ADF_CSR_WR(csr, ADF_WQM_CSR_RPRESETSTS(bank_number), + ADF_WQM_CSR_RPRESETSTS_STATUS); +} + +int adf_gen4_bank_drain_start(struct adf_accel_dev *accel_dev, + u32 bank_number, int timeout_us) +{ + void __iomem *csr = adf_get_etr_base(accel_dev); + int ret; + + dev_dbg(&GET_DEV(accel_dev), "Drain bank %d\n", bank_number); + + ret = drain_bank(csr, bank_number, timeout_us); + if (ret) + dev_err(&GET_DEV(accel_dev), "Bank drain failed (timeout)\n"); + else + dev_dbg(&GET_DEV(accel_dev), "Bank drain successful\n"); + + return ret; +} + +static int adf_gen4_build_comp_block(void *ctx, enum adf_dc_algo algo) +{ + struct icp_qat_fw_comp_req *req_tmpl = ctx; + struct icp_qat_fw_comp_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_hw_comp_20_config_csr_upper hw_comp_upper_csr = { }; + struct icp_qat_hw_comp_20_config_csr_lower hw_comp_lower_csr = { }; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + u32 upper_val; + u32 lower_val; + + switch (algo) { + case QAT_DEFLATE: + header->service_cmd_id = ICP_QAT_FW_COMP_CMD_DYNAMIC; + break; + default: + return -EINVAL; + } + + hw_comp_lower_csr.skip_ctrl = ICP_QAT_HW_COMP_20_BYTE_SKIP_3BYTE_LITERAL; + hw_comp_lower_csr.algo = ICP_QAT_HW_COMP_20_HW_COMP_FORMAT_ILZ77; + hw_comp_lower_csr.lllbd = ICP_QAT_HW_COMP_20_LLLBD_CTRL_LLLBD_ENABLED; + hw_comp_lower_csr.sd = ICP_QAT_HW_COMP_20_SEARCH_DEPTH_LEVEL_1; + hw_comp_lower_csr.hash_update = ICP_QAT_HW_COMP_20_SKIP_HASH_UPDATE_DONT_ALLOW; + hw_comp_lower_csr.edmm = ICP_QAT_HW_COMP_20_EXTENDED_DELAY_MATCH_MODE_EDMM_ENABLED; + hw_comp_upper_csr.nice = ICP_QAT_HW_COMP_20_CONFIG_CSR_NICE_PARAM_DEFAULT_VAL; + hw_comp_upper_csr.lazy = ICP_QAT_HW_COMP_20_CONFIG_CSR_LAZY_PARAM_DEFAULT_VAL; + + upper_val = ICP_QAT_FW_COMP_20_BUILD_CONFIG_UPPER(hw_comp_upper_csr); + lower_val = ICP_QAT_FW_COMP_20_BUILD_CONFIG_LOWER(hw_comp_lower_csr); + + cd_pars->u.sl.comp_slice_cfg_word[0] = lower_val; + cd_pars->u.sl.comp_slice_cfg_word[1] = upper_val; + + return 0; +} + +static int adf_gen4_build_decomp_block(void *ctx, enum adf_dc_algo algo) +{ + struct icp_qat_fw_comp_req *req_tmpl = ctx; + struct icp_qat_hw_decomp_20_config_csr_lower hw_decomp_lower_csr = { }; + struct icp_qat_fw_comp_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + u32 lower_val; + + switch (algo) { + case QAT_DEFLATE: + header->service_cmd_id = ICP_QAT_FW_COMP_CMD_DECOMPRESS; + break; + default: + return -EINVAL; + } + + hw_decomp_lower_csr.algo = ICP_QAT_HW_DECOMP_20_HW_DECOMP_FORMAT_DEFLATE; + lower_val = ICP_QAT_FW_DECOMP_20_BUILD_CONFIG_LOWER(hw_decomp_lower_csr); + + cd_pars->u.sl.comp_slice_cfg_word[0] = lower_val; + cd_pars->u.sl.comp_slice_cfg_word[1] = 0; + + return 0; +} + +void adf_gen4_init_dc_ops(struct adf_dc_ops *dc_ops) +{ + dc_ops->build_comp_block = adf_gen4_build_comp_block; + dc_ops->build_decomp_block = adf_gen4_build_decomp_block; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_dc_ops); + +void adf_gen4_init_num_svc_aes(struct adf_rl_hw_data *device_data) +{ + struct adf_hw_device_data *hw_data; + unsigned int i; + u32 ae_cnt; + + hw_data = container_of(device_data, struct adf_hw_device_data, rl_data); + ae_cnt = hweight32(hw_data->get_ae_mask(hw_data)); + if (!ae_cnt) + return; + + for (i = 0; i < SVC_BASE_COUNT; i++) + device_data->svc_ae_mask[i] = ae_cnt - 1; + + /* + * The decompression service is not supported on QAT GEN4 devices. + * Therefore, set svc_ae_mask to 0. + */ + device_data->svc_ae_mask[SVC_DECOMP] = 0; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_num_svc_aes); + +u32 adf_gen4_get_svc_slice_cnt(struct adf_accel_dev *accel_dev, + enum adf_base_services svc) +{ + struct adf_rl_hw_data *device_data = &accel_dev->hw_device->rl_data; + + switch (svc) { + case SVC_SYM: + return device_data->slices.cph_cnt; + case SVC_ASYM: + return device_data->slices.pke_cnt; + case SVC_DC: + return device_data->slices.dcpr_cnt; + default: + return 0; + } +} +EXPORT_SYMBOL_GPL(adf_gen4_get_svc_slice_cnt); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_data.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_data.h new file mode 100644 index 000000000000..cd26b6724c43 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_hw_data.h @@ -0,0 +1,182 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2020 Intel Corporation */ +#ifndef ADF_GEN4_HW_DATA_H_ +#define ADF_GEN4_HW_DATA_H_ + +#include <linux/units.h> + +#include "adf_accel_devices.h" +#include "adf_cfg_common.h" +#include "adf_dc.h" + +/* PCIe configuration space */ +#define ADF_GEN4_BAR_MASK (BIT(0) | BIT(2) | BIT(4)) +#define ADF_GEN4_SRAM_BAR 0 +#define ADF_GEN4_PMISC_BAR 1 +#define ADF_GEN4_ETR_BAR 2 + +/* Clocks frequency */ +#define ADF_GEN4_KPT_COUNTER_FREQ (100 * HZ_PER_MHZ) + +/* Physical function fuses */ +#define ADF_GEN4_FUSECTL0_OFFSET 0x2C8 +#define ADF_GEN4_FUSECTL1_OFFSET 0x2CC +#define ADF_GEN4_FUSECTL2_OFFSET 0x2D0 +#define ADF_GEN4_FUSECTL3_OFFSET 0x2D4 +#define ADF_GEN4_FUSECTL4_OFFSET 0x2D8 +#define ADF_GEN4_FUSECTL5_OFFSET 0x2DC + +/* Accelerators */ +#define ADF_GEN4_ACCELERATORS_MASK 0x1 +#define ADF_GEN4_MAX_ACCELERATORS 1 +#define ADF_GEN4_ADMIN_ACCELENGINES 1 + +/* MSIX interrupt */ +#define ADF_GEN4_SMIAPF_RP_X0_MASK_OFFSET 0x41A040 +#define ADF_GEN4_SMIAPF_RP_X1_MASK_OFFSET 0x41A044 +#define ADF_GEN4_SMIAPF_MASK_OFFSET 0x41A084 +#define ADF_GEN4_MSIX_RTTABLE_OFFSET(i) (0x409000 + ((i) * 0x04)) + +/* Bank and ring configuration */ +#define ADF_GEN4_MAX_RPS 64 +#define ADF_GEN4_NUM_RINGS_PER_BANK 2 +#define ADF_GEN4_NUM_BANKS_PER_VF 4 +#define ADF_GEN4_ETR_MAX_BANKS 64 +#define ADF_GEN4_RX_RINGS_OFFSET 1 +#define ADF_GEN4_TX_RINGS_MASK 0x1 + +/* Arbiter configuration */ +#define ADF_GEN4_ARB_CONFIG (BIT(31) | BIT(6) | BIT(0)) +#define ADF_GEN4_ARB_OFFSET 0x0 +#define ADF_GEN4_ARB_WRK_2_SER_MAP_OFFSET 0x400 + +/* Admin Interface Reg Offset */ +#define ADF_GEN4_ADMINMSGUR_OFFSET 0x500574 +#define ADF_GEN4_ADMINMSGLR_OFFSET 0x500578 +#define ADF_GEN4_MAILBOX_BASE_OFFSET 0x600970 + +/* Default ring mapping */ +#define ADF_GEN4_DEFAULT_RING_TO_SRV_MAP \ + (ASYM << ADF_CFG_SERV_RING_PAIR_0_SHIFT | \ + SYM << ADF_CFG_SERV_RING_PAIR_1_SHIFT | \ + ASYM << ADF_CFG_SERV_RING_PAIR_2_SHIFT | \ + SYM << ADF_CFG_SERV_RING_PAIR_3_SHIFT) + +/* WDT timers + * + * Timeout is in cycles. Clock speed may vary across products but this + * value should be a few milli-seconds. + */ +#define ADF_SSM_WDT_DEFAULT_VALUE 0x7000000ULL +#define ADF_SSM_WDT_PKE_DEFAULT_VALUE 0x8000000 +#define ADF_SSMWDTL_OFFSET 0x54 +#define ADF_SSMWDTH_OFFSET 0x5C +#define ADF_SSMWDTPKEL_OFFSET 0x58 +#define ADF_SSMWDTPKEH_OFFSET 0x60 + +/* Ring reset */ +#define ADF_RPRESET_POLL_TIMEOUT_US (5 * USEC_PER_SEC) +#define ADF_RPRESET_POLL_DELAY_US 20 +#define ADF_WQM_CSR_RPRESETCTL_RESET BIT(0) +#define ADF_WQM_CSR_RPRESETCTL_DRAIN BIT(2) +#define ADF_WQM_CSR_RPRESETCTL(bank) (0x6000 + ((bank) << 3)) +#define ADF_WQM_CSR_RPRESETSTS_STATUS BIT(0) +#define ADF_WQM_CSR_RPRESETSTS(bank) (ADF_WQM_CSR_RPRESETCTL(bank) + 4) + +/* Ring interrupt */ +#define ADF_COALESCED_POLL_TIMEOUT_US (1 * USEC_PER_SEC) +#define ADF_COALESCED_POLL_DELAY_US 1000 +#define ADF_WQM_CSR_RPINTSOU(bank) (0x200000 + ((bank) << 12)) +#define ADF_WQM_CSR_RP_IDX_RX 1 + +/* Error source registers */ +#define ADF_GEN4_ERRSOU0 (0x41A200) +#define ADF_GEN4_ERRSOU1 (0x41A204) +#define ADF_GEN4_ERRSOU2 (0x41A208) +#define ADF_GEN4_ERRSOU3 (0x41A20C) + +/* Error source mask registers */ +#define ADF_GEN4_ERRMSK0 (0x41A210) +#define ADF_GEN4_ERRMSK1 (0x41A214) +#define ADF_GEN4_ERRMSK2 (0x41A218) +#define ADF_GEN4_ERRMSK3 (0x41A21C) + +#define ADF_GEN4_VFLNOTIFY BIT(7) + +/* Number of heartbeat counter pairs */ +#define ADF_NUM_HB_CNT_PER_AE ADF_NUM_THREADS_PER_AE + +/* Rate Limiting */ +#define ADF_GEN4_RL_R2L_OFFSET 0x508000 +#define ADF_GEN4_RL_L2C_OFFSET 0x509000 +#define ADF_GEN4_RL_C2S_OFFSET 0x508818 +#define ADF_GEN4_RL_TOKEN_PCIEIN_BUCKET_OFFSET 0x508800 +#define ADF_GEN4_RL_TOKEN_PCIEOUT_BUCKET_OFFSET 0x508804 + +/* Arbiter threads mask with error value */ +#define ADF_GEN4_ENA_THD_MASK_ERROR GENMASK(ADF_NUM_THREADS_PER_AE, 0) + +/* PF2VM communication channel */ +#define ADF_GEN4_PF2VM_OFFSET(i) (0x40B010 + (i) * 0x20) +#define ADF_GEN4_VM2PF_OFFSET(i) (0x40B014 + (i) * 0x20) +#define ADF_GEN4_VINTMSKPF2VM_OFFSET(i) (0x40B00C + (i) * 0x20) +#define ADF_GEN4_VINTSOUPF2VM_OFFSET(i) (0x40B008 + (i) * 0x20) +#define ADF_GEN4_VINTMSK_OFFSET(i) (0x40B004 + (i) * 0x20) +#define ADF_GEN4_VINTSOU_OFFSET(i) (0x40B000 + (i) * 0x20) + +struct adf_gen4_vfmig { + struct adf_mstate_mgr *mstate_mgr; + bool bank_stopped[ADF_GEN4_NUM_BANKS_PER_VF]; +}; + +void adf_gen4_set_ssm_wdtimer(struct adf_accel_dev *accel_dev); + +enum icp_qat_gen4_slice_mask { + ICP_ACCEL_GEN4_MASK_CIPHER_SLICE = BIT(0), + ICP_ACCEL_GEN4_MASK_AUTH_SLICE = BIT(1), + ICP_ACCEL_GEN4_MASK_PKE_SLICE = BIT(2), + ICP_ACCEL_GEN4_MASK_COMPRESS_SLICE = BIT(3), + ICP_ACCEL_GEN4_MASK_UCS_SLICE = BIT(4), + ICP_ACCEL_GEN4_MASK_EIA3_SLICE = BIT(5), + ICP_ACCEL_GEN4_MASK_SMX_SLICE = BIT(7), + ICP_ACCEL_GEN4_MASK_WCP_WAT_SLICE = BIT(8), + ICP_ACCEL_GEN4_MASK_ZUC_256_SLICE = BIT(9), +}; + +enum adf_gen4_rp_groups { + RP_GROUP_0, + RP_GROUP_1, + RP_GROUP_COUNT +}; + +void adf_gen4_enable_error_correction(struct adf_accel_dev *accel_dev); +void adf_gen4_enable_ints(struct adf_accel_dev *accel_dev); +u32 adf_gen4_get_accel_mask(struct adf_hw_device_data *self); +void adf_gen4_get_admin_info(struct admin_info *admin_csrs_info); +void adf_gen4_get_arb_info(struct arb_info *arb_info); +u32 adf_gen4_get_etr_bar_id(struct adf_hw_device_data *self); +u32 adf_gen4_get_heartbeat_clock(struct adf_hw_device_data *self); +u32 adf_gen4_get_misc_bar_id(struct adf_hw_device_data *self); +u32 adf_gen4_get_num_accels(struct adf_hw_device_data *self); +u32 adf_gen4_get_num_aes(struct adf_hw_device_data *self); +enum dev_sku_info adf_gen4_get_sku(struct adf_hw_device_data *self); +u32 adf_gen4_get_sram_bar_id(struct adf_hw_device_data *self); +int adf_gen4_init_device(struct adf_accel_dev *accel_dev); +int adf_gen4_ring_pair_reset(struct adf_accel_dev *accel_dev, u32 bank_number); +void adf_gen4_set_msix_default_rttable(struct adf_accel_dev *accel_dev); +void adf_gen4_set_ssm_wdtimer(struct adf_accel_dev *accel_dev); +int adf_gen4_init_thd2arb_map(struct adf_accel_dev *accel_dev); +u16 adf_gen4_get_ring_to_svc_map(struct adf_accel_dev *accel_dev); +int adf_gen4_bank_quiesce_coal_timer(struct adf_accel_dev *accel_dev, + u32 bank_idx, int timeout_ms); +int adf_gen4_bank_drain_start(struct adf_accel_dev *accel_dev, + u32 bank_number, int timeout_us); +void adf_gen4_bank_drain_finish(struct adf_accel_dev *accel_dev, + u32 bank_number); +bool adf_gen4_services_supported(unsigned long service_mask); +void adf_gen4_init_dc_ops(struct adf_dc_ops *dc_ops); +void adf_gen4_init_num_svc_aes(struct adf_rl_hw_data *device_data); +u32 adf_gen4_get_svc_slice_cnt(struct adf_accel_dev *accel_dev, + enum adf_base_services svc); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_pfvf.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_pfvf.c new file mode 100644 index 000000000000..21474d402d09 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_pfvf.c @@ -0,0 +1,145 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2021 Intel Corporation */ +#include <linux/iopoll.h> +#include <linux/mutex.h> +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_gen4_pfvf.h" +#include "adf_gen4_hw_data.h" +#include "adf_pfvf_pf_proto.h" +#include "adf_pfvf_utils.h" + +/* VF2PF interrupt source registers */ +#define ADF_4XXX_VM2PF_SOU 0x41A180 +#define ADF_4XXX_VM2PF_MSK 0x41A1C0 +#define ADF_GEN4_VF_MSK 0xFFFF + +#define ADF_PFVF_GEN4_MSGTYPE_SHIFT 2 +#define ADF_PFVF_GEN4_MSGTYPE_MASK 0x3F +#define ADF_PFVF_GEN4_MSGDATA_SHIFT 8 +#define ADF_PFVF_GEN4_MSGDATA_MASK 0xFFFFFF + +static const struct pfvf_csr_format csr_gen4_fmt = { + { ADF_PFVF_GEN4_MSGTYPE_SHIFT, ADF_PFVF_GEN4_MSGTYPE_MASK }, + { ADF_PFVF_GEN4_MSGDATA_SHIFT, ADF_PFVF_GEN4_MSGDATA_MASK }, +}; + +static u32 adf_gen4_pf_get_pf2vf_offset(u32 i) +{ + return ADF_GEN4_PF2VM_OFFSET(i); +} + +static u32 adf_gen4_pf_get_vf2pf_offset(u32 i) +{ + return ADF_GEN4_VM2PF_OFFSET(i); +} + +static void adf_gen4_enable_vf2pf_interrupts(void __iomem *pmisc_addr, u32 vf_mask) +{ + u32 val; + + val = ADF_CSR_RD(pmisc_addr, ADF_4XXX_VM2PF_MSK) & ~vf_mask; + ADF_CSR_WR(pmisc_addr, ADF_4XXX_VM2PF_MSK, val); +} + +static void adf_gen4_disable_all_vf2pf_interrupts(void __iomem *pmisc_addr) +{ + ADF_CSR_WR(pmisc_addr, ADF_4XXX_VM2PF_MSK, ADF_GEN4_VF_MSK); +} + +static u32 adf_gen4_disable_pending_vf2pf_interrupts(void __iomem *pmisc_addr) +{ + u32 sources, disabled, pending; + + /* Get the interrupt sources triggered by VFs */ + sources = ADF_CSR_RD(pmisc_addr, ADF_4XXX_VM2PF_SOU); + if (!sources) + return 0; + + /* Get the already disabled interrupts */ + disabled = ADF_CSR_RD(pmisc_addr, ADF_4XXX_VM2PF_MSK); + + pending = sources & ~disabled; + if (!pending) + return 0; + + /* Due to HW limitations, when disabling the interrupts, we can't + * just disable the requested sources, as this would lead to missed + * interrupts if VM2PF_SOU changes just before writing to VM2PF_MSK. + * To work around it, disable all and re-enable only the sources that + * are not in vf_mask and were not already disabled. Re-enabling will + * trigger a new interrupt for the sources that have changed in the + * meantime, if any. + */ + ADF_CSR_WR(pmisc_addr, ADF_4XXX_VM2PF_MSK, ADF_GEN4_VF_MSK); + ADF_CSR_WR(pmisc_addr, ADF_4XXX_VM2PF_MSK, disabled | sources); + + /* Return the sources of the (new) interrupt(s) */ + return pending; +} + +static int adf_gen4_pfvf_send(struct adf_accel_dev *accel_dev, + struct pfvf_message msg, u32 pfvf_offset, + struct mutex *csr_lock) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u32 csr_val; + int ret; + + csr_val = adf_pfvf_csr_msg_of(accel_dev, msg, &csr_gen4_fmt); + if (unlikely(!csr_val)) + return -EINVAL; + + mutex_lock(csr_lock); + + ADF_CSR_WR(pmisc_addr, pfvf_offset, csr_val | ADF_PFVF_INT); + + /* Wait for confirmation from remote that it received the message */ + ret = read_poll_timeout(ADF_CSR_RD, csr_val, !(csr_val & ADF_PFVF_INT), + ADF_PFVF_MSG_ACK_DELAY_US, + ADF_PFVF_MSG_ACK_MAX_DELAY_US, + true, pmisc_addr, pfvf_offset); + if (ret < 0) + dev_dbg(&GET_DEV(accel_dev), "ACK not received from remote\n"); + + mutex_unlock(csr_lock); + return ret; +} + +static struct pfvf_message adf_gen4_pfvf_recv(struct adf_accel_dev *accel_dev, + u32 pfvf_offset, u8 compat_ver) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + struct pfvf_message msg = { 0 }; + u32 csr_val; + + /* Read message from the CSR */ + csr_val = ADF_CSR_RD(pmisc_addr, pfvf_offset); + if (!(csr_val & ADF_PFVF_INT)) { + dev_info(&GET_DEV(accel_dev), + "Spurious PFVF interrupt, msg 0x%.8x. Ignored\n", csr_val); + return msg; + } + + /* We can now acknowledge the message reception by clearing the + * interrupt bit + */ + ADF_CSR_WR(pmisc_addr, pfvf_offset, csr_val & ~ADF_PFVF_INT); + + /* Return the pfvf_message format */ + return adf_pfvf_message_of(accel_dev, csr_val, &csr_gen4_fmt); +} + +void adf_gen4_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops) +{ + pfvf_ops->enable_comms = adf_enable_pf2vf_comms; + pfvf_ops->get_pf2vf_offset = adf_gen4_pf_get_pf2vf_offset; + pfvf_ops->get_vf2pf_offset = adf_gen4_pf_get_vf2pf_offset; + pfvf_ops->enable_vf2pf_interrupts = adf_gen4_enable_vf2pf_interrupts; + pfvf_ops->disable_all_vf2pf_interrupts = adf_gen4_disable_all_vf2pf_interrupts; + pfvf_ops->disable_pending_vf2pf_interrupts = adf_gen4_disable_pending_vf2pf_interrupts; + pfvf_ops->send_msg = adf_gen4_pfvf_send; + pfvf_ops->recv_msg = adf_gen4_pfvf_recv; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_pf_pfvf_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_pfvf.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_pfvf.h new file mode 100644 index 000000000000..2c8708117f70 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_pfvf.h @@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2021 Intel Corporation */ +#ifndef ADF_GEN4_PFVF_H +#define ADF_GEN4_PFVF_H + +#include "adf_accel_devices.h" +#include "adf_common_drv.h" + +#ifdef CONFIG_PCI_IOV +void adf_gen4_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops); +#else +static inline void adf_gen4_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops) +{ + pfvf_ops->enable_comms = adf_pfvf_comms_disabled; +} +#endif + +#endif /* ADF_GEN4_PFVF_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_pm.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm.c new file mode 100644 index 000000000000..5dafd9a270db --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm.c @@ -0,0 +1,164 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2022 Intel Corporation */ +#include <linux/bitfield.h> +#include <linux/iopoll.h> +#include <linux/kernel.h> + +#include "adf_accel_devices.h" +#include "adf_admin.h" +#include "adf_common_drv.h" +#include "adf_gen4_pm.h" +#include "adf_cfg_strings.h" +#include "icp_qat_fw_init_admin.h" +#include "adf_gen4_hw_data.h" +#include "adf_cfg.h" + +struct adf_gen4_pm_data { + struct work_struct pm_irq_work; + struct adf_accel_dev *accel_dev; + u32 pm_int_sts; +}; + +static int send_host_msg(struct adf_accel_dev *accel_dev) +{ + char pm_idle_support_cfg[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = {}; + void __iomem *pmisc = adf_get_pmisc_base(accel_dev); + struct adf_pm *pm = &accel_dev->power_management; + bool pm_idle_support; + u32 msg; + int ret; + + msg = ADF_CSR_RD(pmisc, ADF_GEN4_PM_HOST_MSG); + if (msg & ADF_GEN4_PM_MSG_PENDING) + return -EBUSY; + + adf_cfg_get_param_value(accel_dev, ADF_GENERAL_SEC, + ADF_PM_IDLE_SUPPORT, pm_idle_support_cfg); + ret = kstrtobool(pm_idle_support_cfg, &pm_idle_support); + if (ret) + pm_idle_support = true; + + if (pm_idle_support) + pm->host_ack_counter++; + else + pm->host_nack_counter++; + + /* Send HOST_MSG */ + msg = FIELD_PREP(ADF_GEN4_PM_MSG_PAYLOAD_BIT_MASK, + pm_idle_support ? PM_SET_MIN : PM_NO_CHANGE); + msg |= ADF_GEN4_PM_MSG_PENDING; + ADF_CSR_WR(pmisc, ADF_GEN4_PM_HOST_MSG, msg); + + /* Poll status register to make sure the HOST_MSG has been processed */ + return read_poll_timeout(ADF_CSR_RD, msg, + !(msg & ADF_GEN4_PM_MSG_PENDING), + ADF_GEN4_PM_MSG_POLL_DELAY_US, + ADF_GEN4_PM_POLL_TIMEOUT_US, true, pmisc, + ADF_GEN4_PM_HOST_MSG); +} + +static void pm_bh_handler(struct work_struct *work) +{ + struct adf_gen4_pm_data *pm_data = + container_of(work, struct adf_gen4_pm_data, pm_irq_work); + struct adf_accel_dev *accel_dev = pm_data->accel_dev; + void __iomem *pmisc = adf_get_pmisc_base(accel_dev); + struct adf_pm *pm = &accel_dev->power_management; + u32 pm_int_sts = pm_data->pm_int_sts; + u32 val; + + /* PM Idle interrupt */ + if (pm_int_sts & ADF_GEN4_PM_IDLE_STS) { + pm->idle_irq_counters++; + /* Issue host message to FW */ + if (send_host_msg(accel_dev)) + dev_warn_ratelimited(&GET_DEV(accel_dev), + "Failed to send host msg to FW\n"); + } + + /* PM throttle interrupt */ + if (pm_int_sts & ADF_GEN4_PM_THR_STS) + pm->throttle_irq_counters++; + + /* PM fw interrupt */ + if (pm_int_sts & ADF_GEN4_PM_FW_INT_STS) + pm->fw_irq_counters++; + + /* Clear interrupt status */ + ADF_CSR_WR(pmisc, ADF_GEN4_PM_INTERRUPT, pm_int_sts); + + /* Reenable PM interrupt */ + val = ADF_CSR_RD(pmisc, ADF_GEN4_ERRMSK2); + val &= ~ADF_GEN4_PM_SOU; + ADF_CSR_WR(pmisc, ADF_GEN4_ERRMSK2, val); + + kfree(pm_data); +} + +bool adf_gen4_handle_pm_interrupt(struct adf_accel_dev *accel_dev) +{ + void __iomem *pmisc = adf_get_pmisc_base(accel_dev); + struct adf_gen4_pm_data *pm_data = NULL; + u32 errsou2; + u32 errmsk2; + u32 val; + + /* Only handle the interrupt triggered by PM */ + errmsk2 = ADF_CSR_RD(pmisc, ADF_GEN4_ERRMSK2); + if (errmsk2 & ADF_GEN4_PM_SOU) + return false; + + errsou2 = ADF_CSR_RD(pmisc, ADF_GEN4_ERRSOU2); + if (!(errsou2 & ADF_GEN4_PM_SOU)) + return false; + + /* Disable interrupt */ + val = ADF_CSR_RD(pmisc, ADF_GEN4_ERRMSK2); + val |= ADF_GEN4_PM_SOU; + ADF_CSR_WR(pmisc, ADF_GEN4_ERRMSK2, val); + + val = ADF_CSR_RD(pmisc, ADF_GEN4_PM_INTERRUPT); + + pm_data = kzalloc(sizeof(*pm_data), GFP_ATOMIC); + if (!pm_data) + return false; + + pm_data->pm_int_sts = val; + pm_data->accel_dev = accel_dev; + + INIT_WORK(&pm_data->pm_irq_work, pm_bh_handler); + adf_misc_wq_queue_work(&pm_data->pm_irq_work); + + return true; +} +EXPORT_SYMBOL_GPL(adf_gen4_handle_pm_interrupt); + +int adf_gen4_enable_pm(struct adf_accel_dev *accel_dev) +{ + void __iomem *pmisc = adf_get_pmisc_base(accel_dev); + int ret; + u32 val; + + ret = adf_init_admin_pm(accel_dev, ADF_GEN4_PM_DEFAULT_IDLE_FILTER); + if (ret) + return ret; + + /* Initialize PM internal data */ + adf_gen4_init_dev_pm_data(accel_dev); + + /* Enable default PM interrupts: IDLE, THROTTLE */ + val = ADF_CSR_RD(pmisc, ADF_GEN4_PM_INTERRUPT); + val |= ADF_GEN4_PM_INT_EN_DEFAULT; + + /* Clear interrupt status */ + val |= ADF_GEN4_PM_INT_STS_MASK; + ADF_CSR_WR(pmisc, ADF_GEN4_PM_INTERRUPT, val); + + /* Unmask PM Interrupt */ + val = ADF_CSR_RD(pmisc, ADF_GEN4_ERRMSK2); + val &= ~ADF_GEN4_PM_SOU; + ADF_CSR_WR(pmisc, ADF_GEN4_ERRMSK2, val); + + return 0; +} +EXPORT_SYMBOL_GPL(adf_gen4_enable_pm); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_pm.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm.h new file mode 100644 index 000000000000..a49352b79a7a --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm.h @@ -0,0 +1,93 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef ADF_GEN4_PM_H +#define ADF_GEN4_PM_H + +#include <linux/bits.h> + +struct adf_accel_dev; + +enum qat_pm_host_msg { + PM_NO_CHANGE = 0, + PM_SET_MIN, +}; + +/* Power management registers */ +#define ADF_GEN4_PM_HOST_MSG (0x50A01C) + +/* Power management */ +#define ADF_GEN4_PM_POLL_DELAY_US 20 +#define ADF_GEN4_PM_POLL_TIMEOUT_US USEC_PER_SEC +#define ADF_GEN4_PM_MSG_POLL_DELAY_US (10 * USEC_PER_MSEC) +#define ADF_GEN4_PM_STATUS (0x50A00C) +#define ADF_GEN4_PM_INTERRUPT (0x50A028) + +/* Power management source in ERRSOU2 and ERRMSK2 */ +#define ADF_GEN4_PM_SOU BIT(18) + +#define ADF_GEN4_PM_IDLE_INT_EN BIT(18) +#define ADF_GEN4_PM_THROTTLE_INT_EN BIT(19) +#define ADF_GEN4_PM_DRV_ACTIVE BIT(20) +#define ADF_GEN4_PM_INIT_STATE BIT(21) +#define ADF_GEN4_PM_INT_EN_DEFAULT (ADF_GEN4_PM_IDLE_INT_EN | \ + ADF_GEN4_PM_THROTTLE_INT_EN) + +#define ADF_GEN4_PM_THR_STS BIT(0) +#define ADF_GEN4_PM_IDLE_STS BIT(1) +#define ADF_GEN4_PM_FW_INT_STS BIT(2) +#define ADF_GEN4_PM_INT_STS_MASK (ADF_GEN4_PM_THR_STS | \ + ADF_GEN4_PM_IDLE_STS | \ + ADF_GEN4_PM_FW_INT_STS) + +#define ADF_GEN4_PM_MSG_PENDING BIT(0) +#define ADF_GEN4_PM_MSG_PAYLOAD_BIT_MASK GENMASK(28, 1) + +#define ADF_GEN4_PM_DEFAULT_IDLE_FILTER (0x6) +#define ADF_GEN4_PM_MAX_IDLE_FILTER (0x7) +#define ADF_GEN4_PM_DEFAULT_IDLE_SUPPORT (0x1) + +/* PM CSRs fields masks */ +#define ADF_GEN4_PM_DOMAIN_POWER_GATED_MASK GENMASK(15, 0) +#define ADF_GEN4_PM_SSM_PM_ENABLE_MASK GENMASK(15, 0) +#define ADF_GEN4_PM_IDLE_FILTER_MASK GENMASK(5, 3) +#define ADF_GEN4_PM_IDLE_ENABLE_MASK BIT(2) +#define ADF_GEN4_PM_ENABLE_PM_MASK BIT(21) +#define ADF_GEN4_PM_ENABLE_PM_IDLE_MASK BIT(22) +#define ADF_GEN4_PM_ENABLE_DEEP_PM_IDLE_MASK BIT(23) +#define ADF_GEN4_PM_CURRENT_WP_MASK GENMASK(19, 11) +#define ADF_GEN4_PM_CPM_PM_STATE_MASK GENMASK(22, 20) +#define ADF_GEN4_PM_PENDING_WP_MASK GENMASK(31, 23) +#define ADF_GEN4_PM_THR_VALUE_MASK GENMASK(6, 4) +#define ADF_GEN4_PM_MIN_PWR_ACK_MASK BIT(7) +#define ADF_GEN4_PM_MIN_PWR_ACK_PENDING_MASK BIT(17) +#define ADF_GEN4_PM_CPR_ACTIVE_COUNT_MASK BIT(0) +#define ADF_GEN4_PM_CPR_MANAGED_COUNT_MASK BIT(0) +#define ADF_GEN4_PM_XLT_ACTIVE_COUNT_MASK BIT(1) +#define ADF_GEN4_PM_XLT_MANAGED_COUNT_MASK BIT(1) +#define ADF_GEN4_PM_DCPR_ACTIVE_COUNT_MASK GENMASK(3, 2) +#define ADF_GEN4_PM_DCPR_MANAGED_COUNT_MASK GENMASK(3, 2) +#define ADF_GEN4_PM_PKE_ACTIVE_COUNT_MASK GENMASK(8, 4) +#define ADF_GEN4_PM_PKE_MANAGED_COUNT_MASK GENMASK(8, 4) +#define ADF_GEN4_PM_WAT_ACTIVE_COUNT_MASK GENMASK(13, 9) +#define ADF_GEN4_PM_WAT_MANAGED_COUNT_MASK GENMASK(13, 9) +#define ADF_GEN4_PM_WCP_ACTIVE_COUNT_MASK GENMASK(18, 14) +#define ADF_GEN4_PM_WCP_MANAGED_COUNT_MASK GENMASK(18, 14) +#define ADF_GEN4_PM_UCS_ACTIVE_COUNT_MASK GENMASK(20, 19) +#define ADF_GEN4_PM_UCS_MANAGED_COUNT_MASK GENMASK(20, 19) +#define ADF_GEN4_PM_CPH_ACTIVE_COUNT_MASK GENMASK(24, 21) +#define ADF_GEN4_PM_CPH_MANAGED_COUNT_MASK GENMASK(24, 21) +#define ADF_GEN4_PM_ATH_ACTIVE_COUNT_MASK GENMASK(28, 25) +#define ADF_GEN4_PM_ATH_MANAGED_COUNT_MASK GENMASK(28, 25) + +int adf_gen4_enable_pm(struct adf_accel_dev *accel_dev); +bool adf_gen4_handle_pm_interrupt(struct adf_accel_dev *accel_dev); + +#ifdef CONFIG_DEBUG_FS +void adf_gen4_init_dev_pm_data(struct adf_accel_dev *accel_dev); +#else +static inline void adf_gen4_init_dev_pm_data(struct adf_accel_dev *accel_dev) +{ +} +#endif /* CONFIG_DEBUG_FS */ + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c new file mode 100644 index 000000000000..b7e38842a46d --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c @@ -0,0 +1,199 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include <linux/dma-mapping.h> +#include <linux/string_helpers.h> + +#include "adf_accel_devices.h" +#include "adf_admin.h" +#include "adf_common_drv.h" +#include "adf_gen4_pm.h" +#include "adf_pm_dbgfs_utils.h" +#include "icp_qat_fw_init_admin.h" + +#define PM_INFO_REGSET_ENTRY(_reg_, _field_) \ + PM_INFO_REGSET_ENTRY_MASK(_reg_, _field_, ADF_GEN4_PM_##_field_##_MASK) + +static const struct pm_status_row pm_fuse_rows[] = { + PM_INFO_REGSET_ENTRY(fusectl0, ENABLE_PM), + PM_INFO_REGSET_ENTRY(fusectl0, ENABLE_PM_IDLE), + PM_INFO_REGSET_ENTRY(fusectl0, ENABLE_DEEP_PM_IDLE), +}; + +static const struct pm_status_row pm_info_rows[] = { + PM_INFO_REGSET_ENTRY(pm.status, CPM_PM_STATE), + PM_INFO_REGSET_ENTRY(pm.status, PENDING_WP), + PM_INFO_REGSET_ENTRY(pm.status, CURRENT_WP), + PM_INFO_REGSET_ENTRY(pm.fw_init, IDLE_ENABLE), + PM_INFO_REGSET_ENTRY(pm.fw_init, IDLE_FILTER), + PM_INFO_REGSET_ENTRY(pm.main, MIN_PWR_ACK), + PM_INFO_REGSET_ENTRY(pm.thread, MIN_PWR_ACK_PENDING), + PM_INFO_REGSET_ENTRY(pm.main, THR_VALUE), +}; + +static const struct pm_status_row pm_ssm_rows[] = { + PM_INFO_REGSET_ENTRY(ssm.pm_enable, SSM_PM_ENABLE), + PM_INFO_REGSET_ENTRY32(ssm.active_constraint, ACTIVE_CONSTRAINT), + PM_INFO_REGSET_ENTRY(ssm.pm_domain_status, DOMAIN_POWER_GATED), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, ATH_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, CPH_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, PKE_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, CPR_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, DCPR_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, UCS_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, XLT_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, WAT_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_active_status, WCP_ACTIVE_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, ATH_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, CPH_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, PKE_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, CPR_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, DCPR_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, UCS_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, XLT_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, WAT_MANAGED_COUNT), + PM_INFO_REGSET_ENTRY(ssm.pm_managed_status, WCP_MANAGED_COUNT), +}; + +static const struct pm_status_row pm_log_rows[] = { + PM_INFO_REGSET_ENTRY32(event_counters.host_msg, HOST_MSG_EVENT_COUNT), + PM_INFO_REGSET_ENTRY32(event_counters.sys_pm, SYS_PM_EVENT_COUNT), + PM_INFO_REGSET_ENTRY32(event_counters.local_ssm, SSM_EVENT_COUNT), + PM_INFO_REGSET_ENTRY32(event_counters.timer, TIMER_EVENT_COUNT), + PM_INFO_REGSET_ENTRY32(event_counters.unknown, UNKNOWN_EVENT_COUNT), +}; + +static const struct pm_status_row pm_event_rows[ICP_QAT_NUMBER_OF_PM_EVENTS] = { + PM_INFO_REGSET_ENTRY32(event_log[0], EVENT0), + PM_INFO_REGSET_ENTRY32(event_log[1], EVENT1), + PM_INFO_REGSET_ENTRY32(event_log[2], EVENT2), + PM_INFO_REGSET_ENTRY32(event_log[3], EVENT3), + PM_INFO_REGSET_ENTRY32(event_log[4], EVENT4), + PM_INFO_REGSET_ENTRY32(event_log[5], EVENT5), + PM_INFO_REGSET_ENTRY32(event_log[6], EVENT6), + PM_INFO_REGSET_ENTRY32(event_log[7], EVENT7), +}; + +static const struct pm_status_row pm_csrs_rows[] = { + PM_INFO_REGSET_ENTRY32(pm.fw_init, CPM_PM_FW_INIT), + PM_INFO_REGSET_ENTRY32(pm.status, CPM_PM_STATUS), + PM_INFO_REGSET_ENTRY32(pm.main, CPM_PM_MASTER_FW), + PM_INFO_REGSET_ENTRY32(pm.pwrreq, CPM_PM_PWRREQ), +}; + +static_assert(sizeof(struct icp_qat_fw_init_admin_pm_info) < PAGE_SIZE); + +static ssize_t adf_gen4_print_pm_status(struct adf_accel_dev *accel_dev, + char __user *buf, size_t count, + loff_t *pos) +{ + void __iomem *pmisc = adf_get_pmisc_base(accel_dev); + struct adf_pm *pm = &accel_dev->power_management; + struct icp_qat_fw_init_admin_pm_info *pm_info; + dma_addr_t p_state_addr; + u32 *pm_info_regs; + char *pm_kv; + int len = 0; + u32 val; + int ret; + + pm_info = kmalloc(PAGE_SIZE, GFP_KERNEL); + if (!pm_info) + return -ENOMEM; + + pm_kv = kmalloc(PAGE_SIZE, GFP_KERNEL); + if (!pm_kv) { + ret = -ENOMEM; + goto out_free; + } + + p_state_addr = dma_map_single(&GET_DEV(accel_dev), pm_info, PAGE_SIZE, + DMA_FROM_DEVICE); + ret = dma_mapping_error(&GET_DEV(accel_dev), p_state_addr); + if (ret) + goto out_free; + + /* Query PM info from QAT FW */ + ret = adf_get_pm_info(accel_dev, p_state_addr, PAGE_SIZE); + dma_unmap_single(&GET_DEV(accel_dev), p_state_addr, PAGE_SIZE, + DMA_FROM_DEVICE); + if (ret) + goto out_free; + + pm_info_regs = (u32 *)pm_info; + + /* Fusectl related */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "----------- PM Fuse info ---------\n"); + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_fuse_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_fuse_rows)); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "max_pwrreq: %#x\n", + pm_info->max_pwrreq); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "min_pwrreq: %#x\n", + pm_info->min_pwrreq); + + /* PM related */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "------------ PM Info ------------\n"); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "power_level: %s\n", + pm_info->pwr_state == PM_SET_MIN ? "min" : "max"); + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_info_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_info_rows)); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "pm_mode: STATIC\n"); + + /* SSM related */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "----------- SSM_PM Info ----------\n"); + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_ssm_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_ssm_rows)); + + /* Log related */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "------------- PM Log -------------\n"); + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_log_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_log_rows)); + + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_event_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_event_rows)); + + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "idle_irq_count: %#x\n", + pm->idle_irq_counters); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "fw_irq_count: %#x\n", + pm->fw_irq_counters); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "throttle_irq_count: %#x\n", pm->throttle_irq_counters); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "host_ack_count: %#x\n", + pm->host_ack_counter); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "host_nack_count: %#x\n", + pm->host_nack_counter); + + /* CSRs content */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "----------- HW PM CSRs -----------\n"); + len += adf_pm_scnprint_table_upper_keys(&pm_kv[len], pm_csrs_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_csrs_rows)); + + val = ADF_CSR_RD(pmisc, ADF_GEN4_PM_HOST_MSG); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "CPM_PM_HOST_MSG: %#x\n", val); + val = ADF_CSR_RD(pmisc, ADF_GEN4_PM_INTERRUPT); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "CPM_PM_INTERRUPT: %#x\n", val); + ret = simple_read_from_buffer(buf, count, pos, pm_kv, len); + +out_free: + kfree(pm_info); + kfree(pm_kv); + return ret; +} + +void adf_gen4_init_dev_pm_data(struct adf_accel_dev *accel_dev) +{ + accel_dev->power_management.print_pm_status = adf_gen4_print_pm_status; + accel_dev->power_management.present = true; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_ras.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_ras.c new file mode 100644 index 000000000000..0f7f00a19e7d --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_ras.c @@ -0,0 +1,1517 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include "adf_common_drv.h" +#include "adf_gen4_hw_data.h" +#include "adf_gen4_ras.h" +#include "adf_sysfs_ras_counters.h" + +#define BITS_PER_REG(_n_) (sizeof(_n_) * BITS_PER_BYTE) + +static void enable_errsou_reporting(void __iomem *csr) +{ + /* Enable correctable error reporting in ERRSOU0 */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK0, 0); + + /* Enable uncorrectable error reporting in ERRSOU1 */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK1, 0); + + /* + * Enable uncorrectable error reporting in ERRSOU2 + * but disable PM interrupt and CFC attention interrupt by default + */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK2, + ADF_GEN4_ERRSOU2_PM_INT_BIT | + ADF_GEN4_ERRSOU2_CPP_CFC_ATT_INT_BITMASK); + + /* + * Enable uncorrectable error reporting in ERRSOU3 + * but disable RLT error interrupt and VFLR notify interrupt by default + */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK3, + ADF_GEN4_ERRSOU3_RLTERROR_BIT | + ADF_GEN4_ERRSOU3_VFLRNOTIFY_BIT); +} + +static void disable_errsou_reporting(void __iomem *csr) +{ + u32 val = 0; + + /* Disable correctable error reporting in ERRSOU0 */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK0, ADF_GEN4_ERRSOU0_BIT); + + /* Disable uncorrectable error reporting in ERRSOU1 */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK1, ADF_GEN4_ERRSOU1_BITMASK); + + /* Disable uncorrectable error reporting in ERRSOU2 */ + val = ADF_CSR_RD(csr, ADF_GEN4_ERRMSK2); + val |= ADF_GEN4_ERRSOU2_DIS_BITMASK; + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK2, val); + + /* Disable uncorrectable error reporting in ERRSOU3 */ + ADF_CSR_WR(csr, ADF_GEN4_ERRMSK3, ADF_GEN4_ERRSOU3_BITMASK); +} + +static void enable_ae_error_reporting(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + u32 ae_mask = GET_HW_DATA(accel_dev)->ae_mask; + + /* Enable Acceleration Engine correctable error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_HIAECORERRLOGENABLE_CPP0, ae_mask); + + /* Enable Acceleration Engine uncorrectable error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_HIAEUNCERRLOGENABLE_CPP0, ae_mask); +} + +static void disable_ae_error_reporting(void __iomem *csr) +{ + /* Disable Acceleration Engine correctable error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_HIAECORERRLOGENABLE_CPP0, 0); + + /* Disable Acceleration Engine uncorrectable error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_HIAEUNCERRLOGENABLE_CPP0, 0); +} + +static void enable_cpp_error_reporting(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + + /* Enable HI CPP Agents Command Parity Error Reporting */ + ADF_CSR_WR(csr, ADF_GEN4_HICPPAGENTCMDPARERRLOGENABLE, + err_mask->cppagentcmdpar_mask); + + ADF_CSR_WR(csr, ADF_GEN4_CPP_CFC_ERR_CTRL, + ADF_GEN4_CPP_CFC_ERR_CTRL_BITMASK); +} + +static void disable_cpp_error_reporting(void __iomem *csr) +{ + /* Disable HI CPP Agents Command Parity Error Reporting */ + ADF_CSR_WR(csr, ADF_GEN4_HICPPAGENTCMDPARERRLOGENABLE, 0); + + ADF_CSR_WR(csr, ADF_GEN4_CPP_CFC_ERR_CTRL, + ADF_GEN4_CPP_CFC_ERR_CTRL_DIS_BITMASK); +} + +static void enable_ti_ri_error_reporting(void __iomem *csr) +{ + u32 reg; + + /* Enable RI Memory error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_RI_MEM_PAR_ERR_EN0, + ADF_GEN4_RIMEM_PARERR_STS_FATAL_BITMASK | + ADF_GEN4_RIMEM_PARERR_STS_UNCERR_BITMASK); + + /* Enable IOSF Primary Command Parity error Reporting */ + ADF_CSR_WR(csr, ADF_GEN4_RIMISCCTL, ADF_GEN4_RIMISCSTS_BIT); + + /* Enable TI Internal Memory Parity Error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_TI_CI_PAR_ERR_MASK, 0); + ADF_CSR_WR(csr, ADF_GEN4_TI_PULL0FUB_PAR_ERR_MASK, 0); + ADF_CSR_WR(csr, ADF_GEN4_TI_PUSHFUB_PAR_ERR_MASK, 0); + ADF_CSR_WR(csr, ADF_GEN4_TI_CD_PAR_ERR_MASK, 0); + ADF_CSR_WR(csr, ADF_GEN4_TI_TRNSB_PAR_ERR_MASK, 0); + + /* Enable error handling in RI, TI CPP interface control registers */ + ADF_CSR_WR(csr, ADF_GEN4_RICPPINTCTL, ADF_GEN4_RICPPINTCTL_BITMASK); + + ADF_CSR_WR(csr, ADF_GEN4_TICPPINTCTL, ADF_GEN4_TICPPINTCTL_BITMASK); + + /* + * Enable error detection and reporting in TIMISCSTS + * with bits 1, 2 and 30 value preserved + */ + reg = ADF_CSR_RD(csr, ADF_GEN4_TIMISCCTL); + reg &= ADF_GEN4_TIMSCCTL_RELAY_BITMASK; + reg |= ADF_GEN4_TIMISCCTL_BIT; + ADF_CSR_WR(csr, ADF_GEN4_TIMISCCTL, reg); +} + +static void disable_ti_ri_error_reporting(void __iomem *csr) +{ + u32 reg; + + /* Disable RI Memory error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_RI_MEM_PAR_ERR_EN0, 0); + + /* Disable IOSF Primary Command Parity error Reporting */ + ADF_CSR_WR(csr, ADF_GEN4_RIMISCCTL, 0); + + /* Disable TI Internal Memory Parity Error reporting */ + ADF_CSR_WR(csr, ADF_GEN4_TI_CI_PAR_ERR_MASK, + ADF_GEN4_TI_CI_PAR_STS_BITMASK); + ADF_CSR_WR(csr, ADF_GEN4_TI_PULL0FUB_PAR_ERR_MASK, + ADF_GEN4_TI_PULL0FUB_PAR_STS_BITMASK); + ADF_CSR_WR(csr, ADF_GEN4_TI_PUSHFUB_PAR_ERR_MASK, + ADF_GEN4_TI_PUSHFUB_PAR_STS_BITMASK); + ADF_CSR_WR(csr, ADF_GEN4_TI_CD_PAR_ERR_MASK, + ADF_GEN4_TI_CD_PAR_STS_BITMASK); + ADF_CSR_WR(csr, ADF_GEN4_TI_TRNSB_PAR_ERR_MASK, + ADF_GEN4_TI_TRNSB_PAR_STS_BITMASK); + + /* Disable error handling in RI, TI CPP interface control registers */ + ADF_CSR_WR(csr, ADF_GEN4_RICPPINTCTL, 0); + + ADF_CSR_WR(csr, ADF_GEN4_TICPPINTCTL, 0); + + /* + * Disable error detection and reporting in TIMISCSTS + * with bits 1, 2 and 30 value preserved + */ + reg = ADF_CSR_RD(csr, ADF_GEN4_TIMISCCTL); + reg &= ADF_GEN4_TIMSCCTL_RELAY_BITMASK; + ADF_CSR_WR(csr, ADF_GEN4_TIMISCCTL, reg); +} + +static void enable_rf_error_reporting(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + + /* Enable RF parity error in Shared RAM */ + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_SRC, 0); + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_ATH_CPH, 0); + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_CPR_XLT, 0); + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_DCPR_UCS, 0); + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_PKE, 0); + + if (err_mask->parerr_wat_wcp_mask) + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_WAT_WCP, 0); +} + +static void disable_rf_error_reporting(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + + /* Disable RF Parity Error reporting in Shared RAM */ + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_SRC, + ADF_GEN4_SSMSOFTERRORPARITY_SRC_BIT); + + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_ATH_CPH, + err_mask->parerr_ath_cph_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_CPR_XLT, + err_mask->parerr_cpr_xlt_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_DCPR_UCS, + err_mask->parerr_dcpr_ucs_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_PKE, + err_mask->parerr_pke_mask); + + if (err_mask->parerr_wat_wcp_mask) + ADF_CSR_WR(csr, ADF_GEN4_SSMSOFTERRORPARITYMASK_WAT_WCP, + err_mask->parerr_wat_wcp_mask); +} + +static void enable_ssm_error_reporting(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + u32 val = 0; + + /* Enable SSM interrupts */ + ADF_CSR_WR(csr, ADF_GEN4_INTMASKSSM, 0); + + /* Enable shared memory error detection & correction */ + val = ADF_CSR_RD(csr, ADF_GEN4_SSMFEATREN); + val |= err_mask->ssmfeatren_mask; + ADF_CSR_WR(csr, ADF_GEN4_SSMFEATREN, val); + + /* Enable SER detection in SER_err_ssmsh register */ + ADF_CSR_WR(csr, ADF_GEN4_SER_EN_SSMSH, + ADF_GEN4_SER_EN_SSMSH_BITMASK); + + /* Enable SSM soft parity error */ + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_ATH_CPH, 0); + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_CPR_XLT, 0); + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_DCPR_UCS, 0); + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_PKE, 0); + + if (err_mask->parerr_wat_wcp_mask) + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_WAT_WCP, 0); + + /* Enable slice hang interrupt reporting */ + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_ATH_CPH, 0); + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_CPR_XLT, 0); + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_DCPR_UCS, 0); + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_PKE, 0); + + if (err_mask->parerr_wat_wcp_mask) + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_WAT_WCP, 0); +} + +static void disable_ssm_error_reporting(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + u32 val = 0; + + /* Disable SSM interrupts */ + ADF_CSR_WR(csr, ADF_GEN4_INTMASKSSM, + ADF_GEN4_INTMASKSSM_BITMASK); + + /* Disable shared memory error detection & correction */ + val = ADF_CSR_RD(csr, ADF_GEN4_SSMFEATREN); + val &= ADF_GEN4_SSMFEATREN_DIS_BITMASK; + ADF_CSR_WR(csr, ADF_GEN4_SSMFEATREN, val); + + /* Disable SER detection in SER_err_ssmsh register */ + ADF_CSR_WR(csr, ADF_GEN4_SER_EN_SSMSH, 0); + + /* Disable SSM soft parity error */ + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_ATH_CPH, + err_mask->parerr_ath_cph_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_CPR_XLT, + err_mask->parerr_cpr_xlt_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_DCPR_UCS, + err_mask->parerr_dcpr_ucs_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_PKE, + err_mask->parerr_pke_mask); + + if (err_mask->parerr_wat_wcp_mask) + ADF_CSR_WR(csr, ADF_GEN4_SPPPARERRMSK_WAT_WCP, + err_mask->parerr_wat_wcp_mask); + + /* Disable slice hang interrupt reporting */ + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_ATH_CPH, + err_mask->parerr_ath_cph_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_CPR_XLT, + err_mask->parerr_cpr_xlt_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_DCPR_UCS, + err_mask->parerr_dcpr_ucs_mask); + + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_PKE, + err_mask->parerr_pke_mask); + + if (err_mask->parerr_wat_wcp_mask) + ADF_CSR_WR(csr, ADF_GEN4_SHINTMASKSSM_WAT_WCP, + err_mask->parerr_wat_wcp_mask); +} + +static void enable_aram_error_reporting(void __iomem *csr) +{ + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMCERRUERR_EN, + ADF_GEN4_REG_ARAMCERRUERR_EN_BITMASK); + + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMCERR, + ADF_GEN4_REG_ARAMCERR_EN_BITMASK); + + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMUERR, + ADF_GEN4_REG_ARAMUERR_EN_BITMASK); + + ADF_CSR_WR(csr, ADF_GEN4_REG_CPPMEMTGTERR, + ADF_GEN4_REG_CPPMEMTGTERR_EN_BITMASK); +} + +static void disable_aram_error_reporting(void __iomem *csr) +{ + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMCERRUERR_EN, 0); + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMCERR, 0); + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMUERR, 0); + ADF_CSR_WR(csr, ADF_GEN4_REG_CPPMEMTGTERR, 0); +} + +static void adf_gen4_enable_ras(struct adf_accel_dev *accel_dev) +{ + void __iomem *aram_csr = adf_get_aram_base(accel_dev); + void __iomem *csr = adf_get_pmisc_base(accel_dev); + + enable_errsou_reporting(csr); + enable_ae_error_reporting(accel_dev, csr); + enable_cpp_error_reporting(accel_dev, csr); + enable_ti_ri_error_reporting(csr); + enable_rf_error_reporting(accel_dev, csr); + enable_ssm_error_reporting(accel_dev, csr); + enable_aram_error_reporting(aram_csr); +} + +static void adf_gen4_disable_ras(struct adf_accel_dev *accel_dev) +{ + void __iomem *aram_csr = adf_get_aram_base(accel_dev); + void __iomem *csr = adf_get_pmisc_base(accel_dev); + + disable_errsou_reporting(csr); + disable_ae_error_reporting(csr); + disable_cpp_error_reporting(csr); + disable_ti_ri_error_reporting(csr); + disable_rf_error_reporting(accel_dev, csr); + disable_ssm_error_reporting(accel_dev, csr); + disable_aram_error_reporting(aram_csr); +} + +static void adf_gen4_process_errsou0(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + u32 aecorrerr = ADF_CSR_RD(csr, ADF_GEN4_HIAECORERRLOG_CPP0); + + aecorrerr &= GET_HW_DATA(accel_dev)->ae_mask; + + dev_warn(&GET_DEV(accel_dev), + "Correctable error detected in AE: 0x%x\n", + aecorrerr); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + + /* Clear interrupt from ERRSOU0 */ + ADF_CSR_WR(csr, ADF_GEN4_HIAECORERRLOG_CPP0, aecorrerr); +} + +static bool adf_handle_cpp_aeunc(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 aeuncorerr; + + if (!(errsou & ADF_GEN4_ERRSOU1_HIAEUNCERRLOG_CPP0_BIT)) + return false; + + aeuncorerr = ADF_CSR_RD(csr, ADF_GEN4_HIAEUNCERRLOG_CPP0); + aeuncorerr &= GET_HW_DATA(accel_dev)->ae_mask; + + dev_err(&GET_DEV(accel_dev), + "Uncorrectable error detected in AE: 0x%x\n", + aeuncorerr); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_HIAEUNCERRLOG_CPP0, aeuncorerr); + + return false; +} + +static bool adf_handle_cppcmdparerr(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + u32 cmdparerr; + + if (!(errsou & ADF_GEN4_ERRSOU1_HICPPAGENTCMDPARERRLOG_BIT)) + return false; + + cmdparerr = ADF_CSR_RD(csr, ADF_GEN4_HICPPAGENTCMDPARERRLOG); + cmdparerr &= err_mask->cppagentcmdpar_mask; + + dev_err(&GET_DEV(accel_dev), + "HI CPP agent command parity error: 0x%x\n", + cmdparerr); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_HICPPAGENTCMDPARERRLOG, cmdparerr); + + return true; +} + +static bool adf_handle_ri_mem_par_err(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + bool reset_required = false; + u32 rimem_parerr_sts; + + if (!(errsou & ADF_GEN4_ERRSOU1_RIMEM_PARERR_STS_BIT)) + return false; + + rimem_parerr_sts = ADF_CSR_RD(csr, ADF_GEN4_RIMEM_PARERR_STS); + rimem_parerr_sts &= ADF_GEN4_RIMEM_PARERR_STS_UNCERR_BITMASK | + ADF_GEN4_RIMEM_PARERR_STS_FATAL_BITMASK; + + if (rimem_parerr_sts & ADF_GEN4_RIMEM_PARERR_STS_UNCERR_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "RI Memory Parity uncorrectable error: 0x%x\n", + rimem_parerr_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + if (rimem_parerr_sts & ADF_GEN4_RIMEM_PARERR_STS_FATAL_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "RI Memory Parity fatal error: 0x%x\n", + rimem_parerr_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + reset_required = true; + } + + ADF_CSR_WR(csr, ADF_GEN4_RIMEM_PARERR_STS, rimem_parerr_sts); + + return reset_required; +} + +static bool adf_handle_ti_ci_par_sts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 ti_ci_par_sts; + + if (!(errsou & ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT)) + return false; + + ti_ci_par_sts = ADF_CSR_RD(csr, ADF_GEN4_TI_CI_PAR_STS); + ti_ci_par_sts &= ADF_GEN4_TI_CI_PAR_STS_BITMASK; + + if (ti_ci_par_sts) { + dev_err(&GET_DEV(accel_dev), + "TI Memory Parity Error: 0x%x\n", ti_ci_par_sts); + ADF_CSR_WR(csr, ADF_GEN4_TI_CI_PAR_STS, ti_ci_par_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + return false; +} + +static bool adf_handle_ti_pullfub_par_sts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 ti_pullfub_par_sts; + + if (!(errsou & ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT)) + return false; + + ti_pullfub_par_sts = ADF_CSR_RD(csr, ADF_GEN4_TI_PULL0FUB_PAR_STS); + ti_pullfub_par_sts &= ADF_GEN4_TI_PULL0FUB_PAR_STS_BITMASK; + + if (ti_pullfub_par_sts) { + dev_err(&GET_DEV(accel_dev), + "TI Pull Parity Error: 0x%x\n", ti_pullfub_par_sts); + + ADF_CSR_WR(csr, ADF_GEN4_TI_PULL0FUB_PAR_STS, + ti_pullfub_par_sts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + return false; +} + +static bool adf_handle_ti_pushfub_par_sts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 ti_pushfub_par_sts; + + if (!(errsou & ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT)) + return false; + + ti_pushfub_par_sts = ADF_CSR_RD(csr, ADF_GEN4_TI_PUSHFUB_PAR_STS); + ti_pushfub_par_sts &= ADF_GEN4_TI_PUSHFUB_PAR_STS_BITMASK; + + if (ti_pushfub_par_sts) { + dev_err(&GET_DEV(accel_dev), + "TI Push Parity Error: 0x%x\n", ti_pushfub_par_sts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_TI_PUSHFUB_PAR_STS, + ti_pushfub_par_sts); + } + + return false; +} + +static bool adf_handle_ti_cd_par_sts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 ti_cd_par_sts; + + if (!(errsou & ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT)) + return false; + + ti_cd_par_sts = ADF_CSR_RD(csr, ADF_GEN4_TI_CD_PAR_STS); + ti_cd_par_sts &= ADF_GEN4_TI_CD_PAR_STS_BITMASK; + + if (ti_cd_par_sts) { + dev_err(&GET_DEV(accel_dev), + "TI CD Parity Error: 0x%x\n", ti_cd_par_sts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_TI_CD_PAR_STS, ti_cd_par_sts); + } + + return false; +} + +static bool adf_handle_ti_trnsb_par_sts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 ti_trnsb_par_sts; + + if (!(errsou & ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT)) + return false; + + ti_trnsb_par_sts = ADF_CSR_RD(csr, ADF_GEN4_TI_TRNSB_PAR_STS); + ti_trnsb_par_sts &= ADF_GEN4_TI_TRNSB_PAR_STS_BITMASK; + + if (ti_trnsb_par_sts) { + dev_err(&GET_DEV(accel_dev), + "TI TRNSB Parity Error: 0x%x\n", ti_trnsb_par_sts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_TI_TRNSB_PAR_STS, ti_trnsb_par_sts); + } + + return false; +} + +static bool adf_handle_iosfp_cmd_parerr(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 rimiscsts; + + if (!(errsou & ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT)) + return false; + + rimiscsts = ADF_CSR_RD(csr, ADF_GEN4_RIMISCSTS); + rimiscsts &= ADF_GEN4_RIMISCSTS_BIT; + + dev_err(&GET_DEV(accel_dev), + "Command Parity error detected on IOSFP: 0x%x\n", + rimiscsts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_RIMISCSTS, rimiscsts); + + return true; +} + +static void adf_gen4_process_errsou1(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou, + bool *reset_required) +{ + *reset_required |= adf_handle_cpp_aeunc(accel_dev, csr, errsou); + *reset_required |= adf_handle_cppcmdparerr(accel_dev, csr, errsou); + *reset_required |= adf_handle_ri_mem_par_err(accel_dev, csr, errsou); + *reset_required |= adf_handle_ti_ci_par_sts(accel_dev, csr, errsou); + *reset_required |= adf_handle_ti_pullfub_par_sts(accel_dev, csr, errsou); + *reset_required |= adf_handle_ti_pushfub_par_sts(accel_dev, csr, errsou); + *reset_required |= adf_handle_ti_cd_par_sts(accel_dev, csr, errsou); + *reset_required |= adf_handle_ti_trnsb_par_sts(accel_dev, csr, errsou); + *reset_required |= adf_handle_iosfp_cmd_parerr(accel_dev, csr, errsou); +} + +static bool adf_handle_uerrssmsh(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN4_IAINTSTATSSM_UERRSSMSH_BIT)) + return false; + + reg = ADF_CSR_RD(csr, ADF_GEN4_UERRSSMSH); + reg &= ADF_GEN4_UERRSSMSH_BITMASK; + + dev_err(&GET_DEV(accel_dev), + "Uncorrectable error on ssm shared memory: 0x%x\n", + reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_UERRSSMSH, reg); + + return false; +} + +static bool adf_handle_cerrssmsh(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN4_IAINTSTATSSM_CERRSSMSH_BIT)) + return false; + + reg = ADF_CSR_RD(csr, ADF_GEN4_CERRSSMSH); + reg &= ADF_GEN4_CERRSSMSH_ERROR_BIT; + + dev_warn(&GET_DEV(accel_dev), + "Correctable error on ssm shared memory: 0x%x\n", + reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + + ADF_CSR_WR(csr, ADF_GEN4_CERRSSMSH, reg); + + return false; +} + +static bool adf_handle_pperr_err(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN4_IAINTSTATSSM_PPERR_BIT)) + return false; + + reg = ADF_CSR_RD(csr, ADF_GEN4_PPERR); + reg &= ADF_GEN4_PPERR_BITMASK; + + dev_err(&GET_DEV(accel_dev), + "Uncorrectable error CPP transaction on memory target: 0x%x\n", + reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_PPERR, reg); + + return false; +} + +static void adf_poll_slicehang_csr(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 slice_hang_offset, + char *slice_name) +{ + u32 slice_hang_reg = ADF_CSR_RD(csr, slice_hang_offset); + + if (!slice_hang_reg) + return; + + dev_err(&GET_DEV(accel_dev), + "Slice %s hang error encountered\n", slice_name); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); +} + +static bool adf_handle_slice_hang_error(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + + if (!(iastatssm & ADF_GEN4_IAINTSTATSSM_SLICEHANG_ERR_BIT)) + return false; + + adf_poll_slicehang_csr(accel_dev, csr, + ADF_GEN4_SLICEHANGSTATUS_ATH_CPH, "ath_cph"); + adf_poll_slicehang_csr(accel_dev, csr, + ADF_GEN4_SLICEHANGSTATUS_CPR_XLT, "cpr_xlt"); + adf_poll_slicehang_csr(accel_dev, csr, + ADF_GEN4_SLICEHANGSTATUS_DCPR_UCS, "dcpr_ucs"); + adf_poll_slicehang_csr(accel_dev, csr, + ADF_GEN4_SLICEHANGSTATUS_PKE, "pke"); + + if (err_mask->parerr_wat_wcp_mask) + adf_poll_slicehang_csr(accel_dev, csr, + ADF_GEN4_SLICEHANGSTATUS_WAT_WCP, + "wat_wcp"); + + return false; +} + +static bool adf_handle_spp_pullcmd_err(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + bool reset_required = false; + u32 reg; + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLCMDPARERR_ATH_CPH); + reg &= err_mask->parerr_ath_cph_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull command fatal error ATH_CPH: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLCMDPARERR_ATH_CPH, reg); + + reset_required = true; + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLCMDPARERR_CPR_XLT); + reg &= err_mask->parerr_cpr_xlt_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull command fatal error CPR_XLT: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLCMDPARERR_CPR_XLT, reg); + + reset_required = true; + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLCMDPARERR_DCPR_UCS); + reg &= err_mask->parerr_dcpr_ucs_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull command fatal error DCPR_UCS: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLCMDPARERR_DCPR_UCS, reg); + + reset_required = true; + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLCMDPARERR_PKE); + reg &= err_mask->parerr_pke_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull command fatal error PKE: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLCMDPARERR_PKE, reg); + + reset_required = true; + } + + if (err_mask->parerr_wat_wcp_mask) { + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLCMDPARERR_WAT_WCP); + reg &= err_mask->parerr_wat_wcp_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull command fatal error WAT_WCP: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLCMDPARERR_WAT_WCP, reg); + + reset_required = true; + } + } + + return reset_required; +} + +static bool adf_handle_spp_pulldata_err(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + u32 reg; + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLDATAPARERR_ATH_CPH); + reg &= err_mask->parerr_ath_cph_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull data err ATH_CPH: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLDATAPARERR_ATH_CPH, reg); + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLDATAPARERR_CPR_XLT); + reg &= err_mask->parerr_cpr_xlt_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull data err CPR_XLT: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLDATAPARERR_CPR_XLT, reg); + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLDATAPARERR_DCPR_UCS); + reg &= err_mask->parerr_dcpr_ucs_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull data err DCPR_UCS: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLDATAPARERR_DCPR_UCS, reg); + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLDATAPARERR_PKE); + reg &= err_mask->parerr_pke_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull data err PKE: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLDATAPARERR_PKE, reg); + } + + if (err_mask->parerr_wat_wcp_mask) { + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPULLDATAPARERR_WAT_WCP); + reg &= err_mask->parerr_wat_wcp_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP pull data err WAT_WCP: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPULLDATAPARERR_WAT_WCP, reg); + } + } + + return false; +} + +static bool adf_handle_spp_pushcmd_err(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + bool reset_required = false; + u32 reg; + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHCMDPARERR_ATH_CPH); + reg &= err_mask->parerr_ath_cph_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push command fatal error ATH_CPH: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHCMDPARERR_ATH_CPH, reg); + + reset_required = true; + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHCMDPARERR_CPR_XLT); + reg &= err_mask->parerr_cpr_xlt_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push command fatal error CPR_XLT: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHCMDPARERR_CPR_XLT, reg); + + reset_required = true; + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHCMDPARERR_DCPR_UCS); + reg &= err_mask->parerr_dcpr_ucs_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push command fatal error DCPR_UCS: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHCMDPARERR_DCPR_UCS, reg); + + reset_required = true; + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHCMDPARERR_PKE); + reg &= err_mask->parerr_pke_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push command fatal error PKE: 0x%x\n", + reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHCMDPARERR_PKE, reg); + + reset_required = true; + } + + if (err_mask->parerr_wat_wcp_mask) { + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHCMDPARERR_WAT_WCP); + reg &= err_mask->parerr_wat_wcp_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push command fatal error WAT_WCP: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHCMDPARERR_WAT_WCP, reg); + + reset_required = true; + } + } + + return reset_required; +} + +static bool adf_handle_spp_pushdata_err(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + struct adf_dev_err_mask *err_mask = GET_ERR_MASK(accel_dev); + u32 reg; + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHDATAPARERR_ATH_CPH); + reg &= err_mask->parerr_ath_cph_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push data err ATH_CPH: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHDATAPARERR_ATH_CPH, reg); + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHDATAPARERR_CPR_XLT); + reg &= err_mask->parerr_cpr_xlt_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push data err CPR_XLT: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHDATAPARERR_CPR_XLT, reg); + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHDATAPARERR_DCPR_UCS); + reg &= err_mask->parerr_dcpr_ucs_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push data err DCPR_UCS: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHDATAPARERR_DCPR_UCS, reg); + } + + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHDATAPARERR_PKE); + reg &= err_mask->parerr_pke_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push data err PKE: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHDATAPARERR_PKE, reg); + } + + if (err_mask->parerr_wat_wcp_mask) { + reg = ADF_CSR_RD(csr, ADF_GEN4_SPPPUSHDATAPARERR_WAT_WCP); + reg &= err_mask->parerr_wat_wcp_mask; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "SPP push data err WAT_WCP: 0x%x\n", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_SPPPUSHDATAPARERR_WAT_WCP, + reg); + } + } + + return false; +} + +static bool adf_handle_spppar_err(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + bool reset_required; + + if (!(iastatssm & ADF_GEN4_IAINTSTATSSM_SPPPARERR_BIT)) + return false; + + reset_required = adf_handle_spp_pullcmd_err(accel_dev, csr); + reset_required |= adf_handle_spp_pulldata_err(accel_dev, csr); + reset_required |= adf_handle_spp_pushcmd_err(accel_dev, csr); + reset_required |= adf_handle_spp_pushdata_err(accel_dev, csr); + + return reset_required; +} + +static bool adf_handle_ssmcpppar_err(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + u32 reg, bits_num = BITS_PER_REG(reg); + bool reset_required = false; + unsigned long errs_bits; + u32 bit_iterator; + + if (!(iastatssm & ADF_GEN4_IAINTSTATSSM_SSMCPPERR_BIT)) + return false; + + reg = ADF_CSR_RD(csr, ADF_GEN4_SSMCPPERR); + reg &= ADF_GEN4_SSMCPPERR_FATAL_BITMASK | ADF_GEN4_SSMCPPERR_UNCERR_BITMASK; + if (reg & ADF_GEN4_SSMCPPERR_FATAL_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "Fatal SSM CPP parity error: 0x%x\n", reg); + + errs_bits = reg & ADF_GEN4_SSMCPPERR_FATAL_BITMASK; + for_each_set_bit(bit_iterator, &errs_bits, bits_num) { + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + } + reset_required = true; + } + + if (reg & ADF_GEN4_SSMCPPERR_UNCERR_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "non-Fatal SSM CPP parity error: 0x%x\n", reg); + errs_bits = reg & ADF_GEN4_SSMCPPERR_UNCERR_BITMASK; + + for_each_set_bit(bit_iterator, &errs_bits, bits_num) { + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + } + + ADF_CSR_WR(csr, ADF_GEN4_SSMCPPERR, reg); + + return reset_required; +} + +static void adf_handle_rf_parr_err(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + if (!(iastatssm & ADF_GEN4_IAINTSTATSSM_SSMSOFTERRORPARITY_BIT)) + return; + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + dev_err(&GET_DEV(accel_dev), "Slice ssm soft parity error reported"); + + return; +} + +static bool adf_handle_ser_err_ssmsh(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 iastatssm) +{ + u32 reg, bits_num = BITS_PER_REG(reg); + bool reset_required = false; + unsigned long errs_bits; + u32 bit_iterator; + + if (!(iastatssm & (ADF_GEN4_IAINTSTATSSM_SER_ERR_SSMSH_CERR_BIT | + ADF_GEN4_IAINTSTATSSM_SER_ERR_SSMSH_UNCERR_BIT))) + return false; + + reg = ADF_CSR_RD(csr, ADF_GEN4_SER_ERR_SSMSH); + reg &= ADF_GEN4_SER_ERR_SSMSH_FATAL_BITMASK | + ADF_GEN4_SER_ERR_SSMSH_UNCERR_BITMASK | + ADF_GEN4_SER_ERR_SSMSH_CERR_BITMASK; + if (reg & ADF_GEN4_SER_ERR_SSMSH_FATAL_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "Fatal SER_SSMSH_ERR: 0x%x\n", reg); + + errs_bits = reg & ADF_GEN4_SER_ERR_SSMSH_FATAL_BITMASK; + for_each_set_bit(bit_iterator, &errs_bits, bits_num) { + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + } + + reset_required = true; + } + + if (reg & ADF_GEN4_SER_ERR_SSMSH_UNCERR_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "non-fatal SER_SSMSH_ERR: 0x%x\n", reg); + + errs_bits = reg & ADF_GEN4_SER_ERR_SSMSH_UNCERR_BITMASK; + for_each_set_bit(bit_iterator, &errs_bits, bits_num) { + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + } + + if (reg & ADF_GEN4_SER_ERR_SSMSH_CERR_BITMASK) { + dev_warn(&GET_DEV(accel_dev), + "Correctable SER_SSMSH_ERR: 0x%x\n", reg); + + errs_bits = reg & ADF_GEN4_SER_ERR_SSMSH_CERR_BITMASK; + for_each_set_bit(bit_iterator, &errs_bits, bits_num) { + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + } + } + + ADF_CSR_WR(csr, ADF_GEN4_SER_ERR_SSMSH, reg); + + return reset_required; +} + +static bool adf_handle_iaintstatssm(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + u32 iastatssm = ADF_CSR_RD(csr, ADF_GEN4_IAINTSTATSSM); + bool reset_required; + + iastatssm &= ADF_GEN4_IAINTSTATSSM_BITMASK; + if (!iastatssm) + return false; + + reset_required = adf_handle_uerrssmsh(accel_dev, csr, iastatssm); + reset_required |= adf_handle_cerrssmsh(accel_dev, csr, iastatssm); + reset_required |= adf_handle_pperr_err(accel_dev, csr, iastatssm); + reset_required |= adf_handle_slice_hang_error(accel_dev, csr, iastatssm); + reset_required |= adf_handle_spppar_err(accel_dev, csr, iastatssm); + reset_required |= adf_handle_ssmcpppar_err(accel_dev, csr, iastatssm); + reset_required |= adf_handle_ser_err_ssmsh(accel_dev, csr, iastatssm); + adf_handle_rf_parr_err(accel_dev, csr, iastatssm); + + ADF_CSR_WR(csr, ADF_GEN4_IAINTSTATSSM, iastatssm); + + return reset_required; +} + +static bool adf_handle_exprpssmcmpr(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + u32 reg = ADF_CSR_RD(csr, ADF_GEN4_EXPRPSSMCPR); + + reg &= ADF_GEN4_EXPRPSSMCPR_UNCERR_BITMASK; + if (!reg) + return false; + + dev_err(&GET_DEV(accel_dev), + "Uncorrectable error exception in SSM CMP: 0x%x", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_EXPRPSSMCPR, reg); + + return false; +} + +static bool adf_handle_exprpssmxlt(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + u32 reg = ADF_CSR_RD(csr, ADF_GEN4_EXPRPSSMXLT); + + reg &= ADF_GEN4_EXPRPSSMXLT_UNCERR_BITMASK | + ADF_GEN4_EXPRPSSMXLT_CERR_BIT; + if (!reg) + return false; + + if (reg & ADF_GEN4_EXPRPSSMXLT_UNCERR_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "Uncorrectable error exception in SSM XLT: 0x%x", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + if (reg & ADF_GEN4_EXPRPSSMXLT_CERR_BIT) { + dev_warn(&GET_DEV(accel_dev), + "Correctable error exception in SSM XLT: 0x%x", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + } + + ADF_CSR_WR(csr, ADF_GEN4_EXPRPSSMXLT, reg); + + return false; +} + +static bool adf_handle_exprpssmdcpr(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + u32 reg; + int i; + + for (i = 0; i < ADF_GEN4_DCPR_SLICES_NUM; i++) { + reg = ADF_CSR_RD(csr, ADF_GEN4_EXPRPSSMDCPR(i)); + reg &= ADF_GEN4_EXPRPSSMDCPR_UNCERR_BITMASK | + ADF_GEN4_EXPRPSSMDCPR_CERR_BITMASK; + if (!reg) + continue; + + if (reg & ADF_GEN4_EXPRPSSMDCPR_UNCERR_BITMASK) { + dev_err(&GET_DEV(accel_dev), + "Uncorrectable error exception in SSM DCMP: 0x%x", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + if (reg & ADF_GEN4_EXPRPSSMDCPR_CERR_BITMASK) { + dev_warn(&GET_DEV(accel_dev), + "Correctable error exception in SSM DCMP: 0x%x", reg); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + } + + ADF_CSR_WR(csr, ADF_GEN4_EXPRPSSMDCPR(i), reg); + } + + return false; +} + +static bool adf_handle_ssm(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + bool reset_required; + + if (!(errsou & ADF_GEN4_ERRSOU2_SSM_ERR_BIT)) + return false; + + reset_required = adf_handle_iaintstatssm(accel_dev, csr); + reset_required |= adf_handle_exprpssmcmpr(accel_dev, csr); + reset_required |= adf_handle_exprpssmxlt(accel_dev, csr); + reset_required |= adf_handle_exprpssmdcpr(accel_dev, csr); + + return reset_required; +} + +static bool adf_handle_cpp_cfc_err(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + bool reset_required = false; + u32 reg; + + if (!(errsou & ADF_GEN4_ERRSOU2_CPP_CFC_ERR_STATUS_BIT)) + return false; + + reg = ADF_CSR_RD(csr, ADF_GEN4_CPP_CFC_ERR_STATUS); + if (reg & ADF_GEN4_CPP_CFC_ERR_STATUS_DATAPAR_BIT) { + dev_err(&GET_DEV(accel_dev), + "CPP_CFC_ERR: data parity: 0x%x", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + if (reg & ADF_GEN4_CPP_CFC_ERR_STATUS_CMDPAR_BIT) { + dev_err(&GET_DEV(accel_dev), + "CPP_CFC_ERR: command parity: 0x%x", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + reset_required = true; + } + + if (reg & ADF_GEN4_CPP_CFC_ERR_STATUS_MERR_BIT) { + dev_err(&GET_DEV(accel_dev), + "CPP_CFC_ERR: multiple errors: 0x%x", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + reset_required = true; + } + + ADF_CSR_WR(csr, ADF_GEN4_CPP_CFC_ERR_STATUS_CLR, + ADF_GEN4_CPP_CFC_ERR_STATUS_CLR_BITMASK); + + return reset_required; +} + +static void adf_gen4_process_errsou2(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou, + bool *reset_required) +{ + *reset_required |= adf_handle_ssm(accel_dev, csr, errsou); + *reset_required |= adf_handle_cpp_cfc_err(accel_dev, csr, errsou); +} + +static bool adf_handle_timiscsts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 timiscsts; + + if (!(errsou & ADF_GEN4_ERRSOU3_TIMISCSTS_BIT)) + return false; + + timiscsts = ADF_CSR_RD(csr, ADF_GEN4_TIMISCSTS); + + dev_err(&GET_DEV(accel_dev), + "Fatal error in Transmit Interface: 0x%x\n", timiscsts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + return true; +} + +static bool adf_handle_ricppintsts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 ricppintsts; + + if (!(errsou & ADF_GEN4_ERRSOU3_RICPPINTSTS_BITMASK)) + return false; + + ricppintsts = ADF_CSR_RD(csr, ADF_GEN4_RICPPINTSTS); + ricppintsts &= ADF_GEN4_RICPPINTSTS_BITMASK; + + dev_err(&GET_DEV(accel_dev), + "RI CPP Uncorrectable Error: 0x%x\n", ricppintsts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_RICPPINTSTS, ricppintsts); + + return false; +} + +static bool adf_handle_ticppintsts(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 ticppintsts; + + if (!(errsou & ADF_GEN4_ERRSOU3_TICPPINTSTS_BITMASK)) + return false; + + ticppintsts = ADF_CSR_RD(csr, ADF_GEN4_TICPPINTSTS); + ticppintsts &= ADF_GEN4_TICPPINTSTS_BITMASK; + + dev_err(&GET_DEV(accel_dev), + "TI CPP Uncorrectable Error: 0x%x\n", ticppintsts); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_TICPPINTSTS, ticppintsts); + + return false; +} + +static bool adf_handle_aramcerr(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 aram_cerr; + + if (!(errsou & ADF_GEN4_ERRSOU3_REG_ARAMCERR_BIT)) + return false; + + aram_cerr = ADF_CSR_RD(csr, ADF_GEN4_REG_ARAMCERR); + aram_cerr &= ADF_GEN4_REG_ARAMCERR_BIT; + + dev_warn(&GET_DEV(accel_dev), + "ARAM correctable error : 0x%x\n", aram_cerr); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + + aram_cerr |= ADF_GEN4_REG_ARAMCERR_EN_BITMASK; + + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMCERR, aram_cerr); + + return false; +} + +static bool adf_handle_aramuerr(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + bool reset_required = false; + u32 aramuerr; + + if (!(errsou & ADF_GEN4_ERRSOU3_REG_ARAMUERR_BIT)) + return false; + + aramuerr = ADF_CSR_RD(csr, ADF_GEN4_REG_ARAMUERR); + aramuerr &= ADF_GEN4_REG_ARAMUERR_ERROR_BIT | + ADF_GEN4_REG_ARAMUERR_MULTI_ERRORS_BIT; + + if (!aramuerr) + return false; + + if (aramuerr & ADF_GEN4_REG_ARAMUERR_MULTI_ERRORS_BIT) { + dev_err(&GET_DEV(accel_dev), + "ARAM multiple uncorrectable errors: 0x%x\n", aramuerr); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + reset_required = true; + } else { + dev_err(&GET_DEV(accel_dev), + "ARAM uncorrectable error: 0x%x\n", aramuerr); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + aramuerr |= ADF_GEN4_REG_ARAMUERR_EN_BITMASK; + + ADF_CSR_WR(csr, ADF_GEN4_REG_ARAMUERR, aramuerr); + + return reset_required; +} + +static bool adf_handle_reg_cppmemtgterr(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + bool reset_required = false; + u32 cppmemtgterr; + + if (!(errsou & ADF_GEN4_ERRSOU3_REG_ARAMUERR_BIT)) + return false; + + cppmemtgterr = ADF_CSR_RD(csr, ADF_GEN4_REG_CPPMEMTGTERR); + cppmemtgterr &= ADF_GEN4_REG_CPPMEMTGTERR_BITMASK | + ADF_GEN4_REG_CPPMEMTGTERR_MULTI_ERRORS_BIT; + if (!cppmemtgterr) + return false; + + if (cppmemtgterr & ADF_GEN4_REG_CPPMEMTGTERR_MULTI_ERRORS_BIT) { + dev_err(&GET_DEV(accel_dev), + "Misc memory target multiple uncorrectable errors: 0x%x\n", + cppmemtgterr); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + + reset_required = true; + } else { + dev_err(&GET_DEV(accel_dev), + "Misc memory target uncorrectable error: 0x%x\n", cppmemtgterr); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + cppmemtgterr |= ADF_GEN4_REG_CPPMEMTGTERR_EN_BITMASK; + + ADF_CSR_WR(csr, ADF_GEN4_REG_CPPMEMTGTERR, cppmemtgterr); + + return reset_required; +} + +static bool adf_handle_atufaultstatus(struct adf_accel_dev *accel_dev, + void __iomem *csr, u32 errsou) +{ + u32 i; + u32 max_rp_num = GET_HW_DATA(accel_dev)->num_banks; + + if (!(errsou & ADF_GEN4_ERRSOU3_ATUFAULTSTATUS_BIT)) + return false; + + for (i = 0; i < max_rp_num; i++) { + u32 atufaultstatus = ADF_CSR_RD(csr, ADF_GEN4_ATUFAULTSTATUS(i)); + + atufaultstatus &= ADF_GEN4_ATUFAULTSTATUS_BIT; + + if (atufaultstatus) { + dev_err(&GET_DEV(accel_dev), + "Ring Pair (%u) ATU detected fault: 0x%x\n", i, + atufaultstatus); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + + ADF_CSR_WR(csr, ADF_GEN4_ATUFAULTSTATUS(i), atufaultstatus); + } + } + + return false; +} + +static void adf_gen4_process_errsou3(struct adf_accel_dev *accel_dev, + void __iomem *csr, void __iomem *aram_csr, + u32 errsou, bool *reset_required) +{ + *reset_required |= adf_handle_timiscsts(accel_dev, csr, errsou); + *reset_required |= adf_handle_ricppintsts(accel_dev, csr, errsou); + *reset_required |= adf_handle_ticppintsts(accel_dev, csr, errsou); + *reset_required |= adf_handle_aramcerr(accel_dev, aram_csr, errsou); + *reset_required |= adf_handle_aramuerr(accel_dev, aram_csr, errsou); + *reset_required |= adf_handle_reg_cppmemtgterr(accel_dev, aram_csr, errsou); + *reset_required |= adf_handle_atufaultstatus(accel_dev, csr, errsou); +} + +static bool adf_gen4_handle_interrupt(struct adf_accel_dev *accel_dev, + bool *reset_required) +{ + void __iomem *aram_csr = adf_get_aram_base(accel_dev); + void __iomem *csr = adf_get_pmisc_base(accel_dev); + u32 errsou = ADF_CSR_RD(csr, ADF_GEN4_ERRSOU0); + bool handled = false; + + *reset_required = false; + + if (errsou & ADF_GEN4_ERRSOU0_BIT) { + adf_gen4_process_errsou0(accel_dev, csr); + handled = true; + } + + errsou = ADF_CSR_RD(csr, ADF_GEN4_ERRSOU1); + if (errsou & ADF_GEN4_ERRSOU1_BITMASK) { + adf_gen4_process_errsou1(accel_dev, csr, errsou, reset_required); + handled = true; + } + + errsou = ADF_CSR_RD(csr, ADF_GEN4_ERRSOU2); + if (errsou & ADF_GEN4_ERRSOU2_BITMASK) { + adf_gen4_process_errsou2(accel_dev, csr, errsou, reset_required); + handled = true; + } + + errsou = ADF_CSR_RD(csr, ADF_GEN4_ERRSOU3); + if (errsou & ADF_GEN4_ERRSOU3_BITMASK) { + adf_gen4_process_errsou3(accel_dev, csr, aram_csr, errsou, reset_required); + handled = true; + } + + return handled; +} + +void adf_gen4_init_ras_ops(struct adf_ras_ops *ras_ops) +{ + ras_ops->enable_ras_errors = adf_gen4_enable_ras; + ras_ops->disable_ras_errors = adf_gen4_disable_ras; + ras_ops->handle_interrupt = adf_gen4_handle_interrupt; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_ras_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_ras.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_ras.h new file mode 100644 index 000000000000..53352083cd12 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_ras.h @@ -0,0 +1,825 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_GEN4_RAS_H_ +#define ADF_GEN4_RAS_H_ + +#include <linux/bits.h> + +struct adf_ras_ops; + +/* ERRSOU0 Correctable error mask*/ +#define ADF_GEN4_ERRSOU0_BIT BIT(0) + +/* HI AE Correctable error log */ +#define ADF_GEN4_HIAECORERRLOG_CPP0 0x41A308 + +/* HI AE Correctable error log enable */ +#define ADF_GEN4_HIAECORERRLOGENABLE_CPP0 0x41A318 +#define ADF_GEN4_ERRSOU1_HIAEUNCERRLOG_CPP0_BIT BIT(0) +#define ADF_GEN4_ERRSOU1_HICPPAGENTCMDPARERRLOG_BIT BIT(1) +#define ADF_GEN4_ERRSOU1_RIMEM_PARERR_STS_BIT BIT(2) +#define ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT BIT(3) +#define ADF_GEN4_ERRSOU1_RIMISCSTS_BIT BIT(4) + +#define ADF_GEN4_ERRSOU1_BITMASK ( \ + (ADF_GEN4_ERRSOU1_HIAEUNCERRLOG_CPP0_BIT) | \ + (ADF_GEN4_ERRSOU1_HICPPAGENTCMDPARERRLOG_BIT) | \ + (ADF_GEN4_ERRSOU1_RIMEM_PARERR_STS_BIT) | \ + (ADF_GEN4_ERRSOU1_TIMEM_PARERR_STS_BIT) | \ + (ADF_GEN4_ERRSOU1_RIMISCSTS_BIT)) + +/* HI AE Uncorrectable error log */ +#define ADF_GEN4_HIAEUNCERRLOG_CPP0 0x41A300 + +/* HI AE Uncorrectable error log enable */ +#define ADF_GEN4_HIAEUNCERRLOGENABLE_CPP0 0x41A320 + +/* HI CPP Agent Command parity error log */ +#define ADF_GEN4_HICPPAGENTCMDPARERRLOG 0x41A310 + +/* HI CPP Agent Command parity error logging enable */ +#define ADF_GEN4_HICPPAGENTCMDPARERRLOGENABLE 0x41A314 + +/* RI Memory parity error status register */ +#define ADF_GEN4_RIMEM_PARERR_STS 0x41B128 + +/* RI Memory parity error reporting enable */ +#define ADF_GEN4_RI_MEM_PAR_ERR_EN0 0x41B12C + +/* + * RI Memory parity error mask + * BIT(0) - BIT(3) - ri_iosf_pdata_rxq[0:3] parity error + * BIT(4) - ri_tlq_phdr parity error + * BIT(5) - ri_tlq_pdata parity error + * BIT(6) - ri_tlq_nphdr parity error + * BIT(7) - ri_tlq_npdata parity error + * BIT(8) - BIT(9) - ri_tlq_cplhdr[0:1] parity error + * BIT(10) - BIT(17) - ri_tlq_cpldata[0:7] parity error + * BIT(18) - set this bit to 1 to enable logging status to ri_mem_par_err_sts0 + * BIT(19) - ri_cds_cmd_fifo parity error + * BIT(20) - ri_obc_ricpl_fifo parity error + * BIT(21) - ri_obc_tiricpl_fifo parity error + * BIT(22) - ri_obc_cppcpl_fifo parity error + * BIT(23) - ri_obc_pendcpl_fifo parity error + * BIT(24) - ri_cpp_cmd_fifo parity error + * BIT(25) - ri_cds_ticmd_fifo parity error + * BIT(26) - riti_cmd_fifo parity error + * BIT(27) - ri_int_msixtbl parity error + * BIT(28) - ri_int_imstbl parity error + * BIT(30) - ri_kpt_fuses parity error + */ +#define ADF_GEN4_RIMEM_PARERR_STS_UNCERR_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(5) | \ + BIT(7) | BIT(10) | BIT(11) | BIT(12) | BIT(13) | \ + BIT(14) | BIT(15) | BIT(16) | BIT(17) | BIT(18) | BIT(19) | \ + BIT(20) | BIT(21) | BIT(22) | BIT(23) | BIT(24) | BIT(25) | \ + BIT(26) | BIT(27) | BIT(28) | BIT(30)) + +#define ADF_GEN4_RIMEM_PARERR_STS_FATAL_BITMASK \ + (BIT(4) | BIT(6) | BIT(8) | BIT(9)) + +/* TI CI parity status */ +#define ADF_GEN4_TI_CI_PAR_STS 0x50060C + +/* TI CI parity reporting mask */ +#define ADF_GEN4_TI_CI_PAR_ERR_MASK 0x500608 + +/* + * TI CI parity status mask + * BIT(0) - CdCmdQ_sts patiry error status + * BIT(1) - CdDataQ_sts parity error status + * BIT(3) - CPP_SkidQ_sts parity error status + * BIT(7) - CPP_SkidQ_sc_sts parity error status + */ +#define ADF_GEN4_TI_CI_PAR_STS_BITMASK \ + (BIT(0) | BIT(1) | BIT(3) | BIT(7)) + +/* TI PULLFUB parity status */ +#define ADF_GEN4_TI_PULL0FUB_PAR_STS 0x500618 + +/* TI PULLFUB parity error reporting mask */ +#define ADF_GEN4_TI_PULL0FUB_PAR_ERR_MASK 0x500614 + +/* + * TI PULLFUB parity status mask + * BIT(0) - TrnPullReqQ_sts parity status + * BIT(1) - TrnSharedDataQ_sts parity status + * BIT(2) - TrnPullReqDataQ_sts parity status + * BIT(4) - CPP_CiPullReqQ_sts parity status + * BIT(5) - CPP_TrnPullReqQ_sts parity status + * BIT(6) - CPP_PullidQ_sts parity status + * BIT(7) - CPP_WaitDataQ_sts parity status + * BIT(8) - CPP_CdDataQ_sts parity status + * BIT(9) - CPP_TrnDataQP0_sts parity status + * BIT(10) - BIT(11) - CPP_TrnDataQRF[00:01]_sts parity status + * BIT(12) - CPP_TrnDataQP1_sts parity status + * BIT(13) - BIT(14) - CPP_TrnDataQRF[10:11]_sts parity status + */ +#define ADF_GEN4_TI_PULL0FUB_PAR_STS_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(4) | BIT(5) | BIT(6) | BIT(7) | \ + BIT(8) | BIT(9) | BIT(10) | BIT(11) | BIT(12) | BIT(13) | BIT(14)) + +/* TI PUSHUB parity status */ +#define ADF_GEN4_TI_PUSHFUB_PAR_STS 0x500630 + +/* TI PUSHFUB parity error reporting mask */ +#define ADF_GEN4_TI_PUSHFUB_PAR_ERR_MASK 0x50062C + +/* + * TI PUSHUB parity status mask + * BIT(0) - SbPushReqQ_sts parity status + * BIT(1) - BIT(2) - SbPushDataQ[0:1]_sts parity status + * BIT(4) - CPP_CdPushReqQ_sts parity status + * BIT(5) - BIT(6) - CPP_CdPushDataQ[0:1]_sts parity status + * BIT(7) - CPP_SbPushReqQ_sts parity status + * BIT(8) - CPP_SbPushDataQP_sts parity status + * BIT(9) - BIT(10) - CPP_SbPushDataQRF[0:1]_sts parity status + */ +#define ADF_GEN4_TI_PUSHFUB_PAR_STS_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(4) | BIT(5) | \ + BIT(6) | BIT(7) | BIT(8) | BIT(9) | BIT(10)) + +/* TI CD parity status */ +#define ADF_GEN4_TI_CD_PAR_STS 0x50063C + +/* TI CD parity error mask */ +#define ADF_GEN4_TI_CD_PAR_ERR_MASK 0x500638 + +/* + * TI CD parity status mask + * BIT(0) - BIT(15) - CtxMdRam[0:15]_sts parity status + * BIT(16) - Leaf2ClusterRam_sts parity status + * BIT(17) - BIT(18) - Ring2LeafRam[0:1]_sts parity status + * BIT(19) - VirtualQ_sts parity status + * BIT(20) - DtRdQ_sts parity status + * BIT(21) - DtWrQ_sts parity status + * BIT(22) - RiCmdQ_sts parity status + * BIT(23) - BypassQ_sts parity status + * BIT(24) - DtRdQ_sc_sts parity status + * BIT(25) - DtWrQ_sc_sts parity status + */ +#define ADF_GEN4_TI_CD_PAR_STS_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4) | BIT(5) | BIT(6) | \ + BIT(7) | BIT(8) | BIT(9) | BIT(10) | BIT(11) | BIT(12) | BIT(13) | \ + BIT(14) | BIT(15) | BIT(16) | BIT(17) | BIT(18) | BIT(19) | BIT(20) | \ + BIT(21) | BIT(22) | BIT(23) | BIT(24) | BIT(25)) + +/* TI TRNSB parity status */ +#define ADF_GEN4_TI_TRNSB_PAR_STS 0x500648 + +/* TI TRNSB Parity error reporting mask */ +#define ADF_GEN4_TI_TRNSB_PAR_ERR_MASK 0x500644 + +/* + * TI TRNSB parity status mask + * BIT(0) - TrnPHdrQP_sts parity status + * BIT(1) - TrnPHdrQRF_sts parity status + * BIT(2) - TrnPDataQP_sts parity status + * BIT(3) - BIT(6) - TrnPDataQRF[0:3]_sts parity status + * BIT(7) - TrnNpHdrQP_sts parity status + * BIT(8) - BIT(9) - TrnNpHdrQRF[0:1]_sts parity status + * BIT(10) - TrnCplHdrQ_sts parity status + * BIT(11) - TrnPutObsReqQ_sts parity status + * BIT(12) - TrnPushReqQ_sts parity status + * BIT(13) - SbSplitIdRam_sts parity status + * BIT(14) - SbReqCountQ_sts parity status + * BIT(15) - SbCplTrkRam_sts parity status + * BIT(16) - SbGetObsReqQ_sts parity status + * BIT(17) - SbEpochIdQ_sts parity status + * BIT(18) - SbAtCplHdrQ_sts parity status + * BIT(19) - SbAtCplDataQ_sts parity status + * BIT(20) - SbReqCountRam_sts parity status + * BIT(21) - SbAtCplHdrQ_sc_sts parity status + */ +#define ADF_GEN4_TI_TRNSB_PAR_STS_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4) | BIT(5) | BIT(6) | \ + BIT(7) | BIT(8) | BIT(9) | BIT(10) | BIT(11) | BIT(12) | \ + BIT(13) | BIT(14) | BIT(15) | BIT(16) | BIT(17) | BIT(18) | \ + BIT(19) | BIT(20) | BIT(21)) + +/* Status register to log misc error on RI */ +#define ADF_GEN4_RIMISCSTS 0x41B1B8 + +/* Status control register to log misc RI error */ +#define ADF_GEN4_RIMISCCTL 0x41B1BC + +/* + * ERRSOU2 bit mask + * BIT(0) - SSM Interrupt Mask + * BIT(1) - CFC on CPP. ORed of CFC Push error and Pull error + * BIT(2) - BIT(4) - CPP attention interrupts, deprecated on gen4 devices + * BIT(18) - PM interrupt + */ +#define ADF_GEN4_ERRSOU2_SSM_ERR_BIT BIT(0) +#define ADF_GEN4_ERRSOU2_CPP_CFC_ERR_STATUS_BIT BIT(1) +#define ADF_GEN4_ERRSOU2_CPP_CFC_ATT_INT_BITMASK \ + (BIT(2) | BIT(3) | BIT(4)) + +#define ADF_GEN4_ERRSOU2_PM_INT_BIT BIT(18) + +#define ADF_GEN4_ERRSOU2_BITMASK \ + (ADF_GEN4_ERRSOU2_SSM_ERR_BIT | \ + ADF_GEN4_ERRSOU2_CPP_CFC_ERR_STATUS_BIT) + +#define ADF_GEN4_ERRSOU2_DIS_BITMASK \ + (ADF_GEN4_ERRSOU2_SSM_ERR_BIT | \ + ADF_GEN4_ERRSOU2_CPP_CFC_ERR_STATUS_BIT | \ + ADF_GEN4_ERRSOU2_CPP_CFC_ATT_INT_BITMASK) + +#define ADF_GEN4_IAINTSTATSSM 0x28 + +/* IAINTSTATSSM error bit mask definitions */ +#define ADF_GEN4_IAINTSTATSSM_UERRSSMSH_BIT BIT(0) +#define ADF_GEN4_IAINTSTATSSM_CERRSSMSH_BIT BIT(1) +#define ADF_GEN4_IAINTSTATSSM_PPERR_BIT BIT(2) +#define ADF_GEN4_IAINTSTATSSM_SLICEHANG_ERR_BIT BIT(3) +#define ADF_GEN4_IAINTSTATSSM_SPPPARERR_BIT BIT(4) +#define ADF_GEN4_IAINTSTATSSM_SSMCPPERR_BIT BIT(5) +#define ADF_GEN4_IAINTSTATSSM_SSMSOFTERRORPARITY_BIT BIT(6) +#define ADF_GEN4_IAINTSTATSSM_SER_ERR_SSMSH_CERR_BIT BIT(7) +#define ADF_GEN4_IAINTSTATSSM_SER_ERR_SSMSH_UNCERR_BIT BIT(8) + +#define ADF_GEN4_IAINTSTATSSM_BITMASK \ + (ADF_GEN4_IAINTSTATSSM_UERRSSMSH_BIT | \ + ADF_GEN4_IAINTSTATSSM_CERRSSMSH_BIT | \ + ADF_GEN4_IAINTSTATSSM_PPERR_BIT | \ + ADF_GEN4_IAINTSTATSSM_SLICEHANG_ERR_BIT | \ + ADF_GEN4_IAINTSTATSSM_SPPPARERR_BIT | \ + ADF_GEN4_IAINTSTATSSM_SSMCPPERR_BIT | \ + ADF_GEN4_IAINTSTATSSM_SSMSOFTERRORPARITY_BIT | \ + ADF_GEN4_IAINTSTATSSM_SER_ERR_SSMSH_CERR_BIT | \ + ADF_GEN4_IAINTSTATSSM_SER_ERR_SSMSH_UNCERR_BIT) + +#define ADF_GEN4_UERRSSMSH 0x18 + +/* + * UERRSSMSH error bit masks definitions + * + * BIT(0) - Indicates one uncorrectable error + * BIT(15) - Indicates multiple uncorrectable errors + * in device shared memory + */ +#define ADF_GEN4_UERRSSMSH_BITMASK (BIT(0) | BIT(15)) + +#define ADF_GEN4_UERRSSMSHAD 0x1C + +#define ADF_GEN4_CERRSSMSH 0x10 + +/* + * CERRSSMSH error bit + * BIT(0) - Indicates one correctable error + */ +#define ADF_GEN4_CERRSSMSH_ERROR_BIT BIT(0) + +#define ADF_GEN4_CERRSSMSHAD 0x14 + +/* SSM error handling features enable register */ +#define ADF_GEN4_SSMFEATREN 0x198 + +/* + * Disable SSM error detection and reporting features + * enabled by device driver on RAS initialization + * + * following bits should be cleared : + * BIT(4) - Disable parity for CPP parity + * BIT(12) - Disable logging push/pull data error in pperr register. + * BIT(16) - BIT(23) - Disable parity for SPPs + * BIT(24) - BIT(27) - Disable parity for SPPs, if it's supported on the device. + */ +#define ADF_GEN4_SSMFEATREN_DIS_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(5) | BIT(6) | BIT(7) | \ + BIT(8) | BIT(9) | BIT(10) | BIT(11) | BIT(13) | BIT(14) | BIT(15)) + +#define ADF_GEN4_INTMASKSSM 0x0 + +/* + * Error reporting mask in INTMASKSSM + * BIT(0) - Shared memory uncorrectable interrupt mask + * BIT(1) - Shared memory correctable interrupt mask + * BIT(2) - PPERR interrupt mask + * BIT(3) - CPP parity error Interrupt mask + * BIT(4) - SSM interrupt generated by SER correctable error mask + * BIT(5) - SSM interrupt generated by SER uncorrectable error + * - not stop and scream - mask + */ +#define ADF_GEN4_INTMASKSSM_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4) | BIT(5)) + +/* CPP push or pull error */ +#define ADF_GEN4_PPERR 0x8 + +#define ADF_GEN4_PPERR_BITMASK (BIT(0) | BIT(1)) + +#define ADF_GEN4_PPERRID 0xC + +/* Slice hang handling related registers */ +#define ADF_GEN4_SLICEHANGSTATUS_ATH_CPH 0x84 +#define ADF_GEN4_SLICEHANGSTATUS_CPR_XLT 0x88 +#define ADF_GEN4_SLICEHANGSTATUS_DCPR_UCS 0x90 +#define ADF_GEN4_SLICEHANGSTATUS_WAT_WCP 0x8C +#define ADF_GEN4_SLICEHANGSTATUS_PKE 0x94 + +#define ADF_GEN4_SHINTMASKSSM_ATH_CPH 0xF0 +#define ADF_GEN4_SHINTMASKSSM_CPR_XLT 0xF4 +#define ADF_GEN4_SHINTMASKSSM_DCPR_UCS 0xFC +#define ADF_GEN4_SHINTMASKSSM_WAT_WCP 0xF8 +#define ADF_GEN4_SHINTMASKSSM_PKE 0x100 + +/* SPP pull cmd parity err_*slice* CSR */ +#define ADF_GEN4_SPPPULLCMDPARERR_ATH_CPH 0x1A4 +#define ADF_GEN4_SPPPULLCMDPARERR_CPR_XLT 0x1A8 +#define ADF_GEN4_SPPPULLCMDPARERR_DCPR_UCS 0x1B0 +#define ADF_GEN4_SPPPULLCMDPARERR_PKE 0x1B4 +#define ADF_GEN4_SPPPULLCMDPARERR_WAT_WCP 0x1AC + +/* SPP pull data parity err_*slice* CSR */ +#define ADF_GEN4_SPPPULLDATAPARERR_ATH_CPH 0x1BC +#define ADF_GEN4_SPPPULLDATAPARERR_CPR_XLT 0x1C0 +#define ADF_GEN4_SPPPULLDATAPARERR_DCPR_UCS 0x1C8 +#define ADF_GEN4_SPPPULLDATAPARERR_PKE 0x1CC +#define ADF_GEN4_SPPPULLDATAPARERR_WAT_WCP 0x1C4 + +/* SPP push cmd parity err_*slice* CSR */ +#define ADF_GEN4_SPPPUSHCMDPARERR_ATH_CPH 0x1D4 +#define ADF_GEN4_SPPPUSHCMDPARERR_CPR_XLT 0x1D8 +#define ADF_GEN4_SPPPUSHCMDPARERR_DCPR_UCS 0x1E0 +#define ADF_GEN4_SPPPUSHCMDPARERR_PKE 0x1E4 +#define ADF_GEN4_SPPPUSHCMDPARERR_WAT_WCP 0x1DC + +/* SPP push data parity err_*slice* CSR */ +#define ADF_GEN4_SPPPUSHDATAPARERR_ATH_CPH 0x1EC +#define ADF_GEN4_SPPPUSHDATAPARERR_CPR_XLT 0x1F0 +#define ADF_GEN4_SPPPUSHDATAPARERR_DCPR_UCS 0x1F8 +#define ADF_GEN4_SPPPUSHDATAPARERR_PKE 0x1FC +#define ADF_GEN4_SPPPUSHDATAPARERR_WAT_WCP 0x1F4 + +/* Accelerator SPP parity error mask registers */ +#define ADF_GEN4_SPPPARERRMSK_ATH_CPH 0x204 +#define ADF_GEN4_SPPPARERRMSK_CPR_XLT 0x208 +#define ADF_GEN4_SPPPARERRMSK_DCPR_UCS 0x210 +#define ADF_GEN4_SPPPARERRMSK_PKE 0x214 +#define ADF_GEN4_SPPPARERRMSK_WAT_WCP 0x20C + +#define ADF_GEN4_SSMCPPERR 0x224 + +/* + * Uncorrectable error mask in SSMCPPERR + * BIT(0) - indicates CPP command parity error + * BIT(1) - indicates CPP Main Push PPID parity error + * BIT(2) - indicates CPP Main ePPID parity error + * BIT(3) - indicates CPP Main push data parity error + * BIT(4) - indicates CPP Main Pull PPID parity error + * BIT(5) - indicates CPP target pull data parity error + */ +#define ADF_GEN4_SSMCPPERR_FATAL_BITMASK \ + (BIT(0) | BIT(1) | BIT(4)) + +#define ADF_GEN4_SSMCPPERR_UNCERR_BITMASK \ + (BIT(2) | BIT(3) | BIT(5)) + +#define ADF_GEN4_SSMSOFTERRORPARITY_SRC 0x9C +#define ADF_GEN4_SSMSOFTERRORPARITYMASK_SRC 0xB8 + +#define ADF_GEN4_SSMSOFTERRORPARITY_ATH_CPH 0xA0 +#define ADF_GEN4_SSMSOFTERRORPARITYMASK_ATH_CPH 0xBC + +#define ADF_GEN4_SSMSOFTERRORPARITY_CPR_XLT 0xA4 +#define ADF_GEN4_SSMSOFTERRORPARITYMASK_CPR_XLT 0xC0 + +#define ADF_GEN4_SSMSOFTERRORPARITY_DCPR_UCS 0xAC +#define ADF_GEN4_SSMSOFTERRORPARITYMASK_DCPR_UCS 0xC8 + +#define ADF_GEN4_SSMSOFTERRORPARITY_PKE 0xB0 +#define ADF_GEN4_SSMSOFTERRORPARITYMASK_PKE 0xCC + +#define ADF_GEN4_SSMSOFTERRORPARITY_WAT_WCP 0xA8 +#define ADF_GEN4_SSMSOFTERRORPARITYMASK_WAT_WCP 0xC4 + +/* RF parity error detected in SharedRAM */ +#define ADF_GEN4_SSMSOFTERRORPARITY_SRC_BIT BIT(0) + +#define ADF_GEN4_SER_ERR_SSMSH 0x44C + +/* + * Fatal error mask in SER_ERR_SSMSH + * BIT(0) - Indicates an uncorrectable error has occurred in the + * accelerator controller command RFs + * BIT(2) - Parity error occurred in the bank SPP fifos + * BIT(3) - Indicates Parity error occurred in following fifos in + * the design + * BIT(4) - Parity error occurred in flops in the design + * BIT(5) - Uncorrectable error has occurred in the + * target push and pull data register flop + * BIT(7) - Indicates Parity error occurred in the Resource Manager + * pending lock request fifos + * BIT(8) - Indicates Parity error occurred in the Resource Manager + * MECTX command queues logic + * BIT(9) - Indicates Parity error occurred in the Resource Manager + * MECTX sigdone fifo flops + * BIT(10) - Indicates an uncorrectable error has occurred in the + * Resource Manager MECTX command RFs + * BIT(14) - Parity error occurred in Buffer Manager sigdone FIFO + */ + #define ADF_GEN4_SER_ERR_SSMSH_FATAL_BITMASK \ + (BIT(0) | BIT(2) | BIT(3) | BIT(4) | BIT(5) | BIT(7) | \ + BIT(8) | BIT(9) | BIT(10) | BIT(14)) + +/* + * Uncorrectable error mask in SER_ERR_SSMSH + * BIT(12) Parity error occurred in Buffer Manager pool 0 + * BIT(13) Parity error occurred in Buffer Manager pool 1 + */ +#define ADF_GEN4_SER_ERR_SSMSH_UNCERR_BITMASK \ + (BIT(12) | BIT(13)) + +/* + * Correctable error mask in SER_ERR_SSMSH + * BIT(1) - Indicates a correctable Error has occurred + * in the slice controller command RFs + * BIT(6) - Indicates a correctable Error has occurred in + * the target push and pull data RFs + * BIT(11) - Indicates an correctable Error has occurred in + * the Resource Manager MECTX command RFs + */ +#define ADF_GEN4_SER_ERR_SSMSH_CERR_BITMASK \ + (BIT(1) | BIT(6) | BIT(11)) + +/* SSM shared memory SER error reporting mask */ +#define ADF_GEN4_SER_EN_SSMSH 0x450 + +/* + * SSM SER error reporting mask in SER_en_err_ssmsh + * BIT(0) - Enables uncorrectable Error detection in : + * 1) slice controller command RFs. + * 2) target push/pull data registers + * BIT(1) - Enables correctable Error detection in : + * 1) slice controller command RFs + * 2) target push/pull data registers + * BIT(2) - Enables Parity error detection in + * 1) bank SPP fifos + * 2) gen4_pull_id_queue + * 3) gen4_push_id_queue + * 4) AE_pull_sigdn_fifo + * 5) DT_push_sigdn_fifo + * 6) slx_push_sigdn_fifo + * 7) secure_push_cmd_fifo + * 8) secure_pull_cmd_fifo + * 9) Head register in FIFO wrapper + * 10) current_cmd in individual push queue + * 11) current_cmd in individual pull queue + * 12) push_command_rxp arbitrated in ssm_push_cmd_queues + * 13) pull_command_rxp arbitrated in ssm_pull_cmd_queues + * BIT(3) - Enables uncorrectable Error detection in + * the resource manager mectx cmd RFs. + * BIT(4) - Enables correctable error detection in the Resource Manager + * mectx command RFs + * BIT(5) - Enables Parity error detection in + * 1) resource manager lock request fifo + * 2) mectx cmdqueues logic + * 3) mectx sigdone fifo + * BIT(6) - Enables Parity error detection in Buffer Manager pools + * and sigdone fifo + */ +#define ADF_GEN4_SER_EN_SSMSH_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4) | BIT(5) | BIT(6)) + +#define ADF_GEN4_CPP_CFC_ERR_STATUS 0x640C04 + +/* + * BIT(1) - Indicates multiple CPP CFC errors + * BIT(7) - Indicates CPP CFC command parity error type + * BIT(8) - Indicated CPP CFC data parity error type + */ +#define ADF_GEN4_CPP_CFC_ERR_STATUS_MERR_BIT BIT(1) +#define ADF_GEN4_CPP_CFC_ERR_STATUS_CMDPAR_BIT BIT(7) +#define ADF_GEN4_CPP_CFC_ERR_STATUS_DATAPAR_BIT BIT(8) + +/* + * BIT(0) - Enables CFC to detect and log push/pull data error + * BIT(1) - Enables CFC to generate interrupt to PCIEP for CPP error + * BIT(4) - When 1 Parity detection is disabled + * BIT(5) - When 1 Parity detection is disabled on CPP command bus + * BIT(6) - When 1 Parity detection is disabled on CPP push/pull bus + * BIT(9) - When 1 RF parity error detection is disabled + */ +#define ADF_GEN4_CPP_CFC_ERR_CTRL_BITMASK (BIT(0) | BIT(1)) + +#define ADF_GEN4_CPP_CFC_ERR_CTRL_DIS_BITMASK \ + (BIT(4) | BIT(5) | BIT(6) | BIT(9) | BIT(10)) + +#define ADF_GEN4_CPP_CFC_ERR_CTRL 0x640C00 + +/* + * BIT(0) - Clears bit(0) of ADF_GEN4_CPP_CFC_ERR_STATUS + * when an error is reported on CPP + * BIT(1) - Clears bit(1) of ADF_GEN4_CPP_CFC_ERR_STATUS + * when multiple errors are reported on CPP + * BIT(2) - Clears bit(2) of ADF_GEN4_CPP_CFC_ERR_STATUS + * when attention interrupt is reported + */ +#define ADF_GEN4_CPP_CFC_ERR_STATUS_CLR_BITMASK (BIT(0) | BIT(1) | BIT(2)) +#define ADF_GEN4_CPP_CFC_ERR_STATUS_CLR 0x640C08 + +#define ADF_GEN4_CPP_CFC_ERR_PPID_LO 0x640C0C +#define ADF_GEN4_CPP_CFC_ERR_PPID_HI 0x640C10 + +/* Exception reporting in QAT SSM CMP */ +#define ADF_GEN4_EXPRPSSMCPR 0x2000 + +/* + * Uncorrectable error mask in EXPRPSSMCPR + * BIT(2) - Hard fatal error + * BIT(16) - Parity error detected in CPR Push FIFO + * BIT(17) - Parity error detected in CPR Pull FIFO + * BIT(18) - Parity error detected in CPR Hash Table + * BIT(19) - Parity error detected in CPR History Buffer Copy 0 + * BIT(20) - Parity error detected in CPR History Buffer Copy 1 + * BIT(21) - Parity error detected in CPR History Buffer Copy 2 + * BIT(22) - Parity error detected in CPR History Buffer Copy 3 + * BIT(23) - Parity error detected in CPR History Buffer Copy 4 + * BIT(24) - Parity error detected in CPR History Buffer Copy 5 + * BIT(25) - Parity error detected in CPR History Buffer Copy 6 + * BIT(26) - Parity error detected in CPR History Buffer Copy 7 + */ +#define ADF_GEN4_EXPRPSSMCPR_UNCERR_BITMASK \ + (BIT(2) | BIT(16) | BIT(17) | BIT(18) | BIT(19) | BIT(20) | \ + BIT(21) | BIT(22) | BIT(23) | BIT(24) | BIT(25) | BIT(26)) + +/* Exception reporting in QAT SSM XLT */ +#define ADF_GEN4_EXPRPSSMXLT 0xA000 + +/* + * Uncorrectable error mask in EXPRPSSMXLT + * BIT(2) - If set, an Uncorrectable Error event occurred + * BIT(16) - Parity error detected in XLT Push FIFO + * BIT(17) - Parity error detected in XLT Pull FIFO + * BIT(18) - Parity error detected in XLT HCTB0 + * BIT(19) - Parity error detected in XLT HCTB1 + * BIT(20) - Parity error detected in XLT HCTB2 + * BIT(21) - Parity error detected in XLT HCTB3 + * BIT(22) - Parity error detected in XLT CBCL + * BIT(23) - Parity error detected in XLT LITPTR + */ +#define ADF_GEN4_EXPRPSSMXLT_UNCERR_BITMASK \ + (BIT(2) | BIT(16) | BIT(17) | BIT(18) | BIT(19) | BIT(20) | BIT(21) | \ + BIT(22) | BIT(23)) + +/* + * Correctable error mask in EXPRPSSMXLT + * BIT(3) - Correctable error event occurred. + */ +#define ADF_GEN4_EXPRPSSMXLT_CERR_BIT BIT(3) + +/* Exception reporting in QAT SSM DCMP */ +#define ADF_GEN4_EXPRPSSMDCPR(_n_) (0x12000 + (_n_) * 0x80) + +/* + * Uncorrectable error mask in EXPRPSSMDCPR + * BIT(2) - Even hard fatal error + * BIT(4) - Odd hard fatal error + * BIT(6) - decode soft error + * BIT(16) - Parity error detected in CPR Push FIFO + * BIT(17) - Parity error detected in CPR Pull FIFO + * BIT(18) - Parity error detected in the Input Buffer + * BIT(19) - symbuf0parerr + * Parity error detected in CPR Push FIFO + * BIT(20) - symbuf1parerr + * Parity error detected in CPR Push FIFO + */ +#define ADF_GEN4_EXPRPSSMDCPR_UNCERR_BITMASK \ + (BIT(2) | BIT(4) | BIT(6) | BIT(16) | BIT(17) | \ + BIT(18) | BIT(19) | BIT(20)) + +/* + * Correctable error mask in EXPRPSSMDCPR + * BIT(3) - Even ecc correctable error + * BIT(5) - Odd ecc correctable error + */ +#define ADF_GEN4_EXPRPSSMDCPR_CERR_BITMASK (BIT(3) | BIT(5)) + +#define ADF_GEN4_DCPR_SLICES_NUM 3 + +/* + * ERRSOU3 bit masks + * BIT(0) - indicates error Response Order Overflow and/or BME error + * BIT(1) - indicates RI push/pull error + * BIT(2) - indicates TI push/pull error + * BIT(3) - indicates ARAM correctable error + * BIT(4) - indicates ARAM uncorrectable error + * BIT(5) - indicates TI pull parity error + * BIT(6) - indicates RI push parity error + * BIT(7) - indicates VFLR interrupt + * BIT(8) - indicates ring pair interrupts for ATU detected fault + * BIT(9) - indicates error when accessing RLT block + */ +#define ADF_GEN4_ERRSOU3_TIMISCSTS_BIT BIT(0) +#define ADF_GEN4_ERRSOU3_RICPPINTSTS_BITMASK (BIT(1) | BIT(6)) +#define ADF_GEN4_ERRSOU3_TICPPINTSTS_BITMASK (BIT(2) | BIT(5)) +#define ADF_GEN4_ERRSOU3_REG_ARAMCERR_BIT BIT(3) +#define ADF_GEN4_ERRSOU3_REG_ARAMUERR_BIT BIT(4) +#define ADF_GEN4_ERRSOU3_VFLRNOTIFY_BIT BIT(7) +#define ADF_GEN4_ERRSOU3_ATUFAULTSTATUS_BIT BIT(8) +#define ADF_GEN4_ERRSOU3_RLTERROR_BIT BIT(9) + +#define ADF_GEN4_ERRSOU3_BITMASK ( \ + (ADF_GEN4_ERRSOU3_TIMISCSTS_BIT) | \ + (ADF_GEN4_ERRSOU3_RICPPINTSTS_BITMASK) | \ + (ADF_GEN4_ERRSOU3_TICPPINTSTS_BITMASK) | \ + (ADF_GEN4_ERRSOU3_REG_ARAMCERR_BIT) | \ + (ADF_GEN4_ERRSOU3_REG_ARAMUERR_BIT) | \ + (ADF_GEN4_ERRSOU3_VFLRNOTIFY_BIT) | \ + (ADF_GEN4_ERRSOU3_ATUFAULTSTATUS_BIT) | \ + (ADF_GEN4_ERRSOU3_RLTERROR_BIT)) + +/* TI Misc status register */ +#define ADF_GEN4_TIMISCSTS 0x50054C + +/* TI Misc error reporting mask */ +#define ADF_GEN4_TIMISCCTL 0x500548 + +/* + * TI Misc error reporting control mask + * BIT(0) - Enables error detection and logging in TIMISCSTS register + * BIT(1) - It has effect only when SRIOV enabled, this bit is 0 by default + * BIT(2) - Enables the D-F-x counter within the dispatch arbiter + * to start based on the command triggered from + * BIT(30) - Disables VFLR functionality + * By setting this bit will revert to CPM1.x functionality + * bits 1, 2 and 30 value should be preserved and not meant to be changed + * within RAS. + */ +#define ADF_GEN4_TIMISCCTL_BIT BIT(0) +#define ADF_GEN4_TIMSCCTL_RELAY_BITMASK (BIT(1) | BIT(2) | BIT(30)) + +/* RI CPP interface status register */ +#define ADF_GEN4_RICPPINTSTS 0x41A330 + +/* + * Uncorrectable error mask in RICPPINTSTS register + * BIT(0) - RI asserted the CPP error signal during a push + * BIT(1) - RI detected the CPP error signal asserted during a pull + * BIT(2) - RI detected a push data parity error + * BIT(3) - RI detected a push valid parity error + */ +#define ADF_GEN4_RICPPINTSTS_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3)) + +/* RI CPP interface status register control */ +#define ADF_GEN4_RICPPINTCTL 0x41A32C + +/* + * Control bit mask for RICPPINTCTL register + * BIT(0) - value of 1 enables error detection and reporting + * on the RI CPP Push interface + * BIT(1) - value of 1 enables error detection and reporting + * on the RI CPP Pull interface + * BIT(2) - value of 1 enables error detection and reporting + * on the RI Parity + * BIT(3) - value of 1 enable checking parity on CPP + * BIT(4) - value of 1 enables the stop feature of the stop and stream + * for all RI CPP Command RFs + */ +#define ADF_GEN4_RICPPINTCTL_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4)) + +/* Push ID of the command which triggered the transaction error on RI */ +#define ADF_GEN4_RIERRPUSHID 0x41A334 + +/* Pull ID of the command which triggered the transaction error on RI */ +#define ADF_GEN4_RIERRPULLID 0x41A338 + +/* TI CPP interface status register */ +#define ADF_GEN4_TICPPINTSTS 0x50053C + +/* + * Uncorrectable error mask in TICPPINTSTS register + * BIT(0) - value of 1 indicates that the TI asserted + * the CPP error signal during a push + * BIT(1) - value of 1 indicates that the TI detected + * the CPP error signal asserted during a pull + * BIT(2) - value of 1 indicates that the TI detected + * a pull data parity error + */ +#define ADF_GEN4_TICPPINTSTS_BITMASK \ + (BIT(0) | BIT(1) | BIT(2)) + +/* TI CPP interface status register control */ +#define ADF_GEN4_TICPPINTCTL 0x500538 + +/* + * Control bit mask for TICPPINTCTL register + * BIT(0) - value of 1 enables error detection and reporting on + * the TI CPP Push interface + * BIT(1) - value of 1 enables error detection and reporting on + * the TI CPP Push interface + * BIT(2) - value of 1 enables parity error detection and logging on + * the TI CPP Pull interface + * BIT(3) - value of 1 enables CPP CMD and Pull Data parity checking + * BIT(4) - value of 1 enables TI stop part of stop and scream mode on + * CPP/RF Parity error + */ +#define ADF_GEN4_TICPPINTCTL_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4)) + +/* Push ID of the command which triggered the transaction error on TI */ +#define ADF_GEN4_TIERRPUSHID 0x500540 + +/* Pull ID of the command which triggered the transaction error on TI */ +#define ADF_GEN4_TIERRPULLID 0x500544 + +/* Correctable error in ARAM agent register */ +#define ADF_GEN4_REG_ARAMCERR 0x1700 + +#define ADF_GEN4_REG_ARAMCERR_BIT BIT(0) + +/* + * Correctable error enablement in ARAM bit mask + * BIT(3) - enable ARAM RAM to fix and log correctable error + * BIT(26) - enables ARAM agent to generate interrupt for correctable error + */ +#define ADF_GEN4_REG_ARAMCERR_EN_BITMASK (BIT(3) | BIT(26)) + +/* Correctable error address in ARAM agent register */ +#define ADF_GEN4_REG_ARAMCERRAD 0x1708 + +/* Uncorrectable error in ARAM agent register */ +#define ADF_GEN4_REG_ARAMUERR 0x1704 + +/* + * ARAM error bit mask + * BIT(0) - indicates error logged in ARAMCERR or ARAMUCERR + * BIT(18) - indicates uncorrectable multiple errors in ARAM agent + */ +#define ADF_GEN4_REG_ARAMUERR_ERROR_BIT BIT(0) +#define ADF_GEN4_REG_ARAMUERR_MULTI_ERRORS_BIT BIT(18) + +/* + * Uncorrectable error enablement in ARAM bit mask + * BIT(3) - enable ARAM RAM to fix and log uncorrectable error + * BIT(19) - enables ARAM agent to generate interrupt for uncorrectable error + */ +#define ADF_GEN4_REG_ARAMUERR_EN_BITMASK (BIT(3) | BIT(19)) + +/* Unorrectable error address in ARAM agent register */ +#define ADF_GEN4_REG_ARAMUERRAD 0x170C + +/* Uncorrectable error transaction push/pull ID registers*/ +#define ADF_GEN4_REG_ERRPPID_LO 0x1714 +#define ADF_GEN4_REG_ERRPPID_HI 0x1718 + +/* ARAM ECC block error enablement */ +#define ADF_GEN4_REG_ARAMCERRUERR_EN 0x1808 + +/* + * ARAM ECC block error control bit masks + * BIT(0) - enable ARAM CD ECC block error detecting + * BIT(1) - enable ARAM pull request ECC error detecting + * BIT(2) - enable ARAM command dispatch ECC error detecting + * BIT(3) - enable ARAM read datapath push ECC error detecting + * BIT(4) - enable ARAM read datapath pull ECC error detecting + * BIT(5) - enable ARAM RMW ECC error detecting + * BIT(6) - enable ARAM write datapath RMW ECC error detecting + * BIT(7) - enable ARAM write datapath ECC error detecting + */ +#define ADF_GEN4_REG_ARAMCERRUERR_EN_BITMASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4) | \ + BIT(5) | BIT(6) | BIT(7)) + +/* ARAM misc memory target error registers*/ +#define ADF_GEN4_REG_CPPMEMTGTERR 0x1710 + +/* + * ARAM misc memory target error bit masks + * BIT(0) - indicates an error in ARAM target memory + * BIT(1) - indicates multiple errors in ARAM target memory + * BIT(4) - indicates pull error in ARAM target memory + * BIT(5) - indicates parity pull error in ARAM target memory + * BIT(6) - indicates push error in ARAM target memory + */ +#define ADF_GEN4_REG_CPPMEMTGTERR_BITMASK \ + (BIT(0) | BIT(4) | BIT(5) | BIT(6)) + +#define ADF_GEN4_REG_CPPMEMTGTERR_MULTI_ERRORS_BIT BIT(1) + +/* + * ARAM misc memory target error enablement mask + * BIT(2) - enables CPP memory to detect and log push/pull data error + * BIT(7) - enables push/pull error to generate interrupts to RI + * BIT(8) - enables ARAM to check parity on pull data and CPP command buses + * BIT(9) - enables ARAM to autopush to AE when push/parity error is detected + * on lookaside DT + */ +#define ADF_GEN4_REG_CPPMEMTGTERR_EN_BITMASK \ + (BIT(2) | BIT(7) | BIT(8) | BIT(9)) + +/* ATU fault status register */ +#define ADF_GEN4_ATUFAULTSTATUS(i) (0x506000 + ((i) * 0x4)) + +#define ADF_GEN4_ATUFAULTSTATUS_BIT BIT(0) + +/* Command Parity error detected on IOSFP Command to QAT */ +#define ADF_GEN4_RIMISCSTS_BIT BIT(0) + +void adf_gen4_init_ras_ops(struct adf_ras_ops *ras_ops); + +#endif /* ADF_GEN4_RAS_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_tl.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_tl.c new file mode 100644 index 000000000000..c7ad8cf07863 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_tl.c @@ -0,0 +1,154 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright (c) 2023 Intel Corporation. */ +#include <linux/export.h> +#include <linux/kernel.h> + +#include "adf_gen4_tl.h" +#include "adf_telemetry.h" +#include "adf_tl_debugfs.h" + +#define ADF_GEN4_TL_DEV_REG_OFF(reg) ADF_TL_DEV_REG_OFF(reg, gen4) + +#define ADF_GEN4_TL_RP_REG_OFF(reg) ADF_TL_RP_REG_OFF(reg, gen4) + +#define ADF_GEN4_TL_SL_UTIL_COUNTER(_name) \ + ADF_TL_COUNTER("util_" #_name, \ + ADF_TL_SIMPLE_COUNT, \ + ADF_TL_SLICE_REG_OFF(_name, reg_tm_slice_util, gen4)) + +#define ADF_GEN4_TL_SL_EXEC_COUNTER(_name) \ + ADF_TL_COUNTER("exec_" #_name, \ + ADF_TL_SIMPLE_COUNT, \ + ADF_TL_SLICE_REG_OFF(_name, reg_tm_slice_exec_cnt, gen4)) + +/* Device level counters. */ +static const struct adf_tl_dbg_counter dev_counters[] = { + /* PCIe partial transactions. */ + ADF_TL_COUNTER(PCI_TRANS_CNT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_pci_trans_cnt)), + /* Max read latency[ns]. */ + ADF_TL_COUNTER(MAX_RD_LAT_NAME, ADF_TL_COUNTER_NS, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_rd_lat_max)), + /* Read latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(RD_LAT_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_rd_lat_acc), + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_rd_cmpl_cnt)), + /* Max get to put latency[ns]. */ + ADF_TL_COUNTER(MAX_LAT_NAME, ADF_TL_COUNTER_NS, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_gp_lat_max)), + /* Get to put latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(LAT_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_gp_lat_acc), + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_ae_put_cnt)), + /* PCIe write bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_IN_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_bw_in)), + /* PCIe read bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_OUT_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_bw_out)), + /* Page request latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(PAGE_REQ_LAT_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_at_page_req_lat_acc), + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_at_page_req_cnt)), + /* Page translation latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(AT_TRANS_LAT_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_at_trans_lat_acc), + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_at_trans_lat_cnt)), + /* Maximum uTLB used. */ + ADF_TL_COUNTER(AT_MAX_UTLB_USED_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN4_TL_DEV_REG_OFF(reg_tl_at_max_tlb_used)), +}; + +/* Slice utilization counters. */ +static const struct adf_tl_dbg_counter sl_util_counters[ADF_TL_SL_CNT_COUNT] = { + /* Compression slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(cpr), + /* Translator slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(xlt), + /* Decompression slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(dcpr), + /* PKE utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(pke), + /* Wireless Authentication slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(wat), + /* Wireless Cipher slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(wcp), + /* UCS slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(ucs), + /* Cipher slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(cph), + /* Authentication slice utilization. */ + ADF_GEN4_TL_SL_UTIL_COUNTER(ath), +}; + +/* Slice execution counters. */ +static const struct adf_tl_dbg_counter sl_exec_counters[ADF_TL_SL_CNT_COUNT] = { + /* Compression slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(cpr), + /* Translator slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(xlt), + /* Decompression slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(dcpr), + /* PKE execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(pke), + /* Wireless Authentication slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(wat), + /* Wireless Cipher slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(wcp), + /* UCS slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(ucs), + /* Cipher slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(cph), + /* Authentication slice execution count. */ + ADF_GEN4_TL_SL_EXEC_COUNTER(ath), +}; + +/* Ring pair counters. */ +static const struct adf_tl_dbg_counter rp_counters[] = { + /* PCIe partial transactions. */ + ADF_TL_COUNTER(PCI_TRANS_CNT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_pci_trans_cnt)), + /* Get to put latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(LAT_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_gp_lat_acc), + ADF_GEN4_TL_RP_REG_OFF(reg_tl_ae_put_cnt)), + /* PCIe write bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_IN_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_bw_in)), + /* PCIe read bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_OUT_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_bw_out)), + /* Message descriptor DevTLB hit rate. */ + ADF_TL_COUNTER(AT_GLOB_DTLB_HIT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_at_glob_devtlb_hit)), + /* Message descriptor DevTLB miss rate. */ + ADF_TL_COUNTER(AT_GLOB_DTLB_MISS_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_at_glob_devtlb_miss)), + /* Payload DevTLB hit rate. */ + ADF_TL_COUNTER(AT_PAYLD_DTLB_HIT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_at_payld_devtlb_hit)), + /* Payload DevTLB miss rate. */ + ADF_TL_COUNTER(AT_PAYLD_DTLB_MISS_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN4_TL_RP_REG_OFF(reg_tl_at_payld_devtlb_miss)), +}; + +void adf_gen4_init_tl_data(struct adf_tl_hw_data *tl_data) +{ + tl_data->layout_sz = ADF_GEN4_TL_LAYOUT_SZ; + tl_data->slice_reg_sz = ADF_GEN4_TL_SLICE_REG_SZ; + tl_data->rp_reg_sz = ADF_GEN4_TL_RP_REG_SZ; + tl_data->num_hbuff = ADF_GEN4_TL_NUM_HIST_BUFFS; + tl_data->max_rp = ADF_GEN4_TL_MAX_RP_NUM; + tl_data->msg_cnt_off = ADF_GEN4_TL_MSG_CNT_OFF; + tl_data->cpp_ns_per_cycle = ADF_GEN4_CPP_NS_PER_CYCLE; + tl_data->bw_units_to_bytes = ADF_GEN4_TL_BW_HW_UNITS_TO_BYTES; + + tl_data->dev_counters = dev_counters; + tl_data->num_dev_counters = ARRAY_SIZE(dev_counters); + tl_data->sl_util_counters = sl_util_counters; + tl_data->sl_exec_counters = sl_exec_counters; + tl_data->rp_counters = rp_counters; + tl_data->num_rp_counters = ARRAY_SIZE(rp_counters); + tl_data->max_sl_cnt = ADF_GEN4_TL_MAX_SLICES_PER_TYPE; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_tl_data); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_tl.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_tl.h new file mode 100644 index 000000000000..32df4163beb9 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_tl.h @@ -0,0 +1,158 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright (c) 2023 Intel Corporation. */ +#ifndef ADF_GEN4_TL_H +#define ADF_GEN4_TL_H + +#include <linux/stddef.h> +#include <linux/types.h> + +struct adf_tl_hw_data; + +/* Computation constants. */ +#define ADF_GEN4_CPP_NS_PER_CYCLE 2 +#define ADF_GEN4_TL_BW_HW_UNITS_TO_BYTES 64 + +/* Maximum aggregation time. Value in milliseconds. */ +#define ADF_GEN4_TL_MAX_AGGR_TIME_MS 4000 +/* Num of buffers to store historic values. */ +#define ADF_GEN4_TL_NUM_HIST_BUFFS \ + (ADF_GEN4_TL_MAX_AGGR_TIME_MS / ADF_TL_DATA_WR_INTERVAL_MS) + +/* Max number of HW resources of one type. */ +#define ADF_GEN4_TL_MAX_SLICES_PER_TYPE 24 + +/* Max number of simultaneously monitored ring pairs. */ +#define ADF_GEN4_TL_MAX_RP_NUM 4 + +/** + * struct adf_gen4_tl_slice_data_regs - HW slice data as populated by FW. + * @reg_tm_slice_exec_cnt: Slice execution count. + * @reg_tm_slice_util: Slice utilization. + */ +struct adf_gen4_tl_slice_data_regs { + __u32 reg_tm_slice_exec_cnt; + __u32 reg_tm_slice_util; +}; + +#define ADF_GEN4_TL_SLICE_REG_SZ sizeof(struct adf_gen4_tl_slice_data_regs) + +/** + * struct adf_gen4_tl_device_data_regs - This structure stores device telemetry + * counter values as are being populated periodically by device. + * @reg_tl_rd_lat_acc: read latency accumulator + * @reg_tl_gp_lat_acc: get-put latency accumulator + * @reg_tl_at_page_req_lat_acc: AT/DevTLB page request latency accumulator + * @reg_tl_at_trans_lat_acc: DevTLB transaction latency accumulator + * @reg_tl_re_acc: accumulated ring empty time + * @reg_tl_pci_trans_cnt: PCIe partial transactions + * @reg_tl_rd_lat_max: maximum logged read latency + * @reg_tl_rd_cmpl_cnt: read requests completed count + * @reg_tl_gp_lat_max: maximum logged get to put latency + * @reg_tl_ae_put_cnt: Accelerator Engine put counts across all rings + * @reg_tl_bw_in: PCIe write bandwidth + * @reg_tl_bw_out: PCIe read bandwidth + * @reg_tl_at_page_req_cnt: DevTLB page requests count + * @reg_tl_at_trans_lat_cnt: DevTLB transaction latency samples count + * @reg_tl_at_max_tlb_used: maximum uTLB used + * @reg_tl_re_cnt: ring empty time samples count + * @reserved: reserved + * @ath_slices: array of Authentication slices utilization registers + * @cph_slices: array of Cipher slices utilization registers + * @cpr_slices: array of Compression slices utilization registers + * @xlt_slices: array of Translator slices utilization registers + * @dcpr_slices: array of Decompression slices utilization registers + * @pke_slices: array of PKE slices utilization registers + * @ucs_slices: array of UCS slices utilization registers + * @wat_slices: array of Wireless Authentication slices utilization registers + * @wcp_slices: array of Wireless Cipher slices utilization registers + */ +struct adf_gen4_tl_device_data_regs { + __u64 reg_tl_rd_lat_acc; + __u64 reg_tl_gp_lat_acc; + __u64 reg_tl_at_page_req_lat_acc; + __u64 reg_tl_at_trans_lat_acc; + __u64 reg_tl_re_acc; + __u32 reg_tl_pci_trans_cnt; + __u32 reg_tl_rd_lat_max; + __u32 reg_tl_rd_cmpl_cnt; + __u32 reg_tl_gp_lat_max; + __u32 reg_tl_ae_put_cnt; + __u32 reg_tl_bw_in; + __u32 reg_tl_bw_out; + __u32 reg_tl_at_page_req_cnt; + __u32 reg_tl_at_trans_lat_cnt; + __u32 reg_tl_at_max_tlb_used; + __u32 reg_tl_re_cnt; + __u32 reserved; + struct adf_gen4_tl_slice_data_regs ath_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs cph_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs cpr_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs xlt_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs dcpr_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs pke_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs ucs_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs wat_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; + struct adf_gen4_tl_slice_data_regs wcp_slices[ADF_GEN4_TL_MAX_SLICES_PER_TYPE]; +}; + +/** + * struct adf_gen4_tl_ring_pair_data_regs - This structure stores Ring Pair + * telemetry counter values as are being populated periodically by device. + * @reg_tl_gp_lat_acc: get-put latency accumulator + * @reserved: reserved + * @reg_tl_pci_trans_cnt: PCIe partial transactions + * @reg_tl_ae_put_cnt: Accelerator Engine put counts across all rings + * @reg_tl_bw_in: PCIe write bandwidth + * @reg_tl_bw_out: PCIe read bandwidth + * @reg_tl_at_glob_devtlb_hit: Message descriptor DevTLB hit rate + * @reg_tl_at_glob_devtlb_miss: Message descriptor DevTLB miss rate + * @reg_tl_at_payld_devtlb_hit: Payload DevTLB hit rate + * @reg_tl_at_payld_devtlb_miss: Payload DevTLB miss rate + * @reg_tl_re_cnt: ring empty time samples count + * @reserved1: reserved + */ +struct adf_gen4_tl_ring_pair_data_regs { + __u64 reg_tl_gp_lat_acc; + __u64 reserved; + __u32 reg_tl_pci_trans_cnt; + __u32 reg_tl_ae_put_cnt; + __u32 reg_tl_bw_in; + __u32 reg_tl_bw_out; + __u32 reg_tl_at_glob_devtlb_hit; + __u32 reg_tl_at_glob_devtlb_miss; + __u32 reg_tl_at_payld_devtlb_hit; + __u32 reg_tl_at_payld_devtlb_miss; + __u32 reg_tl_re_cnt; + __u32 reserved1; +}; + +#define ADF_GEN4_TL_RP_REG_SZ sizeof(struct adf_gen4_tl_ring_pair_data_regs) + +/** + * struct adf_gen4_tl_layout - This structure represents entire telemetry + * counters data: Device + 4 Ring Pairs as are being populated periodically + * by device. + * @tl_device_data_regs: structure of device telemetry registers + * @tl_ring_pairs_data_regs: array of ring pairs telemetry registers + * @reg_tl_msg_cnt: telemetry messages counter + * @reserved: reserved + */ +struct adf_gen4_tl_layout { + struct adf_gen4_tl_device_data_regs tl_device_data_regs; + struct adf_gen4_tl_ring_pair_data_regs + tl_ring_pairs_data_regs[ADF_GEN4_TL_MAX_RP_NUM]; + __u32 reg_tl_msg_cnt; + __u32 reserved; +}; + +#define ADF_GEN4_TL_LAYOUT_SZ sizeof(struct adf_gen4_tl_layout) +#define ADF_GEN4_TL_MSG_CNT_OFF offsetof(struct adf_gen4_tl_layout, reg_tl_msg_cnt) + +#ifdef CONFIG_DEBUG_FS +void adf_gen4_init_tl_data(struct adf_tl_hw_data *tl_data); +#else +static inline void adf_gen4_init_tl_data(struct adf_tl_hw_data *tl_data) +{ +} +#endif /* CONFIG_DEBUG_FS */ +#endif /* ADF_GEN4_TL_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_vf_mig.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_vf_mig.c new file mode 100644 index 000000000000..adb21656a3ba --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_vf_mig.c @@ -0,0 +1,1011 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2024 Intel Corporation */ +#include <linux/delay.h> +#include <linux/dev_printk.h> +#include <linux/kernel.h> +#include <linux/slab.h> +#include <linux/string.h> +#include <linux/types.h> +#include <asm/errno.h> + +#include "adf_accel_devices.h" +#include "adf_bank_state.h" +#include "adf_common_drv.h" +#include "adf_gen4_hw_data.h" +#include "adf_gen4_pfvf.h" +#include "adf_pfvf_utils.h" +#include "adf_mstate_mgr.h" +#include "adf_gen4_vf_mig.h" + +#define ADF_GEN4_VF_MSTATE_SIZE 4096 +#define ADF_GEN4_PFVF_RSP_TIMEOUT_US 5000 + +static int adf_gen4_vfmig_save_setup(struct qat_mig_dev *mdev); +static int adf_gen4_vfmig_load_setup(struct qat_mig_dev *mdev, int len); + +static int adf_gen4_vfmig_init_device(struct qat_mig_dev *mdev) +{ + u8 *state; + + state = kmalloc(ADF_GEN4_VF_MSTATE_SIZE, GFP_KERNEL); + if (!state) + return -ENOMEM; + + mdev->state = state; + mdev->state_size = ADF_GEN4_VF_MSTATE_SIZE; + mdev->setup_size = 0; + mdev->remote_setup_size = 0; + + return 0; +} + +static void adf_gen4_vfmig_cleanup_device(struct qat_mig_dev *mdev) +{ + kfree(mdev->state); + mdev->state = NULL; +} + +static void adf_gen4_vfmig_reset_device(struct qat_mig_dev *mdev) +{ + mdev->setup_size = 0; + mdev->remote_setup_size = 0; +} + +static int adf_gen4_vfmig_open_device(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vfmig; + + vf_info = &accel_dev->pf.vf_info[mdev->vf_id]; + + vfmig = kzalloc(sizeof(*vfmig), GFP_KERNEL); + if (!vfmig) + return -ENOMEM; + + vfmig->mstate_mgr = adf_mstate_mgr_new(mdev->state, mdev->state_size); + if (!vfmig->mstate_mgr) { + kfree(vfmig); + return -ENOMEM; + } + vf_info->mig_priv = vfmig; + mdev->setup_size = 0; + mdev->remote_setup_size = 0; + + return 0; +} + +static void adf_gen4_vfmig_close_device(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vfmig; + + vf_info = &accel_dev->pf.vf_info[mdev->vf_id]; + if (vf_info->mig_priv) { + vfmig = vf_info->mig_priv; + adf_mstate_mgr_destroy(vfmig->mstate_mgr); + kfree(vfmig); + vf_info->mig_priv = NULL; + } +} + +static int adf_gen4_vfmig_suspend_device(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vf_mig; + u32 vf_nr = mdev->vf_id; + int ret, i; + + vf_info = &accel_dev->pf.vf_info[vf_nr]; + vf_mig = vf_info->mig_priv; + + /* Stop all inflight jobs */ + for (i = 0; i < hw_data->num_banks_per_vf; i++) { + u32 pf_bank_nr = i + vf_nr * hw_data->num_banks_per_vf; + + ret = adf_gen4_bank_drain_start(accel_dev, pf_bank_nr, + ADF_RPRESET_POLL_TIMEOUT_US); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to drain bank %d for vf_nr %d\n", i, + vf_nr); + return ret; + } + vf_mig->bank_stopped[i] = true; + + adf_gen4_bank_quiesce_coal_timer(accel_dev, pf_bank_nr, + ADF_COALESCED_POLL_TIMEOUT_US); + } + + return 0; +} + +static int adf_gen4_vfmig_resume_device(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vf_mig; + u32 vf_nr = mdev->vf_id; + int i; + + vf_info = &accel_dev->pf.vf_info[vf_nr]; + vf_mig = vf_info->mig_priv; + + for (i = 0; i < hw_data->num_banks_per_vf; i++) { + u32 pf_bank_nr = i + vf_nr * hw_data->num_banks_per_vf; + + if (vf_mig->bank_stopped[i]) { + adf_gen4_bank_drain_finish(accel_dev, pf_bank_nr); + vf_mig->bank_stopped[i] = false; + } + } + + return 0; +} + +struct adf_vf_bank_info { + struct adf_accel_dev *accel_dev; + u32 vf_nr; + u32 bank_nr; +}; + +struct mig_user_sla { + enum adf_base_services srv; + u64 rp_mask; + u32 cir; + u32 pir; +}; + +static int adf_mstate_sla_check(struct adf_mstate_mgr *sub_mgr, u8 *src_buf, + u32 src_size, void *opaque) +{ + struct adf_mstate_vreginfo _sinfo = { src_buf, src_size }; + struct adf_mstate_vreginfo *sinfo = &_sinfo, *dinfo = opaque; + u32 src_sla_cnt = sinfo->size / sizeof(struct mig_user_sla); + u32 dst_sla_cnt = dinfo->size / sizeof(struct mig_user_sla); + struct mig_user_sla *src_slas = sinfo->addr; + struct mig_user_sla *dst_slas = dinfo->addr; + int i, j; + + for (i = 0; i < src_sla_cnt; i++) { + for (j = 0; j < dst_sla_cnt; j++) { + if (src_slas[i].srv != dst_slas[j].srv || + src_slas[i].rp_mask != dst_slas[j].rp_mask) + continue; + + if (src_slas[i].cir > dst_slas[j].cir || + src_slas[i].pir > dst_slas[j].pir) { + pr_err("QAT: DST VF rate limiting mismatch.\n"); + return -EINVAL; + } + break; + } + + if (j == dst_sla_cnt) { + pr_err("QAT: SRC VF rate limiting mismatch - SRC srv %d and rp_mask 0x%llx.\n", + src_slas[i].srv, src_slas[i].rp_mask); + return -EINVAL; + } + } + + return 0; +} + +static inline int adf_mstate_check_cap_size(u32 src_sz, u32 dst_sz, u32 max_sz) +{ + if (src_sz > max_sz || dst_sz > max_sz) + return -EINVAL; + else + return 0; +} + +static int adf_mstate_compatver_check(struct adf_mstate_mgr *sub_mgr, + u8 *src_buf, u32 src_sz, void *opaque) +{ + struct adf_mstate_vreginfo *info = opaque; + u8 compat = 0; + u8 *pcompat; + + if (src_sz != info->size) { + pr_debug("QAT: State mismatch (compat version size), current %u, expected %u\n", + src_sz, info->size); + return -EINVAL; + } + + memcpy(info->addr, src_buf, info->size); + pcompat = info->addr; + if (*pcompat == 0) { + pr_warn("QAT: Unable to determine the version of VF\n"); + return 0; + } + + compat = adf_vf_compat_checker(*pcompat); + if (compat == ADF_PF2VF_VF_INCOMPATIBLE) { + pr_debug("QAT: SRC VF driver (ver=%u) is incompatible with DST PF driver (ver=%u)\n", + *pcompat, ADF_PFVF_COMPAT_THIS_VERSION); + return -EINVAL; + } + + if (compat == ADF_PF2VF_VF_COMPAT_UNKNOWN) + pr_debug("QAT: SRC VF driver (ver=%u) is newer than DST PF driver (ver=%u)\n", + *pcompat, ADF_PFVF_COMPAT_THIS_VERSION); + + return 0; +} + +/* + * adf_mstate_capmask_compare() - compare QAT device capability mask + * @sinfo: Pointer to source capability info + * @dinfo: Pointer to target capability info + * + * This function compares the capability mask between source VF and target VF + * + * Returns: 0 if target capability mask is identical to source capability mask, + * 1 if target mask can represent all the capabilities represented by source mask, + * -1 if target mask can't represent all the capabilities represented by source + * mask. + */ +static int adf_mstate_capmask_compare(struct adf_mstate_vreginfo *sinfo, + struct adf_mstate_vreginfo *dinfo) +{ + u64 src = 0, dst = 0; + + if (adf_mstate_check_cap_size(sinfo->size, dinfo->size, sizeof(u64))) { + pr_debug("QAT: Unexpected capability size %u %u %zu\n", + sinfo->size, dinfo->size, sizeof(u64)); + return -1; + } + + memcpy(&src, sinfo->addr, sinfo->size); + memcpy(&dst, dinfo->addr, dinfo->size); + + pr_debug("QAT: Check cap compatibility of cap %llu %llu\n", src, dst); + + if (src == dst) + return 0; + + if ((src | dst) == dst) + return 1; + + return -1; +} + +static int adf_mstate_capmask_superset(struct adf_mstate_mgr *sub_mgr, u8 *buf, + u32 size, void *opa) +{ + struct adf_mstate_vreginfo sinfo = { buf, size }; + + if (adf_mstate_capmask_compare(&sinfo, opa) >= 0) + return 0; + + return -EINVAL; +} + +static int adf_mstate_capmask_equal(struct adf_mstate_mgr *sub_mgr, u8 *buf, + u32 size, void *opa) +{ + struct adf_mstate_vreginfo sinfo = { buf, size }; + + if (adf_mstate_capmask_compare(&sinfo, opa) == 0) + return 0; + + return -EINVAL; +} + +static int adf_mstate_set_vreg(struct adf_mstate_mgr *sub_mgr, u8 *buf, + u32 size, void *opa) +{ + struct adf_mstate_vreginfo *info = opa; + + if (size != info->size) { + pr_debug("QAT: Unexpected cap size %u %u\n", size, info->size); + return -EINVAL; + } + memcpy(info->addr, buf, info->size); + + return 0; +} + +static u32 adf_gen4_vfmig_get_slas(struct adf_accel_dev *accel_dev, u32 vf_nr, + struct mig_user_sla *pmig_slas) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_rl *rl_data = accel_dev->rate_limiting; + struct rl_sla **sla_type_arr = NULL; + u64 rp_mask, rp_index; + u32 max_num_sla; + u32 sla_cnt = 0; + int i, j; + + if (!accel_dev->rate_limiting) + return 0; + + rp_index = vf_nr * hw_data->num_banks_per_vf; + max_num_sla = adf_rl_get_sla_arr_of_type(rl_data, RL_LEAF, &sla_type_arr); + + for (i = 0; i < max_num_sla; i++) { + if (!sla_type_arr[i]) + continue; + + rp_mask = 0; + for (j = 0; j < sla_type_arr[i]->ring_pairs_cnt; j++) + rp_mask |= BIT(sla_type_arr[i]->ring_pairs_ids[j]); + + if (rp_mask & GENMASK_ULL(rp_index + 3, rp_index)) { + pmig_slas->rp_mask = rp_mask; + pmig_slas->cir = sla_type_arr[i]->cir; + pmig_slas->pir = sla_type_arr[i]->pir; + pmig_slas->srv = sla_type_arr[i]->srv; + pmig_slas++; + sla_cnt++; + } + } + + return sla_cnt; +} + +static int adf_gen4_vfmig_load_etr_regs(struct adf_mstate_mgr *sub_mgr, + u8 *state, u32 size, void *opa) +{ + struct adf_vf_bank_info *vf_bank_info = opa; + struct adf_accel_dev *accel_dev = vf_bank_info->accel_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u32 pf_bank_nr; + int ret; + + pf_bank_nr = vf_bank_info->bank_nr + vf_bank_info->vf_nr * hw_data->num_banks_per_vf; + ret = hw_data->bank_state_restore(accel_dev, pf_bank_nr, + (struct adf_bank_state *)state); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to load regs for vf%d bank%d\n", + vf_bank_info->vf_nr, vf_bank_info->bank_nr); + return ret; + } + + return 0; +} + +static int adf_gen4_vfmig_load_etr_bank(struct adf_accel_dev *accel_dev, + u32 vf_nr, u32 bank_nr, + struct adf_mstate_mgr *mstate_mgr) +{ + struct adf_vf_bank_info vf_bank_info = {accel_dev, vf_nr, bank_nr}; + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct adf_mstate_mgr sub_sects_mgr; + char bank_ids[ADF_MSTATE_ID_LEN]; + + snprintf(bank_ids, sizeof(bank_ids), ADF_MSTATE_BANK_IDX_IDS "%x", bank_nr); + subsec = adf_mstate_sect_lookup(mstate_mgr, bank_ids, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), + "Failed to lookup sec %s for vf%d bank%d\n", + ADF_MSTATE_BANK_IDX_IDS, vf_nr, bank_nr); + return -EINVAL; + } + + adf_mstate_mgr_init_from_psect(&sub_sects_mgr, subsec); + l2_subsec = adf_mstate_sect_lookup(&sub_sects_mgr, ADF_MSTATE_ETR_REGS_IDS, + adf_gen4_vfmig_load_etr_regs, + &vf_bank_info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), + "Failed to add sec %s for vf%d bank%d\n", + ADF_MSTATE_ETR_REGS_IDS, vf_nr, bank_nr); + return -EINVAL; + } + + return 0; +} + +static int adf_gen4_vfmig_load_etr(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + struct adf_mstate_mgr sub_sects_mgr; + struct adf_mstate_sect_h *subsec; + int ret, i; + + subsec = adf_mstate_sect_lookup(mstate_mgr, ADF_MSTATE_ETRB_IDS, NULL, + NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to load sec %s\n", + ADF_MSTATE_ETRB_IDS); + return -EINVAL; + } + + adf_mstate_mgr_init_from_psect(&sub_sects_mgr, subsec); + for (i = 0; i < hw_data->num_banks_per_vf; i++) { + ret = adf_gen4_vfmig_load_etr_bank(accel_dev, vf_nr, i, + &sub_sects_mgr); + if (ret) + return ret; + } + + return 0; +} + +static int adf_gen4_vfmig_load_misc(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + void __iomem *csr = adf_get_pmisc_base(accel_dev); + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct adf_mstate_mgr sub_sects_mgr; + struct { + char *id; + u64 ofs; + } misc_states[] = { + {ADF_MSTATE_VINTMSK_IDS, ADF_GEN4_VINTMSK_OFFSET(vf_nr)}, + {ADF_MSTATE_VINTMSK_PF2VM_IDS, ADF_GEN4_VINTMSKPF2VM_OFFSET(vf_nr)}, + {ADF_MSTATE_PF2VM_IDS, ADF_GEN4_PF2VM_OFFSET(vf_nr)}, + {ADF_MSTATE_VM2PF_IDS, ADF_GEN4_VM2PF_OFFSET(vf_nr)}, + }; + int i; + + subsec = adf_mstate_sect_lookup(mstate_mgr, ADF_MSTATE_MISCB_IDS, NULL, + NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to load sec %s\n", + ADF_MSTATE_MISCB_IDS); + return -EINVAL; + } + + adf_mstate_mgr_init_from_psect(&sub_sects_mgr, subsec); + for (i = 0; i < ARRAY_SIZE(misc_states); i++) { + struct adf_mstate_vreginfo info; + u32 regv; + + info.addr = ®v; + info.size = sizeof(regv); + l2_subsec = adf_mstate_sect_lookup(&sub_sects_mgr, + misc_states[i].id, + adf_mstate_set_vreg, + &info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), + "Failed to load sec %s\n", misc_states[i].id); + return -EINVAL; + } + ADF_CSR_WR(csr, misc_states[i].ofs, regv); + } + + return 0; +} + +static int adf_gen4_vfmig_load_generic(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct mig_user_sla dst_slas[RL_RP_CNT_PER_LEAF_MAX] = { }; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct adf_mstate_mgr sub_sects_mgr; + u32 dst_sla_cnt; + struct { + char *id; + int (*action)(struct adf_mstate_mgr *sub_mgr, u8 *buf, u32 size, void *opa); + struct adf_mstate_vreginfo info; + } gen_states[] = { + {ADF_MSTATE_IOV_INIT_IDS, adf_mstate_set_vreg, + {&vf_info->init, sizeof(vf_info->init)}}, + {ADF_MSTATE_COMPAT_VER_IDS, adf_mstate_compatver_check, + {&vf_info->vf_compat_ver, sizeof(vf_info->vf_compat_ver)}}, + {ADF_MSTATE_SLA_IDS, adf_mstate_sla_check, {dst_slas, 0}}, + }; + int i; + + subsec = adf_mstate_sect_lookup(mstate_mgr, ADF_MSTATE_GEN_IDS, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to load sec %s\n", + ADF_MSTATE_GEN_IDS); + return -EINVAL; + } + + adf_mstate_mgr_init_from_psect(&sub_sects_mgr, subsec); + for (i = 0; i < ARRAY_SIZE(gen_states); i++) { + if (gen_states[i].info.addr == dst_slas) { + dst_sla_cnt = adf_gen4_vfmig_get_slas(accel_dev, vf_nr, dst_slas); + gen_states[i].info.size = dst_sla_cnt * sizeof(struct mig_user_sla); + } + + l2_subsec = adf_mstate_sect_lookup(&sub_sects_mgr, + gen_states[i].id, + gen_states[i].action, + &gen_states[i].info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to load sec %s\n", + gen_states[i].id); + return -EINVAL; + } + } + + return 0; +} + +static int adf_gen4_vfmig_load_config(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct adf_mstate_mgr sub_sects_mgr; + struct { + char *id; + int (*action)(struct adf_mstate_mgr *sub_mgr, u8 *buf, u32 size, void *opa); + struct adf_mstate_vreginfo info; + } setups[] = { + {ADF_MSTATE_GEN_CAP_IDS, adf_mstate_capmask_superset, + {&hw_data->accel_capabilities_mask, sizeof(hw_data->accel_capabilities_mask)}}, + {ADF_MSTATE_GEN_SVCMAP_IDS, adf_mstate_capmask_equal, + {&hw_data->ring_to_svc_map, sizeof(hw_data->ring_to_svc_map)}}, + {ADF_MSTATE_GEN_EXTDC_IDS, adf_mstate_capmask_superset, + {&hw_data->extended_dc_capabilities, sizeof(hw_data->extended_dc_capabilities)}}, + }; + int i; + + subsec = adf_mstate_sect_lookup(mstate_mgr, ADF_MSTATE_CONFIG_IDS, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to load sec %s\n", + ADF_MSTATE_CONFIG_IDS); + return -EINVAL; + } + + adf_mstate_mgr_init_from_psect(&sub_sects_mgr, subsec); + for (i = 0; i < ARRAY_SIZE(setups); i++) { + l2_subsec = adf_mstate_sect_lookup(&sub_sects_mgr, setups[i].id, + setups[i].action, &setups[i].info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to load sec %s\n", + setups[i].id); + return -EINVAL; + } + } + + return 0; +} + +static int adf_gen4_vfmig_save_etr_regs(struct adf_mstate_mgr *subs, u8 *state, + u32 size, void *opa) +{ + struct adf_vf_bank_info *vf_bank_info = opa; + struct adf_accel_dev *accel_dev = vf_bank_info->accel_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u32 pf_bank_nr; + int ret; + + pf_bank_nr = vf_bank_info->bank_nr; + pf_bank_nr += vf_bank_info->vf_nr * hw_data->num_banks_per_vf; + + ret = hw_data->bank_state_save(accel_dev, pf_bank_nr, + (struct adf_bank_state *)state); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to save regs for vf%d bank%d\n", + vf_bank_info->vf_nr, vf_bank_info->bank_nr); + return ret; + } + + return sizeof(struct adf_bank_state); +} + +static int adf_gen4_vfmig_save_etr_bank(struct adf_accel_dev *accel_dev, + u32 vf_nr, u32 bank_nr, + struct adf_mstate_mgr *mstate_mgr) +{ + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct adf_vf_bank_info vf_bank_info; + struct adf_mstate_mgr sub_sects_mgr; + char bank_ids[ADF_MSTATE_ID_LEN]; + + snprintf(bank_ids, sizeof(bank_ids), ADF_MSTATE_BANK_IDX_IDS "%x", bank_nr); + + subsec = adf_mstate_sect_add(mstate_mgr, bank_ids, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), + "Failed to add sec %s for vf%d bank%d\n", + ADF_MSTATE_BANK_IDX_IDS, vf_nr, bank_nr); + return -EINVAL; + } + + adf_mstate_mgr_init_from_parent(&sub_sects_mgr, mstate_mgr); + vf_bank_info.accel_dev = accel_dev; + vf_bank_info.vf_nr = vf_nr; + vf_bank_info.bank_nr = bank_nr; + l2_subsec = adf_mstate_sect_add(&sub_sects_mgr, ADF_MSTATE_ETR_REGS_IDS, + adf_gen4_vfmig_save_etr_regs, + &vf_bank_info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), + "Failed to add sec %s for vf%d bank%d\n", + ADF_MSTATE_ETR_REGS_IDS, vf_nr, bank_nr); + return -EINVAL; + } + adf_mstate_sect_update(mstate_mgr, &sub_sects_mgr, subsec); + + return 0; +} + +static int adf_gen4_vfmig_save_etr(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + struct adf_mstate_mgr sub_sects_mgr; + struct adf_mstate_sect_h *subsec; + int ret, i; + + subsec = adf_mstate_sect_add(mstate_mgr, ADF_MSTATE_ETRB_IDS, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to add sec %s\n", + ADF_MSTATE_ETRB_IDS); + return -EINVAL; + } + + adf_mstate_mgr_init_from_parent(&sub_sects_mgr, mstate_mgr); + for (i = 0; i < hw_data->num_banks_per_vf; i++) { + ret = adf_gen4_vfmig_save_etr_bank(accel_dev, vf_nr, i, + &sub_sects_mgr); + if (ret) + return ret; + } + adf_mstate_sect_update(mstate_mgr, &sub_sects_mgr, subsec); + + return 0; +} + +static int adf_gen4_vfmig_save_misc(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + void __iomem *csr = adf_get_pmisc_base(accel_dev); + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct adf_mstate_mgr sub_sects_mgr; + struct { + char *id; + u64 offset; + } misc_states[] = { + {ADF_MSTATE_VINTSRC_IDS, ADF_GEN4_VINTSOU_OFFSET(vf_nr)}, + {ADF_MSTATE_VINTMSK_IDS, ADF_GEN4_VINTMSK_OFFSET(vf_nr)}, + {ADF_MSTATE_VINTSRC_PF2VM_IDS, ADF_GEN4_VINTSOUPF2VM_OFFSET(vf_nr)}, + {ADF_MSTATE_VINTMSK_PF2VM_IDS, ADF_GEN4_VINTMSKPF2VM_OFFSET(vf_nr)}, + {ADF_MSTATE_PF2VM_IDS, ADF_GEN4_PF2VM_OFFSET(vf_nr)}, + {ADF_MSTATE_VM2PF_IDS, ADF_GEN4_VM2PF_OFFSET(vf_nr)}, + }; + ktime_t time_exp; + int i; + + subsec = adf_mstate_sect_add(mstate_mgr, ADF_MSTATE_MISCB_IDS, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to add sec %s\n", + ADF_MSTATE_MISCB_IDS); + return -EINVAL; + } + + time_exp = ktime_add_us(ktime_get(), ADF_GEN4_PFVF_RSP_TIMEOUT_US); + while (!mutex_trylock(&vf_info->pfvf_mig_lock)) { + if (ktime_after(ktime_get(), time_exp)) { + dev_err(&GET_DEV(accel_dev), "Failed to get pfvf mig lock\n"); + return -ETIMEDOUT; + } + usleep_range(500, 1000); + } + + adf_mstate_mgr_init_from_parent(&sub_sects_mgr, mstate_mgr); + for (i = 0; i < ARRAY_SIZE(misc_states); i++) { + struct adf_mstate_vreginfo info; + u32 regv; + + info.addr = ®v; + info.size = sizeof(regv); + regv = ADF_CSR_RD(csr, misc_states[i].offset); + + l2_subsec = adf_mstate_sect_add_vreg(&sub_sects_mgr, + misc_states[i].id, + &info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to add sec %s\n", + misc_states[i].id); + mutex_unlock(&vf_info->pfvf_mig_lock); + return -EINVAL; + } + } + + mutex_unlock(&vf_info->pfvf_mig_lock); + adf_mstate_sect_update(mstate_mgr, &sub_sects_mgr, subsec); + + return 0; +} + +static int adf_gen4_vfmig_save_generic(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + struct adf_mstate_mgr sub_sects_mgr; + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct mig_user_sla src_slas[RL_RP_CNT_PER_LEAF_MAX] = { }; + u32 src_sla_cnt; + struct { + char *id; + struct adf_mstate_vreginfo info; + } gen_states[] = { + {ADF_MSTATE_IOV_INIT_IDS, + {&vf_info->init, sizeof(vf_info->init)}}, + {ADF_MSTATE_COMPAT_VER_IDS, + {&vf_info->vf_compat_ver, sizeof(vf_info->vf_compat_ver)}}, + {ADF_MSTATE_SLA_IDS, {src_slas, 0}}, + }; + int i; + + subsec = adf_mstate_sect_add(mstate_mgr, ADF_MSTATE_GEN_IDS, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to add sec %s\n", + ADF_MSTATE_GEN_IDS); + return -EINVAL; + } + + adf_mstate_mgr_init_from_parent(&sub_sects_mgr, mstate_mgr); + for (i = 0; i < ARRAY_SIZE(gen_states); i++) { + if (gen_states[i].info.addr == src_slas) { + src_sla_cnt = adf_gen4_vfmig_get_slas(accel_dev, vf_nr, src_slas); + gen_states[i].info.size = src_sla_cnt * sizeof(struct mig_user_sla); + } + + l2_subsec = adf_mstate_sect_add_vreg(&sub_sects_mgr, + gen_states[i].id, + &gen_states[i].info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to add sec %s\n", + gen_states[i].id); + return -EINVAL; + } + } + adf_mstate_sect_update(mstate_mgr, &sub_sects_mgr, subsec); + + return 0; +} + +static int adf_gen4_vfmig_save_config(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_gen4_vfmig *vfmig = vf_info->mig_priv; + struct adf_mstate_mgr *mstate_mgr = vfmig->mstate_mgr; + struct adf_mstate_mgr sub_sects_mgr; + struct adf_mstate_sect_h *subsec, *l2_subsec; + struct { + char *id; + struct adf_mstate_vreginfo info; + } setups[] = { + {ADF_MSTATE_GEN_CAP_IDS, + {&hw_data->accel_capabilities_mask, sizeof(hw_data->accel_capabilities_mask)}}, + {ADF_MSTATE_GEN_SVCMAP_IDS, + {&hw_data->ring_to_svc_map, sizeof(hw_data->ring_to_svc_map)}}, + {ADF_MSTATE_GEN_EXTDC_IDS, + {&hw_data->extended_dc_capabilities, sizeof(hw_data->extended_dc_capabilities)}}, + }; + int i; + + subsec = adf_mstate_sect_add(mstate_mgr, ADF_MSTATE_CONFIG_IDS, NULL, NULL); + if (!subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to add sec %s\n", + ADF_MSTATE_CONFIG_IDS); + return -EINVAL; + } + + adf_mstate_mgr_init_from_parent(&sub_sects_mgr, mstate_mgr); + for (i = 0; i < ARRAY_SIZE(setups); i++) { + l2_subsec = adf_mstate_sect_add_vreg(&sub_sects_mgr, setups[i].id, + &setups[i].info); + if (!l2_subsec) { + dev_err(&GET_DEV(accel_dev), "Failed to add sec %s\n", + setups[i].id); + return -EINVAL; + } + } + adf_mstate_sect_update(mstate_mgr, &sub_sects_mgr, subsec); + + return 0; +} + +static int adf_gen4_vfmig_save_state(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vfmig; + u32 vf_nr = mdev->vf_id; + int ret; + + vf_info = &accel_dev->pf.vf_info[vf_nr]; + vfmig = vf_info->mig_priv; + + ret = adf_gen4_vfmig_save_setup(mdev); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to save setup for vf_nr %d\n", vf_nr); + return ret; + } + + adf_mstate_mgr_init(vfmig->mstate_mgr, mdev->state + mdev->setup_size, + mdev->state_size - mdev->setup_size); + if (!adf_mstate_preamble_add(vfmig->mstate_mgr)) + return -EINVAL; + + ret = adf_gen4_vfmig_save_generic(accel_dev, vf_nr); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to save generic state for vf_nr %d\n", vf_nr); + return ret; + } + + ret = adf_gen4_vfmig_save_misc(accel_dev, vf_nr); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to save misc bar state for vf_nr %d\n", vf_nr); + return ret; + } + + ret = adf_gen4_vfmig_save_etr(accel_dev, vf_nr); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to save etr bar state for vf_nr %d\n", vf_nr); + return ret; + } + + adf_mstate_preamble_update(vfmig->mstate_mgr); + + return 0; +} + +static int adf_gen4_vfmig_load_state(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vfmig; + u32 vf_nr = mdev->vf_id; + int ret; + + vf_info = &accel_dev->pf.vf_info[vf_nr]; + vfmig = vf_info->mig_priv; + + ret = adf_gen4_vfmig_load_setup(mdev, mdev->state_size); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Failed to load setup for vf_nr %d\n", + vf_nr); + return ret; + } + + ret = adf_mstate_mgr_init_from_remote(vfmig->mstate_mgr, + mdev->state + mdev->remote_setup_size, + mdev->state_size - mdev->remote_setup_size, + NULL, NULL); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Invalid state for vf_nr %d\n", + vf_nr); + return ret; + } + + ret = adf_gen4_vfmig_load_generic(accel_dev, vf_nr); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to load general state for vf_nr %d\n", vf_nr); + return ret; + } + + ret = adf_gen4_vfmig_load_misc(accel_dev, vf_nr); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to load misc bar state for vf_nr %d\n", vf_nr); + return ret; + } + + ret = adf_gen4_vfmig_load_etr(accel_dev, vf_nr); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to load etr bar state for vf_nr %d\n", vf_nr); + return ret; + } + + return 0; +} + +static int adf_gen4_vfmig_save_setup(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vfmig; + u32 vf_nr = mdev->vf_id; + int ret; + + vf_info = &accel_dev->pf.vf_info[vf_nr]; + vfmig = vf_info->mig_priv; + + if (mdev->setup_size) + return 0; + + adf_mstate_mgr_init(vfmig->mstate_mgr, mdev->state, mdev->state_size); + if (!adf_mstate_preamble_add(vfmig->mstate_mgr)) + return -EINVAL; + + ret = adf_gen4_vfmig_save_config(accel_dev, mdev->vf_id); + if (ret) + return ret; + + adf_mstate_preamble_update(vfmig->mstate_mgr); + mdev->setup_size = adf_mstate_state_size(vfmig->mstate_mgr); + + return 0; +} + +static int adf_gen4_vfmig_load_setup(struct qat_mig_dev *mdev, int len) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + struct adf_accel_vf_info *vf_info; + struct adf_gen4_vfmig *vfmig; + u32 vf_nr = mdev->vf_id; + u32 setup_size; + int ret; + + vf_info = &accel_dev->pf.vf_info[vf_nr]; + vfmig = vf_info->mig_priv; + + if (mdev->remote_setup_size) + return 0; + + if (len < sizeof(struct adf_mstate_preh)) + return -EAGAIN; + + adf_mstate_mgr_init(vfmig->mstate_mgr, mdev->state, mdev->state_size); + setup_size = adf_mstate_state_size_from_remote(vfmig->mstate_mgr); + if (setup_size > mdev->state_size) + return -EINVAL; + + if (len < setup_size) + return -EAGAIN; + + ret = adf_mstate_mgr_init_from_remote(vfmig->mstate_mgr, mdev->state, + setup_size, NULL, NULL); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Invalid setup for vf_nr %d\n", + vf_nr); + return ret; + } + + mdev->remote_setup_size = setup_size; + + ret = adf_gen4_vfmig_load_config(accel_dev, vf_nr); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to load config for vf_nr %d\n", vf_nr); + return ret; + } + + return 0; +} + +void adf_gen4_init_vf_mig_ops(struct qat_migdev_ops *vfmig_ops) +{ + vfmig_ops->init = adf_gen4_vfmig_init_device; + vfmig_ops->cleanup = adf_gen4_vfmig_cleanup_device; + vfmig_ops->reset = adf_gen4_vfmig_reset_device; + vfmig_ops->open = adf_gen4_vfmig_open_device; + vfmig_ops->close = adf_gen4_vfmig_close_device; + vfmig_ops->suspend = adf_gen4_vfmig_suspend_device; + vfmig_ops->resume = adf_gen4_vfmig_resume_device; + vfmig_ops->save_state = adf_gen4_vfmig_save_state; + vfmig_ops->load_state = adf_gen4_vfmig_load_state; + vfmig_ops->load_setup = adf_gen4_vfmig_load_setup; + vfmig_ops->save_setup = adf_gen4_vfmig_save_setup; +} +EXPORT_SYMBOL_GPL(adf_gen4_init_vf_mig_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_vf_mig.h b/drivers/crypto/intel/qat/qat_common/adf_gen4_vf_mig.h new file mode 100644 index 000000000000..72216d078ee1 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_vf_mig.h @@ -0,0 +1,10 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2024 Intel Corporation */ +#ifndef ADF_GEN4_VF_MIG_H_ +#define ADF_GEN4_VF_MIG_H_ + +#include "adf_accel_devices.h" + +void adf_gen4_init_vf_mig_ops(struct qat_migdev_ops *vfmig_ops); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_pm.h b/drivers/crypto/intel/qat/qat_common/adf_gen6_pm.h new file mode 100644 index 000000000000..4c0d576e8c21 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_pm.h @@ -0,0 +1,52 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ADF_GEN6_PM_H +#define ADF_GEN6_PM_H + +#include <linux/bits.h> +#include <linux/time.h> + +struct adf_accel_dev; + +/* Power management */ +#define ADF_GEN6_PM_POLL_DELAY_US 20 +#define ADF_GEN6_PM_POLL_TIMEOUT_US USEC_PER_SEC +#define ADF_GEN6_PM_STATUS 0x50A00C +#define ADF_GEN6_PM_INTERRUPT 0x50A028 + +/* Power management source in ERRSOU2 and ERRMSK2 */ +#define ADF_GEN6_PM_SOU BIT(18) + +/* cpm_pm_interrupt bitfields */ +#define ADF_GEN6_PM_DRV_ACTIVE BIT(20) + +#define ADF_GEN6_PM_DEFAULT_IDLE_FILTER 0x6 + +/* cpm_pm_status bitfields */ +#define ADF_GEN6_PM_INIT_STATE BIT(21) +#define ADF_GEN6_PM_CPM_PM_STATE_MASK GENMASK(22, 20) + +/* fusectl0 bitfields */ +#define ADF_GEN6_PM_ENABLE_PM_MASK BIT(21) +#define ADF_GEN6_PM_ENABLE_PM_IDLE_MASK BIT(22) +#define ADF_GEN6_PM_ENABLE_DEEP_PM_IDLE_MASK BIT(23) + +/* cpm_pm_fw_init bitfields */ +#define ADF_GEN6_PM_IDLE_FILTER_MASK GENMASK(5, 3) +#define ADF_GEN6_PM_IDLE_ENABLE_MASK BIT(2) + +/* ssm_pm_enable bitfield */ +#define ADF_GEN6_PM_SSM_PM_ENABLE_MASK BIT(0) + +/* ssm_pm_domain_status bitfield */ +#define ADF_GEN6_PM_DOMAIN_POWERED_UP_MASK BIT(0) + +#ifdef CONFIG_DEBUG_FS +void adf_gen6_init_dev_pm_data(struct adf_accel_dev *accel_dev); +#else +static inline void adf_gen6_init_dev_pm_data(struct adf_accel_dev *accel_dev) +{ +} +#endif /* CONFIG_DEBUG_FS */ + +#endif /* ADF_GEN6_PM_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_pm_dbgfs.c b/drivers/crypto/intel/qat/qat_common/adf_gen6_pm_dbgfs.c new file mode 100644 index 000000000000..603aefba0fdb --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_pm_dbgfs.c @@ -0,0 +1,124 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2025 Intel Corporation */ +#include <linux/dma-mapping.h> +#include <linux/export.h> +#include <linux/string_helpers.h> + +#include "adf_admin.h" +#include "adf_common_drv.h" +#include "adf_gen6_pm.h" +#include "adf_pm_dbgfs_utils.h" +#include "icp_qat_fw_init_admin.h" + +#define PM_INFO_REGSET_ENTRY(_reg_, _field_) \ + PM_INFO_REGSET_ENTRY_MASK(_reg_, _field_, ADF_GEN6_PM_##_field_##_MASK) + +static struct pm_status_row pm_fuse_rows[] = { + PM_INFO_REGSET_ENTRY(fusectl0, ENABLE_PM), + PM_INFO_REGSET_ENTRY(fusectl0, ENABLE_PM_IDLE), + PM_INFO_REGSET_ENTRY(fusectl0, ENABLE_DEEP_PM_IDLE), +}; + +static struct pm_status_row pm_info_rows[] = { + PM_INFO_REGSET_ENTRY(pm.status, CPM_PM_STATE), + PM_INFO_REGSET_ENTRY(pm.fw_init, IDLE_ENABLE), + PM_INFO_REGSET_ENTRY(pm.fw_init, IDLE_FILTER), +}; + +static struct pm_status_row pm_ssm_rows[] = { + PM_INFO_REGSET_ENTRY(ssm.pm_enable, SSM_PM_ENABLE), + PM_INFO_REGSET_ENTRY(ssm.pm_domain_status, DOMAIN_POWERED_UP), +}; + +static struct pm_status_row pm_csrs_rows[] = { + PM_INFO_REGSET_ENTRY32(pm.fw_init, CPM_PM_FW_INIT), + PM_INFO_REGSET_ENTRY32(pm.status, CPM_PM_STATUS), +}; + +static_assert(sizeof(struct icp_qat_fw_init_admin_pm_info) < PAGE_SIZE); + +static ssize_t adf_gen6_print_pm_status(struct adf_accel_dev *accel_dev, + char __user *buf, size_t count, + loff_t *pos) +{ + void __iomem *pmisc = adf_get_pmisc_base(accel_dev); + struct icp_qat_fw_init_admin_pm_info *pm_info; + dma_addr_t p_state_addr; + u32 *pm_info_regs; + size_t len = 0; + char *pm_kv; + u32 val; + int ret; + + pm_info = kzalloc(PAGE_SIZE, GFP_KERNEL); + if (!pm_info) + return -ENOMEM; + + pm_kv = kzalloc(PAGE_SIZE, GFP_KERNEL); + if (!pm_kv) { + kfree(pm_info); + return -ENOMEM; + } + + p_state_addr = dma_map_single(&GET_DEV(accel_dev), pm_info, PAGE_SIZE, + DMA_FROM_DEVICE); + ret = dma_mapping_error(&GET_DEV(accel_dev), p_state_addr); + if (ret) + goto out_free; + + /* Query power management information from QAT FW */ + ret = adf_get_pm_info(accel_dev, p_state_addr, PAGE_SIZE); + dma_unmap_single(&GET_DEV(accel_dev), p_state_addr, PAGE_SIZE, + DMA_FROM_DEVICE); + if (ret) + goto out_free; + + pm_info_regs = (u32 *)pm_info; + + /* Fuse control register */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "----------- PM Fuse info ---------\n"); + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_fuse_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_fuse_rows)); + + /* Power management */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "----------- PM Info --------------\n"); + + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_info_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_info_rows)); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "pm_mode: ACTIVE\n"); + + /* Shared Slice Module */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "----------- SSM_PM Info ----------\n"); + len += adf_pm_scnprint_table_lower_keys(&pm_kv[len], pm_ssm_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_ssm_rows)); + + /* Control status register content */ + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, + "----------- HW PM CSRs -----------\n"); + len += adf_pm_scnprint_table_upper_keys(&pm_kv[len], pm_csrs_rows, + pm_info_regs, PAGE_SIZE - len, + ARRAY_SIZE(pm_csrs_rows)); + + val = ADF_CSR_RD(pmisc, ADF_GEN6_PM_INTERRUPT); + len += scnprintf(&pm_kv[len], PAGE_SIZE - len, "CPM_PM_INTERRUPT: %#x\n", val); + ret = simple_read_from_buffer(buf, count, pos, pm_kv, len); + +out_free: + kfree(pm_info); + kfree(pm_kv); + + return ret; +} + +void adf_gen6_init_dev_pm_data(struct adf_accel_dev *accel_dev) +{ + accel_dev->power_management.print_pm_status = adf_gen6_print_pm_status; + accel_dev->power_management.present = true; +} +EXPORT_SYMBOL_GPL(adf_gen6_init_dev_pm_data); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_ras.c b/drivers/crypto/intel/qat/qat_common/adf_gen6_ras.c new file mode 100644 index 000000000000..967253082a98 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_ras.c @@ -0,0 +1,818 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2025 Intel Corporation */ +#include <linux/bitfield.h> +#include <linux/types.h> + +#include "adf_common_drv.h" +#include "adf_gen6_ras.h" +#include "adf_sysfs_ras_counters.h" + +static void enable_errsou_reporting(void __iomem *csr) +{ + /* Enable correctable error reporting in ERRSOU0 */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK0, 0); + + /* Enable uncorrectable error reporting in ERRSOU1 */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK1, 0); + + /* + * Enable uncorrectable error reporting in ERRSOU2 + * but disable PM interrupt by default + */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK2, ADF_GEN6_ERRSOU2_PM_INT_BIT); + + /* Enable uncorrectable error reporting in ERRSOU3 */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK3, 0); +} + +static void enable_ae_error_reporting(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 ae_mask = GET_HW_DATA(accel_dev)->ae_mask; + + /* Enable acceleration engine correctable error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_HIAECORERRLOGENABLE_CPP0, ae_mask); + + /* Enable acceleration engine uncorrectable error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_HIAEUNCERRLOGENABLE_CPP0, ae_mask); +} + +static void enable_cpp_error_reporting(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + /* Enable HI CPP agents command parity error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_HICPPAGENTCMDPARERRLOGENABLE, + ADF_6XXX_HICPPAGENTCMDPARERRLOG_MASK); + + ADF_CSR_WR(csr, ADF_GEN6_CPP_CFC_ERR_CTRL, ADF_GEN6_CPP_CFC_ERR_CTRL_MASK); +} + +static void enable_ti_ri_error_reporting(void __iomem *csr) +{ + u32 reg, mask; + + /* Enable RI memory error reporting */ + mask = ADF_GEN6_RIMEM_PARERR_FATAL_MASK | ADF_GEN6_RIMEM_PARERR_CERR_MASK; + ADF_CSR_WR(csr, ADF_GEN6_RI_MEM_PAR_ERR_EN0, mask); + + /* Enable IOSF primary command parity error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_RIMISCCTL, ADF_GEN6_RIMISCSTS_BIT); + + /* Enable TI internal memory parity error reporting */ + reg = ADF_CSR_RD(csr, ADF_GEN6_TI_CI_PAR_ERR_MASK); + reg &= ~ADF_GEN6_TI_CI_PAR_STS_MASK; + ADF_CSR_WR(csr, ADF_GEN6_TI_CI_PAR_ERR_MASK, reg); + + reg = ADF_CSR_RD(csr, ADF_GEN6_TI_PULL0FUB_PAR_ERR_MASK); + reg &= ~ADF_GEN6_TI_PULL0FUB_PAR_STS_MASK; + ADF_CSR_WR(csr, ADF_GEN6_TI_PULL0FUB_PAR_ERR_MASK, reg); + + reg = ADF_CSR_RD(csr, ADF_GEN6_TI_PUSHFUB_PAR_ERR_MASK); + reg &= ~ADF_GEN6_TI_PUSHFUB_PAR_STS_MASK; + ADF_CSR_WR(csr, ADF_GEN6_TI_PUSHFUB_PAR_ERR_MASK, reg); + + reg = ADF_CSR_RD(csr, ADF_GEN6_TI_CD_PAR_ERR_MASK); + reg &= ~ADF_GEN6_TI_CD_PAR_STS_MASK; + ADF_CSR_WR(csr, ADF_GEN6_TI_CD_PAR_ERR_MASK, reg); + + reg = ADF_CSR_RD(csr, ADF_GEN6_TI_TRNSB_PAR_ERR_MASK); + reg &= ~ADF_GEN6_TI_TRNSB_PAR_STS_MASK; + ADF_CSR_WR(csr, ADF_GEN6_TI_TRNSB_PAR_ERR_MASK, reg); + + /* Enable error handling in RI, TI CPP interface control registers */ + ADF_CSR_WR(csr, ADF_GEN6_RICPPINTCTL, ADF_GEN6_RICPPINTCTL_MASK); + ADF_CSR_WR(csr, ADF_GEN6_TICPPINTCTL, ADF_GEN6_TICPPINTCTL_MASK); + + /* + * Enable error detection and reporting in TIMISCSTS + * with bits 1, 2 and 30 value preserved + */ + reg = ADF_CSR_RD(csr, ADF_GEN6_TIMISCCTL); + reg &= ADF_GEN6_TIMSCCTL_RELAY_MASK; + reg |= ADF_GEN6_TIMISCCTL_BIT; + ADF_CSR_WR(csr, ADF_GEN6_TIMISCCTL, reg); +} + +static void enable_ssm_error_reporting(struct adf_accel_dev *accel_dev, + void __iomem *csr) +{ + /* Enable SSM interrupts */ + ADF_CSR_WR(csr, ADF_GEN6_INTMASKSSM, 0); +} + +static void adf_gen6_enable_ras(struct adf_accel_dev *accel_dev) +{ + void __iomem *csr = adf_get_pmisc_base(accel_dev); + + enable_errsou_reporting(csr); + enable_ae_error_reporting(accel_dev, csr); + enable_cpp_error_reporting(accel_dev, csr); + enable_ti_ri_error_reporting(csr); + enable_ssm_error_reporting(accel_dev, csr); +} + +static void disable_errsou_reporting(void __iomem *csr) +{ + u32 val; + + /* Disable correctable error reporting in ERRSOU0 */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK0, ADF_GEN6_ERRSOU0_MASK); + + /* Disable uncorrectable error reporting in ERRSOU1 */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK1, ADF_GEN6_ERRMSK1_MASK); + + /* Disable uncorrectable error reporting in ERRSOU2 */ + val = ADF_CSR_RD(csr, ADF_GEN6_ERRMSK2); + val |= ADF_GEN6_ERRSOU2_DIS_MASK; + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK2, val); + + /* Disable uncorrectable error reporting in ERRSOU3 */ + ADF_CSR_WR(csr, ADF_GEN6_ERRMSK3, ADF_GEN6_ERRSOU3_DIS_MASK); +} + +static void disable_ae_error_reporting(void __iomem *csr) +{ + /* Disable acceleration engine correctable error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_HIAECORERRLOGENABLE_CPP0, 0); + + /* Disable acceleration engine uncorrectable error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_HIAEUNCERRLOGENABLE_CPP0, 0); +} + +static void disable_cpp_error_reporting(void __iomem *csr) +{ + /* Disable HI CPP agents command parity error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_HICPPAGENTCMDPARERRLOGENABLE, 0); + + ADF_CSR_WR(csr, ADF_GEN6_CPP_CFC_ERR_CTRL, ADF_GEN6_CPP_CFC_ERR_CTRL_DIS_MASK); +} + +static void disable_ti_ri_error_reporting(void __iomem *csr) +{ + u32 reg; + + /* Disable RI memory error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_RI_MEM_PAR_ERR_EN0, 0); + + /* Disable IOSF primary command parity error reporting */ + reg = ADF_CSR_RD(csr, ADF_GEN6_RIMISCCTL); + reg &= ~ADF_GEN6_RIMISCSTS_BIT; + ADF_CSR_WR(csr, ADF_GEN6_RIMISCCTL, reg); + + /* Disable TI internal memory parity error reporting */ + ADF_CSR_WR(csr, ADF_GEN6_TI_CI_PAR_ERR_MASK, ADF_GEN6_TI_CI_PAR_STS_MASK); + ADF_CSR_WR(csr, ADF_GEN6_TI_PULL0FUB_PAR_ERR_MASK, ADF_GEN6_TI_PULL0FUB_PAR_STS_MASK); + ADF_CSR_WR(csr, ADF_GEN6_TI_PUSHFUB_PAR_ERR_MASK, ADF_GEN6_TI_PUSHFUB_PAR_STS_MASK); + ADF_CSR_WR(csr, ADF_GEN6_TI_CD_PAR_ERR_MASK, ADF_GEN6_TI_CD_PAR_STS_MASK); + ADF_CSR_WR(csr, ADF_GEN6_TI_TRNSB_PAR_ERR_MASK, ADF_GEN6_TI_TRNSB_PAR_STS_MASK); + + /* Disable error handling in RI, TI CPP interface control registers */ + reg = ADF_CSR_RD(csr, ADF_GEN6_RICPPINTCTL); + reg &= ~ADF_GEN6_RICPPINTCTL_MASK; + ADF_CSR_WR(csr, ADF_GEN6_RICPPINTCTL, reg); + + reg = ADF_CSR_RD(csr, ADF_GEN6_TICPPINTCTL); + reg &= ~ADF_GEN6_TICPPINTCTL_MASK; + ADF_CSR_WR(csr, ADF_GEN6_TICPPINTCTL, reg); + + /* + * Disable error detection and reporting in TIMISCSTS + * with bits 1, 2 and 30 value preserved + */ + reg = ADF_CSR_RD(csr, ADF_GEN6_TIMISCCTL); + reg &= ADF_GEN6_TIMSCCTL_RELAY_MASK; + ADF_CSR_WR(csr, ADF_GEN6_TIMISCCTL, reg); +} + +static void disable_ssm_error_reporting(void __iomem *csr) +{ + /* Disable SSM interrupts */ + ADF_CSR_WR(csr, ADF_GEN6_INTMASKSSM, ADF_GEN6_INTMASKSSM_MASK); +} + +static void adf_gen6_disable_ras(struct adf_accel_dev *accel_dev) +{ + void __iomem *csr = adf_get_pmisc_base(accel_dev); + + disable_errsou_reporting(csr); + disable_ae_error_reporting(csr); + disable_cpp_error_reporting(csr); + disable_ti_ri_error_reporting(csr); + disable_ssm_error_reporting(csr); +} + +static void adf_gen6_process_errsou0(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 ae, errsou; + + ae = ADF_CSR_RD(csr, ADF_GEN6_HIAECORERRLOG_CPP0); + ae &= GET_HW_DATA(accel_dev)->ae_mask; + + dev_warn(&GET_DEV(accel_dev), "Correctable error detected: %#x\n", ae); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + + /* Clear interrupt from ERRSOU0 */ + ADF_CSR_WR(csr, ADF_GEN6_HIAECORERRLOG_CPP0, ae); + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU0); + if (errsou & ADF_GEN6_ERRSOU0_MASK) + dev_warn(&GET_DEV(accel_dev), "errsou0 still set: %#x\n", errsou); +} + +static void adf_handle_cpp_ae_unc(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 ae; + + if (!(errsou & ADF_GEN6_ERRSOU1_CPP0_MEUNC_BIT)) + return; + + ae = ADF_CSR_RD(csr, ADF_GEN6_HIAEUNCERRLOG_CPP0); + ae &= GET_HW_DATA(accel_dev)->ae_mask; + if (ae) { + dev_err(&GET_DEV(accel_dev), "Uncorrectable error detected: %#x\n", ae); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + ADF_CSR_WR(csr, ADF_GEN6_HIAEUNCERRLOG_CPP0, ae); + } +} + +static void adf_handle_cpp_cmd_par_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 cmd_par_err; + + if (!(errsou & ADF_GEN6_ERRSOU1_CPP_CMDPARERR_BIT)) + return; + + cmd_par_err = ADF_CSR_RD(csr, ADF_GEN6_HICPPAGENTCMDPARERRLOG); + cmd_par_err &= ADF_6XXX_HICPPAGENTCMDPARERRLOG_MASK; + if (cmd_par_err) { + dev_err(&GET_DEV(accel_dev), "HI CPP agent command parity error: %#x\n", + cmd_par_err); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_HICPPAGENTCMDPARERRLOG, cmd_par_err); + } +} + +static void adf_handle_ri_mem_par_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 rimem_parerr_sts; + + if (!(errsou & ADF_GEN6_ERRSOU1_RIMEM_PARERR_STS_BIT)) + return; + + rimem_parerr_sts = ADF_CSR_RD(csr, ADF_GEN6_RIMEM_PARERR_STS); + rimem_parerr_sts &= ADF_GEN6_RIMEM_PARERR_CERR_MASK | + ADF_GEN6_RIMEM_PARERR_FATAL_MASK; + if (rimem_parerr_sts & ADF_GEN6_RIMEM_PARERR_CERR_MASK) { + dev_err(&GET_DEV(accel_dev), "RI memory parity correctable error: %#x\n", + rimem_parerr_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + } + + if (rimem_parerr_sts & ADF_GEN6_RIMEM_PARERR_FATAL_MASK) { + dev_err(&GET_DEV(accel_dev), "RI memory parity fatal error: %#x\n", + rimem_parerr_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + } + + ADF_CSR_WR(csr, ADF_GEN6_RIMEM_PARERR_STS, rimem_parerr_sts); +} + +static void adf_handle_ti_ci_par_sts(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 ti_ci_par_sts; + + ti_ci_par_sts = ADF_CSR_RD(csr, ADF_GEN6_TI_CI_PAR_STS); + ti_ci_par_sts &= ADF_GEN6_TI_CI_PAR_STS_MASK; + if (ti_ci_par_sts) { + dev_err(&GET_DEV(accel_dev), "TI memory parity error: %#x\n", ti_ci_par_sts); + ADF_CSR_WR(csr, ADF_GEN6_TI_CI_PAR_STS, ti_ci_par_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } +} + +static void adf_handle_ti_pullfub_par_sts(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 ti_pullfub_par_sts; + + ti_pullfub_par_sts = ADF_CSR_RD(csr, ADF_GEN6_TI_PULL0FUB_PAR_STS); + ti_pullfub_par_sts &= ADF_GEN6_TI_PULL0FUB_PAR_STS_MASK; + if (ti_pullfub_par_sts) { + dev_err(&GET_DEV(accel_dev), "TI pull parity error: %#x\n", ti_pullfub_par_sts); + ADF_CSR_WR(csr, ADF_GEN6_TI_PULL0FUB_PAR_STS, ti_pullfub_par_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } +} + +static void adf_handle_ti_pushfub_par_sts(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 ti_pushfub_par_sts; + + ti_pushfub_par_sts = ADF_CSR_RD(csr, ADF_GEN6_TI_PUSHFUB_PAR_STS); + ti_pushfub_par_sts &= ADF_GEN6_TI_PUSHFUB_PAR_STS_MASK; + if (ti_pushfub_par_sts) { + dev_err(&GET_DEV(accel_dev), "TI push parity error: %#x\n", ti_pushfub_par_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + ADF_CSR_WR(csr, ADF_GEN6_TI_PUSHFUB_PAR_STS, ti_pushfub_par_sts); + } +} + +static void adf_handle_ti_cd_par_sts(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 ti_cd_par_sts; + + ti_cd_par_sts = ADF_CSR_RD(csr, ADF_GEN6_TI_CD_PAR_STS); + ti_cd_par_sts &= ADF_GEN6_TI_CD_PAR_STS_MASK; + if (ti_cd_par_sts) { + dev_err(&GET_DEV(accel_dev), "TI CD parity error: %#x\n", ti_cd_par_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + ADF_CSR_WR(csr, ADF_GEN6_TI_CD_PAR_STS, ti_cd_par_sts); + } +} + +static void adf_handle_ti_trnsb_par_sts(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 ti_trnsb_par_sts; + + ti_trnsb_par_sts = ADF_CSR_RD(csr, ADF_GEN6_TI_TRNSB_PAR_STS); + ti_trnsb_par_sts &= ADF_GEN6_TI_TRNSB_PAR_STS_MASK; + if (ti_trnsb_par_sts) { + dev_err(&GET_DEV(accel_dev), "TI TRNSB parity error: %#x\n", ti_trnsb_par_sts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + ADF_CSR_WR(csr, ADF_GEN6_TI_TRNSB_PAR_STS, ti_trnsb_par_sts); + } +} + +static void adf_handle_iosfp_cmd_parerr(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 rimiscsts; + + rimiscsts = ADF_CSR_RD(csr, ADF_GEN6_RIMISCSTS); + rimiscsts &= ADF_GEN6_RIMISCSTS_BIT; + if (rimiscsts) { + dev_err(&GET_DEV(accel_dev), "Command parity error detected on IOSFP: %#x\n", + rimiscsts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_RIMISCSTS, rimiscsts); + } +} + +static void adf_handle_ti_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU1_TIMEM_PARERR_STS_BIT)) + return; + + adf_handle_ti_ci_par_sts(accel_dev, csr); + adf_handle_ti_pullfub_par_sts(accel_dev, csr); + adf_handle_ti_pushfub_par_sts(accel_dev, csr); + adf_handle_ti_cd_par_sts(accel_dev, csr); + adf_handle_ti_trnsb_par_sts(accel_dev, csr); + adf_handle_iosfp_cmd_parerr(accel_dev, csr); +} + +static void adf_handle_sfi_cmd_parerr(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU1_SFICMD_PARERR_BIT)) + return; + + dev_err(&GET_DEV(accel_dev), + "Command parity error detected on streaming fabric interface\n"); + + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); +} + +static void adf_gen6_process_errsou1(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + adf_handle_cpp_ae_unc(accel_dev, csr, errsou); + adf_handle_cpp_cmd_par_err(accel_dev, csr, errsou); + adf_handle_ri_mem_par_err(accel_dev, csr, errsou); + adf_handle_ti_err(accel_dev, csr, errsou); + adf_handle_sfi_cmd_parerr(accel_dev, csr, errsou); + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU1); + if (errsou & ADF_GEN6_ERRSOU1_MASK) + dev_warn(&GET_DEV(accel_dev), "errsou1 still set: %#x\n", errsou); +} + +static void adf_handle_cerrssmsh(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 reg; + + reg = ADF_CSR_RD(csr, ADF_GEN6_CERRSSMSH); + reg &= ADF_GEN6_CERRSSMSH_ERROR_BIT; + if (reg) { + dev_warn(&GET_DEV(accel_dev), + "Correctable error on ssm shared memory: %#x\n", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_CORR); + ADF_CSR_WR(csr, ADF_GEN6_CERRSSMSH, reg); + } +} + +static void adf_handle_uerrssmsh(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN6_IAINTSTATSSM_SH_ERR_BIT)) + return; + + reg = ADF_CSR_RD(csr, ADF_GEN6_UERRSSMSH); + reg &= ADF_GEN6_UERRSSMSH_MASK; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "Fatal error on ssm shared memory: %#x\n", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_UERRSSMSH, reg); + } +} + +static void adf_handle_pperr_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN6_IAINTSTATSSM_PPERR_BIT)) + return; + + reg = ADF_CSR_RD(csr, ADF_GEN6_PPERR); + reg &= ADF_GEN6_PPERR_MASK; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "Fatal push or pull data error: %#x\n", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_PPERR, reg); + } +} + +static void adf_handle_scmpar_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN6_IAINTSTATSSM_SCMPAR_ERR_BIT)) + return; + + reg = ADF_CSR_RD(csr, ADF_GEN6_SSM_FERR_STATUS); + reg &= ADF_GEN6_SCM_PAR_ERR_MASK; + if (reg) { + dev_err(&GET_DEV(accel_dev), "Fatal error on SCM: %#x\n", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_SSM_FERR_STATUS, reg); + } +} + +static void adf_handle_cpppar_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN6_IAINTSTATSSM_CPPPAR_ERR_BIT)) + return; + + reg = ADF_CSR_RD(csr, ADF_GEN6_SSM_FERR_STATUS); + reg &= ADF_GEN6_CPP_PAR_ERR_MASK; + if (reg) { + dev_err(&GET_DEV(accel_dev), "Fatal error on CPP: %#x\n", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_SSM_FERR_STATUS, reg); + } +} + +static void adf_handle_rfpar_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN6_IAINTSTATSSM_RFPAR_ERR_BIT)) + return; + + reg = ADF_CSR_RD(csr, ADF_GEN6_SSM_FERR_STATUS); + reg &= ADF_GEN6_RF_PAR_ERR_MASK; + if (reg) { + dev_err(&GET_DEV(accel_dev), "Fatal error on RF Parity: %#x\n", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_SSM_FERR_STATUS, reg); + } +} + +static void adf_handle_unexp_cpl_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 iastatssm) +{ + u32 reg; + + if (!(iastatssm & ADF_GEN6_IAINTSTATSSM_UNEXP_CPL_ERR_BIT)) + return; + + reg = ADF_CSR_RD(csr, ADF_GEN6_SSM_FERR_STATUS); + reg &= ADF_GEN6_UNEXP_CPL_ERR_MASK; + if (reg) { + dev_err(&GET_DEV(accel_dev), + "Fatal error for AXI unexpected tag/length: %#x\n", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_SSM_FERR_STATUS, reg); + } +} + +static void adf_handle_iaintstatssm(struct adf_accel_dev *accel_dev, void __iomem *csr) +{ + u32 iastatssm = ADF_CSR_RD(csr, ADF_GEN6_IAINTSTATSSM); + + iastatssm &= ADF_GEN6_IAINTSTATSSM_MASK; + if (!iastatssm) + return; + + adf_handle_uerrssmsh(accel_dev, csr, iastatssm); + adf_handle_pperr_err(accel_dev, csr, iastatssm); + adf_handle_scmpar_err(accel_dev, csr, iastatssm); + adf_handle_cpppar_err(accel_dev, csr, iastatssm); + adf_handle_rfpar_err(accel_dev, csr, iastatssm); + adf_handle_unexp_cpl_err(accel_dev, csr, iastatssm); + + ADF_CSR_WR(csr, ADF_GEN6_IAINTSTATSSM, iastatssm); +} + +static void adf_handle_ssm(struct adf_accel_dev *accel_dev, void __iomem *csr, u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU2_SSM_ERR_BIT)) + return; + + adf_handle_cerrssmsh(accel_dev, csr); + adf_handle_iaintstatssm(accel_dev, csr); +} + +static void adf_handle_cpp_cfc_err(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 reg; + + if (!(errsou & ADF_GEN6_ERRSOU2_CPP_CFC_ERR_STATUS_BIT)) + return; + + reg = ADF_CSR_RD(csr, ADF_GEN6_CPP_CFC_ERR_STATUS); + if (reg & ADF_GEN6_CPP_CFC_ERR_STATUS_DATAPAR_BIT) { + dev_err(&GET_DEV(accel_dev), "CPP_CFC_ERR: data parity: %#x", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + } + + if (reg & ADF_GEN6_CPP_CFC_ERR_STATUS_CMDPAR_BIT) { + dev_err(&GET_DEV(accel_dev), "CPP_CFC_ERR: command parity: %#x", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + } + + if (reg & ADF_GEN6_CPP_CFC_FATAL_ERR_BIT) { + dev_err(&GET_DEV(accel_dev), "CPP_CFC_ERR: errors: %#x", reg); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + } + + ADF_CSR_WR(csr, ADF_GEN6_CPP_CFC_ERR_STATUS_CLR, + ADF_GEN6_CPP_CFC_ERR_STATUS_CLR_MASK); +} + +static void adf_gen6_process_errsou2(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + adf_handle_ssm(accel_dev, csr, errsou); + adf_handle_cpp_cfc_err(accel_dev, csr, errsou); + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU2); + if (errsou & ADF_GEN6_ERRSOU2_MASK) + dev_warn(&GET_DEV(accel_dev), "errsou2 still set: %#x\n", errsou); +} + +static void adf_handle_timiscsts(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 timiscsts; + + if (!(errsou & ADF_GEN6_ERRSOU3_TIMISCSTS_BIT)) + return; + + timiscsts = ADF_CSR_RD(csr, ADF_GEN6_TIMISCSTS); + if (timiscsts) { + dev_err(&GET_DEV(accel_dev), "Fatal error in transmit interface: %#x\n", + timiscsts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + } +} + +static void adf_handle_ricppintsts(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 ricppintsts; + + if (!(errsou & ADF_GEN6_ERRSOU3_RICPPINTSTS_MASK)) + return; + + ricppintsts = ADF_CSR_RD(csr, ADF_GEN6_RICPPINTSTS); + ricppintsts &= ADF_GEN6_RICPPINTSTS_MASK; + if (ricppintsts) { + dev_err(&GET_DEV(accel_dev), "RI push pull error: %#x\n", ricppintsts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + ADF_CSR_WR(csr, ADF_GEN6_RICPPINTSTS, ricppintsts); + } +} + +static void adf_handle_ticppintsts(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 ticppintsts; + + if (!(errsou & ADF_GEN6_ERRSOU3_TICPPINTSTS_MASK)) + return; + + ticppintsts = ADF_CSR_RD(csr, ADF_GEN6_TICPPINTSTS); + ticppintsts &= ADF_GEN6_TICPPINTSTS_MASK; + if (ticppintsts) { + dev_err(&GET_DEV(accel_dev), "TI push pull error: %#x\n", ticppintsts); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); + ADF_CSR_WR(csr, ADF_GEN6_TICPPINTSTS, ticppintsts); + } +} + +static void adf_handle_atufaultstatus(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 max_rp_num = GET_HW_DATA(accel_dev)->num_banks; + u32 atufaultstatus; + u32 i; + + if (!(errsou & ADF_GEN6_ERRSOU3_ATUFAULTSTATUS_BIT)) + return; + + for (i = 0; i < max_rp_num; i++) { + atufaultstatus = ADF_CSR_RD(csr, ADF_GEN6_ATUFAULTSTATUS(i)); + + atufaultstatus &= ADF_GEN6_ATUFAULTSTATUS_BIT; + if (atufaultstatus) { + dev_err(&GET_DEV(accel_dev), "Ring pair (%u) ATU detected fault: %#x\n", i, + atufaultstatus); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + ADF_CSR_WR(csr, ADF_GEN6_ATUFAULTSTATUS(i), atufaultstatus); + } + } +} + +static void adf_handle_rlterror(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + u32 rlterror; + + if (!(errsou & ADF_GEN6_ERRSOU3_RLTERROR_BIT)) + return; + + rlterror = ADF_CSR_RD(csr, ADF_GEN6_RLT_ERRLOG); + rlterror &= ADF_GEN6_RLT_ERRLOG_MASK; + if (rlterror) { + dev_err(&GET_DEV(accel_dev), "Error in rate limiting block: %#x\n", rlterror); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); + ADF_CSR_WR(csr, ADF_GEN6_RLT_ERRLOG, rlterror); + } +} + +static void adf_handle_vflr(struct adf_accel_dev *accel_dev, void __iomem *csr, u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU3_VFLRNOTIFY_BIT)) + return; + + dev_err(&GET_DEV(accel_dev), "Uncorrectable error in VF\n"); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_UNCORR); +} + +static void adf_handle_tc_vc_map_error(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU3_TC_VC_MAP_ERROR_BIT)) + return; + + dev_err(&GET_DEV(accel_dev), "Violation of PCIe TC VC mapping\n"); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); +} + +static void adf_handle_pcie_devhalt(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU3_PCIE_DEVHALT_BIT)) + return; + + dev_err(&GET_DEV(accel_dev), + "DEVHALT due to an error in an incoming transaction\n"); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); +} + +static void adf_handle_pg_req_devhalt(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU3_PG_REQ_DEVHALT_BIT)) + return; + + dev_err(&GET_DEV(accel_dev), + "Error due to response failure in response to a page request\n"); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); +} + +static void adf_handle_xlt_cpl_devhalt(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU3_XLT_CPL_DEVHALT_BIT)) + return; + + dev_err(&GET_DEV(accel_dev), "Error status for a address translation request\n"); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); +} + +static void adf_handle_ti_int_err_devhalt(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + if (!(errsou & ADF_GEN6_ERRSOU3_TI_INT_ERR_DEVHALT_BIT)) + return; + + dev_err(&GET_DEV(accel_dev), "DEVHALT due to a TI internal memory error\n"); + ADF_RAS_ERR_CTR_INC(accel_dev->ras_errors, ADF_RAS_FATAL); +} + +static void adf_gen6_process_errsou3(struct adf_accel_dev *accel_dev, void __iomem *csr, + u32 errsou) +{ + adf_handle_timiscsts(accel_dev, csr, errsou); + adf_handle_ricppintsts(accel_dev, csr, errsou); + adf_handle_ticppintsts(accel_dev, csr, errsou); + adf_handle_atufaultstatus(accel_dev, csr, errsou); + adf_handle_rlterror(accel_dev, csr, errsou); + adf_handle_vflr(accel_dev, csr, errsou); + adf_handle_tc_vc_map_error(accel_dev, csr, errsou); + adf_handle_pcie_devhalt(accel_dev, csr, errsou); + adf_handle_pg_req_devhalt(accel_dev, csr, errsou); + adf_handle_xlt_cpl_devhalt(accel_dev, csr, errsou); + adf_handle_ti_int_err_devhalt(accel_dev, csr, errsou); + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU3); + if (errsou & ADF_GEN6_ERRSOU3_MASK) + dev_warn(&GET_DEV(accel_dev), "errsou3 still set: %#x\n", errsou); +} + +static void adf_gen6_is_reset_required(struct adf_accel_dev *accel_dev, void __iomem *csr, + bool *reset_required) +{ + u8 reset, dev_state; + u32 gensts; + + gensts = ADF_CSR_RD(csr, ADF_GEN6_GENSTS); + dev_state = FIELD_GET(ADF_GEN6_GENSTS_DEVICE_STATE_MASK, gensts); + reset = FIELD_GET(ADF_GEN6_GENSTS_RESET_TYPE_MASK, gensts); + if (dev_state == ADF_GEN6_GENSTS_DEVHALT && reset == ADF_GEN6_GENSTS_PFLR) { + *reset_required = true; + return; + } + + if (reset == ADF_GEN6_GENSTS_COLD_RESET) + dev_err(&GET_DEV(accel_dev), "Fatal error, cold reset required\n"); + + *reset_required = false; +} + +static bool adf_gen6_handle_interrupt(struct adf_accel_dev *accel_dev, bool *reset_required) +{ + void __iomem *csr = adf_get_pmisc_base(accel_dev); + bool handled = false; + u32 errsou; + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU0); + if (errsou & ADF_GEN6_ERRSOU0_MASK) { + adf_gen6_process_errsou0(accel_dev, csr); + handled = true; + } + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU1); + if (errsou & ADF_GEN6_ERRSOU1_MASK) { + adf_gen6_process_errsou1(accel_dev, csr, errsou); + handled = true; + } + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU2); + if (errsou & ADF_GEN6_ERRSOU2_MASK) { + adf_gen6_process_errsou2(accel_dev, csr, errsou); + handled = true; + } + + errsou = ADF_CSR_RD(csr, ADF_GEN6_ERRSOU3); + if (errsou & ADF_GEN6_ERRSOU3_MASK) { + adf_gen6_process_errsou3(accel_dev, csr, errsou); + handled = true; + } + + adf_gen6_is_reset_required(accel_dev, csr, reset_required); + + return handled; +} + +void adf_gen6_init_ras_ops(struct adf_ras_ops *ras_ops) +{ + ras_ops->enable_ras_errors = adf_gen6_enable_ras; + ras_ops->disable_ras_errors = adf_gen6_disable_ras; + ras_ops->handle_interrupt = adf_gen6_handle_interrupt; +} +EXPORT_SYMBOL_GPL(adf_gen6_init_ras_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_ras.h b/drivers/crypto/intel/qat/qat_common/adf_gen6_ras.h new file mode 100644 index 000000000000..66ced271d173 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_ras.h @@ -0,0 +1,504 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ADF_GEN6_RAS_H_ +#define ADF_GEN6_RAS_H_ + +#include <linux/bits.h> + +struct adf_ras_ops; + +/* Error source registers */ +#define ADF_GEN6_ERRSOU0 0x41A200 +#define ADF_GEN6_ERRSOU1 0x41A204 +#define ADF_GEN6_ERRSOU2 0x41A208 +#define ADF_GEN6_ERRSOU3 0x41A20C + +/* Error source mask registers */ +#define ADF_GEN6_ERRMSK0 0x41A210 +#define ADF_GEN6_ERRMSK1 0x41A214 +#define ADF_GEN6_ERRMSK2 0x41A218 +#define ADF_GEN6_ERRMSK3 0x41A21C + +/* ERRSOU0 Correctable error mask */ +#define ADF_GEN6_ERRSOU0_MASK BIT(0) + +#define ADF_GEN6_ERRSOU1_CPP0_MEUNC_BIT BIT(0) +#define ADF_GEN6_ERRSOU1_CPP_CMDPARERR_BIT BIT(1) +#define ADF_GEN6_ERRSOU1_RIMEM_PARERR_STS_BIT BIT(2) +#define ADF_GEN6_ERRSOU1_TIMEM_PARERR_STS_BIT BIT(3) +#define ADF_GEN6_ERRSOU1_SFICMD_PARERR_BIT BIT(4) + +#define ADF_GEN6_ERRSOU1_MASK ( \ + (ADF_GEN6_ERRSOU1_CPP0_MEUNC_BIT) | \ + (ADF_GEN6_ERRSOU1_CPP_CMDPARERR_BIT) | \ + (ADF_GEN6_ERRSOU1_RIMEM_PARERR_STS_BIT) | \ + (ADF_GEN6_ERRSOU1_TIMEM_PARERR_STS_BIT) | \ + (ADF_GEN6_ERRSOU1_SFICMD_PARERR_BIT)) + +#define ADF_GEN6_ERRMSK1_CPP0_MEUNC_BIT BIT(0) +#define ADF_GEN6_ERRMSK1_CPP_CMDPARERR_BIT BIT(1) +#define ADF_GEN6_ERRMSK1_RIMEM_PARERR_STS_BIT BIT(2) +#define ADF_GEN6_ERRMSK1_TIMEM_PARERR_STS_BIT BIT(3) +#define ADF_GEN6_ERRMSK1_IOSFCMD_PARERR_BIT BIT(4) + +#define ADF_GEN6_ERRMSK1_MASK ( \ + (ADF_GEN6_ERRMSK1_CPP0_MEUNC_BIT) | \ + (ADF_GEN6_ERRMSK1_CPP_CMDPARERR_BIT) | \ + (ADF_GEN6_ERRMSK1_RIMEM_PARERR_STS_BIT) | \ + (ADF_GEN6_ERRMSK1_TIMEM_PARERR_STS_BIT) | \ + (ADF_GEN6_ERRMSK1_IOSFCMD_PARERR_BIT)) + +/* HI AE Uncorrectable error log */ +#define ADF_GEN6_HIAEUNCERRLOG_CPP0 0x41A300 + +/* HI AE Uncorrectable error log enable */ +#define ADF_GEN6_HIAEUNCERRLOGENABLE_CPP0 0x41A320 + +/* HI AE Correctable error log */ +#define ADF_GEN6_HIAECORERRLOG_CPP0 0x41A308 + +/* HI AE Correctable error log enable */ +#define ADF_GEN6_HIAECORERRLOGENABLE_CPP0 0x41A318 + +/* HI CPP Agent Command parity error log */ +#define ADF_GEN6_HICPPAGENTCMDPARERRLOG 0x41A310 + +/* HI CPP Agent command parity error logging enable */ +#define ADF_GEN6_HICPPAGENTCMDPARERRLOGENABLE 0x41A314 + +#define ADF_6XXX_HICPPAGENTCMDPARERRLOG_MASK 0x1B + +/* RI Memory parity error status register */ +#define ADF_GEN6_RIMEM_PARERR_STS 0x41B128 + +/* RI Memory parity error reporting enable */ +#define ADF_GEN6_RI_MEM_PAR_ERR_EN0 0x41B12C + +/* + * RI Memory parity error mask + * BIT(4) - ri_tlq_phdr parity error + * BIT(5) - ri_tlq_pdata parity error + * BIT(6) - ri_tlq_nphdr parity error + * BIT(7) - ri_tlq_npdata parity error + * BIT(8) - ri_tlq_cplhdr parity error + * BIT(10) - BIT(13) - ri_tlq_cpldata[0:3] parity error + * BIT(19) - ri_cds_cmd_fifo parity error + * BIT(20) - ri_obc_ricpl_fifo parity error + * BIT(21) - ri_obc_tiricpl_fifo parity error + * BIT(22) - ri_obc_cppcpl_fifo parity error + * BIT(23) - ri_obc_pendcpl_fifo parity error + * BIT(24) - ri_cpp_cmd_fifo parity error + * BIT(25) - ri_cds_ticmd_fifo parity error + * BIT(26) - riti_cmd_fifo parity error + * BIT(27) - ri_int_msixtbl parity error + * BIT(28) - ri_int_imstbl parity error + * BIT(30) - ri_kpt_fuses parity error + */ +#define ADF_GEN6_RIMEM_PARERR_FATAL_MASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(4) | BIT(5) | BIT(6) | \ + BIT(7) | BIT(8) | BIT(18) | BIT(19) | BIT(20) | BIT(21) | \ + BIT(22) | BIT(23) | BIT(24) | BIT(25) | BIT(26) | BIT(27) | \ + BIT(28) | BIT(30)) + +#define ADF_GEN6_RIMEM_PARERR_CERR_MASK \ + (BIT(10) | BIT(11) | BIT(12) | BIT(13)) + +/* TI CI parity status */ +#define ADF_GEN6_TI_CI_PAR_STS 0x50060C + +/* TI CI parity reporting mask */ +#define ADF_GEN6_TI_CI_PAR_ERR_MASK 0x500608 + +/* + * TI CI parity status mask + * BIT(0) - CdCmdQ_sts patiry error status + * BIT(1) - CdDataQ_sts parity error status + * BIT(3) - CPP_SkidQ_sts parity error status + */ +#define ADF_GEN6_TI_CI_PAR_STS_MASK \ + (BIT(0) | BIT(1) | BIT(3)) + +/* TI PULLFUB parity status */ +#define ADF_GEN6_TI_PULL0FUB_PAR_STS 0x500618 + +/* TI PULLFUB parity error reporting mask */ +#define ADF_GEN6_TI_PULL0FUB_PAR_ERR_MASK 0x500614 + +/* + * TI PULLFUB parity status mask + * BIT(0) - TrnPullReqQ_sts parity status + * BIT(1) - TrnSharedDataQ_sts parity status + * BIT(2) - TrnPullReqDataQ_sts parity status + * BIT(4) - CPP_CiPullReqQ_sts parity status + * BIT(5) - CPP_TrnPullReqQ_sts parity status + * BIT(6) - CPP_PullidQ_sts parity status + * BIT(7) - CPP_WaitDataQ_sts parity status + * BIT(8) - CPP_CdDataQ_sts parity status + * BIT(9) - CPP_TrnDataQP0_sts parity status + * BIT(10) - BIT(11) - CPP_TrnDataQRF[00:01]_sts parity status + * BIT(12) - CPP_TrnDataQP1_sts parity status + * BIT(13) - BIT(14) - CPP_TrnDataQRF[10:11]_sts parity status + */ +#define ADF_GEN6_TI_PULL0FUB_PAR_STS_MASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(4) | BIT(5) | BIT(6) | BIT(7) | \ + BIT(8) | BIT(9) | BIT(10) | BIT(11) | BIT(12) | BIT(13) | BIT(14)) + +/* TI PUSHUB parity status */ +#define ADF_GEN6_TI_PUSHFUB_PAR_STS 0x500630 + +/* TI PUSHFUB parity error reporting mask */ +#define ADF_GEN6_TI_PUSHFUB_PAR_ERR_MASK 0x50062C + +/* + * TI PUSHUB parity status mask + * BIT(0) - SbPushReqQ_sts parity status + * BIT(1) - BIT(2) - SbPushDataQ[0:1]_sts parity status + * BIT(4) - CPP_CdPushReqQ_sts parity status + * BIT(5) - BIT(6) - CPP_CdPushDataQ[0:1]_sts parity status + * BIT(7) - CPP_SbPushReqQ_sts parity status + * BIT(8) - CPP_SbPushDataQP_sts parity status + * BIT(9) - BIT(10) - CPP_SbPushDataQRF[0:1]_sts parity status + */ +#define ADF_GEN6_TI_PUSHFUB_PAR_STS_MASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(4) | BIT(5) | \ + BIT(6) | BIT(7) | BIT(8) | BIT(9) | BIT(10)) + +/* TI CD parity status */ +#define ADF_GEN6_TI_CD_PAR_STS 0x50063C + +/* TI CD parity error mask */ +#define ADF_GEN6_TI_CD_PAR_ERR_MASK 0x500638 + +/* + * TI CD parity status mask + * BIT(0) - BIT(15) - CtxMdRam[0:15]_sts parity status + * BIT(16) - Leaf2ClusterRam_sts parity status + * BIT(17) - BIT(18) - Ring2LeafRam[0:1]_sts parity status + * BIT(19) - VirtualQ_sts parity status + * BIT(20) - DtRdQ_sts parity status + * BIT(21) - DtWrQ_sts parity status + * BIT(22) - RiCmdQ_sts parity status + * BIT(23) - BypassQ_sts parity status + * BIT(24) - DtRdQ_sc_sts parity status + * BIT(25) - DtWrQ_sc_sts parity status + */ +#define ADF_GEN6_TI_CD_PAR_STS_MASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4) | BIT(5) | BIT(6) | \ + BIT(7) | BIT(8) | BIT(9) | BIT(10) | BIT(11) | BIT(12) | BIT(13) | \ + BIT(14) | BIT(15) | BIT(16) | BIT(17) | BIT(18) | BIT(19) | BIT(20) | \ + BIT(21) | BIT(22) | BIT(23) | BIT(24) | BIT(25)) + +/* TI TRNSB parity status */ +#define ADF_GEN6_TI_TRNSB_PAR_STS 0x500648 + +/* TI TRNSB parity error reporting mask */ +#define ADF_GEN6_TI_TRNSB_PAR_ERR_MASK 0x500644 + +/* + * TI TRNSB parity status mask + * BIT(0) - TrnPHdrQP_sts parity status + * BIT(1) - TrnPHdrQRF_sts parity status + * BIT(2) - TrnPDataQP_sts parity status + * BIT(3) - BIT(6) - TrnPDataQRF[0:3]_sts parity status + * BIT(7) - TrnNpHdrQP_sts parity status + * BIT(8) - BIT(9) - TrnNpHdrQRF[0:1]_sts parity status + * BIT(10) - TrnCplHdrQ_sts parity status + * BIT(11) - TrnPutObsReqQ_sts parity status + * BIT(12) - TrnPushReqQ_sts parity status + * BIT(13) - SbSplitIdRam_sts parity status + * BIT(14) - SbReqCountQ_sts parity status + * BIT(15) - SbCplTrkRam_sts parity status + * BIT(16) - SbGetObsReqQ_sts parity status + * BIT(17) - SbEpochIdQ_sts parity status + * BIT(18) - SbAtCplHdrQ_sts parity status + * BIT(19) - SbAtCplDataQ_sts parity status + * BIT(20) - SbReqCountRam_sts parity status + * BIT(21) - SbAtCplHdrQ_sc_sts parity status + */ +#define ADF_GEN6_TI_TRNSB_PAR_STS_MASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4) | BIT(5) | BIT(6) | \ + BIT(7) | BIT(8) | BIT(9) | BIT(10) | BIT(11) | BIT(12) | \ + BIT(13) | BIT(14) | BIT(15) | BIT(16) | BIT(17) | BIT(18) | \ + BIT(19) | BIT(20) | BIT(21)) + +/* Status register to log misc error on RI */ +#define ADF_GEN6_RIMISCSTS 0x41B1B8 + +/* Status control register to log misc RI error */ +#define ADF_GEN6_RIMISCCTL 0x41B1BC + +/* + * ERRSOU2 bit mask + * BIT(0) - SSM Interrupt Mask + * BIT(1) - CFC on CPP. ORed of CFC Push error and Pull error + * BIT(2) - BIT(4) - CPP attention interrupts + * BIT(18) - PM interrupt + */ +#define ADF_GEN6_ERRSOU2_SSM_ERR_BIT BIT(0) +#define ADF_GEN6_ERRSOU2_CPP_CFC_ERR_STATUS_BIT BIT(1) +#define ADF_GEN6_ERRSOU2_CPP_CFC_ATT_INT_MASK \ + (BIT(2) | BIT(3) | BIT(4)) + +#define ADF_GEN6_ERRSOU2_PM_INT_BIT BIT(18) + +#define ADF_GEN6_ERRSOU2_MASK \ + (ADF_GEN6_ERRSOU2_SSM_ERR_BIT | \ + ADF_GEN6_ERRSOU2_CPP_CFC_ERR_STATUS_BIT) + +#define ADF_GEN6_ERRSOU2_DIS_MASK \ + (ADF_GEN6_ERRSOU2_SSM_ERR_BIT | \ + ADF_GEN6_ERRSOU2_CPP_CFC_ERR_STATUS_BIT | \ + ADF_GEN6_ERRSOU2_CPP_CFC_ATT_INT_MASK) + +#define ADF_GEN6_IAINTSTATSSM 0x28 + +/* IAINTSTATSSM error bit mask definitions */ +#define ADF_GEN6_IAINTSTATSSM_SH_ERR_BIT BIT(0) +#define ADF_GEN6_IAINTSTATSSM_PPERR_BIT BIT(2) +#define ADF_GEN6_IAINTSTATSSM_SCMPAR_ERR_BIT BIT(4) +#define ADF_GEN6_IAINTSTATSSM_CPPPAR_ERR_BIT BIT(5) +#define ADF_GEN6_IAINTSTATSSM_RFPAR_ERR_BIT BIT(6) +#define ADF_GEN6_IAINTSTATSSM_UNEXP_CPL_ERR_BIT BIT(7) + +#define ADF_GEN6_IAINTSTATSSM_MASK \ + (ADF_GEN6_IAINTSTATSSM_SH_ERR_BIT | \ + ADF_GEN6_IAINTSTATSSM_PPERR_BIT | \ + ADF_GEN6_IAINTSTATSSM_SCMPAR_ERR_BIT | \ + ADF_GEN6_IAINTSTATSSM_CPPPAR_ERR_BIT | \ + ADF_GEN6_IAINTSTATSSM_RFPAR_ERR_BIT | \ + ADF_GEN6_IAINTSTATSSM_UNEXP_CPL_ERR_BIT) + +#define ADF_GEN6_UERRSSMSH 0x18 + +/* + * UERRSSMSH error bit mask definitions + * + * BIT(0) - Indicates one uncorrectable error + * BIT(15) - Indicates multiple uncorrectable errors + * in device shared memory + */ +#define ADF_GEN6_UERRSSMSH_MASK (BIT(0) | BIT(15)) + +/* + * CERRSSMSH error bit + * BIT(0) - Indicates one correctable error + */ +#define ADF_GEN6_CERRSSMSH_ERROR_BIT (BIT(0) | BIT(15) | BIT(24)) +#define ADF_GEN6_CERRSSMSH 0x10 + +#define ADF_GEN6_INTMASKSSM 0x0 + +/* + * Error reporting mask in INTMASKSSM + * BIT(0) - Shared memory uncorrectable interrupt mask + * BIT(2) - PPERR interrupt mask + * BIT(4) - SCM parity error interrupt mask + * BIT(5) - CPP parity error interrupt mask + * BIT(6) - SHRAM RF parity error interrupt mask + * BIT(7) - AXI unexpected completion error mask + */ +#define ADF_GEN6_INTMASKSSM_MASK \ + (BIT(0) | BIT(2) | BIT(4) | BIT(5) | BIT(6) | BIT(7)) + +/* CPP push or pull error */ +#define ADF_GEN6_PPERR 0x8 + +#define ADF_GEN6_PPERR_MASK (BIT(0) | BIT(1)) + +/* + * SSM_FERR_STATUS error bit mask definitions + */ +#define ADF_GEN6_SCM_PAR_ERR_MASK BIT(5) +#define ADF_GEN6_CPP_PAR_ERR_MASK (BIT(0) | BIT(1) | BIT(2)) +#define ADF_GEN6_UNEXP_CPL_ERR_MASK (BIT(3) | BIT(4) | BIT(10) | BIT(11)) +#define ADF_GEN6_RF_PAR_ERR_MASK BIT(16) + +#define ADF_GEN6_SSM_FERR_STATUS 0x9C + +#define ADF_GEN6_CPP_CFC_ERR_STATUS 0x640C04 + +/* + * BIT(0) - Indicates one or more CPP CFC errors + * BIT(1) - Indicates multiple CPP CFC errors + * BIT(7) - Indicates CPP CFC command parity error type + * BIT(8) - Indicates CPP CFC data parity error type + */ +#define ADF_GEN6_CPP_CFC_ERR_STATUS_ERR_BIT BIT(0) +#define ADF_GEN6_CPP_CFC_ERR_STATUS_MERR_BIT BIT(1) +#define ADF_GEN6_CPP_CFC_ERR_STATUS_CMDPAR_BIT BIT(7) +#define ADF_GEN6_CPP_CFC_ERR_STATUS_DATAPAR_BIT BIT(8) +#define ADF_GEN6_CPP_CFC_FATAL_ERR_BIT \ + (ADF_GEN6_CPP_CFC_ERR_STATUS_ERR_BIT | \ + ADF_GEN6_CPP_CFC_ERR_STATUS_MERR_BIT) + +/* + * BIT(0) - Enables CFC to detect and log a push/pull data error + * BIT(1) - Enables CFC to generate interrupt to PCIEP for a CPP error + * BIT(4) - When 1 parity detection is disabled + * BIT(5) - When 1 parity detection is disabled on CPP command bus + * BIT(6) - When 1 parity detection is disabled on CPP push/pull bus + * BIT(9) - When 1 RF parity error detection is disabled + */ +#define ADF_GEN6_CPP_CFC_ERR_CTRL_MASK (BIT(0) | BIT(1)) + +#define ADF_GEN6_CPP_CFC_ERR_CTRL_DIS_MASK \ + (BIT(4) | BIT(5) | BIT(6) | BIT(9) | BIT(10)) + +#define ADF_GEN6_CPP_CFC_ERR_CTRL 0x640C00 + +/* + * BIT(0) - Clears bit(0) of ADF_GEN6_CPP_CFC_ERR_STATUS + * when an error is reported on CPP + * BIT(1) - Clears bit(1) of ADF_GEN6_CPP_CFC_ERR_STATUS + * when multiple errors are reported on CPP + * BIT(2) - Clears bit(2) of ADF_GEN6_CPP_CFC_ERR_STATUS + * when attention interrupt is reported + */ +#define ADF_GEN6_CPP_CFC_ERR_STATUS_CLR_MASK (BIT(0) | BIT(1) | BIT(2)) +#define ADF_GEN6_CPP_CFC_ERR_STATUS_CLR 0x640C08 + +/* + * ERRSOU3 bit masks + * BIT(0) - indicates error response order overflow and/or BME error + * BIT(1) - indicates RI push/pull error + * BIT(2) - indicates TI push/pull error + * BIT(5) - indicates TI pull parity error + * BIT(6) - indicates RI push parity error + * BIT(7) - indicates VFLR interrupt + * BIT(8) - indicates ring pair interrupts for ATU detected fault + * BIT(9) - indicates rate limiting error + */ +#define ADF_GEN6_ERRSOU3_TIMISCSTS_BIT BIT(0) +#define ADF_GEN6_ERRSOU3_RICPPINTSTS_MASK (BIT(1) | BIT(6)) +#define ADF_GEN6_ERRSOU3_TICPPINTSTS_MASK (BIT(2) | BIT(5)) +#define ADF_GEN6_ERRSOU3_VFLRNOTIFY_BIT BIT(7) +#define ADF_GEN6_ERRSOU3_ATUFAULTSTATUS_BIT BIT(8) +#define ADF_GEN6_ERRSOU3_RLTERROR_BIT BIT(9) +#define ADF_GEN6_ERRSOU3_TC_VC_MAP_ERROR_BIT BIT(16) +#define ADF_GEN6_ERRSOU3_PCIE_DEVHALT_BIT BIT(17) +#define ADF_GEN6_ERRSOU3_PG_REQ_DEVHALT_BIT BIT(18) +#define ADF_GEN6_ERRSOU3_XLT_CPL_DEVHALT_BIT BIT(19) +#define ADF_GEN6_ERRSOU3_TI_INT_ERR_DEVHALT_BIT BIT(20) + +#define ADF_GEN6_ERRSOU3_MASK ( \ + (ADF_GEN6_ERRSOU3_TIMISCSTS_BIT) | \ + (ADF_GEN6_ERRSOU3_RICPPINTSTS_MASK) | \ + (ADF_GEN6_ERRSOU3_TICPPINTSTS_MASK) | \ + (ADF_GEN6_ERRSOU3_VFLRNOTIFY_BIT) | \ + (ADF_GEN6_ERRSOU3_ATUFAULTSTATUS_BIT) | \ + (ADF_GEN6_ERRSOU3_RLTERROR_BIT) | \ + (ADF_GEN6_ERRSOU3_TC_VC_MAP_ERROR_BIT) | \ + (ADF_GEN6_ERRSOU3_PCIE_DEVHALT_BIT) | \ + (ADF_GEN6_ERRSOU3_PG_REQ_DEVHALT_BIT) | \ + (ADF_GEN6_ERRSOU3_XLT_CPL_DEVHALT_BIT) | \ + (ADF_GEN6_ERRSOU3_TI_INT_ERR_DEVHALT_BIT)) + +#define ADF_GEN6_ERRSOU3_DIS_MASK ( \ + (ADF_GEN6_ERRSOU3_TIMISCSTS_BIT) | \ + (ADF_GEN6_ERRSOU3_RICPPINTSTS_MASK) | \ + (ADF_GEN6_ERRSOU3_TICPPINTSTS_MASK) | \ + (ADF_GEN6_ERRSOU3_VFLRNOTIFY_BIT) | \ + (ADF_GEN6_ERRSOU3_ATUFAULTSTATUS_BIT) | \ + (ADF_GEN6_ERRSOU3_RLTERROR_BIT) | \ + (ADF_GEN6_ERRSOU3_TC_VC_MAP_ERROR_BIT)) + +/* Rate limiting error log register */ +#define ADF_GEN6_RLT_ERRLOG 0x508814 + +#define ADF_GEN6_RLT_ERRLOG_MASK (BIT(0) | BIT(1) | BIT(2) | BIT(3)) + +/* TI misc status register */ +#define ADF_GEN6_TIMISCSTS 0x50054C + +/* TI misc error reporting mask */ +#define ADF_GEN6_TIMISCCTL 0x500548 + +/* + * TI Misc error reporting control mask + * BIT(0) - Enables error detection and logging in TIMISCSTS register + * BIT(1) - It has effect only when SRIOV enabled, this bit is 0 by default + * BIT(2) - Enables the D-F-x counter within the dispatch arbiter + * to start based on the command triggered from + * BIT(30) - Disables VFLR functionality + * bits 1, 2 and 30 value should be preserved and not meant to be changed + * within RAS. + */ +#define ADF_GEN6_TIMISCCTL_BIT BIT(0) +#define ADF_GEN6_TIMSCCTL_RELAY_MASK (BIT(1) | BIT(2) | BIT(30)) + +/* RI CPP interface status register */ +#define ADF_GEN6_RICPPINTSTS 0x41A330 + +/* + * Uncorrectable error mask in RICPPINTSTS register + * BIT(0) - RI asserted the CPP error signal during a push + * BIT(1) - RI detected the CPP error signal asserted during a pull + * BIT(2) - RI detected a push data parity error + * BIT(3) - RI detected a push valid parity error + */ +#define ADF_GEN6_RICPPINTSTS_MASK (BIT(0) | BIT(1) | BIT(2) | BIT(3)) + +/* RI CPP interface register control */ +#define ADF_GEN6_RICPPINTCTL 0x41A32C + +/* + * Control bit mask for RICPPINTCTL register + * BIT(0) - value of 1 enables error detection and reporting + * on the RI CPP Push interface + * BIT(1) - value of 1 enables error detection and reporting + * on the RI CPP Pull interface + * BIT(2) - value of 1 enables error detection and reporting + * on the RI Parity + * BIT(3) - value of 1 enable checking parity on CPP + */ +#define ADF_GEN6_RICPPINTCTL_MASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4)) + +/* TI CPP interface status register */ +#define ADF_GEN6_TICPPINTSTS 0x50053C + +/* + * Uncorrectable error mask in TICPPINTSTS register + * BIT(0) - value of 1 indicates that the TI asserted + * the CPP error signal during a push + * BIT(1) - value of 1 indicates that the TI detected + * the CPP error signal asserted during a pull + * BIT(2) - value of 1 indicates that the TI detected + * a pull data parity error + */ +#define ADF_GEN6_TICPPINTSTS_MASK (BIT(0) | BIT(1) | BIT(2)) + +/* TI CPP interface status register control */ +#define ADF_GEN6_TICPPINTCTL 0x500538 + +/* + * Control bit mask for TICPPINTCTL register + * BIT(0) - value of 1 enables error detection and reporting on + * the TI CPP Push interface + * BIT(1) - value of 1 enables error detection and reporting on + * the TI CPP Push interface + * BIT(2) - value of 1 enables parity error detection and logging on + * the TI CPP Pull interface + * BIT(3) - value of 1 enables CPP CMD and Pull Data parity checking + */ +#define ADF_GEN6_TICPPINTCTL_MASK \ + (BIT(0) | BIT(1) | BIT(2) | BIT(3) | BIT(4)) + +/* ATU fault status register */ +#define ADF_GEN6_ATUFAULTSTATUS(i) (0x506000 + ((i) * 0x4)) + +#define ADF_GEN6_ATUFAULTSTATUS_BIT BIT(0) + +/* Command parity error detected on IOSFP command to QAT */ +#define ADF_GEN6_RIMISCSTS_BIT BIT(0) + +#define ADF_GEN6_GENSTS 0x41A220 +#define ADF_GEN6_GENSTS_DEVICE_STATE_MASK GENMASK(1, 0) +#define ADF_GEN6_GENSTS_RESET_TYPE_MASK GENMASK(3, 2) +#define ADF_GEN6_GENSTS_PFLR 0x1 +#define ADF_GEN6_GENSTS_COLD_RESET 0x3 +#define ADF_GEN6_GENSTS_DEVHALT 0x1 + +void adf_gen6_init_ras_ops(struct adf_ras_ops *ras_ops); + +#endif /* ADF_GEN6_RAS_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_shared.c b/drivers/crypto/intel/qat/qat_common/adf_gen6_shared.c new file mode 100644 index 000000000000..c9b151006dca --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_shared.c @@ -0,0 +1,56 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2025 Intel Corporation */ +#include <linux/export.h> + +#include "adf_gen4_config.h" +#include "adf_gen4_hw_csr_data.h" +#include "adf_gen4_pfvf.h" +#include "adf_gen4_vf_mig.h" +#include "adf_gen6_shared.h" + +struct adf_accel_dev; +struct adf_pfvf_ops; +struct adf_hw_csr_ops; + +/* + * QAT GEN4 and GEN6 devices often differ in terms of supported features, + * options and internal logic. However, some of the mechanisms and register + * layout are shared between those two GENs. This file serves as an abstraction + * layer that allows to use existing GEN4 implementation that is also + * applicable to GEN6 without additional overhead and complexity. + */ +void adf_gen6_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops) +{ + adf_gen4_init_pf_pfvf_ops(pfvf_ops); +} +EXPORT_SYMBOL_GPL(adf_gen6_init_pf_pfvf_ops); + +void adf_gen6_init_hw_csr_ops(struct adf_hw_csr_ops *csr_ops) +{ + return adf_gen4_init_hw_csr_ops(csr_ops); +} +EXPORT_SYMBOL_GPL(adf_gen6_init_hw_csr_ops); + +int adf_gen6_cfg_dev_init(struct adf_accel_dev *accel_dev) +{ + return adf_gen4_cfg_dev_init(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_gen6_cfg_dev_init); + +int adf_gen6_comp_dev_config(struct adf_accel_dev *accel_dev) +{ + return adf_comp_dev_config(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_gen6_comp_dev_config); + +int adf_gen6_no_dev_config(struct adf_accel_dev *accel_dev) +{ + return adf_no_dev_config(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_gen6_no_dev_config); + +void adf_gen6_init_vf_mig_ops(struct qat_migdev_ops *vfmig_ops) +{ + adf_gen4_init_vf_mig_ops(vfmig_ops); +} +EXPORT_SYMBOL_GPL(adf_gen6_init_vf_mig_ops); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_shared.h b/drivers/crypto/intel/qat/qat_common/adf_gen6_shared.h new file mode 100644 index 000000000000..fc6fad029a70 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_shared.h @@ -0,0 +1,17 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ADF_GEN6_SHARED_H_ +#define ADF_GEN6_SHARED_H_ + +struct adf_hw_csr_ops; +struct qat_migdev_ops; +struct adf_accel_dev; +struct adf_pfvf_ops; + +void adf_gen6_init_pf_pfvf_ops(struct adf_pfvf_ops *pfvf_ops); +void adf_gen6_init_hw_csr_ops(struct adf_hw_csr_ops *csr_ops); +int adf_gen6_cfg_dev_init(struct adf_accel_dev *accel_dev); +int adf_gen6_comp_dev_config(struct adf_accel_dev *accel_dev); +int adf_gen6_no_dev_config(struct adf_accel_dev *accel_dev); +void adf_gen6_init_vf_mig_ops(struct qat_migdev_ops *vfmig_ops); +#endif/* ADF_GEN6_SHARED_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_tl.c b/drivers/crypto/intel/qat/qat_common/adf_gen6_tl.c new file mode 100644 index 000000000000..faa60b04c406 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_tl.c @@ -0,0 +1,258 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright (c) 2025 Intel Corporation. */ +#include <linux/export.h> + +#include "adf_gen6_tl.h" +#include "adf_telemetry.h" +#include "adf_tl_debugfs.h" +#include "icp_qat_fw_init_admin.h" + +#define ADF_GEN6_TL_DEV_REG_OFF(reg) ADF_TL_DEV_REG_OFF(reg, gen6) + +#define ADF_GEN6_TL_RP_REG_OFF(reg) ADF_TL_RP_REG_OFF(reg, gen6) + +#define ADF_GEN6_TL_SL_UTIL_COUNTER(_name) \ + ADF_TL_COUNTER("util_" #_name, ADF_TL_SIMPLE_COUNT, \ + ADF_TL_SLICE_REG_OFF(_name, reg_tm_slice_util, gen6)) + +#define ADF_GEN6_TL_SL_EXEC_COUNTER(_name) \ + ADF_TL_COUNTER("exec_" #_name, ADF_TL_SIMPLE_COUNT, \ + ADF_TL_SLICE_REG_OFF(_name, reg_tm_slice_exec_cnt, gen6)) + +#define SLICE_IDX(sl) offsetof(struct icp_qat_fw_init_admin_slice_cnt, sl##_cnt) + +#define ADF_GEN6_TL_CMDQ_WAIT_COUNTER(_name) \ + ADF_TL_COUNTER("cmdq_wait_" #_name, ADF_TL_SIMPLE_COUNT, \ + ADF_TL_CMDQ_REG_OFF(_name, reg_tm_cmdq_wait_cnt, gen6)) +#define ADF_GEN6_TL_CMDQ_EXEC_COUNTER(_name) \ + ADF_TL_COUNTER("cmdq_exec_" #_name, ADF_TL_SIMPLE_COUNT, \ + ADF_TL_CMDQ_REG_OFF(_name, reg_tm_cmdq_exec_cnt, gen6)) +#define ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(_name) \ + ADF_TL_COUNTER("cmdq_drain_" #_name, ADF_TL_SIMPLE_COUNT, \ + ADF_TL_CMDQ_REG_OFF(_name, reg_tm_cmdq_drain_cnt, \ + gen6)) + +#define CPR_QUEUE_COUNT 5 +#define DCPR_QUEUE_COUNT 3 +#define PKE_QUEUE_COUNT 1 +#define WAT_QUEUE_COUNT 7 +#define WCP_QUEUE_COUNT 7 +#define USC_QUEUE_COUNT 3 +#define ATH_QUEUE_COUNT 2 + +/* Device level counters. */ +static const struct adf_tl_dbg_counter dev_counters[] = { + /* PCIe partial transactions. */ + ADF_TL_COUNTER(PCI_TRANS_CNT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_prt_trans_cnt)), + /* Max read latency[ns]. */ + ADF_TL_COUNTER(MAX_RD_LAT_NAME, ADF_TL_COUNTER_NS, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_rd_lat_max)), + /* Read latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(RD_LAT_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_rd_lat_acc), + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_rd_cmpl_cnt)), + /* Max "get to put" latency[ns]. */ + ADF_TL_COUNTER(MAX_LAT_NAME, ADF_TL_COUNTER_NS, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_gp_lat_max)), + /* "Get to put" latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(LAT_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_gp_lat_acc), + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_ae_put_cnt)), + /* PCIe write bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_IN_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_bw_in)), + /* PCIe read bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_OUT_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_bw_out)), + /* Page request latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(PAGE_REQ_LAT_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_at_page_req_lat_acc), + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_at_page_req_cnt)), + /* Page translation latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(AT_TRANS_LAT_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_at_trans_lat_acc), + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_at_trans_lat_cnt)), + /* Maximum uTLB used. */ + ADF_TL_COUNTER(AT_MAX_UTLB_USED_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_at_max_utlb_used)), + /* Ring Empty average[ns] across all rings */ + ADF_TL_COUNTER_LATENCY(RE_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_re_acc), + ADF_GEN6_TL_DEV_REG_OFF(reg_tl_re_cnt)), +}; + +/* Accelerator utilization counters */ +static const struct adf_tl_dbg_counter sl_util_counters[ADF_TL_SL_CNT_COUNT] = { + /* Compression accelerator utilization. */ + [SLICE_IDX(cpr)] = ADF_GEN6_TL_SL_UTIL_COUNTER(cnv), + /* Decompression accelerator utilization. */ + [SLICE_IDX(dcpr)] = ADF_GEN6_TL_SL_UTIL_COUNTER(dcprz), + /* PKE accelerator utilization. */ + [SLICE_IDX(pke)] = ADF_GEN6_TL_SL_UTIL_COUNTER(pke), + /* Wireless Authentication accelerator utilization. */ + [SLICE_IDX(wat)] = ADF_GEN6_TL_SL_UTIL_COUNTER(wat), + /* Wireless Cipher accelerator utilization. */ + [SLICE_IDX(wcp)] = ADF_GEN6_TL_SL_UTIL_COUNTER(wcp), + /* UCS accelerator utilization. */ + [SLICE_IDX(ucs)] = ADF_GEN6_TL_SL_UTIL_COUNTER(ucs), + /* Authentication accelerator utilization. */ + [SLICE_IDX(ath)] = ADF_GEN6_TL_SL_UTIL_COUNTER(ath), +}; + +/* Accelerator execution counters */ +static const struct adf_tl_dbg_counter sl_exec_counters[ADF_TL_SL_CNT_COUNT] = { + /* Compression accelerator execution count. */ + [SLICE_IDX(cpr)] = ADF_GEN6_TL_SL_EXEC_COUNTER(cnv), + /* Decompression accelerator execution count. */ + [SLICE_IDX(dcpr)] = ADF_GEN6_TL_SL_EXEC_COUNTER(dcprz), + /* PKE execution count. */ + [SLICE_IDX(pke)] = ADF_GEN6_TL_SL_EXEC_COUNTER(pke), + /* Wireless Authentication accelerator execution count. */ + [SLICE_IDX(wat)] = ADF_GEN6_TL_SL_EXEC_COUNTER(wat), + /* Wireless Cipher accelerator execution count. */ + [SLICE_IDX(wcp)] = ADF_GEN6_TL_SL_EXEC_COUNTER(wcp), + /* UCS accelerator execution count. */ + [SLICE_IDX(ucs)] = ADF_GEN6_TL_SL_EXEC_COUNTER(ucs), + /* Authentication accelerator execution count. */ + [SLICE_IDX(ath)] = ADF_GEN6_TL_SL_EXEC_COUNTER(ath), +}; + +static const struct adf_tl_dbg_counter cnv_cmdq_counters[] = { + ADF_GEN6_TL_CMDQ_WAIT_COUNTER(cnv), + ADF_GEN6_TL_CMDQ_EXEC_COUNTER(cnv), + ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(cnv) +}; + +#define NUM_CMDQ_COUNTERS ARRAY_SIZE(cnv_cmdq_counters) + +static const struct adf_tl_dbg_counter dcprz_cmdq_counters[] = { + ADF_GEN6_TL_CMDQ_WAIT_COUNTER(dcprz), + ADF_GEN6_TL_CMDQ_EXEC_COUNTER(dcprz), + ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(dcprz) +}; + +static_assert(ARRAY_SIZE(dcprz_cmdq_counters) == NUM_CMDQ_COUNTERS); + +static const struct adf_tl_dbg_counter pke_cmdq_counters[] = { + ADF_GEN6_TL_CMDQ_WAIT_COUNTER(pke), + ADF_GEN6_TL_CMDQ_EXEC_COUNTER(pke), + ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(pke) +}; + +static_assert(ARRAY_SIZE(pke_cmdq_counters) == NUM_CMDQ_COUNTERS); + +static const struct adf_tl_dbg_counter wat_cmdq_counters[] = { + ADF_GEN6_TL_CMDQ_WAIT_COUNTER(wat), + ADF_GEN6_TL_CMDQ_EXEC_COUNTER(wat), + ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(wat) +}; + +static_assert(ARRAY_SIZE(wat_cmdq_counters) == NUM_CMDQ_COUNTERS); + +static const struct adf_tl_dbg_counter wcp_cmdq_counters[] = { + ADF_GEN6_TL_CMDQ_WAIT_COUNTER(wcp), + ADF_GEN6_TL_CMDQ_EXEC_COUNTER(wcp), + ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(wcp) +}; + +static_assert(ARRAY_SIZE(wcp_cmdq_counters) == NUM_CMDQ_COUNTERS); + +static const struct adf_tl_dbg_counter ucs_cmdq_counters[] = { + ADF_GEN6_TL_CMDQ_WAIT_COUNTER(ucs), + ADF_GEN6_TL_CMDQ_EXEC_COUNTER(ucs), + ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(ucs) +}; + +static_assert(ARRAY_SIZE(ucs_cmdq_counters) == NUM_CMDQ_COUNTERS); + +static const struct adf_tl_dbg_counter ath_cmdq_counters[] = { + ADF_GEN6_TL_CMDQ_WAIT_COUNTER(ath), + ADF_GEN6_TL_CMDQ_EXEC_COUNTER(ath), + ADF_GEN6_TL_CMDQ_DRAIN_COUNTER(ath) +}; + +static_assert(ARRAY_SIZE(ath_cmdq_counters) == NUM_CMDQ_COUNTERS); + +/* CMDQ drain counters. */ +static const struct adf_tl_dbg_counter *cmdq_counters[ADF_TL_SL_CNT_COUNT] = { + /* Compression accelerator execution count. */ + [SLICE_IDX(cpr)] = cnv_cmdq_counters, + /* Decompression accelerator execution count. */ + [SLICE_IDX(dcpr)] = dcprz_cmdq_counters, + /* PKE execution count. */ + [SLICE_IDX(pke)] = pke_cmdq_counters, + /* Wireless Authentication accelerator execution count. */ + [SLICE_IDX(wat)] = wat_cmdq_counters, + /* Wireless Cipher accelerator execution count. */ + [SLICE_IDX(wcp)] = wcp_cmdq_counters, + /* UCS accelerator execution count. */ + [SLICE_IDX(ucs)] = ucs_cmdq_counters, + /* Authentication accelerator execution count. */ + [SLICE_IDX(ath)] = ath_cmdq_counters, +}; + +/* Ring pair counters. */ +static const struct adf_tl_dbg_counter rp_counters[] = { + /* PCIe partial transactions. */ + ADF_TL_COUNTER(PCI_TRANS_CNT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_prt_trans_cnt)), + /* "Get to put" latency average[ns]. */ + ADF_TL_COUNTER_LATENCY(LAT_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_gp_lat_acc), + ADF_GEN6_TL_RP_REG_OFF(reg_tl_ae_put_cnt)), + /* PCIe write bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_IN_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_bw_in)), + /* PCIe read bandwidth[Mbps]. */ + ADF_TL_COUNTER(BW_OUT_NAME, ADF_TL_COUNTER_MBPS, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_bw_out)), + /* Message descriptor DevTLB hit rate. */ + ADF_TL_COUNTER(AT_GLOB_DTLB_HIT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_at_glob_devtlb_hit)), + /* Message descriptor DevTLB miss rate. */ + ADF_TL_COUNTER(AT_GLOB_DTLB_MISS_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_at_glob_devtlb_miss)), + /* Payload DevTLB hit rate. */ + ADF_TL_COUNTER(AT_PAYLD_DTLB_HIT_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_at_payld_devtlb_hit)), + /* Payload DevTLB miss rate. */ + ADF_TL_COUNTER(AT_PAYLD_DTLB_MISS_NAME, ADF_TL_SIMPLE_COUNT, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_at_payld_devtlb_miss)), + /* Ring Empty average[ns]. */ + ADF_TL_COUNTER_LATENCY(RE_ACC_NAME, ADF_TL_COUNTER_NS_AVG, + ADF_GEN6_TL_RP_REG_OFF(reg_tl_re_acc), + ADF_GEN6_TL_RP_REG_OFF(reg_tl_re_cnt)), +}; + +void adf_gen6_init_tl_data(struct adf_tl_hw_data *tl_data) +{ + tl_data->layout_sz = ADF_GEN6_TL_LAYOUT_SZ; + tl_data->slice_reg_sz = ADF_GEN6_TL_SLICE_REG_SZ; + tl_data->cmdq_reg_sz = ADF_GEN6_TL_CMDQ_REG_SZ; + tl_data->rp_reg_sz = ADF_GEN6_TL_RP_REG_SZ; + tl_data->num_hbuff = ADF_GEN6_TL_NUM_HIST_BUFFS; + tl_data->max_rp = ADF_GEN6_TL_MAX_RP_NUM; + tl_data->msg_cnt_off = ADF_GEN6_TL_MSG_CNT_OFF; + tl_data->cpp_ns_per_cycle = ADF_GEN6_CPP_NS_PER_CYCLE; + tl_data->bw_units_to_bytes = ADF_GEN6_TL_BW_HW_UNITS_TO_BYTES; + + tl_data->dev_counters = dev_counters; + tl_data->num_dev_counters = ARRAY_SIZE(dev_counters); + tl_data->sl_util_counters = sl_util_counters; + tl_data->sl_exec_counters = sl_exec_counters; + tl_data->cmdq_counters = cmdq_counters; + tl_data->num_cmdq_counters = NUM_CMDQ_COUNTERS; + tl_data->rp_counters = rp_counters; + tl_data->num_rp_counters = ARRAY_SIZE(rp_counters); + tl_data->max_sl_cnt = ADF_GEN6_TL_MAX_SLICES_PER_TYPE; + + tl_data->multiplier.cpr_cnt = CPR_QUEUE_COUNT; + tl_data->multiplier.dcpr_cnt = DCPR_QUEUE_COUNT; + tl_data->multiplier.pke_cnt = PKE_QUEUE_COUNT; + tl_data->multiplier.wat_cnt = WAT_QUEUE_COUNT; + tl_data->multiplier.wcp_cnt = WCP_QUEUE_COUNT; + tl_data->multiplier.ucs_cnt = USC_QUEUE_COUNT; + tl_data->multiplier.ath_cnt = ATH_QUEUE_COUNT; +} +EXPORT_SYMBOL_GPL(adf_gen6_init_tl_data); diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen6_tl.h b/drivers/crypto/intel/qat/qat_common/adf_gen6_tl.h new file mode 100644 index 000000000000..49db660b8eb9 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_gen6_tl.h @@ -0,0 +1,198 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright (c) 2025 Intel Corporation. */ +#ifndef ADF_GEN6_TL_H +#define ADF_GEN6_TL_H + +#include <linux/types.h> + +struct adf_tl_hw_data; + +/* Computation constants. */ +#define ADF_GEN6_CPP_NS_PER_CYCLE 2 +#define ADF_GEN6_TL_BW_HW_UNITS_TO_BYTES 64 + +/* Maximum aggregation time. Value is in milliseconds. */ +#define ADF_GEN6_TL_MAX_AGGR_TIME_MS 4000 +/* Number of buffers to store historic values. */ +#define ADF_GEN6_TL_NUM_HIST_BUFFS \ + (ADF_GEN6_TL_MAX_AGGR_TIME_MS / ADF_TL_DATA_WR_INTERVAL_MS) + +/* Max number of HW resources of one type */ +#define ADF_GEN6_TL_MAX_SLICES_PER_TYPE 32 +#define MAX_ATH_SL_COUNT 7 +#define MAX_CNV_SL_COUNT 2 +#define MAX_DCPRZ_SL_COUNT 2 +#define MAX_PKE_SL_COUNT 32 +#define MAX_UCS_SL_COUNT 4 +#define MAX_WAT_SL_COUNT 5 +#define MAX_WCP_SL_COUNT 5 + +#define MAX_ATH_CMDQ_COUNT 14 +#define MAX_CNV_CMDQ_COUNT 6 +#define MAX_DCPRZ_CMDQ_COUNT 6 +#define MAX_PKE_CMDQ_COUNT 32 +#define MAX_UCS_CMDQ_COUNT 12 +#define MAX_WAT_CMDQ_COUNT 35 +#define MAX_WCP_CMDQ_COUNT 35 + +/* Max number of simultaneously monitored ring pairs. */ +#define ADF_GEN6_TL_MAX_RP_NUM 4 + +/** + * struct adf_gen6_tl_slice_data_regs - HW slice data as populated by FW. + * @reg_tm_slice_exec_cnt: Slice execution count. + * @reg_tm_slice_util: Slice utilization. + */ +struct adf_gen6_tl_slice_data_regs { + __u32 reg_tm_slice_exec_cnt; + __u32 reg_tm_slice_util; +}; + +#define ADF_GEN6_TL_SLICE_REG_SZ sizeof(struct adf_gen6_tl_slice_data_regs) + +/** + * struct adf_gen6_tl_cmdq_data_regs - HW CMDQ data as populated by FW. + * @reg_tm_cmdq_wait_cnt: CMDQ wait count. + * @reg_tm_cmdq_exec_cnt: CMDQ execution count. + * @reg_tm_cmdq_drain_cnt: CMDQ drain count. + */ +struct adf_gen6_tl_cmdq_data_regs { + __u32 reg_tm_cmdq_wait_cnt; + __u32 reg_tm_cmdq_exec_cnt; + __u32 reg_tm_cmdq_drain_cnt; + __u32 reserved; +}; + +#define ADF_GEN6_TL_CMDQ_REG_SZ sizeof(struct adf_gen6_tl_cmdq_data_regs) + +/** + * struct adf_gen6_tl_device_data_regs - This structure stores device telemetry + * counter values as are being populated periodically by device. + * @reg_tl_rd_lat_acc: read latency accumulator + * @reg_tl_gp_lat_acc: "get to put" latency accumulator + * @reg_tl_at_page_req_lat_acc: AT/DevTLB page request latency accumulator + * @reg_tl_at_trans_lat_acc: DevTLB transaction latency accumulator + * @reg_tl_re_acc: accumulated ring empty time + * @reg_tl_prt_trans_cnt: PCIe partial transactions + * @reg_tl_rd_lat_max: maximum logged read latency + * @reg_tl_rd_cmpl_cnt: read requests completed count + * @reg_tl_gp_lat_max: maximum logged get to put latency + * @reg_tl_ae_put_cnt: Accelerator Engine put counts across all rings + * @reg_tl_bw_in: PCIe write bandwidth + * @reg_tl_bw_out: PCIe read bandwidth + * @reg_tl_at_page_req_cnt: DevTLB page requests count + * @reg_tl_at_trans_lat_cnt: DevTLB transaction latency samples count + * @reg_tl_at_max_utlb_used: maximum uTLB used + * @reg_tl_re_cnt: ring empty time samples count + * @reserved: reserved + * @ath_slices: array of Authentication slices utilization registers + * @cnv_slices: array of Compression slices utilization registers + * @dcprz_slices: array of Decompression slices utilization registers + * @pke_slices: array of PKE slices utilization registers + * @ucs_slices: array of UCS slices utilization registers + * @wat_slices: array of Wireless Authentication slices utilization registers + * @wcp_slices: array of Wireless Cipher slices utilization registers + * @ath_cmdq: array of Authentication cmdq telemetry registers + * @cnv_cmdq: array of Compression cmdq telemetry registers + * @dcprz_cmdq: array of Decomopression cmdq telemetry registers + * @pke_cmdq: array of PKE cmdq telemetry registers + * @ucs_cmdq: array of UCS cmdq telemetry registers + * @wat_cmdq: array of Wireless Authentication cmdq telemetry registers + * @wcp_cmdq: array of Wireless Cipher cmdq telemetry registers + */ +struct adf_gen6_tl_device_data_regs { + __u64 reg_tl_rd_lat_acc; + __u64 reg_tl_gp_lat_acc; + __u64 reg_tl_at_page_req_lat_acc; + __u64 reg_tl_at_trans_lat_acc; + __u64 reg_tl_re_acc; + __u32 reg_tl_prt_trans_cnt; + __u32 reg_tl_rd_lat_max; + __u32 reg_tl_rd_cmpl_cnt; + __u32 reg_tl_gp_lat_max; + __u32 reg_tl_ae_put_cnt; + __u32 reg_tl_bw_in; + __u32 reg_tl_bw_out; + __u32 reg_tl_at_page_req_cnt; + __u32 reg_tl_at_trans_lat_cnt; + __u32 reg_tl_at_max_utlb_used; + __u32 reg_tl_re_cnt; + __u32 reserved; + struct adf_gen6_tl_slice_data_regs ath_slices[MAX_ATH_SL_COUNT]; + struct adf_gen6_tl_slice_data_regs cnv_slices[MAX_CNV_SL_COUNT]; + struct adf_gen6_tl_slice_data_regs dcprz_slices[MAX_DCPRZ_SL_COUNT]; + struct adf_gen6_tl_slice_data_regs pke_slices[MAX_PKE_SL_COUNT]; + struct adf_gen6_tl_slice_data_regs ucs_slices[MAX_UCS_SL_COUNT]; + struct adf_gen6_tl_slice_data_regs wat_slices[MAX_WAT_SL_COUNT]; + struct adf_gen6_tl_slice_data_regs wcp_slices[MAX_WCP_SL_COUNT]; + struct adf_gen6_tl_cmdq_data_regs ath_cmdq[MAX_ATH_CMDQ_COUNT]; + struct adf_gen6_tl_cmdq_data_regs cnv_cmdq[MAX_CNV_CMDQ_COUNT]; + struct adf_gen6_tl_cmdq_data_regs dcprz_cmdq[MAX_DCPRZ_CMDQ_COUNT]; + struct adf_gen6_tl_cmdq_data_regs pke_cmdq[MAX_PKE_CMDQ_COUNT]; + struct adf_gen6_tl_cmdq_data_regs ucs_cmdq[MAX_UCS_CMDQ_COUNT]; + struct adf_gen6_tl_cmdq_data_regs wat_cmdq[MAX_WAT_CMDQ_COUNT]; + struct adf_gen6_tl_cmdq_data_regs wcp_cmdq[MAX_WCP_CMDQ_COUNT]; +}; + +/** + * struct adf_gen6_tl_ring_pair_data_regs - This structure stores ring pair + * telemetry counter values as they are being populated periodically by device. + * @reg_tl_gp_lat_acc: get-put latency accumulator + * @reg_tl_re_acc: accumulated ring empty time + * @reg_tl_pci_trans_cnt: PCIe partial transactions + * @reg_tl_ae_put_cnt: Accelerator Engine put counts across all rings + * @reg_tl_bw_in: PCIe write bandwidth + * @reg_tl_bw_out: PCIe read bandwidth + * @reg_tl_at_glob_devtlb_hit: Message descriptor DevTLB hit rate + * @reg_tl_at_glob_devtlb_miss: Message descriptor DevTLB miss rate + * @reg_tl_at_payld_devtlb_hit: Payload DevTLB hit rate + * @reg_tl_at_payld_devtlb_miss: Payload DevTLB miss rate + * @reg_tl_re_cnt: ring empty time samples count + * @reserved1: reserved + */ +struct adf_gen6_tl_ring_pair_data_regs { + __u64 reg_tl_gp_lat_acc; + __u64 reg_tl_re_acc; + __u32 reg_tl_prt_trans_cnt; + __u32 reg_tl_ae_put_cnt; + __u32 reg_tl_bw_in; + __u32 reg_tl_bw_out; + __u32 reg_tl_at_glob_devtlb_hit; + __u32 reg_tl_at_glob_devtlb_miss; + __u32 reg_tl_at_payld_devtlb_hit; + __u32 reg_tl_at_payld_devtlb_miss; + __u32 reg_tl_re_cnt; + __u32 reserved1; +}; + +#define ADF_GEN6_TL_RP_REG_SZ sizeof(struct adf_gen6_tl_ring_pair_data_regs) + +/** + * struct adf_gen6_tl_layout - This structure represents the entire telemetry + * counters data: Device + 4 Ring Pairs as they are being populated periodically + * by device. + * @tl_device_data_regs: structure of device telemetry registers + * @tl_ring_pairs_data_regs: array of ring pairs telemetry registers + * @reg_tl_msg_cnt: telemetry message counter + * @reserved: reserved + */ +struct adf_gen6_tl_layout { + struct adf_gen6_tl_device_data_regs tl_device_data_regs; + struct adf_gen6_tl_ring_pair_data_regs + tl_ring_pairs_data_regs[ADF_GEN6_TL_MAX_RP_NUM]; + __u32 reg_tl_msg_cnt; + __u32 reserved; +}; + +#define ADF_GEN6_TL_LAYOUT_SZ sizeof(struct adf_gen6_tl_layout) +#define ADF_GEN6_TL_MSG_CNT_OFF \ + offsetof(struct adf_gen6_tl_layout, reg_tl_msg_cnt) + +#ifdef CONFIG_DEBUG_FS +void adf_gen6_init_tl_data(struct adf_tl_hw_data *tl_data); +#else +static inline void adf_gen6_init_tl_data(struct adf_tl_hw_data *tl_data) +{ +} +#endif /* CONFIG_DEBUG_FS */ +#endif /* ADF_GEN6_TL_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_heartbeat.c b/drivers/crypto/intel/qat/qat_common/adf_heartbeat.c new file mode 100644 index 000000000000..b19aa1ef8eee --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_heartbeat.c @@ -0,0 +1,345 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/dev_printk.h> +#include <linux/dma-mapping.h> +#include <linux/export.h> +#include <linux/kernel.h> +#include <linux/kstrtox.h> +#include <linux/overflow.h> +#include <linux/string.h> +#include <linux/slab.h> +#include <linux/types.h> +#include <asm/errno.h> +#include "adf_accel_devices.h" +#include "adf_admin.h" +#include "adf_cfg.h" +#include "adf_cfg_strings.h" +#include "adf_clock.h" +#include "adf_common_drv.h" +#include "adf_heartbeat.h" +#include "adf_transport_internal.h" +#include "icp_qat_fw_init_admin.h" + +#define ADF_HB_EMPTY_SIG 0xA5A5A5A5 + +static int adf_hb_check_polling_freq(struct adf_accel_dev *accel_dev) +{ + u64 curr_time = adf_clock_get_current_time(); + u64 polling_time = curr_time - accel_dev->heartbeat->last_hb_check_time; + + if (polling_time < accel_dev->heartbeat->hb_timer) { + dev_warn(&GET_DEV(accel_dev), + "HB polling too frequent. Configured HB timer %d ms\n", + accel_dev->heartbeat->hb_timer); + return -EINVAL; + } + + accel_dev->heartbeat->last_hb_check_time = curr_time; + return 0; +} + +/** + * validate_hb_ctrs_cnt() - checks if the number of heartbeat counters should + * be updated by one to support the currently loaded firmware. + * @accel_dev: Pointer to acceleration device. + * + * Return: + * * true - hb_ctrs must increased by ADF_NUM_PKE_STRAND + * * false - no changes needed + */ +static bool validate_hb_ctrs_cnt(struct adf_accel_dev *accel_dev) +{ + const size_t hb_ctrs = accel_dev->hw_device->num_hb_ctrs; + const size_t max_aes = accel_dev->hw_device->num_engines; + const size_t hb_struct_size = sizeof(struct hb_cnt_pair); + const size_t exp_diff_size = array3_size(ADF_NUM_PKE_STRAND, max_aes, + hb_struct_size); + const size_t dev_ctrs = size_mul(max_aes, hb_ctrs); + const size_t stats_size = size_mul(dev_ctrs, hb_struct_size); + const u32 exp_diff_cnt = exp_diff_size / sizeof(u32); + const u32 stats_el_cnt = stats_size / sizeof(u32); + struct hb_cnt_pair *hb_stats = accel_dev->heartbeat->dma.virt_addr; + const u32 *mem_to_chk = (u32 *)(hb_stats + dev_ctrs); + u32 el_diff_cnt = 0; + int i; + + /* count how many bytes are different from pattern */ + for (i = 0; i < stats_el_cnt; i++) { + if (mem_to_chk[i] == ADF_HB_EMPTY_SIG) + break; + + el_diff_cnt++; + } + + return el_diff_cnt && el_diff_cnt == exp_diff_cnt; +} + +void adf_heartbeat_check_ctrs(struct adf_accel_dev *accel_dev) +{ + struct hb_cnt_pair *hb_stats = accel_dev->heartbeat->dma.virt_addr; + const size_t hb_ctrs = accel_dev->hw_device->num_hb_ctrs; + const size_t max_aes = accel_dev->hw_device->num_engines; + const size_t dev_ctrs = size_mul(max_aes, hb_ctrs); + const size_t stats_size = size_mul(dev_ctrs, sizeof(struct hb_cnt_pair)); + const size_t mem_items_to_fill = size_mul(stats_size, 2) / sizeof(u32); + + /* fill hb stats memory with pattern */ + memset32((uint32_t *)hb_stats, ADF_HB_EMPTY_SIG, mem_items_to_fill); + accel_dev->heartbeat->ctrs_cnt_checked = false; +} +EXPORT_SYMBOL_GPL(adf_heartbeat_check_ctrs); + +static int get_timer_ticks(struct adf_accel_dev *accel_dev, unsigned int *value) +{ + char timer_str[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = { }; + u32 timer_ms = ADF_CFG_HB_TIMER_DEFAULT_MS; + int cfg_read_status; + u32 ticks; + int ret; + + cfg_read_status = adf_cfg_get_param_value(accel_dev, ADF_GENERAL_SEC, + ADF_HEARTBEAT_TIMER, timer_str); + if (cfg_read_status == 0) { + if (kstrtouint(timer_str, 10, &timer_ms)) + dev_dbg(&GET_DEV(accel_dev), + "kstrtouint failed to parse the %s, param value", + ADF_HEARTBEAT_TIMER); + } + + if (timer_ms < ADF_CFG_HB_TIMER_MIN_MS) { + dev_err(&GET_DEV(accel_dev), "Timer cannot be less than %u\n", + ADF_CFG_HB_TIMER_MIN_MS); + return -EINVAL; + } + + /* + * On 4xxx devices adf_timer is responsible for HB updates and + * its period is fixed to 200ms + */ + if (accel_dev->timer) + timer_ms = ADF_CFG_HB_TIMER_MIN_MS; + + ret = adf_heartbeat_ms_to_ticks(accel_dev, timer_ms, &ticks); + if (ret) + return ret; + + adf_heartbeat_save_cfg_param(accel_dev, timer_ms); + + accel_dev->heartbeat->hb_timer = timer_ms; + *value = ticks; + + return 0; +} + +static int check_ae(struct hb_cnt_pair *curr, struct hb_cnt_pair *prev, + u16 *count, const size_t hb_ctrs) +{ + size_t thr; + + /* loop through all threads in AE */ + for (thr = 0; thr < hb_ctrs; thr++) { + u16 req = curr[thr].req_heartbeat_cnt; + u16 resp = curr[thr].resp_heartbeat_cnt; + u16 last = prev[thr].resp_heartbeat_cnt; + + if ((thr == ADF_AE_ADMIN_THREAD || req != resp) && resp == last) { + u16 retry = ++count[thr]; + + if (retry >= ADF_CFG_HB_COUNT_THRESHOLD) + return -EIO; + + } else { + count[thr] = 0; + } + } + return 0; +} + +static int adf_hb_get_status(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + struct hb_cnt_pair *live_stats, *last_stats, *curr_stats; + const size_t hb_ctrs = hw_device->num_hb_ctrs; + const unsigned long ae_mask = hw_device->ae_mask; + const size_t max_aes = hw_device->num_engines; + const size_t dev_ctrs = size_mul(max_aes, hb_ctrs); + const size_t stats_size = size_mul(dev_ctrs, sizeof(*curr_stats)); + struct hb_cnt_pair *ae_curr_p, *ae_prev_p; + u16 *count_fails, *ae_count_p; + size_t ae_offset; + size_t ae = 0; + int ret = 0; + + if (!accel_dev->heartbeat->ctrs_cnt_checked) { + if (validate_hb_ctrs_cnt(accel_dev)) + hw_device->num_hb_ctrs += ADF_NUM_PKE_STRAND; + + accel_dev->heartbeat->ctrs_cnt_checked = true; + } + + live_stats = accel_dev->heartbeat->dma.virt_addr; + last_stats = live_stats + dev_ctrs; + count_fails = (u16 *)(last_stats + dev_ctrs); + + curr_stats = kmemdup(live_stats, stats_size, GFP_KERNEL); + if (!curr_stats) + return -ENOMEM; + + /* loop through active AEs */ + for_each_set_bit(ae, &ae_mask, max_aes) { + ae_offset = size_mul(ae, hb_ctrs); + ae_curr_p = curr_stats + ae_offset; + ae_prev_p = last_stats + ae_offset; + ae_count_p = count_fails + ae_offset; + + ret = check_ae(ae_curr_p, ae_prev_p, ae_count_p, hb_ctrs); + if (ret) + break; + } + + /* Copy current stats for the next iteration */ + memcpy(last_stats, curr_stats, stats_size); + kfree(curr_stats); + + return ret; +} + +static void adf_heartbeat_reset(struct adf_accel_dev *accel_dev) +{ + u64 curr_time = adf_clock_get_current_time(); + u64 time_since_reset = curr_time - accel_dev->heartbeat->last_hb_reset_time; + + if (time_since_reset < ADF_CFG_HB_RESET_MS) + return; + + accel_dev->heartbeat->last_hb_reset_time = curr_time; + if (adf_notify_fatal_error(accel_dev)) + dev_err(&GET_DEV(accel_dev), "Failed to notify fatal error\n"); +} + +void adf_heartbeat_status(struct adf_accel_dev *accel_dev, + enum adf_device_heartbeat_status *hb_status) +{ + struct adf_heartbeat *hb; + + if (!adf_dev_started(accel_dev) || + test_bit(ADF_STATUS_RESTARTING, &accel_dev->status)) { + *hb_status = HB_DEV_UNRESPONSIVE; + return; + } + + if (adf_hb_check_polling_freq(accel_dev) == -EINVAL) { + *hb_status = HB_DEV_UNSUPPORTED; + return; + } + + hb = accel_dev->heartbeat; + hb->hb_sent_counter++; + + if (adf_hb_get_status(accel_dev)) { + dev_err(&GET_DEV(accel_dev), + "Heartbeat ERROR: QAT is not responding.\n"); + *hb_status = HB_DEV_UNRESPONSIVE; + hb->hb_failed_counter++; + adf_heartbeat_reset(accel_dev); + return; + } + + *hb_status = HB_DEV_ALIVE; +} + +int adf_heartbeat_ms_to_ticks(struct adf_accel_dev *accel_dev, unsigned int time_ms, + u32 *value) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u32 clk_per_sec; + + /* HB clock may be different than AE clock */ + if (!hw_data->get_hb_clock) + return -EINVAL; + + clk_per_sec = hw_data->get_hb_clock(hw_data); + *value = time_ms * (clk_per_sec / MSEC_PER_SEC); + + return 0; +} + +int adf_heartbeat_save_cfg_param(struct adf_accel_dev *accel_dev, + unsigned int timer_ms) +{ + char timer_str[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + + snprintf(timer_str, sizeof(timer_str), "%u", timer_ms); + return adf_cfg_add_key_value_param(accel_dev, ADF_GENERAL_SEC, + ADF_HEARTBEAT_TIMER, timer_str, + ADF_STR); +} +EXPORT_SYMBOL_GPL(adf_heartbeat_save_cfg_param); + +int adf_heartbeat_init(struct adf_accel_dev *accel_dev) +{ + struct adf_heartbeat *hb; + + hb = kzalloc(sizeof(*hb), GFP_KERNEL); + if (!hb) + goto err_ret; + + hb->dma.virt_addr = dma_alloc_coherent(&GET_DEV(accel_dev), PAGE_SIZE, + &hb->dma.phy_addr, GFP_KERNEL); + if (!hb->dma.virt_addr) + goto err_free; + + /* + * Default set this flag as true to avoid unnecessary checks, + * it will be reset on platforms that need such a check + */ + hb->ctrs_cnt_checked = true; + accel_dev->heartbeat = hb; + + return 0; + +err_free: + kfree(hb); +err_ret: + return -ENOMEM; +} + +int adf_heartbeat_start(struct adf_accel_dev *accel_dev) +{ + unsigned int timer_ticks; + int ret; + + if (!accel_dev->heartbeat) { + dev_warn(&GET_DEV(accel_dev), "Heartbeat instance not found!"); + return -EFAULT; + } + + if (accel_dev->hw_device->check_hb_ctrs) + accel_dev->hw_device->check_hb_ctrs(accel_dev); + + ret = get_timer_ticks(accel_dev, &timer_ticks); + if (ret) + return ret; + + ret = adf_send_admin_hb_timer(accel_dev, timer_ticks); + if (ret) + dev_warn(&GET_DEV(accel_dev), "Heartbeat not supported!"); + + return ret; +} + +void adf_heartbeat_shutdown(struct adf_accel_dev *accel_dev) +{ + struct adf_heartbeat *hb = accel_dev->heartbeat; + + if (!hb) + return; + + if (hb->dma.virt_addr) + dma_free_coherent(&GET_DEV(accel_dev), PAGE_SIZE, + hb->dma.virt_addr, hb->dma.phy_addr); + + kfree(hb); + accel_dev->heartbeat = NULL; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_heartbeat.h b/drivers/crypto/intel/qat/qat_common/adf_heartbeat.h new file mode 100644 index 000000000000..16fdfb48b196 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_heartbeat.h @@ -0,0 +1,100 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_HEARTBEAT_H_ +#define ADF_HEARTBEAT_H_ + +#include <linux/types.h> + +struct adf_accel_dev; +struct dentry; + +#define ADF_CFG_HB_TIMER_MIN_MS 200 +#define ADF_CFG_HB_TIMER_DEFAULT_MS 500 +#define ADF_CFG_HB_COUNT_THRESHOLD 3 + +#define ADF_CFG_HB_RESET_MS 5000 + +enum adf_device_heartbeat_status { + HB_DEV_UNRESPONSIVE = 0, + HB_DEV_ALIVE, + HB_DEV_UNSUPPORTED, +}; + +/* Heartbeat counter pair */ +struct hb_cnt_pair { + __u16 resp_heartbeat_cnt; + __u16 req_heartbeat_cnt; +}; + +struct adf_heartbeat { + unsigned int hb_sent_counter; + unsigned int hb_failed_counter; + unsigned int hb_timer; + u64 last_hb_check_time; + u64 last_hb_reset_time; + bool ctrs_cnt_checked; + struct hb_dma_addr { + dma_addr_t phy_addr; + void *virt_addr; + } dma; + struct { + struct dentry *base_dir; + struct dentry *status; + struct dentry *cfg; + struct dentry *sent; + struct dentry *failed; +#ifdef CONFIG_CRYPTO_DEV_QAT_ERROR_INJECTION + struct dentry *inject_error; +#endif + } dbgfs; +}; + +#ifdef CONFIG_DEBUG_FS +int adf_heartbeat_init(struct adf_accel_dev *accel_dev); +int adf_heartbeat_start(struct adf_accel_dev *accel_dev); +void adf_heartbeat_shutdown(struct adf_accel_dev *accel_dev); + +int adf_heartbeat_ms_to_ticks(struct adf_accel_dev *accel_dev, unsigned int time_ms, + uint32_t *value); +int adf_heartbeat_save_cfg_param(struct adf_accel_dev *accel_dev, + unsigned int timer_ms); +void adf_heartbeat_status(struct adf_accel_dev *accel_dev, + enum adf_device_heartbeat_status *hb_status); +void adf_heartbeat_check_ctrs(struct adf_accel_dev *accel_dev); + +#ifdef CONFIG_CRYPTO_DEV_QAT_ERROR_INJECTION +int adf_heartbeat_inject_error(struct adf_accel_dev *accel_dev); +#else +static inline int adf_heartbeat_inject_error(struct adf_accel_dev *accel_dev) +{ + return -EPERM; +} +#endif + +#else +static inline int adf_heartbeat_init(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static inline int adf_heartbeat_start(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static inline void adf_heartbeat_shutdown(struct adf_accel_dev *accel_dev) +{ +} + +static inline int adf_heartbeat_save_cfg_param(struct adf_accel_dev *accel_dev, + unsigned int timer_ms) +{ + return 0; +} + +static inline void adf_heartbeat_check_ctrs(struct adf_accel_dev *accel_dev) +{ +} +#endif +#endif /* ADF_HEARTBEAT_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_heartbeat_dbgfs.c b/drivers/crypto/intel/qat/qat_common/adf_heartbeat_dbgfs.c new file mode 100644 index 000000000000..cccdff24b48d --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_heartbeat_dbgfs.c @@ -0,0 +1,248 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/debugfs.h> +#include <linux/errno.h> +#include <linux/export.h> +#include <linux/fs.h> +#include <linux/kernel.h> +#include <linux/kstrtox.h> +#include <linux/types.h> +#include "adf_admin.h" +#include "adf_cfg.h" +#include "adf_common_drv.h" +#include "adf_heartbeat.h" +#include "adf_heartbeat_dbgfs.h" + +#define HB_OK 0 +#define HB_ERROR -1 +#define HB_STATUS_MAX_STRLEN 4 +#define HB_STATS_MAX_STRLEN 16 + +static ssize_t adf_hb_stats_read(struct file *file, char __user *user_buffer, + size_t count, loff_t *ppos) +{ + char buf[HB_STATS_MAX_STRLEN]; + unsigned int *value; + int len; + + if (*ppos > 0) + return 0; + + value = file->private_data; + len = scnprintf(buf, sizeof(buf), "%u\n", *value); + + return simple_read_from_buffer(user_buffer, count, ppos, buf, len + 1); +} + +static const struct file_operations adf_hb_stats_fops = { + .owner = THIS_MODULE, + .open = simple_open, + .read = adf_hb_stats_read, +}; + +static ssize_t adf_hb_status_read(struct file *file, char __user *user_buf, + size_t count, loff_t *ppos) +{ + enum adf_device_heartbeat_status hb_status; + char ret_str[HB_STATUS_MAX_STRLEN]; + struct adf_accel_dev *accel_dev; + int ret_code; + size_t len; + + if (*ppos > 0) + return 0; + + accel_dev = file->private_data; + ret_code = HB_OK; + + adf_heartbeat_status(accel_dev, &hb_status); + + if (hb_status != HB_DEV_ALIVE) + ret_code = HB_ERROR; + + len = scnprintf(ret_str, sizeof(ret_str), "%d\n", ret_code); + + return simple_read_from_buffer(user_buf, count, ppos, ret_str, len + 1); +} + +static const struct file_operations adf_hb_status_fops = { + .owner = THIS_MODULE, + .open = simple_open, + .read = adf_hb_status_read, +}; + +static ssize_t adf_hb_cfg_read(struct file *file, char __user *user_buf, + size_t count, loff_t *ppos) +{ + char timer_str[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + struct adf_accel_dev *accel_dev; + unsigned int timer_ms; + int len; + + if (*ppos > 0) + return 0; + + accel_dev = file->private_data; + timer_ms = accel_dev->heartbeat->hb_timer; + len = scnprintf(timer_str, sizeof(timer_str), "%u\n", timer_ms); + + return simple_read_from_buffer(user_buf, count, ppos, timer_str, + len + 1); +} + +static ssize_t adf_hb_cfg_write(struct file *file, const char __user *user_buf, + size_t count, loff_t *ppos) +{ + char input_str[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = { }; + struct adf_accel_dev *accel_dev; + int ret, written_chars; + unsigned int timer_ms; + u32 ticks; + + accel_dev = file->private_data; + timer_ms = ADF_CFG_HB_TIMER_DEFAULT_MS; + + /* last byte left as string termination */ + if (count > sizeof(input_str) - 1) + return -EINVAL; + + written_chars = simple_write_to_buffer(input_str, sizeof(input_str) - 1, + ppos, user_buf, count); + if (written_chars > 0) { + ret = kstrtouint(input_str, 10, &timer_ms); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "heartbeat_cfg: Invalid value\n"); + return ret; + } + + if (timer_ms < ADF_CFG_HB_TIMER_MIN_MS) { + dev_err(&GET_DEV(accel_dev), + "heartbeat_cfg: Invalid value\n"); + return -EINVAL; + } + + /* + * On 4xxx devices adf_timer is responsible for HB updates and + * its period is fixed to 200ms + */ + if (accel_dev->timer) + timer_ms = ADF_CFG_HB_TIMER_MIN_MS; + + ret = adf_heartbeat_save_cfg_param(accel_dev, timer_ms); + if (ret) + return ret; + + ret = adf_heartbeat_ms_to_ticks(accel_dev, timer_ms, &ticks); + if (ret) + return ret; + + ret = adf_send_admin_hb_timer(accel_dev, ticks); + if (ret) + return ret; + + accel_dev->heartbeat->hb_timer = timer_ms; + } + + return written_chars; +} + +static const struct file_operations adf_hb_cfg_fops = { + .owner = THIS_MODULE, + .open = simple_open, + .read = adf_hb_cfg_read, + .write = adf_hb_cfg_write, +}; + +static ssize_t adf_hb_error_inject_write(struct file *file, + const char __user *user_buf, + size_t count, loff_t *ppos) +{ + struct adf_accel_dev *accel_dev = file->private_data; + char buf[3]; + int ret; + + /* last byte left as string termination */ + if (*ppos != 0 || count != 2) + return -EINVAL; + + if (copy_from_user(buf, user_buf, count)) + return -EFAULT; + buf[count] = '\0'; + + if (buf[0] != '1') + return -EINVAL; + + ret = adf_heartbeat_inject_error(accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Heartbeat error injection failed with status %d\n", + ret); + return ret; + } + + dev_info(&GET_DEV(accel_dev), "Heartbeat error injection enabled\n"); + + return count; +} + +static const struct file_operations adf_hb_error_inject_fops = { + .owner = THIS_MODULE, + .open = simple_open, + .write = adf_hb_error_inject_write, +}; + +void adf_heartbeat_dbgfs_add(struct adf_accel_dev *accel_dev) +{ + struct adf_heartbeat *hb = accel_dev->heartbeat; + + if (!hb) + return; + + hb->dbgfs.base_dir = debugfs_create_dir("heartbeat", accel_dev->debugfs_dir); + hb->dbgfs.status = debugfs_create_file("status", 0400, hb->dbgfs.base_dir, + accel_dev, &adf_hb_status_fops); + hb->dbgfs.sent = debugfs_create_file("queries_sent", 0400, hb->dbgfs.base_dir, + &hb->hb_sent_counter, &adf_hb_stats_fops); + hb->dbgfs.failed = debugfs_create_file("queries_failed", 0400, hb->dbgfs.base_dir, + &hb->hb_failed_counter, &adf_hb_stats_fops); + hb->dbgfs.cfg = debugfs_create_file("config", 0600, hb->dbgfs.base_dir, + accel_dev, &adf_hb_cfg_fops); + + if (IS_ENABLED(CONFIG_CRYPTO_DEV_QAT_ERROR_INJECTION)) { + struct dentry *inject_error __maybe_unused; + + inject_error = debugfs_create_file("inject_error", 0200, + hb->dbgfs.base_dir, accel_dev, + &adf_hb_error_inject_fops); +#ifdef CONFIG_CRYPTO_DEV_QAT_ERROR_INJECTION + hb->dbgfs.inject_error = inject_error; +#endif + } +} +EXPORT_SYMBOL_GPL(adf_heartbeat_dbgfs_add); + +void adf_heartbeat_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ + struct adf_heartbeat *hb = accel_dev->heartbeat; + + if (!hb) + return; + + debugfs_remove(hb->dbgfs.status); + hb->dbgfs.status = NULL; + debugfs_remove(hb->dbgfs.sent); + hb->dbgfs.sent = NULL; + debugfs_remove(hb->dbgfs.failed); + hb->dbgfs.failed = NULL; + debugfs_remove(hb->dbgfs.cfg); + hb->dbgfs.cfg = NULL; +#ifdef CONFIG_CRYPTO_DEV_QAT_ERROR_INJECTION + debugfs_remove(hb->dbgfs.inject_error); + hb->dbgfs.inject_error = NULL; +#endif + debugfs_remove(hb->dbgfs.base_dir); + hb->dbgfs.base_dir = NULL; +} +EXPORT_SYMBOL_GPL(adf_heartbeat_dbgfs_rm); diff --git a/drivers/crypto/intel/qat/qat_common/adf_heartbeat_dbgfs.h b/drivers/crypto/intel/qat/qat_common/adf_heartbeat_dbgfs.h new file mode 100644 index 000000000000..84dd29ea6454 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_heartbeat_dbgfs.h @@ -0,0 +1,12 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_HEARTBEAT_DBGFS_H_ +#define ADF_HEARTBEAT_DBGFS_H_ + +struct adf_accel_dev; + +void adf_heartbeat_dbgfs_add(struct adf_accel_dev *accel_dev); +void adf_heartbeat_dbgfs_rm(struct adf_accel_dev *accel_dev); + +#endif /* ADF_HEARTBEAT_DBGFS_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_heartbeat_inject.c b/drivers/crypto/intel/qat/qat_common/adf_heartbeat_inject.c new file mode 100644 index 000000000000..a3b474bdef6c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_heartbeat_inject.c @@ -0,0 +1,76 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include <linux/random.h> + +#include "adf_admin.h" +#include "adf_common_drv.h" +#include "adf_heartbeat.h" + +#define MAX_HB_TICKS 0xFFFFFFFF + +static int adf_hb_set_timer_to_max(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + + accel_dev->heartbeat->hb_timer = 0; + + if (hw_data->stop_timer) + hw_data->stop_timer(accel_dev); + + return adf_send_admin_hb_timer(accel_dev, MAX_HB_TICKS); +} + +static void adf_set_hb_counters_fail(struct adf_accel_dev *accel_dev, u32 ae, + u32 thr) +{ + struct hb_cnt_pair *stats = accel_dev->heartbeat->dma.virt_addr; + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + const size_t max_aes = hw_device->get_num_aes(hw_device); + const size_t hb_ctrs = hw_device->num_hb_ctrs; + size_t thr_id = ae * hb_ctrs + thr; + u16 num_rsp = stats[thr_id].resp_heartbeat_cnt; + + /* + * Inject live.req != live.rsp and live.rsp == last.rsp + * to trigger the heartbeat error detection + */ + stats[thr_id].req_heartbeat_cnt++; + stats += (max_aes * hb_ctrs); + stats[thr_id].resp_heartbeat_cnt = num_rsp; +} + +int adf_heartbeat_inject_error(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + const size_t max_aes = hw_device->get_num_aes(hw_device); + const size_t hb_ctrs = hw_device->num_hb_ctrs; + u32 rand, rand_ae, rand_thr; + unsigned long ae_mask; + int ret; + + ae_mask = hw_device->ae_mask; + + do { + /* Ensure we have a valid ae */ + get_random_bytes(&rand, sizeof(rand)); + rand_ae = rand % max_aes; + } while (!test_bit(rand_ae, &ae_mask)); + + get_random_bytes(&rand, sizeof(rand)); + rand_thr = rand % hb_ctrs; + + /* Increase the heartbeat timer to prevent FW updating HB counters */ + ret = adf_hb_set_timer_to_max(accel_dev); + if (ret) + return ret; + + /* Configure worker threads to stop processing any packet */ + ret = adf_disable_arb_thd(accel_dev, rand_ae, rand_thr); + if (ret) + return ret; + + /* Change HB counters memory to simulate a hang */ + adf_set_hb_counters_fail(accel_dev, rand_ae, rand_thr); + + return 0; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_hw_arbiter.c b/drivers/crypto/intel/qat/qat_common/adf_hw_arbiter.c new file mode 100644 index 000000000000..f93d9cca70ce --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_hw_arbiter.c @@ -0,0 +1,126 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_transport_internal.h" + +#define ADF_ARB_NUM 4 +#define ADF_ARB_REG_SIZE 0x4 + +#define WRITE_CSR_ARB_SARCONFIG(csr_addr, arb_offset, index, value) \ + ADF_CSR_WR(csr_addr, (arb_offset) + \ + (ADF_ARB_REG_SIZE * (index)), value) + +#define WRITE_CSR_ARB_WT2SAM(csr_addr, arb_offset, wt_offset, index, value) \ + ADF_CSR_WR(csr_addr, ((arb_offset) + (wt_offset)) + \ + (ADF_ARB_REG_SIZE * (index)), value) + +int adf_init_arb(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + void __iomem *csr = accel_dev->transport->banks[0].csr_addr; + unsigned long ae_mask = hw_data->ae_mask; + u32 arb_off, wt_off, arb_cfg; + const u32 *thd_2_arb_cfg; + struct arb_info info; + int arb, i; + + hw_data->get_arb_info(&info); + arb_cfg = info.arb_cfg; + arb_off = info.arb_offset; + wt_off = info.wt2sam_offset; + + /* Service arb configured for 32 bytes responses and + * ring flow control check enabled. */ + for (arb = 0; arb < ADF_ARB_NUM; arb++) + WRITE_CSR_ARB_SARCONFIG(csr, arb_off, arb, arb_cfg); + + /* Map worker threads to service arbiters */ + thd_2_arb_cfg = hw_data->get_arb_mapping(accel_dev); + + for_each_set_bit(i, &ae_mask, hw_data->num_engines) + WRITE_CSR_ARB_WT2SAM(csr, arb_off, wt_off, i, thd_2_arb_cfg[i]); + + return 0; +} +EXPORT_SYMBOL_GPL(adf_init_arb); + +void adf_update_ring_arb(struct adf_etr_ring_data *ring) +{ + struct adf_accel_dev *accel_dev = ring->bank->accel_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(accel_dev); + u32 tx_ring_mask = hw_data->tx_rings_mask; + u32 shift = hw_data->tx_rx_gap; + u32 arben, arben_tx, arben_rx; + u32 rx_ring_mask; + + /* + * Enable arbitration on a ring only if the TX half of the ring mask + * matches the RX part. This results in writes to CSR on both TX and + * RX update - only one is necessary, but both are done for + * simplicity. + */ + rx_ring_mask = tx_ring_mask << shift; + arben_tx = (ring->bank->ring_mask & tx_ring_mask) >> 0; + arben_rx = (ring->bank->ring_mask & rx_ring_mask) >> shift; + arben = arben_tx & arben_rx; + + csr_ops->write_csr_ring_srv_arb_en(ring->bank->csr_addr, + ring->bank->bank_number, arben); +} + +void adf_exit_arb(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(accel_dev); + u32 arb_off, wt_off; + struct arb_info info; + void __iomem *csr; + unsigned int i; + + hw_data->get_arb_info(&info); + arb_off = info.arb_offset; + wt_off = info.wt2sam_offset; + + if (!accel_dev->transport) + return; + + csr = accel_dev->transport->banks[0].csr_addr; + + hw_data->get_arb_info(&info); + + /* Unmap worker threads to service arbiters */ + for (i = 0; i < hw_data->num_engines; i++) + WRITE_CSR_ARB_WT2SAM(csr, arb_off, wt_off, i, 0); + + /* Disable arbitration on all rings */ + for (i = 0; i < GET_MAX_BANKS(accel_dev); i++) + csr_ops->write_csr_ring_srv_arb_en(csr, i, 0); +} +EXPORT_SYMBOL_GPL(adf_exit_arb); + +int adf_disable_arb_thd(struct adf_accel_dev *accel_dev, u32 ae, u32 thr) +{ + void __iomem *csr = accel_dev->transport->banks[0].csr_addr; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + const u32 *thd_2_arb_cfg; + struct arb_info info; + u32 ae_thr_map; + + if (ADF_AE_STRAND0_THREAD == thr || ADF_AE_STRAND1_THREAD == thr) + thr = ADF_AE_ADMIN_THREAD; + + hw_data->get_arb_info(&info); + thd_2_arb_cfg = hw_data->get_arb_mapping(accel_dev); + if (!thd_2_arb_cfg) + return -EFAULT; + + /* Disable scheduling for this particular AE and thread */ + ae_thr_map = *(thd_2_arb_cfg + ae); + ae_thr_map &= ~(GENMASK(3, 0) << (thr * BIT(2))); + + WRITE_CSR_ARB_WT2SAM(csr, info.arb_offset, info.wt2sam_offset, ae, + ae_thr_map); + return 0; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_init.c b/drivers/crypto/intel/qat/qat_common/adf_init.c new file mode 100644 index 000000000000..46491048e0bb --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_init.c @@ -0,0 +1,516 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/mutex.h> +#include <linux/list.h> +#include <linux/bitops.h> +#include <linux/delay.h> +#include "adf_accel_devices.h" +#include "adf_cfg.h" +#include "adf_common_drv.h" +#include "adf_dbgfs.h" +#include "adf_heartbeat.h" +#include "adf_rl.h" +#include "adf_sysfs_ras_counters.h" +#include "adf_telemetry.h" + +static LIST_HEAD(service_table); +static DEFINE_MUTEX(service_lock); + +static void adf_service_add(struct service_hndl *service) +{ + mutex_lock(&service_lock); + list_add(&service->list, &service_table); + mutex_unlock(&service_lock); +} + +int adf_service_register(struct service_hndl *service) +{ + memset(service->init_status, 0, sizeof(service->init_status)); + memset(service->start_status, 0, sizeof(service->start_status)); + adf_service_add(service); + return 0; +} + +static void adf_service_remove(struct service_hndl *service) +{ + mutex_lock(&service_lock); + list_del(&service->list); + mutex_unlock(&service_lock); +} + +int adf_service_unregister(struct service_hndl *service) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(service->init_status); i++) { + if (service->init_status[i] || service->start_status[i]) { + pr_err("QAT: Could not remove active service\n"); + return -EFAULT; + } + } + adf_service_remove(service); + return 0; +} + +/** + * adf_dev_init() - Init data structures and services for the given accel device + * @accel_dev: Pointer to acceleration device. + * + * Initialize the ring data structures and the admin comms and arbitration + * services. + * + * Return: 0 on success, error code otherwise. + */ +static int adf_dev_init(struct adf_accel_dev *accel_dev) +{ + struct service_hndl *service; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + int ret; + + if (!hw_data) { + dev_err(&GET_DEV(accel_dev), + "Failed to init device - hw_data not set\n"); + return -EFAULT; + } + + if (!test_bit(ADF_STATUS_CONFIGURED, &accel_dev->status) && + !accel_dev->is_vf) { + dev_err(&GET_DEV(accel_dev), "Device not configured\n"); + return -EFAULT; + } + + if (adf_init_etr_data(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed initialize etr\n"); + return -EFAULT; + } + + if (hw_data->init_device && hw_data->init_device(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed to initialize device\n"); + return -EFAULT; + } + + if (hw_data->init_admin_comms && hw_data->init_admin_comms(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed initialize admin comms\n"); + return -EFAULT; + } + + if (hw_data->init_arb && hw_data->init_arb(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed initialize hw arbiter\n"); + return -EFAULT; + } + + if (hw_data->get_ring_to_svc_map) + hw_data->ring_to_svc_map = hw_data->get_ring_to_svc_map(accel_dev); + + if (adf_ae_init(accel_dev)) { + dev_err(&GET_DEV(accel_dev), + "Failed to initialise Acceleration Engine\n"); + return -EFAULT; + } + set_bit(ADF_STATUS_AE_INITIALISED, &accel_dev->status); + + if (adf_ae_fw_load(accel_dev)) { + dev_err(&GET_DEV(accel_dev), + "Failed to load acceleration FW\n"); + return -EFAULT; + } + set_bit(ADF_STATUS_AE_UCODE_LOADED, &accel_dev->status); + + if (hw_data->alloc_irq(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed to allocate interrupts\n"); + return -EFAULT; + } + set_bit(ADF_STATUS_IRQ_ALLOCATED, &accel_dev->status); + + if (hw_data->ras_ops.enable_ras_errors) + hw_data->ras_ops.enable_ras_errors(accel_dev); + + hw_data->enable_ints(accel_dev); + hw_data->enable_error_correction(accel_dev); + + ret = hw_data->pfvf_ops.enable_comms(accel_dev); + if (ret) + return ret; + + if (!test_bit(ADF_STATUS_CONFIGURED, &accel_dev->status) && + accel_dev->is_vf) { + if (qat_crypto_vf_dev_config(accel_dev)) + return -EFAULT; + } + + adf_heartbeat_init(accel_dev); + ret = adf_rl_init(accel_dev); + if (ret && ret != -EOPNOTSUPP) + return ret; + + ret = adf_tl_init(accel_dev); + if (ret && ret != -EOPNOTSUPP) + return ret; + + /* + * Subservice initialisation is divided into two stages: init and start. + * This is to facilitate any ordering dependencies between services + * prior to starting any of the accelerators. + */ + list_for_each_entry(service, &service_table, list) { + if (service->event_hld(accel_dev, ADF_EVENT_INIT)) { + dev_err(&GET_DEV(accel_dev), + "Failed to initialise service %s\n", + service->name); + return -EFAULT; + } + set_bit(accel_dev->accel_id, service->init_status); + } + + return 0; +} + +/** + * adf_dev_start() - Start acceleration service for the given accel device + * @accel_dev: Pointer to acceleration device. + * + * Function notifies all the registered services that the acceleration device + * is ready to be used. + * To be used by QAT device specific drivers. + * + * Return: 0 on success, error code otherwise. + */ +static int adf_dev_start(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct service_hndl *service; + int ret; + + set_bit(ADF_STATUS_STARTING, &accel_dev->status); + + if (adf_ae_start(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "AE Start Failed\n"); + return -EFAULT; + } + set_bit(ADF_STATUS_AE_STARTED, &accel_dev->status); + + if (hw_data->send_admin_init(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed to send init message\n"); + return -EFAULT; + } + + if (hw_data->measure_clock) { + ret = hw_data->measure_clock(accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Failed measure device clock\n"); + return ret; + } + } + + /* Set ssm watch dog timer */ + if (hw_data->set_ssm_wdtimer) + hw_data->set_ssm_wdtimer(accel_dev); + + /* Enable Power Management */ + if (hw_data->enable_pm && hw_data->enable_pm(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Failed to configure Power Management\n"); + return -EFAULT; + } + + if (hw_data->start_timer) { + ret = hw_data->start_timer(accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Failed to start internal sync timer\n"); + return ret; + } + } + + adf_heartbeat_start(accel_dev); + ret = adf_rl_start(accel_dev); + if (ret && ret != -EOPNOTSUPP) + return ret; + + ret = adf_tl_start(accel_dev); + if (ret && ret != -EOPNOTSUPP) + return ret; + + list_for_each_entry(service, &service_table, list) { + if (service->event_hld(accel_dev, ADF_EVENT_START)) { + dev_err(&GET_DEV(accel_dev), + "Failed to start service %s\n", + service->name); + return -EFAULT; + } + set_bit(accel_dev->accel_id, service->start_status); + } + + clear_bit(ADF_STATUS_STARTING, &accel_dev->status); + set_bit(ADF_STATUS_STARTED, &accel_dev->status); + + if (!list_empty(&accel_dev->crypto_list) && + (qat_algs_register() || qat_asym_algs_register())) { + dev_err(&GET_DEV(accel_dev), + "Failed to register crypto algs\n"); + set_bit(ADF_STATUS_STARTING, &accel_dev->status); + clear_bit(ADF_STATUS_STARTED, &accel_dev->status); + return -EFAULT; + } + set_bit(ADF_STATUS_CRYPTO_ALGS_REGISTERED, &accel_dev->status); + + if (!list_empty(&accel_dev->compression_list) && qat_comp_algs_register()) { + dev_err(&GET_DEV(accel_dev), + "Failed to register compression algs\n"); + set_bit(ADF_STATUS_STARTING, &accel_dev->status); + clear_bit(ADF_STATUS_STARTED, &accel_dev->status); + return -EFAULT; + } + set_bit(ADF_STATUS_COMP_ALGS_REGISTERED, &accel_dev->status); + + adf_dbgfs_add(accel_dev); + adf_sysfs_start_ras(accel_dev); + + return 0; +} + +/** + * adf_dev_stop() - Stop acceleration service for the given accel device + * @accel_dev: Pointer to acceleration device. + * + * Function notifies all the registered services that the acceleration device + * is shuting down. + * To be used by QAT device specific drivers. + * + * Return: void + */ +static void adf_dev_stop(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct service_hndl *service; + bool wait = false; + int ret; + + if (!adf_dev_started(accel_dev) && + !test_bit(ADF_STATUS_STARTING, &accel_dev->status)) + return; + + adf_tl_stop(accel_dev); + adf_rl_stop(accel_dev); + adf_dbgfs_rm(accel_dev); + adf_sysfs_stop_ras(accel_dev); + + clear_bit(ADF_STATUS_STARTING, &accel_dev->status); + clear_bit(ADF_STATUS_STARTED, &accel_dev->status); + + if (!list_empty(&accel_dev->crypto_list) && + test_bit(ADF_STATUS_CRYPTO_ALGS_REGISTERED, &accel_dev->status)) { + qat_algs_unregister(); + qat_asym_algs_unregister(); + } + clear_bit(ADF_STATUS_CRYPTO_ALGS_REGISTERED, &accel_dev->status); + + if (!list_empty(&accel_dev->compression_list) && + test_bit(ADF_STATUS_COMP_ALGS_REGISTERED, &accel_dev->status)) + qat_comp_algs_unregister(); + clear_bit(ADF_STATUS_COMP_ALGS_REGISTERED, &accel_dev->status); + + list_for_each_entry(service, &service_table, list) { + if (!test_bit(accel_dev->accel_id, service->start_status)) + continue; + ret = service->event_hld(accel_dev, ADF_EVENT_STOP); + if (!ret) { + clear_bit(accel_dev->accel_id, service->start_status); + } else if (ret == -EAGAIN) { + wait = true; + clear_bit(accel_dev->accel_id, service->start_status); + } + } + + if (hw_data->stop_timer) + hw_data->stop_timer(accel_dev); + + hw_data->disable_iov(accel_dev); + + if (wait) + msleep(100); + + if (test_bit(ADF_STATUS_AE_STARTED, &accel_dev->status)) { + if (adf_ae_stop(accel_dev)) + dev_err(&GET_DEV(accel_dev), "failed to stop AE\n"); + else + clear_bit(ADF_STATUS_AE_STARTED, &accel_dev->status); + } +} + +/** + * adf_dev_shutdown() - shutdown acceleration services and data strucutures + * @accel_dev: Pointer to acceleration device + * + * Cleanup the ring data structures and the admin comms and arbitration + * services. + */ +static void adf_dev_shutdown(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct service_hndl *service; + + if (!hw_data) { + dev_err(&GET_DEV(accel_dev), + "QAT: Failed to shutdown device - hw_data not set\n"); + return; + } + + if (test_bit(ADF_STATUS_AE_UCODE_LOADED, &accel_dev->status)) { + adf_ae_fw_release(accel_dev); + clear_bit(ADF_STATUS_AE_UCODE_LOADED, &accel_dev->status); + } + + if (test_bit(ADF_STATUS_AE_INITIALISED, &accel_dev->status)) { + if (adf_ae_shutdown(accel_dev)) + dev_err(&GET_DEV(accel_dev), + "Failed to shutdown Accel Engine\n"); + else + clear_bit(ADF_STATUS_AE_INITIALISED, + &accel_dev->status); + } + + list_for_each_entry(service, &service_table, list) { + if (!test_bit(accel_dev->accel_id, service->init_status)) + continue; + if (service->event_hld(accel_dev, ADF_EVENT_SHUTDOWN)) + dev_err(&GET_DEV(accel_dev), + "Failed to shutdown service %s\n", + service->name); + else + clear_bit(accel_dev->accel_id, service->init_status); + } + + adf_rl_exit(accel_dev); + + if (hw_data->ras_ops.disable_ras_errors) + hw_data->ras_ops.disable_ras_errors(accel_dev); + + adf_heartbeat_shutdown(accel_dev); + + adf_tl_shutdown(accel_dev); + + if (test_bit(ADF_STATUS_IRQ_ALLOCATED, &accel_dev->status)) { + hw_data->free_irq(accel_dev); + clear_bit(ADF_STATUS_IRQ_ALLOCATED, &accel_dev->status); + } + + /* If not restarting, delete all cfg sections except for GENERAL */ + if (!test_bit(ADF_STATUS_RESTARTING, &accel_dev->status)) + adf_cfg_del_all_except(accel_dev, ADF_GENERAL_SEC); + + if (hw_data->exit_arb) + hw_data->exit_arb(accel_dev); + + if (hw_data->exit_admin_comms) + hw_data->exit_admin_comms(accel_dev); + + adf_cleanup_etr_data(accel_dev); + adf_misc_wq_flush(); + adf_dev_restore(accel_dev); +} + +int adf_dev_restarting_notify(struct adf_accel_dev *accel_dev) +{ + struct service_hndl *service; + + list_for_each_entry(service, &service_table, list) { + if (service->event_hld(accel_dev, ADF_EVENT_RESTARTING)) + dev_err(&GET_DEV(accel_dev), + "Failed to restart service %s.\n", + service->name); + } + return 0; +} + +int adf_dev_restarted_notify(struct adf_accel_dev *accel_dev) +{ + struct service_hndl *service; + + list_for_each_entry(service, &service_table, list) { + if (service->event_hld(accel_dev, ADF_EVENT_RESTARTED)) + dev_err(&GET_DEV(accel_dev), + "Failed to restart service %s.\n", + service->name); + } + return 0; +} + +void adf_error_notifier(struct adf_accel_dev *accel_dev) +{ + struct service_hndl *service; + + list_for_each_entry(service, &service_table, list) { + if (service->event_hld(accel_dev, ADF_EVENT_FATAL_ERROR)) + dev_err(&GET_DEV(accel_dev), + "Failed to send error event to %s.\n", + service->name); + } +} + +int adf_dev_down(struct adf_accel_dev *accel_dev) +{ + int ret = 0; + + if (!accel_dev) + return -EINVAL; + + mutex_lock(&accel_dev->state_lock); + + adf_dev_stop(accel_dev); + adf_dev_shutdown(accel_dev); + + mutex_unlock(&accel_dev->state_lock); + return ret; +} +EXPORT_SYMBOL_GPL(adf_dev_down); + +int adf_dev_up(struct adf_accel_dev *accel_dev, bool config) +{ + int ret = 0; + + if (!accel_dev) + return -EINVAL; + + mutex_lock(&accel_dev->state_lock); + + if (adf_dev_started(accel_dev)) { + dev_info(&GET_DEV(accel_dev), "Device qat_dev%d already up\n", + accel_dev->accel_id); + ret = -EALREADY; + goto out; + } + + if (config && GET_HW_DATA(accel_dev)->dev_config) { + ret = GET_HW_DATA(accel_dev)->dev_config(accel_dev); + if (unlikely(ret)) + goto out; + } + + ret = adf_dev_init(accel_dev); + if (unlikely(ret)) + goto out; + + ret = adf_dev_start(accel_dev); + +out: + mutex_unlock(&accel_dev->state_lock); + return ret; +} +EXPORT_SYMBOL_GPL(adf_dev_up); + +int adf_dev_restart(struct adf_accel_dev *accel_dev) +{ + int ret = 0; + + if (!accel_dev) + return -EFAULT; + + adf_dev_down(accel_dev); + + ret = adf_dev_up(accel_dev, false); + /* if device is already up return success*/ + if (ret == -EALREADY) + return 0; + + return ret; +} +EXPORT_SYMBOL_GPL(adf_dev_restart); diff --git a/drivers/crypto/intel/qat/qat_common/adf_isr.c b/drivers/crypto/intel/qat/qat_common/adf_isr.c new file mode 100644 index 000000000000..4639d7fd93e6 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_isr.c @@ -0,0 +1,415 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/pci.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/interrupt.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_cfg.h" +#include "adf_cfg_strings.h" +#include "adf_cfg_common.h" +#include "adf_transport_access_macros.h" +#include "adf_transport_internal.h" + +#define ADF_MAX_NUM_VFS 32 +static struct workqueue_struct *adf_misc_wq; + +static int adf_enable_msix(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *pci_dev_info = &accel_dev->accel_pci_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u32 msix_num_entries = hw_data->num_banks + 1; + int ret; + + if (hw_data->set_msix_rttable) + hw_data->set_msix_rttable(accel_dev); + + ret = pci_alloc_irq_vectors(pci_dev_info->pci_dev, msix_num_entries, + msix_num_entries, PCI_IRQ_MSIX); + if (unlikely(ret < 0)) { + dev_err(&GET_DEV(accel_dev), + "Failed to allocate %d MSI-X vectors\n", + msix_num_entries); + return ret; + } + return 0; +} + +static void adf_disable_msix(struct adf_accel_pci *pci_dev_info) +{ + pci_free_irq_vectors(pci_dev_info->pci_dev); +} + +static irqreturn_t adf_msix_isr_bundle(int irq, void *bank_ptr) +{ + struct adf_etr_bank_data *bank = bank_ptr; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(bank->accel_dev); + + csr_ops->write_csr_int_flag_and_col(bank->csr_addr, bank->bank_number, + 0); + tasklet_hi_schedule(&bank->resp_handler); + return IRQ_HANDLED; +} + +#ifdef CONFIG_PCI_IOV +void adf_enable_vf2pf_interrupts(struct adf_accel_dev *accel_dev, u32 vf_mask) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + unsigned long flags; + + spin_lock_irqsave(&accel_dev->pf.vf2pf_ints_lock, flags); + GET_PFVF_OPS(accel_dev)->enable_vf2pf_interrupts(pmisc_addr, vf_mask); + spin_unlock_irqrestore(&accel_dev->pf.vf2pf_ints_lock, flags); +} + +void adf_disable_all_vf2pf_interrupts(struct adf_accel_dev *accel_dev) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + unsigned long flags; + + spin_lock_irqsave(&accel_dev->pf.vf2pf_ints_lock, flags); + GET_PFVF_OPS(accel_dev)->disable_all_vf2pf_interrupts(pmisc_addr); + spin_unlock_irqrestore(&accel_dev->pf.vf2pf_ints_lock, flags); +} + +static u32 adf_disable_pending_vf2pf_interrupts(struct adf_accel_dev *accel_dev) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u32 pending; + + spin_lock(&accel_dev->pf.vf2pf_ints_lock); + pending = GET_PFVF_OPS(accel_dev)->disable_pending_vf2pf_interrupts(pmisc_addr); + spin_unlock(&accel_dev->pf.vf2pf_ints_lock); + + return pending; +} + +static bool adf_handle_vf2pf_int(struct adf_accel_dev *accel_dev) +{ + bool irq_handled = false; + unsigned long vf_mask; + + /* Get the interrupt sources triggered by VFs, except for those already disabled */ + vf_mask = adf_disable_pending_vf2pf_interrupts(accel_dev); + if (vf_mask) { + struct adf_accel_vf_info *vf_info; + int i; + + /* + * Handle VF2PF interrupt unless the VF is malicious and + * is attempting to flood the host OS with VF2PF interrupts. + */ + for_each_set_bit(i, &vf_mask, ADF_MAX_NUM_VFS) { + vf_info = accel_dev->pf.vf_info + i; + + if (!__ratelimit(&vf_info->vf2pf_ratelimit)) { + dev_info(&GET_DEV(accel_dev), + "Too many ints from VF%d\n", + vf_info->vf_nr); + continue; + } + + adf_schedule_vf2pf_handler(vf_info); + irq_handled = true; + } + } + return irq_handled; +} +#endif /* CONFIG_PCI_IOV */ + +static bool adf_handle_pm_int(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + + if (hw_data->handle_pm_interrupt && + hw_data->handle_pm_interrupt(accel_dev)) + return true; + + return false; +} + +static bool adf_handle_ras_int(struct adf_accel_dev *accel_dev) +{ + struct adf_ras_ops *ras_ops = &accel_dev->hw_device->ras_ops; + bool reset_required; + + if (ras_ops->handle_interrupt && + ras_ops->handle_interrupt(accel_dev, &reset_required)) { + if (reset_required) { + dev_err(&GET_DEV(accel_dev), "Fatal error, reset required\n"); + if (adf_notify_fatal_error(accel_dev)) + dev_err(&GET_DEV(accel_dev), + "Failed to notify fatal error\n"); + } + + return true; + } + + return false; +} + +static irqreturn_t adf_msix_isr_ae(int irq, void *dev_ptr) +{ + struct adf_accel_dev *accel_dev = dev_ptr; + +#ifdef CONFIG_PCI_IOV + /* If SR-IOV is enabled (vf_info is non-NULL), check for VF->PF ints */ + if (accel_dev->pf.vf_info && adf_handle_vf2pf_int(accel_dev)) + return IRQ_HANDLED; +#endif /* CONFIG_PCI_IOV */ + + if (adf_handle_pm_int(accel_dev)) + return IRQ_HANDLED; + + if (adf_handle_ras_int(accel_dev)) + return IRQ_HANDLED; + + dev_dbg(&GET_DEV(accel_dev), "qat_dev%d spurious AE interrupt\n", + accel_dev->accel_id); + + return IRQ_NONE; +} + +static void adf_free_irqs(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *pci_dev_info = &accel_dev->accel_pci_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_irq *irqs = pci_dev_info->msix_entries.irqs; + struct adf_etr_data *etr_data = accel_dev->transport; + int clust_irq = hw_data->num_banks; + int irq, i = 0; + + if (pci_dev_info->msix_entries.num_entries > 1) { + for (i = 0; i < hw_data->num_banks; i++) { + if (irqs[i].enabled) { + irq = pci_irq_vector(pci_dev_info->pci_dev, i); + irq_set_affinity_hint(irq, NULL); + free_irq(irq, &etr_data->banks[i]); + } + } + } + + if (irqs[i].enabled) { + irq = pci_irq_vector(pci_dev_info->pci_dev, clust_irq); + free_irq(irq, accel_dev); + } +} + +static int adf_request_irqs(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *pci_dev_info = &accel_dev->accel_pci_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_irq *irqs = pci_dev_info->msix_entries.irqs; + struct adf_etr_data *etr_data = accel_dev->transport; + int clust_irq = hw_data->num_banks; + int ret, irq, i = 0; + char *name; + + /* Request msix irq for all banks unless SR-IOV enabled */ + if (!accel_dev->pf.vf_info) { + for (i = 0; i < hw_data->num_banks; i++) { + struct adf_etr_bank_data *bank = &etr_data->banks[i]; + unsigned int cpu, cpus = num_online_cpus(); + + name = irqs[i].name; + snprintf(name, ADF_MAX_MSIX_VECTOR_NAME, + "qat%d-bundle%d", accel_dev->accel_id, i); + irq = pci_irq_vector(pci_dev_info->pci_dev, i); + if (unlikely(irq < 0)) { + dev_err(&GET_DEV(accel_dev), + "Failed to get IRQ number of device vector %d - %s\n", + i, name); + ret = irq; + goto err; + } + ret = request_irq(irq, adf_msix_isr_bundle, 0, + &name[0], bank); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to allocate IRQ %d for %s\n", + irq, name); + goto err; + } + + cpu = ((accel_dev->accel_id * hw_data->num_banks) + + i) % cpus; + irq_set_affinity_hint(irq, get_cpu_mask(cpu)); + irqs[i].enabled = true; + } + } + + /* Request msix irq for AE */ + name = irqs[i].name; + snprintf(name, ADF_MAX_MSIX_VECTOR_NAME, + "qat%d-ae-cluster", accel_dev->accel_id); + irq = pci_irq_vector(pci_dev_info->pci_dev, clust_irq); + if (unlikely(irq < 0)) { + dev_err(&GET_DEV(accel_dev), + "Failed to get IRQ number of device vector %d - %s\n", + i, name); + ret = irq; + goto err; + } + ret = request_irq(irq, adf_msix_isr_ae, 0, &name[0], accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to allocate IRQ %d for %s\n", irq, name); + goto err; + } + irqs[i].enabled = true; + return ret; +err: + adf_free_irqs(accel_dev); + return ret; +} + +static int adf_isr_alloc_msix_vectors_data(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u32 msix_num_entries = 1; + struct adf_irq *irqs; + + /* If SR-IOV is disabled (vf_info is NULL), add entries for each bank */ + if (!accel_dev->pf.vf_info) + msix_num_entries += hw_data->num_banks; + + irqs = kcalloc_node(msix_num_entries, sizeof(*irqs), + GFP_KERNEL, dev_to_node(&GET_DEV(accel_dev))); + if (!irqs) + return -ENOMEM; + + accel_dev->accel_pci_dev.msix_entries.num_entries = msix_num_entries; + accel_dev->accel_pci_dev.msix_entries.irqs = irqs; + return 0; +} + +static void adf_isr_free_msix_vectors_data(struct adf_accel_dev *accel_dev) +{ + kfree(accel_dev->accel_pci_dev.msix_entries.irqs); + accel_dev->accel_pci_dev.msix_entries.irqs = NULL; +} + +static int adf_setup_bh(struct adf_accel_dev *accel_dev) +{ + struct adf_etr_data *priv_data = accel_dev->transport; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + int i; + + for (i = 0; i < hw_data->num_banks; i++) + tasklet_init(&priv_data->banks[i].resp_handler, + adf_response_handler, + (unsigned long)&priv_data->banks[i]); + return 0; +} + +static void adf_cleanup_bh(struct adf_accel_dev *accel_dev) +{ + struct adf_etr_data *priv_data = accel_dev->transport; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + int i; + + for (i = 0; i < hw_data->num_banks; i++) { + tasklet_disable(&priv_data->banks[i].resp_handler); + tasklet_kill(&priv_data->banks[i].resp_handler); + } +} + +/** + * adf_isr_resource_free() - Free IRQ for acceleration device + * @accel_dev: Pointer to acceleration device. + * + * Function frees interrupts for acceleration device. + */ +void adf_isr_resource_free(struct adf_accel_dev *accel_dev) +{ + adf_free_irqs(accel_dev); + adf_cleanup_bh(accel_dev); + adf_disable_msix(&accel_dev->accel_pci_dev); + adf_isr_free_msix_vectors_data(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_isr_resource_free); + +/** + * adf_isr_resource_alloc() - Allocate IRQ for acceleration device + * @accel_dev: Pointer to acceleration device. + * + * Function allocates interrupts for acceleration device. + * + * Return: 0 on success, error code otherwise. + */ +int adf_isr_resource_alloc(struct adf_accel_dev *accel_dev) +{ + int ret; + + ret = adf_isr_alloc_msix_vectors_data(accel_dev); + if (ret) + goto err_out; + + ret = adf_enable_msix(accel_dev); + if (ret) + goto err_free_msix_table; + + ret = adf_setup_bh(accel_dev); + if (ret) + goto err_disable_msix; + + ret = adf_request_irqs(accel_dev); + if (ret) + goto err_cleanup_bh; + + return 0; + +err_cleanup_bh: + adf_cleanup_bh(accel_dev); + +err_disable_msix: + adf_disable_msix(&accel_dev->accel_pci_dev); + +err_free_msix_table: + adf_isr_free_msix_vectors_data(accel_dev); + +err_out: + return ret; +} +EXPORT_SYMBOL_GPL(adf_isr_resource_alloc); + +/** + * adf_init_misc_wq() - Init misc workqueue + * + * Return: 0 on success, error code otherwise. + */ +int __init adf_init_misc_wq(void) +{ + adf_misc_wq = alloc_workqueue("qat_misc_wq", + WQ_MEM_RECLAIM | WQ_PERCPU, 0); + + return !adf_misc_wq ? -ENOMEM : 0; +} + +void adf_exit_misc_wq(void) +{ + if (adf_misc_wq) + destroy_workqueue(adf_misc_wq); + + adf_misc_wq = NULL; +} + +bool adf_misc_wq_queue_work(struct work_struct *work) +{ + return queue_work(adf_misc_wq, work); +} + +bool adf_misc_wq_queue_delayed_work(struct delayed_work *work, + unsigned long delay) +{ + return queue_delayed_work(adf_misc_wq, work, delay); +} + +void adf_misc_wq_flush(void) +{ + flush_workqueue(adf_misc_wq); +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_mstate_mgr.c b/drivers/crypto/intel/qat/qat_common/adf_mstate_mgr.c new file mode 100644 index 000000000000..41cc763a74aa --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_mstate_mgr.c @@ -0,0 +1,318 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2024 Intel Corporation */ + +#include <linux/slab.h> +#include <linux/types.h> +#include "adf_mstate_mgr.h" + +#define ADF_MSTATE_MAGIC 0xADF5CAEA +#define ADF_MSTATE_VERSION 0x1 + +struct adf_mstate_sect_h { + u8 id[ADF_MSTATE_ID_LEN]; + u32 size; + u32 sub_sects; + u8 state[]; +}; + +u32 adf_mstate_state_size(struct adf_mstate_mgr *mgr) +{ + return mgr->state - mgr->buf; +} + +static inline u32 adf_mstate_avail_room(struct adf_mstate_mgr *mgr) +{ + return mgr->buf + mgr->size - mgr->state; +} + +void adf_mstate_mgr_init(struct adf_mstate_mgr *mgr, u8 *buf, u32 size) +{ + mgr->buf = buf; + mgr->state = buf; + mgr->size = size; + mgr->n_sects = 0; +}; + +struct adf_mstate_mgr *adf_mstate_mgr_new(u8 *buf, u32 size) +{ + struct adf_mstate_mgr *mgr; + + mgr = kzalloc(sizeof(*mgr), GFP_KERNEL); + if (!mgr) + return NULL; + + adf_mstate_mgr_init(mgr, buf, size); + + return mgr; +} + +void adf_mstate_mgr_destroy(struct adf_mstate_mgr *mgr) +{ + kfree(mgr); +} + +void adf_mstate_mgr_init_from_parent(struct adf_mstate_mgr *mgr, + struct adf_mstate_mgr *p_mgr) +{ + adf_mstate_mgr_init(mgr, p_mgr->state, + p_mgr->size - adf_mstate_state_size(p_mgr)); +} + +void adf_mstate_mgr_init_from_psect(struct adf_mstate_mgr *mgr, + struct adf_mstate_sect_h *p_sect) +{ + adf_mstate_mgr_init(mgr, p_sect->state, p_sect->size); + mgr->n_sects = p_sect->sub_sects; +} + +static void adf_mstate_preamble_init(struct adf_mstate_preh *preamble) +{ + preamble->magic = ADF_MSTATE_MAGIC; + preamble->version = ADF_MSTATE_VERSION; + preamble->preh_len = sizeof(*preamble); + preamble->size = 0; + preamble->n_sects = 0; +} + +/* default preambles checker */ +static int adf_mstate_preamble_def_checker(struct adf_mstate_preh *preamble, + void *opaque) +{ + struct adf_mstate_mgr *mgr = opaque; + + if (preamble->magic != ADF_MSTATE_MAGIC || + preamble->version > ADF_MSTATE_VERSION || + preamble->preh_len > mgr->size) { + pr_debug("QAT: LM - Invalid state (magic=%#x, version=%#x, hlen=%u), state_size=%u\n", + preamble->magic, preamble->version, preamble->preh_len, + mgr->size); + return -EINVAL; + } + + return 0; +} + +struct adf_mstate_preh *adf_mstate_preamble_add(struct adf_mstate_mgr *mgr) +{ + struct adf_mstate_preh *pre = (struct adf_mstate_preh *)mgr->buf; + + if (adf_mstate_avail_room(mgr) < sizeof(*pre)) { + pr_err("QAT: LM - Not enough space for preamble\n"); + return NULL; + } + + adf_mstate_preamble_init(pre); + mgr->state += pre->preh_len; + + return pre; +} + +int adf_mstate_preamble_update(struct adf_mstate_mgr *mgr) +{ + struct adf_mstate_preh *preamble = (struct adf_mstate_preh *)mgr->buf; + + preamble->size = adf_mstate_state_size(mgr) - preamble->preh_len; + preamble->n_sects = mgr->n_sects; + + return 0; +} + +static void adf_mstate_dump_sect(struct adf_mstate_sect_h *sect, + const char *prefix) +{ + pr_debug("QAT: LM - %s QAT state section %s\n", prefix, sect->id); + print_hex_dump_debug("h-", DUMP_PREFIX_OFFSET, 16, 2, sect, + sizeof(*sect), true); + print_hex_dump_debug("s-", DUMP_PREFIX_OFFSET, 16, 2, sect->state, + sect->size, true); +} + +static inline void __adf_mstate_sect_update(struct adf_mstate_mgr *mgr, + struct adf_mstate_sect_h *sect, + u32 size, + u32 n_subsects) +{ + sect->size += size; + sect->sub_sects += n_subsects; + mgr->n_sects++; + mgr->state += sect->size; + + adf_mstate_dump_sect(sect, "Add"); +} + +void adf_mstate_sect_update(struct adf_mstate_mgr *p_mgr, + struct adf_mstate_mgr *curr_mgr, + struct adf_mstate_sect_h *sect) +{ + __adf_mstate_sect_update(p_mgr, sect, adf_mstate_state_size(curr_mgr), + curr_mgr->n_sects); +} + +static struct adf_mstate_sect_h *adf_mstate_sect_add_header(struct adf_mstate_mgr *mgr, + const char *id) +{ + struct adf_mstate_sect_h *sect = (struct adf_mstate_sect_h *)(mgr->state); + + if (adf_mstate_avail_room(mgr) < sizeof(*sect)) { + pr_debug("QAT: LM - Not enough space for header of QAT state sect %s\n", id); + return NULL; + } + + strscpy(sect->id, id, sizeof(sect->id)); + sect->size = 0; + sect->sub_sects = 0; + mgr->state += sizeof(*sect); + + return sect; +} + +struct adf_mstate_sect_h *adf_mstate_sect_add_vreg(struct adf_mstate_mgr *mgr, + const char *id, + struct adf_mstate_vreginfo *info) +{ + struct adf_mstate_sect_h *sect; + + sect = adf_mstate_sect_add_header(mgr, id); + if (!sect) + return NULL; + + if (adf_mstate_avail_room(mgr) < info->size) { + pr_debug("QAT: LM - Not enough space for QAT state sect %s, requires %u\n", + id, info->size); + return NULL; + } + + memcpy(sect->state, info->addr, info->size); + __adf_mstate_sect_update(mgr, sect, info->size, 0); + + return sect; +} + +struct adf_mstate_sect_h *adf_mstate_sect_add(struct adf_mstate_mgr *mgr, + const char *id, + adf_mstate_populate populate, + void *opaque) +{ + struct adf_mstate_mgr sub_sects_mgr; + struct adf_mstate_sect_h *sect; + int avail_room, size; + + sect = adf_mstate_sect_add_header(mgr, id); + if (!sect) + return NULL; + + if (!populate) + return sect; + + avail_room = adf_mstate_avail_room(mgr); + adf_mstate_mgr_init_from_parent(&sub_sects_mgr, mgr); + + size = (*populate)(&sub_sects_mgr, sect->state, avail_room, opaque); + if (size < 0) + return NULL; + + size += adf_mstate_state_size(&sub_sects_mgr); + if (avail_room < size) { + pr_debug("QAT: LM - Not enough space for QAT state sect %s, requires %u\n", + id, size); + return NULL; + } + __adf_mstate_sect_update(mgr, sect, size, sub_sects_mgr.n_sects); + + return sect; +} + +static int adf_mstate_sect_validate(struct adf_mstate_mgr *mgr) +{ + struct adf_mstate_sect_h *start = (struct adf_mstate_sect_h *)mgr->state; + struct adf_mstate_sect_h *sect = start; + u64 end; + int i; + + end = (uintptr_t)mgr->buf + mgr->size; + for (i = 0; i < mgr->n_sects; i++) { + uintptr_t s_start = (uintptr_t)sect->state; + uintptr_t s_end = s_start + sect->size; + + if (s_end < s_start || s_end > end) { + pr_debug("QAT: LM - Corrupted state section (index=%u, size=%u) in state_mgr (size=%u, secs=%u)\n", + i, sect->size, mgr->size, mgr->n_sects); + return -EINVAL; + } + sect = (struct adf_mstate_sect_h *)s_end; + } + + pr_debug("QAT: LM - Scanned section (last child=%s, size=%lu) in state_mgr (size=%u, secs=%u)\n", + start->id, sizeof(struct adf_mstate_sect_h) * (ulong)(sect - start), + mgr->size, mgr->n_sects); + + return 0; +} + +u32 adf_mstate_state_size_from_remote(struct adf_mstate_mgr *mgr) +{ + struct adf_mstate_preh *preh = (struct adf_mstate_preh *)mgr->buf; + + return preh->preh_len + preh->size; +} + +int adf_mstate_mgr_init_from_remote(struct adf_mstate_mgr *mgr, u8 *buf, u32 size, + adf_mstate_preamble_checker pre_checker, + void *opaque) +{ + struct adf_mstate_preh *pre; + int ret; + + adf_mstate_mgr_init(mgr, buf, size); + pre = (struct adf_mstate_preh *)(mgr->buf); + + pr_debug("QAT: LM - Dump state preambles\n"); + print_hex_dump_debug("", DUMP_PREFIX_OFFSET, 16, 2, pre, pre->preh_len, 0); + + if (pre_checker) + ret = (*pre_checker)(pre, opaque); + else + ret = adf_mstate_preamble_def_checker(pre, mgr); + if (ret) + return ret; + + mgr->state = mgr->buf + pre->preh_len; + mgr->n_sects = pre->n_sects; + + return adf_mstate_sect_validate(mgr); +} + +struct adf_mstate_sect_h *adf_mstate_sect_lookup(struct adf_mstate_mgr *mgr, + const char *id, + adf_mstate_action action, + void *opaque) +{ + struct adf_mstate_sect_h *sect = (struct adf_mstate_sect_h *)mgr->state; + struct adf_mstate_mgr sub_sects_mgr; + int i, ret; + + for (i = 0; i < mgr->n_sects; i++) { + if (!strncmp(sect->id, id, sizeof(sect->id))) + goto found; + + sect = (struct adf_mstate_sect_h *)(sect->state + sect->size); + } + + return NULL; + +found: + adf_mstate_dump_sect(sect, "Found"); + + adf_mstate_mgr_init_from_psect(&sub_sects_mgr, sect); + if (sect->sub_sects && adf_mstate_sect_validate(&sub_sects_mgr)) + return NULL; + + if (!action) + return sect; + + ret = (*action)(&sub_sects_mgr, sect->state, sect->size, opaque); + if (ret) + return NULL; + + return sect; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_mstate_mgr.h b/drivers/crypto/intel/qat/qat_common/adf_mstate_mgr.h new file mode 100644 index 000000000000..81d263a596c5 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_mstate_mgr.h @@ -0,0 +1,89 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2024 Intel Corporation */ + +#ifndef ADF_MSTATE_MGR_H +#define ADF_MSTATE_MGR_H + +#define ADF_MSTATE_ID_LEN 8 + +#define ADF_MSTATE_ETRB_IDS "ETRBAR" +#define ADF_MSTATE_MISCB_IDS "MISCBAR" +#define ADF_MSTATE_EXTB_IDS "EXTBAR" +#define ADF_MSTATE_GEN_IDS "GENER" +#define ADF_MSTATE_CONFIG_IDS "CONFIG" +#define ADF_MSTATE_SECTION_NUM 5 + +#define ADF_MSTATE_BANK_IDX_IDS "bnk" + +#define ADF_MSTATE_ETR_REGS_IDS "mregs" +#define ADF_MSTATE_VINTSRC_IDS "visrc" +#define ADF_MSTATE_VINTMSK_IDS "vimsk" +#define ADF_MSTATE_SLA_IDS "sla" +#define ADF_MSTATE_IOV_INIT_IDS "iovinit" +#define ADF_MSTATE_COMPAT_VER_IDS "compver" +#define ADF_MSTATE_GEN_CAP_IDS "gencap" +#define ADF_MSTATE_GEN_SVCMAP_IDS "svcmap" +#define ADF_MSTATE_GEN_EXTDC_IDS "extdc" +#define ADF_MSTATE_VINTSRC_PF2VM_IDS "vispv" +#define ADF_MSTATE_VINTMSK_PF2VM_IDS "vimpv" +#define ADF_MSTATE_VM2PF_IDS "vm2pf" +#define ADF_MSTATE_PF2VM_IDS "pf2vm" + +struct adf_mstate_mgr { + u8 *buf; + u8 *state; + u32 size; + u32 n_sects; +}; + +struct adf_mstate_preh { + u32 magic; + u32 version; + u16 preh_len; + u16 n_sects; + u32 size; +}; + +struct adf_mstate_vreginfo { + void *addr; + u32 size; +}; + +struct adf_mstate_sect_h; + +typedef int (*adf_mstate_preamble_checker)(struct adf_mstate_preh *preamble, void *opa); +typedef int (*adf_mstate_populate)(struct adf_mstate_mgr *sub_mgr, u8 *buf, + u32 size, void *opa); +typedef int (*adf_mstate_action)(struct adf_mstate_mgr *sub_mgr, u8 *buf, u32 size, + void *opa); + +struct adf_mstate_mgr *adf_mstate_mgr_new(u8 *buf, u32 size); +void adf_mstate_mgr_destroy(struct adf_mstate_mgr *mgr); +void adf_mstate_mgr_init(struct adf_mstate_mgr *mgr, u8 *buf, u32 size); +void adf_mstate_mgr_init_from_parent(struct adf_mstate_mgr *mgr, + struct adf_mstate_mgr *p_mgr); +void adf_mstate_mgr_init_from_psect(struct adf_mstate_mgr *mgr, + struct adf_mstate_sect_h *p_sect); +int adf_mstate_mgr_init_from_remote(struct adf_mstate_mgr *mgr, + u8 *buf, u32 size, + adf_mstate_preamble_checker checker, + void *opaque); +struct adf_mstate_preh *adf_mstate_preamble_add(struct adf_mstate_mgr *mgr); +int adf_mstate_preamble_update(struct adf_mstate_mgr *mgr); +u32 adf_mstate_state_size(struct adf_mstate_mgr *mgr); +u32 adf_mstate_state_size_from_remote(struct adf_mstate_mgr *mgr); +void adf_mstate_sect_update(struct adf_mstate_mgr *p_mgr, + struct adf_mstate_mgr *curr_mgr, + struct adf_mstate_sect_h *sect); +struct adf_mstate_sect_h *adf_mstate_sect_add_vreg(struct adf_mstate_mgr *mgr, + const char *id, + struct adf_mstate_vreginfo *info); +struct adf_mstate_sect_h *adf_mstate_sect_add(struct adf_mstate_mgr *mgr, + const char *id, + adf_mstate_populate populate, + void *opaque); +struct adf_mstate_sect_h *adf_mstate_sect_lookup(struct adf_mstate_mgr *mgr, + const char *id, + adf_mstate_action action, + void *opaque); +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_msg.h b/drivers/crypto/intel/qat/qat_common/adf_pfvf_msg.h new file mode 100644 index 000000000000..d1b3ef9cadac --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_msg.h @@ -0,0 +1,264 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#ifndef ADF_PFVF_MSG_H +#define ADF_PFVF_MSG_H + +#include <linux/bits.h> + +/* + * PF<->VF Gen2 Messaging format + * + * The PF has an array of 32-bit PF2VF registers, one for each VF. The + * PF can access all these registers while each VF can access only the one + * register associated with that particular VF. + * + * The register functionally is split into two parts: + * The bottom half is for PF->VF messages. In particular when the first + * bit of this register (bit 0) gets set an interrupt will be triggered + * in the respective VF. + * The top half is for VF->PF messages. In particular when the first bit + * of this half of register (bit 16) gets set an interrupt will be triggered + * in the PF. + * + * The remaining bits within this register are available to encode messages. + * and implement a collision control mechanism to prevent concurrent use of + * the PF2VF register by both the PF and VF. + * + * 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 + * _______________________________________________ + * | | | | | | | | | | | | | | | | | + * +-----------------------------------------------+ + * \___________________________/ \_________/ ^ ^ + * ^ ^ | | + * | | | VF2PF Int + * | | Message Origin + * | Message Type + * Message-specific Data/Reserved + * + * 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 + * _______________________________________________ + * | | | | | | | | | | | | | | | | | + * +-----------------------------------------------+ + * \___________________________/ \_________/ ^ ^ + * ^ ^ | | + * | | | PF2VF Int + * | | Message Origin + * | Message Type + * Message-specific Data/Reserved + * + * Message Origin (Should always be 1) + * A legacy out-of-tree QAT driver allowed for a set of messages not supported + * by this driver; these had a Msg Origin of 0 and are ignored by this driver. + * + * When a PF or VF attempts to send a message in the lower or upper 16 bits, + * respectively, the other 16 bits are written to first with a defined + * IN_USE_BY pattern as part of a collision control scheme (see function + * adf_gen2_pfvf_send() in adf_pf2vf_msg.c). + * + * + * PF<->VF Gen4 Messaging format + * + * Similarly to the gen2 messaging format, 32-bit long registers are used for + * communication between PF and VFs. However, each VF and PF share a pair of + * 32-bits register to avoid collisions: one for PV to VF messages and one + * for VF to PF messages. + * + * Both the Interrupt bit and the Message Origin bit retain the same position + * and meaning, although non-system messages are now deprecated and not + * expected. + * + * 31 30 9 8 7 6 5 4 3 2 1 0 + * _______________________________________________ + * | | | . . . | | | | | | | | | | | + * +-----------------------------------------------+ + * \_____________________/ \_______________/ ^ ^ + * ^ ^ | | + * | | | PF/VF Int + * | | Message Origin + * | Message Type + * Message-specific Data/Reserved + * + * For both formats, the message reception is acknowledged by lowering the + * interrupt bit on the register where the message was sent. + */ + +/* PFVF message common bits */ +#define ADF_PFVF_INT BIT(0) +#define ADF_PFVF_MSGORIGIN_SYSTEM BIT(1) + +/* Different generations have different CSR layouts, use this struct + * to abstract these differences away + */ +struct pfvf_message { + u8 type; + u32 data; +}; + +/* PF->VF messages */ +enum pf2vf_msgtype { + ADF_PF2VF_MSGTYPE_RESTARTING = 0x01, + ADF_PF2VF_MSGTYPE_VERSION_RESP = 0x02, + ADF_PF2VF_MSGTYPE_BLKMSG_RESP = 0x03, + ADF_PF2VF_MSGTYPE_FATAL_ERROR = 0x04, + ADF_PF2VF_MSGTYPE_RESTARTED = 0x05, +/* Values from 0x10 are Gen4 specific, message type is only 4 bits in Gen2 devices. */ + ADF_PF2VF_MSGTYPE_RP_RESET_RESP = 0x10, +}; + +/* VF->PF messages */ +enum vf2pf_msgtype { + ADF_VF2PF_MSGTYPE_INIT = 0x03, + ADF_VF2PF_MSGTYPE_SHUTDOWN = 0x04, + ADF_VF2PF_MSGTYPE_VERSION_REQ = 0x05, + ADF_VF2PF_MSGTYPE_COMPAT_VER_REQ = 0x06, + ADF_VF2PF_MSGTYPE_LARGE_BLOCK_REQ = 0x07, + ADF_VF2PF_MSGTYPE_MEDIUM_BLOCK_REQ = 0x08, + ADF_VF2PF_MSGTYPE_SMALL_BLOCK_REQ = 0x09, + ADF_VF2PF_MSGTYPE_RESTARTING_COMPLETE = 0x0a, +/* Values from 0x10 are Gen4 specific, message type is only 4 bits in Gen2 devices. */ + ADF_VF2PF_MSGTYPE_RP_RESET = 0x10, +}; + +/* VF/PF compatibility version. */ +enum pfvf_compatibility_version { + /* Support for extended capabilities */ + ADF_PFVF_COMPAT_CAPABILITIES = 0x02, + /* In-use pattern cleared by receiver */ + ADF_PFVF_COMPAT_FAST_ACK = 0x03, + /* Ring to service mapping support for non-standard mappings */ + ADF_PFVF_COMPAT_RING_TO_SVC_MAP = 0x04, + /* Fallback compat */ + ADF_PFVF_COMPAT_FALLBACK = 0x05, + /* Reference to the latest version */ + ADF_PFVF_COMPAT_THIS_VERSION = 0x05, +}; + +/* PF->VF Version Response */ +#define ADF_PF2VF_VERSION_RESP_VERS_MASK GENMASK(7, 0) +#define ADF_PF2VF_VERSION_RESP_RESULT_MASK GENMASK(9, 8) + +enum pf2vf_compat_response { + ADF_PF2VF_VF_COMPATIBLE = 0x01, + ADF_PF2VF_VF_INCOMPATIBLE = 0x02, + ADF_PF2VF_VF_COMPAT_UNKNOWN = 0x03, +}; + +enum ring_reset_result { + RPRESET_SUCCESS = 0x00, + RPRESET_NOT_SUPPORTED = 0x01, + RPRESET_INVAL_BANK = 0x02, + RPRESET_TIMEOUT = 0x03, +}; + +#define ADF_VF2PF_RNG_RESET_RP_MASK GENMASK(1, 0) +#define ADF_VF2PF_RNG_RESET_RSVD_MASK GENMASK(25, 2) + +/* PF->VF Block Responses */ +#define ADF_PF2VF_BLKMSG_RESP_TYPE_MASK GENMASK(1, 0) +#define ADF_PF2VF_BLKMSG_RESP_DATA_MASK GENMASK(9, 2) + +enum pf2vf_blkmsg_resp_type { + ADF_PF2VF_BLKMSG_RESP_TYPE_DATA = 0x00, + ADF_PF2VF_BLKMSG_RESP_TYPE_CRC = 0x01, + ADF_PF2VF_BLKMSG_RESP_TYPE_ERROR = 0x02, +}; + +/* PF->VF Block Error Code */ +enum pf2vf_blkmsg_error { + ADF_PF2VF_INVALID_BLOCK_TYPE = 0x00, + ADF_PF2VF_INVALID_BYTE_NUM_REQ = 0x01, + ADF_PF2VF_PAYLOAD_TRUNCATED = 0x02, + ADF_PF2VF_UNSPECIFIED_ERROR = 0x03, +}; + +/* VF->PF Block Requests */ +#define ADF_VF2PF_LARGE_BLOCK_TYPE_MASK GENMASK(1, 0) +#define ADF_VF2PF_LARGE_BLOCK_BYTE_MASK GENMASK(8, 2) +#define ADF_VF2PF_MEDIUM_BLOCK_TYPE_MASK GENMASK(2, 0) +#define ADF_VF2PF_MEDIUM_BLOCK_BYTE_MASK GENMASK(8, 3) +#define ADF_VF2PF_SMALL_BLOCK_TYPE_MASK GENMASK(3, 0) +#define ADF_VF2PF_SMALL_BLOCK_BYTE_MASK GENMASK(8, 4) +#define ADF_VF2PF_BLOCK_CRC_REQ_MASK BIT(9) + +/* PF->VF Block Request Types + * 0..15 - 32 byte message + * 16..23 - 64 byte message + * 24..27 - 128 byte message + */ +enum vf2pf_blkmsg_req_type { + ADF_VF2PF_BLKMSG_REQ_CAP_SUMMARY = 0x02, + ADF_VF2PF_BLKMSG_REQ_RING_SVC_MAP = 0x03, +}; + +#define ADF_VF2PF_SMALL_BLOCK_TYPE_MAX \ + (FIELD_MAX(ADF_VF2PF_SMALL_BLOCK_TYPE_MASK)) + +#define ADF_VF2PF_MEDIUM_BLOCK_TYPE_MAX \ + (FIELD_MAX(ADF_VF2PF_MEDIUM_BLOCK_TYPE_MASK) + \ + ADF_VF2PF_SMALL_BLOCK_TYPE_MAX + 1) + +#define ADF_VF2PF_LARGE_BLOCK_TYPE_MAX \ + (FIELD_MAX(ADF_VF2PF_LARGE_BLOCK_TYPE_MASK) + \ + ADF_VF2PF_MEDIUM_BLOCK_TYPE_MAX) + +#define ADF_VF2PF_SMALL_BLOCK_BYTE_MAX \ + FIELD_MAX(ADF_VF2PF_SMALL_BLOCK_BYTE_MASK) + +#define ADF_VF2PF_MEDIUM_BLOCK_BYTE_MAX \ + FIELD_MAX(ADF_VF2PF_MEDIUM_BLOCK_BYTE_MASK) + +#define ADF_VF2PF_LARGE_BLOCK_BYTE_MAX \ + FIELD_MAX(ADF_VF2PF_LARGE_BLOCK_BYTE_MASK) + +struct pfvf_blkmsg_header { + u8 version; + u8 payload_size; +} __packed; + +#define ADF_PFVF_BLKMSG_HEADER_SIZE (sizeof(struct pfvf_blkmsg_header)) +#define ADF_PFVF_BLKMSG_PAYLOAD_SIZE(blkmsg) (sizeof(blkmsg) - \ + ADF_PFVF_BLKMSG_HEADER_SIZE) +#define ADF_PFVF_BLKMSG_MSG_SIZE(blkmsg) (ADF_PFVF_BLKMSG_HEADER_SIZE + \ + (blkmsg)->hdr.payload_size) +#define ADF_PFVF_BLKMSG_MSG_MAX_SIZE 128 + +/* PF->VF Block message header bytes */ +#define ADF_PFVF_BLKMSG_VER_BYTE 0 +#define ADF_PFVF_BLKMSG_LEN_BYTE 1 + +/* PF/VF Capabilities message values */ +enum blkmsg_capabilities_versions { + ADF_PFVF_CAPABILITIES_V1_VERSION = 0x01, + ADF_PFVF_CAPABILITIES_V2_VERSION = 0x02, + ADF_PFVF_CAPABILITIES_V3_VERSION = 0x03, +}; + +struct capabilities_v1 { + struct pfvf_blkmsg_header hdr; + u32 ext_dc_caps; +} __packed; + +struct capabilities_v2 { + struct pfvf_blkmsg_header hdr; + u32 ext_dc_caps; + u32 capabilities; +} __packed; + +struct capabilities_v3 { + struct pfvf_blkmsg_header hdr; + u32 ext_dc_caps; + u32 capabilities; + u32 frequency; +} __packed; + +/* PF/VF Ring to service mapping values */ +enum blkmsg_ring_to_svc_versions { + ADF_PFVF_RING_TO_SVC_VERSION = 0x01, +}; + +struct ring_to_svc_map_v1 { + struct pfvf_blkmsg_header hdr; + u16 map; +} __packed; + +#endif /* ADF_PFVF_MSG_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_msg.c b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_msg.c new file mode 100644 index 000000000000..0cee3b23dee9 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_msg.c @@ -0,0 +1,117 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <linux/delay.h> +#include <linux/pci.h> +#include "adf_accel_devices.h" +#include "adf_pfvf_msg.h" +#include "adf_pfvf_pf_msg.h" +#include "adf_pfvf_pf_proto.h" + +#define ADF_PF_WAIT_RESTARTING_COMPLETE_DELAY 100 +#define ADF_VF_SHUTDOWN_RETRY 100 + +void adf_pf2vf_notify_restarting(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_vf_info *vf; + struct pfvf_message msg = { .type = ADF_PF2VF_MSGTYPE_RESTARTING }; + int i, num_vfs = pci_num_vf(accel_to_pci_dev(accel_dev)); + + dev_dbg(&GET_DEV(accel_dev), "pf2vf notify restarting\n"); + for (i = 0, vf = accel_dev->pf.vf_info; i < num_vfs; i++, vf++) { + if (vf->init && vf->vf_compat_ver >= ADF_PFVF_COMPAT_FALLBACK) + vf->restarting = true; + else + vf->restarting = false; + + if (!vf->init) + continue; + + if (adf_send_pf2vf_msg(accel_dev, i, msg)) + dev_err(&GET_DEV(accel_dev), + "Failed to send restarting msg to VF%d\n", i); + } +} + +void adf_pf2vf_wait_for_restarting_complete(struct adf_accel_dev *accel_dev) +{ + int num_vfs = pci_num_vf(accel_to_pci_dev(accel_dev)); + int i, retries = ADF_VF_SHUTDOWN_RETRY; + struct adf_accel_vf_info *vf; + bool vf_running; + + dev_dbg(&GET_DEV(accel_dev), "pf2vf wait for restarting complete\n"); + do { + vf_running = false; + for (i = 0, vf = accel_dev->pf.vf_info; i < num_vfs; i++, vf++) + if (vf->restarting) + vf_running = true; + if (!vf_running) + break; + msleep(ADF_PF_WAIT_RESTARTING_COMPLETE_DELAY); + } while (--retries); + + if (vf_running) + dev_warn(&GET_DEV(accel_dev), "Some VFs are still running\n"); +} + +void adf_pf2vf_notify_restarted(struct adf_accel_dev *accel_dev) +{ + struct pfvf_message msg = { .type = ADF_PF2VF_MSGTYPE_RESTARTED }; + int i, num_vfs = pci_num_vf(accel_to_pci_dev(accel_dev)); + struct adf_accel_vf_info *vf; + + dev_dbg(&GET_DEV(accel_dev), "pf2vf notify restarted\n"); + for (i = 0, vf = accel_dev->pf.vf_info; i < num_vfs; i++, vf++) { + if (vf->init && vf->vf_compat_ver >= ADF_PFVF_COMPAT_FALLBACK && + adf_send_pf2vf_msg(accel_dev, i, msg)) + dev_err(&GET_DEV(accel_dev), + "Failed to send restarted msg to VF%d\n", i); + } +} + +void adf_pf2vf_notify_fatal_error(struct adf_accel_dev *accel_dev) +{ + struct pfvf_message msg = { .type = ADF_PF2VF_MSGTYPE_FATAL_ERROR }; + int i, num_vfs = pci_num_vf(accel_to_pci_dev(accel_dev)); + struct adf_accel_vf_info *vf; + + dev_dbg(&GET_DEV(accel_dev), "pf2vf notify fatal error\n"); + for (i = 0, vf = accel_dev->pf.vf_info; i < num_vfs; i++, vf++) { + if (vf->init && vf->vf_compat_ver >= ADF_PFVF_COMPAT_FALLBACK && + adf_send_pf2vf_msg(accel_dev, i, msg)) + dev_err(&GET_DEV(accel_dev), + "Failed to send fatal error msg to VF%d\n", i); + } +} + +int adf_pf_capabilities_msg_provider(struct adf_accel_dev *accel_dev, + u8 *buffer, u8 compat) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct capabilities_v2 caps_msg; + + caps_msg.ext_dc_caps = hw_data->extended_dc_capabilities; + caps_msg.capabilities = hw_data->accel_capabilities_mask; + + caps_msg.hdr.version = ADF_PFVF_CAPABILITIES_V2_VERSION; + caps_msg.hdr.payload_size = + ADF_PFVF_BLKMSG_PAYLOAD_SIZE(struct capabilities_v2); + + memcpy(buffer, &caps_msg, sizeof(caps_msg)); + + return 0; +} + +int adf_pf_ring_to_svc_msg_provider(struct adf_accel_dev *accel_dev, + u8 *buffer, u8 compat) +{ + struct ring_to_svc_map_v1 rts_map_msg; + + rts_map_msg.map = accel_dev->hw_device->ring_to_svc_map; + rts_map_msg.hdr.version = ADF_PFVF_RING_TO_SVC_VERSION; + rts_map_msg.hdr.payload_size = ADF_PFVF_BLKMSG_PAYLOAD_SIZE(rts_map_msg); + + memcpy(buffer, &rts_map_msg, sizeof(rts_map_msg)); + + return 0; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_msg.h b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_msg.h new file mode 100644 index 000000000000..f203d88c919c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_msg.h @@ -0,0 +1,39 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2021 Intel Corporation */ +#ifndef ADF_PFVF_PF_MSG_H +#define ADF_PFVF_PF_MSG_H + +#include "adf_accel_devices.h" + +#if defined(CONFIG_PCI_IOV) +void adf_pf2vf_notify_restarting(struct adf_accel_dev *accel_dev); +void adf_pf2vf_wait_for_restarting_complete(struct adf_accel_dev *accel_dev); +void adf_pf2vf_notify_restarted(struct adf_accel_dev *accel_dev); +void adf_pf2vf_notify_fatal_error(struct adf_accel_dev *accel_dev); +#else +static inline void adf_pf2vf_notify_restarting(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_pf2vf_wait_for_restarting_complete(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_pf2vf_notify_restarted(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_pf2vf_notify_fatal_error(struct adf_accel_dev *accel_dev) +{ +} +#endif + +typedef int (*adf_pf2vf_blkmsg_provider)(struct adf_accel_dev *accel_dev, + u8 *buffer, u8 compat); + +int adf_pf_capabilities_msg_provider(struct adf_accel_dev *accel_dev, + u8 *buffer, u8 comapt); +int adf_pf_ring_to_svc_msg_provider(struct adf_accel_dev *accel_dev, + u8 *buffer, u8 comapt); + +#endif /* ADF_PFVF_PF_MSG_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_proto.c b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_proto.c new file mode 100644 index 000000000000..b9b5e744a3f1 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_proto.c @@ -0,0 +1,350 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <linux/bitfield.h> +#include <linux/spinlock.h> +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_pfvf_msg.h" +#include "adf_pfvf_pf_msg.h" +#include "adf_pfvf_pf_proto.h" +#include "adf_pfvf_utils.h" + +typedef u8 (*pf2vf_blkmsg_data_getter_fn)(u8 const *blkmsg, u8 byte); + +static const adf_pf2vf_blkmsg_provider pf2vf_blkmsg_providers[] = { + NULL, /* no message type defined for value 0 */ + NULL, /* no message type defined for value 1 */ + adf_pf_capabilities_msg_provider, /* ADF_VF2PF_BLKMSG_REQ_CAP_SUMMARY */ + adf_pf_ring_to_svc_msg_provider, /* ADF_VF2PF_BLKMSG_REQ_RING_SVC_MAP */ +}; + +/** + * adf_send_pf2vf_msg() - send PF to VF message + * @accel_dev: Pointer to acceleration device + * @vf_nr: VF number to which the message will be sent + * @msg: Message to send + * + * This function allows the PF to send a message to a specific VF. + * + * Return: 0 on success, error code otherwise. + */ +int adf_send_pf2vf_msg(struct adf_accel_dev *accel_dev, u8 vf_nr, struct pfvf_message msg) +{ + struct adf_pfvf_ops *pfvf_ops = GET_PFVF_OPS(accel_dev); + u32 pfvf_offset = pfvf_ops->get_pf2vf_offset(vf_nr); + + return pfvf_ops->send_msg(accel_dev, msg, pfvf_offset, + &accel_dev->pf.vf_info[vf_nr].pf2vf_lock); +} + +/** + * adf_recv_vf2pf_msg() - receive a VF to PF message + * @accel_dev: Pointer to acceleration device + * @vf_nr: Number of the VF from where the message will be received + * + * This function allows the PF to receive a message from a specific VF. + * + * Return: a valid message on success, zero otherwise. + */ +static struct pfvf_message adf_recv_vf2pf_msg(struct adf_accel_dev *accel_dev, u8 vf_nr) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + struct adf_pfvf_ops *pfvf_ops = GET_PFVF_OPS(accel_dev); + u32 pfvf_offset = pfvf_ops->get_vf2pf_offset(vf_nr); + + return pfvf_ops->recv_msg(accel_dev, pfvf_offset, vf_info->vf_compat_ver); +} + +static adf_pf2vf_blkmsg_provider get_blkmsg_response_provider(u8 type) +{ + if (type >= ARRAY_SIZE(pf2vf_blkmsg_providers)) + return NULL; + + return pf2vf_blkmsg_providers[type]; +} + +/* Byte pf2vf_blkmsg_data_getter_fn callback */ +static u8 adf_pf2vf_blkmsg_get_byte(u8 const *blkmsg, u8 index) +{ + return blkmsg[index]; +} + +/* CRC pf2vf_blkmsg_data_getter_fn callback */ +static u8 adf_pf2vf_blkmsg_get_crc(u8 const *blkmsg, u8 count) +{ + /* count is 0-based, turn it into a length */ + return adf_pfvf_calc_blkmsg_crc(blkmsg, count + 1); +} + +static int adf_pf2vf_blkmsg_get_data(struct adf_accel_vf_info *vf_info, + u8 type, u8 byte, u8 max_size, u8 *data, + pf2vf_blkmsg_data_getter_fn data_getter) +{ + u8 blkmsg[ADF_PFVF_BLKMSG_MSG_MAX_SIZE] = { 0 }; + struct adf_accel_dev *accel_dev = vf_info->accel_dev; + adf_pf2vf_blkmsg_provider provider; + u8 msg_size; + + provider = get_blkmsg_response_provider(type); + + if (unlikely(!provider)) { + pr_err("QAT: No registered provider for message %d\n", type); + *data = ADF_PF2VF_INVALID_BLOCK_TYPE; + return -EINVAL; + } + + if (unlikely((*provider)(accel_dev, blkmsg, vf_info->vf_compat_ver))) { + pr_err("QAT: unknown error from provider for message %d\n", type); + *data = ADF_PF2VF_UNSPECIFIED_ERROR; + return -EINVAL; + } + + msg_size = ADF_PFVF_BLKMSG_HEADER_SIZE + blkmsg[ADF_PFVF_BLKMSG_LEN_BYTE]; + + if (unlikely(msg_size >= max_size)) { + pr_err("QAT: Invalid size %d provided for message type %d\n", + msg_size, type); + *data = ADF_PF2VF_PAYLOAD_TRUNCATED; + return -EINVAL; + } + + if (unlikely(byte >= msg_size)) { + pr_err("QAT: Out-of-bound byte number %d (msg size %d)\n", + byte, msg_size); + *data = ADF_PF2VF_INVALID_BYTE_NUM_REQ; + return -EINVAL; + } + + *data = data_getter(blkmsg, byte); + return 0; +} + +static struct pfvf_message handle_blkmsg_req(struct adf_accel_vf_info *vf_info, + struct pfvf_message req) +{ + u8 resp_type = ADF_PF2VF_BLKMSG_RESP_TYPE_ERROR; + struct pfvf_message resp = { 0 }; + u8 resp_data = 0; + u8 blk_type; + u8 blk_byte; + u8 byte_max; + + switch (req.type) { + case ADF_VF2PF_MSGTYPE_LARGE_BLOCK_REQ: + blk_type = FIELD_GET(ADF_VF2PF_LARGE_BLOCK_TYPE_MASK, req.data) + + ADF_VF2PF_MEDIUM_BLOCK_TYPE_MAX + 1; + blk_byte = FIELD_GET(ADF_VF2PF_LARGE_BLOCK_BYTE_MASK, req.data); + byte_max = ADF_VF2PF_LARGE_BLOCK_BYTE_MAX; + break; + case ADF_VF2PF_MSGTYPE_MEDIUM_BLOCK_REQ: + blk_type = FIELD_GET(ADF_VF2PF_MEDIUM_BLOCK_TYPE_MASK, req.data) + + ADF_VF2PF_SMALL_BLOCK_TYPE_MAX + 1; + blk_byte = FIELD_GET(ADF_VF2PF_MEDIUM_BLOCK_BYTE_MASK, req.data); + byte_max = ADF_VF2PF_MEDIUM_BLOCK_BYTE_MAX; + break; + case ADF_VF2PF_MSGTYPE_SMALL_BLOCK_REQ: + blk_type = FIELD_GET(ADF_VF2PF_SMALL_BLOCK_TYPE_MASK, req.data); + blk_byte = FIELD_GET(ADF_VF2PF_SMALL_BLOCK_BYTE_MASK, req.data); + byte_max = ADF_VF2PF_SMALL_BLOCK_BYTE_MAX; + break; + } + + /* Is this a request for CRC or data? */ + if (FIELD_GET(ADF_VF2PF_BLOCK_CRC_REQ_MASK, req.data)) { + dev_dbg(&GET_DEV(vf_info->accel_dev), + "BlockMsg of type %d for CRC over %d bytes received from VF%d\n", + blk_type, blk_byte + 1, vf_info->vf_nr); + + if (!adf_pf2vf_blkmsg_get_data(vf_info, blk_type, blk_byte, + byte_max, &resp_data, + adf_pf2vf_blkmsg_get_crc)) + resp_type = ADF_PF2VF_BLKMSG_RESP_TYPE_CRC; + } else { + dev_dbg(&GET_DEV(vf_info->accel_dev), + "BlockMsg of type %d for data byte %d received from VF%d\n", + blk_type, blk_byte, vf_info->vf_nr); + + if (!adf_pf2vf_blkmsg_get_data(vf_info, blk_type, blk_byte, + byte_max, &resp_data, + adf_pf2vf_blkmsg_get_byte)) + resp_type = ADF_PF2VF_BLKMSG_RESP_TYPE_DATA; + } + + resp.type = ADF_PF2VF_MSGTYPE_BLKMSG_RESP; + resp.data = FIELD_PREP(ADF_PF2VF_BLKMSG_RESP_TYPE_MASK, resp_type) | + FIELD_PREP(ADF_PF2VF_BLKMSG_RESP_DATA_MASK, resp_data); + + return resp; +} + +static struct pfvf_message handle_rp_reset_req(struct adf_accel_dev *accel_dev, u8 vf_nr, + struct pfvf_message req) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct pfvf_message resp = { + .type = ADF_PF2VF_MSGTYPE_RP_RESET_RESP, + .data = RPRESET_SUCCESS + }; + u32 bank_number; + u32 rsvd_field; + + bank_number = FIELD_GET(ADF_VF2PF_RNG_RESET_RP_MASK, req.data); + rsvd_field = FIELD_GET(ADF_VF2PF_RNG_RESET_RSVD_MASK, req.data); + + dev_dbg(&GET_DEV(accel_dev), + "Ring Pair Reset Message received from VF%d for bank 0x%x\n", + vf_nr, bank_number); + + if (!hw_data->ring_pair_reset || rsvd_field) { + dev_dbg(&GET_DEV(accel_dev), + "Ring Pair Reset for VF%d is not supported\n", vf_nr); + resp.data = RPRESET_NOT_SUPPORTED; + goto out; + } + + if (bank_number >= hw_data->num_banks_per_vf) { + dev_err(&GET_DEV(accel_dev), + "Invalid bank number (0x%x) from VF%d for Ring Reset\n", + bank_number, vf_nr); + resp.data = RPRESET_INVAL_BANK; + goto out; + } + + /* Convert the VF provided value to PF bank number */ + bank_number = vf_nr * hw_data->num_banks_per_vf + bank_number; + if (hw_data->ring_pair_reset(accel_dev, bank_number)) { + dev_dbg(&GET_DEV(accel_dev), + "Ring pair reset for VF%d failure\n", vf_nr); + resp.data = RPRESET_TIMEOUT; + goto out; + } + + dev_dbg(&GET_DEV(accel_dev), + "Ring pair reset for VF%d successfully\n", vf_nr); + +out: + return resp; +} + +static int adf_handle_vf2pf_msg(struct adf_accel_dev *accel_dev, u8 vf_nr, + struct pfvf_message msg, struct pfvf_message *resp) +{ + struct adf_accel_vf_info *vf_info = &accel_dev->pf.vf_info[vf_nr]; + + switch (msg.type) { + case ADF_VF2PF_MSGTYPE_COMPAT_VER_REQ: + { + u8 vf_compat_ver = msg.data; + u8 compat; + + dev_dbg(&GET_DEV(accel_dev), + "VersionRequest received from VF%d (vers %d) to PF (vers %d)\n", + vf_nr, vf_compat_ver, ADF_PFVF_COMPAT_THIS_VERSION); + + compat = adf_vf_compat_checker(vf_compat_ver); + vf_info->vf_compat_ver = vf_compat_ver; + + resp->type = ADF_PF2VF_MSGTYPE_VERSION_RESP; + resp->data = FIELD_PREP(ADF_PF2VF_VERSION_RESP_VERS_MASK, + ADF_PFVF_COMPAT_THIS_VERSION) | + FIELD_PREP(ADF_PF2VF_VERSION_RESP_RESULT_MASK, compat); + } + break; + case ADF_VF2PF_MSGTYPE_VERSION_REQ: + { + u8 compat; + + dev_dbg(&GET_DEV(accel_dev), + "Legacy VersionRequest received from VF%d to PF (vers 1.1)\n", + vf_nr); + + /* legacy driver, VF compat_ver is 0 */ + vf_info->vf_compat_ver = 0; + + /* PF always newer than legacy VF */ + compat = ADF_PF2VF_VF_COMPATIBLE; + + /* Set legacy major and minor version to the latest, 1.1 */ + resp->type = ADF_PF2VF_MSGTYPE_VERSION_RESP; + resp->data = FIELD_PREP(ADF_PF2VF_VERSION_RESP_VERS_MASK, 0x11) | + FIELD_PREP(ADF_PF2VF_VERSION_RESP_RESULT_MASK, compat); + } + break; + case ADF_VF2PF_MSGTYPE_INIT: + { + dev_dbg(&GET_DEV(accel_dev), + "Init message received from VF%d\n", vf_nr); + vf_info->init = true; + } + break; + case ADF_VF2PF_MSGTYPE_SHUTDOWN: + { + dev_dbg(&GET_DEV(accel_dev), + "Shutdown message received from VF%d\n", vf_nr); + vf_info->init = false; + } + break; + case ADF_VF2PF_MSGTYPE_RESTARTING_COMPLETE: + { + dev_dbg(&GET_DEV(accel_dev), + "Restarting Complete received from VF%d\n", vf_nr); + vf_info->restarting = false; + vf_info->init = false; + } + break; + case ADF_VF2PF_MSGTYPE_LARGE_BLOCK_REQ: + case ADF_VF2PF_MSGTYPE_MEDIUM_BLOCK_REQ: + case ADF_VF2PF_MSGTYPE_SMALL_BLOCK_REQ: + *resp = handle_blkmsg_req(vf_info, msg); + break; + case ADF_VF2PF_MSGTYPE_RP_RESET: + *resp = handle_rp_reset_req(accel_dev, vf_nr, msg); + break; + default: + dev_dbg(&GET_DEV(accel_dev), + "Unknown message from VF%d (type 0x%.4x, data: 0x%.4x)\n", + vf_nr, msg.type, msg.data); + return -ENOMSG; + } + + return 0; +} + +bool adf_recv_and_handle_vf2pf_msg(struct adf_accel_dev *accel_dev, u32 vf_nr) +{ + struct pfvf_message req; + struct pfvf_message resp = {0}; + + req = adf_recv_vf2pf_msg(accel_dev, vf_nr); + if (!req.type) /* Legacy or no message */ + return true; + + if (adf_handle_vf2pf_msg(accel_dev, vf_nr, req, &resp)) + return false; + + if (resp.type && adf_send_pf2vf_msg(accel_dev, vf_nr, resp)) + dev_err(&GET_DEV(accel_dev), + "Failed to send response to VF%d\n", vf_nr); + + return true; +} + +/** + * adf_enable_pf2vf_comms() - Function enables communication from pf to vf + * + * @accel_dev: Pointer to acceleration device virtual function. + * + * This function carries out the necessary steps to setup and start the PFVF + * communication channel, if any. + * + * Return: 0 on success, error code otherwise. + */ +int adf_enable_pf2vf_comms(struct adf_accel_dev *accel_dev) +{ + adf_pfvf_crc_init(); + spin_lock_init(&accel_dev->pf.vf2pf_ints_lock); + + return 0; +} +EXPORT_SYMBOL_GPL(adf_enable_pf2vf_comms); diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_proto.h b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_proto.h new file mode 100644 index 000000000000..165d266d023d --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_pf_proto.h @@ -0,0 +1,13 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2021 Intel Corporation */ +#ifndef ADF_PFVF_PF_PROTO_H +#define ADF_PFVF_PF_PROTO_H + +#include <linux/types.h> +#include "adf_accel_devices.h" + +int adf_send_pf2vf_msg(struct adf_accel_dev *accel_dev, u8 vf_nr, struct pfvf_message msg); + +int adf_enable_pf2vf_comms(struct adf_accel_dev *accel_dev); + +#endif /* ADF_PFVF_PF_PROTO_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_utils.c b/drivers/crypto/intel/qat/qat_common/adf_pfvf_utils.c new file mode 100644 index 000000000000..c5f6d77d4bb8 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_utils.c @@ -0,0 +1,65 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2021 Intel Corporation */ +#include <linux/crc8.h> +#include <linux/pci.h> +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "adf_pfvf_msg.h" +#include "adf_pfvf_utils.h" + +/* CRC Calculation */ +DECLARE_CRC8_TABLE(pfvf_crc8_table); +#define ADF_PFVF_CRC8_POLYNOMIAL 0x97 + +void adf_pfvf_crc_init(void) +{ + crc8_populate_msb(pfvf_crc8_table, ADF_PFVF_CRC8_POLYNOMIAL); +} + +u8 adf_pfvf_calc_blkmsg_crc(u8 const *buf, u8 buf_len) +{ + return crc8(pfvf_crc8_table, buf, buf_len, CRC8_INIT_VALUE); +} + +static bool set_value_on_csr_msg(struct adf_accel_dev *accel_dev, u32 *csr_msg, + u32 value, const struct pfvf_field_format *fmt) +{ + if (unlikely((value & fmt->mask) != value)) { + dev_err(&GET_DEV(accel_dev), + "PFVF message value 0x%X out of range, %u max allowed\n", + value, fmt->mask); + return false; + } + + *csr_msg |= value << fmt->offset; + + return true; +} + +u32 adf_pfvf_csr_msg_of(struct adf_accel_dev *accel_dev, + struct pfvf_message msg, + const struct pfvf_csr_format *fmt) +{ + u32 csr_msg = 0; + + if (!set_value_on_csr_msg(accel_dev, &csr_msg, msg.type, &fmt->type) || + !set_value_on_csr_msg(accel_dev, &csr_msg, msg.data, &fmt->data)) + return 0; + + return csr_msg | ADF_PFVF_MSGORIGIN_SYSTEM; +} + +struct pfvf_message adf_pfvf_message_of(struct adf_accel_dev *accel_dev, u32 csr_msg, + const struct pfvf_csr_format *fmt) +{ + struct pfvf_message msg = { 0 }; + + msg.type = (csr_msg >> fmt->type.offset) & fmt->type.mask; + msg.data = (csr_msg >> fmt->data.offset) & fmt->data.mask; + + if (unlikely(!msg.type)) + dev_err(&GET_DEV(accel_dev), + "Invalid PFVF msg with no type received\n"); + + return msg; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_utils.h b/drivers/crypto/intel/qat/qat_common/adf_pfvf_utils.h new file mode 100644 index 000000000000..1a044297d873 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_utils.h @@ -0,0 +1,42 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2021 Intel Corporation */ +#ifndef ADF_PFVF_UTILS_H +#define ADF_PFVF_UTILS_H + +#include <linux/types.h> +#include "adf_pfvf_msg.h" + +/* How long to wait for far side to acknowledge receipt */ +#define ADF_PFVF_MSG_ACK_DELAY_US 4 +#define ADF_PFVF_MSG_ACK_MAX_DELAY_US (1 * USEC_PER_SEC) + +u8 adf_pfvf_calc_blkmsg_crc(u8 const *buf, u8 buf_len); +void adf_pfvf_crc_init(void); + +struct pfvf_field_format { + u8 offset; + u32 mask; +}; + +struct pfvf_csr_format { + struct pfvf_field_format type; + struct pfvf_field_format data; +}; + +u32 adf_pfvf_csr_msg_of(struct adf_accel_dev *accel_dev, struct pfvf_message msg, + const struct pfvf_csr_format *fmt); +struct pfvf_message adf_pfvf_message_of(struct adf_accel_dev *accel_dev, u32 raw_msg, + const struct pfvf_csr_format *fmt); + +static inline u8 adf_vf_compat_checker(u8 vf_compat_ver) +{ + if (vf_compat_ver == 0) + return ADF_PF2VF_VF_INCOMPATIBLE; + + if (vf_compat_ver <= ADF_PFVF_COMPAT_THIS_VERSION) + return ADF_PF2VF_VF_COMPATIBLE; + + return ADF_PF2VF_VF_COMPAT_UNKNOWN; +} + +#endif /* ADF_PFVF_UTILS_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_msg.c b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_msg.c new file mode 100644 index 000000000000..10c91e56d6be --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_msg.c @@ -0,0 +1,181 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <linux/bitfield.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_pfvf_msg.h" +#include "adf_pfvf_vf_msg.h" +#include "adf_pfvf_vf_proto.h" + +/** + * adf_vf2pf_notify_init() - send init msg to PF + * @accel_dev: Pointer to acceleration VF device. + * + * Function sends an init message from the VF to a PF + * + * Return: 0 on success, error code otherwise. + */ +int adf_vf2pf_notify_init(struct adf_accel_dev *accel_dev) +{ + struct pfvf_message msg = { .type = ADF_VF2PF_MSGTYPE_INIT }; + + if (adf_send_vf2pf_msg(accel_dev, msg)) { + dev_err(&GET_DEV(accel_dev), + "Failed to send Init event to PF\n"); + return -EFAULT; + } + set_bit(ADF_STATUS_PF_RUNNING, &accel_dev->status); + return 0; +} +EXPORT_SYMBOL_GPL(adf_vf2pf_notify_init); + +/** + * adf_vf2pf_notify_shutdown() - send shutdown msg to PF + * @accel_dev: Pointer to acceleration VF device. + * + * Function sends a shutdown message from the VF to a PF + * + * Return: void + */ +void adf_vf2pf_notify_shutdown(struct adf_accel_dev *accel_dev) +{ + struct pfvf_message msg = { .type = ADF_VF2PF_MSGTYPE_SHUTDOWN }; + + if (test_bit(ADF_STATUS_PF_RUNNING, &accel_dev->status)) + if (adf_send_vf2pf_msg(accel_dev, msg)) + dev_err(&GET_DEV(accel_dev), + "Failed to send Shutdown event to PF\n"); +} +EXPORT_SYMBOL_GPL(adf_vf2pf_notify_shutdown); + +void adf_vf2pf_notify_restart_complete(struct adf_accel_dev *accel_dev) +{ + struct pfvf_message msg = { .type = ADF_VF2PF_MSGTYPE_RESTARTING_COMPLETE }; + + /* Check compatibility version */ + if (accel_dev->vf.pf_compat_ver < ADF_PFVF_COMPAT_FALLBACK) + return; + + if (adf_send_vf2pf_msg(accel_dev, msg)) + dev_err(&GET_DEV(accel_dev), + "Failed to send Restarting complete event to PF\n"); +} +EXPORT_SYMBOL_GPL(adf_vf2pf_notify_restart_complete); + +int adf_vf2pf_request_version(struct adf_accel_dev *accel_dev) +{ + u8 pf_version; + int compat; + int ret; + struct pfvf_message resp; + struct pfvf_message msg = { + .type = ADF_VF2PF_MSGTYPE_COMPAT_VER_REQ, + .data = ADF_PFVF_COMPAT_THIS_VERSION, + }; + + BUILD_BUG_ON(ADF_PFVF_COMPAT_THIS_VERSION > 255); + + ret = adf_send_vf2pf_req(accel_dev, msg, &resp); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to send Compatibility Version Request.\n"); + return ret; + } + + pf_version = FIELD_GET(ADF_PF2VF_VERSION_RESP_VERS_MASK, resp.data); + compat = FIELD_GET(ADF_PF2VF_VERSION_RESP_RESULT_MASK, resp.data); + + /* Response from PF received, check compatibility */ + switch (compat) { + case ADF_PF2VF_VF_COMPATIBLE: + break; + case ADF_PF2VF_VF_COMPAT_UNKNOWN: + /* VF is newer than PF - compatible for now */ + break; + case ADF_PF2VF_VF_INCOMPATIBLE: + dev_err(&GET_DEV(accel_dev), + "PF (vers %d) and VF (vers %d) are not compatible\n", + pf_version, ADF_PFVF_COMPAT_THIS_VERSION); + return -EINVAL; + default: + dev_err(&GET_DEV(accel_dev), + "Invalid response from PF; assume not compatible\n"); + return -EINVAL; + } + + accel_dev->vf.pf_compat_ver = pf_version; + return 0; +} + +int adf_vf2pf_get_capabilities(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct capabilities_v3 cap_msg = { 0 }; + unsigned int len = sizeof(cap_msg); + + if (accel_dev->vf.pf_compat_ver < ADF_PFVF_COMPAT_CAPABILITIES) + /* The PF is too old to support the extended capabilities */ + return 0; + + if (adf_send_vf2pf_blkmsg_req(accel_dev, ADF_VF2PF_BLKMSG_REQ_CAP_SUMMARY, + (u8 *)&cap_msg, &len)) { + dev_err(&GET_DEV(accel_dev), + "QAT: Failed to get block message response\n"); + return -EFAULT; + } + + switch (cap_msg.hdr.version) { + default: + /* Newer version received, handle only the know parts */ + fallthrough; + case ADF_PFVF_CAPABILITIES_V3_VERSION: + if (likely(len >= sizeof(struct capabilities_v3))) + hw_data->clock_frequency = cap_msg.frequency; + else + dev_info(&GET_DEV(accel_dev), "Could not get frequency"); + fallthrough; + case ADF_PFVF_CAPABILITIES_V2_VERSION: + if (likely(len >= sizeof(struct capabilities_v2))) + hw_data->accel_capabilities_mask = cap_msg.capabilities; + else + dev_info(&GET_DEV(accel_dev), "Could not get capabilities"); + fallthrough; + case ADF_PFVF_CAPABILITIES_V1_VERSION: + if (likely(len >= sizeof(struct capabilities_v1))) { + hw_data->extended_dc_capabilities = cap_msg.ext_dc_caps; + } else { + dev_err(&GET_DEV(accel_dev), + "Capabilities message truncated to %d bytes\n", len); + return -EFAULT; + } + } + + return 0; +} + +int adf_vf2pf_get_ring_to_svc(struct adf_accel_dev *accel_dev) +{ + struct ring_to_svc_map_v1 rts_map_msg = { 0 }; + unsigned int len = sizeof(rts_map_msg); + + if (accel_dev->vf.pf_compat_ver < ADF_PFVF_COMPAT_RING_TO_SVC_MAP) + /* Use already set default mappings */ + return 0; + + if (adf_send_vf2pf_blkmsg_req(accel_dev, ADF_VF2PF_BLKMSG_REQ_RING_SVC_MAP, + (u8 *)&rts_map_msg, &len)) { + dev_err(&GET_DEV(accel_dev), + "QAT: Failed to get block message response\n"); + return -EFAULT; + } + + if (unlikely(len < sizeof(struct ring_to_svc_map_v1))) { + dev_err(&GET_DEV(accel_dev), + "RING_TO_SVC message truncated to %d bytes\n", len); + return -EFAULT; + } + + /* Only v1 at present */ + accel_dev->hw_device->ring_to_svc_map = rts_map_msg.map; + return 0; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_msg.h b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_msg.h new file mode 100644 index 000000000000..d79340ab3134 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_msg.h @@ -0,0 +1,24 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2021 Intel Corporation */ +#ifndef ADF_PFVF_VF_MSG_H +#define ADF_PFVF_VF_MSG_H + +#if defined(CONFIG_PCI_IOV) +int adf_vf2pf_notify_init(struct adf_accel_dev *accel_dev); +void adf_vf2pf_notify_shutdown(struct adf_accel_dev *accel_dev); +void adf_vf2pf_notify_restart_complete(struct adf_accel_dev *accel_dev); +int adf_vf2pf_request_version(struct adf_accel_dev *accel_dev); +int adf_vf2pf_get_capabilities(struct adf_accel_dev *accel_dev); +int adf_vf2pf_get_ring_to_svc(struct adf_accel_dev *accel_dev); +#else +static inline int adf_vf2pf_notify_init(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static inline void adf_vf2pf_notify_shutdown(struct adf_accel_dev *accel_dev) +{ +} +#endif + +#endif /* ADF_PFVF_VF_MSG_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_proto.c b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_proto.c new file mode 100644 index 000000000000..dc284a089c88 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_proto.c @@ -0,0 +1,374 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <linux/bitfield.h> +#include <linux/completion.h> +#include <linux/minmax.h> +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_pfvf_msg.h" +#include "adf_pfvf_utils.h" +#include "adf_pfvf_vf_msg.h" +#include "adf_pfvf_vf_proto.h" + +#define ADF_PFVF_MSG_COLLISION_DETECT_DELAY 10 +#define ADF_PFVF_MSG_ACK_DELAY 2 +#define ADF_PFVF_MSG_ACK_MAX_RETRY 100 + +/* How often to retry if there is no response */ +#define ADF_PFVF_MSG_RESP_RETRIES 5 +#define ADF_PFVF_MSG_RESP_TIMEOUT (ADF_PFVF_MSG_ACK_DELAY * \ + ADF_PFVF_MSG_ACK_MAX_RETRY + \ + ADF_PFVF_MSG_COLLISION_DETECT_DELAY) + +/** + * adf_send_vf2pf_msg() - send VF to PF message + * @accel_dev: Pointer to acceleration device + * @msg: Message to send + * + * This function allows the VF to send a message to the PF. + * + * Return: 0 on success, error code otherwise. + */ +int adf_send_vf2pf_msg(struct adf_accel_dev *accel_dev, struct pfvf_message msg) +{ + struct adf_pfvf_ops *pfvf_ops = GET_PFVF_OPS(accel_dev); + u32 pfvf_offset = pfvf_ops->get_vf2pf_offset(0); + + return pfvf_ops->send_msg(accel_dev, msg, pfvf_offset, + &accel_dev->vf.vf2pf_lock); +} + +/** + * adf_recv_pf2vf_msg() - receive a PF to VF message + * @accel_dev: Pointer to acceleration device + * + * This function allows the VF to receive a message from the PF. + * + * Return: a valid message on success, zero otherwise. + */ +static struct pfvf_message adf_recv_pf2vf_msg(struct adf_accel_dev *accel_dev) +{ + struct adf_pfvf_ops *pfvf_ops = GET_PFVF_OPS(accel_dev); + u32 pfvf_offset = pfvf_ops->get_pf2vf_offset(0); + + return pfvf_ops->recv_msg(accel_dev, pfvf_offset, accel_dev->vf.pf_compat_ver); +} + +/** + * adf_send_vf2pf_req() - send VF2PF request message + * @accel_dev: Pointer to acceleration device. + * @msg: Request message to send + * @resp: Returned PF response + * + * This function sends a message that requires a response from the VF to the PF + * and waits for a reply. + * + * Return: 0 on success, error code otherwise. + */ +int adf_send_vf2pf_req(struct adf_accel_dev *accel_dev, struct pfvf_message msg, + struct pfvf_message *resp) +{ + unsigned long timeout = msecs_to_jiffies(ADF_PFVF_MSG_RESP_TIMEOUT); + unsigned int retries = ADF_PFVF_MSG_RESP_RETRIES; + int ret; + + reinit_completion(&accel_dev->vf.msg_received); + + /* Send request from VF to PF */ + do { + ret = adf_send_vf2pf_msg(accel_dev, msg); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to send request msg to PF\n"); + return ret; + } + + /* Wait for response, if it times out retry */ + ret = wait_for_completion_timeout(&accel_dev->vf.msg_received, + timeout); + if (ret) { + if (likely(resp)) + *resp = accel_dev->vf.response; + + /* Once copied, set to an invalid value */ + accel_dev->vf.response.type = 0; + + return 0; + } + + dev_err(&GET_DEV(accel_dev), "PFVF response message timeout\n"); + } while (--retries); + + return -EIO; +} + +static int adf_vf2pf_blkmsg_data_req(struct adf_accel_dev *accel_dev, bool crc, + u8 *type, u8 *data) +{ + struct pfvf_message req = { 0 }; + struct pfvf_message resp = { 0 }; + u8 blk_type; + u8 blk_byte; + u8 msg_type; + u8 max_data; + int err; + + /* Convert the block type to {small, medium, large} size category */ + if (*type <= ADF_VF2PF_SMALL_BLOCK_TYPE_MAX) { + msg_type = ADF_VF2PF_MSGTYPE_SMALL_BLOCK_REQ; + blk_type = FIELD_PREP(ADF_VF2PF_SMALL_BLOCK_TYPE_MASK, *type); + blk_byte = FIELD_PREP(ADF_VF2PF_SMALL_BLOCK_BYTE_MASK, *data); + max_data = ADF_VF2PF_SMALL_BLOCK_BYTE_MAX; + } else if (*type <= ADF_VF2PF_MEDIUM_BLOCK_TYPE_MAX) { + msg_type = ADF_VF2PF_MSGTYPE_MEDIUM_BLOCK_REQ; + blk_type = FIELD_PREP(ADF_VF2PF_MEDIUM_BLOCK_TYPE_MASK, + *type - ADF_VF2PF_SMALL_BLOCK_TYPE_MAX); + blk_byte = FIELD_PREP(ADF_VF2PF_MEDIUM_BLOCK_BYTE_MASK, *data); + max_data = ADF_VF2PF_MEDIUM_BLOCK_BYTE_MAX; + } else if (*type <= ADF_VF2PF_LARGE_BLOCK_TYPE_MAX) { + msg_type = ADF_VF2PF_MSGTYPE_LARGE_BLOCK_REQ; + blk_type = FIELD_PREP(ADF_VF2PF_LARGE_BLOCK_TYPE_MASK, + *type - ADF_VF2PF_MEDIUM_BLOCK_TYPE_MAX); + blk_byte = FIELD_PREP(ADF_VF2PF_LARGE_BLOCK_BYTE_MASK, *data); + max_data = ADF_VF2PF_LARGE_BLOCK_BYTE_MAX; + } else { + dev_err(&GET_DEV(accel_dev), "Invalid message type %u\n", *type); + return -EINVAL; + } + + /* Sanity check */ + if (*data > max_data) { + dev_err(&GET_DEV(accel_dev), + "Invalid byte %s %u for message type %u\n", + crc ? "count" : "index", *data, *type); + return -EINVAL; + } + + /* Build the block message */ + req.type = msg_type; + req.data = blk_type | blk_byte | FIELD_PREP(ADF_VF2PF_BLOCK_CRC_REQ_MASK, crc); + + err = adf_send_vf2pf_req(accel_dev, req, &resp); + if (err) + return err; + + *type = FIELD_GET(ADF_PF2VF_BLKMSG_RESP_TYPE_MASK, resp.data); + *data = FIELD_GET(ADF_PF2VF_BLKMSG_RESP_DATA_MASK, resp.data); + + return 0; +} + +static int adf_vf2pf_blkmsg_get_byte(struct adf_accel_dev *accel_dev, u8 type, + u8 index, u8 *data) +{ + int ret; + + ret = adf_vf2pf_blkmsg_data_req(accel_dev, false, &type, &index); + if (ret < 0) + return ret; + + if (unlikely(type != ADF_PF2VF_BLKMSG_RESP_TYPE_DATA)) { + dev_err(&GET_DEV(accel_dev), + "Unexpected BLKMSG response type %u, byte 0x%x\n", + type, index); + return -EFAULT; + } + + *data = index; + return 0; +} + +static int adf_vf2pf_blkmsg_get_crc(struct adf_accel_dev *accel_dev, u8 type, + u8 bytes, u8 *crc) +{ + int ret; + + /* The count of bytes refers to a length, however shift it to a 0-based + * count to avoid overflows. Thus, a request for 0 bytes is technically + * valid. + */ + --bytes; + + ret = adf_vf2pf_blkmsg_data_req(accel_dev, true, &type, &bytes); + if (ret < 0) + return ret; + + if (unlikely(type != ADF_PF2VF_BLKMSG_RESP_TYPE_CRC)) { + dev_err(&GET_DEV(accel_dev), + "Unexpected CRC BLKMSG response type %u, crc 0x%x\n", + type, bytes); + return -EFAULT; + } + + *crc = bytes; + return 0; +} + +/** + * adf_send_vf2pf_blkmsg_req() - retrieve block message + * @accel_dev: Pointer to acceleration VF device. + * @type: The block message type, see adf_pfvf_msg.h for allowed values + * @buffer: input buffer where to place the received data + * @buffer_len: buffer length as input, the amount of written bytes on output + * + * Request a message of type 'type' over the block message transport. + * This function will send the required amount block message requests and + * return the overall content back to the caller through the provided buffer. + * The buffer should be large enough to contain the requested message type, + * otherwise the response will be truncated. + * + * Return: 0 on success, error code otherwise. + */ +int adf_send_vf2pf_blkmsg_req(struct adf_accel_dev *accel_dev, u8 type, + u8 *buffer, unsigned int *buffer_len) +{ + unsigned int index; + unsigned int msg_len; + int ret; + u8 remote_crc; + u8 local_crc; + + if (unlikely(type > ADF_VF2PF_LARGE_BLOCK_TYPE_MAX)) { + dev_err(&GET_DEV(accel_dev), "Invalid block message type %d\n", + type); + return -EINVAL; + } + + if (unlikely(*buffer_len < ADF_PFVF_BLKMSG_HEADER_SIZE)) { + dev_err(&GET_DEV(accel_dev), + "Buffer size too small for a block message\n"); + return -EINVAL; + } + + ret = adf_vf2pf_blkmsg_get_byte(accel_dev, type, + ADF_PFVF_BLKMSG_VER_BYTE, + &buffer[ADF_PFVF_BLKMSG_VER_BYTE]); + if (unlikely(ret)) + return ret; + + if (unlikely(!buffer[ADF_PFVF_BLKMSG_VER_BYTE])) { + dev_err(&GET_DEV(accel_dev), + "Invalid version 0 received for block request %u", type); + return -EFAULT; + } + + ret = adf_vf2pf_blkmsg_get_byte(accel_dev, type, + ADF_PFVF_BLKMSG_LEN_BYTE, + &buffer[ADF_PFVF_BLKMSG_LEN_BYTE]); + if (unlikely(ret)) + return ret; + + if (unlikely(!buffer[ADF_PFVF_BLKMSG_LEN_BYTE])) { + dev_err(&GET_DEV(accel_dev), + "Invalid size 0 received for block request %u", type); + return -EFAULT; + } + + /* We need to pick the minimum since there is no way to request a + * specific version. As a consequence any scenario is possible: + * - PF has a newer (longer) version which doesn't fit in the buffer + * - VF expects a newer (longer) version, so we must not ask for + * bytes in excess + * - PF and VF share the same version, no problem + */ + msg_len = ADF_PFVF_BLKMSG_HEADER_SIZE + buffer[ADF_PFVF_BLKMSG_LEN_BYTE]; + msg_len = min(*buffer_len, msg_len); + + /* Get the payload */ + for (index = ADF_PFVF_BLKMSG_HEADER_SIZE; index < msg_len; index++) { + ret = adf_vf2pf_blkmsg_get_byte(accel_dev, type, index, + &buffer[index]); + if (unlikely(ret)) + return ret; + } + + ret = adf_vf2pf_blkmsg_get_crc(accel_dev, type, msg_len, &remote_crc); + if (unlikely(ret)) + return ret; + + local_crc = adf_pfvf_calc_blkmsg_crc(buffer, msg_len); + if (unlikely(local_crc != remote_crc)) { + dev_err(&GET_DEV(accel_dev), + "CRC error on msg type %d. Local %02X, remote %02X\n", + type, local_crc, remote_crc); + return -EIO; + } + + *buffer_len = msg_len; + return 0; +} + +static bool adf_handle_pf2vf_msg(struct adf_accel_dev *accel_dev, + struct pfvf_message msg) +{ + switch (msg.type) { + case ADF_PF2VF_MSGTYPE_RESTARTING: + dev_dbg(&GET_DEV(accel_dev), "Restarting message received from PF\n"); + + adf_pf2vf_handle_pf_restarting(accel_dev); + return false; + case ADF_PF2VF_MSGTYPE_RESTARTED: + dev_dbg(&GET_DEV(accel_dev), "Restarted message received from PF\n"); + return true; + case ADF_PF2VF_MSGTYPE_FATAL_ERROR: + dev_err(&GET_DEV(accel_dev), "Fatal error received from PF\n"); + return true; + case ADF_PF2VF_MSGTYPE_VERSION_RESP: + case ADF_PF2VF_MSGTYPE_BLKMSG_RESP: + case ADF_PF2VF_MSGTYPE_RP_RESET_RESP: + dev_dbg(&GET_DEV(accel_dev), + "Response Message received from PF (type 0x%.4x, data 0x%.4x)\n", + msg.type, msg.data); + accel_dev->vf.response = msg; + complete(&accel_dev->vf.msg_received); + return true; + default: + dev_err(&GET_DEV(accel_dev), + "Unknown message from PF (type 0x%.4x, data: 0x%.4x)\n", + msg.type, msg.data); + } + + return false; +} + +bool adf_recv_and_handle_pf2vf_msg(struct adf_accel_dev *accel_dev) +{ + struct pfvf_message msg; + + msg = adf_recv_pf2vf_msg(accel_dev); + if (msg.type) /* Invalid or no message */ + return adf_handle_pf2vf_msg(accel_dev, msg); + + /* No replies for PF->VF messages at present */ + + return true; +} + +/** + * adf_enable_vf2pf_comms() - Function enables communication from vf to pf + * + * @accel_dev: Pointer to acceleration device virtual function. + * + * Return: 0 on success, error code otherwise. + */ +int adf_enable_vf2pf_comms(struct adf_accel_dev *accel_dev) +{ + int ret; + + adf_pfvf_crc_init(); + adf_enable_pf2vf_interrupts(accel_dev); + + ret = adf_vf2pf_request_version(accel_dev); + if (ret) + return ret; + + ret = adf_vf2pf_get_capabilities(accel_dev); + if (ret) + return ret; + + ret = adf_vf2pf_get_ring_to_svc(accel_dev); + + return ret; +} +EXPORT_SYMBOL_GPL(adf_enable_vf2pf_comms); diff --git a/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_proto.h b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_proto.h new file mode 100644 index 000000000000..f6ee9b38c0e1 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pfvf_vf_proto.h @@ -0,0 +1,17 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2021 Intel Corporation */ +#ifndef ADF_PFVF_VF_PROTO_H +#define ADF_PFVF_VF_PROTO_H + +#include <linux/types.h> +#include "adf_accel_devices.h" + +int adf_send_vf2pf_msg(struct adf_accel_dev *accel_dev, struct pfvf_message msg); +int adf_send_vf2pf_req(struct adf_accel_dev *accel_dev, struct pfvf_message msg, + struct pfvf_message *resp); +int adf_send_vf2pf_blkmsg_req(struct adf_accel_dev *accel_dev, u8 type, + u8 *buffer, unsigned int *buffer_len); + +int adf_enable_vf2pf_comms(struct adf_accel_dev *accel_dev); + +#endif /* ADF_PFVF_VF_PROTO_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs.c b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs.c new file mode 100644 index 000000000000..f0a13c190196 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs.c @@ -0,0 +1,48 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ +#include <linux/debugfs.h> +#include <linux/fs.h> +#include <linux/kernel.h> + +#include "adf_accel_devices.h" +#include "adf_pm_dbgfs.h" + +static ssize_t pm_status_read(struct file *f, char __user *buf, size_t count, + loff_t *pos) +{ + struct adf_accel_dev *accel_dev = file_inode(f)->i_private; + struct adf_pm pm = accel_dev->power_management; + + if (pm.print_pm_status) + return pm.print_pm_status(accel_dev, buf, count, pos); + + return count; +} + +static const struct file_operations pm_status_fops = { + .owner = THIS_MODULE, + .read = pm_status_read, +}; + +void adf_pm_dbgfs_add(struct adf_accel_dev *accel_dev) +{ + struct adf_pm *pm = &accel_dev->power_management; + + if (!pm->present || !pm->print_pm_status) + return; + + pm->debugfs_pm_status = debugfs_create_file("pm_status", 0400, + accel_dev->debugfs_dir, + accel_dev, &pm_status_fops); +} + +void adf_pm_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ + struct adf_pm *pm = &accel_dev->power_management; + + if (!pm->present) + return; + + debugfs_remove(pm->debugfs_pm_status); + pm->debugfs_pm_status = NULL; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs.h b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs.h new file mode 100644 index 000000000000..83632e5aa097 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs.h @@ -0,0 +1,12 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_PM_DBGFS_H_ +#define ADF_PM_DBGFS_H_ + +struct adf_accel_dev; + +void adf_pm_dbgfs_rm(struct adf_accel_dev *accel_dev); +void adf_pm_dbgfs_add(struct adf_accel_dev *accel_dev); + +#endif /* ADF_PM_DBGFS_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs_utils.c b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs_utils.c new file mode 100644 index 000000000000..4ccc94ed9493 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs_utils.c @@ -0,0 +1,46 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2025 Intel Corporation */ +#include <linux/bitfield.h> +#include <linux/bitops.h> +#include <linux/sprintf.h> +#include <linux/string_helpers.h> + +#include "adf_pm_dbgfs_utils.h" + +#define PM_INFO_MAX_KEY_LEN 21 + +static int pm_scnprint_table(char *buff, const struct pm_status_row *table, + u32 *pm_info_regs, size_t buff_size, int table_len, + bool lowercase) +{ + char key[PM_INFO_MAX_KEY_LEN]; + int wr = 0; + int i; + + for (i = 0; i < table_len; i++) { + if (lowercase) + string_lower(key, table[i].key); + else + string_upper(key, table[i].key); + + wr += scnprintf(&buff[wr], buff_size - wr, "%s: %#x\n", key, + field_get(table[i].field_mask, + pm_info_regs[table[i].reg_offset])); + } + + return wr; +} + +int adf_pm_scnprint_table_upper_keys(char *buff, const struct pm_status_row *table, + u32 *pm_info_regs, size_t buff_size, int table_len) +{ + return pm_scnprint_table(buff, table, pm_info_regs, buff_size, + table_len, false); +} + +int adf_pm_scnprint_table_lower_keys(char *buff, const struct pm_status_row *table, + u32 *pm_info_regs, size_t buff_size, int table_len) +{ + return pm_scnprint_table(buff, table, pm_info_regs, buff_size, + table_len, true); +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs_utils.h b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs_utils.h new file mode 100644 index 000000000000..854f058b35ed --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_pm_dbgfs_utils.h @@ -0,0 +1,36 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ADF_PM_DBGFS_UTILS_H_ +#define ADF_PM_DBGFS_UTILS_H_ + +#include <linux/stddef.h> +#include <linux/stringify.h> +#include <linux/types.h> +#include "icp_qat_fw_init_admin.h" + +#define PM_INFO_MEMBER_OFF(member) \ + (offsetof(struct icp_qat_fw_init_admin_pm_info, member) / sizeof(u32)) + +#define PM_INFO_REGSET_ENTRY_MASK(_reg_, _field_, _mask_) \ +{ \ + .reg_offset = PM_INFO_MEMBER_OFF(_reg_), \ + .key = __stringify(_field_), \ + .field_mask = _mask_, \ +} + +#define PM_INFO_REGSET_ENTRY32(_reg_, _field_) \ + PM_INFO_REGSET_ENTRY_MASK(_reg_, _field_, GENMASK(31, 0)) + +struct pm_status_row { + int reg_offset; + u32 field_mask; + const char *key; +}; + +int adf_pm_scnprint_table_upper_keys(char *buff, const struct pm_status_row *table, + u32 *pm_info_regs, size_t buff_size, int table_len); + +int adf_pm_scnprint_table_lower_keys(char *buff, const struct pm_status_row *table, + u32 *pm_info_regs, size_t buff_size, int table_len); + +#endif /* ADF_PM_DBGFS_UTILS_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_rl.c b/drivers/crypto/intel/qat/qat_common/adf_rl.c new file mode 100644 index 000000000000..c6a54e465931 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_rl.c @@ -0,0 +1,1155 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#define dev_fmt(fmt) "RateLimiting: " fmt + +#include <asm/errno.h> +#include <asm/div64.h> + +#include <linux/dev_printk.h> +#include <linux/kernel.h> +#include <linux/pci.h> +#include <linux/slab.h> +#include <linux/units.h> + +#include "adf_accel_devices.h" +#include "adf_cfg_services.h" +#include "adf_common_drv.h" +#include "adf_rl_admin.h" +#include "adf_rl.h" +#include "adf_sysfs_rl.h" + +#define RL_TOKEN_GRANULARITY_PCIEIN_BUCKET 0U +#define RL_TOKEN_GRANULARITY_PCIEOUT_BUCKET 0U +#define RL_TOKEN_PCIE_SIZE 64 +#define RL_TOKEN_ASYM_SIZE 1024 +#define RL_CSR_SIZE 4U +#define RL_CAPABILITY_MASK GENMASK(6, 4) +#define RL_CAPABILITY_VALUE 0x70 +#define RL_VALIDATE_NON_ZERO(input) ((input) == 0) +#define ROOT_MASK GENMASK(1, 0) +#define CLUSTER_MASK GENMASK(3, 0) +#define LEAF_MASK GENMASK(5, 0) + +static int validate_user_input(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in, + bool is_update) +{ + const unsigned long rp_mask = sla_in->rp_mask; + size_t rp_mask_size; + int i, cnt; + + if (sla_in->pir < sla_in->cir) { + dev_notice(&GET_DEV(accel_dev), + "PIR must be >= CIR, setting PIR to CIR\n"); + sla_in->pir = sla_in->cir; + } + + if (!is_update) { + cnt = 0; + rp_mask_size = sizeof(sla_in->rp_mask) * BITS_PER_BYTE; + for_each_set_bit(i, &rp_mask, rp_mask_size) { + if (++cnt > RL_RP_CNT_PER_LEAF_MAX) { + dev_notice(&GET_DEV(accel_dev), + "Too many ring pairs selected for this SLA\n"); + return -EINVAL; + } + } + + if (sla_in->srv >= SVC_BASE_COUNT) { + dev_notice(&GET_DEV(accel_dev), + "Wrong service type\n"); + return -EINVAL; + } + + if (sla_in->type > RL_LEAF) { + dev_notice(&GET_DEV(accel_dev), + "Wrong node type\n"); + return -EINVAL; + } + + if (sla_in->parent_id < RL_PARENT_DEFAULT_ID || + sla_in->parent_id >= RL_NODES_CNT_MAX) { + dev_notice(&GET_DEV(accel_dev), + "Wrong parent ID\n"); + return -EINVAL; + } + } + + return 0; +} + +static int validate_sla_id(struct adf_accel_dev *accel_dev, int sla_id) +{ + struct rl_sla *sla; + + if (sla_id <= RL_SLA_EMPTY_ID || sla_id >= RL_NODES_CNT_MAX) { + dev_notice(&GET_DEV(accel_dev), "Provided ID is out of bounds\n"); + return -EINVAL; + } + + sla = accel_dev->rate_limiting->sla[sla_id]; + + if (!sla) { + dev_notice(&GET_DEV(accel_dev), "SLA with provided ID does not exist\n"); + return -EINVAL; + } + + if (sla->type != RL_LEAF) { + dev_notice(&GET_DEV(accel_dev), "This ID is reserved for internal use\n"); + return -EINVAL; + } + + return 0; +} + +/** + * find_parent() - Find the parent for a new SLA + * @rl_data: pointer to ratelimiting data + * @sla_in: pointer to user input data for a new SLA + * + * Function returns a pointer to the parent SLA. If the parent ID is provided + * as input in the user data, then such ID is validated and the parent SLA + * is returned. + * Otherwise, it returns the default parent SLA (root or cluster) for + * the new object. + * + * Return: + * * Pointer to the parent SLA object + * * NULL - when parent cannot be found + */ +static struct rl_sla *find_parent(struct adf_rl *rl_data, + struct adf_rl_sla_input_data *sla_in) +{ + int input_parent_id = sla_in->parent_id; + struct rl_sla *root = NULL; + struct rl_sla *parent_sla; + int i; + + if (sla_in->type == RL_ROOT) + return NULL; + + if (input_parent_id > RL_PARENT_DEFAULT_ID) { + parent_sla = rl_data->sla[input_parent_id]; + /* + * SLA can be a parent if it has the same service as the child + * and its type is higher in the hierarchy, + * for example the parent type of a LEAF must be a CLUSTER. + */ + if (parent_sla && parent_sla->srv == sla_in->srv && + parent_sla->type == sla_in->type - 1) + return parent_sla; + + return NULL; + } + + /* If input_parent_id is not valid, get root for this service type. */ + for (i = 0; i < RL_ROOT_MAX; i++) { + if (rl_data->root[i] && rl_data->root[i]->srv == sla_in->srv) { + root = rl_data->root[i]; + break; + } + } + + if (!root) + return NULL; + + /* + * If the type of this SLA is cluster, then return the root. + * Otherwise, find the default (i.e. first) cluster for this service. + */ + if (sla_in->type == RL_CLUSTER) + return root; + + for (i = 0; i < RL_CLUSTER_MAX; i++) { + if (rl_data->cluster[i] && rl_data->cluster[i]->parent == root) + return rl_data->cluster[i]; + } + + return NULL; +} + +/** + * adf_rl_get_sla_arr_of_type() - Returns a pointer to SLA type specific array + * @rl_data: pointer to ratelimiting data + * @type: SLA type + * @sla_arr: pointer to variable where requested pointer will be stored + * + * Return: Max number of elements allowed for the returned array + */ +u32 adf_rl_get_sla_arr_of_type(struct adf_rl *rl_data, enum rl_node_type type, + struct rl_sla ***sla_arr) +{ + switch (type) { + case RL_LEAF: + *sla_arr = rl_data->leaf; + return RL_LEAF_MAX; + case RL_CLUSTER: + *sla_arr = rl_data->cluster; + return RL_CLUSTER_MAX; + case RL_ROOT: + *sla_arr = rl_data->root; + return RL_ROOT_MAX; + default: + *sla_arr = NULL; + return 0; + } +} + +/** + * prepare_rp_ids() - Creates an array of ring pair IDs from bitmask + * @accel_dev: pointer to acceleration device structure + * @sla: SLA object data where result will be written + * @rp_mask: bitmask of ring pair IDs + * + * Function tries to convert provided bitmap to an array of IDs. It checks if + * RPs aren't in use, are assigned to SLA service or if a number of provided + * IDs is not too big. If successful, writes the result into the field + * sla->ring_pairs_cnt. + * + * Return: + * * 0 - ok + * * -EINVAL - ring pairs array cannot be created from provided mask + */ +static int prepare_rp_ids(struct adf_accel_dev *accel_dev, struct rl_sla *sla, + const unsigned long rp_mask) +{ + enum adf_cfg_service_type arb_srv = adf_srv_to_cfg_svc_type(sla->srv); + u16 rps_per_bundle = GET_HW_DATA(accel_dev)->num_banks_per_vf; + bool *rp_in_use = accel_dev->rate_limiting->rp_in_use; + size_t rp_cnt_max = ARRAY_SIZE(sla->ring_pairs_ids); + u16 rp_id_max = GET_HW_DATA(accel_dev)->num_banks; + u16 cnt = 0; + u16 rp_id; + + for_each_set_bit(rp_id, &rp_mask, rp_id_max) { + if (cnt >= rp_cnt_max) { + dev_notice(&GET_DEV(accel_dev), + "Assigned more ring pairs than supported"); + return -EINVAL; + } + + if (rp_in_use[rp_id]) { + dev_notice(&GET_DEV(accel_dev), + "RP %u already assigned to other SLA", rp_id); + return -EINVAL; + } + + if (GET_SRV_TYPE(accel_dev, rp_id % rps_per_bundle) != arb_srv) { + dev_notice(&GET_DEV(accel_dev), + "RP %u does not support SLA service", rp_id); + return -EINVAL; + } + + sla->ring_pairs_ids[cnt++] = rp_id; + } + + sla->ring_pairs_cnt = cnt; + + return 0; +} + +static void mark_rps_usage(struct rl_sla *sla, bool *rp_in_use, bool used) +{ + u16 rp_id; + int i; + + for (i = 0; i < sla->ring_pairs_cnt; i++) { + rp_id = sla->ring_pairs_ids[i]; + rp_in_use[rp_id] = used; + } +} + +static void assign_rps_to_leaf(struct adf_accel_dev *accel_dev, + struct rl_sla *sla, bool clear) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u32 base_offset = hw_data->rl_data.r2l_offset; + u32 node_id = clear ? 0U : (sla->node_id & LEAF_MASK); + u32 offset; + int i; + + for (i = 0; i < sla->ring_pairs_cnt; i++) { + offset = base_offset + (RL_CSR_SIZE * sla->ring_pairs_ids[i]); + ADF_CSR_WR(pmisc_addr, offset, node_id); + } +} + +static void assign_leaf_to_cluster(struct adf_accel_dev *accel_dev, + struct rl_sla *sla, bool clear) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u32 base_offset = hw_data->rl_data.l2c_offset; + u32 node_id = sla->node_id & LEAF_MASK; + u32 parent_id = clear ? 0U : (sla->parent->node_id & CLUSTER_MASK); + u32 offset; + + offset = base_offset + (RL_CSR_SIZE * node_id); + ADF_CSR_WR(pmisc_addr, offset, parent_id); +} + +static void assign_cluster_to_root(struct adf_accel_dev *accel_dev, + struct rl_sla *sla, bool clear) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u32 base_offset = hw_data->rl_data.c2s_offset; + u32 node_id = sla->node_id & CLUSTER_MASK; + u32 parent_id = clear ? 0U : (sla->parent->node_id & ROOT_MASK); + u32 offset; + + offset = base_offset + (RL_CSR_SIZE * node_id); + ADF_CSR_WR(pmisc_addr, offset, parent_id); +} + +static void assign_node_to_parent(struct adf_accel_dev *accel_dev, + struct rl_sla *sla, bool clear_assignment) +{ + switch (sla->type) { + case RL_LEAF: + assign_rps_to_leaf(accel_dev, sla, clear_assignment); + assign_leaf_to_cluster(accel_dev, sla, clear_assignment); + break; + case RL_CLUSTER: + assign_cluster_to_root(accel_dev, sla, clear_assignment); + break; + default: + break; + } +} + +/** + * can_parent_afford_sla() - Verifies if parent allows to create an SLA + * @sla_in: pointer to user input data for a new SLA + * @sla_parent: pointer to parent SLA object + * @sla_cir: current child CIR value (only for update) + * @is_update: request is a update + * + * Algorithm verifies if parent has enough remaining budget to take assignment + * of a child with provided parameters. In update case current CIR value must be + * returned to budget first. + * PIR value cannot exceed the PIR assigned to parent. + * + * Return: + * * true - SLA can be created + * * false - SLA cannot be created + */ +static bool can_parent_afford_sla(struct adf_rl_sla_input_data *sla_in, + struct rl_sla *sla_parent, u32 sla_cir, + bool is_update) +{ + u32 rem_cir = sla_parent->rem_cir; + + if (is_update) + rem_cir += sla_cir; + + if (sla_in->cir > rem_cir || sla_in->pir > sla_parent->pir) + return false; + + return true; +} + +/** + * can_node_afford_update() - Verifies if SLA can be updated with input data + * @sla_in: pointer to user input data for a new SLA + * @sla: pointer to SLA object selected for update + * + * Algorithm verifies if a new CIR value is big enough to satisfy currently + * assigned child SLAs and if PIR can be updated + * + * Return: + * * true - SLA can be updated + * * false - SLA cannot be updated + */ +static bool can_node_afford_update(struct adf_rl_sla_input_data *sla_in, + struct rl_sla *sla) +{ + u32 cir_in_use = sla->cir - sla->rem_cir; + + /* new CIR cannot be smaller then currently consumed value */ + if (cir_in_use > sla_in->cir) + return false; + + /* PIR of root/cluster cannot be reduced in node with assigned children */ + if (sla_in->pir < sla->pir && sla->type != RL_LEAF && cir_in_use > 0) + return false; + + return true; +} + +static bool is_enough_budget(struct adf_rl *rl_data, struct rl_sla *sla, + struct adf_rl_sla_input_data *sla_in, + bool is_update) +{ + u32 max_val = rl_data->device_data->scale_ref; + struct rl_sla *parent = sla->parent; + bool ret = true; + + if (sla_in->cir > max_val || sla_in->pir > max_val) + ret = false; + + switch (sla->type) { + case RL_LEAF: + ret &= can_parent_afford_sla(sla_in, parent, sla->cir, + is_update); + break; + case RL_CLUSTER: + ret &= can_parent_afford_sla(sla_in, parent, sla->cir, + is_update); + + if (is_update) + ret &= can_node_afford_update(sla_in, sla); + + break; + case RL_ROOT: + if (is_update) + ret &= can_node_afford_update(sla_in, sla); + + break; + default: + ret = false; + break; + } + + return ret; +} + +static void update_budget(struct rl_sla *sla, u32 old_cir, bool is_update) +{ + switch (sla->type) { + case RL_LEAF: + if (is_update) + sla->parent->rem_cir += old_cir; + + sla->parent->rem_cir -= sla->cir; + sla->rem_cir = 0; + break; + case RL_CLUSTER: + if (is_update) { + sla->parent->rem_cir += old_cir; + sla->rem_cir = sla->cir - (old_cir - sla->rem_cir); + } else { + sla->rem_cir = sla->cir; + } + + sla->parent->rem_cir -= sla->cir; + break; + case RL_ROOT: + if (is_update) + sla->rem_cir = sla->cir - (old_cir - sla->rem_cir); + else + sla->rem_cir = sla->cir; + break; + default: + break; + } +} + +/** + * get_next_free_sla_id() - finds next free ID in the SLA array + * @rl_data: Pointer to ratelimiting data structure + * + * Return: + * * 0 : RL_NODES_CNT_MAX - correct ID + * * -ENOSPC - all SLA slots are in use + */ +static int get_next_free_sla_id(struct adf_rl *rl_data) +{ + int i = 0; + + while (i < RL_NODES_CNT_MAX && rl_data->sla[i++]) + ; + + if (i == RL_NODES_CNT_MAX) + return -ENOSPC; + + return i - 1; +} + +/** + * get_next_free_node_id() - finds next free ID in the array of that node type + * @rl_data: Pointer to ratelimiting data structure + * @sla: Pointer to SLA object for which the ID is searched + * + * Return: + * * 0 : RL_[NODE_TYPE]_MAX - correct ID + * * -ENOSPC - all slots of that type are in use + */ +static int get_next_free_node_id(struct adf_rl *rl_data, struct rl_sla *sla) +{ + struct adf_hw_device_data *hw_device = GET_HW_DATA(rl_data->accel_dev); + int max_id, i, step, rp_per_leaf; + struct rl_sla **sla_list; + + rp_per_leaf = hw_device->num_banks / hw_device->num_banks_per_vf; + + /* + * Static nodes mapping: + * root0 - cluster[0,4,8,12] - leaf[0-15] + * root1 - cluster[1,5,9,13] - leaf[16-31] + * root2 - cluster[2,6,10,14] - leaf[32-47] + */ + switch (sla->type) { + case RL_LEAF: + i = sla->srv * rp_per_leaf; + step = 1; + max_id = i + rp_per_leaf; + sla_list = rl_data->leaf; + break; + case RL_CLUSTER: + i = sla->srv; + step = 4; + max_id = RL_CLUSTER_MAX; + sla_list = rl_data->cluster; + break; + case RL_ROOT: + return sla->srv; + default: + return -EINVAL; + } + + while (i < max_id && sla_list[i]) + i += step; + + if (i >= max_id) + return -ENOSPC; + + return i; +} + +u32 adf_rl_calculate_slice_tokens(struct adf_accel_dev *accel_dev, u32 sla_val, + enum adf_base_services svc_type) +{ + struct adf_rl_hw_data *device_data = &accel_dev->hw_device->rl_data; + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + u64 avail_slice_cycles, allocated_tokens; + + if (!sla_val) + return 0; + + /* Handle generation specific slice count adjustment */ + avail_slice_cycles = hw_data->clock_frequency; + avail_slice_cycles *= hw_data->get_svc_slice_cnt(accel_dev, svc_type); + + do_div(avail_slice_cycles, device_data->scan_interval); + allocated_tokens = avail_slice_cycles * sla_val; + do_div(allocated_tokens, device_data->scale_ref); + + return allocated_tokens; +} + +static u32 adf_rl_get_num_svc_aes(struct adf_accel_dev *accel_dev, + enum adf_base_services svc) +{ + struct adf_rl_hw_data *device_data = &accel_dev->hw_device->rl_data; + + if (svc >= SVC_BASE_COUNT) + return 0; + + return device_data->svc_ae_mask[svc]; +} + +u32 adf_rl_calculate_ae_cycles(struct adf_accel_dev *accel_dev, u32 sla_val, + enum adf_base_services svc_type) +{ + struct adf_rl_hw_data *device_data = &accel_dev->hw_device->rl_data; + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + u64 allocated_ae_cycles, avail_ae_cycles; + + if (!sla_val) + return 0; + + avail_ae_cycles = hw_data->clock_frequency; + avail_ae_cycles *= adf_rl_get_num_svc_aes(accel_dev, svc_type); + do_div(avail_ae_cycles, device_data->scan_interval); + + sla_val *= device_data->max_tp[svc_type]; + sla_val /= device_data->scale_ref; + + allocated_ae_cycles = (sla_val * avail_ae_cycles); + do_div(allocated_ae_cycles, device_data->max_tp[svc_type]); + + return allocated_ae_cycles; +} + +u32 adf_rl_calculate_pci_bw(struct adf_accel_dev *accel_dev, u32 sla_val, + enum adf_base_services svc_type, bool is_bw_out) +{ + struct adf_rl_hw_data *device_data = &accel_dev->hw_device->rl_data; + u64 sla_to_bytes, allocated_bw, sla_scaled; + + if (!sla_val) + return 0; + + sla_to_bytes = sla_val; + sla_to_bytes *= device_data->max_tp[svc_type]; + do_div(sla_to_bytes, device_data->scale_ref); + + sla_to_bytes *= (svc_type == SVC_ASYM) ? RL_TOKEN_ASYM_SIZE : BYTES_PER_MBIT; + if (svc_type == SVC_DC && is_bw_out) + sla_to_bytes *= device_data->slices.dcpr_cnt - + device_data->dcpr_correction; + + sla_scaled = sla_to_bytes * device_data->pcie_scale_mul; + do_div(sla_scaled, device_data->pcie_scale_div); + allocated_bw = sla_scaled; + do_div(allocated_bw, RL_TOKEN_PCIE_SIZE); + do_div(allocated_bw, device_data->scan_interval); + + return allocated_bw; +} + +/** + * add_new_sla_entry() - creates a new SLA object and fills it with user data + * @accel_dev: pointer to acceleration device structure + * @sla_in: pointer to user input data for a new SLA + * @sla_out: Pointer to variable that will contain the address of a new + * SLA object if the operation succeeds + * + * Return: + * * 0 - ok + * * -ENOMEM - memory allocation failed + * * -EINVAL - invalid user input + * * -ENOSPC - all available SLAs are in use + */ +static int add_new_sla_entry(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in, + struct rl_sla **sla_out) +{ + struct adf_rl *rl_data = accel_dev->rate_limiting; + struct rl_sla *sla; + int ret = 0; + + sla = kzalloc(sizeof(*sla), GFP_KERNEL); + if (!sla) { + ret = -ENOMEM; + goto ret_err; + } + *sla_out = sla; + + if (!adf_is_service_enabled(accel_dev, sla_in->srv)) { + dev_notice(&GET_DEV(accel_dev), + "Provided service is not enabled\n"); + ret = -EINVAL; + goto ret_err; + } + + sla->srv = sla_in->srv; + sla->type = sla_in->type; + ret = get_next_free_node_id(rl_data, sla); + if (ret < 0) { + dev_notice(&GET_DEV(accel_dev), + "Exceeded number of available nodes for that service\n"); + goto ret_err; + } + sla->node_id = ret; + + ret = get_next_free_sla_id(rl_data); + if (ret < 0) { + dev_notice(&GET_DEV(accel_dev), + "Allocated maximum SLAs number\n"); + goto ret_err; + } + sla->sla_id = ret; + + sla->parent = find_parent(rl_data, sla_in); + if (!sla->parent && sla->type != RL_ROOT) { + if (sla_in->parent_id != RL_PARENT_DEFAULT_ID) + dev_notice(&GET_DEV(accel_dev), + "Provided parent ID does not exist or cannot be parent for this SLA."); + else + dev_notice(&GET_DEV(accel_dev), + "Unable to find parent node for this service. Is service enabled?"); + ret = -EINVAL; + goto ret_err; + } + + if (sla->type == RL_LEAF) { + ret = prepare_rp_ids(accel_dev, sla, sla_in->rp_mask); + if (!sla->ring_pairs_cnt || ret) { + dev_notice(&GET_DEV(accel_dev), + "Unable to find ring pairs to assign to the leaf"); + if (!ret) + ret = -EINVAL; + + goto ret_err; + } + } + + return 0; + +ret_err: + kfree(sla); + *sla_out = NULL; + + return ret; +} + +static int initialize_default_nodes(struct adf_accel_dev *accel_dev) +{ + struct adf_rl *rl_data = accel_dev->rate_limiting; + struct adf_rl_hw_data *device_data = rl_data->device_data; + struct adf_rl_sla_input_data sla_in = { }; + int ret = 0; + int i; + + /* Init root for each enabled service */ + sla_in.type = RL_ROOT; + sla_in.parent_id = RL_PARENT_DEFAULT_ID; + + for (i = 0; i < SVC_BASE_COUNT; i++) { + if (!adf_is_service_enabled(accel_dev, i)) + continue; + + sla_in.cir = device_data->scale_ref; + sla_in.pir = sla_in.cir; + sla_in.srv = i; + + ret = adf_rl_add_sla(accel_dev, &sla_in); + if (ret) + return ret; + } + + /* Init default cluster for each root */ + sla_in.type = RL_CLUSTER; + for (i = 0; i < SVC_BASE_COUNT; i++) { + if (!rl_data->root[i]) + continue; + sla_in.cir = rl_data->root[i]->cir; + sla_in.pir = sla_in.cir; + sla_in.srv = rl_data->root[i]->srv; + + ret = adf_rl_add_sla(accel_dev, &sla_in); + if (ret) + return ret; + } + + return 0; +} + +static void clear_sla(struct adf_rl *rl_data, struct rl_sla *sla) +{ + bool *rp_in_use = rl_data->rp_in_use; + struct rl_sla **sla_type_arr = NULL; + int i, sla_id, node_id; + u32 old_cir; + + sla_id = sla->sla_id; + node_id = sla->node_id; + old_cir = sla->cir; + sla->cir = 0; + sla->pir = 0; + + for (i = 0; i < sla->ring_pairs_cnt; i++) + rp_in_use[sla->ring_pairs_ids[i]] = false; + + update_budget(sla, old_cir, true); + adf_rl_get_sla_arr_of_type(rl_data, sla->type, &sla_type_arr); + assign_node_to_parent(rl_data->accel_dev, sla, true); + adf_rl_send_admin_delete_msg(rl_data->accel_dev, node_id, sla->type); + mark_rps_usage(sla, rl_data->rp_in_use, false); + + kfree(sla); + rl_data->sla[sla_id] = NULL; + sla_type_arr[node_id] = NULL; +} + +static void free_all_sla(struct adf_accel_dev *accel_dev) +{ + struct adf_rl *rl_data = accel_dev->rate_limiting; + int sla_id; + + mutex_lock(&rl_data->rl_lock); + + for (sla_id = 0; sla_id < RL_NODES_CNT_MAX; sla_id++) { + if (!rl_data->sla[sla_id]) + continue; + + kfree(rl_data->sla[sla_id]); + rl_data->sla[sla_id] = NULL; + } + + mutex_unlock(&rl_data->rl_lock); +} + +/** + * add_update_sla() - handles the creation and the update of an SLA + * @accel_dev: pointer to acceleration device structure + * @sla_in: pointer to user input data for a new/updated SLA + * @is_update: flag to indicate if this is an update or an add operation + * + * Return: + * * 0 - ok + * * -ENOMEM - memory allocation failed + * * -EINVAL - user input data cannot be used to create SLA + * * -ENOSPC - all available SLAs are in use + */ +static int add_update_sla(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in, bool is_update) +{ + struct adf_rl *rl_data = accel_dev->rate_limiting; + struct rl_sla **sla_type_arr = NULL; + struct rl_sla *sla = NULL; + u32 old_cir = 0; + int ret; + + if (!sla_in) { + dev_warn(&GET_DEV(accel_dev), + "SLA input data pointer is missing\n"); + return -EFAULT; + } + + mutex_lock(&rl_data->rl_lock); + + /* Input validation */ + ret = validate_user_input(accel_dev, sla_in, is_update); + if (ret) + goto ret_err; + + if (is_update) { + ret = validate_sla_id(accel_dev, sla_in->sla_id); + if (ret) + goto ret_err; + + sla = rl_data->sla[sla_in->sla_id]; + old_cir = sla->cir; + } else { + ret = add_new_sla_entry(accel_dev, sla_in, &sla); + if (ret) + goto ret_err; + } + + if (!is_enough_budget(rl_data, sla, sla_in, is_update)) { + dev_notice(&GET_DEV(accel_dev), + "Input value exceeds the remaining budget%s\n", + is_update ? " or more budget is already in use" : ""); + ret = -EINVAL; + goto ret_err; + } + sla->cir = sla_in->cir; + sla->pir = sla_in->pir; + + /* Apply SLA */ + assign_node_to_parent(accel_dev, sla, false); + ret = adf_rl_send_admin_add_update_msg(accel_dev, sla, is_update); + if (ret) { + dev_notice(&GET_DEV(accel_dev), + "Failed to apply an SLA\n"); + goto ret_err; + } + update_budget(sla, old_cir, is_update); + + if (!is_update) { + mark_rps_usage(sla, rl_data->rp_in_use, true); + adf_rl_get_sla_arr_of_type(rl_data, sla->type, &sla_type_arr); + sla_type_arr[sla->node_id] = sla; + rl_data->sla[sla->sla_id] = sla; + } + + sla_in->sla_id = sla->sla_id; + goto ret_ok; + +ret_err: + if (!is_update) { + sla_in->sla_id = -1; + kfree(sla); + } +ret_ok: + mutex_unlock(&rl_data->rl_lock); + return ret; +} + +/** + * adf_rl_add_sla() - handles the creation of an SLA + * @accel_dev: pointer to acceleration device structure + * @sla_in: pointer to user input data required to add an SLA + * + * Return: + * * 0 - ok + * * -ENOMEM - memory allocation failed + * * -EINVAL - invalid user input + * * -ENOSPC - all available SLAs are in use + */ +int adf_rl_add_sla(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in) +{ + return add_update_sla(accel_dev, sla_in, false); +} + +/** + * adf_rl_update_sla() - handles the update of an SLA + * @accel_dev: pointer to acceleration device structure + * @sla_in: pointer to user input data required to update an SLA + * + * Return: + * * 0 - ok + * * -EINVAL - user input data cannot be used to update SLA + */ +int adf_rl_update_sla(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in) +{ + return add_update_sla(accel_dev, sla_in, true); +} + +/** + * adf_rl_get_sla() - returns an existing SLA data + * @accel_dev: pointer to acceleration device structure + * @sla_in: pointer to user data where SLA info will be stored + * + * The sla_id for which data are requested should be set in sla_id structure + * + * Return: + * * 0 - ok + * * -EINVAL - provided sla_id does not exist + */ +int adf_rl_get_sla(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in) +{ + struct rl_sla *sla; + int ret, i; + + ret = validate_sla_id(accel_dev, sla_in->sla_id); + if (ret) + return ret; + + sla = accel_dev->rate_limiting->sla[sla_in->sla_id]; + sla_in->type = sla->type; + sla_in->srv = sla->srv; + sla_in->cir = sla->cir; + sla_in->pir = sla->pir; + sla_in->rp_mask = 0U; + if (sla->parent) + sla_in->parent_id = sla->parent->sla_id; + else + sla_in->parent_id = RL_PARENT_DEFAULT_ID; + + for (i = 0; i < sla->ring_pairs_cnt; i++) + sla_in->rp_mask |= BIT(sla->ring_pairs_ids[i]); + + return 0; +} + +/** + * adf_rl_get_capability_remaining() - returns the remaining SLA value (CIR) for + * selected service or provided sla_id + * @accel_dev: pointer to acceleration device structure + * @srv: service ID for which capability is requested + * @sla_id: ID of the cluster or root to which we want assign a new SLA + * + * Check if the provided SLA id is valid. If it is and the service matches + * the requested service and the type is cluster or root, return the remaining + * capability. + * If the provided ID does not match the service or type, return the remaining + * capacity of the default cluster for that service. + * + * Return: + * * Positive value - correct remaining value + * * -EINVAL - algorithm cannot find a remaining value for provided data + */ +int adf_rl_get_capability_remaining(struct adf_accel_dev *accel_dev, + enum adf_base_services srv, int sla_id) +{ + struct adf_rl *rl_data = accel_dev->rate_limiting; + struct rl_sla *sla = NULL; + int i; + + if (srv >= SVC_BASE_COUNT) + return -EINVAL; + + if (sla_id > RL_SLA_EMPTY_ID && !validate_sla_id(accel_dev, sla_id)) { + sla = rl_data->sla[sla_id]; + + if (sla->srv == srv && sla->type <= RL_CLUSTER) + goto ret_ok; + } + + for (i = 0; i < RL_CLUSTER_MAX; i++) { + if (!rl_data->cluster[i]) + continue; + + if (rl_data->cluster[i]->srv == srv) { + sla = rl_data->cluster[i]; + goto ret_ok; + } + } + + return -EINVAL; +ret_ok: + return sla->rem_cir; +} + +/** + * adf_rl_remove_sla() - removes provided sla_id + * @accel_dev: pointer to acceleration device structure + * @sla_id: ID of the cluster or root to which we want assign an new SLA + * + * Return: + * * 0 - ok + * * -EINVAL - wrong sla_id or it still have assigned children + */ +int adf_rl_remove_sla(struct adf_accel_dev *accel_dev, u32 sla_id) +{ + struct adf_rl *rl_data = accel_dev->rate_limiting; + struct rl_sla *sla; + int ret = 0; + + mutex_lock(&rl_data->rl_lock); + ret = validate_sla_id(accel_dev, sla_id); + if (ret) + goto err_ret; + + sla = rl_data->sla[sla_id]; + + if (sla->type < RL_LEAF && sla->rem_cir != sla->cir) { + dev_notice(&GET_DEV(accel_dev), + "To remove parent SLA all its children must be removed first"); + ret = -EINVAL; + goto err_ret; + } + + clear_sla(rl_data, sla); + +err_ret: + mutex_unlock(&rl_data->rl_lock); + return ret; +} + +/** + * adf_rl_remove_sla_all() - removes all SLAs from device + * @accel_dev: pointer to acceleration device structure + * @incl_default: set to true if default SLAs also should be removed + */ +void adf_rl_remove_sla_all(struct adf_accel_dev *accel_dev, bool incl_default) +{ + struct adf_rl *rl_data = accel_dev->rate_limiting; + int end_type = incl_default ? RL_ROOT : RL_LEAF; + struct rl_sla **sla_type_arr = NULL; + u32 max_id; + int i, j; + + mutex_lock(&rl_data->rl_lock); + + /* Unregister and remove all SLAs */ + for (j = RL_LEAF; j >= end_type; j--) { + max_id = adf_rl_get_sla_arr_of_type(rl_data, j, &sla_type_arr); + + for (i = 0; i < max_id; i++) { + if (!sla_type_arr[i]) + continue; + + clear_sla(rl_data, sla_type_arr[i]); + } + } + + mutex_unlock(&rl_data->rl_lock); +} + +int adf_rl_init(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct adf_rl_hw_data *rl_hw_data = &hw_data->rl_data; + struct adf_rl *rl; + int ret = 0; + + /* Validate device parameters */ + if (RL_VALIDATE_NON_ZERO(rl_hw_data->max_tp[SVC_ASYM]) || + RL_VALIDATE_NON_ZERO(rl_hw_data->max_tp[SVC_SYM]) || + RL_VALIDATE_NON_ZERO(rl_hw_data->max_tp[SVC_DC]) || + RL_VALIDATE_NON_ZERO(rl_hw_data->scan_interval) || + RL_VALIDATE_NON_ZERO(rl_hw_data->pcie_scale_div) || + RL_VALIDATE_NON_ZERO(rl_hw_data->pcie_scale_mul) || + RL_VALIDATE_NON_ZERO(rl_hw_data->scale_ref)) { + ret = -EOPNOTSUPP; + goto err_ret; + } + + rl = kzalloc(sizeof(*rl), GFP_KERNEL); + if (!rl) { + ret = -ENOMEM; + goto err_ret; + } + + mutex_init(&rl->rl_lock); + rl->device_data = &accel_dev->hw_device->rl_data; + rl->accel_dev = accel_dev; + init_rwsem(&rl->user_input.lock); + accel_dev->rate_limiting = rl; + +err_ret: + return ret; +} + +int adf_rl_start(struct adf_accel_dev *accel_dev) +{ + struct adf_rl_hw_data *rl_hw_data = &GET_HW_DATA(accel_dev)->rl_data; + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + u16 fw_caps = GET_HW_DATA(accel_dev)->fw_capabilities; + int ret; + + if (!accel_dev->rate_limiting) { + ret = -EOPNOTSUPP; + goto ret_err; + } + + if ((fw_caps & RL_CAPABILITY_MASK) != RL_CAPABILITY_VALUE) { + dev_info(&GET_DEV(accel_dev), "feature not supported by FW\n"); + ret = -EOPNOTSUPP; + goto ret_free; + } + + ADF_CSR_WR(pmisc_addr, rl_hw_data->pciin_tb_offset, + RL_TOKEN_GRANULARITY_PCIEIN_BUCKET); + ADF_CSR_WR(pmisc_addr, rl_hw_data->pciout_tb_offset, + RL_TOKEN_GRANULARITY_PCIEOUT_BUCKET); + + ret = adf_rl_send_admin_init_msg(accel_dev, &rl_hw_data->slices); + if (ret) { + dev_err(&GET_DEV(accel_dev), "initialization failed\n"); + goto ret_free; + } + + ret = initialize_default_nodes(accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "failed to initialize default SLAs\n"); + goto ret_sla_rm; + } + + ret = adf_sysfs_rl_add(accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), "failed to add sysfs interface\n"); + goto ret_sysfs_rm; + } + + return 0; + +ret_sysfs_rm: + adf_sysfs_rl_rm(accel_dev); +ret_sla_rm: + adf_rl_remove_sla_all(accel_dev, true); +ret_free: + kfree(accel_dev->rate_limiting); + accel_dev->rate_limiting = NULL; +ret_err: + return ret; +} + +void adf_rl_stop(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->rate_limiting) + return; + + adf_sysfs_rl_rm(accel_dev); + free_all_sla(accel_dev); +} + +void adf_rl_exit(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->rate_limiting) + return; + + kfree(accel_dev->rate_limiting); + accel_dev->rate_limiting = NULL; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_rl.h b/drivers/crypto/intel/qat/qat_common/adf_rl.h new file mode 100644 index 000000000000..c1f3f9a51195 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_rl.h @@ -0,0 +1,176 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_RL_H_ +#define ADF_RL_H_ + +#include <linux/mutex.h> +#include <linux/types.h> + +#include "adf_cfg_services.h" + +struct adf_accel_dev; + +#define RL_ROOT_MAX 4 +#define RL_CLUSTER_MAX 16 +#define RL_LEAF_MAX 64 +#define RL_NODES_CNT_MAX (RL_ROOT_MAX + RL_CLUSTER_MAX + RL_LEAF_MAX) +#define RL_RP_CNT_PER_LEAF_MAX 4U +#define RL_RP_CNT_MAX 64 +#define RL_SLA_EMPTY_ID -1 +#define RL_PARENT_DEFAULT_ID -1 + +enum rl_node_type { + RL_ROOT, + RL_CLUSTER, + RL_LEAF, +}; + +/** + * struct adf_rl_sla_input_data - ratelimiting user input data structure + * @rp_mask: 64 bit bitmask of ring pair IDs which will be assigned to SLA. + * Eg. 0x5 -> RP0 and RP2 assigned; 0xA005 -> RP0,2,13,15 assigned. + * @sla_id: ID of current SLA for operations update, rm, get. For the add + * operation, this field will be updated with the ID of the newly + * added SLA + * @parent_id: ID of the SLA to which the current one should be assigned. + * Set to -1 to refer to the default parent. + * @cir: Committed information rate. Rate guaranteed to be achieved. Input value + * is expressed in permille scale, i.e. 1000 refers to the maximum + * device throughput for a selected service. + * @pir: Peak information rate. Maximum rate available that the SLA can achieve. + * Input value is expressed in permille scale, i.e. 1000 refers to + * the maximum device throughput for a selected service. + * @type: SLA type: root, cluster, node + * @srv: Service associated to the SLA: asym, sym dc. + * + * This structure is used to perform operations on an SLA. + * Depending on the operation, some of the parameters are ignored. + * The following list reports which parameters should be set for each operation. + * - add: all except sla_id + * - update: cir, pir, sla_id + * - rm: sla_id + * - rm_all: - + * - get: sla_id + * - get_capability_rem: srv, sla_id + */ +struct adf_rl_sla_input_data { + u64 rp_mask; + int sla_id; + int parent_id; + unsigned int cir; + unsigned int pir; + enum rl_node_type type; + enum adf_base_services srv; +}; + +struct rl_slice_cnt { + u8 dcpr_cnt; + u8 pke_cnt; + u8 cph_cnt; + u8 cpr_cnt; +}; + +struct adf_rl_interface_data { + struct adf_rl_sla_input_data input; + enum adf_base_services cap_rem_srv; + struct rw_semaphore lock; + bool sysfs_added; +}; + +struct adf_rl_hw_data { + u32 scale_ref; + u32 scan_interval; + u32 r2l_offset; + u32 l2c_offset; + u32 c2s_offset; + u32 pciin_tb_offset; + u32 pciout_tb_offset; + u32 pcie_scale_mul; + u32 pcie_scale_div; + u32 dcpr_correction; + u32 max_tp[RL_ROOT_MAX]; + u32 svc_ae_mask[SVC_BASE_COUNT]; + struct rl_slice_cnt slices; +}; + +/** + * struct adf_rl - ratelimiting data structure + * @accel_dev: pointer to acceleration device data + * @device_data: pointer to rate limiting data specific to a device type (or revision) + * @sla: array of pointers to SLA objects + * @root: array of pointers to root type SLAs, element number reflects node_id + * @cluster: array of pointers to cluster type SLAs, element number reflects node_id + * @leaf: array of pointers to leaf type SLAs, element number reflects node_id + * @rp_in_use: array of ring pair IDs already used in one of SLAs + * @rl_lock: mutex object which is protecting data in this structure + * @input: structure which is used for holding the data received from user + */ +struct adf_rl { + struct adf_accel_dev *accel_dev; + struct adf_rl_hw_data *device_data; + /* mapping sla_id to SLA objects */ + struct rl_sla *sla[RL_NODES_CNT_MAX]; + struct rl_sla *root[RL_ROOT_MAX]; + struct rl_sla *cluster[RL_CLUSTER_MAX]; + struct rl_sla *leaf[RL_LEAF_MAX]; + bool rp_in_use[RL_RP_CNT_MAX]; + /* Mutex protecting writing to SLAs lists */ + struct mutex rl_lock; + struct adf_rl_interface_data user_input; +}; + +/** + * struct rl_sla - SLA object data structure + * @parent: pointer to the parent SLA (root/cluster) + * @type: SLA type + * @srv: service associated with this SLA + * @sla_id: ID of the SLA, used as element number in SLA array and as identifier + * shared with the user + * @node_id: ID of node, each of SLA type have a separate ID list + * @cir: committed information rate + * @pir: peak information rate (PIR >= CIR) + * @rem_cir: if this SLA is a parent then this field represents a remaining + * value to be used by child SLAs. + * @ring_pairs_ids: array with numeric ring pairs IDs assigned to this SLA + * @ring_pairs_cnt: number of assigned ring pairs listed in the array above + */ +struct rl_sla { + struct rl_sla *parent; + enum rl_node_type type; + enum adf_base_services srv; + u32 sla_id; + u32 node_id; + u32 cir; + u32 pir; + u32 rem_cir; + u16 ring_pairs_ids[RL_RP_CNT_PER_LEAF_MAX]; + u16 ring_pairs_cnt; +}; + +u32 adf_rl_get_sla_arr_of_type(struct adf_rl *rl_data, enum rl_node_type type, + struct rl_sla ***sla_arr); +int adf_rl_add_sla(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in); +int adf_rl_update_sla(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in); +int adf_rl_get_sla(struct adf_accel_dev *accel_dev, + struct adf_rl_sla_input_data *sla_in); +int adf_rl_get_capability_remaining(struct adf_accel_dev *accel_dev, + enum adf_base_services srv, int sla_id); +int adf_rl_remove_sla(struct adf_accel_dev *accel_dev, u32 sla_id); +void adf_rl_remove_sla_all(struct adf_accel_dev *accel_dev, bool incl_default); + +int adf_rl_init(struct adf_accel_dev *accel_dev); +int adf_rl_start(struct adf_accel_dev *accel_dev); +void adf_rl_stop(struct adf_accel_dev *accel_dev); +void adf_rl_exit(struct adf_accel_dev *accel_dev); + +u32 adf_rl_calculate_pci_bw(struct adf_accel_dev *accel_dev, u32 sla_val, + enum adf_base_services svc_type, bool is_bw_out); +u32 adf_rl_calculate_ae_cycles(struct adf_accel_dev *accel_dev, u32 sla_val, + enum adf_base_services svc_type); +u32 adf_rl_calculate_slice_tokens(struct adf_accel_dev *accel_dev, u32 sla_val, + enum adf_base_services svc_type); + +#endif /* ADF_RL_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_rl_admin.c b/drivers/crypto/intel/qat/qat_common/adf_rl_admin.c new file mode 100644 index 000000000000..4a3e0591fdba --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_rl_admin.c @@ -0,0 +1,98 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/dma-mapping.h> +#include <linux/pci.h> + +#include "adf_admin.h" +#include "adf_accel_devices.h" +#include "adf_rl_admin.h" + +static void +prep_admin_req_msg(struct rl_sla *sla, dma_addr_t dma_addr, + struct icp_qat_fw_init_admin_sla_config_params *fw_params, + struct icp_qat_fw_init_admin_req *req, bool is_update) +{ + req->cmd_id = is_update ? ICP_QAT_FW_RL_UPDATE : ICP_QAT_FW_RL_ADD; + req->init_cfg_ptr = dma_addr; + req->init_cfg_sz = sizeof(*fw_params); + req->node_id = sla->node_id; + req->node_type = sla->type; + req->rp_count = sla->ring_pairs_cnt; + req->svc_type = sla->srv; +} + +static void +prep_admin_req_params(struct adf_accel_dev *accel_dev, struct rl_sla *sla, + struct icp_qat_fw_init_admin_sla_config_params *fw_params) +{ + fw_params->pcie_in_cir = + adf_rl_calculate_pci_bw(accel_dev, sla->cir, sla->srv, false); + fw_params->pcie_in_pir = + adf_rl_calculate_pci_bw(accel_dev, sla->pir, sla->srv, false); + fw_params->pcie_out_cir = + adf_rl_calculate_pci_bw(accel_dev, sla->cir, sla->srv, true); + fw_params->pcie_out_pir = + adf_rl_calculate_pci_bw(accel_dev, sla->pir, sla->srv, true); + + fw_params->slice_util_cir = + adf_rl_calculate_slice_tokens(accel_dev, sla->cir, sla->srv); + fw_params->slice_util_pir = + adf_rl_calculate_slice_tokens(accel_dev, sla->pir, sla->srv); + + fw_params->ae_util_cir = + adf_rl_calculate_ae_cycles(accel_dev, sla->cir, sla->srv); + fw_params->ae_util_pir = + adf_rl_calculate_ae_cycles(accel_dev, sla->pir, sla->srv); + + memcpy(fw_params->rp_ids, sla->ring_pairs_ids, + sizeof(sla->ring_pairs_ids)); +} + +int adf_rl_send_admin_init_msg(struct adf_accel_dev *accel_dev, + struct rl_slice_cnt *slices_int) +{ + struct icp_qat_fw_init_admin_slice_cnt slices_resp = { }; + int ret; + + ret = adf_send_admin_rl_init(accel_dev, &slices_resp); + if (ret) + return ret; + + slices_int->dcpr_cnt = slices_resp.dcpr_cnt; + slices_int->pke_cnt = slices_resp.pke_cnt; + /* For symmetric crypto, slice tokens are relative to the UCS slice */ + slices_int->cph_cnt = slices_resp.ucs_cnt; + slices_int->cpr_cnt = slices_resp.cpr_cnt; + + return 0; +} + +int adf_rl_send_admin_add_update_msg(struct adf_accel_dev *accel_dev, + struct rl_sla *sla, bool is_update) +{ + struct icp_qat_fw_init_admin_sla_config_params *fw_params; + struct icp_qat_fw_init_admin_req req = { }; + dma_addr_t dma_addr; + int ret; + + fw_params = dma_alloc_coherent(&GET_DEV(accel_dev), sizeof(*fw_params), + &dma_addr, GFP_KERNEL); + if (!fw_params) + return -ENOMEM; + + prep_admin_req_params(accel_dev, sla, fw_params); + prep_admin_req_msg(sla, dma_addr, fw_params, &req, is_update); + ret = adf_send_admin_rl_add_update(accel_dev, &req); + + dma_free_coherent(&GET_DEV(accel_dev), sizeof(*fw_params), fw_params, + dma_addr); + + return ret; +} + +int adf_rl_send_admin_delete_msg(struct adf_accel_dev *accel_dev, u16 node_id, + u8 node_type) +{ + return adf_send_admin_rl_delete(accel_dev, node_id, node_type); +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_rl_admin.h b/drivers/crypto/intel/qat/qat_common/adf_rl_admin.h new file mode 100644 index 000000000000..dd5419b7e896 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_rl_admin.h @@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_RL_ADMIN_H_ +#define ADF_RL_ADMIN_H_ + +#include <linux/types.h> + +#include "adf_rl.h" + +int adf_rl_send_admin_init_msg(struct adf_accel_dev *accel_dev, + struct rl_slice_cnt *slices_int); +int adf_rl_send_admin_add_update_msg(struct adf_accel_dev *accel_dev, + struct rl_sla *sla, bool is_update); +int adf_rl_send_admin_delete_msg(struct adf_accel_dev *accel_dev, u16 node_id, + u8 node_type); + +#endif /* ADF_RL_ADMIN_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_sriov.c b/drivers/crypto/intel/qat/qat_common/adf_sriov.c new file mode 100644 index 000000000000..bb904ba4bf84 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_sriov.c @@ -0,0 +1,314 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <linux/workqueue.h> +#include <linux/pci.h> +#include <linux/device.h> +#include "adf_common_drv.h" +#include "adf_cfg.h" +#include "adf_pfvf_pf_msg.h" + +#define ADF_VF2PF_RATELIMIT_INTERVAL 8 +#define ADF_VF2PF_RATELIMIT_BURST 130 + +static struct workqueue_struct *pf2vf_resp_wq; + +struct adf_pf2vf_resp { + struct work_struct pf2vf_resp_work; + struct adf_accel_vf_info *vf_info; +}; + +static void adf_iov_send_resp(struct work_struct *work) +{ + struct adf_pf2vf_resp *pf2vf_resp = + container_of(work, struct adf_pf2vf_resp, pf2vf_resp_work); + struct adf_accel_vf_info *vf_info = pf2vf_resp->vf_info; + struct adf_accel_dev *accel_dev = vf_info->accel_dev; + u32 vf_nr = vf_info->vf_nr; + bool ret; + + mutex_lock(&vf_info->pfvf_mig_lock); + ret = adf_recv_and_handle_vf2pf_msg(accel_dev, vf_nr); + if (ret) + /* re-enable interrupt on PF from this VF */ + adf_enable_vf2pf_interrupts(accel_dev, 1 << vf_nr); + mutex_unlock(&vf_info->pfvf_mig_lock); + + kfree(pf2vf_resp); +} + +void adf_schedule_vf2pf_handler(struct adf_accel_vf_info *vf_info) +{ + struct adf_pf2vf_resp *pf2vf_resp; + + pf2vf_resp = kzalloc(sizeof(*pf2vf_resp), GFP_ATOMIC); + if (!pf2vf_resp) + return; + + pf2vf_resp->vf_info = vf_info; + INIT_WORK(&pf2vf_resp->pf2vf_resp_work, adf_iov_send_resp); + queue_work(pf2vf_resp_wq, &pf2vf_resp->pf2vf_resp_work); +} + +static int adf_enable_sriov(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + int totalvfs = pci_sriov_get_totalvfs(pdev); + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_accel_vf_info *vf_info; + int i; + + for (i = 0, vf_info = accel_dev->pf.vf_info; i < totalvfs; + i++, vf_info++) { + /* This ptr will be populated when VFs will be created */ + vf_info->accel_dev = accel_dev; + vf_info->vf_nr = i; + + mutex_init(&vf_info->pf2vf_lock); + mutex_init(&vf_info->pfvf_mig_lock); + ratelimit_state_init(&vf_info->vf2pf_ratelimit, + ADF_VF2PF_RATELIMIT_INTERVAL, + ADF_VF2PF_RATELIMIT_BURST); + } + + /* Set Valid bits in AE Thread to PCIe Function Mapping */ + if (hw_data->configure_iov_threads) + hw_data->configure_iov_threads(accel_dev, true); + + /* Enable VF to PF interrupts for all VFs */ + adf_enable_vf2pf_interrupts(accel_dev, BIT_ULL(totalvfs) - 1); + + /* + * Due to the hardware design, when SR-IOV and the ring arbiter + * are enabled all the VFs supported in hardware must be enabled in + * order for all the hardware resources (i.e. bundles) to be usable. + * When SR-IOV is enabled, each of the VFs will own one bundle. + */ + return pci_enable_sriov(pdev, totalvfs); +} + +static int adf_add_sriov_configuration(struct adf_accel_dev *accel_dev) +{ + unsigned long val = 0; + int ret; + + ret = adf_cfg_section_add(accel_dev, ADF_KERNEL_SEC); + if (ret) + return ret; + + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_CY, + &val, ADF_DEC); + if (ret) + return ret; + + ret = adf_cfg_add_key_value_param(accel_dev, ADF_KERNEL_SEC, ADF_NUM_DC, + &val, ADF_DEC); + if (ret) + return ret; + + set_bit(ADF_STATUS_CONFIGURED, &accel_dev->status); + + return ret; +} + +static int adf_do_disable_sriov(struct adf_accel_dev *accel_dev) +{ + int ret; + + if (adf_dev_in_use(accel_dev)) { + dev_err(&GET_DEV(accel_dev), + "Cannot disable SR-IOV, device in use\n"); + return -EBUSY; + } + + if (adf_dev_started(accel_dev)) { + if (adf_devmgr_in_reset(accel_dev)) { + dev_err(&GET_DEV(accel_dev), + "Cannot disable SR-IOV, device in reset\n"); + return -EBUSY; + } + + ret = adf_dev_down(accel_dev); + if (ret) + goto err_del_cfg; + } + + adf_disable_sriov(accel_dev); + + ret = adf_dev_up(accel_dev, true); + if (ret) + goto err_del_cfg; + + return 0; + +err_del_cfg: + adf_cfg_del_all_except(accel_dev, ADF_GENERAL_SEC); + return ret; +} + +static int adf_do_enable_sriov(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + int totalvfs = pci_sriov_get_totalvfs(pdev); + unsigned long val; + int ret; + + if (!device_iommu_mapped(&GET_DEV(accel_dev))) { + dev_warn(&GET_DEV(accel_dev), + "IOMMU should be enabled for SR-IOV to work correctly\n"); + } + + if (adf_dev_started(accel_dev)) { + if (adf_devmgr_in_reset(accel_dev) || adf_dev_in_use(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Device busy\n"); + return -EBUSY; + } + + ret = adf_dev_down(accel_dev); + if (ret) + return ret; + } + + ret = adf_add_sriov_configuration(accel_dev); + if (ret) + goto err_del_cfg; + + /* Allocate memory for VF info structs */ + accel_dev->pf.vf_info = kcalloc(totalvfs, sizeof(struct adf_accel_vf_info), + GFP_KERNEL); + ret = -ENOMEM; + if (!accel_dev->pf.vf_info) + goto err_del_cfg; + + ret = adf_dev_up(accel_dev, false); + if (ret) { + dev_err(&GET_DEV(accel_dev), "Failed to start qat_dev%d\n", + accel_dev->accel_id); + goto err_free_vf_info; + } + + ret = adf_enable_sriov(accel_dev); + if (ret) + goto err_free_vf_info; + + val = 1; + ret = adf_cfg_add_key_value_param(accel_dev, ADF_GENERAL_SEC, ADF_SRIOV_ENABLED, + &val, ADF_DEC); + if (ret) + goto err_free_vf_info; + + return totalvfs; + +err_free_vf_info: + adf_dev_down(accel_dev); + kfree(accel_dev->pf.vf_info); + accel_dev->pf.vf_info = NULL; + return ret; +err_del_cfg: + adf_cfg_del_all_except(accel_dev, ADF_GENERAL_SEC); + return ret; +} + +void adf_reenable_sriov(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + char cfg[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = {0}; + + if (adf_cfg_get_param_value(accel_dev, ADF_GENERAL_SEC, + ADF_SRIOV_ENABLED, cfg)) + return; + + if (!accel_dev->pf.vf_info) + return; + + if (adf_add_sriov_configuration(accel_dev)) + return; + + dev_dbg(&pdev->dev, "Re-enabling SRIOV\n"); + adf_enable_sriov(accel_dev); +} + +/** + * adf_disable_sriov() - Disable SRIOV for the device + * @accel_dev: Pointer to accel device. + * + * Function disables SRIOV for the accel device. + * + * Return: 0 on success, error code otherwise. + */ +void adf_disable_sriov(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + int totalvfs = pci_sriov_get_totalvfs(accel_to_pci_dev(accel_dev)); + struct adf_accel_vf_info *vf; + int i; + + if (!accel_dev->pf.vf_info) + return; + + adf_pf2vf_notify_restarting(accel_dev); + adf_pf2vf_wait_for_restarting_complete(accel_dev); + pci_disable_sriov(accel_to_pci_dev(accel_dev)); + + /* Disable VF to PF interrupts */ + adf_disable_all_vf2pf_interrupts(accel_dev); + + /* Clear Valid bits in AE Thread to PCIe Function Mapping */ + if (hw_data->configure_iov_threads) + hw_data->configure_iov_threads(accel_dev, false); + + for (i = 0, vf = accel_dev->pf.vf_info; i < totalvfs; i++, vf++) { + mutex_destroy(&vf->pf2vf_lock); + mutex_destroy(&vf->pfvf_mig_lock); + } + + if (!test_bit(ADF_STATUS_RESTARTING, &accel_dev->status)) { + kfree(accel_dev->pf.vf_info); + accel_dev->pf.vf_info = NULL; + } +} +EXPORT_SYMBOL_GPL(adf_disable_sriov); + +/** + * adf_sriov_configure() - Enable SRIOV for the device + * @pdev: Pointer to PCI device. + * @numvfs: Number of virtual functions (VFs) to enable. + * + * Note that the @numvfs parameter is ignored and all VFs supported by the + * device are enabled due to the design of the hardware. + * + * Function enables SRIOV for the PCI device. + * + * Return: number of VFs enabled on success, error code otherwise. + */ +int adf_sriov_configure(struct pci_dev *pdev, int numvfs) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + dev_err(&pdev->dev, "Failed to find accel_dev\n"); + return -EFAULT; + } + + if (numvfs) + return adf_do_enable_sriov(accel_dev); + else + return adf_do_disable_sriov(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_sriov_configure); + +int __init adf_init_pf_wq(void) +{ + /* Workqueue for PF2VF responses */ + pf2vf_resp_wq = alloc_workqueue("qat_pf2vf_resp_wq", + WQ_MEM_RECLAIM | WQ_PERCPU, 0); + + return !pf2vf_resp_wq ? -ENOMEM : 0; +} + +void adf_exit_pf_wq(void) +{ + if (pf2vf_resp_wq) { + destroy_workqueue(pf2vf_resp_wq); + pf2vf_resp_wq = NULL; + } +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_sysfs.c b/drivers/crypto/intel/qat/qat_common/adf_sysfs.c new file mode 100644 index 000000000000..79c63dfa8ff3 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_sysfs.c @@ -0,0 +1,353 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2022 Intel Corporation */ +#include <linux/device.h> +#include <linux/errno.h> +#include <linux/pci.h> +#include <linux/string_choices.h> +#include "adf_accel_devices.h" +#include "adf_cfg.h" +#include "adf_cfg_services.h" +#include "adf_common_drv.h" + +#define UNSET_RING_NUM -1 + +static const char * const state_operations[] = { + [DEV_DOWN] = "down", + [DEV_UP] = "up", +}; + +static ssize_t state_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct adf_accel_dev *accel_dev; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + return sysfs_emit(buf, "%s\n", str_up_down(adf_dev_started(accel_dev))); +} + +static ssize_t state_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct adf_accel_dev *accel_dev; + u32 accel_id; + int ret; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + accel_id = accel_dev->accel_id; + + if (adf_devmgr_in_reset(accel_dev) || adf_dev_in_use(accel_dev)) { + dev_info(dev, "Device qat_dev%d is busy\n", accel_id); + return -EBUSY; + } + + ret = sysfs_match_string(state_operations, buf); + if (ret < 0) + return ret; + + switch (ret) { + case DEV_DOWN: + dev_info(dev, "Stopping device qat_dev%d\n", accel_id); + + if (!adf_dev_started(accel_dev)) { + dev_info(&GET_DEV(accel_dev), "Device qat_dev%d already down\n", + accel_id); + + break; + } + + ret = adf_dev_down(accel_dev); + if (ret) + return ret; + + break; + case DEV_UP: + dev_info(dev, "Starting device qat_dev%d\n", accel_id); + + ret = adf_dev_up(accel_dev, true); + if (ret == -EALREADY) { + break; + } else if (ret) { + dev_err(dev, "Failed to start device qat_dev%d\n", + accel_id); + adf_dev_down(accel_dev); + return ret; + } + break; + default: + return -EINVAL; + } + + return count; +} + +static ssize_t cfg_services_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + char services[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = {0}; + struct adf_accel_dev *accel_dev; + int ret; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + ret = adf_cfg_get_param_value(accel_dev, ADF_GENERAL_SEC, + ADF_SERVICES_ENABLED, services); + if (ret) + return ret; + + return sysfs_emit(buf, "%s\n", services); +} + +static int adf_sysfs_update_dev_config(struct adf_accel_dev *accel_dev, + const char *services) +{ + return adf_cfg_add_key_value_param(accel_dev, ADF_GENERAL_SEC, + ADF_SERVICES_ENABLED, services, + ADF_STR); +} + +static ssize_t cfg_services_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + char services[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = { }; + struct adf_hw_device_data *hw_data; + struct adf_accel_dev *accel_dev; + int ret; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + ret = adf_parse_service_string(accel_dev, buf, count, services, + ADF_CFG_MAX_VAL_LEN_IN_BYTES); + if (ret) + return ret; + + if (adf_dev_started(accel_dev)) { + dev_info(dev, "Device qat_dev%d must be down to reconfigure the service.\n", + accel_dev->accel_id); + return -EINVAL; + } + + ret = adf_sysfs_update_dev_config(accel_dev, services); + if (ret < 0) + return ret; + + hw_data = GET_HW_DATA(accel_dev); + + /* Update capabilities mask after change in configuration. + * A call to this function is required as capabilities are, at the + * moment, tied to configuration + */ + hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev); + if (!hw_data->accel_capabilities_mask) + return -EINVAL; + + return count; +} + +static ssize_t pm_idle_enabled_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + char pm_idle_enabled[ADF_CFG_MAX_VAL_LEN_IN_BYTES] = {}; + struct adf_accel_dev *accel_dev; + int ret; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + ret = adf_cfg_get_param_value(accel_dev, ADF_GENERAL_SEC, + ADF_PM_IDLE_SUPPORT, pm_idle_enabled); + if (ret) + return sysfs_emit(buf, "1\n"); + + return sysfs_emit(buf, "%s\n", pm_idle_enabled); +} + +static ssize_t pm_idle_enabled_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + unsigned long pm_idle_enabled_cfg_val; + struct adf_accel_dev *accel_dev; + bool pm_idle_enabled; + int ret; + + ret = kstrtobool(buf, &pm_idle_enabled); + if (ret) + return ret; + + pm_idle_enabled_cfg_val = pm_idle_enabled; + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + if (adf_dev_started(accel_dev)) { + dev_info(dev, "Device qat_dev%d must be down to set pm_idle_enabled.\n", + accel_dev->accel_id); + return -EINVAL; + } + + ret = adf_cfg_add_key_value_param(accel_dev, ADF_GENERAL_SEC, + ADF_PM_IDLE_SUPPORT, &pm_idle_enabled_cfg_val, + ADF_DEC); + if (ret) + return ret; + + return count; +} +static DEVICE_ATTR_RW(pm_idle_enabled); + +static ssize_t auto_reset_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct adf_accel_dev *accel_dev; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + return sysfs_emit(buf, "%s\n", str_on_off(accel_dev->autoreset_on_error)); +} + +static ssize_t auto_reset_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct adf_accel_dev *accel_dev; + bool enabled = false; + int ret; + + ret = kstrtobool(buf, &enabled); + if (ret) + return ret; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + accel_dev->autoreset_on_error = enabled; + + return count; +} +static DEVICE_ATTR_RW(auto_reset); + +static DEVICE_ATTR_RW(state); +static DEVICE_ATTR_RW(cfg_services); + +static ssize_t rp2srv_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct adf_hw_device_data *hw_data; + struct adf_accel_dev *accel_dev; + enum adf_cfg_service_type svc; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + hw_data = GET_HW_DATA(accel_dev); + + if (accel_dev->sysfs.ring_num == UNSET_RING_NUM) + return -EINVAL; + + down_read(&accel_dev->sysfs.lock); + svc = GET_SRV_TYPE(accel_dev, accel_dev->sysfs.ring_num % + hw_data->num_banks_per_vf); + up_read(&accel_dev->sysfs.lock); + + switch (svc) { + case COMP: + return sysfs_emit(buf, "%s\n", ADF_CFG_DC); + case SYM: + return sysfs_emit(buf, "%s\n", ADF_CFG_SYM); + case ASYM: + return sysfs_emit(buf, "%s\n", ADF_CFG_ASYM); + case DECOMP: + return sysfs_emit(buf, "%s\n", ADF_CFG_DECOMP); + default: + break; + } + return -EINVAL; +} + +static ssize_t rp2srv_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct adf_accel_dev *accel_dev; + int num_rings, ret; + unsigned int ring; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + ret = kstrtouint(buf, 10, &ring); + if (ret) + return ret; + + num_rings = GET_MAX_BANKS(accel_dev); + if (ring >= num_rings) { + dev_err(&GET_DEV(accel_dev), + "Device does not support more than %u ring pairs\n", + num_rings); + return -EINVAL; + } + + down_write(&accel_dev->sysfs.lock); + accel_dev->sysfs.ring_num = ring; + up_write(&accel_dev->sysfs.lock); + + return count; +} +static DEVICE_ATTR_RW(rp2srv); + +static ssize_t num_rps_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct adf_accel_dev *accel_dev; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + return sysfs_emit(buf, "%u\n", GET_MAX_BANKS(accel_dev)); +} +static DEVICE_ATTR_RO(num_rps); + +static struct attribute *qat_attrs[] = { + &dev_attr_state.attr, + &dev_attr_cfg_services.attr, + &dev_attr_pm_idle_enabled.attr, + &dev_attr_rp2srv.attr, + &dev_attr_num_rps.attr, + &dev_attr_auto_reset.attr, + NULL, +}; + +static struct attribute_group qat_group = { + .attrs = qat_attrs, + .name = "qat", +}; + +int adf_sysfs_init(struct adf_accel_dev *accel_dev) +{ + int ret; + + ret = devm_device_add_group(&GET_DEV(accel_dev), &qat_group); + if (ret) { + dev_err(&GET_DEV(accel_dev), + "Failed to create qat attribute group: %d\n", ret); + } + + accel_dev->sysfs.ring_num = UNSET_RING_NUM; + + return ret; +} +EXPORT_SYMBOL_GPL(adf_sysfs_init); diff --git a/drivers/crypto/intel/qat/qat_common/adf_sysfs_ras_counters.c b/drivers/crypto/intel/qat/qat_common/adf_sysfs_ras_counters.c new file mode 100644 index 000000000000..e97c67c87b3c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_sysfs_ras_counters.c @@ -0,0 +1,117 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/sysfs.h> +#include <linux/pci.h> +#include <linux/string.h> + +#include "adf_common_drv.h" +#include "adf_sysfs_ras_counters.h" + +static ssize_t errors_correctable_show(struct device *dev, + struct device_attribute *dev_attr, + char *buf) +{ + struct adf_accel_dev *accel_dev; + unsigned long counter; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + counter = ADF_RAS_ERR_CTR_READ(accel_dev->ras_errors, ADF_RAS_CORR); + return scnprintf(buf, PAGE_SIZE, "%ld\n", counter); +} + +static ssize_t errors_nonfatal_show(struct device *dev, + struct device_attribute *dev_attr, + char *buf) +{ + struct adf_accel_dev *accel_dev; + unsigned long counter; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + counter = ADF_RAS_ERR_CTR_READ(accel_dev->ras_errors, ADF_RAS_UNCORR); + return scnprintf(buf, PAGE_SIZE, "%ld\n", counter); +} + +static ssize_t errors_fatal_show(struct device *dev, + struct device_attribute *dev_attr, + char *buf) +{ + struct adf_accel_dev *accel_dev; + unsigned long counter; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + counter = ADF_RAS_ERR_CTR_READ(accel_dev->ras_errors, ADF_RAS_FATAL); + return scnprintf(buf, PAGE_SIZE, "%ld\n", counter); +} + +static ssize_t reset_error_counters_store(struct device *dev, + struct device_attribute *dev_attr, + const char *buf, size_t count) +{ + struct adf_accel_dev *accel_dev; + + if (buf[0] != '1' || count != 2) + return -EINVAL; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + ADF_RAS_ERR_CTR_CLEAR(accel_dev->ras_errors); + + return count; +} + +static DEVICE_ATTR_RO(errors_correctable); +static DEVICE_ATTR_RO(errors_nonfatal); +static DEVICE_ATTR_RO(errors_fatal); +static DEVICE_ATTR_WO(reset_error_counters); + +static struct attribute *qat_ras_attrs[] = { + &dev_attr_errors_correctable.attr, + &dev_attr_errors_nonfatal.attr, + &dev_attr_errors_fatal.attr, + &dev_attr_reset_error_counters.attr, + NULL, +}; + +static struct attribute_group qat_ras_group = { + .attrs = qat_ras_attrs, + .name = "qat_ras", +}; + +void adf_sysfs_start_ras(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->ras_errors.enabled) + return; + + ADF_RAS_ERR_CTR_CLEAR(accel_dev->ras_errors); + + if (device_add_group(&GET_DEV(accel_dev), &qat_ras_group)) + dev_err(&GET_DEV(accel_dev), + "Failed to create qat_ras attribute group.\n"); + + accel_dev->ras_errors.sysfs_added = true; +} + +void adf_sysfs_stop_ras(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->ras_errors.enabled) + return; + + if (accel_dev->ras_errors.sysfs_added) { + device_remove_group(&GET_DEV(accel_dev), &qat_ras_group); + accel_dev->ras_errors.sysfs_added = false; + } + + ADF_RAS_ERR_CTR_CLEAR(accel_dev->ras_errors); +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_sysfs_ras_counters.h b/drivers/crypto/intel/qat/qat_common/adf_sysfs_ras_counters.h new file mode 100644 index 000000000000..99e9d9cf57f8 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_sysfs_ras_counters.h @@ -0,0 +1,27 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_RAS_H +#define ADF_RAS_H + +#include <linux/bitops.h> +#include <linux/atomic.h> + +struct adf_accel_dev; + +void adf_sysfs_start_ras(struct adf_accel_dev *accel_dev); +void adf_sysfs_stop_ras(struct adf_accel_dev *accel_dev); + +#define ADF_RAS_ERR_CTR_READ(ras_errors, ERR) \ + atomic_read(&(ras_errors).counter[ERR]) + +#define ADF_RAS_ERR_CTR_CLEAR(ras_errors) \ + do { \ + for (int err = 0; err < ADF_RAS_ERRORS; ++err) \ + atomic_set(&(ras_errors).counter[err], 0); \ + } while (0) + +#define ADF_RAS_ERR_CTR_INC(ras_errors, ERR) \ + atomic_inc(&(ras_errors).counter[ERR]) + +#endif /* ADF_RAS_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_sysfs_rl.c b/drivers/crypto/intel/qat/qat_common/adf_sysfs_rl.c new file mode 100644 index 000000000000..f31556beed8b --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_sysfs_rl.c @@ -0,0 +1,468 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#define dev_fmt(fmt) "RateLimiting: " fmt + +#include <linux/dev_printk.h> +#include <linux/pci.h> +#include <linux/sysfs.h> +#include <linux/types.h> + +#include "adf_common_drv.h" +#include "adf_rl.h" +#include "adf_sysfs_rl.h" + +#define GET_RL_STRUCT(accel_dev) ((accel_dev)->rate_limiting->user_input) + +enum rl_ops { + ADD, + UPDATE, + RM, + RM_ALL, + GET, +}; + +enum rl_params { + RP_MASK, + ID, + CIR, + PIR, + SRV, + CAP_REM_SRV, +}; + +static const char *const rl_services[] = { + [SVC_ASYM] = "asym", + [SVC_SYM] = "sym", + [SVC_DC] = "dc", + [SVC_DECOMP] = "decomp", +}; + +static const char *const rl_operations[] = { + [ADD] = "add", + [UPDATE] = "update", + [RM] = "rm", + [RM_ALL] = "rm_all", + [GET] = "get", +}; + +static int set_param_u(struct device *dev, enum rl_params param, u64 set) +{ + struct adf_rl_interface_data *data; + struct adf_accel_dev *accel_dev; + int ret = 0; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + data = &GET_RL_STRUCT(accel_dev); + + down_write(&data->lock); + switch (param) { + case RP_MASK: + data->input.rp_mask = set; + break; + case CIR: + data->input.cir = set; + break; + case PIR: + data->input.pir = set; + break; + case SRV: + data->input.srv = set; + break; + case CAP_REM_SRV: + data->cap_rem_srv = set; + break; + default: + ret = -EINVAL; + break; + } + up_write(&data->lock); + + return ret; +} + +static int set_param_s(struct device *dev, enum rl_params param, int set) +{ + struct adf_rl_interface_data *data; + struct adf_accel_dev *accel_dev; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev || param != ID) + return -EINVAL; + + data = &GET_RL_STRUCT(accel_dev); + + down_write(&data->lock); + data->input.sla_id = set; + up_write(&data->lock); + + return 0; +} + +static int get_param_u(struct device *dev, enum rl_params param, u64 *get) +{ + struct adf_rl_interface_data *data; + struct adf_accel_dev *accel_dev; + int ret = 0; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + data = &GET_RL_STRUCT(accel_dev); + + down_read(&data->lock); + switch (param) { + case RP_MASK: + *get = data->input.rp_mask; + break; + case CIR: + *get = data->input.cir; + break; + case PIR: + *get = data->input.pir; + break; + case SRV: + *get = data->input.srv; + break; + default: + ret = -EINVAL; + } + up_read(&data->lock); + + return ret; +} + +static int get_param_s(struct device *dev, enum rl_params param) +{ + struct adf_rl_interface_data *data; + struct adf_accel_dev *accel_dev; + int ret = 0; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + data = &GET_RL_STRUCT(accel_dev); + + down_read(&data->lock); + if (param == ID) + ret = data->input.sla_id; + up_read(&data->lock); + + return ret; +} + +static ssize_t rp_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + int ret; + u64 get; + + ret = get_param_u(dev, RP_MASK, &get); + if (ret) + return ret; + + return sysfs_emit(buf, "%#llx\n", get); +} + +static ssize_t rp_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + int err; + u64 val; + + err = kstrtou64(buf, 16, &val); + if (err) + return err; + + err = set_param_u(dev, RP_MASK, val); + if (err) + return err; + + return count; +} +static DEVICE_ATTR_RW(rp); + +static ssize_t id_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + return sysfs_emit(buf, "%d\n", get_param_s(dev, ID)); +} + +static ssize_t id_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + int err; + int val; + + err = kstrtoint(buf, 10, &val); + if (err) + return err; + + err = set_param_s(dev, ID, val); + if (err) + return err; + + return count; +} +static DEVICE_ATTR_RW(id); + +static ssize_t cir_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + int ret; + u64 get; + + ret = get_param_u(dev, CIR, &get); + if (ret) + return ret; + + return sysfs_emit(buf, "%llu\n", get); +} + +static ssize_t cir_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + unsigned int val; + int err; + + err = kstrtouint(buf, 10, &val); + if (err) + return err; + + err = set_param_u(dev, CIR, val); + if (err) + return err; + + return count; +} +static DEVICE_ATTR_RW(cir); + +static ssize_t pir_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + int ret; + u64 get; + + ret = get_param_u(dev, PIR, &get); + if (ret) + return ret; + + return sysfs_emit(buf, "%llu\n", get); +} + +static ssize_t pir_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + unsigned int val; + int err; + + err = kstrtouint(buf, 10, &val); + if (err) + return err; + + err = set_param_u(dev, PIR, val); + if (err) + return err; + + return count; +} +static DEVICE_ATTR_RW(pir); + +static ssize_t srv_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + int ret; + u64 get; + + ret = get_param_u(dev, SRV, &get); + if (ret) + return ret; + + if (get == SVC_BASE_COUNT) + return -EINVAL; + + return sysfs_emit(buf, "%s\n", rl_services[get]); +} + +static ssize_t srv_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct adf_accel_dev *accel_dev; + unsigned int val; + int ret; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + ret = sysfs_match_string(rl_services, buf); + if (ret < 0) + return ret; + + val = ret; + if (!adf_is_service_enabled(accel_dev, val)) + return -EINVAL; + + ret = set_param_u(dev, SRV, val); + if (ret) + return ret; + + return count; +} +static DEVICE_ATTR_RW(srv); + +static ssize_t cap_rem_show(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct adf_rl_interface_data *data; + struct adf_accel_dev *accel_dev; + int ret, rem_cap; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + data = &GET_RL_STRUCT(accel_dev); + + down_read(&data->lock); + rem_cap = adf_rl_get_capability_remaining(accel_dev, data->cap_rem_srv, + RL_SLA_EMPTY_ID); + up_read(&data->lock); + if (rem_cap < 0) + return rem_cap; + + ret = sysfs_emit(buf, "%u\n", rem_cap); + + return ret; +} + +static ssize_t cap_rem_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + unsigned int val; + int ret; + + ret = sysfs_match_string(rl_services, buf); + if (ret < 0) + return ret; + + val = ret; + ret = set_param_u(dev, CAP_REM_SRV, val); + if (ret) + return ret; + + return count; +} +static DEVICE_ATTR_RW(cap_rem); + +static ssize_t sla_op_store(struct device *dev, struct device_attribute *attr, + const char *buf, size_t count) +{ + struct adf_rl_interface_data *data; + struct adf_accel_dev *accel_dev; + int ret; + + accel_dev = adf_devmgr_pci_to_accel_dev(to_pci_dev(dev)); + if (!accel_dev) + return -EINVAL; + + data = &GET_RL_STRUCT(accel_dev); + + ret = sysfs_match_string(rl_operations, buf); + if (ret < 0) + return ret; + + down_write(&data->lock); + switch (ret) { + case ADD: + data->input.parent_id = RL_PARENT_DEFAULT_ID; + data->input.type = RL_LEAF; + data->input.sla_id = 0; + ret = adf_rl_add_sla(accel_dev, &data->input); + if (ret) + goto err_free_lock; + break; + case UPDATE: + ret = adf_rl_update_sla(accel_dev, &data->input); + if (ret) + goto err_free_lock; + break; + case RM: + ret = adf_rl_remove_sla(accel_dev, data->input.sla_id); + if (ret) + goto err_free_lock; + break; + case RM_ALL: + adf_rl_remove_sla_all(accel_dev, false); + break; + case GET: + ret = adf_rl_get_sla(accel_dev, &data->input); + if (ret) + goto err_free_lock; + break; + default: + ret = -EINVAL; + goto err_free_lock; + } + up_write(&data->lock); + + return count; + +err_free_lock: + up_write(&data->lock); + + return ret; +} +static DEVICE_ATTR_WO(sla_op); + +static struct attribute *qat_rl_attrs[] = { + &dev_attr_rp.attr, + &dev_attr_id.attr, + &dev_attr_cir.attr, + &dev_attr_pir.attr, + &dev_attr_srv.attr, + &dev_attr_cap_rem.attr, + &dev_attr_sla_op.attr, + NULL, +}; + +static struct attribute_group qat_rl_group = { + .attrs = qat_rl_attrs, + .name = "qat_rl", +}; + +int adf_sysfs_rl_add(struct adf_accel_dev *accel_dev) +{ + struct adf_rl_interface_data *data; + int ret; + + data = &GET_RL_STRUCT(accel_dev); + + ret = device_add_group(&GET_DEV(accel_dev), &qat_rl_group); + if (ret) + dev_err(&GET_DEV(accel_dev), + "Failed to create qat_rl attribute group\n"); + + data->cap_rem_srv = SVC_BASE_COUNT; + data->input.srv = SVC_BASE_COUNT; + data->sysfs_added = true; + + return ret; +} + +void adf_sysfs_rl_rm(struct adf_accel_dev *accel_dev) +{ + struct adf_rl_interface_data *data; + + data = &GET_RL_STRUCT(accel_dev); + if (!data->sysfs_added) + return; + + device_remove_group(&GET_DEV(accel_dev), &qat_rl_group); + data->sysfs_added = false; +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_sysfs_rl.h b/drivers/crypto/intel/qat/qat_common/adf_sysfs_rl.h new file mode 100644 index 000000000000..22d36aa8a757 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_sysfs_rl.h @@ -0,0 +1,11 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ +#ifndef ADF_SYSFS_RL_H_ +#define ADF_SYSFS_RL_H_ + +struct adf_accel_dev; + +int adf_sysfs_rl_add(struct adf_accel_dev *accel_dev); +void adf_sysfs_rl_rm(struct adf_accel_dev *accel_dev); + +#endif /* ADF_SYSFS_RL_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_telemetry.c b/drivers/crypto/intel/qat/qat_common/adf_telemetry.c new file mode 100644 index 000000000000..b64142db1f0d --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_telemetry.c @@ -0,0 +1,328 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright (c) 2023 Intel Corporation. */ +#define dev_fmt(fmt) "Telemetry: " fmt + +#include <asm/errno.h> +#include <linux/atomic.h> +#include <linux/device.h> +#include <linux/dev_printk.h> +#include <linux/dma-mapping.h> +#include <linux/jiffies.h> +#include <linux/kernel.h> +#include <linux/mutex.h> +#include <linux/slab.h> +#include <linux/string.h> +#include <linux/workqueue.h> + +#include "adf_admin.h" +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_telemetry.h" + +#define TL_IS_ZERO(input) ((input) == 0) + +static bool is_tl_supported(struct adf_accel_dev *accel_dev) +{ + u16 fw_caps = GET_HW_DATA(accel_dev)->fw_capabilities; + + return fw_caps & TL_CAPABILITY_BIT; +} + +static int validate_tl_data(struct adf_tl_hw_data *tl_data) +{ + if (!tl_data->dev_counters || + TL_IS_ZERO(tl_data->num_dev_counters) || + !tl_data->sl_util_counters || + !tl_data->sl_exec_counters || + !tl_data->rp_counters || + TL_IS_ZERO(tl_data->num_rp_counters)) + return -EOPNOTSUPP; + + return 0; +} + +static int validate_tl_slice_counters(struct icp_qat_fw_init_admin_slice_cnt *slice_count, + u8 max_slices_per_type) +{ + u8 *sl_counter = (u8 *)slice_count; + int i; + + for (i = 0; i < ADF_TL_SL_CNT_COUNT; i++) { + if (sl_counter[i] > max_slices_per_type) + return -EINVAL; + } + + return 0; +} + +static int adf_tl_alloc_mem(struct adf_accel_dev *accel_dev) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + struct device *dev = &GET_DEV(accel_dev); + size_t regs_sz = tl_data->layout_sz; + struct adf_telemetry *telemetry; + int node = dev_to_node(dev); + void *tl_data_regs; + unsigned int i; + + telemetry = kzalloc_node(sizeof(*telemetry), GFP_KERNEL, node); + if (!telemetry) + return -ENOMEM; + + telemetry->rp_num_indexes = kmalloc_array(tl_data->max_rp, + sizeof(*telemetry->rp_num_indexes), + GFP_KERNEL); + if (!telemetry->rp_num_indexes) + goto err_free_tl; + + telemetry->regs_hist_buff = kmalloc_array(tl_data->num_hbuff, + sizeof(*telemetry->regs_hist_buff), + GFP_KERNEL); + if (!telemetry->regs_hist_buff) + goto err_free_rp_indexes; + + telemetry->regs_data = dma_alloc_coherent(dev, regs_sz, + &telemetry->regs_data_p, + GFP_KERNEL); + if (!telemetry->regs_data) + goto err_free_regs_hist_buff; + + for (i = 0; i < tl_data->num_hbuff; i++) { + tl_data_regs = kzalloc_node(regs_sz, GFP_KERNEL, node); + if (!tl_data_regs) + goto err_free_dma; + + telemetry->regs_hist_buff[i] = tl_data_regs; + } + + accel_dev->telemetry = telemetry; + + return 0; + +err_free_dma: + dma_free_coherent(dev, regs_sz, telemetry->regs_data, + telemetry->regs_data_p); + + while (i--) + kfree(telemetry->regs_hist_buff[i]); + +err_free_regs_hist_buff: + kfree(telemetry->regs_hist_buff); +err_free_rp_indexes: + kfree(telemetry->rp_num_indexes); +err_free_tl: + kfree(telemetry); + + return -ENOMEM; +} + +static void adf_tl_free_mem(struct adf_accel_dev *accel_dev) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + struct adf_telemetry *telemetry = accel_dev->telemetry; + struct device *dev = &GET_DEV(accel_dev); + size_t regs_sz = tl_data->layout_sz; + unsigned int i; + + for (i = 0; i < tl_data->num_hbuff; i++) + kfree(telemetry->regs_hist_buff[i]); + + dma_free_coherent(dev, regs_sz, telemetry->regs_data, + telemetry->regs_data_p); + + kfree(telemetry->regs_hist_buff); + kfree(telemetry->rp_num_indexes); + kfree(telemetry); + accel_dev->telemetry = NULL; +} + +static unsigned long get_next_timeout(void) +{ + return msecs_to_jiffies(ADF_TL_TIMER_INT_MS); +} + +static void snapshot_regs(struct adf_telemetry *telemetry, size_t size) +{ + void *dst = telemetry->regs_hist_buff[telemetry->hb_num]; + void *src = telemetry->regs_data; + + memcpy(dst, src, size); +} + +static void tl_work_handler(struct work_struct *work) +{ + struct delayed_work *delayed_work; + struct adf_telemetry *telemetry; + struct adf_tl_hw_data *tl_data; + u32 msg_cnt, old_msg_cnt; + size_t layout_sz; + u32 *regs_data; + size_t id; + + delayed_work = to_delayed_work(work); + telemetry = container_of(delayed_work, struct adf_telemetry, work_ctx); + tl_data = &GET_TL_DATA(telemetry->accel_dev); + regs_data = telemetry->regs_data; + + id = tl_data->msg_cnt_off / sizeof(*regs_data); + layout_sz = tl_data->layout_sz; + + if (!atomic_read(&telemetry->state)) { + cancel_delayed_work_sync(&telemetry->work_ctx); + return; + } + + msg_cnt = regs_data[id]; + old_msg_cnt = msg_cnt; + if (msg_cnt == telemetry->msg_cnt) + goto out; + + mutex_lock(&telemetry->regs_hist_lock); + + snapshot_regs(telemetry, layout_sz); + + /* Check if data changed while updating it */ + msg_cnt = regs_data[id]; + if (old_msg_cnt != msg_cnt) + snapshot_regs(telemetry, layout_sz); + + telemetry->msg_cnt = msg_cnt; + telemetry->hb_num++; + telemetry->hb_num %= telemetry->hbuffs; + + mutex_unlock(&telemetry->regs_hist_lock); + +out: + adf_misc_wq_queue_delayed_work(&telemetry->work_ctx, get_next_timeout()); +} + +int adf_tl_halt(struct adf_accel_dev *accel_dev) +{ + struct adf_telemetry *telemetry = accel_dev->telemetry; + struct device *dev = &GET_DEV(accel_dev); + int ret; + + cancel_delayed_work_sync(&telemetry->work_ctx); + atomic_set(&telemetry->state, 0); + + ret = adf_send_admin_tl_stop(accel_dev); + if (ret) + dev_err(dev, "failed to stop telemetry\n"); + + return ret; +} + +static void adf_set_cmdq_cnt(struct adf_accel_dev *accel_dev, + struct adf_tl_hw_data *tl_data) +{ + struct icp_qat_fw_init_admin_slice_cnt *slice_cnt, *cmdq_cnt; + + slice_cnt = &accel_dev->telemetry->slice_cnt; + cmdq_cnt = &accel_dev->telemetry->cmdq_cnt; + + cmdq_cnt->cpr_cnt = slice_cnt->cpr_cnt * tl_data->multiplier.cpr_cnt; + cmdq_cnt->dcpr_cnt = slice_cnt->dcpr_cnt * tl_data->multiplier.dcpr_cnt; + cmdq_cnt->pke_cnt = slice_cnt->pke_cnt * tl_data->multiplier.pke_cnt; + cmdq_cnt->wat_cnt = slice_cnt->wat_cnt * tl_data->multiplier.wat_cnt; + cmdq_cnt->wcp_cnt = slice_cnt->wcp_cnt * tl_data->multiplier.wcp_cnt; + cmdq_cnt->ucs_cnt = slice_cnt->ucs_cnt * tl_data->multiplier.ucs_cnt; + cmdq_cnt->ath_cnt = slice_cnt->ath_cnt * tl_data->multiplier.ath_cnt; +} + +int adf_tl_run(struct adf_accel_dev *accel_dev, int state) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + struct adf_telemetry *telemetry = accel_dev->telemetry; + struct device *dev = &GET_DEV(accel_dev); + size_t layout_sz = tl_data->layout_sz; + int ret; + + ret = adf_send_admin_tl_start(accel_dev, telemetry->regs_data_p, + layout_sz, telemetry->rp_num_indexes, + &telemetry->slice_cnt); + if (ret) { + dev_err(dev, "failed to start telemetry\n"); + return ret; + } + + ret = validate_tl_slice_counters(&telemetry->slice_cnt, tl_data->max_sl_cnt); + if (ret) { + dev_err(dev, "invalid value returned by FW\n"); + adf_send_admin_tl_stop(accel_dev); + return ret; + } + + adf_set_cmdq_cnt(accel_dev, tl_data); + + telemetry->hbuffs = state; + atomic_set(&telemetry->state, state); + + adf_misc_wq_queue_delayed_work(&telemetry->work_ctx, get_next_timeout()); + + return 0; +} + +int adf_tl_init(struct adf_accel_dev *accel_dev) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + u8 max_rp = GET_TL_DATA(accel_dev).max_rp; + struct device *dev = &GET_DEV(accel_dev); + struct adf_telemetry *telemetry; + unsigned int i; + int ret; + + ret = validate_tl_data(tl_data); + if (ret) + return ret; + + ret = adf_tl_alloc_mem(accel_dev); + if (ret) { + dev_err(dev, "failed to initialize: %d\n", ret); + return ret; + } + + telemetry = accel_dev->telemetry; + telemetry->accel_dev = accel_dev; + + mutex_init(&telemetry->wr_lock); + mutex_init(&telemetry->regs_hist_lock); + INIT_DELAYED_WORK(&telemetry->work_ctx, tl_work_handler); + + for (i = 0; i < max_rp; i++) + telemetry->rp_num_indexes[i] = ADF_TL_RP_REGS_DISABLED; + + return 0; +} + +int adf_tl_start(struct adf_accel_dev *accel_dev) +{ + struct device *dev = &GET_DEV(accel_dev); + + if (!accel_dev->telemetry) + return -EOPNOTSUPP; + + if (!is_tl_supported(accel_dev)) { + dev_info(dev, "feature not supported by FW\n"); + adf_tl_free_mem(accel_dev); + return -EOPNOTSUPP; + } + + return 0; +} + +void adf_tl_stop(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->telemetry) + return; + + if (atomic_read(&accel_dev->telemetry->state)) + adf_tl_halt(accel_dev); +} + +void adf_tl_shutdown(struct adf_accel_dev *accel_dev) +{ + if (!accel_dev->telemetry) + return; + + adf_tl_free_mem(accel_dev); +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_telemetry.h b/drivers/crypto/intel/qat/qat_common/adf_telemetry.h new file mode 100644 index 000000000000..02d75c3c214a --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_telemetry.h @@ -0,0 +1,105 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright (c) 2023 Intel Corporation. */ +#ifndef ADF_TELEMETRY_H +#define ADF_TELEMETRY_H + +#include <linux/bits.h> +#include <linux/mutex.h> +#include <linux/types.h> +#include <linux/workqueue.h> + +#include "icp_qat_fw_init_admin.h" + +struct adf_accel_dev; +struct adf_tl_dbg_counter; +struct dentry; + +#define ADF_TL_SL_CNT_COUNT \ + (sizeof(struct icp_qat_fw_init_admin_slice_cnt) / sizeof(__u8)) + +#define TL_CAPABILITY_BIT BIT(1) +/* Interval within device writes data to DMA region. Value in milliseconds. */ +#define ADF_TL_DATA_WR_INTERVAL_MS 1000 +/* Interval within timer interrupt should be handled. Value in milliseconds. */ +#define ADF_TL_TIMER_INT_MS (ADF_TL_DATA_WR_INTERVAL_MS / 2) + +#define ADF_TL_RP_REGS_DISABLED (0xff) + +struct adf_tl_hw_data { + size_t layout_sz; + size_t slice_reg_sz; + size_t cmdq_reg_sz; + size_t rp_reg_sz; + size_t msg_cnt_off; + const struct adf_tl_dbg_counter *dev_counters; + const struct adf_tl_dbg_counter *sl_util_counters; + const struct adf_tl_dbg_counter *sl_exec_counters; + const struct adf_tl_dbg_counter **cmdq_counters; + const struct adf_tl_dbg_counter *rp_counters; + u8 num_hbuff; + u8 cpp_ns_per_cycle; + u8 bw_units_to_bytes; + u8 num_dev_counters; + u8 num_rp_counters; + u8 num_cmdq_counters; + u8 max_rp; + u8 max_sl_cnt; + struct icp_qat_fw_init_admin_slice_cnt multiplier; +}; + +struct adf_telemetry { + struct adf_accel_dev *accel_dev; + atomic_t state; + u32 hbuffs; + int hb_num; + u32 msg_cnt; + dma_addr_t regs_data_p; /* bus address for DMA mapping */ + void *regs_data; /* virtual address for DMA mapping */ + /** + * @regs_hist_buff: array of pointers to copies of the last @hbuffs + * values of @regs_data + */ + void **regs_hist_buff; + struct dentry *dbg_dir; + u8 *rp_num_indexes; + /** + * @regs_hist_lock: protects from race conditions between write and read + * to the copies referenced by @regs_hist_buff + */ + struct mutex regs_hist_lock; + /** + * @wr_lock: protects from concurrent writes to debugfs telemetry files + */ + struct mutex wr_lock; + struct delayed_work work_ctx; + struct icp_qat_fw_init_admin_slice_cnt slice_cnt; + struct icp_qat_fw_init_admin_slice_cnt cmdq_cnt; +}; + +#ifdef CONFIG_DEBUG_FS +int adf_tl_init(struct adf_accel_dev *accel_dev); +int adf_tl_start(struct adf_accel_dev *accel_dev); +void adf_tl_stop(struct adf_accel_dev *accel_dev); +void adf_tl_shutdown(struct adf_accel_dev *accel_dev); +int adf_tl_run(struct adf_accel_dev *accel_dev, int state); +int adf_tl_halt(struct adf_accel_dev *accel_dev); +#else +static inline int adf_tl_init(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static inline int adf_tl_start(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static inline void adf_tl_stop(struct adf_accel_dev *accel_dev) +{ +} + +static inline void adf_tl_shutdown(struct adf_accel_dev *accel_dev) +{ +} +#endif /* CONFIG_DEBUG_FS */ +#endif /* ADF_TELEMETRY_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_timer.c b/drivers/crypto/intel/qat/qat_common/adf_timer.c new file mode 100644 index 000000000000..8962a49f145a --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_timer.c @@ -0,0 +1,71 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2023 Intel Corporation */ + +#include <linux/container_of.h> +#include <linux/dev_printk.h> +#include <linux/export.h> +#include <linux/jiffies.h> +#include <linux/ktime.h> +#include <linux/slab.h> +#include <linux/workqueue.h> + +#include "adf_admin.h" +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_timer.h" + +#define ADF_DEFAULT_TIMER_PERIOD_MS 200 + +/* This periodic update is used to trigger HB, RL & TL fw events */ +static void work_handler(struct work_struct *work) +{ + struct adf_accel_dev *accel_dev; + struct adf_timer *timer_ctx; + u32 time_periods; + + timer_ctx = container_of(to_delayed_work(work), struct adf_timer, work_ctx); + accel_dev = timer_ctx->accel_dev; + + adf_misc_wq_queue_delayed_work(&timer_ctx->work_ctx, + msecs_to_jiffies(ADF_DEFAULT_TIMER_PERIOD_MS)); + + time_periods = div_u64(ktime_ms_delta(ktime_get_real(), timer_ctx->initial_ktime), + ADF_DEFAULT_TIMER_PERIOD_MS); + + if (adf_send_admin_tim_sync(accel_dev, time_periods)) + dev_err(&GET_DEV(accel_dev), "Failed to synchronize qat timer\n"); +} + +int adf_timer_start(struct adf_accel_dev *accel_dev) +{ + struct adf_timer *timer_ctx; + + timer_ctx = kzalloc(sizeof(*timer_ctx), GFP_KERNEL); + if (!timer_ctx) + return -ENOMEM; + + timer_ctx->accel_dev = accel_dev; + accel_dev->timer = timer_ctx; + timer_ctx->initial_ktime = ktime_get_real(); + + INIT_DELAYED_WORK(&timer_ctx->work_ctx, work_handler); + adf_misc_wq_queue_delayed_work(&timer_ctx->work_ctx, + msecs_to_jiffies(ADF_DEFAULT_TIMER_PERIOD_MS)); + + return 0; +} +EXPORT_SYMBOL_GPL(adf_timer_start); + +void adf_timer_stop(struct adf_accel_dev *accel_dev) +{ + struct adf_timer *timer_ctx = accel_dev->timer; + + if (!timer_ctx) + return; + + cancel_delayed_work_sync(&timer_ctx->work_ctx); + + kfree(timer_ctx); + accel_dev->timer = NULL; +} +EXPORT_SYMBOL_GPL(adf_timer_stop); diff --git a/drivers/crypto/intel/qat/qat_common/adf_timer.h b/drivers/crypto/intel/qat/qat_common/adf_timer.h new file mode 100644 index 000000000000..68e5136d6ba1 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_timer.h @@ -0,0 +1,21 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2023 Intel Corporation */ + +#ifndef ADF_TIMER_H_ +#define ADF_TIMER_H_ + +#include <linux/ktime.h> +#include <linux/workqueue.h> + +struct adf_accel_dev; + +struct adf_timer { + struct adf_accel_dev *accel_dev; + struct delayed_work work_ctx; + ktime_t initial_ktime; +}; + +int adf_timer_start(struct adf_accel_dev *accel_dev); +void adf_timer_stop(struct adf_accel_dev *accel_dev); + +#endif /* ADF_TIMER_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_tl_debugfs.c b/drivers/crypto/intel/qat/qat_common/adf_tl_debugfs.c new file mode 100644 index 000000000000..b81f70576683 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_tl_debugfs.c @@ -0,0 +1,737 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright (c) 2023 Intel Corporation. */ +#define dev_fmt(fmt) "Telemetry debugfs: " fmt + +#include <linux/atomic.h> +#include <linux/debugfs.h> +#include <linux/dev_printk.h> +#include <linux/dcache.h> +#include <linux/file.h> +#include <linux/kernel.h> +#include <linux/math64.h> +#include <linux/mutex.h> +#include <linux/seq_file.h> +#include <linux/slab.h> +#include <linux/units.h> + +#include "adf_accel_devices.h" +#include "adf_cfg_strings.h" +#include "adf_telemetry.h" +#include "adf_tl_debugfs.h" + +#define TL_VALUE_MIN_PADDING 20 +#define TL_KEY_MIN_PADDING 23 +#define TL_RP_SRV_UNKNOWN "Unknown" + +static int tl_collect_values_u32(struct adf_telemetry *telemetry, + size_t counter_offset, u64 *arr) +{ + unsigned int samples, hb_idx, i; + u32 *regs_hist_buff; + u32 counter_val; + + samples = min(telemetry->msg_cnt, telemetry->hbuffs); + hb_idx = telemetry->hb_num + telemetry->hbuffs - samples; + + mutex_lock(&telemetry->regs_hist_lock); + + for (i = 0; i < samples; i++) { + regs_hist_buff = telemetry->regs_hist_buff[hb_idx % telemetry->hbuffs]; + counter_val = regs_hist_buff[counter_offset / sizeof(counter_val)]; + arr[i] = counter_val; + hb_idx++; + } + + mutex_unlock(&telemetry->regs_hist_lock); + + return samples; +} + +static int tl_collect_values_u64(struct adf_telemetry *telemetry, + size_t counter_offset, u64 *arr) +{ + unsigned int samples, hb_idx, i; + u64 *regs_hist_buff; + u64 counter_val; + + samples = min(telemetry->msg_cnt, telemetry->hbuffs); + hb_idx = telemetry->hb_num + telemetry->hbuffs - samples; + + mutex_lock(&telemetry->regs_hist_lock); + + for (i = 0; i < samples; i++) { + regs_hist_buff = telemetry->regs_hist_buff[hb_idx % telemetry->hbuffs]; + counter_val = regs_hist_buff[counter_offset / sizeof(counter_val)]; + arr[i] = counter_val; + hb_idx++; + } + + mutex_unlock(&telemetry->regs_hist_lock); + + return samples; +} + +/** + * avg_array() - Return average of values within an array. + * @array: Array of values. + * @len: Number of elements. + * + * This algorithm computes average of an array without running into overflow. + * + * Return: average of values. + */ +#define avg_array(array, len) ( \ +{ \ + typeof(&(array)[0]) _array = (array); \ + __unqual_scalar_typeof(_array[0]) _x = 0; \ + __unqual_scalar_typeof(_array[0]) _y = 0; \ + __unqual_scalar_typeof(_array[0]) _a, _b; \ + typeof(len) _len = (len); \ + size_t _i; \ + \ + for (_i = 0; _i < _len; _i++) { \ + _a = _array[_i]; \ + _b = do_div(_a, _len); \ + _x += _a; \ + if (_y >= _len - _b) { \ + _x++; \ + _y -= _len - _b; \ + } else { \ + _y += _b; \ + } \ + } \ + do_div(_y, _len); \ + (_x + _y); \ +}) + +/* Calculation function for simple counter. */ +static int tl_calc_count(struct adf_telemetry *telemetry, + const struct adf_tl_dbg_counter *ctr, + struct adf_tl_dbg_aggr_values *vals) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(telemetry->accel_dev); + u64 *hist_vals; + int sample_cnt; + int ret = 0; + + hist_vals = kmalloc_array(tl_data->num_hbuff, sizeof(*hist_vals), + GFP_KERNEL); + if (!hist_vals) + return -ENOMEM; + + memset(vals, 0, sizeof(*vals)); + sample_cnt = tl_collect_values_u32(telemetry, ctr->offset1, hist_vals); + if (!sample_cnt) + goto out_free_hist_vals; + + vals->curr = hist_vals[sample_cnt - 1]; + vals->min = min_array(hist_vals, sample_cnt); + vals->max = max_array(hist_vals, sample_cnt); + vals->avg = avg_array(hist_vals, sample_cnt); + +out_free_hist_vals: + kfree(hist_vals); + return ret; +} + +/* Convert CPP bus cycles to ns. */ +static int tl_cycles_to_ns(struct adf_telemetry *telemetry, + const struct adf_tl_dbg_counter *ctr, + struct adf_tl_dbg_aggr_values *vals) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(telemetry->accel_dev); + u8 cpp_ns_per_cycle = tl_data->cpp_ns_per_cycle; + int ret; + + ret = tl_calc_count(telemetry, ctr, vals); + if (ret) + return ret; + + vals->curr *= cpp_ns_per_cycle; + vals->min *= cpp_ns_per_cycle; + vals->max *= cpp_ns_per_cycle; + vals->avg *= cpp_ns_per_cycle; + + return 0; +} + +/* + * Compute latency cumulative average with division of accumulated value + * by sample count. Returned value is in ns. + */ +static int tl_lat_acc_avg(struct adf_telemetry *telemetry, + const struct adf_tl_dbg_counter *ctr, + struct adf_tl_dbg_aggr_values *vals) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(telemetry->accel_dev); + u8 cpp_ns_per_cycle = tl_data->cpp_ns_per_cycle; + u8 num_hbuff = tl_data->num_hbuff; + int sample_cnt, i; + u64 *hist_vals; + u64 *hist_cnt; + int ret = 0; + + hist_vals = kmalloc_array(num_hbuff, sizeof(*hist_vals), GFP_KERNEL); + if (!hist_vals) + return -ENOMEM; + + hist_cnt = kmalloc_array(num_hbuff, sizeof(*hist_cnt), GFP_KERNEL); + if (!hist_cnt) { + ret = -ENOMEM; + goto out_free_hist_vals; + } + + memset(vals, 0, sizeof(*vals)); + sample_cnt = tl_collect_values_u64(telemetry, ctr->offset1, hist_vals); + if (!sample_cnt) + goto out_free_hist_cnt; + + tl_collect_values_u32(telemetry, ctr->offset2, hist_cnt); + + for (i = 0; i < sample_cnt; i++) { + /* Avoid division by 0 if count is 0. */ + if (hist_cnt[i]) + hist_vals[i] = div_u64(hist_vals[i] * cpp_ns_per_cycle, + hist_cnt[i]); + else + hist_vals[i] = 0; + } + + vals->curr = hist_vals[sample_cnt - 1]; + vals->min = min_array(hist_vals, sample_cnt); + vals->max = max_array(hist_vals, sample_cnt); + vals->avg = avg_array(hist_vals, sample_cnt); + +out_free_hist_cnt: + kfree(hist_cnt); +out_free_hist_vals: + kfree(hist_vals); + return ret; +} + +/* Convert HW raw bandwidth units to Mbps. */ +static int tl_bw_hw_units_to_mbps(struct adf_telemetry *telemetry, + const struct adf_tl_dbg_counter *ctr, + struct adf_tl_dbg_aggr_values *vals) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(telemetry->accel_dev); + u16 bw_hw_2_bits = tl_data->bw_units_to_bytes * BITS_PER_BYTE; + u64 *hist_vals; + int sample_cnt; + int ret = 0; + + hist_vals = kmalloc_array(tl_data->num_hbuff, sizeof(*hist_vals), + GFP_KERNEL); + if (!hist_vals) + return -ENOMEM; + + memset(vals, 0, sizeof(*vals)); + sample_cnt = tl_collect_values_u32(telemetry, ctr->offset1, hist_vals); + if (!sample_cnt) + goto out_free_hist_vals; + + vals->curr = div_u64(hist_vals[sample_cnt - 1] * bw_hw_2_bits, MEGA); + vals->min = div_u64(min_array(hist_vals, sample_cnt) * bw_hw_2_bits, MEGA); + vals->max = div_u64(max_array(hist_vals, sample_cnt) * bw_hw_2_bits, MEGA); + vals->avg = div_u64(avg_array(hist_vals, sample_cnt) * bw_hw_2_bits, MEGA); + +out_free_hist_vals: + kfree(hist_vals); + return ret; +} + +static void tl_seq_printf_counter(struct adf_telemetry *telemetry, + struct seq_file *s, const char *name, + struct adf_tl_dbg_aggr_values *vals) +{ + seq_printf(s, "%-*s", TL_KEY_MIN_PADDING, name); + seq_printf(s, "%*llu", TL_VALUE_MIN_PADDING, vals->curr); + if (atomic_read(&telemetry->state) > 1) { + seq_printf(s, "%*llu", TL_VALUE_MIN_PADDING, vals->min); + seq_printf(s, "%*llu", TL_VALUE_MIN_PADDING, vals->max); + seq_printf(s, "%*llu", TL_VALUE_MIN_PADDING, vals->avg); + } + seq_puts(s, "\n"); +} + +static int tl_calc_and_print_counter(struct adf_telemetry *telemetry, + struct seq_file *s, + const struct adf_tl_dbg_counter *ctr, + const char *name) +{ + const char *counter_name = name ? name : ctr->name; + enum adf_tl_counter_type type = ctr->type; + struct adf_tl_dbg_aggr_values vals; + int ret; + + switch (type) { + case ADF_TL_SIMPLE_COUNT: + ret = tl_calc_count(telemetry, ctr, &vals); + break; + case ADF_TL_COUNTER_NS: + ret = tl_cycles_to_ns(telemetry, ctr, &vals); + break; + case ADF_TL_COUNTER_NS_AVG: + ret = tl_lat_acc_avg(telemetry, ctr, &vals); + break; + case ADF_TL_COUNTER_MBPS: + ret = tl_bw_hw_units_to_mbps(telemetry, ctr, &vals); + break; + default: + return -EINVAL; + } + + if (ret) + return ret; + + tl_seq_printf_counter(telemetry, s, counter_name, &vals); + + return 0; +} + +static int tl_print_sl_counter(struct adf_telemetry *telemetry, + const struct adf_tl_dbg_counter *ctr, + struct seq_file *s, u8 cnt_id) +{ + size_t sl_regs_sz = GET_TL_DATA(telemetry->accel_dev).slice_reg_sz; + struct adf_tl_dbg_counter slice_ctr; + size_t offset_inc = cnt_id * sl_regs_sz; + char cnt_name[MAX_COUNT_NAME_SIZE]; + + snprintf(cnt_name, MAX_COUNT_NAME_SIZE, "%s%d", ctr->name, cnt_id); + slice_ctr = *ctr; + slice_ctr.offset1 += offset_inc; + + return tl_calc_and_print_counter(telemetry, s, &slice_ctr, cnt_name); +} + +static int tl_calc_and_print_sl_counters(struct adf_accel_dev *accel_dev, + struct seq_file *s, u8 cnt_type, u8 cnt_id) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + struct adf_telemetry *telemetry = accel_dev->telemetry; + const struct adf_tl_dbg_counter *sl_tl_util_counters; + const struct adf_tl_dbg_counter *sl_tl_exec_counters; + const struct adf_tl_dbg_counter *ctr; + int ret; + + sl_tl_util_counters = tl_data->sl_util_counters; + sl_tl_exec_counters = tl_data->sl_exec_counters; + + ctr = &sl_tl_util_counters[cnt_type]; + + ret = tl_print_sl_counter(telemetry, ctr, s, cnt_id); + if (ret) { + dev_notice(&GET_DEV(accel_dev), + "invalid slice utilization counter type\n"); + return ret; + } + + ctr = &sl_tl_exec_counters[cnt_type]; + + ret = tl_print_sl_counter(telemetry, ctr, s, cnt_id); + if (ret) { + dev_notice(&GET_DEV(accel_dev), + "invalid slice execution counter type\n"); + return ret; + } + + return 0; +} + +static int tl_print_cmdq_counter(struct adf_telemetry *telemetry, + const struct adf_tl_dbg_counter *ctr, + struct seq_file *s, u8 cnt_id, u8 counter) +{ + size_t cmdq_regs_sz = GET_TL_DATA(telemetry->accel_dev).cmdq_reg_sz; + size_t offset_inc = cnt_id * cmdq_regs_sz; + struct adf_tl_dbg_counter slice_ctr; + char cnt_name[MAX_COUNT_NAME_SIZE]; + + slice_ctr = *(ctr + counter); + slice_ctr.offset1 += offset_inc; + snprintf(cnt_name, MAX_COUNT_NAME_SIZE, "%s%d", slice_ctr.name, cnt_id); + + return tl_calc_and_print_counter(telemetry, s, &slice_ctr, cnt_name); +} + +static int tl_calc_and_print_cmdq_counters(struct adf_accel_dev *accel_dev, + struct seq_file *s, u8 cnt_type, + u8 cnt_id) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + struct adf_telemetry *telemetry = accel_dev->telemetry; + const struct adf_tl_dbg_counter **cmdq_tl_counters; + const struct adf_tl_dbg_counter *ctr; + u8 counter; + int ret; + + cmdq_tl_counters = tl_data->cmdq_counters; + ctr = cmdq_tl_counters[cnt_type]; + + for (counter = 0; counter < tl_data->num_cmdq_counters; counter++) { + ret = tl_print_cmdq_counter(telemetry, ctr, s, cnt_id, counter); + if (ret) { + dev_notice(&GET_DEV(accel_dev), + "invalid slice utilization counter type\n"); + return ret; + } + } + + return 0; +} + +static void tl_print_msg_cnt(struct seq_file *s, u32 msg_cnt) +{ + seq_printf(s, "%-*s", TL_KEY_MIN_PADDING, SNAPSHOT_CNT_MSG); + seq_printf(s, "%*u\n", TL_VALUE_MIN_PADDING, msg_cnt); +} + +static int tl_print_dev_data(struct adf_accel_dev *accel_dev, + struct seq_file *s) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + struct adf_telemetry *telemetry = accel_dev->telemetry; + const struct adf_tl_dbg_counter *dev_tl_counters; + u8 num_dev_counters = tl_data->num_dev_counters; + u8 *cmdq_cnt = (u8 *)&telemetry->cmdq_cnt; + u8 *sl_cnt = (u8 *)&telemetry->slice_cnt; + const struct adf_tl_dbg_counter *ctr; + unsigned int i; + int ret; + u8 j; + + if (!atomic_read(&telemetry->state)) { + dev_info(&GET_DEV(accel_dev), "not enabled\n"); + return -EPERM; + } + + dev_tl_counters = tl_data->dev_counters; + + tl_print_msg_cnt(s, telemetry->msg_cnt); + + /* Print device level telemetry. */ + for (i = 0; i < num_dev_counters; i++) { + ctr = &dev_tl_counters[i]; + ret = tl_calc_and_print_counter(telemetry, s, ctr, NULL); + if (ret) { + dev_notice(&GET_DEV(accel_dev), + "invalid counter type\n"); + return ret; + } + } + + /* Print per slice telemetry. */ + for (i = 0; i < ADF_TL_SL_CNT_COUNT; i++) { + for (j = 0; j < sl_cnt[i]; j++) { + ret = tl_calc_and_print_sl_counters(accel_dev, s, i, j); + if (ret) + return ret; + } + } + + /* Print per command queue telemetry. */ + for (i = 0; i < ADF_TL_SL_CNT_COUNT; i++) { + for (j = 0; j < cmdq_cnt[i]; j++) { + ret = tl_calc_and_print_cmdq_counters(accel_dev, s, i, j); + if (ret) + return ret; + } + } + + return 0; +} + +static int tl_dev_data_show(struct seq_file *s, void *unused) +{ + struct adf_accel_dev *accel_dev = s->private; + + if (!accel_dev) + return -EINVAL; + + return tl_print_dev_data(accel_dev, s); +} +DEFINE_SHOW_ATTRIBUTE(tl_dev_data); + +static int tl_control_show(struct seq_file *s, void *unused) +{ + struct adf_accel_dev *accel_dev = s->private; + + if (!accel_dev) + return -EINVAL; + + seq_printf(s, "%d\n", atomic_read(&accel_dev->telemetry->state)); + + return 0; +} + +static ssize_t tl_control_write(struct file *file, const char __user *userbuf, + size_t count, loff_t *ppos) +{ + struct seq_file *seq_f = file->private_data; + struct adf_accel_dev *accel_dev; + struct adf_telemetry *telemetry; + struct adf_tl_hw_data *tl_data; + struct device *dev; + u32 input; + int ret; + + accel_dev = seq_f->private; + if (!accel_dev) + return -EINVAL; + + tl_data = &GET_TL_DATA(accel_dev); + telemetry = accel_dev->telemetry; + dev = &GET_DEV(accel_dev); + + mutex_lock(&telemetry->wr_lock); + + ret = kstrtou32_from_user(userbuf, count, 10, &input); + if (ret) + goto unlock_and_exit; + + if (input > tl_data->num_hbuff) { + dev_info(dev, "invalid control input\n"); + ret = -EINVAL; + goto unlock_and_exit; + } + + /* If input is 0, just stop telemetry. */ + if (!input) { + ret = adf_tl_halt(accel_dev); + if (!ret) + ret = count; + + goto unlock_and_exit; + } + + /* If TL is already enabled, stop it. */ + if (atomic_read(&telemetry->state)) { + dev_info(dev, "already enabled, restarting.\n"); + ret = adf_tl_halt(accel_dev); + if (ret) + goto unlock_and_exit; + } + + ret = adf_tl_run(accel_dev, input); + if (ret) + goto unlock_and_exit; + + ret = count; + +unlock_and_exit: + mutex_unlock(&telemetry->wr_lock); + return ret; +} +DEFINE_SHOW_STORE_ATTRIBUTE(tl_control); + +static int adf_tl_dbg_change_rp_index(struct adf_accel_dev *accel_dev, + unsigned int new_rp_num, + unsigned int rp_regs_index) +{ + struct adf_hw_device_data *hw_data = GET_HW_DATA(accel_dev); + struct adf_telemetry *telemetry = accel_dev->telemetry; + struct device *dev = &GET_DEV(accel_dev); + unsigned int i; + u8 curr_state; + int ret; + + if (new_rp_num >= hw_data->num_rps) { + dev_info(dev, "invalid Ring Pair number selected\n"); + return -EINVAL; + } + + for (i = 0; i < hw_data->tl_data.max_rp; i++) { + if (telemetry->rp_num_indexes[i] == new_rp_num) { + dev_info(dev, "RP nr: %d is already selected in slot rp_%c_data\n", + new_rp_num, ADF_TL_DBG_RP_ALPHA_INDEX(i)); + return 0; + } + } + + dev_dbg(dev, "selecting RP nr %u into slot rp_%c_data\n", + new_rp_num, ADF_TL_DBG_RP_ALPHA_INDEX(rp_regs_index)); + + curr_state = atomic_read(&telemetry->state); + + if (curr_state) { + ret = adf_tl_halt(accel_dev); + if (ret) + return ret; + + telemetry->rp_num_indexes[rp_regs_index] = new_rp_num; + + ret = adf_tl_run(accel_dev, curr_state); + if (ret) + return ret; + } else { + telemetry->rp_num_indexes[rp_regs_index] = new_rp_num; + } + + return 0; +} + +static void tl_print_rp_srv(struct adf_accel_dev *accel_dev, struct seq_file *s, + u8 rp_idx) +{ + u32 banks_per_vf = GET_HW_DATA(accel_dev)->num_banks_per_vf; + enum adf_cfg_service_type svc; + + seq_printf(s, "%-*s", TL_KEY_MIN_PADDING, RP_SERVICE_TYPE); + + svc = GET_SRV_TYPE(accel_dev, rp_idx % banks_per_vf); + switch (svc) { + case COMP: + seq_printf(s, "%*s\n", TL_VALUE_MIN_PADDING, ADF_CFG_DC); + break; + case SYM: + seq_printf(s, "%*s\n", TL_VALUE_MIN_PADDING, ADF_CFG_SYM); + break; + case ASYM: + seq_printf(s, "%*s\n", TL_VALUE_MIN_PADDING, ADF_CFG_ASYM); + break; + case DECOMP: + seq_printf(s, "%*s\n", TL_VALUE_MIN_PADDING, ADF_CFG_DECOMP); + break; + default: + seq_printf(s, "%*s\n", TL_VALUE_MIN_PADDING, TL_RP_SRV_UNKNOWN); + break; + } +} + +static int tl_print_rp_data(struct adf_accel_dev *accel_dev, struct seq_file *s, + u8 rp_regs_index) +{ + struct adf_tl_hw_data *tl_data = &GET_TL_DATA(accel_dev); + struct adf_telemetry *telemetry = accel_dev->telemetry; + const struct adf_tl_dbg_counter *rp_tl_counters; + u8 num_rp_counters = tl_data->num_rp_counters; + size_t rp_regs_sz = tl_data->rp_reg_sz; + struct adf_tl_dbg_counter ctr; + unsigned int i; + u8 rp_idx; + int ret; + + if (!atomic_read(&telemetry->state)) { + dev_info(&GET_DEV(accel_dev), "not enabled\n"); + return -EPERM; + } + + rp_tl_counters = tl_data->rp_counters; + rp_idx = telemetry->rp_num_indexes[rp_regs_index]; + + if (rp_idx == ADF_TL_RP_REGS_DISABLED) { + dev_info(&GET_DEV(accel_dev), "no RP number selected in rp_%c_data\n", + ADF_TL_DBG_RP_ALPHA_INDEX(rp_regs_index)); + return -EPERM; + } + + tl_print_msg_cnt(s, telemetry->msg_cnt); + seq_printf(s, "%-*s", TL_KEY_MIN_PADDING, RP_NUM_INDEX); + seq_printf(s, "%*d\n", TL_VALUE_MIN_PADDING, rp_idx); + tl_print_rp_srv(accel_dev, s, rp_idx); + + for (i = 0; i < num_rp_counters; i++) { + ctr = rp_tl_counters[i]; + ctr.offset1 += rp_regs_sz * rp_regs_index; + ctr.offset2 += rp_regs_sz * rp_regs_index; + ret = tl_calc_and_print_counter(telemetry, s, &ctr, NULL); + if (ret) { + dev_dbg(&GET_DEV(accel_dev), + "invalid RP counter type\n"); + return ret; + } + } + + return 0; +} + +static int tl_rp_data_show(struct seq_file *s, void *unused) +{ + struct adf_accel_dev *accel_dev = s->private; + u8 rp_regs_index; + + if (!accel_dev) + return -EINVAL; + + rp_regs_index = debugfs_get_aux_num(s->file); + + return tl_print_rp_data(accel_dev, s, rp_regs_index); +} + +static ssize_t tl_rp_data_write(struct file *file, const char __user *userbuf, + size_t count, loff_t *ppos) +{ + struct seq_file *seq_f = file->private_data; + struct adf_accel_dev *accel_dev; + struct adf_telemetry *telemetry; + unsigned int new_rp_num; + u8 rp_regs_index; + int ret; + + accel_dev = seq_f->private; + if (!accel_dev) + return -EINVAL; + + telemetry = accel_dev->telemetry; + + mutex_lock(&telemetry->wr_lock); + + rp_regs_index = debugfs_get_aux_num(file); + + ret = kstrtou32_from_user(userbuf, count, 10, &new_rp_num); + if (ret) + goto unlock_and_exit; + + ret = adf_tl_dbg_change_rp_index(accel_dev, new_rp_num, rp_regs_index); + if (ret) + goto unlock_and_exit; + + ret = count; + +unlock_and_exit: + mutex_unlock(&telemetry->wr_lock); + return ret; +} +DEFINE_SHOW_STORE_ATTRIBUTE(tl_rp_data); + +void adf_tl_dbgfs_add(struct adf_accel_dev *accel_dev) +{ + struct adf_telemetry *telemetry = accel_dev->telemetry; + struct dentry *parent = accel_dev->debugfs_dir; + u8 max_rp = GET_TL_DATA(accel_dev).max_rp; + char name[ADF_TL_RP_REGS_FNAME_SIZE]; + struct dentry *dir; + unsigned int i; + + if (!telemetry) + return; + + dir = debugfs_create_dir("telemetry", parent); + accel_dev->telemetry->dbg_dir = dir; + debugfs_create_file("device_data", 0444, dir, accel_dev, &tl_dev_data_fops); + debugfs_create_file("control", 0644, dir, accel_dev, &tl_control_fops); + + for (i = 0; i < max_rp; i++) { + snprintf(name, sizeof(name), ADF_TL_RP_REGS_FNAME, + ADF_TL_DBG_RP_ALPHA_INDEX(i)); + debugfs_create_file_aux_num(name, 0644, dir, accel_dev, i, + &tl_rp_data_fops); + } +} + +void adf_tl_dbgfs_rm(struct adf_accel_dev *accel_dev) +{ + struct adf_telemetry *telemetry = accel_dev->telemetry; + struct dentry *dbg_dir; + + if (!telemetry) + return; + + dbg_dir = telemetry->dbg_dir; + + debugfs_remove_recursive(dbg_dir); + + if (atomic_read(&telemetry->state)) + adf_tl_halt(accel_dev); +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_tl_debugfs.h b/drivers/crypto/intel/qat/qat_common/adf_tl_debugfs.h new file mode 100644 index 000000000000..97c5eeaa1b17 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_tl_debugfs.h @@ -0,0 +1,122 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright (c) 2023 Intel Corporation. */ +#ifndef ADF_TL_DEBUGFS_H +#define ADF_TL_DEBUGFS_H + +#include <linux/types.h> + +struct adf_accel_dev; + +#define MAX_COUNT_NAME_SIZE 32 +#define SNAPSHOT_CNT_MSG "sample_cnt" +#define RP_NUM_INDEX "rp_num" +#define PCI_TRANS_CNT_NAME "pci_trans_cnt" +#define MAX_RD_LAT_NAME "max_rd_lat" +#define RD_LAT_ACC_NAME "rd_lat_acc_avg" +#define MAX_LAT_NAME "max_gp_lat" +#define LAT_ACC_NAME "gp_lat_acc_avg" +#define BW_IN_NAME "bw_in" +#define BW_OUT_NAME "bw_out" +#define RE_ACC_NAME "re_acc_avg" +#define PAGE_REQ_LAT_NAME "at_page_req_lat_avg" +#define AT_TRANS_LAT_NAME "at_trans_lat_avg" +#define AT_MAX_UTLB_USED_NAME "at_max_tlb_used" +#define AT_GLOB_DTLB_HIT_NAME "at_glob_devtlb_hit" +#define AT_GLOB_DTLB_MISS_NAME "at_glob_devtlb_miss" +#define AT_PAYLD_DTLB_HIT_NAME "tl_at_payld_devtlb_hit" +#define AT_PAYLD_DTLB_MISS_NAME "tl_at_payld_devtlb_miss" +#define RP_SERVICE_TYPE "service_type" + +#define ADF_TL_DBG_RP_ALPHA_INDEX(index) ((index) + 'A') +#define ADF_TL_DBG_RP_INDEX_ALPHA(alpha) ((alpha) - 'A') + +#define ADF_TL_RP_REGS_FNAME "rp_%c_data" +#define ADF_TL_RP_REGS_FNAME_SIZE 16 + +#define ADF_TL_DATA_REG_OFF(reg, qat_gen) \ + offsetof(struct adf_##qat_gen##_tl_layout, reg) + +#define ADF_TL_DEV_REG_OFF(reg, qat_gen) \ + (ADF_TL_DATA_REG_OFF(tl_device_data_regs, qat_gen) + \ + offsetof(struct adf_##qat_gen##_tl_device_data_regs, reg)) + +#define ADF_TL_SLICE_REG_OFF(slice, reg, qat_gen) \ + (ADF_TL_DEV_REG_OFF(slice##_slices[0], qat_gen) + \ + offsetof(struct adf_##qat_gen##_tl_slice_data_regs, reg)) + +#define ADF_TL_CMDQ_REG_OFF(slice, reg, qat_gen) \ + (ADF_TL_DEV_REG_OFF(slice##_cmdq[0], qat_gen) + \ + offsetof(struct adf_##qat_gen##_tl_cmdq_data_regs, reg)) + +#define ADF_TL_RP_REG_OFF(reg, qat_gen) \ + (ADF_TL_DATA_REG_OFF(tl_ring_pairs_data_regs[0], qat_gen) + \ + offsetof(struct adf_##qat_gen##_tl_ring_pair_data_regs, reg)) + +/** + * enum adf_tl_counter_type - telemetry counter types + * @ADF_TL_COUNTER_UNSUPPORTED: unsupported counter + * @ADF_TL_SIMPLE_COUNT: simple counter + * @ADF_TL_COUNTER_NS: latency counter, value in ns + * @ADF_TL_COUNTER_NS_AVG: accumulated average latency counter, value in ns + * @ADF_TL_COUNTER_MBPS: bandwidth, value in MBps + */ +enum adf_tl_counter_type { + ADF_TL_COUNTER_UNSUPPORTED, + ADF_TL_SIMPLE_COUNT, + ADF_TL_COUNTER_NS, + ADF_TL_COUNTER_NS_AVG, + ADF_TL_COUNTER_MBPS, +}; + +/** + * struct adf_tl_dbg_counter - telemetry counter definition + * @name: name of the counter as printed in the report + * @adf_tl_counter_type: type of the counter + * @offset1: offset of 1st register + * @offset2: offset of 2nd optional register + */ +struct adf_tl_dbg_counter { + const char *name; + enum adf_tl_counter_type type; + size_t offset1; + size_t offset2; +}; + +#define ADF_TL_COUNTER(_name, _type, _offset) \ +{ .name = _name, \ + .type = _type, \ + .offset1 = _offset \ +} + +#define ADF_TL_COUNTER_LATENCY(_name, _type, _offset1, _offset2) \ +{ .name = _name, \ + .type = _type, \ + .offset1 = _offset1, \ + .offset2 = _offset2 \ +} + +/* Telemetry counter aggregated values. */ +struct adf_tl_dbg_aggr_values { + u64 curr; + u64 min; + u64 max; + u64 avg; +}; + +/** + * adf_tl_dbgfs_add() - Add telemetry's debug fs entries. + * @accel_dev: Pointer to acceleration device. + * + * Creates telemetry's debug fs folder and attributes in QAT debug fs root. + */ +void adf_tl_dbgfs_add(struct adf_accel_dev *accel_dev); + +/** + * adf_tl_dbgfs_rm() - Remove telemetry's debug fs entries. + * @accel_dev: Pointer to acceleration device. + * + * Removes telemetry's debug fs folder and attributes from QAT debug fs root. + */ +void adf_tl_dbgfs_rm(struct adf_accel_dev *accel_dev); + +#endif /* ADF_TL_DEBUGFS_H */ diff --git a/drivers/crypto/intel/qat/qat_common/adf_transport.c b/drivers/crypto/intel/qat/qat_common/adf_transport.c new file mode 100644 index 000000000000..1efdf46490f1 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_transport.c @@ -0,0 +1,575 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/delay.h> +#include <linux/nospec.h> +#include "adf_accel_devices.h" +#include "adf_transport_internal.h" +#include "adf_transport_access_macros.h" +#include "adf_cfg.h" +#include "adf_common_drv.h" + +#define ADF_MAX_RING_THRESHOLD 80 +#define ADF_PERCENT(tot, percent) (((tot) * (percent)) / 100) + +static inline u32 adf_modulo(u32 data, u32 shift) +{ + u32 div = data >> shift; + u32 mult = div << shift; + + return data - mult; +} + +static inline int adf_check_ring_alignment(u64 addr, u64 size) +{ + if (((size - 1) & addr) != 0) + return -EFAULT; + return 0; +} + +static int adf_verify_ring_size(u32 msg_size, u32 msg_num) +{ + int i = ADF_MIN_RING_SIZE; + + for (; i <= ADF_MAX_RING_SIZE; i++) + if ((msg_size * msg_num) == ADF_SIZE_TO_RING_SIZE_IN_BYTES(i)) + return i; + + return ADF_DEFAULT_RING_SIZE; +} + +static int adf_reserve_ring(struct adf_etr_bank_data *bank, u32 ring) +{ + spin_lock(&bank->lock); + if (bank->ring_mask & (1 << ring)) { + spin_unlock(&bank->lock); + return -EFAULT; + } + bank->ring_mask |= (1 << ring); + spin_unlock(&bank->lock); + return 0; +} + +static void adf_unreserve_ring(struct adf_etr_bank_data *bank, u32 ring) +{ + spin_lock(&bank->lock); + bank->ring_mask &= ~(1 << ring); + spin_unlock(&bank->lock); +} + +static void adf_enable_ring_irq(struct adf_etr_bank_data *bank, u32 ring) +{ + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(bank->accel_dev); + + spin_lock_bh(&bank->lock); + bank->irq_mask |= (1 << ring); + spin_unlock_bh(&bank->lock); + csr_ops->write_csr_int_col_en(bank->csr_addr, bank->bank_number, + bank->irq_mask); + csr_ops->write_csr_int_col_ctl(bank->csr_addr, bank->bank_number, + bank->irq_coalesc_timer); +} + +static void adf_disable_ring_irq(struct adf_etr_bank_data *bank, u32 ring) +{ + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(bank->accel_dev); + + spin_lock_bh(&bank->lock); + bank->irq_mask &= ~(1 << ring); + spin_unlock_bh(&bank->lock); + csr_ops->write_csr_int_col_en(bank->csr_addr, bank->bank_number, + bank->irq_mask); +} + +bool adf_ring_nearly_full(struct adf_etr_ring_data *ring) +{ + return atomic_read(ring->inflights) > ring->threshold; +} + +int adf_send_message(struct adf_etr_ring_data *ring, u32 *msg) +{ + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(ring->bank->accel_dev); + + if (atomic_add_return(1, ring->inflights) > + ADF_MAX_INFLIGHTS(ring->ring_size, ring->msg_size)) { + atomic_dec(ring->inflights); + return -EAGAIN; + } + spin_lock_bh(&ring->lock); + memcpy((void *)((uintptr_t)ring->base_addr + ring->tail), msg, + ADF_MSG_SIZE_TO_BYTES(ring->msg_size)); + + ring->tail = adf_modulo(ring->tail + + ADF_MSG_SIZE_TO_BYTES(ring->msg_size), + ADF_RING_SIZE_MODULO(ring->ring_size)); + csr_ops->write_csr_ring_tail(ring->bank->csr_addr, + ring->bank->bank_number, ring->ring_number, + ring->tail); + spin_unlock_bh(&ring->lock); + + return 0; +} + +static int adf_handle_response(struct adf_etr_ring_data *ring) +{ + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(ring->bank->accel_dev); + u32 msg_counter = 0; + u32 *msg = (u32 *)((uintptr_t)ring->base_addr + ring->head); + + while (*msg != ADF_RING_EMPTY_SIG) { + ring->callback((u32 *)msg); + atomic_dec(ring->inflights); + *msg = ADF_RING_EMPTY_SIG; + ring->head = adf_modulo(ring->head + + ADF_MSG_SIZE_TO_BYTES(ring->msg_size), + ADF_RING_SIZE_MODULO(ring->ring_size)); + msg_counter++; + msg = (u32 *)((uintptr_t)ring->base_addr + ring->head); + } + if (msg_counter > 0) { + csr_ops->write_csr_ring_head(ring->bank->csr_addr, + ring->bank->bank_number, + ring->ring_number, ring->head); + } + return 0; +} + +static void adf_configure_tx_ring(struct adf_etr_ring_data *ring) +{ + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(ring->bank->accel_dev); + u32 ring_config = BUILD_RING_CONFIG(ring->ring_size); + + csr_ops->write_csr_ring_config(ring->bank->csr_addr, + ring->bank->bank_number, + ring->ring_number, ring_config); + +} + +static void adf_configure_rx_ring(struct adf_etr_ring_data *ring) +{ + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(ring->bank->accel_dev); + u32 ring_config = + BUILD_RESP_RING_CONFIG(ring->ring_size, + ADF_RING_NEAR_WATERMARK_512, + ADF_RING_NEAR_WATERMARK_0); + + csr_ops->write_csr_ring_config(ring->bank->csr_addr, + ring->bank->bank_number, + ring->ring_number, ring_config); +} + +static int adf_init_ring(struct adf_etr_ring_data *ring) +{ + struct adf_etr_bank_data *bank = ring->bank; + struct adf_accel_dev *accel_dev = bank->accel_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(accel_dev); + u64 ring_base; + u32 ring_size_bytes = + ADF_SIZE_TO_RING_SIZE_IN_BYTES(ring->ring_size); + + ring_size_bytes = ADF_RING_SIZE_BYTES_MIN(ring_size_bytes); + ring->base_addr = dma_alloc_coherent(&GET_DEV(accel_dev), + ring_size_bytes, &ring->dma_addr, + GFP_KERNEL); + if (!ring->base_addr) + return -ENOMEM; + + memset(ring->base_addr, 0x7F, ring_size_bytes); + /* The base_addr has to be aligned to the size of the buffer */ + if (adf_check_ring_alignment(ring->dma_addr, ring_size_bytes)) { + dev_err(&GET_DEV(accel_dev), "Ring address not aligned\n"); + dma_free_coherent(&GET_DEV(accel_dev), ring_size_bytes, + ring->base_addr, ring->dma_addr); + ring->base_addr = NULL; + return -EFAULT; + } + + if (hw_data->tx_rings_mask & (1 << ring->ring_number)) + adf_configure_tx_ring(ring); + + else + adf_configure_rx_ring(ring); + + ring_base = csr_ops->build_csr_ring_base_addr(ring->dma_addr, + ring->ring_size); + + csr_ops->write_csr_ring_base(ring->bank->csr_addr, + ring->bank->bank_number, ring->ring_number, + ring_base); + spin_lock_init(&ring->lock); + return 0; +} + +static void adf_cleanup_ring(struct adf_etr_ring_data *ring) +{ + u32 ring_size_bytes = + ADF_SIZE_TO_RING_SIZE_IN_BYTES(ring->ring_size); + ring_size_bytes = ADF_RING_SIZE_BYTES_MIN(ring_size_bytes); + + if (ring->base_addr) { + memset(ring->base_addr, 0x7F, ring_size_bytes); + dma_free_coherent(&GET_DEV(ring->bank->accel_dev), + ring_size_bytes, ring->base_addr, + ring->dma_addr); + } +} + +int adf_create_ring(struct adf_accel_dev *accel_dev, const char *section, + u32 bank_num, u32 num_msgs, + u32 msg_size, const char *ring_name, + adf_callback_fn callback, int poll_mode, + struct adf_etr_ring_data **ring_ptr) +{ + struct adf_etr_data *transport_data = accel_dev->transport; + u8 num_rings_per_bank = GET_NUM_RINGS_PER_BANK(accel_dev); + struct adf_etr_bank_data *bank; + struct adf_etr_ring_data *ring; + char val[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + int max_inflights; + u32 ring_num; + int ret; + + if (bank_num >= GET_MAX_BANKS(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Invalid bank number\n"); + return -EFAULT; + } + if (msg_size > ADF_MSG_SIZE_TO_BYTES(ADF_MAX_MSG_SIZE)) { + dev_err(&GET_DEV(accel_dev), "Invalid msg size\n"); + return -EFAULT; + } + if (ADF_MAX_INFLIGHTS(adf_verify_ring_size(msg_size, num_msgs), + ADF_BYTES_TO_MSG_SIZE(msg_size)) < 2) { + dev_err(&GET_DEV(accel_dev), + "Invalid ring size for given msg size\n"); + return -EFAULT; + } + if (adf_cfg_get_param_value(accel_dev, section, ring_name, val)) { + dev_err(&GET_DEV(accel_dev), "Section %s, no such entry : %s\n", + section, ring_name); + return -EFAULT; + } + if (kstrtouint(val, 10, &ring_num)) { + dev_err(&GET_DEV(accel_dev), "Can't get ring number\n"); + return -EFAULT; + } + if (ring_num >= num_rings_per_bank) { + dev_err(&GET_DEV(accel_dev), "Invalid ring number\n"); + return -EFAULT; + } + + ring_num = array_index_nospec(ring_num, num_rings_per_bank); + bank = &transport_data->banks[bank_num]; + if (adf_reserve_ring(bank, ring_num)) { + dev_err(&GET_DEV(accel_dev), "Ring %d, %s already exists.\n", + ring_num, ring_name); + return -EFAULT; + } + ring = &bank->rings[ring_num]; + ring->ring_number = ring_num; + ring->bank = bank; + ring->callback = callback; + ring->msg_size = ADF_BYTES_TO_MSG_SIZE(msg_size); + ring->ring_size = adf_verify_ring_size(msg_size, num_msgs); + ring->head = 0; + ring->tail = 0; + max_inflights = ADF_MAX_INFLIGHTS(ring->ring_size, ring->msg_size); + ring->threshold = ADF_PERCENT(max_inflights, ADF_MAX_RING_THRESHOLD); + atomic_set(ring->inflights, 0); + ret = adf_init_ring(ring); + if (ret) + goto err; + + /* Enable HW arbitration for the given ring */ + adf_update_ring_arb(ring); + + if (adf_ring_debugfs_add(ring, ring_name)) { + dev_err(&GET_DEV(accel_dev), + "Couldn't add ring debugfs entry\n"); + ret = -EFAULT; + goto err; + } + + /* Enable interrupts if needed */ + if (callback && (!poll_mode)) + adf_enable_ring_irq(bank, ring->ring_number); + *ring_ptr = ring; + return 0; +err: + adf_cleanup_ring(ring); + adf_unreserve_ring(bank, ring_num); + adf_update_ring_arb(ring); + return ret; +} + +void adf_remove_ring(struct adf_etr_ring_data *ring) +{ + struct adf_etr_bank_data *bank = ring->bank; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(bank->accel_dev); + + /* Disable interrupts for the given ring */ + adf_disable_ring_irq(bank, ring->ring_number); + + /* Clear PCI config space */ + + csr_ops->write_csr_ring_config(bank->csr_addr, bank->bank_number, + ring->ring_number, 0); + csr_ops->write_csr_ring_base(bank->csr_addr, bank->bank_number, + ring->ring_number, 0); + adf_ring_debugfs_rm(ring); + adf_unreserve_ring(bank, ring->ring_number); + /* Disable HW arbitration for the given ring */ + adf_update_ring_arb(ring); + adf_cleanup_ring(ring); +} + +static void adf_ring_response_handler(struct adf_etr_bank_data *bank) +{ + struct adf_accel_dev *accel_dev = bank->accel_dev; + u8 num_rings_per_bank = GET_NUM_RINGS_PER_BANK(accel_dev); + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(accel_dev); + unsigned long empty_rings; + int i; + + empty_rings = csr_ops->read_csr_e_stat(bank->csr_addr, + bank->bank_number); + empty_rings = ~empty_rings & bank->irq_mask; + + for_each_set_bit(i, &empty_rings, num_rings_per_bank) + adf_handle_response(&bank->rings[i]); +} + +void adf_response_handler(uintptr_t bank_addr) +{ + struct adf_etr_bank_data *bank = (void *)bank_addr; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(bank->accel_dev); + + /* Handle all the responses and reenable IRQs */ + adf_ring_response_handler(bank); + + csr_ops->write_csr_int_flag_and_col(bank->csr_addr, bank->bank_number, + bank->irq_mask); +} + +static inline int adf_get_cfg_int(struct adf_accel_dev *accel_dev, + const char *section, const char *format, + u32 key, u32 *value) +{ + char key_buf[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + char val_buf[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + + snprintf(key_buf, ADF_CFG_MAX_KEY_LEN_IN_BYTES, format, key); + + if (adf_cfg_get_param_value(accel_dev, section, key_buf, val_buf)) + return -EFAULT; + + if (kstrtouint(val_buf, 10, value)) + return -EFAULT; + return 0; +} + +static void adf_get_coalesc_timer(struct adf_etr_bank_data *bank, + const char *section, + u32 bank_num_in_accel) +{ + if (adf_get_cfg_int(bank->accel_dev, section, + ADF_ETRMGR_COALESCE_TIMER_FORMAT, + bank_num_in_accel, &bank->irq_coalesc_timer)) + bank->irq_coalesc_timer = ADF_COALESCING_DEF_TIME; + + if (ADF_COALESCING_MAX_TIME < bank->irq_coalesc_timer || + ADF_COALESCING_MIN_TIME > bank->irq_coalesc_timer) + bank->irq_coalesc_timer = ADF_COALESCING_DEF_TIME; +} + +static int adf_init_bank(struct adf_accel_dev *accel_dev, + struct adf_etr_bank_data *bank, + u32 bank_num, void __iomem *csr_addr) +{ + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u8 num_rings_per_bank = hw_data->num_rings_per_bank; + struct adf_hw_csr_ops *csr_ops = &hw_data->csr_ops; + u32 irq_mask = BIT(num_rings_per_bank) - 1; + struct adf_etr_ring_data *ring; + struct adf_etr_ring_data *tx_ring; + u32 i, coalesc_enabled = 0; + unsigned long ring_mask; + int size; + + memset(bank, 0, sizeof(*bank)); + bank->bank_number = bank_num; + bank->csr_addr = csr_addr; + bank->accel_dev = accel_dev; + spin_lock_init(&bank->lock); + + /* Allocate the rings in the bank */ + size = num_rings_per_bank * sizeof(struct adf_etr_ring_data); + bank->rings = kzalloc_node(size, GFP_KERNEL, + dev_to_node(&GET_DEV(accel_dev))); + if (!bank->rings) + return -ENOMEM; + + /* Enable IRQ coalescing always. This will allow to use + * the optimised flag and coalesc register. + * If it is disabled in the config file just use min time value */ + if ((adf_get_cfg_int(accel_dev, "Accelerator0", + ADF_ETRMGR_COALESCING_ENABLED_FORMAT, bank_num, + &coalesc_enabled) == 0) && coalesc_enabled) + adf_get_coalesc_timer(bank, "Accelerator0", bank_num); + else + bank->irq_coalesc_timer = ADF_COALESCING_MIN_TIME; + + for (i = 0; i < num_rings_per_bank; i++) { + csr_ops->write_csr_ring_config(csr_addr, bank_num, i, 0); + csr_ops->write_csr_ring_base(csr_addr, bank_num, i, 0); + + ring = &bank->rings[i]; + if (hw_data->tx_rings_mask & (1 << i)) { + ring->inflights = + kzalloc_node(sizeof(atomic_t), + GFP_KERNEL, + dev_to_node(&GET_DEV(accel_dev))); + if (!ring->inflights) + goto err; + } else { + if (i < hw_data->tx_rx_gap) { + dev_err(&GET_DEV(accel_dev), + "Invalid tx rings mask config\n"); + goto err; + } + tx_ring = &bank->rings[i - hw_data->tx_rx_gap]; + ring->inflights = tx_ring->inflights; + } + } + if (adf_bank_debugfs_add(bank)) { + dev_err(&GET_DEV(accel_dev), + "Failed to add bank debugfs entry\n"); + goto err; + } + + csr_ops->write_csr_int_flag(csr_addr, bank_num, irq_mask); + csr_ops->write_csr_int_srcsel(csr_addr, bank_num); + + return 0; +err: + ring_mask = hw_data->tx_rings_mask; + for_each_set_bit(i, &ring_mask, num_rings_per_bank) { + ring = &bank->rings[i]; + kfree(ring->inflights); + ring->inflights = NULL; + } + kfree(bank->rings); + return -ENOMEM; +} + +/** + * adf_init_etr_data() - Initialize transport rings for acceleration device + * @accel_dev: Pointer to acceleration device. + * + * Function is the initializes the communications channels (rings) to the + * acceleration device accel_dev. + * To be used by QAT device specific drivers. + * + * Return: 0 on success, error code otherwise. + */ +int adf_init_etr_data(struct adf_accel_dev *accel_dev) +{ + struct adf_etr_data *etr_data; + void __iomem *csr_addr; + u32 size; + u32 num_banks = 0; + int i, ret; + + etr_data = kzalloc_node(sizeof(*etr_data), GFP_KERNEL, + dev_to_node(&GET_DEV(accel_dev))); + if (!etr_data) + return -ENOMEM; + + num_banks = GET_MAX_BANKS(accel_dev); + size = num_banks * sizeof(struct adf_etr_bank_data); + etr_data->banks = kzalloc_node(size, GFP_KERNEL, + dev_to_node(&GET_DEV(accel_dev))); + if (!etr_data->banks) { + ret = -ENOMEM; + goto err_bank; + } + + accel_dev->transport = etr_data; + csr_addr = adf_get_etr_base(accel_dev); + + /* accel_dev->debugfs_dir should always be non-NULL here */ + etr_data->debug = debugfs_create_dir("transport", + accel_dev->debugfs_dir); + + for (i = 0; i < num_banks; i++) { + ret = adf_init_bank(accel_dev, &etr_data->banks[i], i, + csr_addr); + if (ret) + goto err_bank_all; + } + + return 0; + +err_bank_all: + debugfs_remove(etr_data->debug); + kfree(etr_data->banks); +err_bank: + kfree(etr_data); + accel_dev->transport = NULL; + return ret; +} +EXPORT_SYMBOL_GPL(adf_init_etr_data); + +static void cleanup_bank(struct adf_etr_bank_data *bank) +{ + struct adf_accel_dev *accel_dev = bank->accel_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + u8 num_rings_per_bank = hw_data->num_rings_per_bank; + u32 i; + + for (i = 0; i < num_rings_per_bank; i++) { + struct adf_etr_ring_data *ring = &bank->rings[i]; + + if (bank->ring_mask & (1 << i)) + adf_cleanup_ring(ring); + + if (hw_data->tx_rings_mask & (1 << i)) + kfree(ring->inflights); + } + kfree(bank->rings); + adf_bank_debugfs_rm(bank); + memset(bank, 0, sizeof(*bank)); +} + +static void adf_cleanup_etr_handles(struct adf_accel_dev *accel_dev) +{ + struct adf_etr_data *etr_data = accel_dev->transport; + u32 i, num_banks = GET_MAX_BANKS(accel_dev); + + for (i = 0; i < num_banks; i++) + cleanup_bank(&etr_data->banks[i]); +} + +/** + * adf_cleanup_etr_data() - Clear transport rings for acceleration device + * @accel_dev: Pointer to acceleration device. + * + * Function is the clears the communications channels (rings) of the + * acceleration device accel_dev. + * To be used by QAT device specific drivers. + * + * Return: void + */ +void adf_cleanup_etr_data(struct adf_accel_dev *accel_dev) +{ + struct adf_etr_data *etr_data = accel_dev->transport; + + if (etr_data) { + adf_cleanup_etr_handles(accel_dev); + debugfs_remove(etr_data->debug); + kfree(etr_data->banks->rings); + kfree(etr_data->banks); + kfree(etr_data); + accel_dev->transport = NULL; + } +} +EXPORT_SYMBOL_GPL(adf_cleanup_etr_data); diff --git a/drivers/crypto/intel/qat/qat_common/adf_transport.h b/drivers/crypto/intel/qat/qat_common/adf_transport.h new file mode 100644 index 000000000000..e6ef6f9b7691 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_transport.h @@ -0,0 +1,20 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_TRANSPORT_H +#define ADF_TRANSPORT_H + +#include "adf_accel_devices.h" + +struct adf_etr_ring_data; + +typedef void (*adf_callback_fn)(void *resp_msg); + +int adf_create_ring(struct adf_accel_dev *accel_dev, const char *section, + u32 bank_num, u32 num_mgs, u32 msg_size, + const char *ring_name, adf_callback_fn callback, + int poll_mode, struct adf_etr_ring_data **ring_ptr); + +bool adf_ring_nearly_full(struct adf_etr_ring_data *ring); +int adf_send_message(struct adf_etr_ring_data *ring, u32 *msg); +void adf_remove_ring(struct adf_etr_ring_data *ring); +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_transport_access_macros.h b/drivers/crypto/intel/qat/qat_common/adf_transport_access_macros.h new file mode 100644 index 000000000000..d3667dbd9826 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_transport_access_macros.h @@ -0,0 +1,58 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_TRANSPORT_ACCESS_MACROS_H +#define ADF_TRANSPORT_ACCESS_MACROS_H + +#include "adf_accel_devices.h" +#define ADF_RING_CONFIG_NEAR_FULL_WM 0x0A +#define ADF_RING_CONFIG_NEAR_EMPTY_WM 0x05 +#define ADF_COALESCING_MIN_TIME 0x1FF +#define ADF_COALESCING_MAX_TIME 0xFFFFF +#define ADF_COALESCING_DEF_TIME 0x27FF +#define ADF_RING_NEAR_WATERMARK_512 0x08 +#define ADF_RING_NEAR_WATERMARK_0 0x00 +#define ADF_RING_EMPTY_SIG 0x7F7F7F7F + +/* Valid internal ring size values */ +#define ADF_RING_SIZE_128 0x01 +#define ADF_RING_SIZE_256 0x02 +#define ADF_RING_SIZE_512 0x03 +#define ADF_RING_SIZE_4K 0x06 +#define ADF_RING_SIZE_16K 0x08 +#define ADF_RING_SIZE_4M 0x10 +#define ADF_MIN_RING_SIZE ADF_RING_SIZE_128 +#define ADF_MAX_RING_SIZE ADF_RING_SIZE_4M +#define ADF_DEFAULT_RING_SIZE ADF_RING_SIZE_16K + +/* Valid internal msg size values */ +#define ADF_MSG_SIZE_32 0x01 +#define ADF_MSG_SIZE_64 0x02 +#define ADF_MSG_SIZE_128 0x04 +#define ADF_MIN_MSG_SIZE ADF_MSG_SIZE_32 +#define ADF_MAX_MSG_SIZE ADF_MSG_SIZE_128 + +/* Size to bytes conversion macros for ring and msg size values */ +#define ADF_MSG_SIZE_TO_BYTES(SIZE) (SIZE << 5) +#define ADF_BYTES_TO_MSG_SIZE(SIZE) (SIZE >> 5) +#define ADF_SIZE_TO_RING_SIZE_IN_BYTES(SIZE) ((1 << (SIZE - 1)) << 7) +#define ADF_RING_SIZE_IN_BYTES_TO_SIZE(SIZE) ((1 << (SIZE - 1)) >> 7) + +/* Minimum ring buffer size for memory allocation */ +#define ADF_RING_SIZE_BYTES_MIN(SIZE) \ + ((SIZE < ADF_SIZE_TO_RING_SIZE_IN_BYTES(ADF_RING_SIZE_4K)) ? \ + ADF_SIZE_TO_RING_SIZE_IN_BYTES(ADF_RING_SIZE_4K) : SIZE) +#define ADF_RING_SIZE_MODULO(SIZE) (SIZE + 0x6) +#define ADF_SIZE_TO_POW(SIZE) ((((SIZE & 0x4) >> 1) | ((SIZE & 0x4) >> 2) | \ + SIZE) & ~0x4) +/* Max outstanding requests */ +#define ADF_MAX_INFLIGHTS(RING_SIZE, MSG_SIZE) \ + ((((1 << (RING_SIZE - 1)) << 3) >> ADF_SIZE_TO_POW(MSG_SIZE)) - 1) +#define BUILD_RING_CONFIG(size) \ + ((ADF_RING_NEAR_WATERMARK_0 << ADF_RING_CONFIG_NEAR_FULL_WM) \ + | (ADF_RING_NEAR_WATERMARK_0 << ADF_RING_CONFIG_NEAR_EMPTY_WM) \ + | size) +#define BUILD_RESP_RING_CONFIG(size, watermark_nf, watermark_ne) \ + ((watermark_nf << ADF_RING_CONFIG_NEAR_FULL_WM) \ + | (watermark_ne << ADF_RING_CONFIG_NEAR_EMPTY_WM) \ + | size) +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_transport_debug.c b/drivers/crypto/intel/qat/qat_common/adf_transport_debug.c new file mode 100644 index 000000000000..6c22bc9b28e4 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_transport_debug.c @@ -0,0 +1,216 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/mutex.h> +#include <linux/slab.h> +#include <linux/seq_file.h> +#include "adf_accel_devices.h" +#include "adf_transport_internal.h" +#include "adf_transport_access_macros.h" + +static DEFINE_MUTEX(ring_read_lock); +static DEFINE_MUTEX(bank_read_lock); + +#define ADF_RING_NUM_MSGS(ring) \ + (ADF_SIZE_TO_RING_SIZE_IN_BYTES(ring->ring_size) / \ + ADF_MSG_SIZE_TO_BYTES(ring->msg_size)) + +static void *adf_ring_start(struct seq_file *sfile, loff_t *pos) +{ + struct adf_etr_ring_data *ring = sfile->private; + unsigned int num_msg = ADF_RING_NUM_MSGS(ring); + loff_t val = *pos; + + mutex_lock(&ring_read_lock); + if (val == 0) + return SEQ_START_TOKEN; + + if (val >= num_msg) + return NULL; + + return ring->base_addr + + (ADF_MSG_SIZE_TO_BYTES(ring->msg_size) * (*pos)++); +} + +static void *adf_ring_next(struct seq_file *sfile, void *v, loff_t *pos) +{ + struct adf_etr_ring_data *ring = sfile->private; + unsigned int num_msg = ADF_RING_NUM_MSGS(ring); + loff_t val = *pos; + + (*pos)++; + + if (val >= num_msg) + return NULL; + + return ring->base_addr + (ADF_MSG_SIZE_TO_BYTES(ring->msg_size) * val); +} + +static int adf_ring_show(struct seq_file *sfile, void *v) +{ + struct adf_etr_ring_data *ring = sfile->private; + struct adf_etr_bank_data *bank = ring->bank; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(bank->accel_dev); + void __iomem *csr = ring->bank->csr_addr; + + if (v == SEQ_START_TOKEN) { + int head, tail, empty; + + head = csr_ops->read_csr_ring_head(csr, bank->bank_number, + ring->ring_number); + tail = csr_ops->read_csr_ring_tail(csr, bank->bank_number, + ring->ring_number); + empty = csr_ops->read_csr_e_stat(csr, bank->bank_number); + + seq_puts(sfile, "------- Ring configuration -------\n"); + seq_printf(sfile, "ring name: %s\n", + ring->ring_debug->ring_name); + seq_printf(sfile, "ring num %d, bank num %d\n", + ring->ring_number, ring->bank->bank_number); + seq_printf(sfile, "head %x, tail %x, empty: %d\n", + head, tail, (empty & 1 << ring->ring_number) + >> ring->ring_number); + seq_printf(sfile, "ring size %lld, msg size %d\n", + (long long)ADF_SIZE_TO_RING_SIZE_IN_BYTES(ring->ring_size), + ADF_MSG_SIZE_TO_BYTES(ring->msg_size)); + seq_puts(sfile, "----------- Ring data ------------\n"); + return 0; + } + seq_hex_dump(sfile, "", DUMP_PREFIX_ADDRESS, 32, 4, + v, ADF_MSG_SIZE_TO_BYTES(ring->msg_size), false); + return 0; +} + +static void adf_ring_stop(struct seq_file *sfile, void *v) +{ + mutex_unlock(&ring_read_lock); +} + +static const struct seq_operations adf_ring_debug_sops = { + .start = adf_ring_start, + .next = adf_ring_next, + .stop = adf_ring_stop, + .show = adf_ring_show +}; + +DEFINE_SEQ_ATTRIBUTE(adf_ring_debug); + +int adf_ring_debugfs_add(struct adf_etr_ring_data *ring, const char *name) +{ + struct adf_etr_ring_debug_entry *ring_debug; + char entry_name[16]; + + ring_debug = kzalloc(sizeof(*ring_debug), GFP_KERNEL); + if (!ring_debug) + return -ENOMEM; + + strscpy(ring_debug->ring_name, name, sizeof(ring_debug->ring_name)); + snprintf(entry_name, sizeof(entry_name), "ring_%02d", + ring->ring_number); + + ring_debug->debug = debugfs_create_file(entry_name, S_IRUSR, + ring->bank->bank_debug_dir, + ring, &adf_ring_debug_fops); + ring->ring_debug = ring_debug; + return 0; +} + +void adf_ring_debugfs_rm(struct adf_etr_ring_data *ring) +{ + if (ring->ring_debug) { + debugfs_remove(ring->ring_debug->debug); + kfree(ring->ring_debug); + ring->ring_debug = NULL; + } +} + +static void *adf_bank_start(struct seq_file *sfile, loff_t *pos) +{ + struct adf_etr_bank_data *bank = sfile->private; + u8 num_rings_per_bank = GET_NUM_RINGS_PER_BANK(bank->accel_dev); + + mutex_lock(&bank_read_lock); + if (*pos == 0) + return SEQ_START_TOKEN; + + if (*pos >= num_rings_per_bank) + return NULL; + + return pos; +} + +static void *adf_bank_next(struct seq_file *sfile, void *v, loff_t *pos) +{ + struct adf_etr_bank_data *bank = sfile->private; + u8 num_rings_per_bank = GET_NUM_RINGS_PER_BANK(bank->accel_dev); + + if (++(*pos) >= num_rings_per_bank) + return NULL; + + return pos; +} + +static int adf_bank_show(struct seq_file *sfile, void *v) +{ + struct adf_etr_bank_data *bank = sfile->private; + struct adf_hw_csr_ops *csr_ops = GET_CSR_OPS(bank->accel_dev); + + if (v == SEQ_START_TOKEN) { + seq_printf(sfile, "------- Bank %d configuration -------\n", + bank->bank_number); + } else { + int ring_id = *((int *)v) - 1; + struct adf_etr_ring_data *ring = &bank->rings[ring_id]; + void __iomem *csr = bank->csr_addr; + int head, tail, empty; + + if (!(bank->ring_mask & 1 << ring_id)) + return 0; + + head = csr_ops->read_csr_ring_head(csr, bank->bank_number, + ring->ring_number); + tail = csr_ops->read_csr_ring_tail(csr, bank->bank_number, + ring->ring_number); + empty = csr_ops->read_csr_e_stat(csr, bank->bank_number); + + seq_printf(sfile, + "ring num %02d, head %04x, tail %04x, empty: %d\n", + ring->ring_number, head, tail, + (empty & 1 << ring->ring_number) >> + ring->ring_number); + } + return 0; +} + +static void adf_bank_stop(struct seq_file *sfile, void *v) +{ + mutex_unlock(&bank_read_lock); +} + +static const struct seq_operations adf_bank_debug_sops = { + .start = adf_bank_start, + .next = adf_bank_next, + .stop = adf_bank_stop, + .show = adf_bank_show +}; + +DEFINE_SEQ_ATTRIBUTE(adf_bank_debug); + +int adf_bank_debugfs_add(struct adf_etr_bank_data *bank) +{ + struct adf_accel_dev *accel_dev = bank->accel_dev; + struct dentry *parent = accel_dev->transport->debug; + char name[16]; + + snprintf(name, sizeof(name), "bank_%02d", bank->bank_number); + bank->bank_debug_dir = debugfs_create_dir(name, parent); + bank->bank_debug_cfg = debugfs_create_file("config", S_IRUSR, + bank->bank_debug_dir, bank, + &adf_bank_debug_fops); + return 0; +} + +void adf_bank_debugfs_rm(struct adf_etr_bank_data *bank) +{ + debugfs_remove(bank->bank_debug_cfg); + debugfs_remove(bank->bank_debug_dir); +} diff --git a/drivers/crypto/intel/qat/qat_common/adf_transport_internal.h b/drivers/crypto/intel/qat/qat_common/adf_transport_internal.h new file mode 100644 index 000000000000..8b2c92ba7ca1 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_transport_internal.h @@ -0,0 +1,73 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_TRANSPORT_INTRN_H +#define ADF_TRANSPORT_INTRN_H + +#include <linux/interrupt.h> +#include <linux/spinlock_types.h> +#include "adf_transport.h" + +struct adf_etr_ring_debug_entry { + char ring_name[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + struct dentry *debug; +}; + +struct adf_etr_ring_data { + void *base_addr; + atomic_t *inflights; + adf_callback_fn callback; + struct adf_etr_bank_data *bank; + dma_addr_t dma_addr; + struct adf_etr_ring_debug_entry *ring_debug; + spinlock_t lock; /* protects ring data struct */ + u16 head; + u16 tail; + u32 threshold; + u8 ring_number; + u8 ring_size; + u8 msg_size; +}; + +struct adf_etr_bank_data { + struct adf_etr_ring_data *rings; + struct tasklet_struct resp_handler; + void __iomem *csr_addr; + u32 irq_coalesc_timer; + u32 bank_number; + u16 ring_mask; + u16 irq_mask; + spinlock_t lock; /* protects bank data struct */ + struct adf_accel_dev *accel_dev; + struct dentry *bank_debug_dir; + struct dentry *bank_debug_cfg; +}; + +struct adf_etr_data { + struct adf_etr_bank_data *banks; + struct dentry *debug; +}; + +void adf_response_handler(uintptr_t bank_addr); +#ifdef CONFIG_DEBUG_FS +#include <linux/debugfs.h> +int adf_bank_debugfs_add(struct adf_etr_bank_data *bank); +void adf_bank_debugfs_rm(struct adf_etr_bank_data *bank); +int adf_ring_debugfs_add(struct adf_etr_ring_data *ring, const char *name); +void adf_ring_debugfs_rm(struct adf_etr_ring_data *ring); +#else +static inline int adf_bank_debugfs_add(struct adf_etr_bank_data *bank) +{ + return 0; +} + +#define adf_bank_debugfs_rm(bank) do {} while (0) + +static inline int adf_ring_debugfs_add(struct adf_etr_ring_data *ring, + const char *name) +{ + return 0; +} + +#define adf_ring_debugfs_rm(ring) do {} while (0) +#endif +#endif diff --git a/drivers/crypto/intel/qat/qat_common/adf_vf_isr.c b/drivers/crypto/intel/qat/qat_common/adf_vf_isr.c new file mode 100644 index 000000000000..d0fef20a3df4 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/adf_vf_isr.c @@ -0,0 +1,314 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/pci.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/interrupt.h> +#include <linux/workqueue.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_cfg.h" +#include "adf_cfg_strings.h" +#include "adf_cfg_common.h" +#include "adf_pfvf_vf_msg.h" +#include "adf_transport_access_macros.h" +#include "adf_transport_internal.h" + +#define ADF_VINTSOU_OFFSET 0x204 +#define ADF_VINTMSK_OFFSET 0x208 +#define ADF_VINTSOU_BUN BIT(0) +#define ADF_VINTSOU_PF2VF BIT(1) + +static struct workqueue_struct *adf_vf_stop_wq; + +struct adf_vf_stop_data { + struct adf_accel_dev *accel_dev; + struct work_struct work; +}; + +void adf_enable_pf2vf_interrupts(struct adf_accel_dev *accel_dev) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + + ADF_CSR_WR(pmisc_addr, ADF_VINTMSK_OFFSET, 0x0); +} + +void adf_disable_pf2vf_interrupts(struct adf_accel_dev *accel_dev) +{ + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + + ADF_CSR_WR(pmisc_addr, ADF_VINTMSK_OFFSET, 0x2); +} +EXPORT_SYMBOL_GPL(adf_disable_pf2vf_interrupts); + +static int adf_enable_msi(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *pci_dev_info = &accel_dev->accel_pci_dev; + int stat = pci_alloc_irq_vectors(pci_dev_info->pci_dev, 1, 1, + PCI_IRQ_MSI); + if (unlikely(stat < 0)) { + dev_err(&GET_DEV(accel_dev), + "Failed to enable MSI interrupt: %d\n", stat); + return stat; + } + + return 0; +} + +static void adf_disable_msi(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + + pci_free_irq_vectors(pdev); +} + +static void adf_dev_stop_async(struct work_struct *work) +{ + struct adf_vf_stop_data *stop_data = + container_of(work, struct adf_vf_stop_data, work); + struct adf_accel_dev *accel_dev = stop_data->accel_dev; + + adf_dev_restarting_notify(accel_dev); + adf_dev_down(accel_dev); + + /* Re-enable PF2VF interrupts */ + adf_enable_pf2vf_interrupts(accel_dev); + adf_vf2pf_notify_restart_complete(accel_dev); + kfree(stop_data); +} + +int adf_pf2vf_handle_pf_restarting(struct adf_accel_dev *accel_dev) +{ + struct adf_vf_stop_data *stop_data; + + clear_bit(ADF_STATUS_PF_RUNNING, &accel_dev->status); + stop_data = kzalloc(sizeof(*stop_data), GFP_ATOMIC); + if (!stop_data) { + dev_err(&GET_DEV(accel_dev), + "Couldn't schedule stop for vf_%d\n", + accel_dev->accel_id); + return -ENOMEM; + } + stop_data->accel_dev = accel_dev; + INIT_WORK(&stop_data->work, adf_dev_stop_async); + queue_work(adf_vf_stop_wq, &stop_data->work); + + return 0; +} + +static void adf_pf2vf_bh_handler(void *data) +{ + struct adf_accel_dev *accel_dev = data; + bool ret; + + ret = adf_recv_and_handle_pf2vf_msg(accel_dev); + if (ret) + /* Re-enable PF2VF interrupts */ + adf_enable_pf2vf_interrupts(accel_dev); + + return; + +} + +static int adf_setup_pf2vf_bh(struct adf_accel_dev *accel_dev) +{ + tasklet_init(&accel_dev->vf.pf2vf_bh_tasklet, + (void *)adf_pf2vf_bh_handler, (unsigned long)accel_dev); + + mutex_init(&accel_dev->vf.vf2pf_lock); + return 0; +} + +static void adf_cleanup_pf2vf_bh(struct adf_accel_dev *accel_dev) +{ + tasklet_disable(&accel_dev->vf.pf2vf_bh_tasklet); + tasklet_kill(&accel_dev->vf.pf2vf_bh_tasklet); + mutex_destroy(&accel_dev->vf.vf2pf_lock); +} + +static irqreturn_t adf_isr(int irq, void *privdata) +{ + struct adf_accel_dev *accel_dev = privdata; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + struct adf_hw_csr_ops *csr_ops = &hw_data->csr_ops; + struct adf_bar *pmisc = + &GET_BARS(accel_dev)[hw_data->get_misc_bar_id(hw_data)]; + void __iomem *pmisc_bar_addr = pmisc->virt_addr; + bool handled = false; + u32 v_int, v_mask; + + /* Read VF INT source CSR to determine the source of VF interrupt */ + v_int = ADF_CSR_RD(pmisc_bar_addr, ADF_VINTSOU_OFFSET); + + /* Read VF INT mask CSR to determine which sources are masked */ + v_mask = ADF_CSR_RD(pmisc_bar_addr, ADF_VINTMSK_OFFSET); + + /* + * Recompute v_int ignoring sources that are masked. This is to + * avoid rescheduling the tasklet for interrupts already handled + */ + v_int &= ~v_mask; + + /* Check for PF2VF interrupt */ + if (v_int & ADF_VINTSOU_PF2VF) { + /* Disable PF to VF interrupt */ + adf_disable_pf2vf_interrupts(accel_dev); + + /* Schedule tasklet to handle interrupt BH */ + tasklet_hi_schedule(&accel_dev->vf.pf2vf_bh_tasklet); + handled = true; + } + + /* Check bundle interrupt */ + if (v_int & ADF_VINTSOU_BUN) { + struct adf_etr_data *etr_data = accel_dev->transport; + struct adf_etr_bank_data *bank = &etr_data->banks[0]; + + /* Disable Flag and Coalesce Ring Interrupts */ + csr_ops->write_csr_int_flag_and_col(bank->csr_addr, + bank->bank_number, 0); + tasklet_hi_schedule(&bank->resp_handler); + handled = true; + } + + return handled ? IRQ_HANDLED : IRQ_NONE; +} + +static int adf_request_msi_irq(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + unsigned int cpu; + int ret; + + snprintf(accel_dev->vf.irq_name, ADF_MAX_MSIX_VECTOR_NAME, + "qat_%02x:%02d.%02d", pdev->bus->number, PCI_SLOT(pdev->devfn), + PCI_FUNC(pdev->devfn)); + ret = request_irq(pdev->irq, adf_isr, 0, accel_dev->vf.irq_name, + (void *)accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), "failed to enable irq for %s\n", + accel_dev->vf.irq_name); + return ret; + } + cpu = accel_dev->accel_id % num_online_cpus(); + irq_set_affinity_hint(pdev->irq, get_cpu_mask(cpu)); + accel_dev->vf.irq_enabled = true; + + return ret; +} + +static int adf_setup_bh(struct adf_accel_dev *accel_dev) +{ + struct adf_etr_data *priv_data = accel_dev->transport; + + tasklet_init(&priv_data->banks[0].resp_handler, adf_response_handler, + (unsigned long)priv_data->banks); + return 0; +} + +static void adf_cleanup_bh(struct adf_accel_dev *accel_dev) +{ + struct adf_etr_data *priv_data = accel_dev->transport; + + tasklet_disable(&priv_data->banks[0].resp_handler); + tasklet_kill(&priv_data->banks[0].resp_handler); +} + +/** + * adf_vf_isr_resource_free() - Free IRQ for acceleration device + * @accel_dev: Pointer to acceleration device. + * + * Function frees interrupts for acceleration device virtual function. + */ +void adf_vf_isr_resource_free(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_to_pci_dev(accel_dev); + + if (accel_dev->vf.irq_enabled) { + irq_set_affinity_hint(pdev->irq, NULL); + free_irq(pdev->irq, accel_dev); + } + adf_cleanup_bh(accel_dev); + adf_cleanup_pf2vf_bh(accel_dev); + adf_disable_msi(accel_dev); +} +EXPORT_SYMBOL_GPL(adf_vf_isr_resource_free); + +/** + * adf_vf_isr_resource_alloc() - Allocate IRQ for acceleration device + * @accel_dev: Pointer to acceleration device. + * + * Function allocates interrupts for acceleration device virtual function. + * + * Return: 0 on success, error code otherwise. + */ +int adf_vf_isr_resource_alloc(struct adf_accel_dev *accel_dev) +{ + if (adf_enable_msi(accel_dev)) + goto err_out; + + if (adf_setup_pf2vf_bh(accel_dev)) + goto err_disable_msi; + + if (adf_setup_bh(accel_dev)) + goto err_cleanup_pf2vf_bh; + + if (adf_request_msi_irq(accel_dev)) + goto err_cleanup_bh; + + return 0; + +err_cleanup_bh: + adf_cleanup_bh(accel_dev); + +err_cleanup_pf2vf_bh: + adf_cleanup_pf2vf_bh(accel_dev); + +err_disable_msi: + adf_disable_msi(accel_dev); + +err_out: + return -EFAULT; +} +EXPORT_SYMBOL_GPL(adf_vf_isr_resource_alloc); + +/** + * adf_flush_vf_wq() - Flush workqueue for VF + * @accel_dev: Pointer to acceleration device. + * + * Function disables the PF/VF interrupts on the VF so that no new messages + * are received and flushes the workqueue 'adf_vf_stop_wq'. + * + * Return: void. + */ +void adf_flush_vf_wq(struct adf_accel_dev *accel_dev) +{ + adf_disable_pf2vf_interrupts(accel_dev); + + flush_workqueue(adf_vf_stop_wq); +} +EXPORT_SYMBOL_GPL(adf_flush_vf_wq); + +/** + * adf_init_vf_wq() - Init workqueue for VF + * + * Return: 0 on success, error code otherwise. + */ +int __init adf_init_vf_wq(void) +{ + adf_vf_stop_wq = alloc_workqueue("adf_vf_stop_wq", + WQ_MEM_RECLAIM | WQ_PERCPU, 0); + + return !adf_vf_stop_wq ? -EFAULT : 0; +} + +void adf_exit_vf_wq(void) +{ + if (adf_vf_stop_wq) + destroy_workqueue(adf_vf_stop_wq); + + adf_vf_stop_wq = NULL; +} diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_fw.h b/drivers/crypto/intel/qat/qat_common/icp_qat_fw.h new file mode 100644 index 000000000000..c141160421e1 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_fw.h @@ -0,0 +1,298 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef _ICP_QAT_FW_H_ +#define _ICP_QAT_FW_H_ +#include <linux/types.h> +#include "icp_qat_hw.h" + +#define QAT_FIELD_SET(flags, val, bitpos, mask) \ +{ (flags) = (((flags) & (~((mask) << (bitpos)))) | \ + (((val) & (mask)) << (bitpos))) ; } + +#define QAT_FIELD_GET(flags, bitpos, mask) \ + (((flags) >> (bitpos)) & (mask)) + +#define ICP_QAT_FW_REQ_DEFAULT_SZ 128 +#define ICP_QAT_FW_RESP_DEFAULT_SZ 32 +#define ICP_QAT_FW_COMN_ONE_BYTE_SHIFT 8 +#define ICP_QAT_FW_COMN_SINGLE_BYTE_MASK 0xFF +#define ICP_QAT_FW_NUM_LONGWORDS_1 1 +#define ICP_QAT_FW_NUM_LONGWORDS_2 2 +#define ICP_QAT_FW_NUM_LONGWORDS_3 3 +#define ICP_QAT_FW_NUM_LONGWORDS_4 4 +#define ICP_QAT_FW_NUM_LONGWORDS_5 5 +#define ICP_QAT_FW_NUM_LONGWORDS_6 6 +#define ICP_QAT_FW_NUM_LONGWORDS_7 7 +#define ICP_QAT_FW_NUM_LONGWORDS_10 10 +#define ICP_QAT_FW_NUM_LONGWORDS_13 13 +#define ICP_QAT_FW_NULL_REQ_SERV_ID 1 + +enum icp_qat_fw_comn_resp_serv_id { + ICP_QAT_FW_COMN_RESP_SERV_NULL, + ICP_QAT_FW_COMN_RESP_SERV_CPM_FW, + ICP_QAT_FW_COMN_RESP_SERV_DELIMITER +}; + +enum icp_qat_fw_comn_request_id { + ICP_QAT_FW_COMN_REQ_NULL = 0, + ICP_QAT_FW_COMN_REQ_CPM_FW_PKE = 3, + ICP_QAT_FW_COMN_REQ_CPM_FW_LA = 4, + ICP_QAT_FW_COMN_REQ_CPM_FW_DMA = 7, + ICP_QAT_FW_COMN_REQ_CPM_FW_COMP = 9, + ICP_QAT_FW_COMN_REQ_DELIMITER +}; + +struct icp_qat_fw_comn_req_hdr_cd_pars { + union { + struct { + __u64 content_desc_addr; + __u16 content_desc_resrvd1; + __u8 content_desc_params_sz; + __u8 content_desc_hdr_resrvd2; + __u32 content_desc_resrvd3; + } s; + struct { + __u32 serv_specif_fields[4]; + } s1; + } u; +}; + +struct icp_qat_fw_comn_req_mid { + __u64 opaque_data; + __u64 src_data_addr; + __u64 dest_data_addr; + __u32 src_length; + __u32 dst_length; +}; + +struct icp_qat_fw_comn_req_cd_ctrl { + __u32 content_desc_ctrl_lw[ICP_QAT_FW_NUM_LONGWORDS_5]; +}; + +struct icp_qat_fw_comn_req_hdr { + __u8 resrvd1; + __u8 service_cmd_id; + __u8 service_type; + __u8 hdr_flags; + __u16 serv_specif_flags; + __u16 comn_req_flags; +}; + +struct icp_qat_fw_comn_req_rqpars { + __u32 serv_specif_rqpars_lw[ICP_QAT_FW_NUM_LONGWORDS_13]; +}; + +struct icp_qat_fw_comn_req { + struct icp_qat_fw_comn_req_hdr comn_hdr; + struct icp_qat_fw_comn_req_hdr_cd_pars cd_pars; + struct icp_qat_fw_comn_req_mid comn_mid; + struct icp_qat_fw_comn_req_rqpars serv_specif_rqpars; + struct icp_qat_fw_comn_req_cd_ctrl cd_ctrl; +}; + +struct icp_qat_fw_comn_error { + __u8 xlat_err_code; + __u8 cmp_err_code; +}; + +struct icp_qat_fw_comn_resp_hdr { + __u8 resrvd1; + __u8 service_id; + __u8 response_type; + __u8 hdr_flags; + struct icp_qat_fw_comn_error comn_error; + __u8 comn_status; + __u8 cmd_id; +}; + +struct icp_qat_fw_comn_resp { + struct icp_qat_fw_comn_resp_hdr comn_hdr; + __u64 opaque_data; + __u32 resrvd[ICP_QAT_FW_NUM_LONGWORDS_4]; +}; + +#define ICP_QAT_FW_COMN_REQ_FLAG_SET 1 +#define ICP_QAT_FW_COMN_REQ_FLAG_CLR 0 +#define ICP_QAT_FW_COMN_VALID_FLAG_BITPOS 7 +#define ICP_QAT_FW_COMN_VALID_FLAG_MASK 0x1 +#define ICP_QAT_FW_COMN_HDR_RESRVD_FLD_MASK 0x7F +#define ICP_QAT_FW_COMN_CNV_FLAG_BITPOS 6 +#define ICP_QAT_FW_COMN_CNV_FLAG_MASK 0x1 +#define ICP_QAT_FW_COMN_CNVNR_FLAG_BITPOS 5 +#define ICP_QAT_FW_COMN_CNVNR_FLAG_MASK 0x1 + +#define ICP_QAT_FW_COMN_OV_SRV_TYPE_GET(icp_qat_fw_comn_req_hdr_t) \ + icp_qat_fw_comn_req_hdr_t.service_type + +#define ICP_QAT_FW_COMN_OV_SRV_TYPE_SET(icp_qat_fw_comn_req_hdr_t, val) \ + icp_qat_fw_comn_req_hdr_t.service_type = val + +#define ICP_QAT_FW_COMN_OV_SRV_CMD_ID_GET(icp_qat_fw_comn_req_hdr_t) \ + icp_qat_fw_comn_req_hdr_t.service_cmd_id + +#define ICP_QAT_FW_COMN_OV_SRV_CMD_ID_SET(icp_qat_fw_comn_req_hdr_t, val) \ + icp_qat_fw_comn_req_hdr_t.service_cmd_id = val + +#define ICP_QAT_FW_COMN_HDR_VALID_FLAG_GET(hdr_t) \ + ICP_QAT_FW_COMN_VALID_FLAG_GET(hdr_t.hdr_flags) + +#define ICP_QAT_FW_COMN_HDR_CNVNR_FLAG_GET(hdr_flags) \ + QAT_FIELD_GET(hdr_flags, \ + ICP_QAT_FW_COMN_CNVNR_FLAG_BITPOS, \ + ICP_QAT_FW_COMN_CNVNR_FLAG_MASK) + +#define ICP_QAT_FW_COMN_HDR_CNVNR_FLAG_SET(hdr_t, val) \ + QAT_FIELD_SET((hdr_t.hdr_flags), (val), \ + ICP_QAT_FW_COMN_CNVNR_FLAG_BITPOS, \ + ICP_QAT_FW_COMN_CNVNR_FLAG_MASK) + +#define ICP_QAT_FW_COMN_HDR_CNV_FLAG_GET(hdr_flags) \ + QAT_FIELD_GET(hdr_flags, \ + ICP_QAT_FW_COMN_CNV_FLAG_BITPOS, \ + ICP_QAT_FW_COMN_CNV_FLAG_MASK) + +#define ICP_QAT_FW_COMN_HDR_CNV_FLAG_SET(hdr_t, val) \ + QAT_FIELD_SET((hdr_t.hdr_flags), (val), \ + ICP_QAT_FW_COMN_CNV_FLAG_BITPOS, \ + ICP_QAT_FW_COMN_CNV_FLAG_MASK) + +#define ICP_QAT_FW_COMN_HDR_VALID_FLAG_SET(hdr_t, val) \ + ICP_QAT_FW_COMN_VALID_FLAG_SET(hdr_t, val) + +#define ICP_QAT_FW_COMN_VALID_FLAG_GET(hdr_flags) \ + QAT_FIELD_GET(hdr_flags, \ + ICP_QAT_FW_COMN_VALID_FLAG_BITPOS, \ + ICP_QAT_FW_COMN_VALID_FLAG_MASK) + +#define ICP_QAT_FW_COMN_HDR_RESRVD_FLD_GET(hdr_flags) \ + (hdr_flags & ICP_QAT_FW_COMN_HDR_RESRVD_FLD_MASK) + +#define ICP_QAT_FW_COMN_VALID_FLAG_SET(hdr_t, val) \ + QAT_FIELD_SET((hdr_t.hdr_flags), (val), \ + ICP_QAT_FW_COMN_VALID_FLAG_BITPOS, \ + ICP_QAT_FW_COMN_VALID_FLAG_MASK) + +#define ICP_QAT_FW_COMN_HDR_FLAGS_BUILD(valid) \ + (((valid) & ICP_QAT_FW_COMN_VALID_FLAG_MASK) << \ + ICP_QAT_FW_COMN_VALID_FLAG_BITPOS) + +#define QAT_COMN_PTR_TYPE_BITPOS 0 +#define QAT_COMN_PTR_TYPE_MASK 0x1 +#define QAT_COMN_CD_FLD_TYPE_BITPOS 1 +#define QAT_COMN_CD_FLD_TYPE_MASK 0x1 +#define QAT_COMN_PTR_TYPE_FLAT 0x0 +#define QAT_COMN_PTR_TYPE_SGL 0x1 +#define QAT_COMN_CD_FLD_TYPE_64BIT_ADR 0x0 +#define QAT_COMN_CD_FLD_TYPE_16BYTE_DATA 0x1 + +#define ICP_QAT_FW_COMN_FLAGS_BUILD(cdt, ptr) \ + ((((cdt) & QAT_COMN_CD_FLD_TYPE_MASK) << QAT_COMN_CD_FLD_TYPE_BITPOS) \ + | (((ptr) & QAT_COMN_PTR_TYPE_MASK) << QAT_COMN_PTR_TYPE_BITPOS)) + +#define ICP_QAT_FW_COMN_PTR_TYPE_GET(flags) \ + QAT_FIELD_GET(flags, QAT_COMN_PTR_TYPE_BITPOS, QAT_COMN_PTR_TYPE_MASK) + +#define ICP_QAT_FW_COMN_CD_FLD_TYPE_GET(flags) \ + QAT_FIELD_GET(flags, QAT_COMN_CD_FLD_TYPE_BITPOS, \ + QAT_COMN_CD_FLD_TYPE_MASK) + +#define ICP_QAT_FW_COMN_PTR_TYPE_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_COMN_PTR_TYPE_BITPOS, \ + QAT_COMN_PTR_TYPE_MASK) + +#define ICP_QAT_FW_COMN_CD_FLD_TYPE_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_COMN_CD_FLD_TYPE_BITPOS, \ + QAT_COMN_CD_FLD_TYPE_MASK) + +#define ICP_QAT_FW_COMN_NEXT_ID_BITPOS 4 +#define ICP_QAT_FW_COMN_NEXT_ID_MASK 0xF0 +#define ICP_QAT_FW_COMN_CURR_ID_BITPOS 0 +#define ICP_QAT_FW_COMN_CURR_ID_MASK 0x0F + +#define ICP_QAT_FW_COMN_NEXT_ID_GET(cd_ctrl_hdr_t) \ + ((((cd_ctrl_hdr_t)->next_curr_id) & ICP_QAT_FW_COMN_NEXT_ID_MASK) \ + >> (ICP_QAT_FW_COMN_NEXT_ID_BITPOS)) + +#define ICP_QAT_FW_COMN_NEXT_ID_SET(cd_ctrl_hdr_t, val) \ + { ((cd_ctrl_hdr_t)->next_curr_id) = ((((cd_ctrl_hdr_t)->next_curr_id) \ + & ICP_QAT_FW_COMN_CURR_ID_MASK) | \ + ((val << ICP_QAT_FW_COMN_NEXT_ID_BITPOS) \ + & ICP_QAT_FW_COMN_NEXT_ID_MASK)); } + +#define ICP_QAT_FW_COMN_CURR_ID_GET(cd_ctrl_hdr_t) \ + (((cd_ctrl_hdr_t)->next_curr_id) & ICP_QAT_FW_COMN_CURR_ID_MASK) + +#define ICP_QAT_FW_COMN_CURR_ID_SET(cd_ctrl_hdr_t, val) \ + { ((cd_ctrl_hdr_t)->next_curr_id) = ((((cd_ctrl_hdr_t)->next_curr_id) \ + & ICP_QAT_FW_COMN_NEXT_ID_MASK) | \ + ((val) & ICP_QAT_FW_COMN_CURR_ID_MASK)); } + +#define QAT_COMN_RESP_CRYPTO_STATUS_BITPOS 7 +#define QAT_COMN_RESP_CRYPTO_STATUS_MASK 0x1 +#define QAT_COMN_RESP_PKE_STATUS_BITPOS 6 +#define QAT_COMN_RESP_PKE_STATUS_MASK 0x1 +#define QAT_COMN_RESP_CMP_STATUS_BITPOS 5 +#define QAT_COMN_RESP_CMP_STATUS_MASK 0x1 +#define QAT_COMN_RESP_XLAT_STATUS_BITPOS 4 +#define QAT_COMN_RESP_XLAT_STATUS_MASK 0x1 +#define QAT_COMN_RESP_CMP_END_OF_LAST_BLK_BITPOS 3 +#define QAT_COMN_RESP_CMP_END_OF_LAST_BLK_MASK 0x1 + +#define ICP_QAT_FW_COMN_RESP_STATUS_BUILD(crypto, comp, xlat, eolb) \ + ((((crypto) & QAT_COMN_RESP_CRYPTO_STATUS_MASK) << \ + QAT_COMN_RESP_CRYPTO_STATUS_BITPOS) | \ + (((comp) & QAT_COMN_RESP_CMP_STATUS_MASK) << \ + QAT_COMN_RESP_CMP_STATUS_BITPOS) | \ + (((xlat) & QAT_COMN_RESP_XLAT_STATUS_MASK) << \ + QAT_COMN_RESP_XLAT_STATUS_BITPOS) | \ + (((eolb) & QAT_COMN_RESP_CMP_END_OF_LAST_BLK_MASK) << \ + QAT_COMN_RESP_CMP_END_OF_LAST_BLK_BITPOS)) + +#define ICP_QAT_FW_COMN_RESP_CRYPTO_STAT_GET(status) \ + QAT_FIELD_GET(status, QAT_COMN_RESP_CRYPTO_STATUS_BITPOS, \ + QAT_COMN_RESP_CRYPTO_STATUS_MASK) + +#define ICP_QAT_FW_COMN_RESP_CMP_STAT_GET(status) \ + QAT_FIELD_GET(status, QAT_COMN_RESP_CMP_STATUS_BITPOS, \ + QAT_COMN_RESP_CMP_STATUS_MASK) + +#define ICP_QAT_FW_COMN_RESP_XLAT_STAT_GET(status) \ + QAT_FIELD_GET(status, QAT_COMN_RESP_XLAT_STATUS_BITPOS, \ + QAT_COMN_RESP_XLAT_STATUS_MASK) + +#define ICP_QAT_FW_COMN_RESP_CMP_END_OF_LAST_BLK_FLAG_GET(status) \ + QAT_FIELD_GET(status, QAT_COMN_RESP_CMP_END_OF_LAST_BLK_BITPOS, \ + QAT_COMN_RESP_CMP_END_OF_LAST_BLK_MASK) + +#define ICP_QAT_FW_COMN_STATUS_FLAG_OK 0 +#define ICP_QAT_FW_COMN_STATUS_FLAG_ERROR 1 +#define ICP_QAT_FW_COMN_STATUS_CMP_END_OF_LAST_BLK_FLAG_CLR 0 +#define ICP_QAT_FW_COMN_STATUS_CMP_END_OF_LAST_BLK_FLAG_SET 1 +#define ERR_CODE_NO_ERROR 0 +#define ERR_CODE_INVALID_BLOCK_TYPE -1 +#define ERR_CODE_NO_MATCH_ONES_COMP -2 +#define ERR_CODE_TOO_MANY_LEN_OR_DIS -3 +#define ERR_CODE_INCOMPLETE_LEN -4 +#define ERR_CODE_RPT_LEN_NO_FIRST_LEN -5 +#define ERR_CODE_RPT_GT_SPEC_LEN -6 +#define ERR_CODE_INV_LIT_LEN_CODE_LEN -7 +#define ERR_CODE_INV_DIS_CODE_LEN -8 +#define ERR_CODE_INV_LIT_LEN_DIS_IN_BLK -9 +#define ERR_CODE_DIS_TOO_FAR_BACK -10 +#define ERR_CODE_OVERFLOW_ERROR -11 +#define ERR_CODE_SOFT_ERROR -12 +#define ERR_CODE_FATAL_ERROR -13 +#define ERR_CODE_SSM_ERROR -14 +#define ERR_CODE_ENDPOINT_ERROR -15 + +enum icp_qat_fw_slice { + ICP_QAT_FW_SLICE_NULL = 0, + ICP_QAT_FW_SLICE_CIPHER = 1, + ICP_QAT_FW_SLICE_AUTH = 2, + ICP_QAT_FW_SLICE_DRAM_RD = 3, + ICP_QAT_FW_SLICE_DRAM_WR = 4, + ICP_QAT_FW_SLICE_COMP = 5, + ICP_QAT_FW_SLICE_XLAT = 6, + ICP_QAT_FW_SLICE_DELIMITER +}; +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_fw_comp.h b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_comp.h new file mode 100644 index 000000000000..81969c515a17 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_comp.h @@ -0,0 +1,421 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef _ICP_QAT_FW_COMP_H_ +#define _ICP_QAT_FW_COMP_H_ +#include "icp_qat_fw.h" + +enum icp_qat_fw_comp_cmd_id { + ICP_QAT_FW_COMP_CMD_STATIC = 0, + ICP_QAT_FW_COMP_CMD_DYNAMIC = 1, + ICP_QAT_FW_COMP_CMD_DECOMPRESS = 2, + ICP_QAT_FW_COMP_CMD_DELIMITER +}; + +enum icp_qat_fw_comp_20_cmd_id { + ICP_QAT_FW_COMP_20_CMD_LZ4_COMPRESS = 3, + ICP_QAT_FW_COMP_20_CMD_LZ4_DECOMPRESS = 4, + ICP_QAT_FW_COMP_20_CMD_LZ4S_COMPRESS = 5, + ICP_QAT_FW_COMP_20_CMD_LZ4S_DECOMPRESS = 6, + ICP_QAT_FW_COMP_20_CMD_RESERVED_7 = 7, + ICP_QAT_FW_COMP_20_CMD_RESERVED_8 = 8, + ICP_QAT_FW_COMP_20_CMD_RESERVED_9 = 9, + ICP_QAT_FW_COMP_23_CMD_ZSTD_COMPRESS = 10, + ICP_QAT_FW_COMP_23_CMD_ZSTD_DECOMPRESS = 11, + ICP_QAT_FW_COMP_20_CMD_DELIMITER +}; + +#define ICP_QAT_FW_COMP_STATELESS_SESSION 0 +#define ICP_QAT_FW_COMP_STATEFUL_SESSION 1 +#define ICP_QAT_FW_COMP_NOT_AUTO_SELECT_BEST 0 +#define ICP_QAT_FW_COMP_AUTO_SELECT_BEST 1 +#define ICP_QAT_FW_COMP_NOT_ENH_AUTO_SELECT_BEST 0 +#define ICP_QAT_FW_COMP_ENH_AUTO_SELECT_BEST 1 +#define ICP_QAT_FW_COMP_NOT_DISABLE_TYPE0_ENH_AUTO_SELECT_BEST 0 +#define ICP_QAT_FW_COMP_DISABLE_TYPE0_ENH_AUTO_SELECT_BEST 1 +#define ICP_QAT_FW_COMP_DISABLE_SECURE_RAM_USED_AS_INTMD_BUF 1 +#define ICP_QAT_FW_COMP_ENABLE_SECURE_RAM_USED_AS_INTMD_BUF 0 +#define ICP_QAT_FW_COMP_SESSION_TYPE_BITPOS 2 +#define ICP_QAT_FW_COMP_SESSION_TYPE_MASK 0x1 +#define ICP_QAT_FW_COMP_AUTO_SELECT_BEST_BITPOS 3 +#define ICP_QAT_FW_COMP_AUTO_SELECT_BEST_MASK 0x1 +#define ICP_QAT_FW_COMP_ENHANCED_AUTO_SELECT_BEST_BITPOS 4 +#define ICP_QAT_FW_COMP_ENHANCED_AUTO_SELECT_BEST_MASK 0x1 +#define ICP_QAT_FW_COMP_RET_DISABLE_TYPE0_HEADER_DATA_BITPOS 5 +#define ICP_QAT_FW_COMP_RET_DISABLE_TYPE0_HEADER_DATA_MASK 0x1 +#define ICP_QAT_FW_COMP_DISABLE_SECURE_RAM_AS_INTMD_BUF_BITPOS 7 +#define ICP_QAT_FW_COMP_DISABLE_SECURE_RAM_AS_INTMD_BUF_MASK 0x1 +#define ICP_QAT_FW_COMP_AUTO_SELECT_BEST_MAX_VALUE 0xFFFFFFFF + +#define ICP_QAT_FW_COMP_FLAGS_BUILD(sesstype, autoselect, enhanced_asb, \ + ret_uncomp, secure_ram) \ + ((((sesstype) & ICP_QAT_FW_COMP_SESSION_TYPE_MASK) << \ + ICP_QAT_FW_COMP_SESSION_TYPE_BITPOS) | \ + (((autoselect) & ICP_QAT_FW_COMP_AUTO_SELECT_BEST_MASK) << \ + ICP_QAT_FW_COMP_AUTO_SELECT_BEST_BITPOS) | \ + (((enhanced_asb) & ICP_QAT_FW_COMP_ENHANCED_AUTO_SELECT_BEST_MASK) << \ + ICP_QAT_FW_COMP_ENHANCED_AUTO_SELECT_BEST_BITPOS) | \ + (((ret_uncomp) & ICP_QAT_FW_COMP_RET_DISABLE_TYPE0_HEADER_DATA_MASK) << \ + ICP_QAT_FW_COMP_RET_DISABLE_TYPE0_HEADER_DATA_BITPOS) | \ + (((secure_ram) & ICP_QAT_FW_COMP_DISABLE_SECURE_RAM_AS_INTMD_BUF_MASK) << \ + ICP_QAT_FW_COMP_DISABLE_SECURE_RAM_AS_INTMD_BUF_BITPOS)) + +#define ICP_QAT_FW_COMP_SESSION_TYPE_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_SESSION_TYPE_BITPOS, \ + ICP_QAT_FW_COMP_SESSION_TYPE_MASK) + +#define ICP_QAT_FW_COMP_SESSION_TYPE_SET(flags, val) \ + QAT_FIELD_SET(flags, val, ICP_QAT_FW_COMP_SESSION_TYPE_BITPOS, \ + ICP_QAT_FW_COMP_SESSION_TYPE_MASK) + +#define ICP_QAT_FW_COMP_AUTO_SELECT_BEST_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_AUTO_SELECT_BEST_BITPOS, \ + ICP_QAT_FW_COMP_AUTO_SELECT_BEST_MASK) + +#define ICP_QAT_FW_COMP_EN_ASB_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_ENHANCED_AUTO_SELECT_BEST_BITPOS, \ + ICP_QAT_FW_COMP_ENHANCED_AUTO_SELECT_BEST_MASK) + +#define ICP_QAT_FW_COMP_RET_UNCOMP_GET(flags) \ + QAT_FIELD_GET(flags, \ + ICP_QAT_FW_COMP_RET_DISABLE_TYPE0_HEADER_DATA_BITPOS, \ + ICP_QAT_FW_COMP_RET_DISABLE_TYPE0_HEADER_DATA_MASK) + +#define ICP_QAT_FW_COMP_SECURE_RAM_USE_GET(flags) \ + QAT_FIELD_GET(flags, \ + ICP_QAT_FW_COMP_DISABLE_SECURE_RAM_AS_INTMD_BUF_BITPOS, \ + ICP_QAT_FW_COMP_DISABLE_SECURE_RAM_AS_INTMD_BUF_MASK) + +struct icp_qat_fw_comp_req_hdr_cd_pars { + union { + struct { + __u64 content_desc_addr; + __u16 content_desc_resrvd1; + __u8 content_desc_params_sz; + __u8 content_desc_hdr_resrvd2; + __u32 content_desc_resrvd3; + } s; + struct { + __u32 comp_slice_cfg_word[ICP_QAT_FW_NUM_LONGWORDS_2]; + __u32 content_desc_resrvd4; + } sl; + } u; +}; + +struct icp_qat_fw_comp_req_params { + __u32 comp_len; + __u32 out_buffer_sz; + union { + struct { + __u32 initial_crc32; + __u32 initial_adler; + } legacy; + __u64 crc_data_addr; + } crc; + __u32 req_par_flags; + __u32 rsrvd; +}; + +#define ICP_QAT_FW_COMP_REQ_PARAM_FLAGS_BUILD(sop, eop, bfinal, cnv, cnvnr, \ + cnvdfx, crc, xxhash_acc, \ + cnv_error_type, append_crc, \ + drop_data, partial_decomp) \ + ((((sop) & ICP_QAT_FW_COMP_SOP_MASK) << \ + ICP_QAT_FW_COMP_SOP_BITPOS) | \ + (((eop) & ICP_QAT_FW_COMP_EOP_MASK) << \ + ICP_QAT_FW_COMP_EOP_BITPOS) | \ + (((bfinal) & ICP_QAT_FW_COMP_BFINAL_MASK) \ + << ICP_QAT_FW_COMP_BFINAL_BITPOS) | \ + (((cnv) & ICP_QAT_FW_COMP_CNV_MASK) << \ + ICP_QAT_FW_COMP_CNV_BITPOS) | \ + (((cnvnr) & ICP_QAT_FW_COMP_CNVNR_MASK) \ + << ICP_QAT_FW_COMP_CNVNR_BITPOS) | \ + (((cnvdfx) & ICP_QAT_FW_COMP_CNV_DFX_MASK) \ + << ICP_QAT_FW_COMP_CNV_DFX_BITPOS) | \ + (((crc) & ICP_QAT_FW_COMP_CRC_MODE_MASK) \ + << ICP_QAT_FW_COMP_CRC_MODE_BITPOS) | \ + (((xxhash_acc) & ICP_QAT_FW_COMP_XXHASH_ACC_MODE_MASK) \ + << ICP_QAT_FW_COMP_XXHASH_ACC_MODE_BITPOS) | \ + (((cnv_error_type) & ICP_QAT_FW_COMP_CNV_ERROR_MASK) \ + << ICP_QAT_FW_COMP_CNV_ERROR_BITPOS) | \ + (((append_crc) & ICP_QAT_FW_COMP_APPEND_CRC_MASK) \ + << ICP_QAT_FW_COMP_APPEND_CRC_BITPOS) | \ + (((drop_data) & ICP_QAT_FW_COMP_DROP_DATA_MASK) \ + << ICP_QAT_FW_COMP_DROP_DATA_BITPOS) | \ + (((partial_decomp) & ICP_QAT_FW_COMP_PARTIAL_DECOMP_MASK) \ + << ICP_QAT_FW_COMP_PARTIAL_DECOMP_BITPOS)) + +#define ICP_QAT_FW_COMP_NOT_SOP 0 +#define ICP_QAT_FW_COMP_SOP 1 +#define ICP_QAT_FW_COMP_NOT_EOP 0 +#define ICP_QAT_FW_COMP_EOP 1 +#define ICP_QAT_FW_COMP_NOT_BFINAL 0 +#define ICP_QAT_FW_COMP_BFINAL 1 +#define ICP_QAT_FW_COMP_NO_CNV 0 +#define ICP_QAT_FW_COMP_CNV 1 +#define ICP_QAT_FW_COMP_NO_CNV_RECOVERY 0 +#define ICP_QAT_FW_COMP_CNV_RECOVERY 1 +#define ICP_QAT_FW_COMP_NO_CNV_DFX 0 +#define ICP_QAT_FW_COMP_CNV_DFX 1 +#define ICP_QAT_FW_COMP_CRC_MODE_LEGACY 0 +#define ICP_QAT_FW_COMP_CRC_MODE_E2E 1 +#define ICP_QAT_FW_COMP_NO_XXHASH_ACC 0 +#define ICP_QAT_FW_COMP_XXHASH_ACC 1 +#define ICP_QAT_FW_COMP_APPEND_CRC 1 +#define ICP_QAT_FW_COMP_NO_APPEND_CRC 0 +#define ICP_QAT_FW_COMP_DROP_DATA 1 +#define ICP_QAT_FW_COMP_NO_DROP_DATA 0 +#define ICP_QAT_FW_COMP_PARTIAL_DECOMPRESS 1 +#define ICP_QAT_FW_COMP_NO_PARTIAL_DECOMPRESS 0 +#define ICP_QAT_FW_COMP_SOP_BITPOS 0 +#define ICP_QAT_FW_COMP_SOP_MASK 0x1 +#define ICP_QAT_FW_COMP_EOP_BITPOS 1 +#define ICP_QAT_FW_COMP_EOP_MASK 0x1 +#define ICP_QAT_FW_COMP_BFINAL_BITPOS 6 +#define ICP_QAT_FW_COMP_BFINAL_MASK 0x1 +#define ICP_QAT_FW_COMP_CNV_BITPOS 16 +#define ICP_QAT_FW_COMP_CNV_MASK 0x1 +#define ICP_QAT_FW_COMP_CNVNR_BITPOS 17 +#define ICP_QAT_FW_COMP_CNVNR_MASK 0x1 +#define ICP_QAT_FW_COMP_CNV_DFX_BITPOS 18 +#define ICP_QAT_FW_COMP_CNV_DFX_MASK 0x1 +#define ICP_QAT_FW_COMP_CRC_MODE_BITPOS 19 +#define ICP_QAT_FW_COMP_CRC_MODE_MASK 0x1 +#define ICP_QAT_FW_COMP_XXHASH_ACC_MODE_BITPOS 20 +#define ICP_QAT_FW_COMP_XXHASH_ACC_MODE_MASK 0x1 +#define ICP_QAT_FW_COMP_CNV_ERROR_BITPOS 21 +#define ICP_QAT_FW_COMP_CNV_ERROR_MASK 0b111 +#define ICP_QAT_FW_COMP_CNV_ERROR_NONE 0b000 +#define ICP_QAT_FW_COMP_CNV_ERROR_CHECKSUM 0b001 +#define ICP_QAT_FW_COMP_CNV_ERROR_DCPR_OBC_DIFF 0b010 +#define ICP_QAT_FW_COMP_CNV_ERROR_DCPR 0b011 +#define ICP_QAT_FW_COMP_CNV_ERROR_XLT 0b100 +#define ICP_QAT_FW_COMP_CNV_ERROR_DCPR_IBC_DIFF 0b101 +#define ICP_QAT_FW_COMP_APPEND_CRC_BITPOS 24 +#define ICP_QAT_FW_COMP_APPEND_CRC_MASK 0x1 +#define ICP_QAT_FW_COMP_DROP_DATA_BITPOS 25 +#define ICP_QAT_FW_COMP_DROP_DATA_MASK 0x1 +#define ICP_QAT_FW_COMP_PARTIAL_DECOMP_BITPOS 27 +#define ICP_QAT_FW_COMP_PARTIAL_DECOMP_MASK 0x1 + +#define ICP_QAT_FW_COMP_SOP_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_SOP_BITPOS, \ + ICP_QAT_FW_COMP_SOP_MASK) + +#define ICP_QAT_FW_COMP_SOP_SET(flags, val) \ + QAT_FIELD_SET(flags, val, ICP_QAT_FW_COMP_SOP_BITPOS, \ + ICP_QAT_FW_COMP_SOP_MASK) + +#define ICP_QAT_FW_COMP_EOP_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_EOP_BITPOS, \ + ICP_QAT_FW_COMP_EOP_MASK) + +#define ICP_QAT_FW_COMP_EOP_SET(flags, val) \ + QAT_FIELD_SET(flags, val, ICP_QAT_FW_COMP_EOP_BITPOS, \ + ICP_QAT_FW_COMP_EOP_MASK) + +#define ICP_QAT_FW_COMP_BFINAL_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_BFINAL_BITPOS, \ + ICP_QAT_FW_COMP_BFINAL_MASK) + +#define ICP_QAT_FW_COMP_BFINAL_SET(flags, val) \ + QAT_FIELD_SET(flags, val, ICP_QAT_FW_COMP_BFINAL_BITPOS, \ + ICP_QAT_FW_COMP_BFINAL_MASK) + +#define ICP_QAT_FW_COMP_CNV_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_CNV_BITPOS, \ + ICP_QAT_FW_COMP_CNV_MASK) + +#define ICP_QAT_FW_COMP_CNVNR_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_CNVNR_BITPOS, \ + ICP_QAT_FW_COMP_CNVNR_MASK) + +#define ICP_QAT_FW_COMP_CNV_DFX_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_CNV_DFX_BITPOS, \ + ICP_QAT_FW_COMP_CNV_DFX_MASK) + +#define ICP_QAT_FW_COMP_CNV_DFX_SET(flags, val) \ + QAT_FIELD_SET(flags, val, ICP_QAT_FW_COMP_CNV_DFX_BITPOS, \ + ICP_QAT_FW_COMP_CNV_DFX_MASK) + +#define ICP_QAT_FW_COMP_CRC_MODE_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_CRC_MODE_BITPOS, \ + ICP_QAT_FW_COMP_CRC_MODE_MASK) + +#define ICP_QAT_FW_COMP_XXHASH_ACC_MODE_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_XXHASH_ACC_MODE_BITPOS, \ + ICP_QAT_FW_COMP_XXHASH_ACC_MODE_MASK) + +#define ICP_QAT_FW_COMP_XXHASH_ACC_MODE_SET(flags, val) \ + QAT_FIELD_SET(flags, val, ICP_QAT_FW_COMP_XXHASH_ACC_MODE_BITPOS, \ + ICP_QAT_FW_COMP_XXHASH_ACC_MODE_MASK) + +#define ICP_QAT_FW_COMP_CNV_ERROR_TYPE_GET(flags) \ + QAT_FIELD_GET(flags, ICP_QAT_FW_COMP_CNV_ERROR_BITPOS, \ + ICP_QAT_FW_COMP_CNV_ERROR_MASK) + +#define ICP_QAT_FW_COMP_CNV_ERROR_TYPE_SET(flags, val) \ + QAT_FIELD_SET(flags, val, ICP_QAT_FW_COMP_CNV_ERROR_BITPOS, \ + ICP_QAT_FW_COMP_CNV_ERROR_MASK) + +struct icp_qat_fw_xlt_req_params { + __u64 inter_buff_ptr; +}; + +struct icp_qat_fw_comp_cd_hdr { + __u16 ram_bank_flags; + __u8 comp_cfg_offset; + __u8 next_curr_id; + __u32 resrvd; + __u64 comp_state_addr; + __u64 ram_banks_addr; +}; + +#define COMP_CPR_INITIAL_CRC 0 +#define COMP_CPR_INITIAL_ADLER 1 + +struct icp_qat_fw_xlt_cd_hdr { + __u16 resrvd1; + __u8 resrvd2; + __u8 next_curr_id; + __u32 resrvd3; +}; + +struct icp_qat_fw_comp_req { + struct icp_qat_fw_comn_req_hdr comn_hdr; + struct icp_qat_fw_comp_req_hdr_cd_pars cd_pars; + struct icp_qat_fw_comn_req_mid comn_mid; + struct icp_qat_fw_comp_req_params comp_pars; + union { + struct icp_qat_fw_xlt_req_params xlt_pars; + __u32 resrvd1[ICP_QAT_FW_NUM_LONGWORDS_2]; + struct { + __u32 partial_decompress_length; + __u32 partial_decompress_offset; + } partial_decompress; + } u1; + union { + __u32 resrvd2[ICP_QAT_FW_NUM_LONGWORDS_2]; + struct { + __u32 asb_value; + __u32 reserved; + } asb_threshold; + } u3; + struct icp_qat_fw_comp_cd_hdr comp_cd_ctrl; + union { + struct icp_qat_fw_xlt_cd_hdr xlt_cd_ctrl; + __u32 resrvd3[ICP_QAT_FW_NUM_LONGWORDS_2]; + } u2; +}; + +struct icp_qat_fw_resp_comp_pars { + __u32 input_byte_counter; + __u32 output_byte_counter; + union { + struct { + __u32 curr_crc32; + __u32 curr_adler_32; + } legacy; + __u32 resrvd[ICP_QAT_FW_NUM_LONGWORDS_2]; + } crc; +}; + +struct icp_qat_fw_comp_state { + __u32 rd8_counter; + __u32 status_flags; + __u32 in_counter; + __u32 out_counter; + __u64 intermediate_state; + __u32 lobc; + __u32 replaybc; + __u64 pcrc64_poly; + __u32 crc32; + __u32 adler_xxhash32; + __u64 pcrc64_xorout; + __u32 out_buf_size; + __u32 in_buf_size; + __u64 in_pcrc64; + __u64 out_pcrc64; + __u32 lobs; + __u32 libc; + __u64 reserved; + __u32 xxhash_state[4]; + __u32 cleartext[4]; +}; + +struct icp_qat_fw_comp_resp { + struct icp_qat_fw_comn_resp_hdr comn_resp; + __u64 opaque_data; + struct icp_qat_fw_resp_comp_pars comp_resp_pars; +}; + +#define QAT_FW_COMP_BANK_FLAG_MASK 0x1 +#define QAT_FW_COMP_BANK_I_BITPOS 8 +#define QAT_FW_COMP_BANK_H_BITPOS 7 +#define QAT_FW_COMP_BANK_G_BITPOS 6 +#define QAT_FW_COMP_BANK_F_BITPOS 5 +#define QAT_FW_COMP_BANK_E_BITPOS 4 +#define QAT_FW_COMP_BANK_D_BITPOS 3 +#define QAT_FW_COMP_BANK_C_BITPOS 2 +#define QAT_FW_COMP_BANK_B_BITPOS 1 +#define QAT_FW_COMP_BANK_A_BITPOS 0 + +enum icp_qat_fw_comp_bank_enabled { + ICP_QAT_FW_COMP_BANK_DISABLED = 0, + ICP_QAT_FW_COMP_BANK_ENABLED = 1, + ICP_QAT_FW_COMP_BANK_DELIMITER = 2 +}; + +#define ICP_QAT_FW_COMP_RAM_FLAGS_BUILD(bank_i_enable, bank_h_enable, \ + bank_g_enable, bank_f_enable, \ + bank_e_enable, bank_d_enable, \ + bank_c_enable, bank_b_enable, \ + bank_a_enable) \ + ((((bank_i_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_I_BITPOS) | \ + (((bank_h_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_H_BITPOS) | \ + (((bank_g_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_G_BITPOS) | \ + (((bank_f_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_F_BITPOS) | \ + (((bank_e_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_E_BITPOS) | \ + (((bank_d_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_D_BITPOS) | \ + (((bank_c_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_C_BITPOS) | \ + (((bank_b_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_B_BITPOS) | \ + (((bank_a_enable) & QAT_FW_COMP_BANK_FLAG_MASK) << \ + QAT_FW_COMP_BANK_A_BITPOS)) + +struct icp_qat_fw_comp_crc_data_struct { + __u32 crc32; + union { + __u32 adler; + __u32 xxhash; + } adler_xxhash_u; + __u32 cpr_in_crc_lo; + __u32 cpr_in_crc_hi; + __u32 cpr_out_crc_lo; + __u32 cpr_out_crc_hi; + __u32 xlt_in_crc_lo; + __u32 xlt_in_crc_hi; + __u32 xlt_out_crc_lo; + __u32 xlt_out_crc_hi; + __u32 prog_crc_poly_lo; + __u32 prog_crc_poly_hi; + __u32 xor_out_lo; + __u32 xor_out_hi; + __u32 append_crc_lo; + __u32 append_crc_hi; +}; + +struct xxhash_acc_state_buff { + __u32 in_counter; + __u32 out_counter; + __u32 xxhash_state[4]; + __u32 clear_txt[4]; +}; + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_fw_init_admin.h b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_init_admin.h new file mode 100644 index 000000000000..63cf18e2a4e5 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_init_admin.h @@ -0,0 +1,202 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef _ICP_QAT_FW_INIT_ADMIN_H_ +#define _ICP_QAT_FW_INIT_ADMIN_H_ + +#include "icp_qat_fw.h" + +#define RL_MAX_RP_IDS 16 + +enum icp_qat_fw_init_admin_cmd_id { + ICP_QAT_FW_INIT_AE = 0, + ICP_QAT_FW_TRNG_ENABLE = 1, + ICP_QAT_FW_TRNG_DISABLE = 2, + ICP_QAT_FW_CONSTANTS_CFG = 3, + ICP_QAT_FW_STATUS_GET = 4, + ICP_QAT_FW_COUNTERS_GET = 5, + ICP_QAT_FW_LOOPBACK = 6, + ICP_QAT_FW_HEARTBEAT_SYNC = 7, + ICP_QAT_FW_HEARTBEAT_GET = 8, + ICP_QAT_FW_COMP_CAPABILITY_GET = 9, + ICP_QAT_FW_CRYPTO_CAPABILITY_GET = 10, + ICP_QAT_FW_DC_CHAIN_INIT = 11, + ICP_QAT_FW_HEARTBEAT_TIMER_SET = 13, + ICP_QAT_FW_RL_INIT = 15, + ICP_QAT_FW_TIMER_GET = 19, + ICP_QAT_FW_CNV_STATS_GET = 20, + ICP_QAT_FW_PM_STATE_CONFIG = 128, + ICP_QAT_FW_PM_INFO = 129, + ICP_QAT_FW_RL_ADD = 134, + ICP_QAT_FW_RL_UPDATE = 135, + ICP_QAT_FW_RL_REMOVE = 136, + ICP_QAT_FW_TL_START = 137, + ICP_QAT_FW_TL_STOP = 138, +}; + +enum icp_qat_fw_init_admin_resp_status { + ICP_QAT_FW_INIT_RESP_STATUS_SUCCESS = 0, + ICP_QAT_FW_INIT_RESP_STATUS_FAIL +}; + +struct icp_qat_fw_init_admin_tl_rp_indexes { + __u8 rp_num_index_0; + __u8 rp_num_index_1; + __u8 rp_num_index_2; + __u8 rp_num_index_3; +}; + +struct icp_qat_fw_init_admin_slice_cnt { + __u8 cpr_cnt; + __u8 xlt_cnt; + __u8 dcpr_cnt; + __u8 pke_cnt; + __u8 wat_cnt; + __u8 wcp_cnt; + __u8 ucs_cnt; + __u8 cph_cnt; + __u8 ath_cnt; +}; + +struct icp_qat_fw_init_admin_sla_config_params { + __u32 pcie_in_cir; + __u32 pcie_in_pir; + __u32 pcie_out_cir; + __u32 pcie_out_pir; + __u32 slice_util_cir; + __u32 slice_util_pir; + __u32 ae_util_cir; + __u32 ae_util_pir; + __u16 rp_ids[RL_MAX_RP_IDS]; +}; + +struct icp_qat_fw_init_admin_req { + __u16 init_cfg_sz; + __u8 resrvd1; + __u8 cmd_id; + __u32 resrvd2; + __u64 opaque_data; + __u64 init_cfg_ptr; + + union { + struct { + __u16 ibuf_size_in_kb; + __u16 resrvd3; + }; + struct { + __u32 int_timer_ticks; + }; + struct { + __u32 heartbeat_ticks; + }; + struct { + __u16 node_id; + __u8 node_type; + __u8 svc_type; + __u8 resrvd5[3]; + __u8 rp_count; + }; + __u32 idle_filter; + struct icp_qat_fw_init_admin_tl_rp_indexes rp_indexes; + }; + + __u32 resrvd4; +} __packed; + +struct icp_qat_fw_init_admin_resp { + __u8 flags; + __u8 resrvd1; + __u8 status; + __u8 cmd_id; + union { + __u32 resrvd2; + struct { + __u16 version_minor_num; + __u16 version_major_num; + }; + __u32 extended_features; + struct { + __u16 error_count; + __u16 latest_error; + }; + }; + __u64 opaque_data; + union { + __u32 resrvd3[ICP_QAT_FW_NUM_LONGWORDS_4]; + struct { + __u32 version_patch_num; + __u8 context_id; + __u8 ae_id; + __u16 resrvd4; + __u64 resrvd5; + }; + struct { + __u64 req_rec_count; + __u64 resp_sent_count; + }; + struct { + __u16 compression_algos; + __u16 checksum_algos; + __u32 deflate_capabilities; + __u32 resrvd6; + __u32 lzs_capabilities; + }; + struct { + __u32 cipher_algos; + __u32 hash_algos; + __u16 keygen_algos; + __u16 other; + __u16 public_key_algos; + __u16 prime_algos; + }; + struct { + __u64 timestamp; + __u64 resrvd7; + }; + struct { + __u32 successful_count; + __u32 unsuccessful_count; + __u64 resrvd8; + }; + struct icp_qat_fw_init_admin_slice_cnt slices; + __u16 fw_capabilities; + }; +} __packed; + +#define ICP_QAT_FW_SYNC ICP_QAT_FW_HEARTBEAT_SYNC +#define ICP_QAT_FW_CAPABILITIES_GET ICP_QAT_FW_CRYPTO_CAPABILITY_GET + +#define ICP_QAT_NUMBER_OF_PM_EVENTS 8 + +struct icp_qat_fw_init_admin_pm_info { + __u16 max_pwrreq; + __u16 min_pwrreq; + __u16 resvrd1; + __u8 pwr_state; + __u8 resvrd2; + __u32 fusectl0; + struct_group(event_counters, + __u32 sys_pm; + __u32 host_msg; + __u32 unknown; + __u32 local_ssm; + __u32 timer; + ); + __u32 event_log[ICP_QAT_NUMBER_OF_PM_EVENTS]; + struct_group(pm, + __u32 fw_init; + __u32 pwrreq; + __u32 status; + __u32 main; + __u32 thread; + ); + struct_group(ssm, + __u32 pm_enable; + __u32 pm_active_status; + __u32 pm_managed_status; + __u32 pm_domain_status; + __u32 active_constraint; + ); + __u32 resvrd3[6]; +}; + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_fw_la.h b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_la.h new file mode 100644 index 000000000000..28fa17f14be4 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_la.h @@ -0,0 +1,367 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef _ICP_QAT_FW_LA_H_ +#define _ICP_QAT_FW_LA_H_ +#include "icp_qat_fw.h" + +enum icp_qat_fw_la_cmd_id { + ICP_QAT_FW_LA_CMD_CIPHER = 0, + ICP_QAT_FW_LA_CMD_AUTH = 1, + ICP_QAT_FW_LA_CMD_CIPHER_HASH = 2, + ICP_QAT_FW_LA_CMD_HASH_CIPHER = 3, + ICP_QAT_FW_LA_CMD_TRNG_GET_RANDOM = 4, + ICP_QAT_FW_LA_CMD_TRNG_TEST = 5, + ICP_QAT_FW_LA_CMD_SSL3_KEY_DERIVE = 6, + ICP_QAT_FW_LA_CMD_TLS_V1_1_KEY_DERIVE = 7, + ICP_QAT_FW_LA_CMD_TLS_V1_2_KEY_DERIVE = 8, + ICP_QAT_FW_LA_CMD_MGF1 = 9, + ICP_QAT_FW_LA_CMD_AUTH_PRE_COMP = 10, + ICP_QAT_FW_LA_CMD_CIPHER_PRE_COMP = 11, + ICP_QAT_FW_LA_CMD_DELIMITER = 12 +}; + +#define ICP_QAT_FW_LA_ICV_VER_STATUS_PASS ICP_QAT_FW_COMN_STATUS_FLAG_OK +#define ICP_QAT_FW_LA_ICV_VER_STATUS_FAIL ICP_QAT_FW_COMN_STATUS_FLAG_ERROR +#define ICP_QAT_FW_LA_TRNG_STATUS_PASS ICP_QAT_FW_COMN_STATUS_FLAG_OK +#define ICP_QAT_FW_LA_TRNG_STATUS_FAIL ICP_QAT_FW_COMN_STATUS_FLAG_ERROR + +struct icp_qat_fw_la_bulk_req { + struct icp_qat_fw_comn_req_hdr comn_hdr; + struct icp_qat_fw_comn_req_hdr_cd_pars cd_pars; + struct icp_qat_fw_comn_req_mid comn_mid; + struct icp_qat_fw_comn_req_rqpars serv_specif_rqpars; + struct icp_qat_fw_comn_req_cd_ctrl cd_ctrl; +}; + +#define ICP_QAT_FW_LA_USE_UCS_SLICE_TYPE 1 +#define QAT_LA_SLICE_TYPE_BITPOS 14 +#define QAT_LA_SLICE_TYPE_MASK 0x3 +#define ICP_QAT_FW_LA_GCM_IV_LEN_12_OCTETS 1 +#define ICP_QAT_FW_LA_GCM_IV_LEN_NOT_12_OCTETS 0 +#define QAT_FW_LA_ZUC_3G_PROTO_FLAG_BITPOS 12 +#define ICP_QAT_FW_LA_ZUC_3G_PROTO 1 +#define QAT_FW_LA_ZUC_3G_PROTO_FLAG_MASK 0x1 +#define QAT_LA_GCM_IV_LEN_FLAG_BITPOS 11 +#define QAT_LA_GCM_IV_LEN_FLAG_MASK 0x1 +#define ICP_QAT_FW_LA_DIGEST_IN_BUFFER 1 +#define ICP_QAT_FW_LA_NO_DIGEST_IN_BUFFER 0 +#define QAT_LA_DIGEST_IN_BUFFER_BITPOS 10 +#define QAT_LA_DIGEST_IN_BUFFER_MASK 0x1 +#define ICP_QAT_FW_LA_SNOW_3G_PROTO 4 +#define ICP_QAT_FW_LA_GCM_PROTO 2 +#define ICP_QAT_FW_LA_CCM_PROTO 1 +#define ICP_QAT_FW_LA_NO_PROTO 0 +#define QAT_LA_PROTO_BITPOS 7 +#define QAT_LA_PROTO_MASK 0x7 +#define ICP_QAT_FW_LA_CMP_AUTH_RES 1 +#define ICP_QAT_FW_LA_NO_CMP_AUTH_RES 0 +#define QAT_LA_CMP_AUTH_RES_BITPOS 6 +#define QAT_LA_CMP_AUTH_RES_MASK 0x1 +#define ICP_QAT_FW_LA_RET_AUTH_RES 1 +#define ICP_QAT_FW_LA_NO_RET_AUTH_RES 0 +#define QAT_LA_RET_AUTH_RES_BITPOS 5 +#define QAT_LA_RET_AUTH_RES_MASK 0x1 +#define ICP_QAT_FW_LA_UPDATE_STATE 1 +#define ICP_QAT_FW_LA_NO_UPDATE_STATE 0 +#define QAT_LA_UPDATE_STATE_BITPOS 4 +#define QAT_LA_UPDATE_STATE_MASK 0x1 +#define ICP_QAT_FW_CIPH_AUTH_CFG_OFFSET_IN_CD_SETUP 0 +#define ICP_QAT_FW_CIPH_AUTH_CFG_OFFSET_IN_SHRAM_CP 1 +#define QAT_LA_CIPH_AUTH_CFG_OFFSET_BITPOS 3 +#define QAT_LA_CIPH_AUTH_CFG_OFFSET_MASK 0x1 +#define ICP_QAT_FW_CIPH_IV_64BIT_PTR 0 +#define ICP_QAT_FW_CIPH_IV_16BYTE_DATA 1 +#define QAT_LA_CIPH_IV_FLD_BITPOS 2 +#define QAT_LA_CIPH_IV_FLD_MASK 0x1 +#define ICP_QAT_FW_LA_PARTIAL_NONE 0 +#define ICP_QAT_FW_LA_PARTIAL_START 1 +#define ICP_QAT_FW_LA_PARTIAL_MID 3 +#define ICP_QAT_FW_LA_PARTIAL_END 2 +#define QAT_LA_PARTIAL_BITPOS 0 +#define QAT_LA_PARTIAL_MASK 0x3 +#define ICP_QAT_FW_LA_FLAGS_BUILD(zuc_proto, gcm_iv_len, auth_rslt, proto, \ + cmp_auth, ret_auth, update_state, \ + ciph_iv, ciphcfg, partial) \ + (((zuc_proto & QAT_FW_LA_ZUC_3G_PROTO_FLAG_MASK) << \ + QAT_FW_LA_ZUC_3G_PROTO_FLAG_BITPOS) | \ + ((gcm_iv_len & QAT_LA_GCM_IV_LEN_FLAG_MASK) << \ + QAT_LA_GCM_IV_LEN_FLAG_BITPOS) | \ + ((auth_rslt & QAT_LA_DIGEST_IN_BUFFER_MASK) << \ + QAT_LA_DIGEST_IN_BUFFER_BITPOS) | \ + ((proto & QAT_LA_PROTO_MASK) << \ + QAT_LA_PROTO_BITPOS) | \ + ((cmp_auth & QAT_LA_CMP_AUTH_RES_MASK) << \ + QAT_LA_CMP_AUTH_RES_BITPOS) | \ + ((ret_auth & QAT_LA_RET_AUTH_RES_MASK) << \ + QAT_LA_RET_AUTH_RES_BITPOS) | \ + ((update_state & QAT_LA_UPDATE_STATE_MASK) << \ + QAT_LA_UPDATE_STATE_BITPOS) | \ + ((ciph_iv & QAT_LA_CIPH_IV_FLD_MASK) << \ + QAT_LA_CIPH_IV_FLD_BITPOS) | \ + ((ciphcfg & QAT_LA_CIPH_AUTH_CFG_OFFSET_MASK) << \ + QAT_LA_CIPH_AUTH_CFG_OFFSET_BITPOS) | \ + ((partial & QAT_LA_PARTIAL_MASK) << \ + QAT_LA_PARTIAL_BITPOS)) + +#define ICP_QAT_FW_LA_CIPH_IV_FLD_FLAG_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_CIPH_IV_FLD_BITPOS, \ + QAT_LA_CIPH_IV_FLD_MASK) + +#define ICP_QAT_FW_LA_CIPH_AUTH_CFG_OFFSET_FLAG_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_CIPH_AUTH_CFG_OFFSET_BITPOS, \ + QAT_LA_CIPH_AUTH_CFG_OFFSET_MASK) + +#define ICP_QAT_FW_LA_ZUC_3G_PROTO_FLAG_GET(flags) \ + QAT_FIELD_GET(flags, QAT_FW_LA_ZUC_3G_PROTO_FLAG_BITPOS, \ + QAT_FW_LA_ZUC_3G_PROTO_FLAG_MASK) + +#define ICP_QAT_FW_LA_GCM_IV_LEN_FLAG_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_GCM_IV_LEN_FLAG_BITPOS, \ + QAT_LA_GCM_IV_LEN_FLAG_MASK) + +#define ICP_QAT_FW_LA_PROTO_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_PROTO_BITPOS, QAT_LA_PROTO_MASK) + +#define ICP_QAT_FW_LA_CMP_AUTH_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_CMP_AUTH_RES_BITPOS, \ + QAT_LA_CMP_AUTH_RES_MASK) + +#define ICP_QAT_FW_LA_RET_AUTH_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_RET_AUTH_RES_BITPOS, \ + QAT_LA_RET_AUTH_RES_MASK) + +#define ICP_QAT_FW_LA_DIGEST_IN_BUFFER_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_DIGEST_IN_BUFFER_BITPOS, \ + QAT_LA_DIGEST_IN_BUFFER_MASK) + +#define ICP_QAT_FW_LA_UPDATE_STATE_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_UPDATE_STATE_BITPOS, \ + QAT_LA_UPDATE_STATE_MASK) + +#define ICP_QAT_FW_LA_PARTIAL_GET(flags) \ + QAT_FIELD_GET(flags, QAT_LA_PARTIAL_BITPOS, \ + QAT_LA_PARTIAL_MASK) + +#define ICP_QAT_FW_LA_CIPH_IV_FLD_FLAG_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_CIPH_IV_FLD_BITPOS, \ + QAT_LA_CIPH_IV_FLD_MASK) + +#define ICP_QAT_FW_LA_CIPH_AUTH_CFG_OFFSET_FLAG_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_CIPH_AUTH_CFG_OFFSET_BITPOS, \ + QAT_LA_CIPH_AUTH_CFG_OFFSET_MASK) + +#define ICP_QAT_FW_LA_ZUC_3G_PROTO_FLAG_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_FW_LA_ZUC_3G_PROTO_FLAG_BITPOS, \ + QAT_FW_LA_ZUC_3G_PROTO_FLAG_MASK) + +#define ICP_QAT_FW_LA_GCM_IV_LEN_FLAG_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_GCM_IV_LEN_FLAG_BITPOS, \ + QAT_LA_GCM_IV_LEN_FLAG_MASK) + +#define ICP_QAT_FW_LA_PROTO_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_PROTO_BITPOS, \ + QAT_LA_PROTO_MASK) + +#define ICP_QAT_FW_LA_CMP_AUTH_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_CMP_AUTH_RES_BITPOS, \ + QAT_LA_CMP_AUTH_RES_MASK) + +#define ICP_QAT_FW_LA_RET_AUTH_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_RET_AUTH_RES_BITPOS, \ + QAT_LA_RET_AUTH_RES_MASK) + +#define ICP_QAT_FW_LA_DIGEST_IN_BUFFER_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_DIGEST_IN_BUFFER_BITPOS, \ + QAT_LA_DIGEST_IN_BUFFER_MASK) + +#define ICP_QAT_FW_LA_UPDATE_STATE_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_UPDATE_STATE_BITPOS, \ + QAT_LA_UPDATE_STATE_MASK) + +#define ICP_QAT_FW_LA_PARTIAL_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_PARTIAL_BITPOS, \ + QAT_LA_PARTIAL_MASK) + +#define ICP_QAT_FW_LA_SLICE_TYPE_SET(flags, val) \ + QAT_FIELD_SET(flags, val, QAT_LA_SLICE_TYPE_BITPOS, \ + QAT_LA_SLICE_TYPE_MASK) + +struct icp_qat_fw_cipher_req_hdr_cd_pars { + union { + struct { + __u64 content_desc_addr; + __u16 content_desc_resrvd1; + __u8 content_desc_params_sz; + __u8 content_desc_hdr_resrvd2; + __u32 content_desc_resrvd3; + } s; + struct { + __u32 cipher_key_array[ICP_QAT_FW_NUM_LONGWORDS_4]; + } s1; + } u; +}; + +struct icp_qat_fw_cipher_auth_req_hdr_cd_pars { + union { + struct { + __u64 content_desc_addr; + __u16 content_desc_resrvd1; + __u8 content_desc_params_sz; + __u8 content_desc_hdr_resrvd2; + __u32 content_desc_resrvd3; + } s; + struct { + __u32 cipher_key_array[ICP_QAT_FW_NUM_LONGWORDS_4]; + } sl; + } u; +}; + +struct icp_qat_fw_cipher_cd_ctrl_hdr { + __u8 cipher_state_sz; + __u8 cipher_key_sz; + __u8 cipher_cfg_offset; + __u8 next_curr_id; + __u8 cipher_padding_sz; + __u8 resrvd1; + __u16 resrvd2; + __u32 resrvd3[ICP_QAT_FW_NUM_LONGWORDS_3]; +}; + +struct icp_qat_fw_auth_cd_ctrl_hdr { + __u32 resrvd1; + __u8 resrvd2; + __u8 hash_flags; + __u8 hash_cfg_offset; + __u8 next_curr_id; + __u8 resrvd3; + __u8 outer_prefix_sz; + __u8 final_sz; + __u8 inner_res_sz; + __u8 resrvd4; + __u8 inner_state1_sz; + __u8 inner_state2_offset; + __u8 inner_state2_sz; + __u8 outer_config_offset; + __u8 outer_state1_sz; + __u8 outer_res_sz; + __u8 outer_prefix_offset; +}; + +struct icp_qat_fw_cipher_auth_cd_ctrl_hdr { + __u8 cipher_state_sz; + __u8 cipher_key_sz; + __u8 cipher_cfg_offset; + __u8 next_curr_id_cipher; + __u8 cipher_padding_sz; + __u8 hash_flags; + __u8 hash_cfg_offset; + __u8 next_curr_id_auth; + __u8 resrvd1; + __u8 outer_prefix_sz; + __u8 final_sz; + __u8 inner_res_sz; + __u8 resrvd2; + __u8 inner_state1_sz; + __u8 inner_state2_offset; + __u8 inner_state2_sz; + __u8 outer_config_offset; + __u8 outer_state1_sz; + __u8 outer_res_sz; + __u8 outer_prefix_offset; +}; + +#define ICP_QAT_FW_AUTH_HDR_FLAG_DO_NESTED 1 +#define ICP_QAT_FW_AUTH_HDR_FLAG_NO_NESTED 0 +#define ICP_QAT_FW_CCM_GCM_AAD_SZ_MAX 240 +#define ICP_QAT_FW_HASH_REQUEST_PARAMETERS_OFFSET \ + (sizeof(struct icp_qat_fw_la_cipher_req_params_t)) +#define ICP_QAT_FW_CIPHER_REQUEST_PARAMETERS_OFFSET (0) + +struct icp_qat_fw_la_cipher_req_params { + __u32 cipher_offset; + __u32 cipher_length; + union { + __u32 cipher_IV_array[ICP_QAT_FW_NUM_LONGWORDS_4]; + struct { + __u64 cipher_IV_ptr; + __u64 resrvd1; + } s; + } u; +}; + +struct icp_qat_fw_la_auth_req_params { + __u32 auth_off; + __u32 auth_len; + union { + __u64 auth_partial_st_prefix; + __u64 aad_adr; + } u1; + __u64 auth_res_addr; + union { + __u8 inner_prefix_sz; + __u8 aad_sz; + } u2; + __u8 resrvd1; + __u8 hash_state_sz; + __u8 auth_res_sz; +} __packed; + +struct icp_qat_fw_la_auth_req_params_resrvd_flds { + __u32 resrvd[ICP_QAT_FW_NUM_LONGWORDS_6]; + union { + __u8 inner_prefix_sz; + __u8 aad_sz; + } u2; + __u8 resrvd1; + __u16 resrvd2; +}; + +struct icp_qat_fw_la_resp { + struct icp_qat_fw_comn_resp_hdr comn_resp; + __u64 opaque_data; + __u32 resrvd[ICP_QAT_FW_NUM_LONGWORDS_4]; +}; + +#define ICP_QAT_FW_CIPHER_NEXT_ID_GET(cd_ctrl_hdr_t) \ + ((((cd_ctrl_hdr_t)->next_curr_id_cipher) & \ + ICP_QAT_FW_COMN_NEXT_ID_MASK) >> (ICP_QAT_FW_COMN_NEXT_ID_BITPOS)) + +#define ICP_QAT_FW_CIPHER_NEXT_ID_SET(cd_ctrl_hdr_t, val) \ +{ (cd_ctrl_hdr_t)->next_curr_id_cipher = \ + ((((cd_ctrl_hdr_t)->next_curr_id_cipher) \ + & ICP_QAT_FW_COMN_CURR_ID_MASK) | \ + ((val << ICP_QAT_FW_COMN_NEXT_ID_BITPOS) \ + & ICP_QAT_FW_COMN_NEXT_ID_MASK)) } + +#define ICP_QAT_FW_CIPHER_CURR_ID_GET(cd_ctrl_hdr_t) \ + (((cd_ctrl_hdr_t)->next_curr_id_cipher) \ + & ICP_QAT_FW_COMN_CURR_ID_MASK) + +#define ICP_QAT_FW_CIPHER_CURR_ID_SET(cd_ctrl_hdr_t, val) \ +{ (cd_ctrl_hdr_t)->next_curr_id_cipher = \ + ((((cd_ctrl_hdr_t)->next_curr_id_cipher) \ + & ICP_QAT_FW_COMN_NEXT_ID_MASK) | \ + ((val) & ICP_QAT_FW_COMN_CURR_ID_MASK)) } + +#define ICP_QAT_FW_AUTH_NEXT_ID_GET(cd_ctrl_hdr_t) \ + ((((cd_ctrl_hdr_t)->next_curr_id_auth) & ICP_QAT_FW_COMN_NEXT_ID_MASK) \ + >> (ICP_QAT_FW_COMN_NEXT_ID_BITPOS)) + +#define ICP_QAT_FW_AUTH_NEXT_ID_SET(cd_ctrl_hdr_t, val) \ +{ (cd_ctrl_hdr_t)->next_curr_id_auth = \ + ((((cd_ctrl_hdr_t)->next_curr_id_auth) \ + & ICP_QAT_FW_COMN_CURR_ID_MASK) | \ + ((val << ICP_QAT_FW_COMN_NEXT_ID_BITPOS) \ + & ICP_QAT_FW_COMN_NEXT_ID_MASK)) } + +#define ICP_QAT_FW_AUTH_CURR_ID_GET(cd_ctrl_hdr_t) \ + (((cd_ctrl_hdr_t)->next_curr_id_auth) \ + & ICP_QAT_FW_COMN_CURR_ID_MASK) + +#define ICP_QAT_FW_AUTH_CURR_ID_SET(cd_ctrl_hdr_t, val) \ +{ (cd_ctrl_hdr_t)->next_curr_id_auth = \ + ((((cd_ctrl_hdr_t)->next_curr_id_auth) \ + & ICP_QAT_FW_COMN_NEXT_ID_MASK) | \ + ((val) & ICP_QAT_FW_COMN_CURR_ID_MASK)) } + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_fw_loader_handle.h b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_loader_handle.h new file mode 100644 index 000000000000..6887930c7995 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_loader_handle.h @@ -0,0 +1,69 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef __ICP_QAT_FW_LOADER_HANDLE_H__ +#define __ICP_QAT_FW_LOADER_HANDLE_H__ +#include "icp_qat_uclo.h" + +struct icp_qat_fw_loader_ae_data { + unsigned int state; + unsigned int ustore_size; + unsigned int free_addr; + unsigned int free_size; + unsigned int live_ctx_mask; +}; + +struct icp_qat_fw_loader_hal_handle { + struct icp_qat_fw_loader_ae_data aes[ICP_QAT_UCLO_MAX_AE]; + unsigned int ae_mask; + unsigned int admin_ae_mask; + unsigned int slice_mask; + unsigned int revision_id; + unsigned int ae_max_num; + unsigned int upc_mask; + unsigned int max_ustore; +}; + +struct icp_qat_fw_loader_chip_info { + int mmp_sram_size; + bool nn; + bool lm2lm3; + u32 lm_size; + u32 icp_rst_csr; + u32 icp_rst_mask; + u32 glb_clk_enable_csr; + u32 misc_ctl_csr; + u32 wakeup_event_val; + bool fw_auth; + bool css_3k; + bool dual_sign; + bool tgroup_share_ustore; + u32 fcu_ctl_csr; + u32 fcu_sts_csr; + u32 fcu_dram_addr_hi; + u32 fcu_dram_addr_lo; + u32 fcu_loaded_ae_csr; + u8 fcu_loaded_ae_pos; +}; + +struct icp_qat_fw_loader_handle { + struct icp_qat_fw_loader_hal_handle *hal_handle; + struct icp_qat_fw_loader_chip_info *chip_info; + struct pci_dev *pci_dev; + void *obj_handle; + void *sobj_handle; + void *mobj_handle; + unsigned int cfg_ae_mask; + void __iomem *hal_sram_addr_v; + void __iomem *hal_cap_g_ctl_csr_addr_v; + void __iomem *hal_cap_ae_xfer_csr_addr_v; + void __iomem *hal_cap_ae_local_csr_addr_v; + void __iomem *hal_ep_csr_addr_v; +}; + +struct icp_firml_dram_desc { + void __iomem *dram_base_addr; + void *dram_base_addr_v; + dma_addr_t dram_bus_addr; + u64 dram_size; +}; +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_fw_pke.h b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_pke.h new file mode 100644 index 000000000000..9dddae0009fc --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_fw_pke.h @@ -0,0 +1,68 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef _ICP_QAT_FW_PKE_ +#define _ICP_QAT_FW_PKE_ + +#include "icp_qat_fw.h" + +struct icp_qat_fw_req_hdr_pke_cd_pars { + __u64 content_desc_addr; + __u32 content_desc_resrvd; + __u32 func_id; +}; + +struct icp_qat_fw_req_pke_mid { + __u64 opaque; + __u64 src_data_addr; + __u64 dest_data_addr; +}; + +struct icp_qat_fw_req_pke_hdr { + __u8 resrvd1; + __u8 resrvd2; + __u8 service_type; + __u8 hdr_flags; + __u16 comn_req_flags; + __u16 resrvd4; + struct icp_qat_fw_req_hdr_pke_cd_pars cd_pars; +}; + +struct icp_qat_fw_pke_request { + struct icp_qat_fw_req_pke_hdr pke_hdr; + struct icp_qat_fw_req_pke_mid pke_mid; + __u8 output_param_count; + __u8 input_param_count; + __u16 resrvd1; + __u32 resrvd2; + __u64 next_req_adr; +}; + +struct icp_qat_fw_resp_pke_hdr { + __u8 resrvd1; + __u8 resrvd2; + __u8 response_type; + __u8 hdr_flags; + __u16 comn_resp_flags; + __u16 resrvd4; +}; + +struct icp_qat_fw_pke_resp { + struct icp_qat_fw_resp_pke_hdr pke_resp_hdr; + __u64 opaque; + __u64 src_data_addr; + __u64 dest_data_addr; +}; + +#define ICP_QAT_FW_PKE_HDR_VALID_FLAG_BITPOS 7 +#define ICP_QAT_FW_PKE_HDR_VALID_FLAG_MASK 0x1 +#define ICP_QAT_FW_PKE_RESP_PKE_STAT_GET(status_word) \ + QAT_FIELD_GET(((status_word >> ICP_QAT_FW_COMN_ONE_BYTE_SHIFT) & \ + ICP_QAT_FW_COMN_SINGLE_BYTE_MASK), \ + QAT_COMN_RESP_PKE_STATUS_BITPOS, \ + QAT_COMN_RESP_PKE_STATUS_MASK) + +#define ICP_QAT_FW_PKE_HDR_VALID_FLAG_SET(hdr_t, val) \ + QAT_FIELD_SET((hdr_t.hdr_flags), (val), \ + ICP_QAT_FW_PKE_HDR_VALID_FLAG_BITPOS, \ + ICP_QAT_FW_PKE_HDR_VALID_FLAG_MASK) +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_hal.h b/drivers/crypto/intel/qat/qat_common/icp_qat_hal.h new file mode 100644 index 000000000000..20b2ee1fc65a --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_hal.h @@ -0,0 +1,143 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef __ICP_QAT_HAL_H +#define __ICP_QAT_HAL_H +#include "icp_qat_fw_loader_handle.h" + +enum hal_global_csr { + MISC_CONTROL = 0xA04, + ICP_RESET = 0xA0c, + ICP_GLOBAL_CLK_ENABLE = 0xA50 +}; + +enum { + MISC_CONTROL_C4XXX = 0xAA0, + ICP_RESET_CPP0 = 0x938, + ICP_RESET_CPP1 = 0x93c, + ICP_GLOBAL_CLK_ENABLE_CPP0 = 0x964, + ICP_GLOBAL_CLK_ENABLE_CPP1 = 0x968 +}; + +enum hal_ae_csr { + USTORE_ADDRESS = 0x000, + USTORE_DATA_LOWER = 0x004, + USTORE_DATA_UPPER = 0x008, + ALU_OUT = 0x010, + CTX_ARB_CNTL = 0x014, + CTX_ENABLES = 0x018, + CC_ENABLE = 0x01c, + CSR_CTX_POINTER = 0x020, + CTX_STS_INDIRECT = 0x040, + ACTIVE_CTX_STATUS = 0x044, + CTX_SIG_EVENTS_INDIRECT = 0x048, + CTX_SIG_EVENTS_ACTIVE = 0x04c, + CTX_WAKEUP_EVENTS_INDIRECT = 0x050, + LM_ADDR_0_INDIRECT = 0x060, + LM_ADDR_1_INDIRECT = 0x068, + LM_ADDR_2_INDIRECT = 0x0cc, + LM_ADDR_3_INDIRECT = 0x0d4, + INDIRECT_LM_ADDR_0_BYTE_INDEX = 0x0e0, + INDIRECT_LM_ADDR_1_BYTE_INDEX = 0x0e8, + INDIRECT_LM_ADDR_2_BYTE_INDEX = 0x10c, + INDIRECT_LM_ADDR_3_BYTE_INDEX = 0x114, + INDIRECT_T_INDEX = 0x0f8, + INDIRECT_T_INDEX_BYTE_INDEX = 0x0fc, + FUTURE_COUNT_SIGNAL_INDIRECT = 0x078, + TIMESTAMP_LOW = 0x0c0, + TIMESTAMP_HIGH = 0x0c4, + PROFILE_COUNT = 0x144, + SIGNATURE_ENABLE = 0x150, + AE_MISC_CONTROL = 0x160, + LOCAL_CSR_STATUS = 0x180, +}; + +enum fcu_csr { + FCU_CONTROL = 0x8c0, + FCU_STATUS = 0x8c4, + FCU_STATUS1 = 0x8c8, + FCU_DRAM_ADDR_LO = 0x8cc, + FCU_DRAM_ADDR_HI = 0x8d0, + FCU_RAMBASE_ADDR_HI = 0x8d4, + FCU_RAMBASE_ADDR_LO = 0x8d8 +}; + +enum fcu_csr_4xxx { + FCU_CONTROL_4XXX = 0x1000, + FCU_STATUS_4XXX = 0x1004, + FCU_ME_BROADCAST_MASK_TYPE = 0x1008, + FCU_AE_LOADED_4XXX = 0x1010, + FCU_DRAM_ADDR_LO_4XXX = 0x1014, + FCU_DRAM_ADDR_HI_4XXX = 0x1018, +}; + +enum fcu_cmd { + FCU_CTRL_CMD_NOOP = 0, + FCU_CTRL_CMD_AUTH = 1, + FCU_CTRL_CMD_LOAD = 2, + FCU_CTRL_CMD_START = 3 +}; + +enum fcu_sts { + FCU_STS_NO_STS = 0, + FCU_STS_VERI_DONE = 1, + FCU_STS_LOAD_DONE = 2, + FCU_STS_VERI_FAIL = 3, + FCU_STS_LOAD_FAIL = 4, + FCU_STS_BUSY = 5 +}; + +#define ALL_AE_MASK 0xFFFFFFFF +#define UA_ECS (0x1 << 31) +#define ACS_ABO_BITPOS 31 +#define ACS_ACNO 0x7 +#define CE_ENABLE_BITPOS 0x8 +#define CE_LMADDR_0_GLOBAL_BITPOS 16 +#define CE_LMADDR_1_GLOBAL_BITPOS 17 +#define CE_LMADDR_2_GLOBAL_BITPOS 22 +#define CE_LMADDR_3_GLOBAL_BITPOS 23 +#define CE_T_INDEX_GLOBAL_BITPOS 21 +#define CE_NN_MODE_BITPOS 20 +#define CE_REG_PAR_ERR_BITPOS 25 +#define CE_BREAKPOINT_BITPOS 27 +#define CE_CNTL_STORE_PARITY_ERROR_BITPOS 29 +#define CE_INUSE_CONTEXTS_BITPOS 31 +#define CE_NN_MODE (0x1 << CE_NN_MODE_BITPOS) +#define CE_INUSE_CONTEXTS (0x1 << CE_INUSE_CONTEXTS_BITPOS) +#define XCWE_VOLUNTARY (0x1) +#define LCS_STATUS (0x1) +#define MMC_SHARE_CS_BITPOS 2 +#define WAKEUP_EVENT 0x10000 +#define FCU_CTRL_BROADCAST_POS 0x4 +#define FCU_CTRL_AE_POS 0x8 +#define FCU_AUTH_STS_MASK 0x7 +#define FCU_STS_DONE_POS 0x9 +#define FCU_STS_AUTHFWLD_POS 0X8 +#define FCU_LOADED_AE_POS 0x16 +#define FW_AUTH_WAIT_PERIOD 10 +#define FW_AUTH_MAX_RETRY 300 +#define ICP_QAT_AE_OFFSET 0x20000 +#define ICP_QAT_CAP_OFFSET (ICP_QAT_AE_OFFSET + 0x10000) +#define LOCAL_TO_XFER_REG_OFFSET 0x800 +#define ICP_QAT_EP_OFFSET 0x3a000 +#define ICP_QAT_EP_OFFSET_4XXX 0x200000 /* HI MMIO CSRs */ +#define ICP_QAT_AE_OFFSET_4XXX 0x600000 +#define ICP_QAT_CAP_OFFSET_4XXX 0x640000 +#define SET_CAP_CSR(handle, csr, val) \ + ADF_CSR_WR((handle)->hal_cap_g_ctl_csr_addr_v, csr, val) +#define GET_CAP_CSR(handle, csr) \ + ADF_CSR_RD((handle)->hal_cap_g_ctl_csr_addr_v, csr) +#define AE_CSR(handle, ae) \ + ((char __iomem *)(handle)->hal_cap_ae_local_csr_addr_v + ((ae) << 12)) +#define AE_CSR_ADDR(handle, ae, csr) (AE_CSR(handle, ae) + (0x3ff & (csr))) +#define SET_AE_CSR(handle, ae, csr, val) \ + ADF_CSR_WR(AE_CSR_ADDR(handle, ae, csr), 0, val) +#define GET_AE_CSR(handle, ae, csr) ADF_CSR_RD(AE_CSR_ADDR(handle, ae, csr), 0) +#define AE_XFER(handle, ae) \ + ((char __iomem *)(handle)->hal_cap_ae_xfer_csr_addr_v + ((ae) << 12)) +#define AE_XFER_ADDR(handle, ae, reg) (AE_XFER(handle, ae) + \ + (((reg) & 0xff) << 2)) +#define SET_AE_XFER(handle, ae, reg, val) \ + ADF_CSR_WR(AE_XFER_ADDR(handle, ae, reg), 0, val) +#define SRAM_WRITE(handle, addr, val) \ + ADF_CSR_WR((handle)->hal_sram_addr_v, addr, val) +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_hw.h b/drivers/crypto/intel/qat/qat_common/icp_qat_hw.h new file mode 100644 index 000000000000..b8f1c4ffb8b5 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_hw.h @@ -0,0 +1,388 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef _ICP_QAT_HW_H_ +#define _ICP_QAT_HW_H_ + +#include <linux/bits.h> + +enum icp_qat_hw_ae_id { + ICP_QAT_HW_AE_0 = 0, + ICP_QAT_HW_AE_1 = 1, + ICP_QAT_HW_AE_2 = 2, + ICP_QAT_HW_AE_3 = 3, + ICP_QAT_HW_AE_4 = 4, + ICP_QAT_HW_AE_5 = 5, + ICP_QAT_HW_AE_6 = 6, + ICP_QAT_HW_AE_7 = 7, + ICP_QAT_HW_AE_8 = 8, + ICP_QAT_HW_AE_9 = 9, + ICP_QAT_HW_AE_10 = 10, + ICP_QAT_HW_AE_11 = 11, + ICP_QAT_HW_AE_12 = 12, + ICP_QAT_HW_AE_13 = 13, + ICP_QAT_HW_AE_14 = 14, + ICP_QAT_HW_AE_15 = 15, + ICP_QAT_HW_AE_16 = 16, + ICP_QAT_HW_AE_DELIMITER = 17 +}; + +enum icp_qat_hw_qat_id { + ICP_QAT_HW_QAT_0 = 0, + ICP_QAT_HW_QAT_1 = 1, + ICP_QAT_HW_QAT_2 = 2, + ICP_QAT_HW_QAT_3 = 3, + ICP_QAT_HW_QAT_4 = 4, + ICP_QAT_HW_QAT_5 = 5, + ICP_QAT_HW_QAT_DELIMITER = 6 +}; + +enum icp_qat_hw_auth_algo { + ICP_QAT_HW_AUTH_ALGO_NULL = 0, + ICP_QAT_HW_AUTH_ALGO_SHA1 = 1, + ICP_QAT_HW_AUTH_ALGO_MD5 = 2, + ICP_QAT_HW_AUTH_ALGO_SHA224 = 3, + ICP_QAT_HW_AUTH_ALGO_SHA256 = 4, + ICP_QAT_HW_AUTH_ALGO_SHA384 = 5, + ICP_QAT_HW_AUTH_ALGO_SHA512 = 6, + ICP_QAT_HW_AUTH_ALGO_AES_XCBC_MAC = 7, + ICP_QAT_HW_AUTH_ALGO_AES_CBC_MAC = 8, + ICP_QAT_HW_AUTH_ALGO_AES_F9 = 9, + ICP_QAT_HW_AUTH_ALGO_GALOIS_128 = 10, + ICP_QAT_HW_AUTH_ALGO_GALOIS_64 = 11, + ICP_QAT_HW_AUTH_ALGO_KASUMI_F9 = 12, + ICP_QAT_HW_AUTH_ALGO_SNOW_3G_UIA2 = 13, + ICP_QAT_HW_AUTH_ALGO_ZUC_3G_128_EIA3 = 14, + ICP_QAT_HW_AUTH_RESERVED_1 = 15, + ICP_QAT_HW_AUTH_RESERVED_2 = 16, + ICP_QAT_HW_AUTH_ALGO_SHA3_256 = 17, + ICP_QAT_HW_AUTH_RESERVED_3 = 18, + ICP_QAT_HW_AUTH_ALGO_SHA3_512 = 19, + ICP_QAT_HW_AUTH_ALGO_DELIMITER = 20 +}; + +enum icp_qat_hw_auth_mode { + ICP_QAT_HW_AUTH_MODE0 = 0, + ICP_QAT_HW_AUTH_MODE1 = 1, + ICP_QAT_HW_AUTH_MODE2 = 2, + ICP_QAT_HW_AUTH_MODE_DELIMITER = 3 +}; + +struct icp_qat_hw_auth_config { + __u32 config; + __u32 reserved; +}; + +struct icp_qat_hw_ucs_cipher_config { + __u32 val; + __u32 reserved[3]; +}; + +enum icp_qat_slice_mask { + ICP_ACCEL_MASK_CIPHER_SLICE = BIT(0), + ICP_ACCEL_MASK_AUTH_SLICE = BIT(1), + ICP_ACCEL_MASK_PKE_SLICE = BIT(2), + ICP_ACCEL_MASK_COMPRESS_SLICE = BIT(3), + ICP_ACCEL_MASK_LZS_SLICE = BIT(4), + ICP_ACCEL_MASK_EIA3_SLICE = BIT(5), + ICP_ACCEL_MASK_SHA3_SLICE = BIT(6), +}; + +enum icp_qat_capabilities_mask { + ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC = BIT(0), + ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC = BIT(1), + ICP_ACCEL_CAPABILITIES_CIPHER = BIT(2), + ICP_ACCEL_CAPABILITIES_AUTHENTICATION = BIT(3), + ICP_ACCEL_CAPABILITIES_RESERVED_1 = BIT(4), + ICP_ACCEL_CAPABILITIES_COMPRESSION = BIT(5), + /* Bits 6-7 are currently reserved */ + ICP_ACCEL_CAPABILITIES_ZUC = BIT(8), + ICP_ACCEL_CAPABILITIES_SHA3 = BIT(9), + /* Bits 10-11 are currently reserved */ + ICP_ACCEL_CAPABILITIES_HKDF = BIT(12), + ICP_ACCEL_CAPABILITIES_ECEDMONT = BIT(13), + ICP_ACCEL_CAPABILITIES_EXT_ALGCHAIN = BIT(14), + ICP_ACCEL_CAPABILITIES_SHA3_EXT = BIT(15), + ICP_ACCEL_CAPABILITIES_AESGCM_SPC = BIT(16), + ICP_ACCEL_CAPABILITIES_CHACHA_POLY = BIT(17), + ICP_ACCEL_CAPABILITIES_SM2 = BIT(18), + ICP_ACCEL_CAPABILITIES_SM3 = BIT(19), + ICP_ACCEL_CAPABILITIES_SM4 = BIT(20), + /* Bit 21 is currently reserved */ + ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY = BIT(22), + ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64 = BIT(23), + ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION = BIT(24), + ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION = BIT(25), + ICP_ACCEL_CAPABILITIES_AES_V2 = BIT(26), + /* Bits 27-28 are currently reserved */ + ICP_ACCEL_CAPABILITIES_ZUC_256 = BIT(29), + ICP_ACCEL_CAPABILITIES_WIRELESS_CRYPTO_EXT = BIT(30), +}; + +#define QAT_AUTH_MODE_BITPOS 4 +#define QAT_AUTH_MODE_MASK 0xF +#define QAT_AUTH_ALGO_BITPOS 0 +#define QAT_AUTH_ALGO_MASK 0xF +#define QAT_AUTH_CMP_BITPOS 8 +#define QAT_AUTH_CMP_MASK 0x7F +#define QAT_AUTH_SHA3_PADDING_BITPOS 16 +#define QAT_AUTH_SHA3_PADDING_MASK 0x1 +#define QAT_AUTH_ALGO_SHA3_BITPOS 22 +#define QAT_AUTH_ALGO_SHA3_MASK 0x3 +#define ICP_QAT_HW_AUTH_CONFIG_BUILD(mode, algo, cmp_len) \ + (((mode & QAT_AUTH_MODE_MASK) << QAT_AUTH_MODE_BITPOS) | \ + ((algo & QAT_AUTH_ALGO_MASK) << QAT_AUTH_ALGO_BITPOS) | \ + (((algo >> 4) & QAT_AUTH_ALGO_SHA3_MASK) << \ + QAT_AUTH_ALGO_SHA3_BITPOS) | \ + (((((algo == ICP_QAT_HW_AUTH_ALGO_SHA3_256) || \ + (algo == ICP_QAT_HW_AUTH_ALGO_SHA3_512)) ? 1 : 0) \ + & QAT_AUTH_SHA3_PADDING_MASK) << QAT_AUTH_SHA3_PADDING_BITPOS) | \ + ((cmp_len & QAT_AUTH_CMP_MASK) << QAT_AUTH_CMP_BITPOS)) + +struct icp_qat_hw_auth_counter { + __be32 counter; + __u32 reserved; +}; + +#define QAT_AUTH_COUNT_MASK 0xFFFFFFFF +#define QAT_AUTH_COUNT_BITPOS 0 +#define ICP_QAT_HW_AUTH_COUNT_BUILD(val) \ + (((val) & QAT_AUTH_COUNT_MASK) << QAT_AUTH_COUNT_BITPOS) + +struct icp_qat_hw_auth_setup { + struct icp_qat_hw_auth_config auth_config; + struct icp_qat_hw_auth_counter auth_counter; +}; + +#define QAT_HW_DEFAULT_ALIGNMENT 8 +#define QAT_HW_ROUND_UP(val, n) (((val) + ((n) - 1)) & (~(n - 1))) +#define ICP_QAT_HW_NULL_STATE1_SZ 32 +#define ICP_QAT_HW_MD5_STATE1_SZ 16 +#define ICP_QAT_HW_SHA1_STATE1_SZ 20 +#define ICP_QAT_HW_SHA224_STATE1_SZ 32 +#define ICP_QAT_HW_SHA256_STATE1_SZ 32 +#define ICP_QAT_HW_SHA3_256_STATE1_SZ 32 +#define ICP_QAT_HW_SHA384_STATE1_SZ 64 +#define ICP_QAT_HW_SHA512_STATE1_SZ 64 +#define ICP_QAT_HW_SHA3_512_STATE1_SZ 64 +#define ICP_QAT_HW_SHA3_224_STATE1_SZ 28 +#define ICP_QAT_HW_SHA3_384_STATE1_SZ 48 +#define ICP_QAT_HW_AES_XCBC_MAC_STATE1_SZ 16 +#define ICP_QAT_HW_AES_CBC_MAC_STATE1_SZ 16 +#define ICP_QAT_HW_AES_F9_STATE1_SZ 32 +#define ICP_QAT_HW_KASUMI_F9_STATE1_SZ 16 +#define ICP_QAT_HW_GALOIS_128_STATE1_SZ 16 +#define ICP_QAT_HW_SNOW_3G_UIA2_STATE1_SZ 8 +#define ICP_QAT_HW_ZUC_3G_EIA3_STATE1_SZ 8 +#define ICP_QAT_HW_NULL_STATE2_SZ 32 +#define ICP_QAT_HW_MD5_STATE2_SZ 16 +#define ICP_QAT_HW_SHA1_STATE2_SZ 20 +#define ICP_QAT_HW_SHA224_STATE2_SZ 32 +#define ICP_QAT_HW_SHA256_STATE2_SZ 32 +#define ICP_QAT_HW_SHA3_256_STATE2_SZ 0 +#define ICP_QAT_HW_SHA384_STATE2_SZ 64 +#define ICP_QAT_HW_SHA512_STATE2_SZ 64 +#define ICP_QAT_HW_SHA3_512_STATE2_SZ 0 +#define ICP_QAT_HW_SHA3_224_STATE2_SZ 0 +#define ICP_QAT_HW_SHA3_384_STATE2_SZ 0 +#define ICP_QAT_HW_AES_XCBC_MAC_KEY_SZ 16 +#define ICP_QAT_HW_AES_CBC_MAC_KEY_SZ 16 +#define ICP_QAT_HW_AES_CCM_CBC_E_CTR0_SZ 16 +#define ICP_QAT_HW_F9_IK_SZ 16 +#define ICP_QAT_HW_F9_FK_SZ 16 +#define ICP_QAT_HW_KASUMI_F9_STATE2_SZ (ICP_QAT_HW_F9_IK_SZ + \ + ICP_QAT_HW_F9_FK_SZ) +#define ICP_QAT_HW_AES_F9_STATE2_SZ ICP_QAT_HW_KASUMI_F9_STATE2_SZ +#define ICP_QAT_HW_SNOW_3G_UIA2_STATE2_SZ 24 +#define ICP_QAT_HW_ZUC_3G_EIA3_STATE2_SZ 32 +#define ICP_QAT_HW_GALOIS_H_SZ 16 +#define ICP_QAT_HW_GALOIS_LEN_A_SZ 8 +#define ICP_QAT_HW_GALOIS_E_CTR0_SZ 16 + +struct icp_qat_hw_auth_sha512 { + struct icp_qat_hw_auth_setup inner_setup; + __u8 state1[ICP_QAT_HW_SHA512_STATE1_SZ]; + struct icp_qat_hw_auth_setup outer_setup; + __u8 state2[ICP_QAT_HW_SHA512_STATE2_SZ]; +}; + +struct icp_qat_hw_auth_algo_blk { + struct icp_qat_hw_auth_sha512 sha; +}; + +#define ICP_QAT_HW_GALOIS_LEN_A_BITPOS 0 +#define ICP_QAT_HW_GALOIS_LEN_A_MASK 0xFFFFFFFF + +enum icp_qat_hw_cipher_algo { + ICP_QAT_HW_CIPHER_ALGO_NULL = 0, + ICP_QAT_HW_CIPHER_ALGO_DES = 1, + ICP_QAT_HW_CIPHER_ALGO_3DES = 2, + ICP_QAT_HW_CIPHER_ALGO_AES128 = 3, + ICP_QAT_HW_CIPHER_ALGO_AES192 = 4, + ICP_QAT_HW_CIPHER_ALGO_AES256 = 5, + ICP_QAT_HW_CIPHER_ALGO_ARC4 = 6, + ICP_QAT_HW_CIPHER_ALGO_KASUMI = 7, + ICP_QAT_HW_CIPHER_ALGO_SNOW_3G_UEA2 = 8, + ICP_QAT_HW_CIPHER_ALGO_ZUC_3G_128_EEA3 = 9, + ICP_QAT_HW_CIPHER_DELIMITER = 10 +}; + +enum icp_qat_hw_cipher_mode { + ICP_QAT_HW_CIPHER_ECB_MODE = 0, + ICP_QAT_HW_CIPHER_CBC_MODE = 1, + ICP_QAT_HW_CIPHER_CTR_MODE = 2, + ICP_QAT_HW_CIPHER_F8_MODE = 3, + ICP_QAT_HW_CIPHER_XTS_MODE = 6, + ICP_QAT_HW_CIPHER_MODE_DELIMITER = 7 +}; + +struct icp_qat_hw_cipher_config { + __u32 val; + __u32 reserved; +}; + +enum icp_qat_hw_cipher_dir { + ICP_QAT_HW_CIPHER_ENCRYPT = 0, + ICP_QAT_HW_CIPHER_DECRYPT = 1, +}; + +enum icp_qat_hw_cipher_convert { + ICP_QAT_HW_CIPHER_NO_CONVERT = 0, + ICP_QAT_HW_CIPHER_KEY_CONVERT = 1, +}; + +#define QAT_CIPHER_MODE_BITPOS 4 +#define QAT_CIPHER_MODE_MASK 0xF +#define QAT_CIPHER_ALGO_BITPOS 0 +#define QAT_CIPHER_ALGO_MASK 0xF +#define QAT_CIPHER_CONVERT_BITPOS 9 +#define QAT_CIPHER_CONVERT_MASK 0x1 +#define QAT_CIPHER_DIR_BITPOS 8 +#define QAT_CIPHER_DIR_MASK 0x1 +#define QAT_CIPHER_MODE_F8_KEY_SZ_MULT 2 +#define QAT_CIPHER_MODE_XTS_KEY_SZ_MULT 2 +#define ICP_QAT_HW_CIPHER_CONFIG_BUILD(mode, algo, convert, dir) \ + (((mode & QAT_CIPHER_MODE_MASK) << QAT_CIPHER_MODE_BITPOS) | \ + ((algo & QAT_CIPHER_ALGO_MASK) << QAT_CIPHER_ALGO_BITPOS) | \ + ((convert & QAT_CIPHER_CONVERT_MASK) << QAT_CIPHER_CONVERT_BITPOS) | \ + ((dir & QAT_CIPHER_DIR_MASK) << QAT_CIPHER_DIR_BITPOS)) +#define ICP_QAT_HW_DES_BLK_SZ 8 +#define ICP_QAT_HW_3DES_BLK_SZ 8 +#define ICP_QAT_HW_NULL_BLK_SZ 8 +#define ICP_QAT_HW_AES_BLK_SZ 16 +#define ICP_QAT_HW_KASUMI_BLK_SZ 8 +#define ICP_QAT_HW_SNOW_3G_BLK_SZ 8 +#define ICP_QAT_HW_ZUC_3G_BLK_SZ 8 +#define ICP_QAT_HW_NULL_KEY_SZ 256 +#define ICP_QAT_HW_DES_KEY_SZ 8 +#define ICP_QAT_HW_3DES_KEY_SZ 24 +#define ICP_QAT_HW_AES_128_KEY_SZ 16 +#define ICP_QAT_HW_AES_192_KEY_SZ 24 +#define ICP_QAT_HW_AES_256_KEY_SZ 32 +#define ICP_QAT_HW_AES_128_F8_KEY_SZ (ICP_QAT_HW_AES_128_KEY_SZ * \ + QAT_CIPHER_MODE_F8_KEY_SZ_MULT) +#define ICP_QAT_HW_AES_192_F8_KEY_SZ (ICP_QAT_HW_AES_192_KEY_SZ * \ + QAT_CIPHER_MODE_F8_KEY_SZ_MULT) +#define ICP_QAT_HW_AES_256_F8_KEY_SZ (ICP_QAT_HW_AES_256_KEY_SZ * \ + QAT_CIPHER_MODE_F8_KEY_SZ_MULT) +#define ICP_QAT_HW_AES_128_XTS_KEY_SZ (ICP_QAT_HW_AES_128_KEY_SZ * \ + QAT_CIPHER_MODE_XTS_KEY_SZ_MULT) +#define ICP_QAT_HW_AES_256_XTS_KEY_SZ (ICP_QAT_HW_AES_256_KEY_SZ * \ + QAT_CIPHER_MODE_XTS_KEY_SZ_MULT) +#define ICP_QAT_HW_KASUMI_KEY_SZ 16 +#define ICP_QAT_HW_KASUMI_F8_KEY_SZ (ICP_QAT_HW_KASUMI_KEY_SZ * \ + QAT_CIPHER_MODE_F8_KEY_SZ_MULT) +#define ICP_QAT_HW_AES_128_XTS_KEY_SZ (ICP_QAT_HW_AES_128_KEY_SZ * \ + QAT_CIPHER_MODE_XTS_KEY_SZ_MULT) +#define ICP_QAT_HW_AES_256_XTS_KEY_SZ (ICP_QAT_HW_AES_256_KEY_SZ * \ + QAT_CIPHER_MODE_XTS_KEY_SZ_MULT) +#define ICP_QAT_HW_ARC4_KEY_SZ 256 +#define ICP_QAT_HW_SNOW_3G_UEA2_KEY_SZ 16 +#define ICP_QAT_HW_SNOW_3G_UEA2_IV_SZ 16 +#define ICP_QAT_HW_ZUC_3G_EEA3_KEY_SZ 16 +#define ICP_QAT_HW_ZUC_3G_EEA3_IV_SZ 16 +#define ICP_QAT_HW_MODE_F8_NUM_REG_TO_CLEAR 2 +#define INIT_SHRAM_CONSTANTS_TABLE_SZ 1024 + +struct icp_qat_hw_cipher_aes256_f8 { + struct icp_qat_hw_cipher_config cipher_config; + __u8 key[ICP_QAT_HW_AES_256_F8_KEY_SZ]; +}; + +struct icp_qat_hw_ucs_cipher_aes256_f8 { + struct icp_qat_hw_ucs_cipher_config cipher_config; + __u8 key[ICP_QAT_HW_AES_256_F8_KEY_SZ]; +}; + +struct icp_qat_hw_cipher_algo_blk { + union { + struct icp_qat_hw_cipher_aes256_f8 aes; + struct icp_qat_hw_ucs_cipher_aes256_f8 ucs_aes; + }; +} __aligned(64); + +enum icp_qat_hw_compression_direction { + ICP_QAT_HW_COMPRESSION_DIR_COMPRESS = 0, + ICP_QAT_HW_COMPRESSION_DIR_DECOMPRESS = 1, + ICP_QAT_HW_COMPRESSION_DIR_DELIMITER = 2 +}; + +enum icp_qat_hw_compression_delayed_match { + ICP_QAT_HW_COMPRESSION_DELAYED_MATCH_DISABLED = 0, + ICP_QAT_HW_COMPRESSION_DELAYED_MATCH_ENABLED = 1, + ICP_QAT_HW_COMPRESSION_DELAYED_MATCH_DELIMITER = 2 +}; + +enum icp_qat_hw_compression_algo { + ICP_QAT_HW_COMPRESSION_ALGO_DEFLATE = 0, + ICP_QAT_HW_COMPRESSION_ALGO_LZS = 1, + ICP_QAT_HW_COMPRESSION_ALGO_DELIMITER = 2 +}; + +enum icp_qat_hw_compression_depth { + ICP_QAT_HW_COMPRESSION_DEPTH_1 = 0, + ICP_QAT_HW_COMPRESSION_DEPTH_4 = 1, + ICP_QAT_HW_COMPRESSION_DEPTH_8 = 2, + ICP_QAT_HW_COMPRESSION_DEPTH_16 = 3, + ICP_QAT_HW_COMPRESSION_DEPTH_128 = 4, + ICP_QAT_HW_COMPRESSION_DEPTH_DELIMITER = 5 +}; + +enum icp_qat_hw_compression_file_type { + ICP_QAT_HW_COMPRESSION_FILE_TYPE_0 = 0, + ICP_QAT_HW_COMPRESSION_FILE_TYPE_1 = 1, + ICP_QAT_HW_COMPRESSION_FILE_TYPE_2 = 2, + ICP_QAT_HW_COMPRESSION_FILE_TYPE_3 = 3, + ICP_QAT_HW_COMPRESSION_FILE_TYPE_4 = 4, + ICP_QAT_HW_COMPRESSION_FILE_TYPE_DELIMITER = 5 +}; + +struct icp_qat_hw_compression_config { + __u32 lower_val; + __u32 upper_val; +}; + +#define QAT_COMPRESSION_DIR_BITPOS 4 +#define QAT_COMPRESSION_DIR_MASK 0x7 +#define QAT_COMPRESSION_DELAYED_MATCH_BITPOS 16 +#define QAT_COMPRESSION_DELAYED_MATCH_MASK 0x1 +#define QAT_COMPRESSION_ALGO_BITPOS 31 +#define QAT_COMPRESSION_ALGO_MASK 0x1 +#define QAT_COMPRESSION_DEPTH_BITPOS 28 +#define QAT_COMPRESSION_DEPTH_MASK 0x7 +#define QAT_COMPRESSION_FILE_TYPE_BITPOS 24 +#define QAT_COMPRESSION_FILE_TYPE_MASK 0xF + +#define ICP_QAT_HW_COMPRESSION_CONFIG_BUILD(dir, delayed, \ + algo, depth, filetype) \ + ((((dir) & QAT_COMPRESSION_DIR_MASK) << \ + QAT_COMPRESSION_DIR_BITPOS) | \ + (((delayed) & QAT_COMPRESSION_DELAYED_MATCH_MASK) << \ + QAT_COMPRESSION_DELAYED_MATCH_BITPOS) | \ + (((algo) & QAT_COMPRESSION_ALGO_MASK) << \ + QAT_COMPRESSION_ALGO_BITPOS) | \ + (((depth) & QAT_COMPRESSION_DEPTH_MASK) << \ + QAT_COMPRESSION_DEPTH_BITPOS) | \ + (((filetype) & QAT_COMPRESSION_FILE_TYPE_MASK) << \ + QAT_COMPRESSION_FILE_TYPE_BITPOS)) + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_hw_20_comp.h b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_20_comp.h new file mode 100644 index 000000000000..7ea8962272f2 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_20_comp.h @@ -0,0 +1,164 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef _ICP_QAT_HW_20_COMP_H_ +#define _ICP_QAT_HW_20_COMP_H_ + +#include "icp_qat_hw_20_comp_defs.h" +#include "icp_qat_fw.h" + +struct icp_qat_hw_comp_20_config_csr_lower { + enum icp_qat_hw_comp_20_extended_delay_match_mode edmm; + enum icp_qat_hw_comp_20_hw_comp_format algo; + enum icp_qat_hw_comp_20_search_depth sd; + enum icp_qat_hw_comp_20_hbs_control hbs; + enum icp_qat_hw_comp_20_abd abd; + enum icp_qat_hw_comp_20_lllbd_ctrl lllbd; + enum icp_qat_hw_comp_20_min_match_control mmctrl; + enum icp_qat_hw_comp_20_skip_hash_collision hash_col; + enum icp_qat_hw_comp_20_skip_hash_update hash_update; + enum icp_qat_hw_comp_20_byte_skip skip_ctrl; +}; + +static inline __u32 +ICP_QAT_FW_COMP_20_BUILD_CONFIG_LOWER(struct icp_qat_hw_comp_20_config_csr_lower csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.algo, + ICP_QAT_HW_COMP_20_CONFIG_CSR_HW_COMP_FORMAT_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_HW_COMP_FORMAT_MASK); + QAT_FIELD_SET(val32, csr.sd, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SEARCH_DEPTH_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SEARCH_DEPTH_MASK); + QAT_FIELD_SET(val32, csr.edmm, + ICP_QAT_HW_COMP_20_CONFIG_CSR_EXTENDED_DELAY_MATCH_MODE_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_EXTENDED_DELAY_MATCH_MODE_MASK); + QAT_FIELD_SET(val32, csr.hbs, + ICP_QAT_HW_COMP_20_CONFIG_CSR_HBS_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_HBS_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.lllbd, + ICP_QAT_HW_COMP_20_CONFIG_CSR_LLLBD_CTRL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_LLLBD_CTRL_MASK); + QAT_FIELD_SET(val32, csr.mmctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.hash_col, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_COLLISION_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_COLLISION_MASK); + QAT_FIELD_SET(val32, csr.hash_update, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_UPDATE_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_UPDATE_MASK); + QAT_FIELD_SET(val32, csr.skip_ctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_BYTE_SKIP_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_BYTE_SKIP_MASK); + QAT_FIELD_SET(val32, csr.abd, ICP_QAT_HW_COMP_20_CONFIG_CSR_ABD_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_ABD_MASK); + + return __builtin_bswap32(val32); +} + +struct icp_qat_hw_comp_20_config_csr_upper { + enum icp_qat_hw_comp_20_scb_control scb_ctrl; + enum icp_qat_hw_comp_20_rmb_control rmb_ctrl; + enum icp_qat_hw_comp_20_som_control som_ctrl; + enum icp_qat_hw_comp_20_skip_hash_rd_control skip_hash_ctrl; + enum icp_qat_hw_comp_20_scb_unload_control scb_unload_ctrl; + enum icp_qat_hw_comp_20_disable_token_fusion_control disable_token_fusion_ctrl; + enum icp_qat_hw_comp_20_lbms lbms; + enum icp_qat_hw_comp_20_scb_mode_reset_mask scb_mode_reset; + __u16 lazy; + __u16 nice; +}; + +static inline __u32 +ICP_QAT_FW_COMP_20_BUILD_CONFIG_UPPER(struct icp_qat_hw_comp_20_config_csr_upper csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.scb_ctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.rmb_ctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_RMB_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_RMB_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.som_ctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SOM_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SOM_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.skip_hash_ctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_RD_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_RD_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.scb_unload_ctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_UNLOAD_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_UNLOAD_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.disable_token_fusion_ctrl, + ICP_QAT_HW_COMP_20_CONFIG_CSR_DISABLE_TOKEN_FUSION_CONTROL_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_DISABLE_TOKEN_FUSION_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.lbms, + ICP_QAT_HW_COMP_20_CONFIG_CSR_LBMS_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_LBMS_MASK); + QAT_FIELD_SET(val32, csr.scb_mode_reset, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_MODE_RESET_MASK_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_MODE_RESET_MASK_MASK); + QAT_FIELD_SET(val32, csr.lazy, + ICP_QAT_HW_COMP_20_CONFIG_CSR_LAZY_PARAM_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_LAZY_PARAM_MASK); + QAT_FIELD_SET(val32, csr.nice, + ICP_QAT_HW_COMP_20_CONFIG_CSR_NICE_PARAM_BITPOS, + ICP_QAT_HW_COMP_20_CONFIG_CSR_NICE_PARAM_MASK); + + return __builtin_bswap32(val32); +} + +struct icp_qat_hw_decomp_20_config_csr_lower { + enum icp_qat_hw_decomp_20_hbs_control hbs; + enum icp_qat_hw_decomp_20_lbms lbms; + enum icp_qat_hw_decomp_20_hw_comp_format algo; + enum icp_qat_hw_decomp_20_min_match_control mmctrl; + enum icp_qat_hw_decomp_20_lz4_block_checksum_present lbc; +}; + +static inline __u32 +ICP_QAT_FW_DECOMP_20_BUILD_CONFIG_LOWER(struct icp_qat_hw_decomp_20_config_csr_lower csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.hbs, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HBS_CONTROL_BITPOS, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HBS_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.lbms, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LBMS_BITPOS, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LBMS_MASK); + QAT_FIELD_SET(val32, csr.algo, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HW_DECOMP_FORMAT_BITPOS, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HW_DECOMP_FORMAT_MASK); + QAT_FIELD_SET(val32, csr.mmctrl, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_BITPOS, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.lbc, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_PRESENT_BITPOS, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_PRESENT_MASK); + + return __builtin_bswap32(val32); +} + +struct icp_qat_hw_decomp_20_config_csr_upper { + enum icp_qat_hw_decomp_20_speculative_decoder_control sdc; + enum icp_qat_hw_decomp_20_mini_cam_control mcc; +}; + +static inline __u32 +ICP_QAT_FW_DECOMP_20_BUILD_CONFIG_UPPER(struct icp_qat_hw_decomp_20_config_csr_upper csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.sdc, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_SPECULATIVE_DECODER_CONTROL_BITPOS, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_SPECULATIVE_DECODER_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.mcc, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MINI_CAM_CONTROL_BITPOS, + ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MINI_CAM_CONTROL_MASK); + + return __builtin_bswap32(val32); +} + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_hw_20_comp_defs.h b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_20_comp_defs.h new file mode 100644 index 000000000000..208d4554283b --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_20_comp_defs.h @@ -0,0 +1,300 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef _ICP_QAT_HW_20_COMP_DEFS_H +#define _ICP_QAT_HW_20_COMP_DEFS_H + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_CONTROL_BITPOS 31 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_CONTROL_MASK 0x1 + +enum icp_qat_hw_comp_20_scb_control { + ICP_QAT_HW_COMP_20_SCB_CONTROL_ENABLE = 0x0, + ICP_QAT_HW_COMP_20_SCB_CONTROL_DISABLE = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SCB_CONTROL_DISABLE + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_RMB_CONTROL_BITPOS 30 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_RMB_CONTROL_MASK 0x1 + +enum icp_qat_hw_comp_20_rmb_control { + ICP_QAT_HW_COMP_20_RMB_CONTROL_RESET_ALL = 0x0, + ICP_QAT_HW_COMP_20_RMB_CONTROL_RESET_FC_ONLY = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_RMB_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_RMB_CONTROL_RESET_ALL + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SOM_CONTROL_BITPOS 28 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SOM_CONTROL_MASK 0x3 + +enum icp_qat_hw_comp_20_som_control { + ICP_QAT_HW_COMP_20_SOM_CONTROL_NORMAL_MODE = 0x0, + ICP_QAT_HW_COMP_20_SOM_CONTROL_REPLAY_MODE = 0x1, + ICP_QAT_HW_COMP_20_SOM_CONTROL_INPUT_CRC = 0x2, + ICP_QAT_HW_COMP_20_SOM_CONTROL_RESERVED_MODE = 0x3, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SOM_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SOM_CONTROL_NORMAL_MODE + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_RD_CONTROL_BITPOS 27 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_RD_CONTROL_MASK 0x1 + +enum icp_qat_hw_comp_20_skip_hash_rd_control { + ICP_QAT_HW_COMP_20_SKIP_HASH_RD_CONTROL_NO_SKIP = 0x0, + ICP_QAT_HW_COMP_20_SKIP_HASH_RD_CONTROL_SKIP_HASH_READS = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_RD_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SKIP_HASH_RD_CONTROL_NO_SKIP + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_UNLOAD_CONTROL_BITPOS 26 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_UNLOAD_CONTROL_MASK 0x1 + +enum icp_qat_hw_comp_20_scb_unload_control { + ICP_QAT_HW_COMP_20_SCB_UNLOAD_CONTROL_UNLOAD = 0x0, + ICP_QAT_HW_COMP_20_SCB_UNLOAD_CONTROL_NO_UNLOAD = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_UNLOAD_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SCB_UNLOAD_CONTROL_UNLOAD + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_DISABLE_TOKEN_FUSION_CONTROL_BITPOS 21 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_DISABLE_TOKEN_FUSION_CONTROL_MASK 0x1 + +enum icp_qat_hw_comp_20_disable_token_fusion_control { + ICP_QAT_HW_COMP_20_DISABLE_TOKEN_FUSION_CONTROL_ENABLE = 0x0, + ICP_QAT_HW_COMP_20_DISABLE_TOKEN_FUSION_CONTROL_DISABLE = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_DISABLE_TOKEN_FUSION_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_DISABLE_TOKEN_FUSION_CONTROL_ENABLE + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LBMS_BITPOS 19 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LBMS_MASK 0x3 + +enum icp_qat_hw_comp_20_lbms { + ICP_QAT_HW_COMP_20_LBMS_LBMS_64KB = 0x0, + ICP_QAT_HW_COMP_20_LBMS_LBMS_256KB = 0x1, + ICP_QAT_HW_COMP_20_LBMS_LBMS_1MB = 0x2, + ICP_QAT_HW_COMP_20_LBMS_LBMS_4MB = 0x3, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LBMS_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_LBMS_LBMS_64KB + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_MODE_RESET_MASK_BITPOS 18 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_MODE_RESET_MASK_MASK 0x1 + +enum icp_qat_hw_comp_20_scb_mode_reset_mask { + ICP_QAT_HW_COMP_20_SCB_MODE_RESET_MASK_RESET_COUNTERS = 0x0, + ICP_QAT_HW_COMP_20_SCB_MODE_RESET_MASK_RESET_COUNTERS_AND_HISTORY = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SCB_MODE_RESET_MASK_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SCB_MODE_RESET_MASK_RESET_COUNTERS + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LAZY_PARAM_BITPOS 9 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LAZY_PARAM_MASK 0x1ff +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LAZY_PARAM_DEFAULT_VAL 258 + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_NICE_PARAM_BITPOS 0 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_NICE_PARAM_MASK 0x1ff +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_NICE_PARAM_DEFAULT_VAL 259 + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_HBS_CONTROL_BITPOS 14 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_HBS_CONTROL_MASK 0x7 + +enum icp_qat_hw_comp_20_hbs_control { + ICP_QAT_HW_COMP_20_HBS_CONTROL_HBS_IS_32KB = 0x0, + ICP_QAT_HW_COMP_23_HBS_CONTROL_HBS_IS_64KB = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_HBS_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_HBS_CONTROL_HBS_IS_32KB + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_ABD_BITPOS 13 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_ABD_MASK 0x1 + +enum icp_qat_hw_comp_20_abd { + ICP_QAT_HW_COMP_20_ABD_ABD_ENABLED = 0x0, + ICP_QAT_HW_COMP_20_ABD_ABD_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_ABD_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_ABD_ABD_ENABLED + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LLLBD_CTRL_BITPOS 12 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LLLBD_CTRL_MASK 0x1 + +enum icp_qat_hw_comp_20_lllbd_ctrl { + ICP_QAT_HW_COMP_20_LLLBD_CTRL_LLLBD_ENABLED = 0x0, + ICP_QAT_HW_COMP_20_LLLBD_CTRL_LLLBD_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_LLLBD_CTRL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_LLLBD_CTRL_LLLBD_ENABLED + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SEARCH_DEPTH_BITPOS 8 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SEARCH_DEPTH_MASK 0xf + +enum icp_qat_hw_comp_20_search_depth { + ICP_QAT_HW_COMP_20_SEARCH_DEPTH_LEVEL_1 = 0x1, + ICP_QAT_HW_COMP_20_SEARCH_DEPTH_LEVEL_6 = 0x3, + ICP_QAT_HW_COMP_20_SEARCH_DEPTH_LEVEL_9 = 0x4, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SEARCH_DEPTH_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SEARCH_DEPTH_LEVEL_1 + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_HW_COMP_FORMAT_BITPOS 5 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_HW_COMP_FORMAT_MASK 0x7 + +enum icp_qat_hw_comp_20_hw_comp_format { + ICP_QAT_HW_COMP_20_HW_COMP_FORMAT_ILZ77 = 0x0, + ICP_QAT_HW_COMP_20_HW_COMP_FORMAT_DEFLATE = 0x1, + ICP_QAT_HW_COMP_20_HW_COMP_FORMAT_LZ4 = 0x2, + ICP_QAT_HW_COMP_20_HW_COMP_FORMAT_LZ4S = 0x3, + ICP_QAT_HW_COMP_23_HW_COMP_FORMAT_ZSTD = 0x4, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_HW_COMP_FORMAT_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_HW_COMP_FORMAT_DEFLATE + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_BITPOS 4 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_MASK 0x1 + +enum icp_qat_hw_comp_20_min_match_control { + ICP_QAT_HW_COMP_20_MIN_MATCH_CONTROL_MATCH_3B = 0x0, + ICP_QAT_HW_COMP_20_MIN_MATCH_CONTROL_MATCH_4B = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_MIN_MATCH_CONTROL_MATCH_3B + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_COLLISION_BITPOS 3 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_COLLISION_MASK 0x1 + +enum icp_qat_hw_comp_20_skip_hash_collision { + ICP_QAT_HW_COMP_20_SKIP_HASH_COLLISION_ALLOW = 0x0, + ICP_QAT_HW_COMP_20_SKIP_HASH_COLLISION_DONT_ALLOW = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_COLLISION_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SKIP_HASH_COLLISION_ALLOW + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_UPDATE_BITPOS 2 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_UPDATE_MASK 0x1 + +enum icp_qat_hw_comp_20_skip_hash_update { + ICP_QAT_HW_COMP_20_SKIP_HASH_UPDATE_ALLOW = 0x0, + ICP_QAT_HW_COMP_20_SKIP_HASH_UPDATE_DONT_ALLOW = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_SKIP_HASH_UPDATE_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_SKIP_HASH_UPDATE_ALLOW + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_BYTE_SKIP_BITPOS 1 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_BYTE_SKIP_MASK 0x1 + +enum icp_qat_hw_comp_20_byte_skip { + ICP_QAT_HW_COMP_20_BYTE_SKIP_3BYTE_TOKEN = 0x0, + ICP_QAT_HW_COMP_20_BYTE_SKIP_3BYTE_LITERAL = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_BYTE_SKIP_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_BYTE_SKIP_3BYTE_TOKEN + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_EXTENDED_DELAY_MATCH_MODE_BITPOS 0 +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_EXTENDED_DELAY_MATCH_MODE_MASK 0x1 + +enum icp_qat_hw_comp_20_extended_delay_match_mode { + ICP_QAT_HW_COMP_20_EXTENDED_DELAY_MATCH_MODE_EDMM_DISABLED = 0x0, + ICP_QAT_HW_COMP_20_EXTENDED_DELAY_MATCH_MODE_EDMM_ENABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_20_CONFIG_CSR_EXTENDED_DELAY_MATCH_MODE_DEFAULT_VAL \ + ICP_QAT_HW_COMP_20_EXTENDED_DELAY_MATCH_MODE_EDMM_DISABLED + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_SPECULATIVE_DECODER_CONTROL_BITPOS 31 +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_SPECULATIVE_DECODER_CONTROL_MASK 0x1 + +enum icp_qat_hw_decomp_20_speculative_decoder_control { + ICP_QAT_HW_DECOMP_20_SPECULATIVE_DECODER_CONTROL_ENABLE = 0x0, + ICP_QAT_HW_DECOMP_20_SPECULATIVE_DECODER_CONTROL_DISABLE = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_SPECULATIVE_DECODER_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_20_SPECULATIVE_DECODER_CONTROL_ENABLE + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MINI_CAM_CONTROL_BITPOS 30 +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MINI_CAM_CONTROL_MASK 0x1 + +enum icp_qat_hw_decomp_20_mini_cam_control { + ICP_QAT_HW_DECOMP_20_MINI_CAM_CONTROL_ENABLE = 0x0, + ICP_QAT_HW_DECOMP_20_MINI_CAM_CONTROL_DISABLE = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MINI_CAM_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_20_MINI_CAM_CONTROL_ENABLE + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HBS_CONTROL_BITPOS 14 +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HBS_CONTROL_MASK 0x7 + +enum icp_qat_hw_decomp_20_hbs_control { + ICP_QAT_HW_DECOMP_20_HBS_CONTROL_HBS_IS_32KB = 0x0, +}; + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HBS_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_20_HBS_CONTROL_HBS_IS_32KB + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LBMS_BITPOS 8 +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LBMS_MASK 0x3 + +enum icp_qat_hw_decomp_20_lbms { + ICP_QAT_HW_DECOMP_20_LBMS_LBMS_64KB = 0x0, + ICP_QAT_HW_DECOMP_20_LBMS_LBMS_256KB = 0x1, + ICP_QAT_HW_DECOMP_20_LBMS_LBMS_1MB = 0x2, + ICP_QAT_HW_DECOMP_20_LBMS_LBMS_4MB = 0x3, +}; + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LBMS_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_20_LBMS_LBMS_64KB + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HW_DECOMP_FORMAT_BITPOS 5 +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HW_DECOMP_FORMAT_MASK 0x7 + +enum icp_qat_hw_decomp_20_hw_comp_format { + ICP_QAT_HW_DECOMP_20_HW_DECOMP_FORMAT_DEFLATE = 0x1, + ICP_QAT_HW_DECOMP_20_HW_DECOMP_FORMAT_LZ4 = 0x2, + ICP_QAT_HW_DECOMP_20_HW_DECOMP_FORMAT_LZ4S = 0x3, + ICP_QAT_HW_DECOMP_23_HW_DECOMP_FORMAT_ZSTD = 0x4, +}; + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_HW_DECOMP_FORMAT_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_20_HW_DECOMP_FORMAT_DEFLATE + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_BITPOS 4 +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_MASK 0x1 + +enum icp_qat_hw_decomp_20_min_match_control { + ICP_QAT_HW_DECOMP_20_MIN_MATCH_CONTROL_MATCH_3B = 0x0, + ICP_QAT_HW_DECOMP_20_MIN_MATCH_CONTROL_MATCH_4B = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_MIN_MATCH_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_20_MIN_MATCH_CONTROL_MATCH_3B + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_PRESENT_BITPOS 3 +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_PRESENT_MASK 0x1 + +enum icp_qat_hw_decomp_20_lz4_block_checksum_present { + ICP_QAT_HW_DECOMP_20_LZ4_BLOCK_CHKSUM_ABSENT = 0x0, + ICP_QAT_HW_DECOMP_20_LZ4_BLOCK_CHKSUM_PRESENT = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_20_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_PRESENT_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_20_LZ4_BLOCK_CHKSUM_ABSENT + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_hw_51_comp.h b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_51_comp.h new file mode 100644 index 000000000000..dce639152345 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_51_comp.h @@ -0,0 +1,99 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ICP_QAT_HW_51_COMP_H_ +#define ICP_QAT_HW_51_COMP_H_ + +#include <linux/types.h> + +#include "icp_qat_fw.h" +#include "icp_qat_hw_51_comp_defs.h" + +struct icp_qat_hw_comp_51_config_csr_lower { + enum icp_qat_hw_comp_51_abd abd; + enum icp_qat_hw_comp_51_lllbd_ctrl lllbd; + enum icp_qat_hw_comp_51_search_depth sd; + enum icp_qat_hw_comp_51_min_match_control mmctrl; + enum icp_qat_hw_comp_51_lz4_block_checksum lbc; +}; + +static inline u32 +ICP_QAT_FW_COMP_51_BUILD_CONFIG_LOWER(struct icp_qat_hw_comp_51_config_csr_lower csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.abd, + ICP_QAT_HW_COMP_51_CONFIG_CSR_ABD_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_ABD_MASK); + QAT_FIELD_SET(val32, csr.lllbd, + ICP_QAT_HW_COMP_51_CONFIG_CSR_LLLBD_CTRL_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_LLLBD_CTRL_MASK); + QAT_FIELD_SET(val32, csr.sd, + ICP_QAT_HW_COMP_51_CONFIG_CSR_SEARCH_DEPTH_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_SEARCH_DEPTH_MASK); + QAT_FIELD_SET(val32, csr.mmctrl, + ICP_QAT_HW_COMP_51_CONFIG_CSR_MIN_MATCH_CONTROL_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_MIN_MATCH_CONTROL_MASK); + QAT_FIELD_SET(val32, csr.lbc, + ICP_QAT_HW_COMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_MASK); + + return val32; +} + +struct icp_qat_hw_comp_51_config_csr_upper { + enum icp_qat_hw_comp_51_dmm_algorithm edmm; + enum icp_qat_hw_comp_51_bms bms; + enum icp_qat_hw_comp_51_scb_mode_reset_mask scb_mode_reset; +}; + +static inline u32 +ICP_QAT_FW_COMP_51_BUILD_CONFIG_UPPER(struct icp_qat_hw_comp_51_config_csr_upper csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.edmm, + ICP_QAT_HW_COMP_51_CONFIG_CSR_DMM_ALGORITHM_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_DMM_ALGORITHM_MASK); + QAT_FIELD_SET(val32, csr.bms, + ICP_QAT_HW_COMP_51_CONFIG_CSR_BMS_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_BMS_MASK); + QAT_FIELD_SET(val32, csr.scb_mode_reset, + ICP_QAT_HW_COMP_51_CONFIG_CSR_SCB_MODE_RESET_MASK_BITPOS, + ICP_QAT_HW_COMP_51_CONFIG_CSR_SCB_MODE_RESET_MASK_MASK); + + return val32; +} + +struct icp_qat_hw_decomp_51_config_csr_lower { + enum icp_qat_hw_decomp_51_lz4_block_checksum lbc; +}; + +static inline u32 +ICP_QAT_FW_DECOMP_51_BUILD_CONFIG_LOWER(struct icp_qat_hw_decomp_51_config_csr_lower csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.lbc, + ICP_QAT_HW_DECOMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_BITPOS, + ICP_QAT_HW_DECOMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_MASK); + + return val32; +} + +struct icp_qat_hw_decomp_51_config_csr_upper { + enum icp_qat_hw_decomp_51_bms bms; +}; + +static inline u32 +ICP_QAT_FW_DECOMP_51_BUILD_CONFIG_UPPER(struct icp_qat_hw_decomp_51_config_csr_upper csr) +{ + u32 val32 = 0; + + QAT_FIELD_SET(val32, csr.bms, + ICP_QAT_HW_DECOMP_51_CONFIG_CSR_BMS_BITPOS, + ICP_QAT_HW_DECOMP_51_CONFIG_CSR_BMS_MASK); + + return val32; +} + +#endif /* ICP_QAT_HW_51_COMP_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_hw_51_comp_defs.h b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_51_comp_defs.h new file mode 100644 index 000000000000..e745688c5da4 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_hw_51_comp_defs.h @@ -0,0 +1,318 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2025 Intel Corporation */ +#ifndef ICP_QAT_HW_51_COMP_DEFS_H_ +#define ICP_QAT_HW_51_COMP_DEFS_H_ + +#include <linux/bits.h> + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SOM_CONTROL_BITPOS 28 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SOM_CONTROL_MASK GENMASK(1, 0) +enum icp_qat_hw_comp_51_som_control { + ICP_QAT_HW_COMP_51_SOM_CONTROL_NORMAL_MODE = 0x0, + ICP_QAT_HW_COMP_51_SOM_CONTROL_DICTIONARY_MODE = 0x1, + ICP_QAT_HW_COMP_51_SOM_CONTROL_INPUT_CRC = 0x2, + ICP_QAT_HW_COMP_51_SOM_CONTROL_RESERVED_MODE = 0x3, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SOM_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SOM_CONTROL_NORMAL_MODE +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_RD_CONTROL_BITPOS 27 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_RD_CONTROL_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_skip_hash_rd_control { + ICP_QAT_HW_COMP_51_SKIP_HASH_RD_CONTROL_NO_SKIP = 0x0, + ICP_QAT_HW_COMP_51_SKIP_HASH_RD_CONTROL_SKIP_HASH_READS = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_RD_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SKIP_HASH_RD_CONTROL_NO_SKIP +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BYPASS_COMPRESSION_BITPOS 25 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BYPASS_COMPRESSION_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_bypass_compression { + ICP_QAT_HW_COMP_51_BYPASS_COMPRESSION_DISABLED = 0x0, + ICP_QAT_HW_COMP_51_BYPASS_COMPRESSION_ENABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BYPASS_COMPRESSION_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_BYPASS_COMPRESSION_DISABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_DMM_ALGORITHM_BITPOS 22 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_DMM_ALGORITHM_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_dmm_algorithm { + ICP_QAT_HW_COMP_51_DMM_ALGORITHM_EDMM_ENABLED = 0x0, + ICP_QAT_HW_COMP_51_DMM_ALGORITHM_ZSTD_DMM_LITE = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_DMM_ALGORITHM_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_DMM_ALGORITHM_EDMM_ENABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_TOKEN_FUSION_INTERNAL_ONLY_BITPOS 21 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_TOKEN_FUSION_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_token_fusion_internal_only { + ICP_QAT_HW_COMP_51_TOKEN_FUSION_INTERNAL_ONLY_ENABLED = 0x0, + ICP_QAT_HW_COMP_51_TOKEN_FUSION_INTERNAL_ONLY_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_TOKEN_FUSION_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_TOKEN_FUSION_INTERNAL_ONLY_ENABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BMS_BITPOS 19 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BMS_MASK GENMASK(1, 0) +enum icp_qat_hw_comp_51_bms { + ICP_QAT_HW_COMP_51_BMS_BMS_64KB = 0x0, + ICP_QAT_HW_COMP_51_BMS_BMS_256KB = 0x1, + ICP_QAT_HW_COMP_51_BMS_BMS_1MB = 0x2, + ICP_QAT_HW_COMP_51_BMS_BMS_4MB = 0x3, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BMS_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_BMS_BMS_64KB +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SCB_MODE_RESET_MASK_BITPOS 18 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SCB_MODE_RESET_MASK_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_scb_mode_reset_mask { + ICP_QAT_HW_COMP_51_SCB_MODE_RESET_MASK_DO_NOT_RESET_HB_HT = 0x0, + ICP_QAT_HW_COMP_51_SCB_MODE_RESET_MASK_RESET_HB_HT = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SCB_MODE_RESET_MASK_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SCB_MODE_RESET_MASK_DO_NOT_RESET_HB_HT +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ZSTD_FRAME_GEN_DEC_EN_BITPOS 2 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ZSTD_FRAME_GEN_DEC_EN_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_zstd_frame_gen_dec_en { + ICP_QAT_HW_COMP_51_ZSTD_FRAME_GEN_DEC_EN_ZSTD_FRAME_HDR_DISABLE = 0x0, + ICP_QAT_HW_COMP_51_ZSTD_FRAME_GEN_DEC_EN_ZSTD_FRAME_HDR_ENABLE = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ZSTD_FRAME_GEN_DEC_EN_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_ZSTD_FRAME_GEN_DEC_EN_ZSTD_FRAME_HDR_ENABLE +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_CNV_DISABLE_BITPOS 1 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_CNV_DISABLE_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_cnv_disable { + ICP_QAT_HW_COMP_51_CNV_DISABLE_CNV_ENABLED = 0x0, + ICP_QAT_HW_COMP_51_CNV_DISABLE_CNV_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_CNV_DISABLE_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_CNV_DISABLE_CNV_ENABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ASB_DISABLE_BITPOS 0 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ASB_DISABLE_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_asb_disable { + ICP_QAT_HW_COMP_51_ASB_DISABLE_ASB_ENABLED = 0x0, + ICP_QAT_HW_COMP_51_ASB_DISABLE_ASB_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ASB_DISABLE_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_ASB_DISABLE_ASB_ENABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SPEC_DECODER_INTERNAL_ONLY_BITPOS 21 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SPEC_DECODER_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_spec_decoder_internal_only { + ICP_QAT_HW_COMP_51_SPEC_DECODER_INTERNAL_ONLY_NORMAL = 0x0, + ICP_QAT_HW_COMP_51_SPEC_DECODER_INTERNAL_ONLY_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SPEC_DECODER_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SPEC_DECODER_INTERNAL_ONLY_NORMAL +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_MINI_XCAM_INTERNAL_ONLY_BITPOS 20 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_MINI_XCAM_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_mini_xcam_internal_only { + ICP_QAT_HW_COMP_51_MINI_XCAM_INTERNAL_ONLY_NORMAL = 0x0, + ICP_QAT_HW_COMP_51_MINI_XCAM_INTERNAL_ONLY_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_MINI_XCAM_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_MINI_XCAM_INTERNAL_ONLY_NORMAL +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_REP_OFF_ENC_INTERNAL_ONLY_BITPOS 19 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_REP_OFF_ENC_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_rep_off_enc_internal_only { + ICP_QAT_HW_COMP_51_REP_OFF_ENC_INTERNAL_ONLY_ENABLED = 0x0, + ICP_QAT_HW_COMP_51_REP_OFF_ENC_INTERNAL_ONLY_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_REP_OFF_ENC_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_REP_OFF_ENC_INTERNAL_ONLY_ENABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_PROG_BLOCK_DROP_INTERNAL_ONLY_BITPOS 18 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_PROG_BLOCK_DROP_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_prog_block_drop_internal_only { + ICP_QAT_HW_COMP_51_PROG_BLOCK_DROP_INTERNAL_ONLY_DISABLE = 0x0, + ICP_QAT_HW_COMP_51_PROG_BLOCK_DROP_INTERNAL_ONLY_ENABLE = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_PROG_BLOCK_DROP_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_PROG_BLOCK_DROP_INTERNAL_ONLY_DISABLE +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_OVERRIDE_INTERNAL_ONLY_BITPOS 17 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_OVERRIDE_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_skip_hash_override_internal_only { + ICP_QAT_HW_COMP_51_SKIP_HASH_OVERRIDE_INTERNAL_ONLY_DETERMINE_HASH_PARAMS = 0x0, + ICP_QAT_HW_COMP_51_SKIP_HASH_OVERRIDE_INTERNAL_ONLY_OVERRIDE_HASH_PARAMS = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_OVERRIDE_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SKIP_HASH_OVERRIDE_INTERNAL_ONLY_DETERMINE_HASH_PARAMS +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_HBS_BITPOS 14 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_HBS_MASK GENMASK(2, 0) +enum icp_qat_hw_comp_51_hbs { + ICP_QAT_HW_COMP_51_HBS_32KB = 0x0, + ICP_QAT_HW_COMP_51_HBS_64KB = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_HBS_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_HBS_32KB +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ABD_BITPOS 13 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ABD_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_abd { + ICP_QAT_HW_COMP_51_ABD_ABD_ENABLED = 0x0, + ICP_QAT_HW_COMP_51_ABD_ABD_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_ABD_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_ABD_ABD_ENABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_LLLBD_CTRL_BITPOS 12 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_LLLBD_CTRL_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_lllbd_ctrl { + ICP_QAT_HW_COMP_51_LLLBD_CTRL_LLLBD_ENABLED = 0x0, + ICP_QAT_HW_COMP_51_LLLBD_CTRL_LLLBD_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_LLLBD_CTRL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_LLLBD_CTRL_LLLBD_ENABLED +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SEARCH_DEPTH_BITPOS 8 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SEARCH_DEPTH_MASK GENMASK(3, 0) +enum icp_qat_hw_comp_51_search_depth { + ICP_QAT_HW_COMP_51_SEARCH_DEPTH_LEVEL_1 = 0x1, + ICP_QAT_HW_COMP_51_SEARCH_DEPTH_LEVEL_6 = 0x3, + ICP_QAT_HW_COMP_51_SEARCH_DEPTH_LEVEL_9 = 0x4, + ICP_QAT_HW_COMP_51_SEARCH_DEPTH_LEVEL_10 = 0x4, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SEARCH_DEPTH_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SEARCH_DEPTH_LEVEL_1 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_FORMAT_BITPOS 5 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_FORMAT_MASK GENMASK(2, 0) +enum icp_qat_hw_comp_51_format { + ICP_QAT_HW_COMP_51_FORMAT_ILZ77 = 0x1, + ICP_QAT_HW_COMP_51_FORMAT_LZ4 = 0x2, + ICP_QAT_HW_COMP_51_FORMAT_LZ4s = 0x3, + ICP_QAT_HW_COMP_51_FORMAT_ZSTD = 0x4, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_FORMAT_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_FORMAT_ILZ77 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_MIN_MATCH_CONTROL_BITPOS 4 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_MIN_MATCH_CONTROL_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_min_match_control { + ICP_QAT_HW_COMP_51_MIN_MATCH_CONTROL_MATCH_3B = 0x0, + ICP_QAT_HW_COMP_51_MIN_MATCH_CONTROL_MATCH_4B = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_MIN_MATCH_CONTROL_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_MIN_MATCH_CONTROL_MATCH_3B +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_COLLISION_BITPOS 3 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_COLLISION_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_skip_hash_collision { + ICP_QAT_HW_COMP_51_SKIP_HASH_COLLISION_ALLOW = 0x0, + ICP_QAT_HW_COMP_51_SKIP_HASH_COLLISION_DONT_ALLOW = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_COLLISION_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SKIP_HASH_COLLISION_ALLOW +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_UPDATE_BITPOS 2 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_UPDATE_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_skip_hash_update { + ICP_QAT_HW_COMP_51_SKIP_HASH_UPDATE_ALLOW = 0x0, + ICP_QAT_HW_COMP_51_SKIP_HASH_UPDATE_DONT_ALLOW = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_SKIP_HASH_UPDATE_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_SKIP_HASH_UPDATE_ALLOW +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BYTE_SKIP_BITPOS 1 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BYTE_SKIP_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_byte_skip { + ICP_QAT_HW_COMP_51_BYTE_SKIP_3BYTE_TOKEN = 0x0, + ICP_QAT_HW_COMP_51_BYTE_SKIP_3BYTE_LITERAL = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_BYTE_SKIP_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_BYTE_SKIP_3BYTE_TOKEN +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_BITPOS 0 +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_MASK GENMASK(0, 0) +enum icp_qat_hw_comp_51_lz4_block_checksum { + ICP_QAT_HW_COMP_51_LZ4_BLOCK_CHECKSUM_ABSENT = 0x0, + ICP_QAT_HW_COMP_51_LZ4_BLOCK_CHECKSUM_PRESENT = 0x1, +}; + +#define ICP_QAT_HW_COMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_DEFAULT_VAL \ + ICP_QAT_HW_COMP_51_LZ4_BLOCK_CHECKSUM_ABSENT +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_DISCARD_DATA_BITPOS 26 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_DISCARD_DATA_MASK GENMASK(0, 0) +enum icp_qat_hw_decomp_51_discard_data { + ICP_QAT_HW_DECOMP_51_DISCARD_DATA_DISABLED = 0x0, + ICP_QAT_HW_DECOMP_51_DISCARD_DATA_ENABLED = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_DISCARD_DATA_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_DISCARD_DATA_DISABLED +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_BMS_BITPOS 19 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_BMS_MASK GENMASK(1, 0) +enum icp_qat_hw_decomp_51_bms { + ICP_QAT_HW_DECOMP_51_BMS_BMS_64KB = 0x0, + ICP_QAT_HW_DECOMP_51_BMS_BMS_256KB = 0x1, + ICP_QAT_HW_DECOMP_51_BMS_BMS_1MB = 0x2, + ICP_QAT_HW_DECOMP_51_BMS_BMS_4MB = 0x3, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_BMS_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_BMS_BMS_64KB +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_ZSTD_FRAME_GEN_DEC_EN_BITPOS 2 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_ZSTD_FRAME_GEN_DEC_EN_MASK GENMASK(0, 0) +enum icp_qat_hw_decomp_51_zstd_frame_gen_dec_en { + ICP_QAT_HW_DECOMP_51_ZSTD_FRAME_GEN_DEC_EN_ZSTD_FRAME_HDR_DISABLE = 0x0, + ICP_QAT_HW_DECOMP_51_ZSTD_FRAME_GEN_DEC_EN_ZSTD_FRAME_HDR_ENABLE = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_ZSTD_FRAME_GEN_DEC_EN_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_ZSTD_FRAME_GEN_DEC_EN_ZSTD_FRAME_HDR_ENABLE +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_SPEC_DECODER_INTERNAL_ONLY_BITPOS 21 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_SPEC_DECODER_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_decomp_51_spec_decoder_internal_only { + ICP_QAT_HW_DECOMP_51_SPEC_DECODER_INTERNAL_ONLY_NORMAL = 0x0, + ICP_QAT_HW_DECOMP_51_SPEC_DECODER_INTERNAL_ONLY_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_SPEC_DECODER_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_SPEC_DECODER_INTERNAL_ONLY_NORMAL +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_MINI_XCAM_INTERNAL_ONLY_BITPOS 20 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_MINI_XCAM_INTERNAL_ONLY_MASK GENMASK(0, 0) +enum icp_qat_hw_decomp_51_mini_xcam_internal_only { + ICP_QAT_HW_DECOMP_51_MINI_XCAM_INTERNAL_ONLY_NORMAL = 0x0, + ICP_QAT_HW_DECOMP_51_MINI_XCAM_INTERNAL_ONLY_DISABLED = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_MINI_XCAM_INTERNAL_ONLY_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_MINI_XCAM_INTERNAL_ONLY_NORMAL +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_HBS_BITPOS 14 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_HBS_MASK GENMASK(2, 0) +enum icp_qat_hw_decomp_51_hbs { + ICP_QAT_HW_DECOMP_51_HBS_32KB = 0x0, + ICP_QAT_HW_DECOMP_51_HBS_64KB = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_HBS_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_HBS_32KB +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_FORMAT_BITPOS 5 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_FORMAT_MASK GENMASK(2, 0) +enum icp_qat_hw_decomp_51_format { + ICP_QAT_HW_DECOMP_51_FORMAT_ILZ77 = 0x1, + ICP_QAT_HW_DECOMP_51_FORMAT_LZ4 = 0x2, + ICP_QAT_HW_DECOMP_51_FORMAT_RESERVED = 0x3, + ICP_QAT_HW_DECOMP_51_FORMAT_ZSTD = 0x4, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_FORMAT_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_FORMAT_ILZ77 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_BITPOS 0 +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_MASK GENMASK(0, 0) +enum icp_qat_hw_decomp_51_lz4_block_checksum { + ICP_QAT_HW_DECOMP_51_LZ4_BLOCK_CHECKSUM_ABSENT = 0x0, + ICP_QAT_HW_DECOMP_51_LZ4_BLOCK_CHECKSUM_PRESENT = 0x1, +}; + +#define ICP_QAT_HW_DECOMP_51_CONFIG_CSR_LZ4_BLOCK_CHECKSUM_DEFAULT_VAL \ + ICP_QAT_HW_DECOMP_51_LZ4_BLOCK_CHECKSUM_ABSENT + +#endif /* ICP_QAT_HW_51_COMP_DEFS_H_ */ diff --git a/drivers/crypto/intel/qat/qat_common/icp_qat_uclo.h b/drivers/crypto/intel/qat/qat_common/icp_qat_uclo.h new file mode 100644 index 000000000000..6313c35eff0c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/icp_qat_uclo.h @@ -0,0 +1,598 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef __ICP_QAT_UCLO_H__ +#define __ICP_QAT_UCLO_H__ + +#define ICP_QAT_AC_895XCC_DEV_TYPE 0x00400000 +#define ICP_QAT_AC_C62X_DEV_TYPE 0x01000000 +#define ICP_QAT_AC_C3XXX_DEV_TYPE 0x02000000 +#define ICP_QAT_AC_4XXX_A_DEV_TYPE 0x08000000 +#define ICP_QAT_AC_6XXX_DEV_TYPE 0x80000000 +#define ICP_QAT_UCLO_MAX_AE 17 +#define ICP_QAT_UCLO_MAX_CTX 8 +#define ICP_QAT_UCLO_MAX_UIMAGE (ICP_QAT_UCLO_MAX_AE * ICP_QAT_UCLO_MAX_CTX) +#define ICP_QAT_UCLO_MAX_USTORE 0x4000 +#define ICP_QAT_UCLO_MAX_XFER_REG 128 +#define ICP_QAT_UCLO_MAX_GPR_REG 128 +#define ICP_QAT_UCLO_MAX_LMEM_REG 1024 +#define ICP_QAT_UCLO_MAX_LMEM_REG_2X 1280 +#define ICP_QAT_UCLO_AE_ALL_CTX 0xff +#define ICP_QAT_UOF_OBJID_LEN 8 +#define ICP_QAT_UOF_FID 0xc6c2 +#define ICP_QAT_UOF_MAJVER 0x4 +#define ICP_QAT_UOF_MINVER 0x11 +#define ICP_QAT_UOF_OBJS "UOF_OBJS" +#define ICP_QAT_UOF_STRT "UOF_STRT" +#define ICP_QAT_UOF_IMAG "UOF_IMAG" +#define ICP_QAT_UOF_IMEM "UOF_IMEM" +#define ICP_QAT_UOF_LOCAL_SCOPE 1 +#define ICP_QAT_UOF_INIT_EXPR 0 +#define ICP_QAT_UOF_INIT_REG 1 +#define ICP_QAT_UOF_INIT_REG_CTX 2 +#define ICP_QAT_UOF_INIT_EXPR_ENDIAN_SWAP 3 +#define ICP_QAT_SUOF_OBJ_ID_LEN 8 +#define ICP_QAT_SUOF_FID 0x53554f46 +#define ICP_QAT_SUOF_MAJVER 0x0 +#define ICP_QAT_SUOF_MINVER 0x1 +#define ICP_QAT_SUOF_OBJ_NAME_LEN 128 +#define ICP_QAT_MOF_OBJ_ID_LEN 8 +#define ICP_QAT_MOF_OBJ_CHUNKID_LEN 8 +#define ICP_QAT_MOF_FID 0x00666f6d +#define ICP_QAT_MOF_MAJVER 0x0 +#define ICP_QAT_MOF_MINVER 0x1 +#define ICP_QAT_MOF_SYM_OBJS "SYM_OBJS" +#define ICP_QAT_SUOF_OBJS "SUF_OBJS" +#define ICP_QAT_SUOF_IMAG "SUF_IMAG" +#define ICP_QAT_SIMG_AE_INIT_SEQ_LEN (50 * sizeof(unsigned long long)) + +#define DSS_FWSK_MODULUS_LEN 384 /* RSA3K */ +#define DSS_FWSK_EXPONENT_LEN 4 +#define DSS_FWSK_PADDING_LEN 380 +#define DSS_SIGNATURE_LEN 384 /* RSA3K */ + +#define CSS_FWSK_MODULUS_LEN 256 /* RSA2K */ +#define CSS_FWSK_EXPONENT_LEN 4 +#define CSS_FWSK_PADDING_LEN 252 +#define CSS_SIGNATURE_LEN 256 /* RSA2K */ + +#define ICP_QAT_CSS_FWSK_MODULUS_LEN(handle) ((handle)->chip_info->css_3k ? \ + DSS_FWSK_MODULUS_LEN : \ + CSS_FWSK_MODULUS_LEN) + +#define ICP_QAT_CSS_FWSK_EXPONENT_LEN(handle) ((handle)->chip_info->css_3k ? \ + DSS_FWSK_EXPONENT_LEN : \ + CSS_FWSK_EXPONENT_LEN) + +#define ICP_QAT_CSS_FWSK_PAD_LEN(handle) ((handle)->chip_info->css_3k ? \ + DSS_FWSK_PADDING_LEN : \ + CSS_FWSK_PADDING_LEN) + +#define ICP_QAT_CSS_FWSK_PUB_LEN(handle) (ICP_QAT_CSS_FWSK_MODULUS_LEN(handle) + \ + ICP_QAT_CSS_FWSK_EXPONENT_LEN(handle) + \ + ICP_QAT_CSS_FWSK_PAD_LEN(handle)) + +#define ICP_QAT_CSS_SIGNATURE_LEN(handle) ((handle)->chip_info->css_3k ? \ + DSS_SIGNATURE_LEN : \ + CSS_SIGNATURE_LEN) + +#define ICP_QAT_AE_IMG_OFFSET(handle) (sizeof(struct icp_qat_css_hdr) + \ + ICP_QAT_CSS_FWSK_MODULUS_LEN(handle) + \ + ICP_QAT_CSS_FWSK_EXPONENT_LEN(handle) + \ + ICP_QAT_CSS_SIGNATURE_LEN(handle)) +#define ICP_QAT_CSS_RSA4K_MAX_IMAGE_LEN 0x40000 +#define ICP_QAT_CSS_RSA3K_MAX_IMAGE_LEN 0x30000 + +/* All lengths below are in bytes */ +#define ICP_QAT_DUALSIGN_OPAQUE_HDR_LEN 12 +#define ICP_QAT_DUALSIGN_OPAQUE_HDR_ALIGN_LEN 16 +#define ICP_QAT_DUALSIGN_OPAQUE_DATA_LEN 3540 +#define ICP_QAT_DUALSIGN_XMSS_PUBKEY_LEN 64 +#define ICP_QAT_DUALSIGN_XMSS_SIG_LEN 2692 +#define ICP_QAT_DUALSIGN_XMSS_SIG_ALIGN_LEN 2696 +#define ICP_QAT_DUALSIGN_MISC_INFO_LEN 16 +#define ICP_QAT_DUALSIGN_FW_TYPE_LEN 7 +#define ICP_QAT_DUALSIGN_MODULE_TYPE 0x14 +#define ICP_QAT_DUALSIGN_HDR_LEN 0x375 +#define ICP_QAT_DUALSIGN_HDR_VER 0x40001 +#define ICP_QAT_DUALSIGN_HDR_LEN_OFFSET 4 +#define ICP_QAT_DUALSIGN_HDR_VER_OFFSET 8 + +#define ICP_QAT_CTX_MODE(ae_mode) ((ae_mode) & 0xf) +#define ICP_QAT_NN_MODE(ae_mode) (((ae_mode) >> 0x4) & 0xf) +#define ICP_QAT_SHARED_USTORE_MODE(ae_mode) (((ae_mode) >> 0xb) & 0x1) +#define RELOADABLE_CTX_SHARED_MODE(ae_mode) (((ae_mode) >> 0xc) & 0x1) + +#define ICP_QAT_LOC_MEM0_MODE(ae_mode) (((ae_mode) >> 0x8) & 0x1) +#define ICP_QAT_LOC_MEM1_MODE(ae_mode) (((ae_mode) >> 0x9) & 0x1) +#define ICP_QAT_LOC_MEM2_MODE(ae_mode) (((ae_mode) >> 0x6) & 0x1) +#define ICP_QAT_LOC_MEM3_MODE(ae_mode) (((ae_mode) >> 0x7) & 0x1) +#define ICP_QAT_LOC_TINDEX_MODE(ae_mode) (((ae_mode) >> 0xe) & 0x1) + +enum icp_qat_uof_mem_region { + ICP_QAT_UOF_SRAM_REGION = 0x0, + ICP_QAT_UOF_LMEM_REGION = 0x3, + ICP_QAT_UOF_UMEM_REGION = 0x5 +}; + +enum icp_qat_uof_regtype { + ICP_NO_DEST = 0, + ICP_GPA_REL = 1, + ICP_GPA_ABS = 2, + ICP_GPB_REL = 3, + ICP_GPB_ABS = 4, + ICP_SR_REL = 5, + ICP_SR_RD_REL = 6, + ICP_SR_WR_REL = 7, + ICP_SR_ABS = 8, + ICP_SR_RD_ABS = 9, + ICP_SR_WR_ABS = 10, + ICP_DR_REL = 19, + ICP_DR_RD_REL = 20, + ICP_DR_WR_REL = 21, + ICP_DR_ABS = 22, + ICP_DR_RD_ABS = 23, + ICP_DR_WR_ABS = 24, + ICP_LMEM = 26, + ICP_LMEM0 = 27, + ICP_LMEM1 = 28, + ICP_NEIGH_REL = 31, + ICP_LMEM2 = 61, + ICP_LMEM3 = 62, +}; + +enum icp_qat_css_fwtype { + CSS_AE_FIRMWARE = 0, + CSS_MMP_FIRMWARE = 1 +}; + +struct icp_qat_uclo_page { + struct icp_qat_uclo_encap_page *encap_page; + struct icp_qat_uclo_region *region; + unsigned int flags; +}; + +struct icp_qat_uclo_region { + struct icp_qat_uclo_page *loaded; + struct icp_qat_uclo_page *page; +}; + +struct icp_qat_uclo_aeslice { + struct icp_qat_uclo_region *region; + struct icp_qat_uclo_page *page; + struct icp_qat_uclo_page *cur_page[ICP_QAT_UCLO_MAX_CTX]; + struct icp_qat_uclo_encapme *encap_image; + unsigned int ctx_mask_assigned; + unsigned int new_uaddr[ICP_QAT_UCLO_MAX_CTX]; +}; + +struct icp_qat_uclo_aedata { + unsigned int slice_num; + unsigned int eff_ustore_size; + struct icp_qat_uclo_aeslice ae_slices[ICP_QAT_UCLO_MAX_CTX]; +}; + +struct icp_qat_uof_encap_obj { + char *beg_uof; + struct icp_qat_uof_objhdr *obj_hdr; + struct icp_qat_uof_chunkhdr *chunk_hdr; + struct icp_qat_uof_varmem_seg *var_mem_seg; +}; + +struct icp_qat_uclo_encap_uwblock { + unsigned int start_addr; + unsigned int words_num; + u64 micro_words; +}; + +struct icp_qat_uclo_encap_page { + unsigned int def_page; + unsigned int page_region; + unsigned int beg_addr_v; + unsigned int beg_addr_p; + unsigned int micro_words_num; + unsigned int uwblock_num; + struct icp_qat_uclo_encap_uwblock *uwblock; +}; + +struct icp_qat_uclo_encapme { + struct icp_qat_uof_image *img_ptr; + struct icp_qat_uclo_encap_page *page; + unsigned int ae_reg_num; + struct icp_qat_uof_ae_reg *ae_reg; + unsigned int init_regsym_num; + struct icp_qat_uof_init_regsym *init_regsym; + unsigned int sbreak_num; + struct icp_qat_uof_sbreak *sbreak; + unsigned int uwords_num; +}; + +struct icp_qat_uclo_init_mem_table { + unsigned int entry_num; + struct icp_qat_uof_initmem *init_mem; +}; + +struct icp_qat_uclo_objhdr { + char *file_buff; + unsigned int checksum; + unsigned int size; +}; + +struct icp_qat_uof_strtable { + unsigned int table_len; + unsigned int reserved; + u64 strings; +}; + +struct icp_qat_uclo_objhandle { + unsigned int prod_type; + unsigned int prod_rev; + struct icp_qat_uclo_objhdr *obj_hdr; + struct icp_qat_uof_encap_obj encap_uof_obj; + struct icp_qat_uof_strtable str_table; + struct icp_qat_uclo_encapme ae_uimage[ICP_QAT_UCLO_MAX_UIMAGE]; + struct icp_qat_uclo_aedata ae_data[ICP_QAT_UCLO_MAX_AE]; + struct icp_qat_uclo_init_mem_table init_mem_tab; + struct icp_qat_uof_batch_init *lm_init_tab[ICP_QAT_UCLO_MAX_AE]; + struct icp_qat_uof_batch_init *umem_init_tab[ICP_QAT_UCLO_MAX_AE]; + int uimage_num; + int uword_in_bytes; + int global_inited; + unsigned int ae_num; + unsigned int ustore_phy_size; + void *obj_buf; + u64 *uword_buf; +}; + +struct icp_qat_uof_uword_block { + unsigned int start_addr; + unsigned int words_num; + unsigned int uword_offset; + unsigned int reserved; +}; + +struct icp_qat_uof_filehdr { + unsigned short file_id; + unsigned short reserved1; + char min_ver; + char maj_ver; + unsigned short reserved2; + unsigned short max_chunks; + unsigned short num_chunks; +}; + +struct icp_qat_uof_filechunkhdr { + char chunk_id[ICP_QAT_UOF_OBJID_LEN]; + unsigned int checksum; + unsigned int offset; + unsigned int size; +}; + +struct icp_qat_uof_objhdr { + unsigned int ac_dev_type; + unsigned short min_cpu_ver; + unsigned short max_cpu_ver; + short max_chunks; + short num_chunks; + unsigned int reserved1; + unsigned int reserved2; +}; + +struct icp_qat_uof_chunkhdr { + char chunk_id[ICP_QAT_UOF_OBJID_LEN]; + unsigned int offset; + unsigned int size; +}; + +struct icp_qat_uof_memvar_attr { + unsigned int offset_in_byte; + unsigned int value; +}; + +struct icp_qat_uof_initmem { + unsigned int sym_name; + char region; + char scope; + unsigned short reserved1; + unsigned int addr; + unsigned int num_in_bytes; + unsigned int val_attr_num; +}; + +struct icp_qat_uof_init_regsym { + unsigned int sym_name; + char init_type; + char value_type; + char reg_type; + unsigned char ctx; + unsigned int reg_addr; + unsigned int value; +}; + +struct icp_qat_uof_varmem_seg { + unsigned int sram_base; + unsigned int sram_size; + unsigned int sram_alignment; + unsigned int sdram_base; + unsigned int sdram_size; + unsigned int sdram_alignment; + unsigned int sdram1_base; + unsigned int sdram1_size; + unsigned int sdram1_alignment; + unsigned int scratch_base; + unsigned int scratch_size; + unsigned int scratch_alignment; +}; + +struct icp_qat_uof_gtid { + char tool_id[ICP_QAT_UOF_OBJID_LEN]; + int tool_ver; + unsigned int reserved1; + unsigned int reserved2; +}; + +struct icp_qat_uof_sbreak { + unsigned int page_num; + unsigned int virt_uaddr; + unsigned char sbreak_type; + unsigned char reg_type; + unsigned short reserved1; + unsigned int addr_offset; + unsigned int reg_addr; +}; + +struct icp_qat_uof_code_page { + unsigned int page_region; + unsigned int page_num; + unsigned char def_page; + unsigned char reserved2; + unsigned short reserved1; + unsigned int beg_addr_v; + unsigned int beg_addr_p; + unsigned int neigh_reg_tab_offset; + unsigned int uc_var_tab_offset; + unsigned int imp_var_tab_offset; + unsigned int imp_expr_tab_offset; + unsigned int code_area_offset; +}; + +struct icp_qat_uof_image { + unsigned int img_name; + unsigned int ae_assigned; + unsigned int ctx_assigned; + unsigned int ac_dev_type; + unsigned int entry_address; + unsigned int fill_pattern[2]; + unsigned int reloadable_size; + unsigned char sensitivity; + unsigned char reserved; + unsigned short ae_mode; + unsigned short max_ver; + unsigned short min_ver; + unsigned short image_attrib; + unsigned short reserved2; + unsigned short page_region_num; + unsigned short numpages; + unsigned int reg_tab_offset; + unsigned int init_reg_sym_tab; + unsigned int sbreak_tab; + unsigned int app_metadata; +}; + +struct icp_qat_uof_objtable { + unsigned int entry_num; +}; + +struct icp_qat_uof_ae_reg { + unsigned int name; + unsigned int vis_name; + unsigned short type; + unsigned short addr; + unsigned short access_mode; + unsigned char visible; + unsigned char reserved1; + unsigned short ref_count; + unsigned short reserved2; + unsigned int xo_id; +}; + +struct icp_qat_uof_code_area { + unsigned int micro_words_num; + unsigned int uword_block_tab; +}; + +struct icp_qat_uof_batch_init { + unsigned int ae; + unsigned int addr; + unsigned int *value; + unsigned int size; + struct icp_qat_uof_batch_init *next; +}; + +struct icp_qat_suof_img_hdr { + char *simg_buf; + unsigned long simg_len; + char *css_header; + char *css_simg; + unsigned long simg_size; + unsigned int ae_num; + unsigned int ae_mask; + unsigned int fw_type; + unsigned long simg_name; + unsigned long appmeta_data; +}; + +struct icp_qat_suof_img_tbl { + unsigned int num_simgs; + struct icp_qat_suof_img_hdr *simg_hdr; +}; + +struct icp_qat_suof_handle { + unsigned int file_id; + unsigned int check_sum; + char min_ver; + char maj_ver; + char fw_type; + char *suof_buf; + unsigned int suof_size; + char *sym_str; + unsigned int sym_size; + struct icp_qat_suof_img_tbl img_table; +}; + +struct icp_qat_fw_auth_desc { + unsigned int img_len; + unsigned int ae_mask; + unsigned int css_hdr_high; + unsigned int css_hdr_low; + unsigned int img_high; + unsigned int img_low; + unsigned int signature_high; + unsigned int signature_low; + unsigned int fwsk_pub_high; + unsigned int fwsk_pub_low; + unsigned int img_ae_mode_data_high; + unsigned int img_ae_mode_data_low; + unsigned int img_ae_init_data_high; + unsigned int img_ae_init_data_low; + unsigned int img_ae_insts_high; + unsigned int img_ae_insts_low; + unsigned int cpp_mask; + unsigned int reserved; + unsigned int xmss_pubkey_high; + unsigned int xmss_pubkey_low; + unsigned int xmss_sig_high; + unsigned int xmss_sig_low; + unsigned int reserved2[2]; +}; + +struct icp_qat_auth_chunk { + struct icp_qat_fw_auth_desc fw_auth_desc; + u64 chunk_size; + u64 chunk_bus_addr; +}; + +struct icp_qat_css_hdr { + unsigned int module_type; + unsigned int header_len; + unsigned int header_ver; + unsigned int module_id; + unsigned int module_vendor; + unsigned int date; + unsigned int size; + unsigned int key_size; + unsigned int module_size; + unsigned int exponent_size; + unsigned int fw_type; + unsigned int reserved[21]; +}; + +struct icp_qat_simg_ae_mode { + unsigned int file_id; + unsigned short maj_ver; + unsigned short min_ver; + unsigned int dev_type; + unsigned short devmax_ver; + unsigned short devmin_ver; + unsigned int ae_mask; + unsigned int ctx_enables; + char fw_type; + char ctx_mode; + char nn_mode; + char lm0_mode; + char lm1_mode; + char scs_mode; + char lm2_mode; + char lm3_mode; + char tindex_mode; + unsigned char reserved[7]; + char simg_name[256]; + char appmeta_data[256]; +}; + +struct icp_qat_suof_filehdr { + unsigned int file_id; + unsigned int check_sum; + char min_ver; + char maj_ver; + char fw_type; + char reserved; + unsigned short max_chunks; + unsigned short num_chunks; +}; + +struct icp_qat_suof_chunk_hdr { + char chunk_id[ICP_QAT_SUOF_OBJ_ID_LEN]; + u64 offset; + u64 size; +}; + +struct icp_qat_suof_strtable { + unsigned int tab_length; + unsigned int strings; +}; + +struct icp_qat_suof_objhdr { + unsigned int img_length; + unsigned int reserved; +}; + +struct icp_qat_mof_file_hdr { + unsigned int file_id; + unsigned int checksum; + char min_ver; + char maj_ver; + unsigned short reserved; + unsigned short max_chunks; + unsigned short num_chunks; +}; + +struct icp_qat_mof_chunkhdr { + char chunk_id[ICP_QAT_MOF_OBJ_ID_LEN]; + u64 offset; + u64 size; +}; + +struct icp_qat_mof_str_table { + unsigned int tab_len; + unsigned int strings; +}; + +struct icp_qat_mof_obj_hdr { + unsigned short max_chunks; + unsigned short num_chunks; + unsigned int reserved; +}; + +struct icp_qat_mof_obj_chunkhdr { + char chunk_id[ICP_QAT_MOF_OBJ_CHUNKID_LEN]; + u64 offset; + u64 size; + unsigned int name; + unsigned int reserved; +}; + +struct icp_qat_mof_objhdr { + char *obj_name; + char *obj_buf; + unsigned int obj_size; +}; + +struct icp_qat_mof_table { + unsigned int num_objs; + struct icp_qat_mof_objhdr *obj_hdr; +}; + +struct icp_qat_mof_handle { + unsigned int file_id; + unsigned int checksum; + char min_ver; + char maj_ver; + char *mof_buf; + u32 mof_size; + char *sym_str; + unsigned int sym_size; + char *uobjs_hdr; + char *sobjs_hdr; + struct icp_qat_mof_table obj_table; +}; +#endif diff --git a/drivers/crypto/intel/qat/qat_common/qat_algs.c b/drivers/crypto/intel/qat/qat_common/qat_algs.c new file mode 100644 index 000000000000..7f638a62e3ad --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_algs.c @@ -0,0 +1,1348 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/module.h> +#include <linux/slab.h> +#include <linux/crypto.h> +#include <crypto/internal/aead.h> +#include <crypto/internal/cipher.h> +#include <crypto/internal/skcipher.h> +#include <crypto/aes.h> +#include <crypto/sha1.h> +#include <crypto/sha2.h> +#include <crypto/algapi.h> +#include <crypto/authenc.h> +#include <crypto/scatterwalk.h> +#include <crypto/xts.h> +#include <linux/dma-mapping.h> +#include "adf_accel_devices.h" +#include "qat_algs_send.h" +#include "adf_common_drv.h" +#include "qat_crypto.h" +#include "icp_qat_hw.h" +#include "icp_qat_fw.h" +#include "icp_qat_fw_la.h" +#include "qat_bl.h" + +#define QAT_AES_HW_CONFIG_ENC(alg, mode) \ + ICP_QAT_HW_CIPHER_CONFIG_BUILD(mode, alg, \ + ICP_QAT_HW_CIPHER_NO_CONVERT, \ + ICP_QAT_HW_CIPHER_ENCRYPT) + +#define QAT_AES_HW_CONFIG_DEC(alg, mode) \ + ICP_QAT_HW_CIPHER_CONFIG_BUILD(mode, alg, \ + ICP_QAT_HW_CIPHER_KEY_CONVERT, \ + ICP_QAT_HW_CIPHER_DECRYPT) + +#define QAT_AES_HW_CONFIG_DEC_NO_CONV(alg, mode) \ + ICP_QAT_HW_CIPHER_CONFIG_BUILD(mode, alg, \ + ICP_QAT_HW_CIPHER_NO_CONVERT, \ + ICP_QAT_HW_CIPHER_DECRYPT) + +#define HW_CAP_AES_V2(accel_dev) \ + (GET_HW_DATA(accel_dev)->accel_capabilities_mask & \ + ICP_ACCEL_CAPABILITIES_AES_V2) + +static DEFINE_MUTEX(algs_lock); +static unsigned int active_devs; + +/* Common content descriptor */ +struct qat_alg_cd { + union { + struct qat_enc { /* Encrypt content desc */ + struct icp_qat_hw_cipher_algo_blk cipher; + struct icp_qat_hw_auth_algo_blk hash; + } qat_enc_cd; + struct qat_dec { /* Decrypt content desc */ + struct icp_qat_hw_auth_algo_blk hash; + struct icp_qat_hw_cipher_algo_blk cipher; + } qat_dec_cd; + }; +} __aligned(64); + +struct qat_alg_aead_ctx { + struct qat_alg_cd *enc_cd; + struct qat_alg_cd *dec_cd; + dma_addr_t enc_cd_paddr; + dma_addr_t dec_cd_paddr; + struct icp_qat_fw_la_bulk_req enc_fw_req; + struct icp_qat_fw_la_bulk_req dec_fw_req; + enum icp_qat_hw_auth_algo qat_hash_alg; + unsigned int hash_digestsize; + unsigned int hash_blocksize; + struct qat_crypto_instance *inst; +}; + +struct qat_alg_skcipher_ctx { + struct icp_qat_hw_cipher_algo_blk *enc_cd; + struct icp_qat_hw_cipher_algo_blk *dec_cd; + dma_addr_t enc_cd_paddr; + dma_addr_t dec_cd_paddr; + struct icp_qat_fw_la_bulk_req enc_fw_req; + struct icp_qat_fw_la_bulk_req dec_fw_req; + struct qat_crypto_instance *inst; + struct crypto_skcipher *ftfm; + struct crypto_cipher *tweak; + bool fallback; + int mode; +}; + +static int qat_alg_do_precomputes(struct icp_qat_hw_auth_algo_blk *hash, + struct qat_alg_aead_ctx *ctx, + const u8 *auth_key, + unsigned int auth_keylen) +{ + switch (ctx->qat_hash_alg) { + case ICP_QAT_HW_AUTH_ALGO_SHA1: { + struct hmac_sha1_key key; + __be32 *istate = (__be32 *)hash->sha.state1; + __be32 *ostate = (__be32 *)(hash->sha.state1 + + round_up(sizeof(key.istate.h), 8)); + + hmac_sha1_preparekey(&key, auth_key, auth_keylen); + for (int i = 0; i < ARRAY_SIZE(key.istate.h); i++) { + istate[i] = cpu_to_be32(key.istate.h[i]); + ostate[i] = cpu_to_be32(key.ostate.h[i]); + } + memzero_explicit(&key, sizeof(key)); + return 0; + } + case ICP_QAT_HW_AUTH_ALGO_SHA256: { + struct hmac_sha256_key key; + __be32 *istate = (__be32 *)hash->sha.state1; + __be32 *ostate = (__be32 *)(hash->sha.state1 + + sizeof(key.key.istate.h)); + + hmac_sha256_preparekey(&key, auth_key, auth_keylen); + for (int i = 0; i < ARRAY_SIZE(key.key.istate.h); i++) { + istate[i] = cpu_to_be32(key.key.istate.h[i]); + ostate[i] = cpu_to_be32(key.key.ostate.h[i]); + } + memzero_explicit(&key, sizeof(key)); + return 0; + } + case ICP_QAT_HW_AUTH_ALGO_SHA512: { + struct hmac_sha512_key key; + __be64 *istate = (__be64 *)hash->sha.state1; + __be64 *ostate = (__be64 *)(hash->sha.state1 + + sizeof(key.key.istate.h)); + + hmac_sha512_preparekey(&key, auth_key, auth_keylen); + for (int i = 0; i < ARRAY_SIZE(key.key.istate.h); i++) { + istate[i] = cpu_to_be64(key.key.istate.h[i]); + ostate[i] = cpu_to_be64(key.key.ostate.h[i]); + } + memzero_explicit(&key, sizeof(key)); + return 0; + } + default: + return -EFAULT; + } +} + +static void qat_alg_init_common_hdr(struct icp_qat_fw_comn_req_hdr *header) +{ + header->hdr_flags = + ICP_QAT_FW_COMN_HDR_FLAGS_BUILD(ICP_QAT_FW_COMN_REQ_FLAG_SET); + header->service_type = ICP_QAT_FW_COMN_REQ_CPM_FW_LA; + header->comn_req_flags = + ICP_QAT_FW_COMN_FLAGS_BUILD(QAT_COMN_CD_FLD_TYPE_64BIT_ADR, + QAT_COMN_PTR_TYPE_SGL); + ICP_QAT_FW_LA_PARTIAL_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_PARTIAL_NONE); + ICP_QAT_FW_LA_CIPH_IV_FLD_FLAG_SET(header->serv_specif_flags, + ICP_QAT_FW_CIPH_IV_16BYTE_DATA); + ICP_QAT_FW_LA_PROTO_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_NO_PROTO); + ICP_QAT_FW_LA_UPDATE_STATE_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_NO_UPDATE_STATE); +} + +static int qat_alg_aead_init_enc_session(struct crypto_aead *aead_tfm, + int alg, + struct crypto_authenc_keys *keys, + int mode) +{ + struct qat_alg_aead_ctx *ctx = crypto_aead_ctx(aead_tfm); + unsigned int digestsize = crypto_aead_authsize(aead_tfm); + struct qat_enc *enc_ctx = &ctx->enc_cd->qat_enc_cd; + struct icp_qat_hw_cipher_algo_blk *cipher = &enc_ctx->cipher; + struct icp_qat_hw_auth_algo_blk *hash = + (struct icp_qat_hw_auth_algo_blk *)((char *)enc_ctx + + sizeof(struct icp_qat_hw_auth_setup) + keys->enckeylen); + struct icp_qat_fw_la_bulk_req *req_tmpl = &ctx->enc_fw_req; + struct icp_qat_fw_comn_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + void *ptr = &req_tmpl->cd_ctrl; + struct icp_qat_fw_cipher_cd_ctrl_hdr *cipher_cd_ctrl = ptr; + struct icp_qat_fw_auth_cd_ctrl_hdr *hash_cd_ctrl = ptr; + + /* CD setup */ + cipher->aes.cipher_config.val = QAT_AES_HW_CONFIG_ENC(alg, mode); + memcpy(cipher->aes.key, keys->enckey, keys->enckeylen); + hash->sha.inner_setup.auth_config.config = + ICP_QAT_HW_AUTH_CONFIG_BUILD(ICP_QAT_HW_AUTH_MODE1, + ctx->qat_hash_alg, digestsize); + hash->sha.inner_setup.auth_counter.counter = + cpu_to_be32(ctx->hash_blocksize); + + if (qat_alg_do_precomputes(hash, ctx, keys->authkey, keys->authkeylen)) + return -EFAULT; + + /* Request setup */ + qat_alg_init_common_hdr(header); + header->service_cmd_id = ICP_QAT_FW_LA_CMD_CIPHER_HASH; + ICP_QAT_FW_LA_DIGEST_IN_BUFFER_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_DIGEST_IN_BUFFER); + ICP_QAT_FW_LA_RET_AUTH_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_RET_AUTH_RES); + ICP_QAT_FW_LA_CMP_AUTH_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_NO_CMP_AUTH_RES); + cd_pars->u.s.content_desc_addr = ctx->enc_cd_paddr; + cd_pars->u.s.content_desc_params_sz = sizeof(struct qat_alg_cd) >> 3; + + /* Cipher CD config setup */ + cipher_cd_ctrl->cipher_key_sz = keys->enckeylen >> 3; + cipher_cd_ctrl->cipher_state_sz = AES_BLOCK_SIZE >> 3; + cipher_cd_ctrl->cipher_cfg_offset = 0; + ICP_QAT_FW_COMN_CURR_ID_SET(cipher_cd_ctrl, ICP_QAT_FW_SLICE_CIPHER); + ICP_QAT_FW_COMN_NEXT_ID_SET(cipher_cd_ctrl, ICP_QAT_FW_SLICE_AUTH); + /* Auth CD config setup */ + hash_cd_ctrl->hash_cfg_offset = ((char *)hash - (char *)cipher) >> 3; + hash_cd_ctrl->hash_flags = ICP_QAT_FW_AUTH_HDR_FLAG_NO_NESTED; + hash_cd_ctrl->inner_res_sz = digestsize; + hash_cd_ctrl->final_sz = digestsize; + + switch (ctx->qat_hash_alg) { + case ICP_QAT_HW_AUTH_ALGO_SHA1: + hash_cd_ctrl->inner_state1_sz = + round_up(ICP_QAT_HW_SHA1_STATE1_SZ, 8); + hash_cd_ctrl->inner_state2_sz = + round_up(ICP_QAT_HW_SHA1_STATE2_SZ, 8); + break; + case ICP_QAT_HW_AUTH_ALGO_SHA256: + hash_cd_ctrl->inner_state1_sz = ICP_QAT_HW_SHA256_STATE1_SZ; + hash_cd_ctrl->inner_state2_sz = ICP_QAT_HW_SHA256_STATE2_SZ; + break; + case ICP_QAT_HW_AUTH_ALGO_SHA512: + hash_cd_ctrl->inner_state1_sz = ICP_QAT_HW_SHA512_STATE1_SZ; + hash_cd_ctrl->inner_state2_sz = ICP_QAT_HW_SHA512_STATE2_SZ; + break; + default: + break; + } + hash_cd_ctrl->inner_state2_offset = hash_cd_ctrl->hash_cfg_offset + + ((sizeof(struct icp_qat_hw_auth_setup) + + round_up(hash_cd_ctrl->inner_state1_sz, 8)) >> 3); + ICP_QAT_FW_COMN_CURR_ID_SET(hash_cd_ctrl, ICP_QAT_FW_SLICE_AUTH); + ICP_QAT_FW_COMN_NEXT_ID_SET(hash_cd_ctrl, ICP_QAT_FW_SLICE_DRAM_WR); + return 0; +} + +static int qat_alg_aead_init_dec_session(struct crypto_aead *aead_tfm, + int alg, + struct crypto_authenc_keys *keys, + int mode) +{ + struct qat_alg_aead_ctx *ctx = crypto_aead_ctx(aead_tfm); + unsigned int digestsize = crypto_aead_authsize(aead_tfm); + struct qat_dec *dec_ctx = &ctx->dec_cd->qat_dec_cd; + struct icp_qat_hw_auth_algo_blk *hash = &dec_ctx->hash; + struct icp_qat_hw_cipher_algo_blk *cipher = + (struct icp_qat_hw_cipher_algo_blk *)((char *)dec_ctx + + sizeof(struct icp_qat_hw_auth_setup) + + roundup(ctx->hash_digestsize, 8) * 2); + struct icp_qat_fw_la_bulk_req *req_tmpl = &ctx->dec_fw_req; + struct icp_qat_fw_comn_req_hdr_cd_pars *cd_pars = &req_tmpl->cd_pars; + struct icp_qat_fw_comn_req_hdr *header = &req_tmpl->comn_hdr; + void *ptr = &req_tmpl->cd_ctrl; + struct icp_qat_fw_cipher_cd_ctrl_hdr *cipher_cd_ctrl = ptr; + struct icp_qat_fw_auth_cd_ctrl_hdr *hash_cd_ctrl = ptr; + struct icp_qat_fw_la_auth_req_params *auth_param = + (struct icp_qat_fw_la_auth_req_params *) + ((char *)&req_tmpl->serv_specif_rqpars + + sizeof(struct icp_qat_fw_la_cipher_req_params)); + + /* CD setup */ + cipher->aes.cipher_config.val = QAT_AES_HW_CONFIG_DEC(alg, mode); + memcpy(cipher->aes.key, keys->enckey, keys->enckeylen); + hash->sha.inner_setup.auth_config.config = + ICP_QAT_HW_AUTH_CONFIG_BUILD(ICP_QAT_HW_AUTH_MODE1, + ctx->qat_hash_alg, + digestsize); + hash->sha.inner_setup.auth_counter.counter = + cpu_to_be32(ctx->hash_blocksize); + + if (qat_alg_do_precomputes(hash, ctx, keys->authkey, keys->authkeylen)) + return -EFAULT; + + /* Request setup */ + qat_alg_init_common_hdr(header); + header->service_cmd_id = ICP_QAT_FW_LA_CMD_HASH_CIPHER; + ICP_QAT_FW_LA_DIGEST_IN_BUFFER_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_DIGEST_IN_BUFFER); + ICP_QAT_FW_LA_RET_AUTH_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_NO_RET_AUTH_RES); + ICP_QAT_FW_LA_CMP_AUTH_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_CMP_AUTH_RES); + cd_pars->u.s.content_desc_addr = ctx->dec_cd_paddr; + cd_pars->u.s.content_desc_params_sz = sizeof(struct qat_alg_cd) >> 3; + + /* Cipher CD config setup */ + cipher_cd_ctrl->cipher_key_sz = keys->enckeylen >> 3; + cipher_cd_ctrl->cipher_state_sz = AES_BLOCK_SIZE >> 3; + cipher_cd_ctrl->cipher_cfg_offset = + (sizeof(struct icp_qat_hw_auth_setup) + + roundup(ctx->hash_digestsize, 8) * 2) >> 3; + ICP_QAT_FW_COMN_CURR_ID_SET(cipher_cd_ctrl, ICP_QAT_FW_SLICE_CIPHER); + ICP_QAT_FW_COMN_NEXT_ID_SET(cipher_cd_ctrl, ICP_QAT_FW_SLICE_DRAM_WR); + + /* Auth CD config setup */ + hash_cd_ctrl->hash_cfg_offset = 0; + hash_cd_ctrl->hash_flags = ICP_QAT_FW_AUTH_HDR_FLAG_NO_NESTED; + hash_cd_ctrl->inner_res_sz = digestsize; + hash_cd_ctrl->final_sz = digestsize; + + switch (ctx->qat_hash_alg) { + case ICP_QAT_HW_AUTH_ALGO_SHA1: + hash_cd_ctrl->inner_state1_sz = + round_up(ICP_QAT_HW_SHA1_STATE1_SZ, 8); + hash_cd_ctrl->inner_state2_sz = + round_up(ICP_QAT_HW_SHA1_STATE2_SZ, 8); + break; + case ICP_QAT_HW_AUTH_ALGO_SHA256: + hash_cd_ctrl->inner_state1_sz = ICP_QAT_HW_SHA256_STATE1_SZ; + hash_cd_ctrl->inner_state2_sz = ICP_QAT_HW_SHA256_STATE2_SZ; + break; + case ICP_QAT_HW_AUTH_ALGO_SHA512: + hash_cd_ctrl->inner_state1_sz = ICP_QAT_HW_SHA512_STATE1_SZ; + hash_cd_ctrl->inner_state2_sz = ICP_QAT_HW_SHA512_STATE2_SZ; + break; + default: + break; + } + + hash_cd_ctrl->inner_state2_offset = hash_cd_ctrl->hash_cfg_offset + + ((sizeof(struct icp_qat_hw_auth_setup) + + round_up(hash_cd_ctrl->inner_state1_sz, 8)) >> 3); + auth_param->auth_res_sz = digestsize; + ICP_QAT_FW_COMN_CURR_ID_SET(hash_cd_ctrl, ICP_QAT_FW_SLICE_AUTH); + ICP_QAT_FW_COMN_NEXT_ID_SET(hash_cd_ctrl, ICP_QAT_FW_SLICE_CIPHER); + return 0; +} + +static void qat_alg_skcipher_init_com(struct qat_alg_skcipher_ctx *ctx, + struct icp_qat_fw_la_bulk_req *req, + struct icp_qat_hw_cipher_algo_blk *cd, + const u8 *key, unsigned int keylen) +{ + struct icp_qat_fw_comn_req_hdr_cd_pars *cd_pars = &req->cd_pars; + struct icp_qat_fw_comn_req_hdr *header = &req->comn_hdr; + struct icp_qat_fw_cipher_cd_ctrl_hdr *cd_ctrl = (void *)&req->cd_ctrl; + bool aes_v2_capable = HW_CAP_AES_V2(ctx->inst->accel_dev); + int mode = ctx->mode; + + qat_alg_init_common_hdr(header); + header->service_cmd_id = ICP_QAT_FW_LA_CMD_CIPHER; + cd_pars->u.s.content_desc_params_sz = + sizeof(struct icp_qat_hw_cipher_algo_blk) >> 3; + + if (aes_v2_capable && mode == ICP_QAT_HW_CIPHER_XTS_MODE) { + ICP_QAT_FW_LA_SLICE_TYPE_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_USE_UCS_SLICE_TYPE); + + /* Store both XTS keys in CD, only the first key is sent + * to the HW, the second key is used for tweak calculation + */ + memcpy(cd->ucs_aes.key, key, keylen); + keylen = keylen / 2; + } else if (aes_v2_capable && mode == ICP_QAT_HW_CIPHER_CTR_MODE) { + ICP_QAT_FW_LA_SLICE_TYPE_SET(header->serv_specif_flags, + ICP_QAT_FW_LA_USE_UCS_SLICE_TYPE); + memcpy(cd->ucs_aes.key, key, keylen); + keylen = round_up(keylen, 16); + } else { + memcpy(cd->aes.key, key, keylen); + } + + /* Cipher CD config setup */ + cd_ctrl->cipher_key_sz = keylen >> 3; + cd_ctrl->cipher_state_sz = AES_BLOCK_SIZE >> 3; + cd_ctrl->cipher_cfg_offset = 0; + ICP_QAT_FW_COMN_CURR_ID_SET(cd_ctrl, ICP_QAT_FW_SLICE_CIPHER); + ICP_QAT_FW_COMN_NEXT_ID_SET(cd_ctrl, ICP_QAT_FW_SLICE_DRAM_WR); +} + +static void qat_alg_skcipher_init_enc(struct qat_alg_skcipher_ctx *ctx, + int alg, const u8 *key, + unsigned int keylen, int mode) +{ + struct icp_qat_hw_cipher_algo_blk *enc_cd = ctx->enc_cd; + struct icp_qat_fw_la_bulk_req *req = &ctx->enc_fw_req; + struct icp_qat_fw_comn_req_hdr_cd_pars *cd_pars = &req->cd_pars; + + qat_alg_skcipher_init_com(ctx, req, enc_cd, key, keylen); + cd_pars->u.s.content_desc_addr = ctx->enc_cd_paddr; + enc_cd->aes.cipher_config.val = QAT_AES_HW_CONFIG_ENC(alg, mode); +} + +static void qat_alg_xts_reverse_key(const u8 *key_forward, unsigned int keylen, + u8 *key_reverse) +{ + struct crypto_aes_ctx aes_expanded; + int nrounds; + u8 *key; + + aes_expandkey(&aes_expanded, key_forward, keylen); + if (keylen == AES_KEYSIZE_128) { + nrounds = 10; + key = (u8 *)aes_expanded.key_enc + (AES_BLOCK_SIZE * nrounds); + memcpy(key_reverse, key, AES_BLOCK_SIZE); + } else { + /* AES_KEYSIZE_256 */ + nrounds = 14; + key = (u8 *)aes_expanded.key_enc + (AES_BLOCK_SIZE * nrounds); + memcpy(key_reverse, key, AES_BLOCK_SIZE); + memcpy(key_reverse + AES_BLOCK_SIZE, key - AES_BLOCK_SIZE, + AES_BLOCK_SIZE); + } +} + +static void qat_alg_skcipher_init_dec(struct qat_alg_skcipher_ctx *ctx, + int alg, const u8 *key, + unsigned int keylen, int mode) +{ + struct icp_qat_hw_cipher_algo_blk *dec_cd = ctx->dec_cd; + struct icp_qat_fw_la_bulk_req *req = &ctx->dec_fw_req; + struct icp_qat_fw_comn_req_hdr_cd_pars *cd_pars = &req->cd_pars; + bool aes_v2_capable = HW_CAP_AES_V2(ctx->inst->accel_dev); + + qat_alg_skcipher_init_com(ctx, req, dec_cd, key, keylen); + cd_pars->u.s.content_desc_addr = ctx->dec_cd_paddr; + + if (aes_v2_capable && mode == ICP_QAT_HW_CIPHER_XTS_MODE) { + /* Key reversing not supported, set no convert */ + dec_cd->aes.cipher_config.val = + QAT_AES_HW_CONFIG_DEC_NO_CONV(alg, mode); + + /* In-place key reversal */ + qat_alg_xts_reverse_key(dec_cd->ucs_aes.key, keylen / 2, + dec_cd->ucs_aes.key); + } else if (mode != ICP_QAT_HW_CIPHER_CTR_MODE) { + dec_cd->aes.cipher_config.val = + QAT_AES_HW_CONFIG_DEC(alg, mode); + } else { + dec_cd->aes.cipher_config.val = + QAT_AES_HW_CONFIG_ENC(alg, mode); + } +} + +static int qat_alg_validate_key(int key_len, int *alg, int mode) +{ + if (mode != ICP_QAT_HW_CIPHER_XTS_MODE) { + switch (key_len) { + case AES_KEYSIZE_128: + *alg = ICP_QAT_HW_CIPHER_ALGO_AES128; + break; + case AES_KEYSIZE_192: + *alg = ICP_QAT_HW_CIPHER_ALGO_AES192; + break; + case AES_KEYSIZE_256: + *alg = ICP_QAT_HW_CIPHER_ALGO_AES256; + break; + default: + return -EINVAL; + } + } else { + switch (key_len) { + case AES_KEYSIZE_128 << 1: + *alg = ICP_QAT_HW_CIPHER_ALGO_AES128; + break; + case AES_KEYSIZE_256 << 1: + *alg = ICP_QAT_HW_CIPHER_ALGO_AES256; + break; + default: + return -EINVAL; + } + } + return 0; +} + +static int qat_alg_aead_init_sessions(struct crypto_aead *tfm, const u8 *key, + unsigned int keylen, int mode) +{ + struct crypto_authenc_keys keys; + int alg; + + if (crypto_authenc_extractkeys(&keys, key, keylen)) + goto bad_key; + + if (qat_alg_validate_key(keys.enckeylen, &alg, mode)) + goto bad_key; + + if (qat_alg_aead_init_enc_session(tfm, alg, &keys, mode)) + goto error; + + if (qat_alg_aead_init_dec_session(tfm, alg, &keys, mode)) + goto error; + + memzero_explicit(&keys, sizeof(keys)); + return 0; +bad_key: + memzero_explicit(&keys, sizeof(keys)); + return -EINVAL; +error: + memzero_explicit(&keys, sizeof(keys)); + return -EFAULT; +} + +static int qat_alg_skcipher_init_sessions(struct qat_alg_skcipher_ctx *ctx, + const u8 *key, + unsigned int keylen, + int mode) +{ + int alg; + + if (qat_alg_validate_key(keylen, &alg, mode)) + return -EINVAL; + + qat_alg_skcipher_init_enc(ctx, alg, key, keylen, mode); + qat_alg_skcipher_init_dec(ctx, alg, key, keylen, mode); + return 0; +} + +static int qat_alg_aead_rekey(struct crypto_aead *tfm, const u8 *key, + unsigned int keylen) +{ + struct qat_alg_aead_ctx *ctx = crypto_aead_ctx(tfm); + + memset(ctx->enc_cd, 0, sizeof(*ctx->enc_cd)); + memset(ctx->dec_cd, 0, sizeof(*ctx->dec_cd)); + memset(&ctx->enc_fw_req, 0, sizeof(ctx->enc_fw_req)); + memset(&ctx->dec_fw_req, 0, sizeof(ctx->dec_fw_req)); + + return qat_alg_aead_init_sessions(tfm, key, keylen, + ICP_QAT_HW_CIPHER_CBC_MODE); +} + +static int qat_alg_aead_newkey(struct crypto_aead *tfm, const u8 *key, + unsigned int keylen) +{ + struct qat_alg_aead_ctx *ctx = crypto_aead_ctx(tfm); + struct qat_crypto_instance *inst = NULL; + int node = numa_node_id(); + struct device *dev; + int ret; + + inst = qat_crypto_get_instance_node(node); + if (!inst) + return -EINVAL; + dev = &GET_DEV(inst->accel_dev); + ctx->inst = inst; + ctx->enc_cd = dma_alloc_coherent(dev, sizeof(*ctx->enc_cd), + &ctx->enc_cd_paddr, + GFP_ATOMIC); + if (!ctx->enc_cd) { + ret = -ENOMEM; + goto out_free_inst; + } + ctx->dec_cd = dma_alloc_coherent(dev, sizeof(*ctx->dec_cd), + &ctx->dec_cd_paddr, + GFP_ATOMIC); + if (!ctx->dec_cd) { + ret = -ENOMEM; + goto out_free_enc; + } + + ret = qat_alg_aead_init_sessions(tfm, key, keylen, + ICP_QAT_HW_CIPHER_CBC_MODE); + if (ret) + goto out_free_all; + + return 0; + +out_free_all: + memset(ctx->dec_cd, 0, sizeof(struct qat_alg_cd)); + dma_free_coherent(dev, sizeof(struct qat_alg_cd), + ctx->dec_cd, ctx->dec_cd_paddr); + ctx->dec_cd = NULL; +out_free_enc: + memset(ctx->enc_cd, 0, sizeof(struct qat_alg_cd)); + dma_free_coherent(dev, sizeof(struct qat_alg_cd), + ctx->enc_cd, ctx->enc_cd_paddr); + ctx->enc_cd = NULL; +out_free_inst: + ctx->inst = NULL; + qat_crypto_put_instance(inst); + return ret; +} + +static int qat_alg_aead_setkey(struct crypto_aead *tfm, const u8 *key, + unsigned int keylen) +{ + struct qat_alg_aead_ctx *ctx = crypto_aead_ctx(tfm); + + if (ctx->enc_cd) + return qat_alg_aead_rekey(tfm, key, keylen); + else + return qat_alg_aead_newkey(tfm, key, keylen); +} + +static void qat_aead_alg_callback(struct icp_qat_fw_la_resp *qat_resp, + struct qat_crypto_request *qat_req) +{ + struct qat_alg_aead_ctx *ctx = qat_req->aead_ctx; + struct qat_crypto_instance *inst = ctx->inst; + struct aead_request *areq = qat_req->aead_req; + u8 stat_filed = qat_resp->comn_resp.comn_status; + int res = 0, qat_res = ICP_QAT_FW_COMN_RESP_CRYPTO_STAT_GET(stat_filed); + + qat_bl_free_bufl(inst->accel_dev, &qat_req->buf); + if (unlikely(qat_res != ICP_QAT_FW_COMN_STATUS_FLAG_OK)) + res = -EBADMSG; + aead_request_complete(areq, res); +} + +static void qat_alg_update_iv_ctr_mode(struct qat_crypto_request *qat_req) +{ + struct skcipher_request *sreq = qat_req->skcipher_req; + u64 iv_lo_prev; + u64 iv_lo; + u64 iv_hi; + + memcpy(qat_req->iv, sreq->iv, AES_BLOCK_SIZE); + + iv_lo = be64_to_cpu(qat_req->iv_lo); + iv_hi = be64_to_cpu(qat_req->iv_hi); + + iv_lo_prev = iv_lo; + iv_lo += DIV_ROUND_UP(sreq->cryptlen, AES_BLOCK_SIZE); + if (iv_lo < iv_lo_prev) + iv_hi++; + + qat_req->iv_lo = cpu_to_be64(iv_lo); + qat_req->iv_hi = cpu_to_be64(iv_hi); +} + +static void qat_alg_update_iv_cbc_mode(struct qat_crypto_request *qat_req) +{ + struct skcipher_request *sreq = qat_req->skcipher_req; + int offset = sreq->cryptlen - AES_BLOCK_SIZE; + struct scatterlist *sgl; + + if (qat_req->encryption) + sgl = sreq->dst; + else + sgl = sreq->src; + + scatterwalk_map_and_copy(qat_req->iv, sgl, offset, AES_BLOCK_SIZE, 0); +} + +static void qat_alg_update_iv(struct qat_crypto_request *qat_req) +{ + struct qat_alg_skcipher_ctx *ctx = qat_req->skcipher_ctx; + struct device *dev = &GET_DEV(ctx->inst->accel_dev); + + switch (ctx->mode) { + case ICP_QAT_HW_CIPHER_CTR_MODE: + qat_alg_update_iv_ctr_mode(qat_req); + break; + case ICP_QAT_HW_CIPHER_CBC_MODE: + qat_alg_update_iv_cbc_mode(qat_req); + break; + case ICP_QAT_HW_CIPHER_XTS_MODE: + break; + default: + dev_warn(dev, "Unsupported IV update for cipher mode %d\n", + ctx->mode); + } +} + +static void qat_skcipher_alg_callback(struct icp_qat_fw_la_resp *qat_resp, + struct qat_crypto_request *qat_req) +{ + struct qat_alg_skcipher_ctx *ctx = qat_req->skcipher_ctx; + struct qat_crypto_instance *inst = ctx->inst; + struct skcipher_request *sreq = qat_req->skcipher_req; + u8 stat_filed = qat_resp->comn_resp.comn_status; + int res = 0, qat_res = ICP_QAT_FW_COMN_RESP_CRYPTO_STAT_GET(stat_filed); + + qat_bl_free_bufl(inst->accel_dev, &qat_req->buf); + if (unlikely(qat_res != ICP_QAT_FW_COMN_STATUS_FLAG_OK)) + res = -EINVAL; + + if (qat_req->encryption) + qat_alg_update_iv(qat_req); + + memcpy(sreq->iv, qat_req->iv, AES_BLOCK_SIZE); + + skcipher_request_complete(sreq, res); +} + +void qat_alg_callback(void *resp) +{ + struct icp_qat_fw_la_resp *qat_resp = resp; + struct qat_crypto_request *qat_req = + (void *)(__force long)qat_resp->opaque_data; + struct qat_instance_backlog *backlog = qat_req->alg_req.backlog; + + qat_req->cb(qat_resp, qat_req); + + qat_alg_send_backlog(backlog); +} + +static int qat_alg_send_sym_message(struct qat_crypto_request *qat_req, + struct qat_crypto_instance *inst, + struct crypto_async_request *base) +{ + struct qat_alg_req *alg_req = &qat_req->alg_req; + + alg_req->fw_req = (u32 *)&qat_req->req; + alg_req->tx_ring = inst->sym_tx; + alg_req->base = base; + alg_req->backlog = &inst->backlog; + + return qat_alg_send_message(alg_req); +} + +static int qat_alg_aead_dec(struct aead_request *areq) +{ + struct crypto_aead *aead_tfm = crypto_aead_reqtfm(areq); + struct crypto_tfm *tfm = crypto_aead_tfm(aead_tfm); + struct qat_alg_aead_ctx *ctx = crypto_tfm_ctx(tfm); + struct qat_crypto_request *qat_req = aead_request_ctx(areq); + struct icp_qat_fw_la_cipher_req_params *cipher_param; + struct icp_qat_fw_la_auth_req_params *auth_param; + struct icp_qat_fw_la_bulk_req *msg; + int digst_size = crypto_aead_authsize(aead_tfm); + gfp_t f = qat_algs_alloc_flags(&areq->base); + int ret; + u32 cipher_len; + + cipher_len = areq->cryptlen - digst_size; + if (cipher_len % AES_BLOCK_SIZE != 0) + return -EINVAL; + + ret = qat_bl_sgl_to_bufl(ctx->inst->accel_dev, areq->src, areq->dst, + &qat_req->buf, NULL, f); + if (unlikely(ret)) + return ret; + + msg = &qat_req->req; + *msg = ctx->dec_fw_req; + qat_req->aead_ctx = ctx; + qat_req->aead_req = areq; + qat_req->cb = qat_aead_alg_callback; + qat_req->req.comn_mid.opaque_data = (u64)(__force long)qat_req; + qat_req->req.comn_mid.src_data_addr = qat_req->buf.blp; + qat_req->req.comn_mid.dest_data_addr = qat_req->buf.bloutp; + cipher_param = (void *)&qat_req->req.serv_specif_rqpars; + cipher_param->cipher_length = cipher_len; + cipher_param->cipher_offset = areq->assoclen; + memcpy(cipher_param->u.cipher_IV_array, areq->iv, AES_BLOCK_SIZE); + auth_param = (void *)((u8 *)cipher_param + sizeof(*cipher_param)); + auth_param->auth_off = 0; + auth_param->auth_len = areq->assoclen + cipher_param->cipher_length; + + ret = qat_alg_send_sym_message(qat_req, ctx->inst, &areq->base); + if (ret == -ENOSPC) + qat_bl_free_bufl(ctx->inst->accel_dev, &qat_req->buf); + + return ret; +} + +static int qat_alg_aead_enc(struct aead_request *areq) +{ + struct crypto_aead *aead_tfm = crypto_aead_reqtfm(areq); + struct crypto_tfm *tfm = crypto_aead_tfm(aead_tfm); + struct qat_alg_aead_ctx *ctx = crypto_tfm_ctx(tfm); + struct qat_crypto_request *qat_req = aead_request_ctx(areq); + struct icp_qat_fw_la_cipher_req_params *cipher_param; + struct icp_qat_fw_la_auth_req_params *auth_param; + gfp_t f = qat_algs_alloc_flags(&areq->base); + struct icp_qat_fw_la_bulk_req *msg; + u8 *iv = areq->iv; + int ret; + + if (areq->cryptlen % AES_BLOCK_SIZE != 0) + return -EINVAL; + + ret = qat_bl_sgl_to_bufl(ctx->inst->accel_dev, areq->src, areq->dst, + &qat_req->buf, NULL, f); + if (unlikely(ret)) + return ret; + + msg = &qat_req->req; + *msg = ctx->enc_fw_req; + qat_req->aead_ctx = ctx; + qat_req->aead_req = areq; + qat_req->cb = qat_aead_alg_callback; + qat_req->req.comn_mid.opaque_data = (u64)(__force long)qat_req; + qat_req->req.comn_mid.src_data_addr = qat_req->buf.blp; + qat_req->req.comn_mid.dest_data_addr = qat_req->buf.bloutp; + cipher_param = (void *)&qat_req->req.serv_specif_rqpars; + auth_param = (void *)((u8 *)cipher_param + sizeof(*cipher_param)); + + memcpy(cipher_param->u.cipher_IV_array, iv, AES_BLOCK_SIZE); + cipher_param->cipher_length = areq->cryptlen; + cipher_param->cipher_offset = areq->assoclen; + + auth_param->auth_off = 0; + auth_param->auth_len = areq->assoclen + areq->cryptlen; + + ret = qat_alg_send_sym_message(qat_req, ctx->inst, &areq->base); + if (ret == -ENOSPC) + qat_bl_free_bufl(ctx->inst->accel_dev, &qat_req->buf); + + return ret; +} + +static int qat_alg_skcipher_rekey(struct qat_alg_skcipher_ctx *ctx, + const u8 *key, unsigned int keylen, + int mode) +{ + memset(ctx->enc_cd, 0, sizeof(*ctx->enc_cd)); + memset(ctx->dec_cd, 0, sizeof(*ctx->dec_cd)); + memset(&ctx->enc_fw_req, 0, sizeof(ctx->enc_fw_req)); + memset(&ctx->dec_fw_req, 0, sizeof(ctx->dec_fw_req)); + + return qat_alg_skcipher_init_sessions(ctx, key, keylen, mode); +} + +static int qat_alg_skcipher_newkey(struct qat_alg_skcipher_ctx *ctx, + const u8 *key, unsigned int keylen, + int mode) +{ + struct qat_crypto_instance *inst = NULL; + struct device *dev; + int node = numa_node_id(); + int ret; + + inst = qat_crypto_get_instance_node(node); + if (!inst) + return -EINVAL; + dev = &GET_DEV(inst->accel_dev); + ctx->inst = inst; + ctx->enc_cd = dma_alloc_coherent(dev, sizeof(*ctx->enc_cd), + &ctx->enc_cd_paddr, + GFP_ATOMIC); + if (!ctx->enc_cd) { + ret = -ENOMEM; + goto out_free_instance; + } + ctx->dec_cd = dma_alloc_coherent(dev, sizeof(*ctx->dec_cd), + &ctx->dec_cd_paddr, + GFP_ATOMIC); + if (!ctx->dec_cd) { + ret = -ENOMEM; + goto out_free_enc; + } + + ret = qat_alg_skcipher_init_sessions(ctx, key, keylen, mode); + if (ret) + goto out_free_all; + + return 0; + +out_free_all: + memset(ctx->dec_cd, 0, sizeof(*ctx->dec_cd)); + dma_free_coherent(dev, sizeof(*ctx->dec_cd), + ctx->dec_cd, ctx->dec_cd_paddr); + ctx->dec_cd = NULL; +out_free_enc: + memset(ctx->enc_cd, 0, sizeof(*ctx->enc_cd)); + dma_free_coherent(dev, sizeof(*ctx->enc_cd), + ctx->enc_cd, ctx->enc_cd_paddr); + ctx->enc_cd = NULL; +out_free_instance: + ctx->inst = NULL; + qat_crypto_put_instance(inst); + return ret; +} + +static int qat_alg_skcipher_setkey(struct crypto_skcipher *tfm, + const u8 *key, unsigned int keylen, + int mode) +{ + struct qat_alg_skcipher_ctx *ctx = crypto_skcipher_ctx(tfm); + + ctx->mode = mode; + + if (ctx->enc_cd) + return qat_alg_skcipher_rekey(ctx, key, keylen, mode); + else + return qat_alg_skcipher_newkey(ctx, key, keylen, mode); +} + +static int qat_alg_skcipher_cbc_setkey(struct crypto_skcipher *tfm, + const u8 *key, unsigned int keylen) +{ + return qat_alg_skcipher_setkey(tfm, key, keylen, + ICP_QAT_HW_CIPHER_CBC_MODE); +} + +static int qat_alg_skcipher_ctr_setkey(struct crypto_skcipher *tfm, + const u8 *key, unsigned int keylen) +{ + return qat_alg_skcipher_setkey(tfm, key, keylen, + ICP_QAT_HW_CIPHER_CTR_MODE); +} + +static int qat_alg_skcipher_xts_setkey(struct crypto_skcipher *tfm, + const u8 *key, unsigned int keylen) +{ + struct qat_alg_skcipher_ctx *ctx = crypto_skcipher_ctx(tfm); + int ret; + + ret = xts_verify_key(tfm, key, keylen); + if (ret) + return ret; + + if (keylen >> 1 == AES_KEYSIZE_192) { + ret = crypto_skcipher_setkey(ctx->ftfm, key, keylen); + if (ret) + return ret; + + ctx->fallback = true; + + return 0; + } + + ctx->fallback = false; + + ret = qat_alg_skcipher_setkey(tfm, key, keylen, + ICP_QAT_HW_CIPHER_XTS_MODE); + if (ret) + return ret; + + if (HW_CAP_AES_V2(ctx->inst->accel_dev)) + ret = crypto_cipher_setkey(ctx->tweak, key + (keylen / 2), + keylen / 2); + + return ret; +} + +static void qat_alg_set_req_iv(struct qat_crypto_request *qat_req) +{ + struct icp_qat_fw_la_cipher_req_params *cipher_param; + struct qat_alg_skcipher_ctx *ctx = qat_req->skcipher_ctx; + bool aes_v2_capable = HW_CAP_AES_V2(ctx->inst->accel_dev); + u8 *iv = qat_req->skcipher_req->iv; + + cipher_param = (void *)&qat_req->req.serv_specif_rqpars; + + if (aes_v2_capable && ctx->mode == ICP_QAT_HW_CIPHER_XTS_MODE) + crypto_cipher_encrypt_one(ctx->tweak, + (u8 *)cipher_param->u.cipher_IV_array, + iv); + else + memcpy(cipher_param->u.cipher_IV_array, iv, AES_BLOCK_SIZE); +} + +static int qat_alg_skcipher_encrypt(struct skcipher_request *req) +{ + struct crypto_skcipher *stfm = crypto_skcipher_reqtfm(req); + struct crypto_tfm *tfm = crypto_skcipher_tfm(stfm); + struct qat_alg_skcipher_ctx *ctx = crypto_tfm_ctx(tfm); + struct qat_crypto_request *qat_req = skcipher_request_ctx(req); + struct icp_qat_fw_la_cipher_req_params *cipher_param; + gfp_t f = qat_algs_alloc_flags(&req->base); + struct icp_qat_fw_la_bulk_req *msg; + int ret; + + if (req->cryptlen == 0) + return 0; + + ret = qat_bl_sgl_to_bufl(ctx->inst->accel_dev, req->src, req->dst, + &qat_req->buf, NULL, f); + if (unlikely(ret)) + return ret; + + msg = &qat_req->req; + *msg = ctx->enc_fw_req; + qat_req->skcipher_ctx = ctx; + qat_req->skcipher_req = req; + qat_req->cb = qat_skcipher_alg_callback; + qat_req->req.comn_mid.opaque_data = (u64)(__force long)qat_req; + qat_req->req.comn_mid.src_data_addr = qat_req->buf.blp; + qat_req->req.comn_mid.dest_data_addr = qat_req->buf.bloutp; + qat_req->encryption = true; + cipher_param = (void *)&qat_req->req.serv_specif_rqpars; + cipher_param->cipher_length = req->cryptlen; + cipher_param->cipher_offset = 0; + + qat_alg_set_req_iv(qat_req); + + ret = qat_alg_send_sym_message(qat_req, ctx->inst, &req->base); + if (ret == -ENOSPC) + qat_bl_free_bufl(ctx->inst->accel_dev, &qat_req->buf); + + return ret; +} + +static int qat_alg_skcipher_blk_encrypt(struct skcipher_request *req) +{ + if (req->cryptlen % AES_BLOCK_SIZE != 0) + return -EINVAL; + + return qat_alg_skcipher_encrypt(req); +} + +static int qat_alg_skcipher_xts_encrypt(struct skcipher_request *req) +{ + struct crypto_skcipher *stfm = crypto_skcipher_reqtfm(req); + struct qat_alg_skcipher_ctx *ctx = crypto_skcipher_ctx(stfm); + struct skcipher_request *nreq = skcipher_request_ctx(req); + + if (req->cryptlen < XTS_BLOCK_SIZE) + return -EINVAL; + + if (ctx->fallback) { + memcpy(nreq, req, sizeof(*req)); + skcipher_request_set_tfm(nreq, ctx->ftfm); + return crypto_skcipher_encrypt(nreq); + } + + return qat_alg_skcipher_encrypt(req); +} + +static int qat_alg_skcipher_decrypt(struct skcipher_request *req) +{ + struct crypto_skcipher *stfm = crypto_skcipher_reqtfm(req); + struct crypto_tfm *tfm = crypto_skcipher_tfm(stfm); + struct qat_alg_skcipher_ctx *ctx = crypto_tfm_ctx(tfm); + struct qat_crypto_request *qat_req = skcipher_request_ctx(req); + struct icp_qat_fw_la_cipher_req_params *cipher_param; + gfp_t f = qat_algs_alloc_flags(&req->base); + struct icp_qat_fw_la_bulk_req *msg; + int ret; + + if (req->cryptlen == 0) + return 0; + + ret = qat_bl_sgl_to_bufl(ctx->inst->accel_dev, req->src, req->dst, + &qat_req->buf, NULL, f); + if (unlikely(ret)) + return ret; + + msg = &qat_req->req; + *msg = ctx->dec_fw_req; + qat_req->skcipher_ctx = ctx; + qat_req->skcipher_req = req; + qat_req->cb = qat_skcipher_alg_callback; + qat_req->req.comn_mid.opaque_data = (u64)(__force long)qat_req; + qat_req->req.comn_mid.src_data_addr = qat_req->buf.blp; + qat_req->req.comn_mid.dest_data_addr = qat_req->buf.bloutp; + qat_req->encryption = false; + cipher_param = (void *)&qat_req->req.serv_specif_rqpars; + cipher_param->cipher_length = req->cryptlen; + cipher_param->cipher_offset = 0; + + qat_alg_set_req_iv(qat_req); + qat_alg_update_iv(qat_req); + + ret = qat_alg_send_sym_message(qat_req, ctx->inst, &req->base); + if (ret == -ENOSPC) + qat_bl_free_bufl(ctx->inst->accel_dev, &qat_req->buf); + + return ret; +} + +static int qat_alg_skcipher_blk_decrypt(struct skcipher_request *req) +{ + if (req->cryptlen % AES_BLOCK_SIZE != 0) + return -EINVAL; + + return qat_alg_skcipher_decrypt(req); +} + +static int qat_alg_skcipher_xts_decrypt(struct skcipher_request *req) +{ + struct crypto_skcipher *stfm = crypto_skcipher_reqtfm(req); + struct qat_alg_skcipher_ctx *ctx = crypto_skcipher_ctx(stfm); + struct skcipher_request *nreq = skcipher_request_ctx(req); + + if (req->cryptlen < XTS_BLOCK_SIZE) + return -EINVAL; + + if (ctx->fallback) { + memcpy(nreq, req, sizeof(*req)); + skcipher_request_set_tfm(nreq, ctx->ftfm); + return crypto_skcipher_decrypt(nreq); + } + + return qat_alg_skcipher_decrypt(req); +} + +static int qat_alg_aead_init(struct crypto_aead *tfm, + enum icp_qat_hw_auth_algo hash_alg, + unsigned int hash_digestsize, + unsigned int hash_blocksize) +{ + struct qat_alg_aead_ctx *ctx = crypto_aead_ctx(tfm); + + ctx->qat_hash_alg = hash_alg; + ctx->hash_digestsize = hash_digestsize; + ctx->hash_blocksize = hash_blocksize; + crypto_aead_set_reqsize(tfm, sizeof(struct qat_crypto_request)); + return 0; +} + +static int qat_alg_aead_sha1_init(struct crypto_aead *tfm) +{ + return qat_alg_aead_init(tfm, ICP_QAT_HW_AUTH_ALGO_SHA1, + SHA1_DIGEST_SIZE, SHA1_BLOCK_SIZE); +} + +static int qat_alg_aead_sha256_init(struct crypto_aead *tfm) +{ + return qat_alg_aead_init(tfm, ICP_QAT_HW_AUTH_ALGO_SHA256, + SHA256_DIGEST_SIZE, SHA256_BLOCK_SIZE); +} + +static int qat_alg_aead_sha512_init(struct crypto_aead *tfm) +{ + return qat_alg_aead_init(tfm, ICP_QAT_HW_AUTH_ALGO_SHA512, + SHA512_DIGEST_SIZE, SHA512_BLOCK_SIZE); +} + +static void qat_alg_aead_exit(struct crypto_aead *tfm) +{ + struct qat_alg_aead_ctx *ctx = crypto_aead_ctx(tfm); + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev; + + if (!inst) + return; + + dev = &GET_DEV(inst->accel_dev); + if (ctx->enc_cd) { + memset(ctx->enc_cd, 0, sizeof(struct qat_alg_cd)); + dma_free_coherent(dev, sizeof(struct qat_alg_cd), + ctx->enc_cd, ctx->enc_cd_paddr); + } + if (ctx->dec_cd) { + memset(ctx->dec_cd, 0, sizeof(struct qat_alg_cd)); + dma_free_coherent(dev, sizeof(struct qat_alg_cd), + ctx->dec_cd, ctx->dec_cd_paddr); + } + qat_crypto_put_instance(inst); +} + +static int qat_alg_skcipher_init_tfm(struct crypto_skcipher *tfm) +{ + crypto_skcipher_set_reqsize(tfm, sizeof(struct qat_crypto_request)); + return 0; +} + +static int qat_alg_skcipher_init_xts_tfm(struct crypto_skcipher *tfm) +{ + struct qat_alg_skcipher_ctx *ctx = crypto_skcipher_ctx(tfm); + int reqsize; + + ctx->ftfm = crypto_alloc_skcipher("xts(aes)", 0, + CRYPTO_ALG_NEED_FALLBACK); + if (IS_ERR(ctx->ftfm)) + return PTR_ERR(ctx->ftfm); + + ctx->tweak = crypto_alloc_cipher("aes", 0, 0); + if (IS_ERR(ctx->tweak)) { + crypto_free_skcipher(ctx->ftfm); + return PTR_ERR(ctx->tweak); + } + + reqsize = max(sizeof(struct qat_crypto_request), + sizeof(struct skcipher_request) + + crypto_skcipher_reqsize(ctx->ftfm)); + crypto_skcipher_set_reqsize(tfm, reqsize); + + return 0; +} + +static void qat_alg_skcipher_exit_tfm(struct crypto_skcipher *tfm) +{ + struct qat_alg_skcipher_ctx *ctx = crypto_skcipher_ctx(tfm); + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev; + + if (!inst) + return; + + dev = &GET_DEV(inst->accel_dev); + if (ctx->enc_cd) { + memset(ctx->enc_cd, 0, + sizeof(struct icp_qat_hw_cipher_algo_blk)); + dma_free_coherent(dev, + sizeof(struct icp_qat_hw_cipher_algo_blk), + ctx->enc_cd, ctx->enc_cd_paddr); + } + if (ctx->dec_cd) { + memset(ctx->dec_cd, 0, + sizeof(struct icp_qat_hw_cipher_algo_blk)); + dma_free_coherent(dev, + sizeof(struct icp_qat_hw_cipher_algo_blk), + ctx->dec_cd, ctx->dec_cd_paddr); + } + qat_crypto_put_instance(inst); +} + +static void qat_alg_skcipher_exit_xts_tfm(struct crypto_skcipher *tfm) +{ + struct qat_alg_skcipher_ctx *ctx = crypto_skcipher_ctx(tfm); + + if (ctx->ftfm) + crypto_free_skcipher(ctx->ftfm); + + if (ctx->tweak) + crypto_free_cipher(ctx->tweak); + + qat_alg_skcipher_exit_tfm(tfm); +} + +static struct aead_alg qat_aeads[] = { { + .base = { + .cra_name = "authenc(hmac(sha1),cbc(aes))", + .cra_driver_name = "qat_aes_cbc_hmac_sha1", + .cra_priority = 100, + .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_ALLOCATES_MEMORY, + .cra_blocksize = AES_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct qat_alg_aead_ctx), + .cra_module = THIS_MODULE, + }, + .init = qat_alg_aead_sha1_init, + .exit = qat_alg_aead_exit, + .setkey = qat_alg_aead_setkey, + .decrypt = qat_alg_aead_dec, + .encrypt = qat_alg_aead_enc, + .ivsize = AES_BLOCK_SIZE, + .maxauthsize = SHA1_DIGEST_SIZE, +}, { + .base = { + .cra_name = "authenc(hmac(sha256),cbc(aes))", + .cra_driver_name = "qat_aes_cbc_hmac_sha256", + .cra_priority = 100, + .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_ALLOCATES_MEMORY, + .cra_blocksize = AES_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct qat_alg_aead_ctx), + .cra_module = THIS_MODULE, + }, + .init = qat_alg_aead_sha256_init, + .exit = qat_alg_aead_exit, + .setkey = qat_alg_aead_setkey, + .decrypt = qat_alg_aead_dec, + .encrypt = qat_alg_aead_enc, + .ivsize = AES_BLOCK_SIZE, + .maxauthsize = SHA256_DIGEST_SIZE, +}, { + .base = { + .cra_name = "authenc(hmac(sha512),cbc(aes))", + .cra_driver_name = "qat_aes_cbc_hmac_sha512", + .cra_priority = 100, + .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_ALLOCATES_MEMORY, + .cra_blocksize = AES_BLOCK_SIZE, + .cra_ctxsize = sizeof(struct qat_alg_aead_ctx), + .cra_module = THIS_MODULE, + }, + .init = qat_alg_aead_sha512_init, + .exit = qat_alg_aead_exit, + .setkey = qat_alg_aead_setkey, + .decrypt = qat_alg_aead_dec, + .encrypt = qat_alg_aead_enc, + .ivsize = AES_BLOCK_SIZE, + .maxauthsize = SHA512_DIGEST_SIZE, +} }; + +static struct skcipher_alg qat_skciphers[] = { { + .base.cra_name = "cbc(aes)", + .base.cra_driver_name = "qat_aes_cbc", + .base.cra_priority = 100, + .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_ALLOCATES_MEMORY, + .base.cra_blocksize = AES_BLOCK_SIZE, + .base.cra_ctxsize = sizeof(struct qat_alg_skcipher_ctx), + .base.cra_alignmask = 0, + .base.cra_module = THIS_MODULE, + + .init = qat_alg_skcipher_init_tfm, + .exit = qat_alg_skcipher_exit_tfm, + .setkey = qat_alg_skcipher_cbc_setkey, + .decrypt = qat_alg_skcipher_blk_decrypt, + .encrypt = qat_alg_skcipher_blk_encrypt, + .min_keysize = AES_MIN_KEY_SIZE, + .max_keysize = AES_MAX_KEY_SIZE, + .ivsize = AES_BLOCK_SIZE, +}, { + .base.cra_name = "ctr(aes)", + .base.cra_driver_name = "qat_aes_ctr", + .base.cra_priority = 100, + .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_ALLOCATES_MEMORY, + .base.cra_blocksize = 1, + .base.cra_ctxsize = sizeof(struct qat_alg_skcipher_ctx), + .base.cra_alignmask = 0, + .base.cra_module = THIS_MODULE, + + .init = qat_alg_skcipher_init_tfm, + .exit = qat_alg_skcipher_exit_tfm, + .setkey = qat_alg_skcipher_ctr_setkey, + .decrypt = qat_alg_skcipher_decrypt, + .encrypt = qat_alg_skcipher_encrypt, + .min_keysize = AES_MIN_KEY_SIZE, + .max_keysize = AES_MAX_KEY_SIZE, + .ivsize = AES_BLOCK_SIZE, +}, { + .base.cra_name = "xts(aes)", + .base.cra_driver_name = "qat_aes_xts", + .base.cra_priority = 100, + .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK | + CRYPTO_ALG_ALLOCATES_MEMORY, + .base.cra_blocksize = AES_BLOCK_SIZE, + .base.cra_ctxsize = sizeof(struct qat_alg_skcipher_ctx), + .base.cra_alignmask = 0, + .base.cra_module = THIS_MODULE, + + .init = qat_alg_skcipher_init_xts_tfm, + .exit = qat_alg_skcipher_exit_xts_tfm, + .setkey = qat_alg_skcipher_xts_setkey, + .decrypt = qat_alg_skcipher_xts_decrypt, + .encrypt = qat_alg_skcipher_xts_encrypt, + .min_keysize = 2 * AES_MIN_KEY_SIZE, + .max_keysize = 2 * AES_MAX_KEY_SIZE, + .ivsize = AES_BLOCK_SIZE, +} }; + +int qat_algs_register(void) +{ + int ret = 0; + + mutex_lock(&algs_lock); + if (++active_devs != 1) + goto unlock; + + ret = crypto_register_skciphers(qat_skciphers, + ARRAY_SIZE(qat_skciphers)); + if (ret) + goto unlock; + + ret = crypto_register_aeads(qat_aeads, ARRAY_SIZE(qat_aeads)); + if (ret) + goto unreg_algs; + +unlock: + mutex_unlock(&algs_lock); + return ret; + +unreg_algs: + crypto_unregister_skciphers(qat_skciphers, ARRAY_SIZE(qat_skciphers)); + goto unlock; +} + +void qat_algs_unregister(void) +{ + mutex_lock(&algs_lock); + if (--active_devs != 0) + goto unlock; + + crypto_unregister_aeads(qat_aeads, ARRAY_SIZE(qat_aeads)); + crypto_unregister_skciphers(qat_skciphers, ARRAY_SIZE(qat_skciphers)); + +unlock: + mutex_unlock(&algs_lock); +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_algs_send.c b/drivers/crypto/intel/qat/qat_common/qat_algs_send.c new file mode 100644 index 000000000000..b97b678823a9 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_algs_send.c @@ -0,0 +1,91 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2022 Intel Corporation */ +#include <crypto/algapi.h> +#include "adf_transport.h" +#include "qat_algs_send.h" +#include "qat_crypto.h" + +#define ADF_MAX_RETRIES 20 + +static int qat_alg_send_message_retry(struct qat_alg_req *req) +{ + int ret = 0, ctr = 0; + + do { + ret = adf_send_message(req->tx_ring, req->fw_req); + } while (ret == -EAGAIN && ctr++ < ADF_MAX_RETRIES); + + if (ret == -EAGAIN) + return -ENOSPC; + + return -EINPROGRESS; +} + +void qat_alg_send_backlog(struct qat_instance_backlog *backlog) +{ + struct qat_alg_req *req, *tmp; + + spin_lock_bh(&backlog->lock); + list_for_each_entry_safe(req, tmp, &backlog->list, list) { + if (adf_send_message(req->tx_ring, req->fw_req)) { + /* The HW ring is full. Do nothing. + * qat_alg_send_backlog() will be invoked again by + * another callback. + */ + break; + } + list_del(&req->list); + crypto_request_complete(req->base, -EINPROGRESS); + } + spin_unlock_bh(&backlog->lock); +} + +static bool qat_alg_try_enqueue(struct qat_alg_req *req) +{ + struct qat_instance_backlog *backlog = req->backlog; + struct adf_etr_ring_data *tx_ring = req->tx_ring; + u32 *fw_req = req->fw_req; + + /* Check if any request is already backlogged */ + if (!list_empty(&backlog->list)) + return false; + + /* Check if ring is nearly full */ + if (adf_ring_nearly_full(tx_ring)) + return false; + + /* Try to enqueue to HW ring */ + if (adf_send_message(tx_ring, fw_req)) + return false; + + return true; +} + + +static int qat_alg_send_message_maybacklog(struct qat_alg_req *req) +{ + struct qat_instance_backlog *backlog = req->backlog; + int ret = -EINPROGRESS; + + if (qat_alg_try_enqueue(req)) + return ret; + + spin_lock_bh(&backlog->lock); + if (!qat_alg_try_enqueue(req)) { + list_add_tail(&req->list, &backlog->list); + ret = -EBUSY; + } + spin_unlock_bh(&backlog->lock); + + return ret; +} + +int qat_alg_send_message(struct qat_alg_req *req) +{ + u32 flags = req->base->flags; + + if (flags & CRYPTO_TFM_REQ_MAY_BACKLOG) + return qat_alg_send_message_maybacklog(req); + else + return qat_alg_send_message_retry(req); +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_algs_send.h b/drivers/crypto/intel/qat/qat_common/qat_algs_send.h new file mode 100644 index 000000000000..0baca16e1eff --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_algs_send.h @@ -0,0 +1,25 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef QAT_ALGS_SEND_H +#define QAT_ALGS_SEND_H + +#include <linux/list.h> +#include "adf_transport_internal.h" + +struct qat_instance_backlog { + struct list_head list; + spinlock_t lock; /* protects backlog list */ +}; + +struct qat_alg_req { + u32 *fw_req; + struct adf_etr_ring_data *tx_ring; + struct crypto_async_request *base; + struct list_head list; + struct qat_instance_backlog *backlog; +}; + +int qat_alg_send_message(struct qat_alg_req *req); +void qat_alg_send_backlog(struct qat_instance_backlog *backlog); + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/qat_asym_algs.c b/drivers/crypto/intel/qat/qat_common/qat_asym_algs.c new file mode 100644 index 000000000000..85c682e248fb --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_asym_algs.c @@ -0,0 +1,1361 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/module.h> +#include <crypto/internal/rsa.h> +#include <crypto/internal/akcipher.h> +#include <crypto/akcipher.h> +#include <crypto/kpp.h> +#include <crypto/internal/kpp.h> +#include <crypto/dh.h> +#include <linux/dma-mapping.h> +#include <linux/fips.h> +#include <crypto/scatterwalk.h> +#include "icp_qat_fw_pke.h" +#include "adf_accel_devices.h" +#include "qat_algs_send.h" +#include "adf_transport.h" +#include "adf_common_drv.h" +#include "qat_crypto.h" + +static DEFINE_MUTEX(algs_lock); +static unsigned int active_devs; + +struct qat_rsa_input_params { + union { + struct { + dma_addr_t m; + dma_addr_t e; + dma_addr_t n; + } enc; + struct { + dma_addr_t c; + dma_addr_t d; + dma_addr_t n; + } dec; + struct { + dma_addr_t c; + dma_addr_t p; + dma_addr_t q; + dma_addr_t dp; + dma_addr_t dq; + dma_addr_t qinv; + } dec_crt; + u64 in_tab[8]; + }; +} __packed __aligned(64); + +struct qat_rsa_output_params { + union { + struct { + dma_addr_t c; + } enc; + struct { + dma_addr_t m; + } dec; + u64 out_tab[8]; + }; +} __packed __aligned(64); + +struct qat_rsa_ctx { + char *n; + char *e; + char *d; + char *p; + char *q; + char *dp; + char *dq; + char *qinv; + dma_addr_t dma_n; + dma_addr_t dma_e; + dma_addr_t dma_d; + dma_addr_t dma_p; + dma_addr_t dma_q; + dma_addr_t dma_dp; + dma_addr_t dma_dq; + dma_addr_t dma_qinv; + unsigned int key_sz; + bool crt_mode; + struct qat_crypto_instance *inst; +} __packed __aligned(64); + +struct qat_dh_input_params { + union { + struct { + dma_addr_t b; + dma_addr_t xa; + dma_addr_t p; + } in; + struct { + dma_addr_t xa; + dma_addr_t p; + } in_g2; + u64 in_tab[8]; + }; +} __packed __aligned(64); + +struct qat_dh_output_params { + union { + dma_addr_t r; + u64 out_tab[8]; + }; +} __packed __aligned(64); + +struct qat_dh_ctx { + char *g; + char *xa; + char *p; + dma_addr_t dma_g; + dma_addr_t dma_xa; + dma_addr_t dma_p; + unsigned int p_size; + bool g2; + struct qat_crypto_instance *inst; + struct crypto_kpp *ftfm; + bool fallback; +} __packed __aligned(64); + +struct qat_asym_request { + union { + struct qat_rsa_input_params rsa; + struct qat_dh_input_params dh; + } in; + union { + struct qat_rsa_output_params rsa; + struct qat_dh_output_params dh; + } out; + dma_addr_t phy_in; + dma_addr_t phy_out; + char *src_align; + char *dst_align; + struct icp_qat_fw_pke_request req; + union { + struct qat_rsa_ctx *rsa; + struct qat_dh_ctx *dh; + } ctx; + union { + struct akcipher_request *rsa; + struct kpp_request *dh; + } areq; + int err; + void (*cb)(struct icp_qat_fw_pke_resp *resp); + struct qat_alg_req alg_req; +} __aligned(64); + +static int qat_alg_send_asym_message(struct qat_asym_request *qat_req, + struct qat_crypto_instance *inst, + struct crypto_async_request *base) +{ + struct qat_alg_req *alg_req = &qat_req->alg_req; + + alg_req->fw_req = (u32 *)&qat_req->req; + alg_req->tx_ring = inst->pke_tx; + alg_req->base = base; + alg_req->backlog = &inst->backlog; + + return qat_alg_send_message(alg_req); +} + +static void qat_dh_cb(struct icp_qat_fw_pke_resp *resp) +{ + struct qat_asym_request *req = (void *)(__force long)resp->opaque; + struct kpp_request *areq = req->areq.dh; + struct device *dev = &GET_DEV(req->ctx.dh->inst->accel_dev); + int err = ICP_QAT_FW_PKE_RESP_PKE_STAT_GET( + resp->pke_resp_hdr.comn_resp_flags); + + err = (err == ICP_QAT_FW_COMN_STATUS_FLAG_OK) ? 0 : -EINVAL; + + if (areq->src) { + dma_unmap_single(dev, req->in.dh.in.b, req->ctx.dh->p_size, + DMA_TO_DEVICE); + kfree_sensitive(req->src_align); + } + + areq->dst_len = req->ctx.dh->p_size; + dma_unmap_single(dev, req->out.dh.r, req->ctx.dh->p_size, + DMA_FROM_DEVICE); + if (req->dst_align) { + scatterwalk_map_and_copy(req->dst_align, areq->dst, 0, + areq->dst_len, 1); + kfree_sensitive(req->dst_align); + } + + dma_unmap_single(dev, req->phy_in, sizeof(struct qat_dh_input_params), + DMA_TO_DEVICE); + dma_unmap_single(dev, req->phy_out, + sizeof(struct qat_dh_output_params), + DMA_TO_DEVICE); + + kpp_request_complete(areq, err); +} + +#define PKE_DH_1536 0x390c1a49 +#define PKE_DH_G2_1536 0x2e0b1a3e +#define PKE_DH_2048 0x4d0c1a60 +#define PKE_DH_G2_2048 0x3e0b1a55 +#define PKE_DH_3072 0x510c1a77 +#define PKE_DH_G2_3072 0x3a0b1a6c +#define PKE_DH_4096 0x690c1a8e +#define PKE_DH_G2_4096 0x4a0b1a83 + +static unsigned long qat_dh_fn_id(unsigned int len, bool g2) +{ + unsigned int bitslen = len << 3; + + switch (bitslen) { + case 1536: + return g2 ? PKE_DH_G2_1536 : PKE_DH_1536; + case 2048: + return g2 ? PKE_DH_G2_2048 : PKE_DH_2048; + case 3072: + return g2 ? PKE_DH_G2_3072 : PKE_DH_3072; + case 4096: + return g2 ? PKE_DH_G2_4096 : PKE_DH_4096; + default: + return 0; + } +} + +static int qat_dh_compute_value(struct kpp_request *req) +{ + struct crypto_kpp *tfm = crypto_kpp_reqtfm(req); + struct qat_dh_ctx *ctx = kpp_tfm_ctx(tfm); + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + struct qat_asym_request *qat_req = + PTR_ALIGN(kpp_request_ctx(req), 64); + struct icp_qat_fw_pke_request *msg = &qat_req->req; + gfp_t flags = qat_algs_alloc_flags(&req->base); + int n_input_params = 0; + u8 *vaddr; + int ret; + + if (unlikely(!ctx->xa)) + return -EINVAL; + + if (req->dst_len < ctx->p_size) { + req->dst_len = ctx->p_size; + return -EOVERFLOW; + } + + if (req->src_len > ctx->p_size) + return -EINVAL; + + memset(msg, '\0', sizeof(*msg)); + ICP_QAT_FW_PKE_HDR_VALID_FLAG_SET(msg->pke_hdr, + ICP_QAT_FW_COMN_REQ_FLAG_SET); + + msg->pke_hdr.cd_pars.func_id = qat_dh_fn_id(ctx->p_size, + !req->src && ctx->g2); + if (unlikely(!msg->pke_hdr.cd_pars.func_id)) + return -EINVAL; + + qat_req->cb = qat_dh_cb; + qat_req->ctx.dh = ctx; + qat_req->areq.dh = req; + msg->pke_hdr.service_type = ICP_QAT_FW_COMN_REQ_CPM_FW_PKE; + msg->pke_hdr.comn_req_flags = + ICP_QAT_FW_COMN_FLAGS_BUILD(QAT_COMN_PTR_TYPE_FLAT, + QAT_COMN_CD_FLD_TYPE_64BIT_ADR); + + /* + * If no source is provided use g as base + */ + if (req->src) { + qat_req->in.dh.in.xa = ctx->dma_xa; + qat_req->in.dh.in.p = ctx->dma_p; + n_input_params = 3; + } else { + if (ctx->g2) { + qat_req->in.dh.in_g2.xa = ctx->dma_xa; + qat_req->in.dh.in_g2.p = ctx->dma_p; + n_input_params = 2; + } else { + qat_req->in.dh.in.b = ctx->dma_g; + qat_req->in.dh.in.xa = ctx->dma_xa; + qat_req->in.dh.in.p = ctx->dma_p; + n_input_params = 3; + } + } + + ret = -ENOMEM; + if (req->src) { + /* + * src can be of any size in valid range, but HW expects it to + * be the same as modulo p so in case it is different we need + * to allocate a new buf and copy src data. + * In other case we just need to map the user provided buffer. + * Also need to make sure that it is in contiguous buffer. + */ + if (sg_is_last(req->src) && req->src_len == ctx->p_size) { + qat_req->src_align = NULL; + vaddr = sg_virt(req->src); + } else { + int shift = ctx->p_size - req->src_len; + + qat_req->src_align = kzalloc(ctx->p_size, flags); + if (unlikely(!qat_req->src_align)) + return ret; + + scatterwalk_map_and_copy(qat_req->src_align + shift, + req->src, 0, req->src_len, 0); + + vaddr = qat_req->src_align; + } + + qat_req->in.dh.in.b = dma_map_single(dev, vaddr, ctx->p_size, + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->in.dh.in.b))) + goto unmap_src; + } + /* + * dst can be of any size in valid range, but HW expects it to be the + * same as modulo m so in case it is different we need to allocate a + * new buf and copy src data. + * In other case we just need to map the user provided buffer. + * Also need to make sure that it is in contiguous buffer. + */ + if (sg_is_last(req->dst) && req->dst_len == ctx->p_size) { + qat_req->dst_align = NULL; + vaddr = sg_virt(req->dst); + } else { + qat_req->dst_align = kzalloc(ctx->p_size, flags); + if (unlikely(!qat_req->dst_align)) + goto unmap_src; + + vaddr = qat_req->dst_align; + } + qat_req->out.dh.r = dma_map_single(dev, vaddr, ctx->p_size, + DMA_FROM_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->out.dh.r))) + goto unmap_dst; + + qat_req->in.dh.in_tab[n_input_params] = 0; + qat_req->out.dh.out_tab[1] = 0; + /* Mapping in.in.b or in.in_g2.xa is the same */ + qat_req->phy_in = dma_map_single(dev, &qat_req->in.dh, + sizeof(struct qat_dh_input_params), + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->phy_in))) + goto unmap_dst; + + qat_req->phy_out = dma_map_single(dev, &qat_req->out.dh, + sizeof(struct qat_dh_output_params), + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->phy_out))) + goto unmap_in_params; + + msg->pke_mid.src_data_addr = qat_req->phy_in; + msg->pke_mid.dest_data_addr = qat_req->phy_out; + msg->pke_mid.opaque = (u64)(__force long)qat_req; + msg->input_param_count = n_input_params; + msg->output_param_count = 1; + + ret = qat_alg_send_asym_message(qat_req, inst, &req->base); + if (ret == -ENOSPC) + goto unmap_all; + + return ret; + +unmap_all: + if (!dma_mapping_error(dev, qat_req->phy_out)) + dma_unmap_single(dev, qat_req->phy_out, + sizeof(struct qat_dh_output_params), + DMA_TO_DEVICE); +unmap_in_params: + if (!dma_mapping_error(dev, qat_req->phy_in)) + dma_unmap_single(dev, qat_req->phy_in, + sizeof(struct qat_dh_input_params), + DMA_TO_DEVICE); +unmap_dst: + if (!dma_mapping_error(dev, qat_req->out.dh.r)) + dma_unmap_single(dev, qat_req->out.dh.r, ctx->p_size, + DMA_FROM_DEVICE); + kfree_sensitive(qat_req->dst_align); +unmap_src: + if (req->src) { + if (!dma_mapping_error(dev, qat_req->in.dh.in.b)) + dma_unmap_single(dev, qat_req->in.dh.in.b, + ctx->p_size, + DMA_TO_DEVICE); + kfree_sensitive(qat_req->src_align); + } + return ret; +} + +static int qat_dh_generate_public_key(struct kpp_request *req) +{ + struct kpp_request *nreq = kpp_request_ctx(req); + struct crypto_kpp *tfm = crypto_kpp_reqtfm(req); + struct qat_dh_ctx *ctx = kpp_tfm_ctx(tfm); + + if (ctx->fallback) { + memcpy(nreq, req, sizeof(*req)); + kpp_request_set_tfm(nreq, ctx->ftfm); + return crypto_kpp_generate_public_key(nreq); + } + + return qat_dh_compute_value(req); +} + +static int qat_dh_compute_shared_secret(struct kpp_request *req) +{ + struct kpp_request *nreq = kpp_request_ctx(req); + struct crypto_kpp *tfm = crypto_kpp_reqtfm(req); + struct qat_dh_ctx *ctx = kpp_tfm_ctx(tfm); + + if (ctx->fallback) { + memcpy(nreq, req, sizeof(*req)); + kpp_request_set_tfm(nreq, ctx->ftfm); + return crypto_kpp_compute_shared_secret(nreq); + } + + return qat_dh_compute_value(req); +} + +static int qat_dh_check_params_length(unsigned int p_len) +{ + switch (p_len) { + case 1536: + case 2048: + case 3072: + case 4096: + return 0; + } + return -EINVAL; +} + +static int qat_dh_set_params(struct qat_dh_ctx *ctx, struct dh *params) +{ + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + + ctx->p_size = params->p_size; + ctx->p = dma_alloc_coherent(dev, ctx->p_size, &ctx->dma_p, GFP_KERNEL); + if (!ctx->p) + return -ENOMEM; + memcpy(ctx->p, params->p, ctx->p_size); + + /* If g equals 2 don't copy it */ + if (params->g_size == 1 && *(char *)params->g == 0x02) { + ctx->g2 = true; + return 0; + } + + ctx->g = dma_alloc_coherent(dev, ctx->p_size, &ctx->dma_g, GFP_KERNEL); + if (!ctx->g) + return -ENOMEM; + memcpy(ctx->g + (ctx->p_size - params->g_size), params->g, + params->g_size); + + return 0; +} + +static void qat_dh_clear_ctx(struct device *dev, struct qat_dh_ctx *ctx) +{ + if (ctx->g) { + memset(ctx->g, 0, ctx->p_size); + dma_free_coherent(dev, ctx->p_size, ctx->g, ctx->dma_g); + ctx->g = NULL; + } + if (ctx->xa) { + memset(ctx->xa, 0, ctx->p_size); + dma_free_coherent(dev, ctx->p_size, ctx->xa, ctx->dma_xa); + ctx->xa = NULL; + } + if (ctx->p) { + memset(ctx->p, 0, ctx->p_size); + dma_free_coherent(dev, ctx->p_size, ctx->p, ctx->dma_p); + ctx->p = NULL; + } + ctx->p_size = 0; + ctx->g2 = false; +} + +static int qat_dh_set_secret(struct crypto_kpp *tfm, const void *buf, + unsigned int len) +{ + struct qat_dh_ctx *ctx = kpp_tfm_ctx(tfm); + struct device *dev = &GET_DEV(ctx->inst->accel_dev); + struct dh params; + int ret; + + if (crypto_dh_decode_key(buf, len, ¶ms) < 0) + return -EINVAL; + + if (qat_dh_check_params_length(params.p_size << 3)) { + ctx->fallback = true; + return crypto_kpp_set_secret(ctx->ftfm, buf, len); + } + + ctx->fallback = false; + + /* Free old secret if any */ + qat_dh_clear_ctx(dev, ctx); + + ret = qat_dh_set_params(ctx, ¶ms); + if (ret < 0) + goto err_clear_ctx; + + ctx->xa = dma_alloc_coherent(dev, ctx->p_size, &ctx->dma_xa, + GFP_KERNEL); + if (!ctx->xa) { + ret = -ENOMEM; + goto err_clear_ctx; + } + memcpy(ctx->xa + (ctx->p_size - params.key_size), params.key, + params.key_size); + + return 0; + +err_clear_ctx: + qat_dh_clear_ctx(dev, ctx); + return ret; +} + +static unsigned int qat_dh_max_size(struct crypto_kpp *tfm) +{ + struct qat_dh_ctx *ctx = kpp_tfm_ctx(tfm); + + if (ctx->fallback) + return crypto_kpp_maxsize(ctx->ftfm); + + return ctx->p_size; +} + +static int qat_dh_init_tfm(struct crypto_kpp *tfm) +{ + struct qat_dh_ctx *ctx = kpp_tfm_ctx(tfm); + struct qat_crypto_instance *inst = + qat_crypto_get_instance_node(numa_node_id()); + const char *alg = kpp_alg_name(tfm); + unsigned int reqsize; + + if (!inst) + return -EINVAL; + + ctx->ftfm = crypto_alloc_kpp(alg, 0, CRYPTO_ALG_NEED_FALLBACK); + if (IS_ERR(ctx->ftfm)) + return PTR_ERR(ctx->ftfm); + + crypto_kpp_set_flags(ctx->ftfm, crypto_kpp_get_flags(tfm)); + + reqsize = max(sizeof(struct qat_asym_request) + 64, + sizeof(struct kpp_request) + crypto_kpp_reqsize(ctx->ftfm)); + + kpp_set_reqsize(tfm, reqsize); + + ctx->p_size = 0; + ctx->g2 = false; + ctx->inst = inst; + return 0; +} + +static void qat_dh_exit_tfm(struct crypto_kpp *tfm) +{ + struct qat_dh_ctx *ctx = kpp_tfm_ctx(tfm); + struct device *dev = &GET_DEV(ctx->inst->accel_dev); + + if (ctx->ftfm) + crypto_free_kpp(ctx->ftfm); + + qat_dh_clear_ctx(dev, ctx); + qat_crypto_put_instance(ctx->inst); +} + +static void qat_rsa_cb(struct icp_qat_fw_pke_resp *resp) +{ + struct qat_asym_request *req = (void *)(__force long)resp->opaque; + struct akcipher_request *areq = req->areq.rsa; + struct device *dev = &GET_DEV(req->ctx.rsa->inst->accel_dev); + int err = ICP_QAT_FW_PKE_RESP_PKE_STAT_GET( + resp->pke_resp_hdr.comn_resp_flags); + + err = (err == ICP_QAT_FW_COMN_STATUS_FLAG_OK) ? 0 : -EINVAL; + + dma_unmap_single(dev, req->in.rsa.enc.m, req->ctx.rsa->key_sz, + DMA_TO_DEVICE); + + kfree_sensitive(req->src_align); + + areq->dst_len = req->ctx.rsa->key_sz; + dma_unmap_single(dev, req->out.rsa.enc.c, req->ctx.rsa->key_sz, + DMA_FROM_DEVICE); + if (req->dst_align) { + scatterwalk_map_and_copy(req->dst_align, areq->dst, 0, + areq->dst_len, 1); + + kfree_sensitive(req->dst_align); + } + + dma_unmap_single(dev, req->phy_in, sizeof(struct qat_rsa_input_params), + DMA_TO_DEVICE); + dma_unmap_single(dev, req->phy_out, + sizeof(struct qat_rsa_output_params), + DMA_TO_DEVICE); + + akcipher_request_complete(areq, err); +} + +void qat_alg_asym_callback(void *_resp) +{ + struct icp_qat_fw_pke_resp *resp = _resp; + struct qat_asym_request *areq = (void *)(__force long)resp->opaque; + struct qat_instance_backlog *backlog = areq->alg_req.backlog; + + areq->cb(resp); + + qat_alg_send_backlog(backlog); +} + +#define PKE_RSA_EP_512 0x1c161b21 +#define PKE_RSA_EP_1024 0x35111bf7 +#define PKE_RSA_EP_1536 0x4d111cdc +#define PKE_RSA_EP_2048 0x6e111dba +#define PKE_RSA_EP_3072 0x7d111ea3 +#define PKE_RSA_EP_4096 0xa5101f7e + +static unsigned long qat_rsa_enc_fn_id(unsigned int len) +{ + unsigned int bitslen = len << 3; + + switch (bitslen) { + case 512: + return PKE_RSA_EP_512; + case 1024: + return PKE_RSA_EP_1024; + case 1536: + return PKE_RSA_EP_1536; + case 2048: + return PKE_RSA_EP_2048; + case 3072: + return PKE_RSA_EP_3072; + case 4096: + return PKE_RSA_EP_4096; + default: + return 0; + } +} + +#define PKE_RSA_DP1_512 0x1c161b3c +#define PKE_RSA_DP1_1024 0x35111c12 +#define PKE_RSA_DP1_1536 0x4d111cf7 +#define PKE_RSA_DP1_2048 0x6e111dda +#define PKE_RSA_DP1_3072 0x7d111ebe +#define PKE_RSA_DP1_4096 0xa5101f98 + +static unsigned long qat_rsa_dec_fn_id(unsigned int len) +{ + unsigned int bitslen = len << 3; + + switch (bitslen) { + case 512: + return PKE_RSA_DP1_512; + case 1024: + return PKE_RSA_DP1_1024; + case 1536: + return PKE_RSA_DP1_1536; + case 2048: + return PKE_RSA_DP1_2048; + case 3072: + return PKE_RSA_DP1_3072; + case 4096: + return PKE_RSA_DP1_4096; + default: + return 0; + } +} + +#define PKE_RSA_DP2_512 0x1c131b57 +#define PKE_RSA_DP2_1024 0x26131c2d +#define PKE_RSA_DP2_1536 0x45111d12 +#define PKE_RSA_DP2_2048 0x59121dfa +#define PKE_RSA_DP2_3072 0x81121ed9 +#define PKE_RSA_DP2_4096 0xb1111fb2 + +static unsigned long qat_rsa_dec_fn_id_crt(unsigned int len) +{ + unsigned int bitslen = len << 3; + + switch (bitslen) { + case 512: + return PKE_RSA_DP2_512; + case 1024: + return PKE_RSA_DP2_1024; + case 1536: + return PKE_RSA_DP2_1536; + case 2048: + return PKE_RSA_DP2_2048; + case 3072: + return PKE_RSA_DP2_3072; + case 4096: + return PKE_RSA_DP2_4096; + default: + return 0; + } +} + +static int qat_rsa_enc(struct akcipher_request *req) +{ + struct crypto_akcipher *tfm = crypto_akcipher_reqtfm(req); + struct qat_rsa_ctx *ctx = akcipher_tfm_ctx(tfm); + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + struct qat_asym_request *qat_req = + PTR_ALIGN(akcipher_request_ctx(req), 64); + struct icp_qat_fw_pke_request *msg = &qat_req->req; + gfp_t flags = qat_algs_alloc_flags(&req->base); + u8 *vaddr; + int ret; + + if (unlikely(!ctx->n || !ctx->e)) + return -EINVAL; + + if (req->dst_len < ctx->key_sz) { + req->dst_len = ctx->key_sz; + return -EOVERFLOW; + } + + if (req->src_len > ctx->key_sz) + return -EINVAL; + + memset(msg, '\0', sizeof(*msg)); + ICP_QAT_FW_PKE_HDR_VALID_FLAG_SET(msg->pke_hdr, + ICP_QAT_FW_COMN_REQ_FLAG_SET); + msg->pke_hdr.cd_pars.func_id = qat_rsa_enc_fn_id(ctx->key_sz); + if (unlikely(!msg->pke_hdr.cd_pars.func_id)) + return -EINVAL; + + qat_req->cb = qat_rsa_cb; + qat_req->ctx.rsa = ctx; + qat_req->areq.rsa = req; + msg->pke_hdr.service_type = ICP_QAT_FW_COMN_REQ_CPM_FW_PKE; + msg->pke_hdr.comn_req_flags = + ICP_QAT_FW_COMN_FLAGS_BUILD(QAT_COMN_PTR_TYPE_FLAT, + QAT_COMN_CD_FLD_TYPE_64BIT_ADR); + + qat_req->in.rsa.enc.e = ctx->dma_e; + qat_req->in.rsa.enc.n = ctx->dma_n; + ret = -ENOMEM; + + /* + * src can be of any size in valid range, but HW expects it to be the + * same as modulo n so in case it is different we need to allocate a + * new buf and copy src data. + * In other case we just need to map the user provided buffer. + * Also need to make sure that it is in contiguous buffer. + */ + if (sg_is_last(req->src) && req->src_len == ctx->key_sz) { + qat_req->src_align = NULL; + vaddr = sg_virt(req->src); + } else { + int shift = ctx->key_sz - req->src_len; + + qat_req->src_align = kzalloc(ctx->key_sz, flags); + if (unlikely(!qat_req->src_align)) + return ret; + + scatterwalk_map_and_copy(qat_req->src_align + shift, req->src, + 0, req->src_len, 0); + vaddr = qat_req->src_align; + } + + qat_req->in.rsa.enc.m = dma_map_single(dev, vaddr, ctx->key_sz, + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->in.rsa.enc.m))) + goto unmap_src; + + if (sg_is_last(req->dst) && req->dst_len == ctx->key_sz) { + qat_req->dst_align = NULL; + vaddr = sg_virt(req->dst); + } else { + qat_req->dst_align = kzalloc(ctx->key_sz, flags); + if (unlikely(!qat_req->dst_align)) + goto unmap_src; + vaddr = qat_req->dst_align; + } + + qat_req->out.rsa.enc.c = dma_map_single(dev, vaddr, ctx->key_sz, + DMA_FROM_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->out.rsa.enc.c))) + goto unmap_dst; + + qat_req->in.rsa.in_tab[3] = 0; + qat_req->out.rsa.out_tab[1] = 0; + qat_req->phy_in = dma_map_single(dev, &qat_req->in.rsa, + sizeof(struct qat_rsa_input_params), + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->phy_in))) + goto unmap_dst; + + qat_req->phy_out = dma_map_single(dev, &qat_req->out.rsa, + sizeof(struct qat_rsa_output_params), + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->phy_out))) + goto unmap_in_params; + + msg->pke_mid.src_data_addr = qat_req->phy_in; + msg->pke_mid.dest_data_addr = qat_req->phy_out; + msg->pke_mid.opaque = (u64)(__force long)qat_req; + msg->input_param_count = 3; + msg->output_param_count = 1; + + ret = qat_alg_send_asym_message(qat_req, inst, &req->base); + if (ret == -ENOSPC) + goto unmap_all; + + return ret; + +unmap_all: + if (!dma_mapping_error(dev, qat_req->phy_out)) + dma_unmap_single(dev, qat_req->phy_out, + sizeof(struct qat_rsa_output_params), + DMA_TO_DEVICE); +unmap_in_params: + if (!dma_mapping_error(dev, qat_req->phy_in)) + dma_unmap_single(dev, qat_req->phy_in, + sizeof(struct qat_rsa_input_params), + DMA_TO_DEVICE); +unmap_dst: + if (!dma_mapping_error(dev, qat_req->out.rsa.enc.c)) + dma_unmap_single(dev, qat_req->out.rsa.enc.c, + ctx->key_sz, DMA_FROM_DEVICE); + kfree_sensitive(qat_req->dst_align); +unmap_src: + if (!dma_mapping_error(dev, qat_req->in.rsa.enc.m)) + dma_unmap_single(dev, qat_req->in.rsa.enc.m, ctx->key_sz, + DMA_TO_DEVICE); + kfree_sensitive(qat_req->src_align); + return ret; +} + +static int qat_rsa_dec(struct akcipher_request *req) +{ + struct crypto_akcipher *tfm = crypto_akcipher_reqtfm(req); + struct qat_rsa_ctx *ctx = akcipher_tfm_ctx(tfm); + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + struct qat_asym_request *qat_req = + PTR_ALIGN(akcipher_request_ctx(req), 64); + struct icp_qat_fw_pke_request *msg = &qat_req->req; + gfp_t flags = qat_algs_alloc_flags(&req->base); + u8 *vaddr; + int ret; + + if (unlikely(!ctx->n || !ctx->d)) + return -EINVAL; + + if (req->dst_len < ctx->key_sz) { + req->dst_len = ctx->key_sz; + return -EOVERFLOW; + } + + if (req->src_len > ctx->key_sz) + return -EINVAL; + + memset(msg, '\0', sizeof(*msg)); + ICP_QAT_FW_PKE_HDR_VALID_FLAG_SET(msg->pke_hdr, + ICP_QAT_FW_COMN_REQ_FLAG_SET); + msg->pke_hdr.cd_pars.func_id = ctx->crt_mode ? + qat_rsa_dec_fn_id_crt(ctx->key_sz) : + qat_rsa_dec_fn_id(ctx->key_sz); + if (unlikely(!msg->pke_hdr.cd_pars.func_id)) + return -EINVAL; + + qat_req->cb = qat_rsa_cb; + qat_req->ctx.rsa = ctx; + qat_req->areq.rsa = req; + msg->pke_hdr.service_type = ICP_QAT_FW_COMN_REQ_CPM_FW_PKE; + msg->pke_hdr.comn_req_flags = + ICP_QAT_FW_COMN_FLAGS_BUILD(QAT_COMN_PTR_TYPE_FLAT, + QAT_COMN_CD_FLD_TYPE_64BIT_ADR); + + if (ctx->crt_mode) { + qat_req->in.rsa.dec_crt.p = ctx->dma_p; + qat_req->in.rsa.dec_crt.q = ctx->dma_q; + qat_req->in.rsa.dec_crt.dp = ctx->dma_dp; + qat_req->in.rsa.dec_crt.dq = ctx->dma_dq; + qat_req->in.rsa.dec_crt.qinv = ctx->dma_qinv; + } else { + qat_req->in.rsa.dec.d = ctx->dma_d; + qat_req->in.rsa.dec.n = ctx->dma_n; + } + ret = -ENOMEM; + + /* + * src can be of any size in valid range, but HW expects it to be the + * same as modulo n so in case it is different we need to allocate a + * new buf and copy src data. + * In other case we just need to map the user provided buffer. + * Also need to make sure that it is in contiguous buffer. + */ + if (sg_is_last(req->src) && req->src_len == ctx->key_sz) { + qat_req->src_align = NULL; + vaddr = sg_virt(req->src); + } else { + int shift = ctx->key_sz - req->src_len; + + qat_req->src_align = kzalloc(ctx->key_sz, flags); + if (unlikely(!qat_req->src_align)) + return ret; + + scatterwalk_map_and_copy(qat_req->src_align + shift, req->src, + 0, req->src_len, 0); + vaddr = qat_req->src_align; + } + + qat_req->in.rsa.dec.c = dma_map_single(dev, vaddr, ctx->key_sz, + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->in.rsa.dec.c))) + goto unmap_src; + + if (sg_is_last(req->dst) && req->dst_len == ctx->key_sz) { + qat_req->dst_align = NULL; + vaddr = sg_virt(req->dst); + } else { + qat_req->dst_align = kzalloc(ctx->key_sz, flags); + if (unlikely(!qat_req->dst_align)) + goto unmap_src; + vaddr = qat_req->dst_align; + } + qat_req->out.rsa.dec.m = dma_map_single(dev, vaddr, ctx->key_sz, + DMA_FROM_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->out.rsa.dec.m))) + goto unmap_dst; + + if (ctx->crt_mode) + qat_req->in.rsa.in_tab[6] = 0; + else + qat_req->in.rsa.in_tab[3] = 0; + qat_req->out.rsa.out_tab[1] = 0; + qat_req->phy_in = dma_map_single(dev, &qat_req->in.rsa, + sizeof(struct qat_rsa_input_params), + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->phy_in))) + goto unmap_dst; + + qat_req->phy_out = dma_map_single(dev, &qat_req->out.rsa, + sizeof(struct qat_rsa_output_params), + DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, qat_req->phy_out))) + goto unmap_in_params; + + msg->pke_mid.src_data_addr = qat_req->phy_in; + msg->pke_mid.dest_data_addr = qat_req->phy_out; + msg->pke_mid.opaque = (u64)(__force long)qat_req; + if (ctx->crt_mode) + msg->input_param_count = 6; + else + msg->input_param_count = 3; + + msg->output_param_count = 1; + + ret = qat_alg_send_asym_message(qat_req, inst, &req->base); + if (ret == -ENOSPC) + goto unmap_all; + + return ret; + +unmap_all: + if (!dma_mapping_error(dev, qat_req->phy_out)) + dma_unmap_single(dev, qat_req->phy_out, + sizeof(struct qat_rsa_output_params), + DMA_TO_DEVICE); +unmap_in_params: + if (!dma_mapping_error(dev, qat_req->phy_in)) + dma_unmap_single(dev, qat_req->phy_in, + sizeof(struct qat_rsa_input_params), + DMA_TO_DEVICE); +unmap_dst: + if (!dma_mapping_error(dev, qat_req->out.rsa.dec.m)) + dma_unmap_single(dev, qat_req->out.rsa.dec.m, + ctx->key_sz, DMA_FROM_DEVICE); + kfree_sensitive(qat_req->dst_align); +unmap_src: + if (!dma_mapping_error(dev, qat_req->in.rsa.dec.c)) + dma_unmap_single(dev, qat_req->in.rsa.dec.c, ctx->key_sz, + DMA_TO_DEVICE); + kfree_sensitive(qat_req->src_align); + return ret; +} + +static int qat_rsa_set_n(struct qat_rsa_ctx *ctx, const char *value, + size_t vlen) +{ + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + const char *ptr = value; + int ret; + + while (!*ptr && vlen) { + ptr++; + vlen--; + } + + ctx->key_sz = vlen; + ret = -EINVAL; + /* invalid key size provided */ + if (!qat_rsa_enc_fn_id(ctx->key_sz)) + goto err; + + ret = -ENOMEM; + ctx->n = dma_alloc_coherent(dev, ctx->key_sz, &ctx->dma_n, GFP_KERNEL); + if (!ctx->n) + goto err; + + memcpy(ctx->n, ptr, ctx->key_sz); + return 0; +err: + ctx->key_sz = 0; + ctx->n = NULL; + return ret; +} + +static int qat_rsa_set_e(struct qat_rsa_ctx *ctx, const char *value, + size_t vlen) +{ + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + const char *ptr = value; + + while (!*ptr && vlen) { + ptr++; + vlen--; + } + + if (!ctx->key_sz || !vlen || vlen > ctx->key_sz) { + ctx->e = NULL; + return -EINVAL; + } + + ctx->e = dma_alloc_coherent(dev, ctx->key_sz, &ctx->dma_e, GFP_KERNEL); + if (!ctx->e) + return -ENOMEM; + + memcpy(ctx->e + (ctx->key_sz - vlen), ptr, vlen); + return 0; +} + +static int qat_rsa_set_d(struct qat_rsa_ctx *ctx, const char *value, + size_t vlen) +{ + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + const char *ptr = value; + int ret; + + while (!*ptr && vlen) { + ptr++; + vlen--; + } + + ret = -EINVAL; + if (!ctx->key_sz || !vlen || vlen > ctx->key_sz) + goto err; + + ret = -ENOMEM; + ctx->d = dma_alloc_coherent(dev, ctx->key_sz, &ctx->dma_d, GFP_KERNEL); + if (!ctx->d) + goto err; + + memcpy(ctx->d + (ctx->key_sz - vlen), ptr, vlen); + return 0; +err: + ctx->d = NULL; + return ret; +} + +static void qat_rsa_drop_leading_zeros(const char **ptr, unsigned int *len) +{ + while (!**ptr && *len) { + (*ptr)++; + (*len)--; + } +} + +static void qat_rsa_setkey_crt(struct qat_rsa_ctx *ctx, struct rsa_key *rsa_key) +{ + struct qat_crypto_instance *inst = ctx->inst; + struct device *dev = &GET_DEV(inst->accel_dev); + const char *ptr; + unsigned int len; + unsigned int half_key_sz = ctx->key_sz / 2; + + /* p */ + ptr = rsa_key->p; + len = rsa_key->p_sz; + qat_rsa_drop_leading_zeros(&ptr, &len); + if (!len) + goto err; + ctx->p = dma_alloc_coherent(dev, half_key_sz, &ctx->dma_p, GFP_KERNEL); + if (!ctx->p) + goto err; + memcpy(ctx->p + (half_key_sz - len), ptr, len); + + /* q */ + ptr = rsa_key->q; + len = rsa_key->q_sz; + qat_rsa_drop_leading_zeros(&ptr, &len); + if (!len) + goto free_p; + ctx->q = dma_alloc_coherent(dev, half_key_sz, &ctx->dma_q, GFP_KERNEL); + if (!ctx->q) + goto free_p; + memcpy(ctx->q + (half_key_sz - len), ptr, len); + + /* dp */ + ptr = rsa_key->dp; + len = rsa_key->dp_sz; + qat_rsa_drop_leading_zeros(&ptr, &len); + if (!len) + goto free_q; + ctx->dp = dma_alloc_coherent(dev, half_key_sz, &ctx->dma_dp, + GFP_KERNEL); + if (!ctx->dp) + goto free_q; + memcpy(ctx->dp + (half_key_sz - len), ptr, len); + + /* dq */ + ptr = rsa_key->dq; + len = rsa_key->dq_sz; + qat_rsa_drop_leading_zeros(&ptr, &len); + if (!len) + goto free_dp; + ctx->dq = dma_alloc_coherent(dev, half_key_sz, &ctx->dma_dq, + GFP_KERNEL); + if (!ctx->dq) + goto free_dp; + memcpy(ctx->dq + (half_key_sz - len), ptr, len); + + /* qinv */ + ptr = rsa_key->qinv; + len = rsa_key->qinv_sz; + qat_rsa_drop_leading_zeros(&ptr, &len); + if (!len) + goto free_dq; + ctx->qinv = dma_alloc_coherent(dev, half_key_sz, &ctx->dma_qinv, + GFP_KERNEL); + if (!ctx->qinv) + goto free_dq; + memcpy(ctx->qinv + (half_key_sz - len), ptr, len); + + ctx->crt_mode = true; + return; + +free_dq: + memset(ctx->dq, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->dq, ctx->dma_dq); + ctx->dq = NULL; +free_dp: + memset(ctx->dp, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->dp, ctx->dma_dp); + ctx->dp = NULL; +free_q: + memset(ctx->q, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->q, ctx->dma_q); + ctx->q = NULL; +free_p: + memset(ctx->p, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->p, ctx->dma_p); + ctx->p = NULL; +err: + ctx->crt_mode = false; +} + +static void qat_rsa_clear_ctx(struct device *dev, struct qat_rsa_ctx *ctx) +{ + unsigned int half_key_sz = ctx->key_sz / 2; + + /* Free the old key if any */ + if (ctx->n) + dma_free_coherent(dev, ctx->key_sz, ctx->n, ctx->dma_n); + if (ctx->e) + dma_free_coherent(dev, ctx->key_sz, ctx->e, ctx->dma_e); + if (ctx->d) { + memset(ctx->d, '\0', ctx->key_sz); + dma_free_coherent(dev, ctx->key_sz, ctx->d, ctx->dma_d); + } + if (ctx->p) { + memset(ctx->p, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->p, ctx->dma_p); + } + if (ctx->q) { + memset(ctx->q, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->q, ctx->dma_q); + } + if (ctx->dp) { + memset(ctx->dp, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->dp, ctx->dma_dp); + } + if (ctx->dq) { + memset(ctx->dq, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->dq, ctx->dma_dq); + } + if (ctx->qinv) { + memset(ctx->qinv, '\0', half_key_sz); + dma_free_coherent(dev, half_key_sz, ctx->qinv, ctx->dma_qinv); + } + + ctx->n = NULL; + ctx->e = NULL; + ctx->d = NULL; + ctx->p = NULL; + ctx->q = NULL; + ctx->dp = NULL; + ctx->dq = NULL; + ctx->qinv = NULL; + ctx->crt_mode = false; + ctx->key_sz = 0; +} + +static int qat_rsa_setkey(struct crypto_akcipher *tfm, const void *key, + unsigned int keylen, bool private) +{ + struct qat_rsa_ctx *ctx = akcipher_tfm_ctx(tfm); + struct device *dev = &GET_DEV(ctx->inst->accel_dev); + struct rsa_key rsa_key; + int ret; + + qat_rsa_clear_ctx(dev, ctx); + + if (private) + ret = rsa_parse_priv_key(&rsa_key, key, keylen); + else + ret = rsa_parse_pub_key(&rsa_key, key, keylen); + if (ret < 0) + goto free; + + ret = qat_rsa_set_n(ctx, rsa_key.n, rsa_key.n_sz); + if (ret < 0) + goto free; + ret = qat_rsa_set_e(ctx, rsa_key.e, rsa_key.e_sz); + if (ret < 0) + goto free; + if (private) { + ret = qat_rsa_set_d(ctx, rsa_key.d, rsa_key.d_sz); + if (ret < 0) + goto free; + qat_rsa_setkey_crt(ctx, &rsa_key); + } + + if (!ctx->n || !ctx->e) { + /* invalid key provided */ + ret = -EINVAL; + goto free; + } + if (private && !ctx->d) { + /* invalid private key provided */ + ret = -EINVAL; + goto free; + } + + return 0; +free: + qat_rsa_clear_ctx(dev, ctx); + return ret; +} + +static int qat_rsa_setpubkey(struct crypto_akcipher *tfm, const void *key, + unsigned int keylen) +{ + return qat_rsa_setkey(tfm, key, keylen, false); +} + +static int qat_rsa_setprivkey(struct crypto_akcipher *tfm, const void *key, + unsigned int keylen) +{ + return qat_rsa_setkey(tfm, key, keylen, true); +} + +static unsigned int qat_rsa_max_size(struct crypto_akcipher *tfm) +{ + struct qat_rsa_ctx *ctx = akcipher_tfm_ctx(tfm); + + return ctx->key_sz; +} + +static int qat_rsa_init_tfm(struct crypto_akcipher *tfm) +{ + struct qat_rsa_ctx *ctx = akcipher_tfm_ctx(tfm); + struct qat_crypto_instance *inst = + qat_crypto_get_instance_node(numa_node_id()); + + if (!inst) + return -EINVAL; + + akcipher_set_reqsize(tfm, sizeof(struct qat_asym_request) + 64); + + ctx->key_sz = 0; + ctx->inst = inst; + return 0; +} + +static void qat_rsa_exit_tfm(struct crypto_akcipher *tfm) +{ + struct qat_rsa_ctx *ctx = akcipher_tfm_ctx(tfm); + struct device *dev = &GET_DEV(ctx->inst->accel_dev); + + qat_rsa_clear_ctx(dev, ctx); + qat_crypto_put_instance(ctx->inst); +} + +static struct akcipher_alg rsa = { + .encrypt = qat_rsa_enc, + .decrypt = qat_rsa_dec, + .set_pub_key = qat_rsa_setpubkey, + .set_priv_key = qat_rsa_setprivkey, + .max_size = qat_rsa_max_size, + .init = qat_rsa_init_tfm, + .exit = qat_rsa_exit_tfm, + .base = { + .cra_name = "rsa", + .cra_driver_name = "qat-rsa", + .cra_priority = 1000, + .cra_module = THIS_MODULE, + .cra_ctxsize = sizeof(struct qat_rsa_ctx), + }, +}; + +static struct kpp_alg dh = { + .set_secret = qat_dh_set_secret, + .generate_public_key = qat_dh_generate_public_key, + .compute_shared_secret = qat_dh_compute_shared_secret, + .max_size = qat_dh_max_size, + .init = qat_dh_init_tfm, + .exit = qat_dh_exit_tfm, + .base = { + .cra_name = "dh", + .cra_driver_name = "qat-dh", + .cra_priority = 1000, + .cra_module = THIS_MODULE, + .cra_ctxsize = sizeof(struct qat_dh_ctx), + .cra_flags = CRYPTO_ALG_NEED_FALLBACK, + }, +}; + +int qat_asym_algs_register(void) +{ + int ret = 0; + + mutex_lock(&algs_lock); + if (++active_devs == 1) { + rsa.base.cra_flags = 0; + ret = crypto_register_akcipher(&rsa); + if (ret) + goto unlock; + ret = crypto_register_kpp(&dh); + } +unlock: + mutex_unlock(&algs_lock); + return ret; +} + +void qat_asym_algs_unregister(void) +{ + mutex_lock(&algs_lock); + if (--active_devs == 0) { + crypto_unregister_akcipher(&rsa); + crypto_unregister_kpp(&dh); + } + mutex_unlock(&algs_lock); +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_bl.c b/drivers/crypto/intel/qat/qat_common/qat_bl.c new file mode 100644 index 000000000000..9b2338f58d97 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_bl.c @@ -0,0 +1,253 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2014 - 2022 Intel Corporation */ +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/pci.h> +#include <linux/scatterlist.h> +#include <linux/slab.h> +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "qat_bl.h" +#include "qat_crypto.h" + +void qat_bl_free_bufl(struct adf_accel_dev *accel_dev, + struct qat_request_buffs *buf) +{ + struct device *dev = &GET_DEV(accel_dev); + struct qat_alg_buf_list *bl = buf->bl; + struct qat_alg_buf_list *blout = buf->blout; + dma_addr_t blp = buf->blp; + dma_addr_t blpout = buf->bloutp; + size_t sz = buf->sz; + size_t sz_out = buf->sz_out; + int bl_dma_dir; + int i; + + bl_dma_dir = blp != blpout ? DMA_TO_DEVICE : DMA_BIDIRECTIONAL; + + for (i = 0; i < bl->num_bufs; i++) + dma_unmap_single(dev, bl->buffers[i].addr, + bl->buffers[i].len, bl_dma_dir); + + dma_unmap_single(dev, blp, sz, DMA_TO_DEVICE); + + if (!buf->sgl_src_valid) + kfree(bl); + + if (blp != blpout) { + for (i = 0; i < blout->num_mapped_bufs; i++) { + dma_unmap_single(dev, blout->buffers[i].addr, + blout->buffers[i].len, + DMA_BIDIRECTIONAL); + } + dma_unmap_single(dev, blpout, sz_out, DMA_TO_DEVICE); + + if (!buf->sgl_dst_valid) + kfree(blout); + } +} + +static int __qat_bl_sgl_to_bufl(struct adf_accel_dev *accel_dev, + struct scatterlist *sgl, + struct scatterlist *sglout, + struct qat_request_buffs *buf, + dma_addr_t extra_dst_buff, + size_t sz_extra_dst_buff, + unsigned int sskip, + unsigned int dskip, + gfp_t flags) +{ + struct device *dev = &GET_DEV(accel_dev); + int i, sg_nctr = 0; + int n = sg_nents(sgl); + struct qat_alg_buf_list *bufl; + struct qat_alg_buf_list *buflout = NULL; + dma_addr_t blp = DMA_MAPPING_ERROR; + dma_addr_t bloutp = DMA_MAPPING_ERROR; + struct scatterlist *sg; + size_t sz_out, sz = struct_size(bufl, buffers, n); + int node = dev_to_node(&GET_DEV(accel_dev)); + unsigned int left; + int bufl_dma_dir; + + if (unlikely(!n)) + return -EINVAL; + + buf->sgl_src_valid = false; + buf->sgl_dst_valid = false; + + if (n > QAT_MAX_BUFF_DESC) { + bufl = kzalloc_node(sz, flags, node); + if (unlikely(!bufl)) + return -ENOMEM; + } else { + bufl = container_of(&buf->sgl_src.sgl_hdr, + struct qat_alg_buf_list, hdr); + memset(bufl, 0, sizeof(struct qat_alg_buf_list)); + buf->sgl_src_valid = true; + } + + bufl_dma_dir = sgl != sglout ? DMA_TO_DEVICE : DMA_BIDIRECTIONAL; + + for (i = 0; i < n; i++) + bufl->buffers[i].addr = DMA_MAPPING_ERROR; + + left = sskip; + + for_each_sg(sgl, sg, n, i) { + int y = sg_nctr; + + if (!sg->length) + continue; + + if (left >= sg->length) { + left -= sg->length; + continue; + } + bufl->buffers[y].addr = dma_map_single(dev, sg_virt(sg) + left, + sg->length - left, + bufl_dma_dir); + bufl->buffers[y].len = sg->length; + if (unlikely(dma_mapping_error(dev, bufl->buffers[y].addr))) + goto err_in; + sg_nctr++; + if (left) { + bufl->buffers[y].len -= left; + left = 0; + } + } + bufl->num_bufs = sg_nctr; + blp = dma_map_single(dev, bufl, sz, DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, blp))) + goto err_in; + buf->bl = bufl; + buf->blp = blp; + buf->sz = sz; + /* Handle out of place operation */ + if (sgl != sglout) { + struct qat_alg_buf *buffers; + int extra_buff = extra_dst_buff ? 1 : 0; + int n_sglout = sg_nents(sglout); + + n = n_sglout + extra_buff; + sz_out = struct_size(buflout, buffers, n); + left = dskip; + + sg_nctr = 0; + + if (n > QAT_MAX_BUFF_DESC) { + buflout = kzalloc_node(sz_out, flags, node); + if (unlikely(!buflout)) + goto err_in; + } else { + buflout = container_of(&buf->sgl_dst.sgl_hdr, + struct qat_alg_buf_list, hdr); + memset(buflout, 0, sizeof(struct qat_alg_buf_list)); + buf->sgl_dst_valid = true; + } + + buffers = buflout->buffers; + for (i = 0; i < n; i++) + buffers[i].addr = DMA_MAPPING_ERROR; + + for_each_sg(sglout, sg, n_sglout, i) { + int y = sg_nctr; + + if (!sg->length) + continue; + + if (left >= sg->length) { + left -= sg->length; + continue; + } + buffers[y].addr = dma_map_single(dev, sg_virt(sg) + left, + sg->length - left, + DMA_BIDIRECTIONAL); + if (unlikely(dma_mapping_error(dev, buffers[y].addr))) + goto err_out; + buffers[y].len = sg->length; + sg_nctr++; + if (left) { + buffers[y].len -= left; + left = 0; + } + } + if (extra_buff) { + buffers[sg_nctr].addr = extra_dst_buff; + buffers[sg_nctr].len = sz_extra_dst_buff; + } + + buflout->num_bufs = sg_nctr; + buflout->num_bufs += extra_buff; + buflout->num_mapped_bufs = sg_nctr; + bloutp = dma_map_single(dev, buflout, sz_out, DMA_TO_DEVICE); + if (unlikely(dma_mapping_error(dev, bloutp))) + goto err_out; + buf->blout = buflout; + buf->bloutp = bloutp; + buf->sz_out = sz_out; + } else { + /* Otherwise set the src and dst to the same address */ + buf->bloutp = buf->blp; + buf->sz_out = 0; + } + return 0; + +err_out: + if (!dma_mapping_error(dev, bloutp)) + dma_unmap_single(dev, bloutp, sz_out, DMA_TO_DEVICE); + + n = sg_nents(sglout); + for (i = 0; i < n; i++) { + if (buflout->buffers[i].addr == extra_dst_buff) + break; + if (!dma_mapping_error(dev, buflout->buffers[i].addr)) + dma_unmap_single(dev, buflout->buffers[i].addr, + buflout->buffers[i].len, + DMA_BIDIRECTIONAL); + } + + if (!buf->sgl_dst_valid) + kfree(buflout); + +err_in: + if (!dma_mapping_error(dev, blp)) + dma_unmap_single(dev, blp, sz, DMA_TO_DEVICE); + + n = sg_nents(sgl); + for (i = 0; i < n; i++) + if (!dma_mapping_error(dev, bufl->buffers[i].addr)) + dma_unmap_single(dev, bufl->buffers[i].addr, + bufl->buffers[i].len, + bufl_dma_dir); + + if (!buf->sgl_src_valid) + kfree(bufl); + + dev_err(dev, "Failed to map buf for dma\n"); + return -ENOMEM; +} + +int qat_bl_sgl_to_bufl(struct adf_accel_dev *accel_dev, + struct scatterlist *sgl, + struct scatterlist *sglout, + struct qat_request_buffs *buf, + struct qat_sgl_to_bufl_params *params, + gfp_t flags) +{ + dma_addr_t extra_dst_buff = 0; + size_t sz_extra_dst_buff = 0; + unsigned int sskip = 0; + unsigned int dskip = 0; + + if (params) { + extra_dst_buff = params->extra_dst_buff; + sz_extra_dst_buff = params->sz_extra_dst_buff; + sskip = params->sskip; + dskip = params->dskip; + } + + return __qat_bl_sgl_to_bufl(accel_dev, sgl, sglout, buf, + extra_dst_buff, sz_extra_dst_buff, + sskip, dskip, flags); +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_bl.h b/drivers/crypto/intel/qat/qat_common/qat_bl.h new file mode 100644 index 000000000000..2827d5055d3c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_bl.h @@ -0,0 +1,68 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2014 - 2022 Intel Corporation */ +#ifndef QAT_BL_H +#define QAT_BL_H +#include <linux/crypto.h> +#include <linux/scatterlist.h> +#include <linux/types.h> + +#define QAT_MAX_BUFF_DESC 4 + +struct qat_alg_buf { + u32 len; + u32 resrvd; + u64 addr; +} __packed; + +struct qat_alg_buf_list { + /* New members must be added within the __struct_group() macro below. */ + __struct_group(qat_alg_buf_list_hdr, hdr, __packed, + u64 resrvd; + u32 num_bufs; + u32 num_mapped_bufs; + ); + struct qat_alg_buf buffers[]; +} __packed; +static_assert(offsetof(struct qat_alg_buf_list, buffers) == sizeof(struct qat_alg_buf_list_hdr), + "struct member likely outside of __struct_group()"); + +struct qat_alg_fixed_buf_list { + struct qat_alg_buf_list_hdr sgl_hdr; + struct qat_alg_buf descriptors[QAT_MAX_BUFF_DESC]; +} __packed __aligned(64); + +struct qat_request_buffs { + struct qat_alg_buf_list *bl; + dma_addr_t blp; + struct qat_alg_buf_list *blout; + dma_addr_t bloutp; + size_t sz; + size_t sz_out; + bool sgl_src_valid; + bool sgl_dst_valid; + struct qat_alg_fixed_buf_list sgl_src; + struct qat_alg_fixed_buf_list sgl_dst; +}; + +struct qat_sgl_to_bufl_params { + dma_addr_t extra_dst_buff; + size_t sz_extra_dst_buff; + unsigned int sskip; + unsigned int dskip; +}; + +void qat_bl_free_bufl(struct adf_accel_dev *accel_dev, + struct qat_request_buffs *buf); +int qat_bl_sgl_to_bufl(struct adf_accel_dev *accel_dev, + struct scatterlist *sgl, + struct scatterlist *sglout, + struct qat_request_buffs *buf, + struct qat_sgl_to_bufl_params *params, + gfp_t flags); + +static inline gfp_t qat_algs_alloc_flags(struct crypto_async_request *req) +{ + return req->flags & CRYPTO_TFM_REQ_MAY_SLEEP ? GFP_KERNEL : GFP_ATOMIC; +} + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/qat_comp_algs.c b/drivers/crypto/intel/qat/qat_common/qat_comp_algs.c new file mode 100644 index 000000000000..8b123472b71c --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_comp_algs.c @@ -0,0 +1,269 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2022 Intel Corporation */ +#include <linux/crypto.h> +#include <crypto/acompress.h> +#include <crypto/internal/acompress.h> +#include <crypto/scatterwalk.h> +#include <linux/dma-mapping.h> +#include <linux/workqueue.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_dc.h" +#include "qat_bl.h" +#include "qat_comp_req.h" +#include "qat_compression.h" +#include "qat_algs_send.h" + +static DEFINE_MUTEX(algs_lock); +static unsigned int active_devs; + +enum direction { + DECOMPRESSION = 0, + COMPRESSION = 1, +}; + +struct qat_compression_req; + +struct qat_compression_ctx { + u8 comp_ctx[QAT_COMP_CTX_SIZE]; + struct qat_compression_instance *inst; + int (*qat_comp_callback)(struct qat_compression_req *qat_req, void *resp); +}; + +struct qat_compression_req { + u8 req[QAT_COMP_REQ_SIZE]; + struct qat_compression_ctx *qat_compression_ctx; + struct acomp_req *acompress_req; + struct qat_request_buffs buf; + enum direction dir; + int actual_dlen; + struct qat_alg_req alg_req; +}; + +static int qat_alg_send_dc_message(struct qat_compression_req *qat_req, + struct qat_compression_instance *inst, + struct crypto_async_request *base) +{ + struct qat_alg_req *alg_req = &qat_req->alg_req; + + alg_req->fw_req = (u32 *)&qat_req->req; + alg_req->tx_ring = inst->dc_tx; + alg_req->base = base; + alg_req->backlog = &inst->backlog; + + return qat_alg_send_message(alg_req); +} + +static void qat_comp_generic_callback(struct qat_compression_req *qat_req, + void *resp) +{ + struct acomp_req *areq = qat_req->acompress_req; + struct qat_compression_ctx *ctx = qat_req->qat_compression_ctx; + struct adf_accel_dev *accel_dev = ctx->inst->accel_dev; + struct crypto_acomp *tfm = crypto_acomp_reqtfm(areq); + struct qat_compression_instance *inst = ctx->inst; + int consumed, produced; + s8 cmp_err, xlt_err; + int res = -EBADMSG; + int status; + u8 cnv; + + status = qat_comp_get_cmp_status(resp); + status |= qat_comp_get_xlt_status(resp); + cmp_err = qat_comp_get_cmp_err(resp); + xlt_err = qat_comp_get_xlt_err(resp); + + consumed = qat_comp_get_consumed_ctr(resp); + produced = qat_comp_get_produced_ctr(resp); + + dev_dbg(&GET_DEV(accel_dev), + "[%s][%s][%s] slen = %8d dlen = %8d consumed = %8d produced = %8d cmp_err = %3d xlt_err = %3d", + crypto_tfm_alg_driver_name(crypto_acomp_tfm(tfm)), + qat_req->dir == COMPRESSION ? "comp " : "decomp", + status ? "ERR" : "OK ", + areq->slen, areq->dlen, consumed, produced, cmp_err, xlt_err); + + areq->dlen = 0; + + if (unlikely(status != ICP_QAT_FW_COMN_STATUS_FLAG_OK)) + goto end; + + if (qat_req->dir == COMPRESSION) { + cnv = qat_comp_get_cmp_cnv_flag(resp); + if (unlikely(!cnv)) { + dev_err(&GET_DEV(accel_dev), + "Verified compression not supported\n"); + goto end; + } + + if (unlikely(produced > qat_req->actual_dlen)) { + memset(inst->dc_data->ovf_buff, 0, + inst->dc_data->ovf_buff_sz); + dev_dbg(&GET_DEV(accel_dev), + "Actual buffer overflow: produced=%d, dlen=%d\n", + produced, qat_req->actual_dlen); + goto end; + } + } + + res = 0; + areq->dlen = produced; + + if (ctx->qat_comp_callback) + res = ctx->qat_comp_callback(qat_req, resp); + +end: + qat_bl_free_bufl(accel_dev, &qat_req->buf); + acomp_request_complete(areq, res); +} + +void qat_comp_alg_callback(void *resp) +{ + struct qat_compression_req *qat_req = + (void *)(__force long)qat_comp_get_opaque(resp); + struct qat_instance_backlog *backlog = qat_req->alg_req.backlog; + + qat_comp_generic_callback(qat_req, resp); + + qat_alg_send_backlog(backlog); +} + +static int qat_comp_alg_init_tfm(struct crypto_acomp *acomp_tfm) +{ + struct crypto_tfm *tfm = crypto_acomp_tfm(acomp_tfm); + struct qat_compression_ctx *ctx = crypto_tfm_ctx(tfm); + struct qat_compression_instance *inst; + int node; + + if (tfm->node == NUMA_NO_NODE) + node = numa_node_id(); + else + node = tfm->node; + + memset(ctx, 0, sizeof(*ctx)); + inst = qat_compression_get_instance_node(node); + if (!inst) + return -EINVAL; + ctx->inst = inst; + + return qat_comp_build_ctx(inst->accel_dev, ctx->comp_ctx, QAT_DEFLATE); +} + +static void qat_comp_alg_exit_tfm(struct crypto_acomp *acomp_tfm) +{ + struct crypto_tfm *tfm = crypto_acomp_tfm(acomp_tfm); + struct qat_compression_ctx *ctx = crypto_tfm_ctx(tfm); + + qat_compression_put_instance(ctx->inst); + memset(ctx, 0, sizeof(*ctx)); +} + +static int qat_comp_alg_compress_decompress(struct acomp_req *areq, enum direction dir, + unsigned int shdr, unsigned int sftr, + unsigned int dhdr, unsigned int dftr) +{ + struct qat_compression_req *qat_req = acomp_request_ctx(areq); + struct crypto_acomp *acomp_tfm = crypto_acomp_reqtfm(areq); + struct crypto_tfm *tfm = crypto_acomp_tfm(acomp_tfm); + struct qat_compression_ctx *ctx = crypto_tfm_ctx(tfm); + struct qat_compression_instance *inst = ctx->inst; + gfp_t f = qat_algs_alloc_flags(&areq->base); + struct qat_sgl_to_bufl_params params = {0}; + int slen = areq->slen - shdr - sftr; + int dlen = areq->dlen - dhdr - dftr; + dma_addr_t sfbuf, dfbuf; + u8 *req = qat_req->req; + size_t ovf_buff_sz; + int ret; + + params.sskip = shdr; + params.dskip = dhdr; + + if (!areq->src || !slen) + return -EINVAL; + + if (!areq->dst || !dlen) + return -EINVAL; + + if (dir == COMPRESSION) { + params.extra_dst_buff = inst->dc_data->ovf_buff_p; + ovf_buff_sz = inst->dc_data->ovf_buff_sz; + params.sz_extra_dst_buff = ovf_buff_sz; + } + + ret = qat_bl_sgl_to_bufl(ctx->inst->accel_dev, areq->src, areq->dst, + &qat_req->buf, ¶ms, f); + if (unlikely(ret)) + return ret; + + sfbuf = qat_req->buf.blp; + dfbuf = qat_req->buf.bloutp; + qat_req->qat_compression_ctx = ctx; + qat_req->acompress_req = areq; + qat_req->dir = dir; + + if (dir == COMPRESSION) { + qat_req->actual_dlen = dlen; + dlen += ovf_buff_sz; + qat_comp_create_compression_req(ctx->comp_ctx, req, + (u64)(__force long)sfbuf, slen, + (u64)(__force long)dfbuf, dlen, + (u64)(__force long)qat_req); + } else { + qat_comp_create_decompression_req(ctx->comp_ctx, req, + (u64)(__force long)sfbuf, slen, + (u64)(__force long)dfbuf, dlen, + (u64)(__force long)qat_req); + } + + ret = qat_alg_send_dc_message(qat_req, inst, &areq->base); + if (ret == -ENOSPC) + qat_bl_free_bufl(inst->accel_dev, &qat_req->buf); + + return ret; +} + +static int qat_comp_alg_compress(struct acomp_req *req) +{ + return qat_comp_alg_compress_decompress(req, COMPRESSION, 0, 0, 0, 0); +} + +static int qat_comp_alg_decompress(struct acomp_req *req) +{ + return qat_comp_alg_compress_decompress(req, DECOMPRESSION, 0, 0, 0, 0); +} + +static struct acomp_alg qat_acomp[] = { { + .base = { + .cra_name = "deflate", + .cra_driver_name = "qat_deflate", + .cra_priority = 4001, + .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_ALLOCATES_MEMORY, + .cra_ctxsize = sizeof(struct qat_compression_ctx), + .cra_reqsize = sizeof(struct qat_compression_req), + .cra_module = THIS_MODULE, + }, + .init = qat_comp_alg_init_tfm, + .exit = qat_comp_alg_exit_tfm, + .compress = qat_comp_alg_compress, + .decompress = qat_comp_alg_decompress, +}}; + +int qat_comp_algs_register(void) +{ + int ret = 0; + + mutex_lock(&algs_lock); + if (++active_devs == 1) + ret = crypto_register_acomps(qat_acomp, ARRAY_SIZE(qat_acomp)); + mutex_unlock(&algs_lock); + return ret; +} + +void qat_comp_algs_unregister(void) +{ + mutex_lock(&algs_lock); + if (--active_devs == 0) + crypto_unregister_acomps(qat_acomp, ARRAY_SIZE(qat_acomp)); + mutex_unlock(&algs_lock); +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_comp_req.h b/drivers/crypto/intel/qat/qat_common/qat_comp_req.h new file mode 100644 index 000000000000..18a1f33a6db9 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_comp_req.h @@ -0,0 +1,113 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef _QAT_COMP_REQ_H_ +#define _QAT_COMP_REQ_H_ + +#include "icp_qat_fw_comp.h" + +#define QAT_COMP_REQ_SIZE (sizeof(struct icp_qat_fw_comp_req)) +#define QAT_COMP_CTX_SIZE (QAT_COMP_REQ_SIZE * 2) + +static inline void qat_comp_create_req(void *ctx, void *req, u64 src, u32 slen, + u64 dst, u32 dlen, u64 opaque) +{ + struct icp_qat_fw_comp_req *fw_tmpl = ctx; + struct icp_qat_fw_comp_req *fw_req = req; + struct icp_qat_fw_comp_req_params *req_pars = &fw_req->comp_pars; + + memcpy(fw_req, fw_tmpl, sizeof(*fw_req)); + fw_req->comn_mid.src_data_addr = src; + fw_req->comn_mid.src_length = slen; + fw_req->comn_mid.dest_data_addr = dst; + fw_req->comn_mid.dst_length = dlen; + fw_req->comn_mid.opaque_data = opaque; + req_pars->comp_len = slen; + req_pars->out_buffer_sz = dlen; +} + +static inline void qat_comp_create_compression_req(void *ctx, void *req, + u64 src, u32 slen, + u64 dst, u32 dlen, + u64 opaque) +{ + qat_comp_create_req(ctx, req, src, slen, dst, dlen, opaque); +} + +static inline void qat_comp_create_decompression_req(void *ctx, void *req, + u64 src, u32 slen, + u64 dst, u32 dlen, + u64 opaque) +{ + struct icp_qat_fw_comp_req *fw_tmpl = ctx; + + fw_tmpl++; + qat_comp_create_req(fw_tmpl, req, src, slen, dst, dlen, opaque); +} + +static inline u32 qat_comp_get_consumed_ctr(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + + return qat_resp->comp_resp_pars.input_byte_counter; +} + +static inline u32 qat_comp_get_produced_ctr(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + + return qat_resp->comp_resp_pars.output_byte_counter; +} + +static inline u32 qat_comp_get_produced_adler32(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + + return qat_resp->comp_resp_pars.crc.legacy.curr_adler_32; +} + +static inline u64 qat_comp_get_opaque(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + + return qat_resp->opaque_data; +} + +static inline s8 qat_comp_get_cmp_err(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + + return qat_resp->comn_resp.comn_error.cmp_err_code; +} + +static inline s8 qat_comp_get_xlt_err(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + + return qat_resp->comn_resp.comn_error.xlat_err_code; +} + +static inline s8 qat_comp_get_cmp_status(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + u8 stat_filed = qat_resp->comn_resp.comn_status; + + return ICP_QAT_FW_COMN_RESP_CMP_STAT_GET(stat_filed); +} + +static inline s8 qat_comp_get_xlt_status(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + u8 stat_filed = qat_resp->comn_resp.comn_status; + + return ICP_QAT_FW_COMN_RESP_XLAT_STAT_GET(stat_filed); +} + +static inline u8 qat_comp_get_cmp_cnv_flag(void *resp) +{ + struct icp_qat_fw_comp_resp *qat_resp = resp; + u8 flags = qat_resp->comn_resp.hdr_flags; + + return ICP_QAT_FW_COMN_HDR_CNV_FLAG_GET(flags); +} + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/qat_compression.c b/drivers/crypto/intel/qat/qat_common/qat_compression.c new file mode 100644 index 000000000000..53a4db5507ec --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_compression.c @@ -0,0 +1,295 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2022 Intel Corporation */ +#include <linux/module.h> +#include <linux/slab.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_transport.h" +#include "adf_transport_access_macros.h" +#include "adf_cfg.h" +#include "adf_cfg_strings.h" +#include "qat_compression.h" +#include "icp_qat_fw.h" + +#define SEC ADF_KERNEL_SEC + +static struct service_hndl qat_compression; + +void qat_compression_put_instance(struct qat_compression_instance *inst) +{ + atomic_dec(&inst->refctr); + adf_dev_put(inst->accel_dev); +} + +static int qat_compression_free_instances(struct adf_accel_dev *accel_dev) +{ + struct qat_compression_instance *inst; + struct list_head *list_ptr, *tmp; + int i; + + list_for_each_safe(list_ptr, tmp, &accel_dev->compression_list) { + inst = list_entry(list_ptr, + struct qat_compression_instance, list); + + for (i = 0; i < atomic_read(&inst->refctr); i++) + qat_compression_put_instance(inst); + + if (inst->dc_tx) + adf_remove_ring(inst->dc_tx); + + if (inst->dc_rx) + adf_remove_ring(inst->dc_rx); + + list_del(list_ptr); + kfree(inst); + } + return 0; +} + +struct qat_compression_instance *qat_compression_get_instance_node(int node) +{ + struct qat_compression_instance *inst = NULL; + struct adf_accel_dev *accel_dev = NULL; + unsigned long best = ~0; + struct list_head *itr; + + list_for_each(itr, adf_devmgr_get_head()) { + struct adf_accel_dev *tmp_dev; + unsigned long ctr; + int tmp_dev_node; + + tmp_dev = list_entry(itr, struct adf_accel_dev, list); + tmp_dev_node = dev_to_node(&GET_DEV(tmp_dev)); + + if ((node == tmp_dev_node || tmp_dev_node < 0) && + adf_dev_started(tmp_dev) && !list_empty(&tmp_dev->compression_list)) { + ctr = atomic_read(&tmp_dev->ref_count); + if (best > ctr) { + accel_dev = tmp_dev; + best = ctr; + } + } + } + + if (!accel_dev) { + pr_debug_ratelimited("QAT: Could not find a device on node %d\n", node); + /* Get any started device */ + list_for_each(itr, adf_devmgr_get_head()) { + struct adf_accel_dev *tmp_dev; + + tmp_dev = list_entry(itr, struct adf_accel_dev, list); + if (adf_dev_started(tmp_dev) && + !list_empty(&tmp_dev->compression_list)) { + accel_dev = tmp_dev; + break; + } + } + } + + if (!accel_dev) + return NULL; + + best = ~0; + list_for_each(itr, &accel_dev->compression_list) { + struct qat_compression_instance *tmp_inst; + unsigned long ctr; + + tmp_inst = list_entry(itr, struct qat_compression_instance, list); + ctr = atomic_read(&tmp_inst->refctr); + if (best > ctr) { + inst = tmp_inst; + best = ctr; + } + } + if (inst) { + if (adf_dev_get(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Could not increment dev refctr\n"); + return NULL; + } + atomic_inc(&inst->refctr); + } + return inst; +} + +static int qat_compression_create_instances(struct adf_accel_dev *accel_dev) +{ + struct qat_compression_instance *inst; + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + char val[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + unsigned long num_inst, num_msg_dc; + unsigned long bank; + int msg_size; + int ret; + int i; + + INIT_LIST_HEAD(&accel_dev->compression_list); + strscpy(key, ADF_NUM_DC, sizeof(key)); + ret = adf_cfg_get_param_value(accel_dev, SEC, key, val); + if (ret) + return ret; + + ret = kstrtoul(val, 10, &num_inst); + if (ret) + return ret; + + for (i = 0; i < num_inst; i++) { + inst = kzalloc_node(sizeof(*inst), GFP_KERNEL, + dev_to_node(&GET_DEV(accel_dev))); + if (!inst) { + ret = -ENOMEM; + goto err; + } + + list_add_tail(&inst->list, &accel_dev->compression_list); + inst->id = i; + atomic_set(&inst->refctr, 0); + inst->accel_dev = accel_dev; + + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_BANK_NUM, i); + ret = adf_cfg_get_param_value(accel_dev, SEC, key, val); + if (ret) + return ret; + + ret = kstrtoul(val, 10, &bank); + if (ret) + return ret; + + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_SIZE, i); + ret = adf_cfg_get_param_value(accel_dev, SEC, key, val); + if (ret) + return ret; + + ret = kstrtoul(val, 10, &num_msg_dc); + if (ret) + return ret; + + msg_size = ICP_QAT_FW_REQ_DEFAULT_SZ; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_TX, i); + ret = adf_create_ring(accel_dev, SEC, bank, num_msg_dc, + msg_size, key, NULL, 0, &inst->dc_tx); + if (ret) + return ret; + + msg_size = ICP_QAT_FW_RESP_DEFAULT_SZ; + snprintf(key, sizeof(key), ADF_DC "%d" ADF_RING_DC_RX, i); + ret = adf_create_ring(accel_dev, SEC, bank, num_msg_dc, + msg_size, key, qat_comp_alg_callback, 0, + &inst->dc_rx); + if (ret) + return ret; + + inst->dc_data = accel_dev->dc_data; + INIT_LIST_HEAD(&inst->backlog.list); + spin_lock_init(&inst->backlog.lock); + } + return 0; +err: + qat_compression_free_instances(accel_dev); + return ret; +} + +static int qat_compression_alloc_dc_data(struct adf_accel_dev *accel_dev) +{ + struct device *dev = &GET_DEV(accel_dev); + dma_addr_t obuff_p = DMA_MAPPING_ERROR; + size_t ovf_buff_sz = QAT_COMP_MAX_SKID; + struct adf_dc_data *dc_data = NULL; + u8 *obuff = NULL; + + dc_data = kzalloc_node(sizeof(*dc_data), GFP_KERNEL, dev_to_node(dev)); + if (!dc_data) + goto err; + + obuff = kzalloc_node(ovf_buff_sz, GFP_KERNEL, dev_to_node(dev)); + if (!obuff) + goto err; + + obuff_p = dma_map_single(dev, obuff, ovf_buff_sz, DMA_BIDIRECTIONAL); + if (unlikely(dma_mapping_error(dev, obuff_p))) + goto err; + + dc_data->ovf_buff = obuff; + dc_data->ovf_buff_p = obuff_p; + dc_data->ovf_buff_sz = ovf_buff_sz; + + accel_dev->dc_data = dc_data; + + return 0; + +err: + accel_dev->dc_data = NULL; + kfree(obuff); + devm_kfree(dev, dc_data); + return -ENOMEM; +} + +static void qat_free_dc_data(struct adf_accel_dev *accel_dev) +{ + struct adf_dc_data *dc_data = accel_dev->dc_data; + struct device *dev = &GET_DEV(accel_dev); + + if (!dc_data) + return; + + dma_unmap_single(dev, dc_data->ovf_buff_p, dc_data->ovf_buff_sz, + DMA_BIDIRECTIONAL); + kfree_sensitive(dc_data->ovf_buff); + kfree(dc_data); + accel_dev->dc_data = NULL; +} + +static int qat_compression_init(struct adf_accel_dev *accel_dev) +{ + int ret; + + ret = qat_compression_alloc_dc_data(accel_dev); + if (ret) + return ret; + + ret = qat_compression_create_instances(accel_dev); + if (ret) + qat_free_dc_data(accel_dev); + + return ret; +} + +static int qat_compression_shutdown(struct adf_accel_dev *accel_dev) +{ + qat_free_dc_data(accel_dev); + return qat_compression_free_instances(accel_dev); +} + +static int qat_compression_event_handler(struct adf_accel_dev *accel_dev, + enum adf_event event) +{ + int ret; + + switch (event) { + case ADF_EVENT_INIT: + ret = qat_compression_init(accel_dev); + break; + case ADF_EVENT_SHUTDOWN: + ret = qat_compression_shutdown(accel_dev); + break; + case ADF_EVENT_RESTARTING: + case ADF_EVENT_RESTARTED: + case ADF_EVENT_START: + case ADF_EVENT_STOP: + default: + ret = 0; + } + return ret; +} + +int qat_compression_register(void) +{ + memset(&qat_compression, 0, sizeof(qat_compression)); + qat_compression.event_hld = qat_compression_event_handler; + qat_compression.name = "qat_compression"; + return adf_service_register(&qat_compression); +} + +int qat_compression_unregister(void) +{ + return adf_service_unregister(&qat_compression); +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_compression.h b/drivers/crypto/intel/qat/qat_common/qat_compression.h new file mode 100644 index 000000000000..5ced3ed0e5ea --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_compression.h @@ -0,0 +1,36 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright(c) 2022 Intel Corporation */ +#ifndef _QAT_COMPRESSION_H_ +#define _QAT_COMPRESSION_H_ + +#include <linux/list.h> +#include <linux/types.h> +#include "adf_accel_devices.h" +#include "qat_algs_send.h" + +#define QAT_COMP_MAX_SKID 4096 + +struct qat_compression_instance { + struct adf_etr_ring_data *dc_tx; + struct adf_etr_ring_data *dc_rx; + struct adf_accel_dev *accel_dev; + struct list_head list; + unsigned long state; + int id; + atomic_t refctr; + struct qat_instance_backlog backlog; + struct adf_dc_data *dc_data; +}; + +static inline bool adf_hw_dev_has_compression(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + u32 mask = ~hw_device->accel_capabilities_mask; + + if (mask & ADF_ACCEL_CAPABILITIES_COMPRESSION) + return false; + + return true; +} + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/qat_crypto.c b/drivers/crypto/intel/qat/qat_common/qat_crypto.c new file mode 100644 index 000000000000..101c6ea41673 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_crypto.c @@ -0,0 +1,287 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/module.h> +#include <linux/slab.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "adf_transport.h" +#include "adf_cfg.h" +#include "adf_cfg_strings.h" +#include "adf_gen2_hw_data.h" +#include "qat_crypto.h" +#include "icp_qat_fw.h" + +#define SEC ADF_KERNEL_SEC + +static struct service_hndl qat_crypto; + +void qat_crypto_put_instance(struct qat_crypto_instance *inst) +{ + atomic_dec(&inst->refctr); + adf_dev_put(inst->accel_dev); +} + +static int qat_crypto_free_instances(struct adf_accel_dev *accel_dev) +{ + struct qat_crypto_instance *inst, *tmp; + int i; + + list_for_each_entry_safe(inst, tmp, &accel_dev->crypto_list, list) { + for (i = 0; i < atomic_read(&inst->refctr); i++) + qat_crypto_put_instance(inst); + + if (inst->sym_tx) + adf_remove_ring(inst->sym_tx); + + if (inst->sym_rx) + adf_remove_ring(inst->sym_rx); + + if (inst->pke_tx) + adf_remove_ring(inst->pke_tx); + + if (inst->pke_rx) + adf_remove_ring(inst->pke_rx); + + list_del(&inst->list); + kfree(inst); + } + return 0; +} + +struct qat_crypto_instance *qat_crypto_get_instance_node(int node) +{ + struct adf_accel_dev *accel_dev = NULL, *tmp_dev; + struct qat_crypto_instance *inst = NULL, *tmp_inst; + unsigned long best = ~0; + + list_for_each_entry(tmp_dev, adf_devmgr_get_head(), list) { + unsigned long ctr; + + if ((node == dev_to_node(&GET_DEV(tmp_dev)) || + dev_to_node(&GET_DEV(tmp_dev)) < 0) && + adf_dev_started(tmp_dev) && + !list_empty(&tmp_dev->crypto_list)) { + ctr = atomic_read(&tmp_dev->ref_count); + if (best > ctr) { + accel_dev = tmp_dev; + best = ctr; + } + } + } + + if (!accel_dev) { + pr_debug_ratelimited("QAT: Could not find a device on node %d\n", node); + /* Get any started device */ + list_for_each_entry(tmp_dev, adf_devmgr_get_head(), list) { + if (adf_dev_started(tmp_dev) && + !list_empty(&tmp_dev->crypto_list)) { + accel_dev = tmp_dev; + break; + } + } + } + + if (!accel_dev) + return NULL; + + best = ~0; + list_for_each_entry(tmp_inst, &accel_dev->crypto_list, list) { + unsigned long ctr; + + ctr = atomic_read(&tmp_inst->refctr); + if (best > ctr) { + inst = tmp_inst; + best = ctr; + } + } + if (inst) { + if (adf_dev_get(accel_dev)) { + dev_err(&GET_DEV(accel_dev), "Could not increment dev refctr\n"); + return NULL; + } + atomic_inc(&inst->refctr); + } + return inst; +} + +/** + * qat_crypto_vf_dev_config() - create dev config required to create + * crypto inst. + * + * @accel_dev: Pointer to acceleration device. + * + * Function creates device configuration required to create + * asym, sym or, crypto instances + * + * Return: 0 on success, error code otherwise. + */ +int qat_crypto_vf_dev_config(struct adf_accel_dev *accel_dev) +{ + u16 ring_to_svc_map = GET_HW_DATA(accel_dev)->ring_to_svc_map; + + if (ring_to_svc_map != ADF_GEN2_DEFAULT_RING_TO_SRV_MAP) { + dev_err(&GET_DEV(accel_dev), + "Unsupported ring/service mapping present on PF"); + return -EFAULT; + } + + return GET_HW_DATA(accel_dev)->dev_config(accel_dev); +} + +static int qat_crypto_create_instances(struct adf_accel_dev *accel_dev) +{ + unsigned long num_inst, num_msg_sym, num_msg_asym; + char key[ADF_CFG_MAX_KEY_LEN_IN_BYTES]; + char val[ADF_CFG_MAX_VAL_LEN_IN_BYTES]; + unsigned long sym_bank, asym_bank; + struct qat_crypto_instance *inst; + int msg_size; + int ret; + int i; + + INIT_LIST_HEAD(&accel_dev->crypto_list); + ret = adf_cfg_get_param_value(accel_dev, SEC, ADF_NUM_CY, val); + if (ret) + return ret; + + ret = kstrtoul(val, 0, &num_inst); + if (ret) + return ret; + + for (i = 0; i < num_inst; i++) { + inst = kzalloc_node(sizeof(*inst), GFP_KERNEL, + dev_to_node(&GET_DEV(accel_dev))); + if (!inst) { + ret = -ENOMEM; + goto err; + } + + list_add_tail(&inst->list, &accel_dev->crypto_list); + inst->id = i; + atomic_set(&inst->refctr, 0); + inst->accel_dev = accel_dev; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_BANK_NUM, i); + ret = adf_cfg_get_param_value(accel_dev, SEC, key, val); + if (ret) + goto err; + + ret = kstrtoul(val, 10, &sym_bank); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_BANK_NUM, i); + ret = adf_cfg_get_param_value(accel_dev, SEC, key, val); + if (ret) + goto err; + + ret = kstrtoul(val, 10, &asym_bank); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_SIZE, i); + ret = adf_cfg_get_param_value(accel_dev, SEC, key, val); + if (ret) + goto err; + + ret = kstrtoul(val, 10, &num_msg_sym); + if (ret) + goto err; + + num_msg_sym = num_msg_sym >> 1; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_SIZE, i); + ret = adf_cfg_get_param_value(accel_dev, SEC, key, val); + if (ret) + goto err; + + ret = kstrtoul(val, 10, &num_msg_asym); + if (ret) + goto err; + num_msg_asym = num_msg_asym >> 1; + + msg_size = ICP_QAT_FW_REQ_DEFAULT_SZ; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_TX, i); + ret = adf_create_ring(accel_dev, SEC, sym_bank, num_msg_sym, + msg_size, key, NULL, 0, &inst->sym_tx); + if (ret) + goto err; + + msg_size = msg_size >> 1; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_TX, i); + ret = adf_create_ring(accel_dev, SEC, asym_bank, num_msg_asym, + msg_size, key, NULL, 0, &inst->pke_tx); + if (ret) + goto err; + + msg_size = ICP_QAT_FW_RESP_DEFAULT_SZ; + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_SYM_RX, i); + ret = adf_create_ring(accel_dev, SEC, sym_bank, num_msg_sym, + msg_size, key, qat_alg_callback, 0, + &inst->sym_rx); + if (ret) + goto err; + + snprintf(key, sizeof(key), ADF_CY "%d" ADF_RING_ASYM_RX, i); + ret = adf_create_ring(accel_dev, SEC, asym_bank, num_msg_asym, + msg_size, key, qat_alg_asym_callback, 0, + &inst->pke_rx); + if (ret) + goto err; + + INIT_LIST_HEAD(&inst->backlog.list); + spin_lock_init(&inst->backlog.lock); + } + return 0; +err: + qat_crypto_free_instances(accel_dev); + return ret; +} + +static int qat_crypto_init(struct adf_accel_dev *accel_dev) +{ + if (qat_crypto_create_instances(accel_dev)) + return -EFAULT; + + return 0; +} + +static int qat_crypto_shutdown(struct adf_accel_dev *accel_dev) +{ + return qat_crypto_free_instances(accel_dev); +} + +static int qat_crypto_event_handler(struct adf_accel_dev *accel_dev, + enum adf_event event) +{ + int ret; + + switch (event) { + case ADF_EVENT_INIT: + ret = qat_crypto_init(accel_dev); + break; + case ADF_EVENT_SHUTDOWN: + ret = qat_crypto_shutdown(accel_dev); + break; + case ADF_EVENT_RESTARTING: + case ADF_EVENT_RESTARTED: + case ADF_EVENT_START: + case ADF_EVENT_STOP: + default: + ret = 0; + } + return ret; +} + +int qat_crypto_register(void) +{ + memset(&qat_crypto, 0, sizeof(qat_crypto)); + qat_crypto.event_hld = qat_crypto_event_handler; + qat_crypto.name = "qat_crypto"; + return adf_service_register(&qat_crypto); +} + +int qat_crypto_unregister(void) +{ + return adf_service_unregister(&qat_crypto); +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_crypto.h b/drivers/crypto/intel/qat/qat_common/qat_crypto.h new file mode 100644 index 000000000000..6a0e961bb9dc --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_crypto.h @@ -0,0 +1,68 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef _QAT_CRYPTO_INSTANCE_H_ +#define _QAT_CRYPTO_INSTANCE_H_ + +#include <crypto/aes.h> +#include <linux/list.h> +#include <linux/slab.h> +#include "adf_accel_devices.h" +#include "icp_qat_fw_la.h" +#include "qat_algs_send.h" +#include "qat_bl.h" + +struct qat_crypto_instance { + struct adf_etr_ring_data *sym_tx; + struct adf_etr_ring_data *sym_rx; + struct adf_etr_ring_data *pke_tx; + struct adf_etr_ring_data *pke_rx; + struct adf_accel_dev *accel_dev; + struct list_head list; + unsigned long state; + int id; + atomic_t refctr; + struct qat_instance_backlog backlog; +}; + +struct qat_crypto_request; + +struct qat_crypto_request { + struct icp_qat_fw_la_bulk_req req; + union { + struct qat_alg_aead_ctx *aead_ctx; + struct qat_alg_skcipher_ctx *skcipher_ctx; + }; + union { + struct aead_request *aead_req; + struct skcipher_request *skcipher_req; + }; + struct qat_request_buffs buf; + void (*cb)(struct icp_qat_fw_la_resp *resp, + struct qat_crypto_request *req); + union { + struct { + __be64 iv_hi; + __be64 iv_lo; + }; + u8 iv[AES_BLOCK_SIZE]; + }; + bool encryption; + struct qat_alg_req alg_req; +}; + +static inline bool adf_hw_dev_has_crypto(struct adf_accel_dev *accel_dev) +{ + struct adf_hw_device_data *hw_device = accel_dev->hw_device; + u32 mask = ~hw_device->accel_capabilities_mask; + + if (mask & ADF_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC) + return false; + if (mask & ADF_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC) + return false; + if (mask & ADF_ACCEL_CAPABILITIES_AUTHENTICATION) + return false; + + return true; +} + +#endif diff --git a/drivers/crypto/intel/qat/qat_common/qat_hal.c b/drivers/crypto/intel/qat/qat_common/qat_hal.c new file mode 100644 index 000000000000..da4eca6e1633 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_hal.c @@ -0,0 +1,1601 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/slab.h> +#include <linux/delay.h> +#include <linux/pci_ids.h> + +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "icp_qat_hal.h" +#include "icp_qat_uclo.h" + +#define BAD_REGADDR 0xffff +#define MAX_RETRY_TIMES 10000 +#define INIT_CTX_ARB_VALUE 0x0 +#define INIT_CTX_ENABLE_VALUE 0x0 +#define INIT_PC_VALUE 0x0 +#define INIT_WAKEUP_EVENTS_VALUE 0x1 +#define INIT_SIG_EVENTS_VALUE 0x1 +#define INIT_CCENABLE_VALUE 0x2000 +#define RST_CSR_QAT_LSB 20 +#define RST_CSR_AE_LSB 0 +#define MC_TIMESTAMP_ENABLE (0x1 << 7) + +#define IGNORE_W1C_MASK ((~(1 << CE_BREAKPOINT_BITPOS)) & \ + (~(1 << CE_CNTL_STORE_PARITY_ERROR_BITPOS)) & \ + (~(1 << CE_REG_PAR_ERR_BITPOS))) +#define INSERT_IMMED_GPRA_CONST(inst, const_val) \ + (inst = ((inst & 0xFFFF00C03FFull) | \ + ((((const_val) << 12) & 0x0FF00000ull) | \ + (((const_val) << 10) & 0x0003FC00ull)))) +#define INSERT_IMMED_GPRB_CONST(inst, const_val) \ + (inst = ((inst & 0xFFFF00FFF00ull) | \ + ((((const_val) << 12) & 0x0FF00000ull) | \ + (((const_val) << 0) & 0x000000FFull)))) + +#define AE(handle, ae) ((handle)->hal_handle->aes[ae]) + +static const u64 inst_4b[] = { + 0x0F0400C0000ull, 0x0F4400C0000ull, 0x0F040000300ull, 0x0F440000300ull, + 0x0FC066C0000ull, 0x0F0000C0300ull, 0x0F0000C0300ull, 0x0F0000C0300ull, + 0x0A021000000ull +}; + +static const u64 inst[] = { + 0x0F0000C0000ull, 0x0F000000380ull, 0x0D805000011ull, 0x0FC082C0300ull, + 0x0F0000C0300ull, 0x0F0000C0300ull, 0x0F0000C0300ull, 0x0F0000C0300ull, + 0x0A0643C0000ull, 0x0BAC0000301ull, 0x0D802000101ull, 0x0F0000C0001ull, + 0x0FC066C0001ull, 0x0F0000C0300ull, 0x0F0000C0300ull, 0x0F0000C0300ull, + 0x0F000400300ull, 0x0A0610C0000ull, 0x0BAC0000301ull, 0x0D804400101ull, + 0x0A0580C0000ull, 0x0A0581C0000ull, 0x0A0582C0000ull, 0x0A0583C0000ull, + 0x0A0584C0000ull, 0x0A0585C0000ull, 0x0A0586C0000ull, 0x0A0587C0000ull, + 0x0A0588C0000ull, 0x0A0589C0000ull, 0x0A058AC0000ull, 0x0A058BC0000ull, + 0x0A058CC0000ull, 0x0A058DC0000ull, 0x0A058EC0000ull, 0x0A058FC0000ull, + 0x0A05C0C0000ull, 0x0A05C1C0000ull, 0x0A05C2C0000ull, 0x0A05C3C0000ull, + 0x0A05C4C0000ull, 0x0A05C5C0000ull, 0x0A05C6C0000ull, 0x0A05C7C0000ull, + 0x0A05C8C0000ull, 0x0A05C9C0000ull, 0x0A05CAC0000ull, 0x0A05CBC0000ull, + 0x0A05CCC0000ull, 0x0A05CDC0000ull, 0x0A05CEC0000ull, 0x0A05CFC0000ull, + 0x0A0400C0000ull, 0x0B0400C0000ull, 0x0A0401C0000ull, 0x0B0401C0000ull, + 0x0A0402C0000ull, 0x0B0402C0000ull, 0x0A0403C0000ull, 0x0B0403C0000ull, + 0x0A0404C0000ull, 0x0B0404C0000ull, 0x0A0405C0000ull, 0x0B0405C0000ull, + 0x0A0406C0000ull, 0x0B0406C0000ull, 0x0A0407C0000ull, 0x0B0407C0000ull, + 0x0A0408C0000ull, 0x0B0408C0000ull, 0x0A0409C0000ull, 0x0B0409C0000ull, + 0x0A040AC0000ull, 0x0B040AC0000ull, 0x0A040BC0000ull, 0x0B040BC0000ull, + 0x0A040CC0000ull, 0x0B040CC0000ull, 0x0A040DC0000ull, 0x0B040DC0000ull, + 0x0A040EC0000ull, 0x0B040EC0000ull, 0x0A040FC0000ull, 0x0B040FC0000ull, + 0x0D81581C010ull, 0x0E000010000ull, 0x0E000010000ull, +}; + +void qat_hal_set_live_ctx(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask) +{ + AE(handle, ae).live_ctx_mask = ctx_mask; +} + +#define CSR_RETRY_TIMES 500 +static int qat_hal_rd_ae_csr(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int csr) +{ + unsigned int iterations = CSR_RETRY_TIMES; + int value; + + do { + value = GET_AE_CSR(handle, ae, csr); + if (!(GET_AE_CSR(handle, ae, LOCAL_CSR_STATUS) & LCS_STATUS)) + return value; + } while (iterations--); + + pr_err("QAT: Read CSR timeout\n"); + return 0; +} + +static int qat_hal_wr_ae_csr(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int csr, + unsigned int value) +{ + unsigned int iterations = CSR_RETRY_TIMES; + + do { + SET_AE_CSR(handle, ae, csr, value); + if (!(GET_AE_CSR(handle, ae, LOCAL_CSR_STATUS) & LCS_STATUS)) + return 0; + } while (iterations--); + + pr_err("QAT: Write CSR Timeout\n"); + return -EFAULT; +} + +static void qat_hal_get_wakeup_event(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + unsigned int *events) +{ + unsigned int cur_ctx; + + cur_ctx = qat_hal_rd_ae_csr(handle, ae, CSR_CTX_POINTER); + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, ctx); + *events = qat_hal_rd_ae_csr(handle, ae, CTX_WAKEUP_EVENTS_INDIRECT); + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, cur_ctx); +} + +static int qat_hal_wait_cycles(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int cycles, + int chk_inactive) +{ + unsigned int base_cnt = 0, cur_cnt = 0; + unsigned int csr = (1 << ACS_ABO_BITPOS); + int times = MAX_RETRY_TIMES; + int elapsed_cycles = 0; + + base_cnt = qat_hal_rd_ae_csr(handle, ae, PROFILE_COUNT); + base_cnt &= 0xffff; + while ((int)cycles > elapsed_cycles && times--) { + if (chk_inactive) + csr = qat_hal_rd_ae_csr(handle, ae, ACTIVE_CTX_STATUS); + + cur_cnt = qat_hal_rd_ae_csr(handle, ae, PROFILE_COUNT); + cur_cnt &= 0xffff; + elapsed_cycles = cur_cnt - base_cnt; + + if (elapsed_cycles < 0) + elapsed_cycles += 0x10000; + + /* ensure at least 8 time cycles elapsed in wait_cycles */ + if (elapsed_cycles >= 8 && !(csr & (1 << ACS_ABO_BITPOS))) + return 0; + } + if (times < 0) { + pr_err("QAT: wait_num_cycles time out\n"); + return -EFAULT; + } + return 0; +} + +#define CLR_BIT(wrd, bit) ((wrd) & ~(1 << (bit))) +#define SET_BIT(wrd, bit) ((wrd) | 1 << (bit)) + +int qat_hal_set_ae_ctx_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char mode) +{ + unsigned int csr, new_csr; + + if (mode != 4 && mode != 8) { + pr_err("QAT: bad ctx mode=%d\n", mode); + return -EINVAL; + } + + /* Sets the acceleration engine context mode to either four or eight */ + csr = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + csr = IGNORE_W1C_MASK & csr; + new_csr = (mode == 4) ? + SET_BIT(csr, CE_INUSE_CONTEXTS_BITPOS) : + CLR_BIT(csr, CE_INUSE_CONTEXTS_BITPOS); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, new_csr); + return 0; +} + +int qat_hal_set_ae_nn_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char mode) +{ + unsigned int csr, new_csr; + + csr = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + csr &= IGNORE_W1C_MASK; + + new_csr = (mode) ? + SET_BIT(csr, CE_NN_MODE_BITPOS) : + CLR_BIT(csr, CE_NN_MODE_BITPOS); + + if (new_csr != csr) + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, new_csr); + + return 0; +} + +int qat_hal_set_ae_lm_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, enum icp_qat_uof_regtype lm_type, + unsigned char mode) +{ + unsigned int csr, new_csr; + + csr = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + csr &= IGNORE_W1C_MASK; + switch (lm_type) { + case ICP_LMEM0: + new_csr = (mode) ? + SET_BIT(csr, CE_LMADDR_0_GLOBAL_BITPOS) : + CLR_BIT(csr, CE_LMADDR_0_GLOBAL_BITPOS); + break; + case ICP_LMEM1: + new_csr = (mode) ? + SET_BIT(csr, CE_LMADDR_1_GLOBAL_BITPOS) : + CLR_BIT(csr, CE_LMADDR_1_GLOBAL_BITPOS); + break; + case ICP_LMEM2: + new_csr = (mode) ? + SET_BIT(csr, CE_LMADDR_2_GLOBAL_BITPOS) : + CLR_BIT(csr, CE_LMADDR_2_GLOBAL_BITPOS); + break; + case ICP_LMEM3: + new_csr = (mode) ? + SET_BIT(csr, CE_LMADDR_3_GLOBAL_BITPOS) : + CLR_BIT(csr, CE_LMADDR_3_GLOBAL_BITPOS); + break; + default: + pr_err("QAT: lmType = 0x%x\n", lm_type); + return -EINVAL; + } + + if (new_csr != csr) + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, new_csr); + return 0; +} + +void qat_hal_set_ae_tindex_mode(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char mode) +{ + unsigned int csr, new_csr; + + csr = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + csr &= IGNORE_W1C_MASK; + new_csr = (mode) ? + SET_BIT(csr, CE_T_INDEX_GLOBAL_BITPOS) : + CLR_BIT(csr, CE_T_INDEX_GLOBAL_BITPOS); + if (new_csr != csr) + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, new_csr); +} + +static unsigned short qat_hal_get_reg_addr(unsigned int type, + unsigned short reg_num) +{ + unsigned short reg_addr; + + switch (type) { + case ICP_GPA_ABS: + case ICP_GPB_ABS: + reg_addr = 0x80 | (reg_num & 0x7f); + break; + case ICP_GPA_REL: + case ICP_GPB_REL: + reg_addr = reg_num & 0x1f; + break; + case ICP_SR_RD_REL: + case ICP_SR_WR_REL: + case ICP_SR_REL: + reg_addr = 0x180 | (reg_num & 0x1f); + break; + case ICP_SR_ABS: + reg_addr = 0x140 | ((reg_num & 0x3) << 1); + break; + case ICP_DR_RD_REL: + case ICP_DR_WR_REL: + case ICP_DR_REL: + reg_addr = 0x1c0 | (reg_num & 0x1f); + break; + case ICP_DR_ABS: + reg_addr = 0x100 | ((reg_num & 0x3) << 1); + break; + case ICP_NEIGH_REL: + reg_addr = 0x280 | (reg_num & 0x1f); + break; + case ICP_LMEM0: + reg_addr = 0x200; + break; + case ICP_LMEM1: + reg_addr = 0x220; + break; + case ICP_LMEM2: + reg_addr = 0x2c0; + break; + case ICP_LMEM3: + reg_addr = 0x2e0; + break; + case ICP_NO_DEST: + reg_addr = 0x300 | (reg_num & 0xff); + break; + default: + reg_addr = BAD_REGADDR; + break; + } + return reg_addr; +} + +void qat_hal_reset(struct icp_qat_fw_loader_handle *handle) +{ + unsigned int reset_mask = handle->chip_info->icp_rst_mask; + unsigned int reset_csr = handle->chip_info->icp_rst_csr; + unsigned int csr_val; + + csr_val = GET_CAP_CSR(handle, reset_csr); + csr_val |= reset_mask; + SET_CAP_CSR(handle, reset_csr, csr_val); +} + +static void qat_hal_wr_indr_csr(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask, + unsigned int ae_csr, unsigned int csr_val) +{ + unsigned int ctx, cur_ctx; + + cur_ctx = qat_hal_rd_ae_csr(handle, ae, CSR_CTX_POINTER); + + for (ctx = 0; ctx < ICP_QAT_UCLO_MAX_CTX; ctx++) { + if (!(ctx_mask & (1 << ctx))) + continue; + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, ctx); + qat_hal_wr_ae_csr(handle, ae, ae_csr, csr_val); + } + + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, cur_ctx); +} + +static unsigned int qat_hal_rd_indr_csr(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + unsigned int ae_csr) +{ + unsigned int cur_ctx, csr_val; + + cur_ctx = qat_hal_rd_ae_csr(handle, ae, CSR_CTX_POINTER); + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, ctx); + csr_val = qat_hal_rd_ae_csr(handle, ae, ae_csr); + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, cur_ctx); + + return csr_val; +} + +static void qat_hal_put_sig_event(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask, + unsigned int events) +{ + unsigned int ctx, cur_ctx; + + cur_ctx = qat_hal_rd_ae_csr(handle, ae, CSR_CTX_POINTER); + for (ctx = 0; ctx < ICP_QAT_UCLO_MAX_CTX; ctx++) { + if (!(ctx_mask & (1 << ctx))) + continue; + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, ctx); + qat_hal_wr_ae_csr(handle, ae, CTX_SIG_EVENTS_INDIRECT, events); + } + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, cur_ctx); +} + +static void qat_hal_put_wakeup_event(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask, + unsigned int events) +{ + unsigned int ctx, cur_ctx; + + cur_ctx = qat_hal_rd_ae_csr(handle, ae, CSR_CTX_POINTER); + for (ctx = 0; ctx < ICP_QAT_UCLO_MAX_CTX; ctx++) { + if (!(ctx_mask & (1 << ctx))) + continue; + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, ctx); + qat_hal_wr_ae_csr(handle, ae, CTX_WAKEUP_EVENTS_INDIRECT, + events); + } + qat_hal_wr_ae_csr(handle, ae, CSR_CTX_POINTER, cur_ctx); +} + +static int qat_hal_check_ae_alive(struct icp_qat_fw_loader_handle *handle) +{ + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned int base_cnt, cur_cnt; + unsigned char ae; + int times = MAX_RETRY_TIMES; + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + base_cnt = qat_hal_rd_ae_csr(handle, ae, PROFILE_COUNT); + base_cnt &= 0xffff; + + do { + cur_cnt = qat_hal_rd_ae_csr(handle, ae, PROFILE_COUNT); + cur_cnt &= 0xffff; + } while (times-- && (cur_cnt == base_cnt)); + + if (times < 0) { + pr_err("QAT: AE%d is inactive!!\n", ae); + return -EFAULT; + } + } + + return 0; +} + +int qat_hal_check_ae_active(struct icp_qat_fw_loader_handle *handle, + unsigned int ae) +{ + unsigned int enable = 0, active = 0; + + enable = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + active = qat_hal_rd_ae_csr(handle, ae, ACTIVE_CTX_STATUS); + if ((enable & (0xff << CE_ENABLE_BITPOS)) || + (active & (1 << ACS_ABO_BITPOS))) + return 1; + else + return 0; +} + +static void qat_hal_reset_timestamp(struct icp_qat_fw_loader_handle *handle) +{ + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned int misc_ctl_csr, misc_ctl; + unsigned char ae; + + misc_ctl_csr = handle->chip_info->misc_ctl_csr; + /* stop the timestamp timers */ + misc_ctl = GET_CAP_CSR(handle, misc_ctl_csr); + if (misc_ctl & MC_TIMESTAMP_ENABLE) + SET_CAP_CSR(handle, misc_ctl_csr, misc_ctl & + (~MC_TIMESTAMP_ENABLE)); + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + qat_hal_wr_ae_csr(handle, ae, TIMESTAMP_LOW, 0); + qat_hal_wr_ae_csr(handle, ae, TIMESTAMP_HIGH, 0); + } + /* start timestamp timers */ + SET_CAP_CSR(handle, misc_ctl_csr, misc_ctl | MC_TIMESTAMP_ENABLE); +} + +#define ESRAM_AUTO_TINIT BIT(2) +#define ESRAM_AUTO_TINIT_DONE BIT(3) +#define ESRAM_AUTO_INIT_USED_CYCLES (1640) +#define ESRAM_AUTO_INIT_CSR_OFFSET 0xC1C +static int qat_hal_init_esram(struct icp_qat_fw_loader_handle *handle) +{ + void __iomem *csr_addr = + (void __iomem *)((uintptr_t)handle->hal_ep_csr_addr_v + + ESRAM_AUTO_INIT_CSR_OFFSET); + unsigned int csr_val; + int times = 30; + + if (handle->pci_dev->device != PCI_DEVICE_ID_INTEL_QAT_DH895XCC) + return 0; + + csr_val = ADF_CSR_RD(csr_addr, 0); + if ((csr_val & ESRAM_AUTO_TINIT) && (csr_val & ESRAM_AUTO_TINIT_DONE)) + return 0; + + csr_val = ADF_CSR_RD(csr_addr, 0); + csr_val |= ESRAM_AUTO_TINIT; + ADF_CSR_WR(csr_addr, 0, csr_val); + + do { + qat_hal_wait_cycles(handle, 0, ESRAM_AUTO_INIT_USED_CYCLES, 0); + csr_val = ADF_CSR_RD(csr_addr, 0); + } while (!(csr_val & ESRAM_AUTO_TINIT_DONE) && times--); + if (times < 0) { + pr_err("QAT: Fail to init eSram!\n"); + return -EFAULT; + } + return 0; +} + +#define SHRAM_INIT_CYCLES 2060 +int qat_hal_clr_reset(struct icp_qat_fw_loader_handle *handle) +{ + unsigned int clk_csr = handle->chip_info->glb_clk_enable_csr; + unsigned int reset_mask = handle->chip_info->icp_rst_mask; + unsigned int reset_csr = handle->chip_info->icp_rst_csr; + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned char ae = 0; + unsigned int times = 100; + unsigned int csr_val; + + /* write to the reset csr */ + csr_val = GET_CAP_CSR(handle, reset_csr); + csr_val &= ~reset_mask; + do { + SET_CAP_CSR(handle, reset_csr, csr_val); + if (!(times--)) + goto out_err; + csr_val = GET_CAP_CSR(handle, reset_csr); + csr_val &= reset_mask; + } while (csr_val); + /* enable clock */ + csr_val = GET_CAP_CSR(handle, clk_csr); + csr_val |= reset_mask; + SET_CAP_CSR(handle, clk_csr, csr_val); + if (qat_hal_check_ae_alive(handle)) + goto out_err; + + /* Set undefined power-up/reset states to reasonable default values */ + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, + INIT_CTX_ENABLE_VALUE); + qat_hal_wr_indr_csr(handle, ae, ICP_QAT_UCLO_AE_ALL_CTX, + CTX_STS_INDIRECT, + handle->hal_handle->upc_mask & + INIT_PC_VALUE); + qat_hal_wr_ae_csr(handle, ae, CTX_ARB_CNTL, INIT_CTX_ARB_VALUE); + qat_hal_wr_ae_csr(handle, ae, CC_ENABLE, INIT_CCENABLE_VALUE); + qat_hal_put_wakeup_event(handle, ae, + ICP_QAT_UCLO_AE_ALL_CTX, + INIT_WAKEUP_EVENTS_VALUE); + qat_hal_put_sig_event(handle, ae, + ICP_QAT_UCLO_AE_ALL_CTX, + INIT_SIG_EVENTS_VALUE); + } + if (qat_hal_init_esram(handle)) + goto out_err; + if (qat_hal_wait_cycles(handle, 0, SHRAM_INIT_CYCLES, 0)) + goto out_err; + qat_hal_reset_timestamp(handle); + + return 0; +out_err: + pr_err("QAT: failed to get device out of reset\n"); + return -EFAULT; +} + +static void qat_hal_disable_ctx(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask) +{ + unsigned int ctx; + + ctx = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + ctx &= IGNORE_W1C_MASK & + (~((ctx_mask & ICP_QAT_UCLO_AE_ALL_CTX) << CE_ENABLE_BITPOS)); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx); +} + +static u64 qat_hal_parity_64bit(u64 word) +{ + word ^= word >> 1; + word ^= word >> 2; + word ^= word >> 4; + word ^= word >> 8; + word ^= word >> 16; + word ^= word >> 32; + return word & 1; +} + +static u64 qat_hal_set_uword_ecc(u64 uword) +{ + u64 bit0_mask = 0xff800007fffULL, bit1_mask = 0x1f801ff801fULL, + bit2_mask = 0xe387e0781e1ULL, bit3_mask = 0x7cb8e388e22ULL, + bit4_mask = 0xaf5b2c93244ULL, bit5_mask = 0xf56d5525488ULL, + bit6_mask = 0xdaf69a46910ULL; + + /* clear the ecc bits */ + uword &= ~(0x7fULL << 0x2C); + uword |= qat_hal_parity_64bit(bit0_mask & uword) << 0x2C; + uword |= qat_hal_parity_64bit(bit1_mask & uword) << 0x2D; + uword |= qat_hal_parity_64bit(bit2_mask & uword) << 0x2E; + uword |= qat_hal_parity_64bit(bit3_mask & uword) << 0x2F; + uword |= qat_hal_parity_64bit(bit4_mask & uword) << 0x30; + uword |= qat_hal_parity_64bit(bit5_mask & uword) << 0x31; + uword |= qat_hal_parity_64bit(bit6_mask & uword) << 0x32; + return uword; +} + +void qat_hal_wr_uwords(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int uaddr, + unsigned int words_num, u64 *uword) +{ + unsigned int ustore_addr; + unsigned int i; + + ustore_addr = qat_hal_rd_ae_csr(handle, ae, USTORE_ADDRESS); + uaddr |= UA_ECS; + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, uaddr); + for (i = 0; i < words_num; i++) { + unsigned int uwrd_lo, uwrd_hi; + u64 tmp; + + tmp = qat_hal_set_uword_ecc(uword[i]); + uwrd_lo = (unsigned int)(tmp & 0xffffffff); + uwrd_hi = (unsigned int)(tmp >> 0x20); + qat_hal_wr_ae_csr(handle, ae, USTORE_DATA_LOWER, uwrd_lo); + qat_hal_wr_ae_csr(handle, ae, USTORE_DATA_UPPER, uwrd_hi); + } + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, ustore_addr); +} + +static void qat_hal_enable_ctx(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask) +{ + unsigned int ctx; + + ctx = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + ctx &= IGNORE_W1C_MASK; + ctx_mask &= (ctx & CE_INUSE_CONTEXTS) ? 0x55 : 0xFF; + ctx |= (ctx_mask << CE_ENABLE_BITPOS); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx); +} + +static void qat_hal_clear_xfer(struct icp_qat_fw_loader_handle *handle) +{ + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned char ae; + unsigned short reg; + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + for (reg = 0; reg < ICP_QAT_UCLO_MAX_GPR_REG; reg++) { + qat_hal_init_rd_xfer(handle, ae, 0, ICP_SR_RD_ABS, + reg, 0); + qat_hal_init_rd_xfer(handle, ae, 0, ICP_DR_RD_ABS, + reg, 0); + } + } +} + +static int qat_hal_clear_gpr(struct icp_qat_fw_loader_handle *handle) +{ + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned char ae; + unsigned int ctx_mask = ICP_QAT_UCLO_AE_ALL_CTX; + int times = MAX_RETRY_TIMES; + unsigned int csr_val = 0; + unsigned int savctx = 0; + int ret = 0; + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + csr_val = qat_hal_rd_ae_csr(handle, ae, AE_MISC_CONTROL); + csr_val &= ~(1 << MMC_SHARE_CS_BITPOS); + qat_hal_wr_ae_csr(handle, ae, AE_MISC_CONTROL, csr_val); + csr_val = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + csr_val &= IGNORE_W1C_MASK; + if (handle->chip_info->nn) + csr_val |= CE_NN_MODE; + + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, csr_val); + qat_hal_wr_uwords(handle, ae, 0, ARRAY_SIZE(inst), + (u64 *)inst); + qat_hal_wr_indr_csr(handle, ae, ctx_mask, CTX_STS_INDIRECT, + handle->hal_handle->upc_mask & + INIT_PC_VALUE); + savctx = qat_hal_rd_ae_csr(handle, ae, ACTIVE_CTX_STATUS); + qat_hal_wr_ae_csr(handle, ae, ACTIVE_CTX_STATUS, 0); + qat_hal_put_wakeup_event(handle, ae, ctx_mask, XCWE_VOLUNTARY); + qat_hal_wr_indr_csr(handle, ae, ctx_mask, + CTX_SIG_EVENTS_INDIRECT, 0); + qat_hal_wr_ae_csr(handle, ae, CTX_SIG_EVENTS_ACTIVE, 0); + qat_hal_enable_ctx(handle, ae, ctx_mask); + } + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + /* wait for AE to finish */ + do { + ret = qat_hal_wait_cycles(handle, ae, 20, 1); + } while (ret && times--); + + if (times < 0) { + pr_err("QAT: clear GPR of AE %d failed", ae); + return -EINVAL; + } + qat_hal_disable_ctx(handle, ae, ctx_mask); + qat_hal_wr_ae_csr(handle, ae, ACTIVE_CTX_STATUS, + savctx & ACS_ACNO); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, + INIT_CTX_ENABLE_VALUE); + qat_hal_wr_indr_csr(handle, ae, ctx_mask, CTX_STS_INDIRECT, + handle->hal_handle->upc_mask & + INIT_PC_VALUE); + qat_hal_wr_ae_csr(handle, ae, CTX_ARB_CNTL, INIT_CTX_ARB_VALUE); + qat_hal_wr_ae_csr(handle, ae, CC_ENABLE, INIT_CCENABLE_VALUE); + qat_hal_put_wakeup_event(handle, ae, ctx_mask, + INIT_WAKEUP_EVENTS_VALUE); + qat_hal_put_sig_event(handle, ae, ctx_mask, + INIT_SIG_EVENTS_VALUE); + } + return 0; +} + +static int qat_hal_chip_init(struct icp_qat_fw_loader_handle *handle, + struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *pci_info = &accel_dev->accel_pci_dev; + struct adf_hw_device_data *hw_data = accel_dev->hw_device; + void __iomem *pmisc_addr = adf_get_pmisc_base(accel_dev); + unsigned int max_en_ae_id = 0; + struct adf_bar *sram_bar; + unsigned int csr_val = 0; + unsigned long ae_mask; + unsigned char ae = 0; + int ret = 0; + + handle->pci_dev = pci_info->pci_dev; + switch (handle->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_4XXX: + case PCI_DEVICE_ID_INTEL_QAT_401XX: + case PCI_DEVICE_ID_INTEL_QAT_402XX: + case PCI_DEVICE_ID_INTEL_QAT_420XX: + case PCI_DEVICE_ID_INTEL_QAT_6XXX: + handle->chip_info->mmp_sram_size = 0; + handle->chip_info->nn = false; + handle->chip_info->lm2lm3 = true; + handle->chip_info->lm_size = ICP_QAT_UCLO_MAX_LMEM_REG_2X; + handle->chip_info->icp_rst_csr = ICP_RESET_CPP0; + if (handle->pci_dev->device == PCI_DEVICE_ID_INTEL_QAT_420XX) + handle->chip_info->icp_rst_mask = 0x100155; + else + handle->chip_info->icp_rst_mask = 0x100015; + handle->chip_info->glb_clk_enable_csr = ICP_GLOBAL_CLK_ENABLE_CPP0; + handle->chip_info->misc_ctl_csr = MISC_CONTROL_C4XXX; + handle->chip_info->wakeup_event_val = 0x80000000; + handle->chip_info->fw_auth = true; + handle->chip_info->css_3k = true; + if (handle->pci_dev->device == PCI_DEVICE_ID_INTEL_QAT_6XXX) + handle->chip_info->dual_sign = true; + handle->chip_info->tgroup_share_ustore = true; + handle->chip_info->fcu_ctl_csr = FCU_CONTROL_4XXX; + handle->chip_info->fcu_sts_csr = FCU_STATUS_4XXX; + handle->chip_info->fcu_dram_addr_hi = FCU_DRAM_ADDR_HI_4XXX; + handle->chip_info->fcu_dram_addr_lo = FCU_DRAM_ADDR_LO_4XXX; + handle->chip_info->fcu_loaded_ae_csr = FCU_AE_LOADED_4XXX; + handle->chip_info->fcu_loaded_ae_pos = 0; + + handle->hal_cap_g_ctl_csr_addr_v = pmisc_addr + ICP_QAT_CAP_OFFSET_4XXX; + handle->hal_cap_ae_xfer_csr_addr_v = pmisc_addr + ICP_QAT_AE_OFFSET_4XXX; + handle->hal_ep_csr_addr_v = pmisc_addr + ICP_QAT_EP_OFFSET_4XXX; + handle->hal_cap_ae_local_csr_addr_v = + (void __iomem *)((uintptr_t)handle->hal_cap_ae_xfer_csr_addr_v + + LOCAL_TO_XFER_REG_OFFSET); + break; + case PCI_DEVICE_ID_INTEL_QAT_C62X: + case PCI_DEVICE_ID_INTEL_QAT_C3XXX: + handle->chip_info->mmp_sram_size = 0; + handle->chip_info->nn = true; + handle->chip_info->lm2lm3 = false; + handle->chip_info->lm_size = ICP_QAT_UCLO_MAX_LMEM_REG; + handle->chip_info->icp_rst_csr = ICP_RESET; + handle->chip_info->icp_rst_mask = (hw_data->ae_mask << RST_CSR_AE_LSB) | + (hw_data->accel_mask << RST_CSR_QAT_LSB); + handle->chip_info->glb_clk_enable_csr = ICP_GLOBAL_CLK_ENABLE; + handle->chip_info->misc_ctl_csr = MISC_CONTROL; + handle->chip_info->wakeup_event_val = WAKEUP_EVENT; + handle->chip_info->fw_auth = true; + handle->chip_info->css_3k = false; + handle->chip_info->tgroup_share_ustore = false; + handle->chip_info->fcu_ctl_csr = FCU_CONTROL; + handle->chip_info->fcu_sts_csr = FCU_STATUS; + handle->chip_info->fcu_dram_addr_hi = FCU_DRAM_ADDR_HI; + handle->chip_info->fcu_dram_addr_lo = FCU_DRAM_ADDR_LO; + handle->chip_info->fcu_loaded_ae_csr = FCU_STATUS; + handle->chip_info->fcu_loaded_ae_pos = FCU_LOADED_AE_POS; + handle->hal_cap_g_ctl_csr_addr_v = pmisc_addr + ICP_QAT_CAP_OFFSET; + handle->hal_cap_ae_xfer_csr_addr_v = pmisc_addr + ICP_QAT_AE_OFFSET; + handle->hal_ep_csr_addr_v = pmisc_addr + ICP_QAT_EP_OFFSET; + handle->hal_cap_ae_local_csr_addr_v = + (void __iomem *)((uintptr_t)handle->hal_cap_ae_xfer_csr_addr_v + + LOCAL_TO_XFER_REG_OFFSET); + break; + case PCI_DEVICE_ID_INTEL_QAT_DH895XCC: + handle->chip_info->mmp_sram_size = 0x40000; + handle->chip_info->nn = true; + handle->chip_info->lm2lm3 = false; + handle->chip_info->lm_size = ICP_QAT_UCLO_MAX_LMEM_REG; + handle->chip_info->icp_rst_csr = ICP_RESET; + handle->chip_info->icp_rst_mask = (hw_data->ae_mask << RST_CSR_AE_LSB) | + (hw_data->accel_mask << RST_CSR_QAT_LSB); + handle->chip_info->glb_clk_enable_csr = ICP_GLOBAL_CLK_ENABLE; + handle->chip_info->misc_ctl_csr = MISC_CONTROL; + handle->chip_info->wakeup_event_val = WAKEUP_EVENT; + handle->chip_info->fw_auth = false; + handle->chip_info->css_3k = false; + handle->chip_info->tgroup_share_ustore = false; + handle->chip_info->fcu_ctl_csr = 0; + handle->chip_info->fcu_sts_csr = 0; + handle->chip_info->fcu_dram_addr_hi = 0; + handle->chip_info->fcu_dram_addr_lo = 0; + handle->chip_info->fcu_loaded_ae_csr = 0; + handle->chip_info->fcu_loaded_ae_pos = 0; + handle->hal_cap_g_ctl_csr_addr_v = pmisc_addr + ICP_QAT_CAP_OFFSET; + handle->hal_cap_ae_xfer_csr_addr_v = pmisc_addr + ICP_QAT_AE_OFFSET; + handle->hal_ep_csr_addr_v = pmisc_addr + ICP_QAT_EP_OFFSET; + handle->hal_cap_ae_local_csr_addr_v = + (void __iomem *)((uintptr_t)handle->hal_cap_ae_xfer_csr_addr_v + + LOCAL_TO_XFER_REG_OFFSET); + break; + default: + ret = -EINVAL; + goto out_err; + } + + if (handle->chip_info->mmp_sram_size > 0) { + sram_bar = + &pci_info->pci_bars[hw_data->get_sram_bar_id(hw_data)]; + handle->hal_sram_addr_v = sram_bar->virt_addr; + } + handle->hal_handle->revision_id = accel_dev->accel_pci_dev.revid; + handle->hal_handle->ae_mask = hw_data->ae_mask; + handle->hal_handle->admin_ae_mask = hw_data->admin_ae_mask; + handle->hal_handle->slice_mask = hw_data->accel_mask; + handle->cfg_ae_mask = ALL_AE_MASK; + /* create AE objects */ + handle->hal_handle->upc_mask = 0x1ffff; + handle->hal_handle->max_ustore = 0x4000; + + ae_mask = handle->hal_handle->ae_mask; + for_each_set_bit(ae, &ae_mask, ICP_QAT_UCLO_MAX_AE) { + handle->hal_handle->aes[ae].free_addr = 0; + handle->hal_handle->aes[ae].free_size = + handle->hal_handle->max_ustore; + handle->hal_handle->aes[ae].ustore_size = + handle->hal_handle->max_ustore; + handle->hal_handle->aes[ae].live_ctx_mask = + ICP_QAT_UCLO_AE_ALL_CTX; + max_en_ae_id = ae; + } + handle->hal_handle->ae_max_num = max_en_ae_id + 1; + + /* Set SIGNATURE_ENABLE[0] to 0x1 in order to enable ALU_OUT csr */ + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + csr_val = qat_hal_rd_ae_csr(handle, ae, SIGNATURE_ENABLE); + csr_val |= 0x1; + qat_hal_wr_ae_csr(handle, ae, SIGNATURE_ENABLE, csr_val); + } +out_err: + return ret; +} + +int qat_hal_init(struct adf_accel_dev *accel_dev) +{ + struct icp_qat_fw_loader_handle *handle; + int ret = 0; + + handle = kzalloc(sizeof(*handle), GFP_KERNEL); + if (!handle) + return -ENOMEM; + + handle->hal_handle = kzalloc(sizeof(*handle->hal_handle), GFP_KERNEL); + if (!handle->hal_handle) { + ret = -ENOMEM; + goto out_hal_handle; + } + + handle->chip_info = kzalloc(sizeof(*handle->chip_info), GFP_KERNEL); + if (!handle->chip_info) { + ret = -ENOMEM; + goto out_chip_info; + } + + ret = qat_hal_chip_init(handle, accel_dev); + if (ret) { + dev_err(&GET_DEV(accel_dev), "qat_hal_chip_init error\n"); + goto out_err; + } + + /* take all AEs out of reset */ + ret = qat_hal_clr_reset(handle); + if (ret) { + dev_err(&GET_DEV(accel_dev), "qat_hal_clr_reset error\n"); + goto out_err; + } + + qat_hal_clear_xfer(handle); + if (!handle->chip_info->fw_auth) { + ret = qat_hal_clear_gpr(handle); + if (ret) + goto out_err; + } + + accel_dev->fw_loader->fw_loader = handle; + return 0; + +out_err: + kfree(handle->chip_info); +out_chip_info: + kfree(handle->hal_handle); +out_hal_handle: + kfree(handle); + return ret; +} + +void qat_hal_deinit(struct icp_qat_fw_loader_handle *handle) +{ + if (!handle) + return; + kfree(handle->chip_info); + kfree(handle->hal_handle); + kfree(handle); +} + +int qat_hal_start(struct icp_qat_fw_loader_handle *handle) +{ + unsigned long ae_mask = handle->hal_handle->ae_mask; + u32 wakeup_val = handle->chip_info->wakeup_event_val; + u32 fcu_ctl_csr, fcu_sts_csr; + unsigned int fcu_sts; + unsigned char ae; + u32 ae_ctr = 0; + int retry = 0; + + if (handle->chip_info->fw_auth) { + fcu_ctl_csr = handle->chip_info->fcu_ctl_csr; + fcu_sts_csr = handle->chip_info->fcu_sts_csr; + ae_ctr = hweight32(ae_mask); + SET_CAP_CSR(handle, fcu_ctl_csr, FCU_CTRL_CMD_START); + do { + msleep(FW_AUTH_WAIT_PERIOD); + fcu_sts = GET_CAP_CSR(handle, fcu_sts_csr); + if (((fcu_sts >> FCU_STS_DONE_POS) & 0x1)) + return ae_ctr; + } while (retry++ < FW_AUTH_MAX_RETRY); + pr_err("QAT: start error (FCU_STS = 0x%x)\n", fcu_sts); + return 0; + } else { + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + qat_hal_put_wakeup_event(handle, ae, 0, wakeup_val); + qat_hal_enable_ctx(handle, ae, ICP_QAT_UCLO_AE_ALL_CTX); + ae_ctr++; + } + return ae_ctr; + } +} + +void qat_hal_stop(struct icp_qat_fw_loader_handle *handle, unsigned char ae, + unsigned int ctx_mask) +{ + if (!handle->chip_info->fw_auth) + qat_hal_disable_ctx(handle, ae, ctx_mask); +} + +void qat_hal_set_pc(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int ctx_mask, unsigned int upc) +{ + qat_hal_wr_indr_csr(handle, ae, ctx_mask, CTX_STS_INDIRECT, + handle->hal_handle->upc_mask & upc); +} + +static void qat_hal_get_uwords(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int uaddr, + unsigned int words_num, u64 *uword) +{ + unsigned int i, uwrd_lo, uwrd_hi; + unsigned int ustore_addr, misc_control; + + misc_control = qat_hal_rd_ae_csr(handle, ae, AE_MISC_CONTROL); + qat_hal_wr_ae_csr(handle, ae, AE_MISC_CONTROL, + misc_control & 0xfffffffb); + ustore_addr = qat_hal_rd_ae_csr(handle, ae, USTORE_ADDRESS); + uaddr |= UA_ECS; + for (i = 0; i < words_num; i++) { + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, uaddr); + uaddr++; + uwrd_lo = qat_hal_rd_ae_csr(handle, ae, USTORE_DATA_LOWER); + uwrd_hi = qat_hal_rd_ae_csr(handle, ae, USTORE_DATA_UPPER); + uword[i] = uwrd_hi; + uword[i] = (uword[i] << 0x20) | uwrd_lo; + } + qat_hal_wr_ae_csr(handle, ae, AE_MISC_CONTROL, misc_control); + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, ustore_addr); +} + +void qat_hal_wr_umem(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int uaddr, + unsigned int words_num, unsigned int *data) +{ + unsigned int i, ustore_addr; + + ustore_addr = qat_hal_rd_ae_csr(handle, ae, USTORE_ADDRESS); + uaddr |= UA_ECS; + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, uaddr); + for (i = 0; i < words_num; i++) { + unsigned int uwrd_lo, uwrd_hi, tmp; + + uwrd_lo = ((data[i] & 0xfff0000) << 4) | (0x3 << 18) | + ((data[i] & 0xff00) << 2) | + (0x3 << 8) | (data[i] & 0xff); + uwrd_hi = (0xf << 4) | ((data[i] & 0xf0000000) >> 28); + uwrd_hi |= (hweight32(data[i] & 0xffff) & 0x1) << 8; + tmp = ((data[i] >> 0x10) & 0xffff); + uwrd_hi |= (hweight32(tmp) & 0x1) << 9; + qat_hal_wr_ae_csr(handle, ae, USTORE_DATA_LOWER, uwrd_lo); + qat_hal_wr_ae_csr(handle, ae, USTORE_DATA_UPPER, uwrd_hi); + } + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, ustore_addr); +} + +#define MAX_EXEC_INST 100 +static int qat_hal_exec_micro_inst(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + u64 *micro_inst, unsigned int inst_num, + int code_off, unsigned int max_cycle, + unsigned int *endpc) +{ + unsigned int ind_lm_addr_byte0 = 0, ind_lm_addr_byte1 = 0; + unsigned int ind_lm_addr_byte2 = 0, ind_lm_addr_byte3 = 0; + unsigned int ind_t_index = 0, ind_t_index_byte = 0; + unsigned int ind_lm_addr0 = 0, ind_lm_addr1 = 0; + unsigned int ind_lm_addr2 = 0, ind_lm_addr3 = 0; + u64 savuwords[MAX_EXEC_INST]; + unsigned int ind_cnt_sig; + unsigned int ind_sig, act_sig; + unsigned int csr_val = 0, newcsr_val; + unsigned int savctx; + unsigned int savcc, wakeup_events, savpc; + unsigned int ctxarb_ctl, ctx_enables; + + if ((inst_num > handle->hal_handle->max_ustore) || !micro_inst) { + pr_err("QAT: invalid instruction num %d\n", inst_num); + return -EINVAL; + } + /* save current context */ + ind_lm_addr0 = qat_hal_rd_indr_csr(handle, ae, ctx, LM_ADDR_0_INDIRECT); + ind_lm_addr1 = qat_hal_rd_indr_csr(handle, ae, ctx, LM_ADDR_1_INDIRECT); + ind_lm_addr_byte0 = qat_hal_rd_indr_csr(handle, ae, ctx, + INDIRECT_LM_ADDR_0_BYTE_INDEX); + ind_lm_addr_byte1 = qat_hal_rd_indr_csr(handle, ae, ctx, + INDIRECT_LM_ADDR_1_BYTE_INDEX); + if (handle->chip_info->lm2lm3) { + ind_lm_addr2 = qat_hal_rd_indr_csr(handle, ae, ctx, + LM_ADDR_2_INDIRECT); + ind_lm_addr3 = qat_hal_rd_indr_csr(handle, ae, ctx, + LM_ADDR_3_INDIRECT); + ind_lm_addr_byte2 = qat_hal_rd_indr_csr(handle, ae, ctx, + INDIRECT_LM_ADDR_2_BYTE_INDEX); + ind_lm_addr_byte3 = qat_hal_rd_indr_csr(handle, ae, ctx, + INDIRECT_LM_ADDR_3_BYTE_INDEX); + ind_t_index = qat_hal_rd_indr_csr(handle, ae, ctx, + INDIRECT_T_INDEX); + ind_t_index_byte = qat_hal_rd_indr_csr(handle, ae, ctx, + INDIRECT_T_INDEX_BYTE_INDEX); + } + if (inst_num <= MAX_EXEC_INST) + qat_hal_get_uwords(handle, ae, 0, inst_num, savuwords); + qat_hal_get_wakeup_event(handle, ae, ctx, &wakeup_events); + savpc = qat_hal_rd_indr_csr(handle, ae, ctx, CTX_STS_INDIRECT); + savpc = (savpc & handle->hal_handle->upc_mask) >> 0; + ctx_enables = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + ctx_enables &= IGNORE_W1C_MASK; + savcc = qat_hal_rd_ae_csr(handle, ae, CC_ENABLE); + savctx = qat_hal_rd_ae_csr(handle, ae, ACTIVE_CTX_STATUS); + ctxarb_ctl = qat_hal_rd_ae_csr(handle, ae, CTX_ARB_CNTL); + ind_cnt_sig = qat_hal_rd_indr_csr(handle, ae, ctx, + FUTURE_COUNT_SIGNAL_INDIRECT); + ind_sig = qat_hal_rd_indr_csr(handle, ae, ctx, + CTX_SIG_EVENTS_INDIRECT); + act_sig = qat_hal_rd_ae_csr(handle, ae, CTX_SIG_EVENTS_ACTIVE); + /* execute micro codes */ + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx_enables); + qat_hal_wr_uwords(handle, ae, 0, inst_num, micro_inst); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), CTX_STS_INDIRECT, 0); + qat_hal_wr_ae_csr(handle, ae, ACTIVE_CTX_STATUS, ctx & ACS_ACNO); + if (code_off) + qat_hal_wr_ae_csr(handle, ae, CC_ENABLE, savcc & 0xffffdfff); + qat_hal_put_wakeup_event(handle, ae, (1 << ctx), XCWE_VOLUNTARY); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), CTX_SIG_EVENTS_INDIRECT, 0); + qat_hal_wr_ae_csr(handle, ae, CTX_SIG_EVENTS_ACTIVE, 0); + qat_hal_enable_ctx(handle, ae, (1 << ctx)); + /* wait for micro codes to finish */ + if (qat_hal_wait_cycles(handle, ae, max_cycle, 1) != 0) + return -EFAULT; + if (endpc) { + unsigned int ctx_status; + + ctx_status = qat_hal_rd_indr_csr(handle, ae, ctx, + CTX_STS_INDIRECT); + *endpc = ctx_status & handle->hal_handle->upc_mask; + } + /* retore to saved context */ + qat_hal_disable_ctx(handle, ae, (1 << ctx)); + if (inst_num <= MAX_EXEC_INST) + qat_hal_wr_uwords(handle, ae, 0, inst_num, savuwords); + qat_hal_put_wakeup_event(handle, ae, (1 << ctx), wakeup_events); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), CTX_STS_INDIRECT, + handle->hal_handle->upc_mask & savpc); + csr_val = qat_hal_rd_ae_csr(handle, ae, AE_MISC_CONTROL); + newcsr_val = CLR_BIT(csr_val, MMC_SHARE_CS_BITPOS); + qat_hal_wr_ae_csr(handle, ae, AE_MISC_CONTROL, newcsr_val); + qat_hal_wr_ae_csr(handle, ae, CC_ENABLE, savcc); + qat_hal_wr_ae_csr(handle, ae, ACTIVE_CTX_STATUS, savctx & ACS_ACNO); + qat_hal_wr_ae_csr(handle, ae, CTX_ARB_CNTL, ctxarb_ctl); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), + LM_ADDR_0_INDIRECT, ind_lm_addr0); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), + LM_ADDR_1_INDIRECT, ind_lm_addr1); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), + INDIRECT_LM_ADDR_0_BYTE_INDEX, ind_lm_addr_byte0); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), + INDIRECT_LM_ADDR_1_BYTE_INDEX, ind_lm_addr_byte1); + if (handle->chip_info->lm2lm3) { + qat_hal_wr_indr_csr(handle, ae, BIT(ctx), LM_ADDR_2_INDIRECT, + ind_lm_addr2); + qat_hal_wr_indr_csr(handle, ae, BIT(ctx), LM_ADDR_3_INDIRECT, + ind_lm_addr3); + qat_hal_wr_indr_csr(handle, ae, BIT(ctx), + INDIRECT_LM_ADDR_2_BYTE_INDEX, + ind_lm_addr_byte2); + qat_hal_wr_indr_csr(handle, ae, BIT(ctx), + INDIRECT_LM_ADDR_3_BYTE_INDEX, + ind_lm_addr_byte3); + qat_hal_wr_indr_csr(handle, ae, BIT(ctx), + INDIRECT_T_INDEX, ind_t_index); + qat_hal_wr_indr_csr(handle, ae, BIT(ctx), + INDIRECT_T_INDEX_BYTE_INDEX, + ind_t_index_byte); + } + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), + FUTURE_COUNT_SIGNAL_INDIRECT, ind_cnt_sig); + qat_hal_wr_indr_csr(handle, ae, (1 << ctx), + CTX_SIG_EVENTS_INDIRECT, ind_sig); + qat_hal_wr_ae_csr(handle, ae, CTX_SIG_EVENTS_ACTIVE, act_sig); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx_enables); + + return 0; +} + +static int qat_hal_rd_rel_reg(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int *data) +{ + unsigned int savctx, uaddr, uwrd_lo, uwrd_hi; + unsigned int ctxarb_cntl, ustore_addr, ctx_enables; + unsigned short reg_addr; + int status = 0; + u64 insts, savuword; + + reg_addr = qat_hal_get_reg_addr(reg_type, reg_num); + if (reg_addr == BAD_REGADDR) { + pr_err("QAT: bad regaddr=0x%x\n", reg_addr); + return -EINVAL; + } + switch (reg_type) { + case ICP_GPA_REL: + insts = 0xA070000000ull | (reg_addr & 0x3ff); + break; + default: + insts = (u64)0xA030000000ull | ((reg_addr & 0x3ff) << 10); + break; + } + savctx = qat_hal_rd_ae_csr(handle, ae, ACTIVE_CTX_STATUS); + ctxarb_cntl = qat_hal_rd_ae_csr(handle, ae, CTX_ARB_CNTL); + ctx_enables = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + ctx_enables &= IGNORE_W1C_MASK; + if (ctx != (savctx & ACS_ACNO)) + qat_hal_wr_ae_csr(handle, ae, ACTIVE_CTX_STATUS, + ctx & ACS_ACNO); + qat_hal_get_uwords(handle, ae, 0, 1, &savuword); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx_enables); + ustore_addr = qat_hal_rd_ae_csr(handle, ae, USTORE_ADDRESS); + uaddr = UA_ECS; + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, uaddr); + insts = qat_hal_set_uword_ecc(insts); + uwrd_lo = (unsigned int)(insts & 0xffffffff); + uwrd_hi = (unsigned int)(insts >> 0x20); + qat_hal_wr_ae_csr(handle, ae, USTORE_DATA_LOWER, uwrd_lo); + qat_hal_wr_ae_csr(handle, ae, USTORE_DATA_UPPER, uwrd_hi); + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, uaddr); + /* delay for at least 8 cycles */ + qat_hal_wait_cycles(handle, ae, 0x8, 0); + /* + * read ALU output + * the instruction should have been executed + * prior to clearing the ECS in putUwords + */ + *data = qat_hal_rd_ae_csr(handle, ae, ALU_OUT); + qat_hal_wr_ae_csr(handle, ae, USTORE_ADDRESS, ustore_addr); + qat_hal_wr_uwords(handle, ae, 0, 1, &savuword); + if (ctx != (savctx & ACS_ACNO)) + qat_hal_wr_ae_csr(handle, ae, ACTIVE_CTX_STATUS, + savctx & ACS_ACNO); + qat_hal_wr_ae_csr(handle, ae, CTX_ARB_CNTL, ctxarb_cntl); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx_enables); + + return status; +} + +static int qat_hal_wr_rel_reg(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int data) +{ + unsigned short src_hiaddr, src_lowaddr, dest_addr, data16hi, data16lo; + u64 insts[] = { + 0x0F440000000ull, + 0x0F040000000ull, + 0x0F0000C0300ull, + 0x0E000010000ull + }; + const int num_inst = ARRAY_SIZE(insts), code_off = 1; + const int imm_w1 = 0, imm_w0 = 1; + + dest_addr = qat_hal_get_reg_addr(reg_type, reg_num); + if (dest_addr == BAD_REGADDR) { + pr_err("QAT: bad destAddr=0x%x\n", dest_addr); + return -EINVAL; + } + + data16lo = 0xffff & data; + data16hi = 0xffff & (data >> 0x10); + src_hiaddr = qat_hal_get_reg_addr(ICP_NO_DEST, (unsigned short) + (0xff & data16hi)); + src_lowaddr = qat_hal_get_reg_addr(ICP_NO_DEST, (unsigned short) + (0xff & data16lo)); + switch (reg_type) { + case ICP_GPA_REL: + insts[imm_w1] = insts[imm_w1] | ((data16hi >> 8) << 20) | + ((src_hiaddr & 0x3ff) << 10) | (dest_addr & 0x3ff); + insts[imm_w0] = insts[imm_w0] | ((data16lo >> 8) << 20) | + ((src_lowaddr & 0x3ff) << 10) | (dest_addr & 0x3ff); + break; + default: + insts[imm_w1] = insts[imm_w1] | ((data16hi >> 8) << 20) | + ((dest_addr & 0x3ff) << 10) | (src_hiaddr & 0x3ff); + + insts[imm_w0] = insts[imm_w0] | ((data16lo >> 8) << 20) | + ((dest_addr & 0x3ff) << 10) | (src_lowaddr & 0x3ff); + break; + } + + return qat_hal_exec_micro_inst(handle, ae, ctx, insts, num_inst, + code_off, num_inst * 0x5, NULL); +} + +int qat_hal_get_ins_num(void) +{ + return ARRAY_SIZE(inst_4b); +} + +static int qat_hal_concat_micro_code(u64 *micro_inst, + unsigned int inst_num, unsigned int size, + unsigned int addr, unsigned int *value) +{ + int i; + unsigned int cur_value; + const u64 *inst_arr; + int fixup_offset; + int usize = 0; + int orig_num; + + orig_num = inst_num; + cur_value = value[0]; + inst_arr = inst_4b; + usize = ARRAY_SIZE(inst_4b); + fixup_offset = inst_num; + for (i = 0; i < usize; i++) + micro_inst[inst_num++] = inst_arr[i]; + INSERT_IMMED_GPRA_CONST(micro_inst[fixup_offset], (addr)); + fixup_offset++; + INSERT_IMMED_GPRA_CONST(micro_inst[fixup_offset], 0); + fixup_offset++; + INSERT_IMMED_GPRB_CONST(micro_inst[fixup_offset], (cur_value >> 0)); + fixup_offset++; + INSERT_IMMED_GPRB_CONST(micro_inst[fixup_offset], (cur_value >> 0x10)); + + return inst_num - orig_num; +} + +static int qat_hal_exec_micro_init_lm(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + int *pfirst_exec, u64 *micro_inst, + unsigned int inst_num) +{ + int stat = 0; + unsigned int gpra0 = 0, gpra1 = 0, gpra2 = 0; + unsigned int gprb0 = 0, gprb1 = 0; + + if (*pfirst_exec) { + qat_hal_rd_rel_reg(handle, ae, ctx, ICP_GPA_REL, 0, &gpra0); + qat_hal_rd_rel_reg(handle, ae, ctx, ICP_GPA_REL, 0x1, &gpra1); + qat_hal_rd_rel_reg(handle, ae, ctx, ICP_GPA_REL, 0x2, &gpra2); + qat_hal_rd_rel_reg(handle, ae, ctx, ICP_GPB_REL, 0, &gprb0); + qat_hal_rd_rel_reg(handle, ae, ctx, ICP_GPB_REL, 0x1, &gprb1); + *pfirst_exec = 0; + } + stat = qat_hal_exec_micro_inst(handle, ae, ctx, micro_inst, inst_num, 1, + inst_num * 0x5, NULL); + if (stat != 0) + return -EFAULT; + qat_hal_wr_rel_reg(handle, ae, ctx, ICP_GPA_REL, 0, gpra0); + qat_hal_wr_rel_reg(handle, ae, ctx, ICP_GPA_REL, 0x1, gpra1); + qat_hal_wr_rel_reg(handle, ae, ctx, ICP_GPA_REL, 0x2, gpra2); + qat_hal_wr_rel_reg(handle, ae, ctx, ICP_GPB_REL, 0, gprb0); + qat_hal_wr_rel_reg(handle, ae, ctx, ICP_GPB_REL, 0x1, gprb1); + + return 0; +} + +int qat_hal_batch_wr_lm(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, + struct icp_qat_uof_batch_init *lm_init_header) +{ + struct icp_qat_uof_batch_init *plm_init; + u64 *micro_inst_arry; + int micro_inst_num; + int alloc_inst_size; + int first_exec = 1; + int stat = 0; + + plm_init = lm_init_header->next; + alloc_inst_size = lm_init_header->size; + if ((unsigned int)alloc_inst_size > handle->hal_handle->max_ustore) + alloc_inst_size = handle->hal_handle->max_ustore; + micro_inst_arry = kmalloc_array(alloc_inst_size, sizeof(u64), + GFP_KERNEL); + if (!micro_inst_arry) + return -ENOMEM; + micro_inst_num = 0; + while (plm_init) { + unsigned int addr, *value, size; + + ae = plm_init->ae; + addr = plm_init->addr; + value = plm_init->value; + size = plm_init->size; + micro_inst_num += qat_hal_concat_micro_code(micro_inst_arry, + micro_inst_num, + size, addr, value); + plm_init = plm_init->next; + } + /* exec micro codes */ + if (micro_inst_arry && micro_inst_num > 0) { + micro_inst_arry[micro_inst_num++] = 0x0E000010000ull; + stat = qat_hal_exec_micro_init_lm(handle, ae, 0, &first_exec, + micro_inst_arry, + micro_inst_num); + } + kfree(micro_inst_arry); + return stat; +} + +static int qat_hal_put_rel_rd_xfer(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int val) +{ + int status = 0; + unsigned int reg_addr; + unsigned int ctx_enables; + unsigned short mask; + unsigned short dr_offset = 0x10; + + ctx_enables = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + if (CE_INUSE_CONTEXTS & ctx_enables) { + if (ctx & 0x1) { + pr_err("QAT: bad 4-ctx mode,ctx=0x%x\n", ctx); + return -EINVAL; + } + mask = 0x1f; + dr_offset = 0x20; + } else { + mask = 0x0f; + } + if (reg_num & ~mask) + return -EINVAL; + reg_addr = reg_num + (ctx << 0x5); + switch (reg_type) { + case ICP_SR_RD_REL: + case ICP_SR_REL: + SET_AE_XFER(handle, ae, reg_addr, val); + break; + case ICP_DR_RD_REL: + case ICP_DR_REL: + SET_AE_XFER(handle, ae, (reg_addr + dr_offset), val); + break; + default: + status = -EINVAL; + break; + } + return status; +} + +static int qat_hal_put_rel_wr_xfer(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int data) +{ + unsigned int gprval, ctx_enables; + unsigned short src_hiaddr, src_lowaddr, gpr_addr, xfr_addr, data16hi, + data16low; + unsigned short reg_mask; + int status = 0; + u64 micro_inst[] = { + 0x0F440000000ull, + 0x0F040000000ull, + 0x0A000000000ull, + 0x0F0000C0300ull, + 0x0E000010000ull + }; + const int num_inst = ARRAY_SIZE(micro_inst), code_off = 1; + const unsigned short gprnum = 0, dly = num_inst * 0x5; + + ctx_enables = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + if (CE_INUSE_CONTEXTS & ctx_enables) { + if (ctx & 0x1) { + pr_err("QAT: 4-ctx mode,ctx=0x%x\n", ctx); + return -EINVAL; + } + reg_mask = (unsigned short)~0x1f; + } else { + reg_mask = (unsigned short)~0xf; + } + if (reg_num & reg_mask) + return -EINVAL; + xfr_addr = qat_hal_get_reg_addr(reg_type, reg_num); + if (xfr_addr == BAD_REGADDR) { + pr_err("QAT: bad xfrAddr=0x%x\n", xfr_addr); + return -EINVAL; + } + status = qat_hal_rd_rel_reg(handle, ae, ctx, ICP_GPB_REL, gprnum, &gprval); + if (status) { + pr_err("QAT: failed to read register"); + return status; + } + gpr_addr = qat_hal_get_reg_addr(ICP_GPB_REL, gprnum); + data16low = 0xffff & data; + data16hi = 0xffff & (data >> 0x10); + src_hiaddr = qat_hal_get_reg_addr(ICP_NO_DEST, + (unsigned short)(0xff & data16hi)); + src_lowaddr = qat_hal_get_reg_addr(ICP_NO_DEST, + (unsigned short)(0xff & data16low)); + micro_inst[0] = micro_inst[0x0] | ((data16hi >> 8) << 20) | + ((gpr_addr & 0x3ff) << 10) | (src_hiaddr & 0x3ff); + micro_inst[1] = micro_inst[0x1] | ((data16low >> 8) << 20) | + ((gpr_addr & 0x3ff) << 10) | (src_lowaddr & 0x3ff); + micro_inst[0x2] = micro_inst[0x2] | + ((xfr_addr & 0x3ff) << 20) | ((gpr_addr & 0x3ff) << 10); + status = qat_hal_exec_micro_inst(handle, ae, ctx, micro_inst, num_inst, + code_off, dly, NULL); + qat_hal_wr_rel_reg(handle, ae, ctx, ICP_GPB_REL, gprnum, gprval); + return status; +} + +static int qat_hal_put_rel_nn(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx, + unsigned short nn, unsigned int val) +{ + unsigned int ctx_enables; + int stat = 0; + + ctx_enables = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + ctx_enables &= IGNORE_W1C_MASK; + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx_enables | CE_NN_MODE); + + stat = qat_hal_put_rel_wr_xfer(handle, ae, ctx, ICP_NEIGH_REL, nn, val); + qat_hal_wr_ae_csr(handle, ae, CTX_ENABLES, ctx_enables); + return stat; +} + +static int qat_hal_convert_abs_to_rel(struct icp_qat_fw_loader_handle + *handle, unsigned char ae, + unsigned short absreg_num, + unsigned short *relreg, + unsigned char *ctx) +{ + unsigned int ctx_enables; + + ctx_enables = qat_hal_rd_ae_csr(handle, ae, CTX_ENABLES); + if (ctx_enables & CE_INUSE_CONTEXTS) { + /* 4-ctx mode */ + *relreg = absreg_num & 0x1F; + *ctx = (absreg_num >> 0x4) & 0x6; + } else { + /* 8-ctx mode */ + *relreg = absreg_num & 0x0F; + *ctx = (absreg_num >> 0x4) & 0x7; + } + return 0; +} + +int qat_hal_init_gpr(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int regdata) +{ + int stat = 0; + unsigned short reg; + unsigned char ctx = 0; + enum icp_qat_uof_regtype type; + + if (reg_num >= ICP_QAT_UCLO_MAX_GPR_REG) + return -EINVAL; + + do { + if (ctx_mask == 0) { + qat_hal_convert_abs_to_rel(handle, ae, reg_num, ®, + &ctx); + type = reg_type - 1; + } else { + reg = reg_num; + type = reg_type; + if (!test_bit(ctx, &ctx_mask)) + continue; + } + stat = qat_hal_wr_rel_reg(handle, ae, ctx, type, reg, regdata); + if (stat) { + pr_err("QAT: write gpr fail\n"); + return -EINVAL; + } + } while (ctx_mask && (ctx++ < ICP_QAT_UCLO_MAX_CTX)); + + return 0; +} + +int qat_hal_init_wr_xfer(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int regdata) +{ + int stat = 0; + unsigned short reg; + unsigned char ctx = 0; + enum icp_qat_uof_regtype type; + + if (reg_num >= ICP_QAT_UCLO_MAX_XFER_REG) + return -EINVAL; + + do { + if (ctx_mask == 0) { + qat_hal_convert_abs_to_rel(handle, ae, reg_num, ®, + &ctx); + type = reg_type - 3; + } else { + reg = reg_num; + type = reg_type; + if (!test_bit(ctx, &ctx_mask)) + continue; + } + stat = qat_hal_put_rel_wr_xfer(handle, ae, ctx, type, reg, + regdata); + if (stat) { + pr_err("QAT: write wr xfer fail\n"); + return -EINVAL; + } + } while (ctx_mask && (ctx++ < ICP_QAT_UCLO_MAX_CTX)); + + return 0; +} + +int qat_hal_init_rd_xfer(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_num, unsigned int regdata) +{ + int stat = 0; + unsigned short reg; + unsigned char ctx = 0; + enum icp_qat_uof_regtype type; + + if (reg_num >= ICP_QAT_UCLO_MAX_XFER_REG) + return -EINVAL; + + do { + if (ctx_mask == 0) { + qat_hal_convert_abs_to_rel(handle, ae, reg_num, ®, + &ctx); + type = reg_type - 3; + } else { + reg = reg_num; + type = reg_type; + if (!test_bit(ctx, &ctx_mask)) + continue; + } + stat = qat_hal_put_rel_rd_xfer(handle, ae, ctx, type, reg, + regdata); + if (stat) { + pr_err("QAT: write rd xfer fail\n"); + return -EINVAL; + } + } while (ctx_mask && (ctx++ < ICP_QAT_UCLO_MAX_CTX)); + + return 0; +} + +int qat_hal_init_nn(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned long ctx_mask, + unsigned short reg_num, unsigned int regdata) +{ + int stat = 0; + unsigned char ctx; + if (!handle->chip_info->nn) { + dev_err(&handle->pci_dev->dev, "QAT: No next neigh in 0x%x\n", + handle->pci_dev->device); + return -EINVAL; + } + + if (ctx_mask == 0) + return -EINVAL; + + for (ctx = 0; ctx < ICP_QAT_UCLO_MAX_CTX; ctx++) { + if (!test_bit(ctx, &ctx_mask)) + continue; + stat = qat_hal_put_rel_nn(handle, ae, ctx, reg_num, regdata); + if (stat) { + pr_err("QAT: write neigh error\n"); + return -EINVAL; + } + } + + return 0; +} diff --git a/drivers/crypto/intel/qat/qat_common/qat_mig_dev.c b/drivers/crypto/intel/qat/qat_common/qat_mig_dev.c new file mode 100644 index 000000000000..892c2283a50e --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_mig_dev.c @@ -0,0 +1,130 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* Copyright(c) 2024 Intel Corporation */ +#include <linux/dev_printk.h> +#include <linux/export.h> +#include <linux/pci.h> +#include <linux/types.h> +#include <linux/qat/qat_mig_dev.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" + +struct qat_mig_dev *qat_vfmig_create(struct pci_dev *pdev, int vf_id) +{ + struct adf_accel_dev *accel_dev; + struct qat_migdev_ops *ops; + struct qat_mig_dev *mdev; + + accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + if (!accel_dev) + return ERR_PTR(-ENODEV); + + ops = GET_VFMIG_OPS(accel_dev); + if (!ops || !ops->init || !ops->cleanup || !ops->reset || !ops->open || + !ops->close || !ops->suspend || !ops->resume || !ops->save_state || + !ops->load_state || !ops->save_setup || !ops->load_setup) + return ERR_PTR(-EINVAL); + + mdev = kmalloc(sizeof(*mdev), GFP_KERNEL); + if (!mdev) + return ERR_PTR(-ENOMEM); + + mdev->vf_id = vf_id; + mdev->parent_accel_dev = accel_dev; + + return mdev; +} +EXPORT_SYMBOL_GPL(qat_vfmig_create); + +int qat_vfmig_init(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->init(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_init); + +void qat_vfmig_cleanup(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->cleanup(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_cleanup); + +void qat_vfmig_reset(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->reset(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_reset); + +int qat_vfmig_open(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->open(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_open); + +void qat_vfmig_close(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + GET_VFMIG_OPS(accel_dev)->close(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_close); + +int qat_vfmig_suspend(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->suspend(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_suspend); + +int qat_vfmig_resume(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->resume(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_resume); + +int qat_vfmig_save_state(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->save_state(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_save_state); + +int qat_vfmig_save_setup(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->save_setup(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_save_setup); + +int qat_vfmig_load_state(struct qat_mig_dev *mdev) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->load_state(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_load_state); + +int qat_vfmig_load_setup(struct qat_mig_dev *mdev, int size) +{ + struct adf_accel_dev *accel_dev = mdev->parent_accel_dev; + + return GET_VFMIG_OPS(accel_dev)->load_setup(mdev, size); +} +EXPORT_SYMBOL_GPL(qat_vfmig_load_setup); + +void qat_vfmig_destroy(struct qat_mig_dev *mdev) +{ + kfree(mdev); +} +EXPORT_SYMBOL_GPL(qat_vfmig_destroy); diff --git a/drivers/crypto/intel/qat/qat_common/qat_uclo.c b/drivers/crypto/intel/qat/qat_common/qat_uclo.c new file mode 100644 index 000000000000..06d49cb781ae --- /dev/null +++ b/drivers/crypto/intel/qat/qat_common/qat_uclo.c @@ -0,0 +1,2276 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ + +#define pr_fmt(fmt) "QAT: " fmt + +#include <linux/align.h> +#include <linux/bitops.h> +#include <linux/slab.h> +#include <linux/ctype.h> +#include <linux/kernel.h> +#include <linux/delay.h> +#include <linux/pci_ids.h> +#include <linux/wordpart.h> +#include "adf_accel_devices.h" +#include "adf_common_drv.h" +#include "icp_qat_uclo.h" +#include "icp_qat_hal.h" +#include "icp_qat_fw_loader_handle.h" + +#define UWORD_CPYBUF_SIZE 1024U +#define INVLD_UWORD 0xffffffffffull +#define PID_MINOR_REV 0xf +#define PID_MAJOR_REV (0xf << 4) + +static int qat_uclo_init_ae_data(struct icp_qat_uclo_objhandle *obj_handle, + unsigned int ae, unsigned int image_num) +{ + struct icp_qat_uclo_aedata *ae_data; + struct icp_qat_uclo_encapme *encap_image; + struct icp_qat_uclo_page *page = NULL; + struct icp_qat_uclo_aeslice *ae_slice = NULL; + + ae_data = &obj_handle->ae_data[ae]; + encap_image = &obj_handle->ae_uimage[image_num]; + ae_slice = &ae_data->ae_slices[ae_data->slice_num]; + ae_slice->encap_image = encap_image; + + if (encap_image->img_ptr) { + ae_slice->ctx_mask_assigned = + encap_image->img_ptr->ctx_assigned; + ae_data->eff_ustore_size = obj_handle->ustore_phy_size; + } else { + ae_slice->ctx_mask_assigned = 0; + } + ae_slice->region = kzalloc(sizeof(*ae_slice->region), GFP_KERNEL); + if (!ae_slice->region) + return -ENOMEM; + ae_slice->page = kzalloc(sizeof(*ae_slice->page), GFP_KERNEL); + if (!ae_slice->page) + goto out_err; + page = ae_slice->page; + page->encap_page = encap_image->page; + ae_slice->page->region = ae_slice->region; + ae_data->slice_num++; + return 0; +out_err: + kfree(ae_slice->region); + ae_slice->region = NULL; + return -ENOMEM; +} + +static int qat_uclo_free_ae_data(struct icp_qat_uclo_aedata *ae_data) +{ + unsigned int i; + + if (!ae_data) { + pr_err("bad argument, ae_data is NULL\n"); + return -EINVAL; + } + + for (i = 0; i < ae_data->slice_num; i++) { + kfree(ae_data->ae_slices[i].region); + ae_data->ae_slices[i].region = NULL; + kfree(ae_data->ae_slices[i].page); + ae_data->ae_slices[i].page = NULL; + } + return 0; +} + +static char *qat_uclo_get_string(struct icp_qat_uof_strtable *str_table, + unsigned int str_offset) +{ + if (!str_table->table_len || str_offset > str_table->table_len) + return NULL; + return (char *)(((uintptr_t)(str_table->strings)) + str_offset); +} + +static int qat_uclo_check_uof_format(struct icp_qat_uof_filehdr *hdr) +{ + int maj = hdr->maj_ver & 0xff; + int min = hdr->min_ver & 0xff; + + if (hdr->file_id != ICP_QAT_UOF_FID) { + pr_err("Invalid header 0x%x\n", hdr->file_id); + return -EINVAL; + } + if (min != ICP_QAT_UOF_MINVER || maj != ICP_QAT_UOF_MAJVER) { + pr_err("bad UOF version, major 0x%x, minor 0x%x\n", maj, min); + return -EINVAL; + } + return 0; +} + +static int qat_uclo_check_suof_format(struct icp_qat_suof_filehdr *suof_hdr) +{ + int maj = suof_hdr->maj_ver & 0xff; + int min = suof_hdr->min_ver & 0xff; + + if (suof_hdr->file_id != ICP_QAT_SUOF_FID) { + pr_err("invalid header 0x%x\n", suof_hdr->file_id); + return -EINVAL; + } + if (suof_hdr->fw_type != 0) { + pr_err("unsupported firmware type\n"); + return -EINVAL; + } + if (suof_hdr->num_chunks <= 0x1) { + pr_err("SUOF chunk amount is incorrect\n"); + return -EINVAL; + } + if (maj != ICP_QAT_SUOF_MAJVER || min != ICP_QAT_SUOF_MINVER) { + pr_err("bad SUOF version, major 0x%x, minor 0x%x\n", maj, min); + return -EINVAL; + } + return 0; +} + +static void qat_uclo_wr_sram_by_words(struct icp_qat_fw_loader_handle *handle, + unsigned int addr, unsigned int *val, + unsigned int num_in_bytes) +{ + unsigned int outval; + unsigned char *ptr = (unsigned char *)val; + + while (num_in_bytes) { + memcpy(&outval, ptr, 4); + SRAM_WRITE(handle, addr, outval); + num_in_bytes -= 4; + ptr += 4; + addr += 4; + } +} + +static void qat_uclo_wr_umem_by_words(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned int addr, + unsigned int *val, + unsigned int num_in_bytes) +{ + unsigned int outval; + unsigned char *ptr = (unsigned char *)val; + + addr >>= 0x2; /* convert to uword address */ + + while (num_in_bytes) { + memcpy(&outval, ptr, 4); + qat_hal_wr_umem(handle, ae, addr++, 1, &outval); + num_in_bytes -= 4; + ptr += 4; + } +} + +static void qat_uclo_batch_wr_umem(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, + struct icp_qat_uof_batch_init + *umem_init_header) +{ + struct icp_qat_uof_batch_init *umem_init; + + if (!umem_init_header) + return; + umem_init = umem_init_header->next; + while (umem_init) { + unsigned int addr, *value, size; + + ae = umem_init->ae; + addr = umem_init->addr; + value = umem_init->value; + size = umem_init->size; + qat_uclo_wr_umem_by_words(handle, ae, addr, value, size); + umem_init = umem_init->next; + } +} + +static void +qat_uclo_cleanup_batch_init_list(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uof_batch_init **base) +{ + struct icp_qat_uof_batch_init *umem_init; + + umem_init = *base; + while (umem_init) { + struct icp_qat_uof_batch_init *pre; + + pre = umem_init; + umem_init = umem_init->next; + kfree(pre); + } + *base = NULL; +} + +static int qat_uclo_parse_num(char *str, unsigned int *num) +{ + unsigned long long ae; + char *end; + + ae = simple_strtoull(str, &end, 10); + if (ae > UINT_MAX || str == end || (end - str) > 19) + return -EINVAL; + *num = (unsigned int)ae; + return 0; +} + +static int qat_uclo_fetch_initmem_ae(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uof_initmem *init_mem, + unsigned int size_range, unsigned int *ae) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + char *str; + + if ((init_mem->addr + init_mem->num_in_bytes) > (size_range << 0x2)) { + pr_err("initmem is out of range"); + return -EINVAL; + } + if (init_mem->scope != ICP_QAT_UOF_LOCAL_SCOPE) { + pr_err("Memory scope for init_mem error\n"); + return -EINVAL; + } + str = qat_uclo_get_string(&obj_handle->str_table, init_mem->sym_name); + if (!str) { + pr_err("AE name assigned in UOF init table is NULL\n"); + return -EINVAL; + } + if (qat_uclo_parse_num(str, ae)) { + pr_err("Parse num for AE number failed\n"); + return -EINVAL; + } + if (*ae >= ICP_QAT_UCLO_MAX_AE) { + pr_err("ae %d out of range\n", *ae); + return -EINVAL; + } + return 0; +} + +static int qat_uclo_create_batch_init_list(struct icp_qat_fw_loader_handle + *handle, struct icp_qat_uof_initmem + *init_mem, unsigned int ae, + struct icp_qat_uof_batch_init + **init_tab_base) +{ + struct icp_qat_uof_batch_init *init_header, *tail; + struct icp_qat_uof_batch_init *mem_init, *tail_old; + struct icp_qat_uof_memvar_attr *mem_val_attr; + unsigned int i, flag = 0; + + mem_val_attr = + (struct icp_qat_uof_memvar_attr *)((uintptr_t)init_mem + + sizeof(struct icp_qat_uof_initmem)); + + init_header = *init_tab_base; + if (!init_header) { + init_header = kzalloc(sizeof(*init_header), GFP_KERNEL); + if (!init_header) + return -ENOMEM; + init_header->size = 1; + *init_tab_base = init_header; + flag = 1; + } + tail_old = init_header; + while (tail_old->next) + tail_old = tail_old->next; + tail = tail_old; + for (i = 0; i < init_mem->val_attr_num; i++) { + mem_init = kzalloc(sizeof(*mem_init), GFP_KERNEL); + if (!mem_init) + goto out_err; + mem_init->ae = ae; + mem_init->addr = init_mem->addr + mem_val_attr->offset_in_byte; + mem_init->value = &mem_val_attr->value; + mem_init->size = 4; + mem_init->next = NULL; + tail->next = mem_init; + tail = mem_init; + init_header->size += qat_hal_get_ins_num(); + mem_val_attr++; + } + return 0; +out_err: + /* Do not free the list head unless we allocated it. */ + tail_old = tail_old->next; + if (flag) { + kfree(*init_tab_base); + *init_tab_base = NULL; + } + + while (tail_old) { + mem_init = tail_old->next; + kfree(tail_old); + tail_old = mem_init; + } + return -ENOMEM; +} + +static int qat_uclo_init_lmem_seg(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uof_initmem *init_mem) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned int ae; + + if (qat_uclo_fetch_initmem_ae(handle, init_mem, + handle->chip_info->lm_size, &ae)) + return -EINVAL; + if (qat_uclo_create_batch_init_list(handle, init_mem, ae, + &obj_handle->lm_init_tab[ae])) + return -EINVAL; + return 0; +} + +static int qat_uclo_init_umem_seg(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uof_initmem *init_mem) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned int ae, ustore_size, uaddr, i; + struct icp_qat_uclo_aedata *aed; + + ustore_size = obj_handle->ustore_phy_size; + if (qat_uclo_fetch_initmem_ae(handle, init_mem, ustore_size, &ae)) + return -EINVAL; + if (qat_uclo_create_batch_init_list(handle, init_mem, ae, + &obj_handle->umem_init_tab[ae])) + return -EINVAL; + /* set the highest ustore address referenced */ + uaddr = (init_mem->addr + init_mem->num_in_bytes) >> 0x2; + aed = &obj_handle->ae_data[ae]; + for (i = 0; i < aed->slice_num; i++) { + if (aed->ae_slices[i].encap_image->uwords_num < uaddr) + aed->ae_slices[i].encap_image->uwords_num = uaddr; + } + return 0; +} + +static int qat_uclo_init_ae_memory(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uof_initmem *init_mem) +{ + switch (init_mem->region) { + case ICP_QAT_UOF_LMEM_REGION: + if (qat_uclo_init_lmem_seg(handle, init_mem)) + return -EINVAL; + break; + case ICP_QAT_UOF_UMEM_REGION: + if (qat_uclo_init_umem_seg(handle, init_mem)) + return -EINVAL; + break; + default: + pr_err("initmem region error. region type=0x%x\n", init_mem->region); + return -EINVAL; + } + return 0; +} + +static int qat_uclo_init_ustore(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uclo_encapme *image) +{ + unsigned int i; + struct icp_qat_uclo_encap_page *page; + struct icp_qat_uof_image *uof_image; + unsigned char ae; + unsigned int ustore_size; + unsigned int patt_pos; + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned long cfg_ae_mask = handle->cfg_ae_mask; + u64 *fill_data; + + uof_image = image->img_ptr; + fill_data = kcalloc(ICP_QAT_UCLO_MAX_USTORE, sizeof(u64), + GFP_KERNEL); + if (!fill_data) + return -ENOMEM; + for (i = 0; i < ICP_QAT_UCLO_MAX_USTORE; i++) + memcpy(&fill_data[i], &uof_image->fill_pattern, + sizeof(u64)); + page = image->page; + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + unsigned long ae_assigned = uof_image->ae_assigned; + + if (!test_bit(ae, &ae_assigned)) + continue; + + if (!test_bit(ae, &cfg_ae_mask)) + continue; + + ustore_size = obj_handle->ae_data[ae].eff_ustore_size; + patt_pos = page->beg_addr_p + page->micro_words_num; + + qat_hal_wr_uwords(handle, (unsigned char)ae, 0, + page->beg_addr_p, &fill_data[0]); + qat_hal_wr_uwords(handle, (unsigned char)ae, patt_pos, + ustore_size - patt_pos + 1, + &fill_data[page->beg_addr_p]); + } + kfree(fill_data); + return 0; +} + +static int qat_uclo_init_memory(struct icp_qat_fw_loader_handle *handle) +{ + int i, ae; + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + struct icp_qat_uof_initmem *initmem = obj_handle->init_mem_tab.init_mem; + unsigned long ae_mask = handle->hal_handle->ae_mask; + + for (i = 0; i < obj_handle->init_mem_tab.entry_num; i++) { + if (initmem->num_in_bytes) { + if (qat_uclo_init_ae_memory(handle, initmem)) + return -EINVAL; + } + initmem = (struct icp_qat_uof_initmem *)((uintptr_t)( + (uintptr_t)initmem + + sizeof(struct icp_qat_uof_initmem)) + + (sizeof(struct icp_qat_uof_memvar_attr) * + initmem->val_attr_num)); + } + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + if (qat_hal_batch_wr_lm(handle, ae, + obj_handle->lm_init_tab[ae])) { + pr_err("fail to batch init lmem for AE %d\n", ae); + return -EINVAL; + } + qat_uclo_cleanup_batch_init_list(handle, + &obj_handle->lm_init_tab[ae]); + qat_uclo_batch_wr_umem(handle, ae, + obj_handle->umem_init_tab[ae]); + qat_uclo_cleanup_batch_init_list(handle, + &obj_handle-> + umem_init_tab[ae]); + } + return 0; +} + +static void *qat_uclo_find_chunk(struct icp_qat_uof_objhdr *obj_hdr, + char *chunk_id, void *cur) +{ + int i; + struct icp_qat_uof_chunkhdr *chunk_hdr = + (struct icp_qat_uof_chunkhdr *) + ((uintptr_t)obj_hdr + sizeof(struct icp_qat_uof_objhdr)); + + for (i = 0; i < obj_hdr->num_chunks; i++) { + if ((cur < (void *)&chunk_hdr[i]) && + !strncmp(chunk_hdr[i].chunk_id, chunk_id, + ICP_QAT_UOF_OBJID_LEN)) { + return &chunk_hdr[i]; + } + } + return NULL; +} + +static unsigned int qat_uclo_calc_checksum(unsigned int reg, int ch) +{ + int i; + unsigned int topbit = 1 << 0xF; + unsigned int inbyte = (unsigned int)((reg >> 0x18) ^ ch); + + reg ^= inbyte << 0x8; + for (i = 0; i < 0x8; i++) { + if (reg & topbit) + reg = (reg << 1) ^ 0x1021; + else + reg <<= 1; + } + return reg & 0xFFFF; +} + +static unsigned int qat_uclo_calc_str_checksum(char *ptr, int num) +{ + unsigned int chksum = 0; + + if (ptr) + while (num--) + chksum = qat_uclo_calc_checksum(chksum, *ptr++); + return chksum; +} + +static struct icp_qat_uclo_objhdr * +qat_uclo_map_chunk(char *buf, struct icp_qat_uof_filehdr *file_hdr, + char *chunk_id) +{ + struct icp_qat_uof_filechunkhdr *file_chunk; + struct icp_qat_uclo_objhdr *obj_hdr; + char *chunk; + int i; + + file_chunk = (struct icp_qat_uof_filechunkhdr *) + (buf + sizeof(struct icp_qat_uof_filehdr)); + for (i = 0; i < file_hdr->num_chunks; i++) { + if (!strncmp(file_chunk->chunk_id, chunk_id, + ICP_QAT_UOF_OBJID_LEN)) { + chunk = buf + file_chunk->offset; + if (file_chunk->checksum != qat_uclo_calc_str_checksum( + chunk, file_chunk->size)) + break; + obj_hdr = kzalloc(sizeof(*obj_hdr), GFP_KERNEL); + if (!obj_hdr) + break; + obj_hdr->file_buff = chunk; + obj_hdr->checksum = file_chunk->checksum; + obj_hdr->size = file_chunk->size; + return obj_hdr; + } + file_chunk++; + } + return NULL; +} + +static int +qat_uclo_check_image_compat(struct icp_qat_uof_encap_obj *encap_uof_obj, + struct icp_qat_uof_image *image) +{ + struct icp_qat_uof_objtable *uc_var_tab, *imp_var_tab, *imp_expr_tab; + struct icp_qat_uof_objtable *neigh_reg_tab; + struct icp_qat_uof_code_page *code_page; + + code_page = (struct icp_qat_uof_code_page *) + ((char *)image + sizeof(struct icp_qat_uof_image)); + uc_var_tab = (struct icp_qat_uof_objtable *)(encap_uof_obj->beg_uof + + code_page->uc_var_tab_offset); + imp_var_tab = (struct icp_qat_uof_objtable *)(encap_uof_obj->beg_uof + + code_page->imp_var_tab_offset); + imp_expr_tab = (struct icp_qat_uof_objtable *) + (encap_uof_obj->beg_uof + + code_page->imp_expr_tab_offset); + if (uc_var_tab->entry_num || imp_var_tab->entry_num || + imp_expr_tab->entry_num) { + pr_err("UOF can't contain imported variable to be parsed\n"); + return -EINVAL; + } + neigh_reg_tab = (struct icp_qat_uof_objtable *) + (encap_uof_obj->beg_uof + + code_page->neigh_reg_tab_offset); + if (neigh_reg_tab->entry_num) { + pr_err("UOF can't contain neighbor register table\n"); + return -EINVAL; + } + if (image->numpages > 1) { + pr_err("UOF can't contain multiple pages\n"); + return -EINVAL; + } + if (ICP_QAT_SHARED_USTORE_MODE(image->ae_mode)) { + pr_err("UOF can't use shared control store feature\n"); + return -EFAULT; + } + if (RELOADABLE_CTX_SHARED_MODE(image->ae_mode)) { + pr_err("UOF can't use reloadable feature\n"); + return -EFAULT; + } + return 0; +} + +static void qat_uclo_map_image_page(struct icp_qat_uof_encap_obj + *encap_uof_obj, + struct icp_qat_uof_image *img, + struct icp_qat_uclo_encap_page *page) +{ + struct icp_qat_uof_code_page *code_page; + struct icp_qat_uof_code_area *code_area; + struct icp_qat_uof_objtable *uword_block_tab; + struct icp_qat_uof_uword_block *uwblock; + int i; + + code_page = (struct icp_qat_uof_code_page *) + ((char *)img + sizeof(struct icp_qat_uof_image)); + page->def_page = code_page->def_page; + page->page_region = code_page->page_region; + page->beg_addr_v = code_page->beg_addr_v; + page->beg_addr_p = code_page->beg_addr_p; + code_area = (struct icp_qat_uof_code_area *)(encap_uof_obj->beg_uof + + code_page->code_area_offset); + page->micro_words_num = code_area->micro_words_num; + uword_block_tab = (struct icp_qat_uof_objtable *) + (encap_uof_obj->beg_uof + + code_area->uword_block_tab); + page->uwblock_num = uword_block_tab->entry_num; + uwblock = (struct icp_qat_uof_uword_block *)((char *)uword_block_tab + + sizeof(struct icp_qat_uof_objtable)); + page->uwblock = (struct icp_qat_uclo_encap_uwblock *)uwblock; + for (i = 0; i < uword_block_tab->entry_num; i++) + page->uwblock[i].micro_words = + (uintptr_t)encap_uof_obj->beg_uof + uwblock[i].uword_offset; +} + +static int qat_uclo_map_uimage(struct icp_qat_uclo_objhandle *obj_handle, + struct icp_qat_uclo_encapme *ae_uimage, + int max_image) +{ + int i, j; + struct icp_qat_uof_chunkhdr *chunk_hdr = NULL; + struct icp_qat_uof_image *image; + struct icp_qat_uof_objtable *ae_regtab; + struct icp_qat_uof_objtable *init_reg_sym_tab; + struct icp_qat_uof_objtable *sbreak_tab; + struct icp_qat_uof_encap_obj *encap_uof_obj = + &obj_handle->encap_uof_obj; + + for (j = 0; j < max_image; j++) { + chunk_hdr = qat_uclo_find_chunk(encap_uof_obj->obj_hdr, + ICP_QAT_UOF_IMAG, chunk_hdr); + if (!chunk_hdr) + break; + image = (struct icp_qat_uof_image *)(encap_uof_obj->beg_uof + + chunk_hdr->offset); + ae_regtab = (struct icp_qat_uof_objtable *) + (image->reg_tab_offset + + obj_handle->obj_hdr->file_buff); + ae_uimage[j].ae_reg_num = ae_regtab->entry_num; + ae_uimage[j].ae_reg = (struct icp_qat_uof_ae_reg *) + (((char *)ae_regtab) + + sizeof(struct icp_qat_uof_objtable)); + init_reg_sym_tab = (struct icp_qat_uof_objtable *) + (image->init_reg_sym_tab + + obj_handle->obj_hdr->file_buff); + ae_uimage[j].init_regsym_num = init_reg_sym_tab->entry_num; + ae_uimage[j].init_regsym = (struct icp_qat_uof_init_regsym *) + (((char *)init_reg_sym_tab) + + sizeof(struct icp_qat_uof_objtable)); + sbreak_tab = (struct icp_qat_uof_objtable *) + (image->sbreak_tab + obj_handle->obj_hdr->file_buff); + ae_uimage[j].sbreak_num = sbreak_tab->entry_num; + ae_uimage[j].sbreak = (struct icp_qat_uof_sbreak *) + (((char *)sbreak_tab) + + sizeof(struct icp_qat_uof_objtable)); + ae_uimage[j].img_ptr = image; + if (qat_uclo_check_image_compat(encap_uof_obj, image)) + goto out_err; + ae_uimage[j].page = + kzalloc(sizeof(struct icp_qat_uclo_encap_page), + GFP_KERNEL); + if (!ae_uimage[j].page) + goto out_err; + qat_uclo_map_image_page(encap_uof_obj, image, + ae_uimage[j].page); + } + return j; +out_err: + for (i = 0; i < j; i++) + kfree(ae_uimage[i].page); + return 0; +} + +static int qat_uclo_map_ae(struct icp_qat_fw_loader_handle *handle, int max_ae) +{ + int i, ae; + int mflag = 0; + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned long cfg_ae_mask = handle->cfg_ae_mask; + + for_each_set_bit(ae, &ae_mask, max_ae) { + if (!test_bit(ae, &cfg_ae_mask)) + continue; + + for (i = 0; i < obj_handle->uimage_num; i++) { + unsigned long ae_assigned = obj_handle->ae_uimage[i].img_ptr->ae_assigned; + + if (!test_bit(ae, &ae_assigned)) + continue; + mflag = 1; + if (qat_uclo_init_ae_data(obj_handle, ae, i)) + return -EINVAL; + } + } + if (!mflag) { + pr_err("uimage uses AE not set\n"); + return -EINVAL; + } + return 0; +} + +static struct icp_qat_uof_strtable * +qat_uclo_map_str_table(struct icp_qat_uclo_objhdr *obj_hdr, + char *tab_name, struct icp_qat_uof_strtable *str_table) +{ + struct icp_qat_uof_chunkhdr *chunk_hdr; + + chunk_hdr = qat_uclo_find_chunk((struct icp_qat_uof_objhdr *) + obj_hdr->file_buff, tab_name, NULL); + if (chunk_hdr) { + int hdr_size; + + memcpy(&str_table->table_len, obj_hdr->file_buff + + chunk_hdr->offset, sizeof(str_table->table_len)); + hdr_size = (char *)&str_table->strings - (char *)str_table; + str_table->strings = (uintptr_t)obj_hdr->file_buff + + chunk_hdr->offset + hdr_size; + return str_table; + } + return NULL; +} + +static void +qat_uclo_map_initmem_table(struct icp_qat_uof_encap_obj *encap_uof_obj, + struct icp_qat_uclo_init_mem_table *init_mem_tab) +{ + struct icp_qat_uof_chunkhdr *chunk_hdr; + + chunk_hdr = qat_uclo_find_chunk(encap_uof_obj->obj_hdr, + ICP_QAT_UOF_IMEM, NULL); + if (chunk_hdr) { + memmove(&init_mem_tab->entry_num, encap_uof_obj->beg_uof + + chunk_hdr->offset, sizeof(unsigned int)); + init_mem_tab->init_mem = (struct icp_qat_uof_initmem *) + (encap_uof_obj->beg_uof + chunk_hdr->offset + + sizeof(unsigned int)); + } +} + +static unsigned int +qat_uclo_get_dev_type(struct icp_qat_fw_loader_handle *handle) +{ + switch (handle->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_DH895XCC: + return ICP_QAT_AC_895XCC_DEV_TYPE; + case PCI_DEVICE_ID_INTEL_QAT_C62X: + return ICP_QAT_AC_C62X_DEV_TYPE; + case PCI_DEVICE_ID_INTEL_QAT_C3XXX: + return ICP_QAT_AC_C3XXX_DEV_TYPE; + case PCI_DEVICE_ID_INTEL_QAT_4XXX: + case PCI_DEVICE_ID_INTEL_QAT_401XX: + case PCI_DEVICE_ID_INTEL_QAT_402XX: + case PCI_DEVICE_ID_INTEL_QAT_420XX: + return ICP_QAT_AC_4XXX_A_DEV_TYPE; + case PCI_DEVICE_ID_INTEL_QAT_6XXX: + return ICP_QAT_AC_6XXX_DEV_TYPE; + default: + pr_err("unsupported device 0x%x\n", handle->pci_dev->device); + return 0; + } +} + +static int qat_uclo_check_uof_compat(struct icp_qat_uclo_objhandle *obj_handle) +{ + unsigned int maj_ver, prod_type = obj_handle->prod_type; + + if (!(prod_type & obj_handle->encap_uof_obj.obj_hdr->ac_dev_type)) { + pr_err("UOF type 0x%x doesn't match with platform 0x%x\n", + obj_handle->encap_uof_obj.obj_hdr->ac_dev_type, + prod_type); + return -EINVAL; + } + maj_ver = obj_handle->prod_rev & 0xff; + if (obj_handle->encap_uof_obj.obj_hdr->max_cpu_ver < maj_ver || + obj_handle->encap_uof_obj.obj_hdr->min_cpu_ver > maj_ver) { + pr_err("UOF majVer 0x%x out of range\n", maj_ver); + return -EINVAL; + } + return 0; +} + +static int qat_uclo_init_reg(struct icp_qat_fw_loader_handle *handle, + unsigned char ae, unsigned char ctx_mask, + enum icp_qat_uof_regtype reg_type, + unsigned short reg_addr, unsigned int value) +{ + switch (reg_type) { + case ICP_GPA_ABS: + case ICP_GPB_ABS: + ctx_mask = 0; + fallthrough; + case ICP_GPA_REL: + case ICP_GPB_REL: + return qat_hal_init_gpr(handle, ae, ctx_mask, reg_type, + reg_addr, value); + case ICP_SR_ABS: + case ICP_DR_ABS: + case ICP_SR_RD_ABS: + case ICP_DR_RD_ABS: + ctx_mask = 0; + fallthrough; + case ICP_SR_REL: + case ICP_DR_REL: + case ICP_SR_RD_REL: + case ICP_DR_RD_REL: + return qat_hal_init_rd_xfer(handle, ae, ctx_mask, reg_type, + reg_addr, value); + case ICP_SR_WR_ABS: + case ICP_DR_WR_ABS: + ctx_mask = 0; + fallthrough; + case ICP_SR_WR_REL: + case ICP_DR_WR_REL: + return qat_hal_init_wr_xfer(handle, ae, ctx_mask, reg_type, + reg_addr, value); + case ICP_NEIGH_REL: + return qat_hal_init_nn(handle, ae, ctx_mask, reg_addr, value); + default: + pr_err("UOF uses not supported reg type 0x%x\n", reg_type); + return -EFAULT; + } + return 0; +} + +static int qat_uclo_init_reg_sym(struct icp_qat_fw_loader_handle *handle, + unsigned int ae, + struct icp_qat_uclo_encapme *encap_ae) +{ + unsigned int i; + unsigned char ctx_mask; + struct icp_qat_uof_init_regsym *init_regsym; + + if (ICP_QAT_CTX_MODE(encap_ae->img_ptr->ae_mode) == + ICP_QAT_UCLO_MAX_CTX) + ctx_mask = 0xff; + else + ctx_mask = 0x55; + + for (i = 0; i < encap_ae->init_regsym_num; i++) { + unsigned int exp_res; + + init_regsym = &encap_ae->init_regsym[i]; + exp_res = init_regsym->value; + switch (init_regsym->init_type) { + case ICP_QAT_UOF_INIT_REG: + qat_uclo_init_reg(handle, ae, ctx_mask, + (enum icp_qat_uof_regtype) + init_regsym->reg_type, + (unsigned short)init_regsym->reg_addr, + exp_res); + break; + case ICP_QAT_UOF_INIT_REG_CTX: + /* check if ctx is appropriate for the ctxMode */ + if (!((1 << init_regsym->ctx) & ctx_mask)) { + pr_err("invalid ctx num = 0x%x\n", init_regsym->ctx); + return -EINVAL; + } + qat_uclo_init_reg(handle, ae, + (unsigned char) + (1 << init_regsym->ctx), + (enum icp_qat_uof_regtype) + init_regsym->reg_type, + (unsigned short)init_regsym->reg_addr, + exp_res); + break; + case ICP_QAT_UOF_INIT_EXPR: + pr_err("INIT_EXPR feature not supported\n"); + return -EINVAL; + case ICP_QAT_UOF_INIT_EXPR_ENDIAN_SWAP: + pr_err("INIT_EXPR_ENDIAN_SWAP feature not supported\n"); + return -EINVAL; + default: + break; + } + } + return 0; +} + +static int qat_uclo_init_globals(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned long ae_mask = handle->hal_handle->ae_mask; + struct icp_qat_uclo_aedata *aed; + unsigned int s, ae; + + if (obj_handle->global_inited) + return 0; + if (obj_handle->init_mem_tab.entry_num) { + if (qat_uclo_init_memory(handle)) { + pr_err("initialize memory failed\n"); + return -EINVAL; + } + } + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + aed = &obj_handle->ae_data[ae]; + for (s = 0; s < aed->slice_num; s++) { + if (!aed->ae_slices[s].encap_image) + continue; + if (qat_uclo_init_reg_sym(handle, ae, aed->ae_slices[s].encap_image)) + return -EINVAL; + } + } + obj_handle->global_inited = 1; + return 0; +} + +static int qat_hal_set_modes(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uclo_objhandle *obj_handle, + unsigned char ae, + struct icp_qat_uof_image *uof_image) +{ + unsigned char mode; + int ret; + + mode = ICP_QAT_CTX_MODE(uof_image->ae_mode); + ret = qat_hal_set_ae_ctx_mode(handle, ae, mode); + if (ret) { + pr_err("qat_hal_set_ae_ctx_mode error\n"); + return ret; + } + if (handle->chip_info->nn) { + mode = ICP_QAT_NN_MODE(uof_image->ae_mode); + ret = qat_hal_set_ae_nn_mode(handle, ae, mode); + if (ret) { + pr_err("qat_hal_set_ae_nn_mode error\n"); + return ret; + } + } + mode = ICP_QAT_LOC_MEM0_MODE(uof_image->ae_mode); + ret = qat_hal_set_ae_lm_mode(handle, ae, ICP_LMEM0, mode); + if (ret) { + pr_err("qat_hal_set_ae_lm_mode LMEM0 error\n"); + return ret; + } + mode = ICP_QAT_LOC_MEM1_MODE(uof_image->ae_mode); + ret = qat_hal_set_ae_lm_mode(handle, ae, ICP_LMEM1, mode); + if (ret) { + pr_err("qat_hal_set_ae_lm_mode LMEM1 error\n"); + return ret; + } + if (handle->chip_info->lm2lm3) { + mode = ICP_QAT_LOC_MEM2_MODE(uof_image->ae_mode); + ret = qat_hal_set_ae_lm_mode(handle, ae, ICP_LMEM2, mode); + if (ret) { + pr_err("qat_hal_set_ae_lm_mode LMEM2 error\n"); + return ret; + } + mode = ICP_QAT_LOC_MEM3_MODE(uof_image->ae_mode); + ret = qat_hal_set_ae_lm_mode(handle, ae, ICP_LMEM3, mode); + if (ret) { + pr_err("qat_hal_set_ae_lm_mode LMEM3 error\n"); + return ret; + } + mode = ICP_QAT_LOC_TINDEX_MODE(uof_image->ae_mode); + qat_hal_set_ae_tindex_mode(handle, ae, mode); + } + return 0; +} + +static int qat_uclo_set_ae_mode(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_uof_image *uof_image; + struct icp_qat_uclo_aedata *ae_data; + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned long cfg_ae_mask = handle->cfg_ae_mask; + unsigned char ae, s; + int error; + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + if (!test_bit(ae, &cfg_ae_mask)) + continue; + + ae_data = &obj_handle->ae_data[ae]; + for (s = 0; s < min_t(unsigned int, ae_data->slice_num, + ICP_QAT_UCLO_MAX_CTX); s++) { + if (!obj_handle->ae_data[ae].ae_slices[s].encap_image) + continue; + uof_image = ae_data->ae_slices[s].encap_image->img_ptr; + error = qat_hal_set_modes(handle, obj_handle, ae, + uof_image); + if (error) + return error; + } + } + return 0; +} + +static void qat_uclo_init_uword_num(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + struct icp_qat_uclo_encapme *image; + int a; + + for (a = 0; a < obj_handle->uimage_num; a++) { + image = &obj_handle->ae_uimage[a]; + image->uwords_num = image->page->beg_addr_p + + image->page->micro_words_num; + } +} + +static int qat_uclo_parse_uof_obj(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned int ae; + + obj_handle->encap_uof_obj.beg_uof = obj_handle->obj_hdr->file_buff; + obj_handle->encap_uof_obj.obj_hdr = (struct icp_qat_uof_objhdr *) + obj_handle->obj_hdr->file_buff; + obj_handle->uword_in_bytes = 6; + obj_handle->prod_type = qat_uclo_get_dev_type(handle); + obj_handle->prod_rev = PID_MAJOR_REV | + (PID_MINOR_REV & handle->hal_handle->revision_id); + if (qat_uclo_check_uof_compat(obj_handle)) { + pr_err("UOF incompatible\n"); + return -EINVAL; + } + obj_handle->uword_buf = kcalloc(UWORD_CPYBUF_SIZE, sizeof(u64), + GFP_KERNEL); + if (!obj_handle->uword_buf) + return -ENOMEM; + obj_handle->ustore_phy_size = ICP_QAT_UCLO_MAX_USTORE; + if (!obj_handle->obj_hdr->file_buff || + !qat_uclo_map_str_table(obj_handle->obj_hdr, ICP_QAT_UOF_STRT, + &obj_handle->str_table)) { + pr_err("UOF doesn't have effective images\n"); + goto out_err; + } + obj_handle->uimage_num = + qat_uclo_map_uimage(obj_handle, obj_handle->ae_uimage, + ICP_QAT_UCLO_MAX_AE * ICP_QAT_UCLO_MAX_CTX); + if (!obj_handle->uimage_num) + goto out_err; + if (qat_uclo_map_ae(handle, handle->hal_handle->ae_max_num)) { + pr_err("Bad object\n"); + goto out_check_uof_aemask_err; + } + qat_uclo_init_uword_num(handle); + qat_uclo_map_initmem_table(&obj_handle->encap_uof_obj, + &obj_handle->init_mem_tab); + if (qat_uclo_set_ae_mode(handle)) + goto out_check_uof_aemask_err; + return 0; +out_check_uof_aemask_err: + for (ae = 0; ae < obj_handle->uimage_num; ae++) + kfree(obj_handle->ae_uimage[ae].page); +out_err: + kfree(obj_handle->uword_buf); + return -EFAULT; +} + +static unsigned int qat_uclo_simg_hdr2sign_len(struct icp_qat_fw_loader_handle *handle) +{ + if (handle->chip_info->dual_sign) + return ICP_QAT_DUALSIGN_OPAQUE_DATA_LEN; + + return ICP_QAT_AE_IMG_OFFSET(handle); +} + +static unsigned int qat_uclo_simg_hdr2cont_len(struct icp_qat_fw_loader_handle *handle) +{ + if (handle->chip_info->dual_sign) + return ICP_QAT_DUALSIGN_OPAQUE_DATA_LEN + ICP_QAT_DUALSIGN_MISC_INFO_LEN; + + return ICP_QAT_AE_IMG_OFFSET(handle); +} + +static unsigned int qat_uclo_simg_fw_type(struct icp_qat_fw_loader_handle *handle, void *img_ptr) +{ + struct icp_qat_css_hdr *hdr = img_ptr; + char *fw_hdr = img_ptr; + unsigned int offset; + + if (handle->chip_info->dual_sign) { + offset = qat_uclo_simg_hdr2sign_len(handle) + ICP_QAT_DUALSIGN_FW_TYPE_LEN; + return *(fw_hdr + offset); + } + + return hdr->fw_type; +} + +static int qat_uclo_map_suof_file_hdr(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_suof_filehdr *suof_ptr, + int suof_size) +{ + unsigned int check_sum = 0; + unsigned int min_ver_offset = 0; + struct icp_qat_suof_handle *suof_handle = handle->sobj_handle; + + suof_handle->file_id = ICP_QAT_SUOF_FID; + suof_handle->suof_buf = (char *)suof_ptr; + suof_handle->suof_size = suof_size; + min_ver_offset = suof_size - offsetof(struct icp_qat_suof_filehdr, + min_ver); + check_sum = qat_uclo_calc_str_checksum((char *)&suof_ptr->min_ver, + min_ver_offset); + if (check_sum != suof_ptr->check_sum) { + pr_err("incorrect SUOF checksum\n"); + return -EINVAL; + } + suof_handle->check_sum = suof_ptr->check_sum; + suof_handle->min_ver = suof_ptr->min_ver; + suof_handle->maj_ver = suof_ptr->maj_ver; + suof_handle->fw_type = suof_ptr->fw_type; + return 0; +} + +static void qat_uclo_map_simg(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_suof_img_hdr *suof_img_hdr, + struct icp_qat_suof_chunk_hdr *suof_chunk_hdr) +{ + struct icp_qat_suof_handle *suof_handle = handle->sobj_handle; + unsigned int offset = qat_uclo_simg_hdr2cont_len(handle); + struct icp_qat_suof_objhdr *suof_objhdr; + struct icp_qat_simg_ae_mode *ae_mode; + + suof_img_hdr->simg_buf = (suof_handle->suof_buf + + suof_chunk_hdr->offset + + sizeof(*suof_objhdr)); + suof_img_hdr->simg_len = ((struct icp_qat_suof_objhdr *)(uintptr_t) + (suof_handle->suof_buf + + suof_chunk_hdr->offset))->img_length; + + suof_img_hdr->css_header = suof_img_hdr->simg_buf; + suof_img_hdr->css_simg = suof_img_hdr->css_header + offset; + + ae_mode = (struct icp_qat_simg_ae_mode *)(suof_img_hdr->css_simg); + suof_img_hdr->ae_mask = ae_mode->ae_mask; + suof_img_hdr->simg_name = (unsigned long)&ae_mode->simg_name; + suof_img_hdr->appmeta_data = (unsigned long)&ae_mode->appmeta_data; + suof_img_hdr->fw_type = ae_mode->fw_type; +} + +static void +qat_uclo_map_suof_symobjs(struct icp_qat_suof_handle *suof_handle, + struct icp_qat_suof_chunk_hdr *suof_chunk_hdr) +{ + char **sym_str = (char **)&suof_handle->sym_str; + unsigned int *sym_size = &suof_handle->sym_size; + struct icp_qat_suof_strtable *str_table_obj; + + *sym_size = *(unsigned int *)(uintptr_t) + (suof_chunk_hdr->offset + suof_handle->suof_buf); + *sym_str = (char *)(uintptr_t) + (suof_handle->suof_buf + suof_chunk_hdr->offset + + sizeof(str_table_obj->tab_length)); +} + +static int qat_uclo_check_simg_compat(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_suof_img_hdr *img_hdr) +{ + struct icp_qat_simg_ae_mode *img_ae_mode = NULL; + unsigned int prod_rev, maj_ver, prod_type; + + prod_type = qat_uclo_get_dev_type(handle); + img_ae_mode = (struct icp_qat_simg_ae_mode *)img_hdr->css_simg; + prod_rev = PID_MAJOR_REV | + (PID_MINOR_REV & handle->hal_handle->revision_id); + if (img_ae_mode->dev_type != prod_type) { + pr_err("incompatible product type %x\n", img_ae_mode->dev_type); + return -EINVAL; + } + maj_ver = prod_rev & 0xff; + if (maj_ver > img_ae_mode->devmax_ver || + maj_ver < img_ae_mode->devmin_ver) { + pr_err("incompatible device majver 0x%x\n", maj_ver); + return -EINVAL; + } + return 0; +} + +static void qat_uclo_del_suof(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_suof_handle *sobj_handle = handle->sobj_handle; + + kfree(sobj_handle->img_table.simg_hdr); + sobj_handle->img_table.simg_hdr = NULL; + kfree(handle->sobj_handle); + handle->sobj_handle = NULL; +} + +static void qat_uclo_tail_img(struct icp_qat_suof_img_hdr *suof_img_hdr, + unsigned int img_id, unsigned int num_simgs) +{ + struct icp_qat_suof_img_hdr img_header; + + if (img_id != num_simgs - 1) { + memcpy(&img_header, &suof_img_hdr[num_simgs - 1], + sizeof(*suof_img_hdr)); + memcpy(&suof_img_hdr[num_simgs - 1], &suof_img_hdr[img_id], + sizeof(*suof_img_hdr)); + memcpy(&suof_img_hdr[img_id], &img_header, + sizeof(*suof_img_hdr)); + } +} + +static int qat_uclo_map_suof(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_suof_filehdr *suof_ptr, + int suof_size) +{ + struct icp_qat_suof_handle *suof_handle = handle->sobj_handle; + struct icp_qat_suof_chunk_hdr *suof_chunk_hdr = NULL; + struct icp_qat_suof_img_hdr *suof_img_hdr = NULL; + int ret = 0, ae0_img = ICP_QAT_UCLO_MAX_AE; + unsigned int i = 0; + struct icp_qat_suof_img_hdr img_header; + + if (!suof_ptr || suof_size == 0) { + pr_err("input parameter SUOF pointer/size is NULL\n"); + return -EINVAL; + } + if (qat_uclo_check_suof_format(suof_ptr)) + return -EINVAL; + ret = qat_uclo_map_suof_file_hdr(handle, suof_ptr, suof_size); + if (ret) + return ret; + suof_chunk_hdr = (struct icp_qat_suof_chunk_hdr *) + ((uintptr_t)suof_ptr + sizeof(*suof_ptr)); + + qat_uclo_map_suof_symobjs(suof_handle, suof_chunk_hdr); + suof_handle->img_table.num_simgs = suof_ptr->num_chunks - 1; + + if (suof_handle->img_table.num_simgs != 0) { + suof_img_hdr = kcalloc(suof_handle->img_table.num_simgs, + sizeof(img_header), + GFP_KERNEL); + if (!suof_img_hdr) + return -ENOMEM; + suof_handle->img_table.simg_hdr = suof_img_hdr; + + for (i = 0; i < suof_handle->img_table.num_simgs; i++) { + qat_uclo_map_simg(handle, &suof_img_hdr[i], + &suof_chunk_hdr[1 + i]); + ret = qat_uclo_check_simg_compat(handle, + &suof_img_hdr[i]); + if (ret) + return ret; + suof_img_hdr[i].ae_mask &= handle->cfg_ae_mask; + if ((suof_img_hdr[i].ae_mask & 0x1) != 0) + ae0_img = i; + } + + if (!handle->chip_info->tgroup_share_ustore) { + qat_uclo_tail_img(suof_img_hdr, ae0_img, + suof_handle->img_table.num_simgs); + } + } + return 0; +} + +#define ADD_ADDR(high, low) ((((u64)high) << 32) + low) + +static int qat_uclo_auth_fw(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_fw_auth_desc *desc) +{ + u32 fcu_sts, retry = 0; + u32 fcu_ctl_csr, fcu_sts_csr; + u32 fcu_dram_hi_csr, fcu_dram_lo_csr; + u64 bus_addr; + + bus_addr = ADD_ADDR(desc->css_hdr_high, desc->css_hdr_low) + - sizeof(struct icp_qat_auth_chunk); + + fcu_ctl_csr = handle->chip_info->fcu_ctl_csr; + fcu_sts_csr = handle->chip_info->fcu_sts_csr; + fcu_dram_hi_csr = handle->chip_info->fcu_dram_addr_hi; + fcu_dram_lo_csr = handle->chip_info->fcu_dram_addr_lo; + + SET_CAP_CSR(handle, fcu_dram_hi_csr, bus_addr >> BITS_PER_TYPE(u32)); + SET_CAP_CSR(handle, fcu_dram_lo_csr, bus_addr); + SET_CAP_CSR(handle, fcu_ctl_csr, FCU_CTRL_CMD_AUTH); + + do { + msleep(FW_AUTH_WAIT_PERIOD); + fcu_sts = GET_CAP_CSR(handle, fcu_sts_csr); + if ((fcu_sts & FCU_AUTH_STS_MASK) == FCU_STS_VERI_FAIL) + goto auth_fail; + if (((fcu_sts >> FCU_STS_AUTHFWLD_POS) & 0x1)) + if ((fcu_sts & FCU_AUTH_STS_MASK) == FCU_STS_VERI_DONE) + return 0; + } while (retry++ < FW_AUTH_MAX_RETRY); +auth_fail: + pr_err("authentication error (FCU_STATUS = 0x%x),retry = %d\n", + fcu_sts & FCU_AUTH_STS_MASK, retry); + return -EINVAL; +} + +static bool qat_uclo_is_broadcast(struct icp_qat_fw_loader_handle *handle, + int imgid) +{ + struct icp_qat_suof_handle *sobj_handle; + + if (!handle->chip_info->tgroup_share_ustore) + return false; + + sobj_handle = (struct icp_qat_suof_handle *)handle->sobj_handle; + if (handle->hal_handle->admin_ae_mask & + sobj_handle->img_table.simg_hdr[imgid].ae_mask) + return false; + + return true; +} + +static int qat_uclo_broadcast_load_fw(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_fw_auth_desc *desc) +{ + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned long desc_ae_mask = desc->ae_mask; + u32 fcu_sts, ae_broadcast_mask = 0; + u32 fcu_loaded_csr, ae_loaded; + u32 fcu_sts_csr, fcu_ctl_csr; + unsigned int ae, retry = 0; + + if (handle->chip_info->tgroup_share_ustore) { + fcu_ctl_csr = handle->chip_info->fcu_ctl_csr; + fcu_sts_csr = handle->chip_info->fcu_sts_csr; + fcu_loaded_csr = handle->chip_info->fcu_loaded_ae_csr; + } else { + pr_err("Chip 0x%x doesn't support broadcast load\n", handle->pci_dev->device); + return -EINVAL; + } + + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + if (qat_hal_check_ae_active(handle, (unsigned char)ae)) { + pr_err("Broadcast load failed. AE is not enabled or active.\n"); + return -EINVAL; + } + + if (test_bit(ae, &desc_ae_mask)) + ae_broadcast_mask |= 1 << ae; + } + + if (ae_broadcast_mask) { + SET_CAP_CSR(handle, FCU_ME_BROADCAST_MASK_TYPE, + ae_broadcast_mask); + + SET_CAP_CSR(handle, fcu_ctl_csr, FCU_CTRL_CMD_LOAD); + + do { + msleep(FW_AUTH_WAIT_PERIOD); + fcu_sts = GET_CAP_CSR(handle, fcu_sts_csr); + fcu_sts &= FCU_AUTH_STS_MASK; + + if (fcu_sts == FCU_STS_LOAD_FAIL) { + pr_err("Broadcast load failed: 0x%x)\n", fcu_sts); + return -EINVAL; + } else if (fcu_sts == FCU_STS_LOAD_DONE) { + ae_loaded = GET_CAP_CSR(handle, fcu_loaded_csr); + ae_loaded >>= handle->chip_info->fcu_loaded_ae_pos; + + if ((ae_loaded & ae_broadcast_mask) == ae_broadcast_mask) + break; + } + } while (retry++ < FW_AUTH_MAX_RETRY); + + if (retry > FW_AUTH_MAX_RETRY) { + pr_err("broadcast load failed timeout %d\n", retry); + return -EINVAL; + } + } + return 0; +} + +static int qat_uclo_simg_alloc(struct icp_qat_fw_loader_handle *handle, + struct icp_firml_dram_desc *dram_desc, + unsigned int size) +{ + void *vptr; + dma_addr_t ptr; + + vptr = dma_alloc_coherent(&handle->pci_dev->dev, + size, &ptr, GFP_KERNEL); + if (!vptr) + return -ENOMEM; + dram_desc->dram_base_addr_v = vptr; + dram_desc->dram_bus_addr = ptr; + dram_desc->dram_size = size; + return 0; +} + +static void qat_uclo_simg_free(struct icp_qat_fw_loader_handle *handle, + struct icp_firml_dram_desc *dram_desc) +{ + if (handle && dram_desc && dram_desc->dram_base_addr_v) { + dma_free_coherent(&handle->pci_dev->dev, + (size_t)(dram_desc->dram_size), + dram_desc->dram_base_addr_v, + dram_desc->dram_bus_addr); + } + + if (dram_desc) + memset(dram_desc, 0, sizeof(*dram_desc)); +} + +static void qat_uclo_ummap_auth_fw(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_fw_auth_desc **desc) +{ + struct icp_firml_dram_desc dram_desc; + + if (*desc) { + dram_desc.dram_base_addr_v = *desc; + dram_desc.dram_bus_addr = ((struct icp_qat_auth_chunk *) + (*desc))->chunk_bus_addr; + dram_desc.dram_size = ((struct icp_qat_auth_chunk *) + (*desc))->chunk_size; + qat_uclo_simg_free(handle, &dram_desc); + } +} + +static int qat_uclo_check_image(struct icp_qat_fw_loader_handle *handle, + void *image, unsigned int size, + unsigned int fw_type) +{ + char *fw_type_name = fw_type ? "MMP" : "AE"; + unsigned int css_dword_size = sizeof(u32); + unsigned int header_len, simg_type; + struct icp_qat_css_hdr *css_hdr; + + if (handle->chip_info->fw_auth) { + header_len = qat_uclo_simg_hdr2sign_len(handle); + simg_type = qat_uclo_simg_fw_type(handle, image); + css_hdr = image; + + if (handle->chip_info->dual_sign) { + if (css_hdr->module_type != ICP_QAT_DUALSIGN_MODULE_TYPE) + goto err; + if (css_hdr->header_len != ICP_QAT_DUALSIGN_HDR_LEN) + goto err; + if (css_hdr->header_ver != ICP_QAT_DUALSIGN_HDR_VER) + goto err; + } else { + if (css_hdr->header_len * css_dword_size != header_len) + goto err; + if (css_hdr->size * css_dword_size != size) + goto err; + if (size <= header_len) + goto err; + } + + if (fw_type != simg_type) + goto err; + + size -= header_len; + } + + if (fw_type == CSS_AE_FIRMWARE) { + if (size < sizeof(struct icp_qat_simg_ae_mode *) + + ICP_QAT_SIMG_AE_INIT_SEQ_LEN) + goto err; + if (size > ICP_QAT_CSS_RSA4K_MAX_IMAGE_LEN) + goto err; + } else if (fw_type == CSS_MMP_FIRMWARE) { + if (size > ICP_QAT_CSS_RSA3K_MAX_IMAGE_LEN) + goto err; + } else { + pr_err("Unsupported firmware type\n"); + return -EINVAL; + } + return 0; + +err: + pr_err("Invalid %s firmware image\n", fw_type_name); + return -EINVAL; +} + +static int qat_uclo_build_auth_desc_RSA(struct icp_qat_fw_loader_handle *handle, + char *image, unsigned int size, + struct icp_firml_dram_desc *dram_desc, + unsigned int fw_type, struct icp_qat_fw_auth_desc **desc) +{ + struct icp_qat_css_hdr *css_hdr = (struct icp_qat_css_hdr *)image; + struct icp_qat_simg_ae_mode *simg_ae_mode; + struct icp_qat_fw_auth_desc *auth_desc; + char *virt_addr, *virt_base; + u64 bus_addr; + + virt_base = dram_desc->dram_base_addr_v; + virt_base += sizeof(struct icp_qat_auth_chunk); + bus_addr = dram_desc->dram_bus_addr + sizeof(struct icp_qat_auth_chunk); + auth_desc = dram_desc->dram_base_addr_v; + auth_desc->css_hdr_high = upper_32_bits(bus_addr); + auth_desc->css_hdr_low = lower_32_bits(bus_addr); + virt_addr = virt_base; + + memcpy(virt_addr, image, sizeof(*css_hdr)); + /* pub key */ + bus_addr = ADD_ADDR(auth_desc->css_hdr_high, auth_desc->css_hdr_low) + + sizeof(*css_hdr); + virt_addr = virt_addr + sizeof(*css_hdr); + + auth_desc->fwsk_pub_high = upper_32_bits(bus_addr); + auth_desc->fwsk_pub_low = lower_32_bits(bus_addr); + + memcpy(virt_addr, image + sizeof(*css_hdr), ICP_QAT_CSS_FWSK_MODULUS_LEN(handle)); + /* padding */ + memset((void *)(uintptr_t)(virt_addr + ICP_QAT_CSS_FWSK_MODULUS_LEN(handle)), + 0, ICP_QAT_CSS_FWSK_PAD_LEN(handle)); + + /* exponent */ + memcpy(virt_addr + ICP_QAT_CSS_FWSK_MODULUS_LEN(handle) + + ICP_QAT_CSS_FWSK_PAD_LEN(handle), image + sizeof(*css_hdr) + + ICP_QAT_CSS_FWSK_MODULUS_LEN(handle), sizeof(unsigned int)); + + /* signature */ + bus_addr = ADD_ADDR(auth_desc->fwsk_pub_high, + auth_desc->fwsk_pub_low) + + ICP_QAT_CSS_FWSK_PUB_LEN(handle); + virt_addr = virt_addr + ICP_QAT_CSS_FWSK_PUB_LEN(handle); + auth_desc->signature_high = upper_32_bits(bus_addr); + auth_desc->signature_low = lower_32_bits(bus_addr); + + memcpy(virt_addr, image + sizeof(*css_hdr) + ICP_QAT_CSS_FWSK_MODULUS_LEN(handle) + + ICP_QAT_CSS_FWSK_EXPONENT_LEN(handle), ICP_QAT_CSS_SIGNATURE_LEN(handle)); + + bus_addr = ADD_ADDR(auth_desc->signature_high, + auth_desc->signature_low) + + ICP_QAT_CSS_SIGNATURE_LEN(handle); + virt_addr += ICP_QAT_CSS_SIGNATURE_LEN(handle); + + auth_desc->img_high = upper_32_bits(bus_addr); + auth_desc->img_low = lower_32_bits(bus_addr); + auth_desc->img_len = size - qat_uclo_simg_hdr2sign_len(handle); + if (bus_addr + auth_desc->img_len > + dram_desc->dram_bus_addr + ICP_QAT_CSS_RSA4K_MAX_IMAGE_LEN) { + pr_err("insufficient memory size for authentication data\n"); + qat_uclo_simg_free(handle, dram_desc); + return -ENOMEM; + } + + memcpy(virt_addr, image + qat_uclo_simg_hdr2sign_len(handle), auth_desc->img_len); + virt_addr = virt_base; + /* AE firmware */ + if (fw_type == CSS_AE_FIRMWARE) { + auth_desc->img_ae_mode_data_high = auth_desc->img_high; + auth_desc->img_ae_mode_data_low = auth_desc->img_low; + bus_addr = ADD_ADDR(auth_desc->img_ae_mode_data_high, + auth_desc->img_ae_mode_data_low) + + sizeof(struct icp_qat_simg_ae_mode); + + auth_desc->img_ae_init_data_high = upper_32_bits(bus_addr); + auth_desc->img_ae_init_data_low = lower_32_bits(bus_addr); + bus_addr += ICP_QAT_SIMG_AE_INIT_SEQ_LEN; + auth_desc->img_ae_insts_high = upper_32_bits(bus_addr); + auth_desc->img_ae_insts_low = lower_32_bits(bus_addr); + virt_addr += sizeof(struct icp_qat_css_hdr); + virt_addr += ICP_QAT_CSS_FWSK_PUB_LEN(handle); + virt_addr += ICP_QAT_CSS_SIGNATURE_LEN(handle); + simg_ae_mode = (struct icp_qat_simg_ae_mode *)(uintptr_t)virt_addr; + auth_desc->ae_mask = simg_ae_mode->ae_mask & handle->cfg_ae_mask; + } else { + auth_desc->img_ae_insts_high = auth_desc->img_high; + auth_desc->img_ae_insts_low = auth_desc->img_low; + } + *desc = auth_desc; + return 0; +} + +static int qat_uclo_build_auth_desc_dualsign(struct icp_qat_fw_loader_handle *handle, + char *image, unsigned int size, + struct icp_firml_dram_desc *dram_desc, + unsigned int fw_type, + struct icp_qat_fw_auth_desc **desc) +{ + struct icp_qat_simg_ae_mode *simg_ae_mode; + struct icp_qat_fw_auth_desc *auth_desc; + unsigned int chunk_offset, img_offset; + u64 bus_addr, addr; + char *virt_addr; + + virt_addr = dram_desc->dram_base_addr_v; + virt_addr += sizeof(struct icp_qat_auth_chunk); + bus_addr = dram_desc->dram_bus_addr + sizeof(struct icp_qat_auth_chunk); + + auth_desc = dram_desc->dram_base_addr_v; + auth_desc->img_len = size - qat_uclo_simg_hdr2sign_len(handle); + auth_desc->css_hdr_high = upper_32_bits(bus_addr); + auth_desc->css_hdr_low = lower_32_bits(bus_addr); + memcpy(virt_addr, image, ICP_QAT_DUALSIGN_OPAQUE_HDR_LEN); + + img_offset = ICP_QAT_DUALSIGN_OPAQUE_HDR_LEN; + chunk_offset = ICP_QAT_DUALSIGN_OPAQUE_HDR_ALIGN_LEN; + + /* RSA pub key */ + addr = bus_addr + chunk_offset; + auth_desc->fwsk_pub_high = upper_32_bits(addr); + auth_desc->fwsk_pub_low = lower_32_bits(addr); + memcpy(virt_addr + chunk_offset, image + img_offset, ICP_QAT_CSS_FWSK_MODULUS_LEN(handle)); + + img_offset += ICP_QAT_CSS_FWSK_MODULUS_LEN(handle); + chunk_offset += ICP_QAT_CSS_FWSK_MODULUS_LEN(handle); + /* RSA padding */ + memset(virt_addr + chunk_offset, 0, ICP_QAT_CSS_FWSK_PAD_LEN(handle)); + + chunk_offset += ICP_QAT_CSS_FWSK_PAD_LEN(handle); + /* RSA exponent */ + memcpy(virt_addr + chunk_offset, image + img_offset, ICP_QAT_CSS_FWSK_EXPONENT_LEN(handle)); + + img_offset += ICP_QAT_CSS_FWSK_EXPONENT_LEN(handle); + chunk_offset += ICP_QAT_CSS_FWSK_EXPONENT_LEN(handle); + /* RSA signature */ + addr = bus_addr + chunk_offset; + auth_desc->signature_high = upper_32_bits(addr); + auth_desc->signature_low = lower_32_bits(addr); + memcpy(virt_addr + chunk_offset, image + img_offset, ICP_QAT_CSS_SIGNATURE_LEN(handle)); + + img_offset += ICP_QAT_CSS_SIGNATURE_LEN(handle); + chunk_offset += ICP_QAT_CSS_SIGNATURE_LEN(handle); + /* XMSS pubkey */ + addr = bus_addr + chunk_offset; + auth_desc->xmss_pubkey_high = upper_32_bits(addr); + auth_desc->xmss_pubkey_low = lower_32_bits(addr); + memcpy(virt_addr + chunk_offset, image + img_offset, ICP_QAT_DUALSIGN_XMSS_PUBKEY_LEN); + + img_offset += ICP_QAT_DUALSIGN_XMSS_PUBKEY_LEN; + chunk_offset += ICP_QAT_DUALSIGN_XMSS_PUBKEY_LEN; + /* XMSS signature */ + addr = bus_addr + chunk_offset; + auth_desc->xmss_sig_high = upper_32_bits(addr); + auth_desc->xmss_sig_low = lower_32_bits(addr); + memcpy(virt_addr + chunk_offset, image + img_offset, ICP_QAT_DUALSIGN_XMSS_SIG_LEN); + + img_offset += ICP_QAT_DUALSIGN_XMSS_SIG_LEN; + chunk_offset += ICP_QAT_DUALSIGN_XMSS_SIG_ALIGN_LEN; + + if (dram_desc->dram_size < (chunk_offset + auth_desc->img_len)) { + pr_err("auth chunk memory size is not enough to store data\n"); + return -ENOMEM; + } + + /* Signed data */ + addr = bus_addr + chunk_offset; + auth_desc->img_high = upper_32_bits(addr); + auth_desc->img_low = lower_32_bits(addr); + memcpy(virt_addr + chunk_offset, image + img_offset, auth_desc->img_len); + + chunk_offset += ICP_QAT_DUALSIGN_MISC_INFO_LEN; + /* AE firmware */ + if (fw_type == CSS_AE_FIRMWARE) { + /* AE mode data */ + addr = bus_addr + chunk_offset; + auth_desc->img_ae_mode_data_high = upper_32_bits(addr); + auth_desc->img_ae_mode_data_low = lower_32_bits(addr); + simg_ae_mode = + (struct icp_qat_simg_ae_mode *)(virt_addr + chunk_offset); + auth_desc->ae_mask = simg_ae_mode->ae_mask & handle->cfg_ae_mask; + + chunk_offset += sizeof(struct icp_qat_simg_ae_mode); + /* AE init seq */ + addr = bus_addr + chunk_offset; + auth_desc->img_ae_init_data_high = upper_32_bits(addr); + auth_desc->img_ae_init_data_low = lower_32_bits(addr); + + chunk_offset += ICP_QAT_SIMG_AE_INIT_SEQ_LEN; + /* AE instructions */ + addr = bus_addr + chunk_offset; + auth_desc->img_ae_insts_high = upper_32_bits(addr); + auth_desc->img_ae_insts_low = lower_32_bits(addr); + } else { + addr = bus_addr + chunk_offset; + auth_desc->img_ae_insts_high = upper_32_bits(addr); + auth_desc->img_ae_insts_low = lower_32_bits(addr); + } + *desc = auth_desc; + return 0; +} + +static int qat_uclo_map_auth_fw(struct icp_qat_fw_loader_handle *handle, + char *image, unsigned int size, + struct icp_qat_fw_auth_desc **desc) +{ + struct icp_qat_auth_chunk *auth_chunk; + struct icp_firml_dram_desc img_desc; + unsigned int simg_fw_type; + int ret; + + ret = qat_uclo_simg_alloc(handle, &img_desc, ICP_QAT_CSS_RSA4K_MAX_IMAGE_LEN); + if (ret) + return ret; + + simg_fw_type = qat_uclo_simg_fw_type(handle, image); + auth_chunk = img_desc.dram_base_addr_v; + auth_chunk->chunk_size = img_desc.dram_size; + auth_chunk->chunk_bus_addr = img_desc.dram_bus_addr; + + if (handle->chip_info->dual_sign) + return qat_uclo_build_auth_desc_dualsign(handle, image, size, &img_desc, + simg_fw_type, desc); + + return qat_uclo_build_auth_desc_RSA(handle, image, size, &img_desc, + simg_fw_type, desc); +} + +static int qat_uclo_load_fw(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_fw_auth_desc *desc) +{ + unsigned long ae_mask = handle->hal_handle->ae_mask; + u32 fcu_sts_csr, fcu_ctl_csr; + u32 loaded_aes, loaded_csr; + unsigned int i; + u32 fcu_sts; + + fcu_ctl_csr = handle->chip_info->fcu_ctl_csr; + fcu_sts_csr = handle->chip_info->fcu_sts_csr; + loaded_csr = handle->chip_info->fcu_loaded_ae_csr; + + for_each_set_bit(i, &ae_mask, handle->hal_handle->ae_max_num) { + int retry = 0; + + if (!((desc->ae_mask >> i) & 0x1)) + continue; + if (qat_hal_check_ae_active(handle, i)) { + pr_err("AE %d is active\n", i); + return -EINVAL; + } + SET_CAP_CSR(handle, fcu_ctl_csr, + (FCU_CTRL_CMD_LOAD | + (1 << FCU_CTRL_BROADCAST_POS) | + (i << FCU_CTRL_AE_POS))); + + do { + msleep(FW_AUTH_WAIT_PERIOD); + fcu_sts = GET_CAP_CSR(handle, fcu_sts_csr); + if ((fcu_sts & FCU_AUTH_STS_MASK) == + FCU_STS_LOAD_DONE) { + loaded_aes = GET_CAP_CSR(handle, loaded_csr); + loaded_aes >>= handle->chip_info->fcu_loaded_ae_pos; + if (loaded_aes & (1 << i)) + break; + } + } while (retry++ < FW_AUTH_MAX_RETRY); + if (retry > FW_AUTH_MAX_RETRY) { + pr_err("firmware load failed timeout %x\n", retry); + return -EINVAL; + } + } + return 0; +} + +static int qat_uclo_map_suof_obj(struct icp_qat_fw_loader_handle *handle, + void *addr_ptr, int mem_size) +{ + struct icp_qat_suof_handle *suof_handle; + + suof_handle = kzalloc(sizeof(*suof_handle), GFP_KERNEL); + if (!suof_handle) + return -ENOMEM; + handle->sobj_handle = suof_handle; + if (qat_uclo_map_suof(handle, addr_ptr, mem_size)) { + qat_uclo_del_suof(handle); + pr_err("map SUOF failed\n"); + return -EINVAL; + } + return 0; +} + +int qat_uclo_wr_mimage(struct icp_qat_fw_loader_handle *handle, + void *addr_ptr, int mem_size) +{ + struct icp_qat_fw_auth_desc *desc = NULL; + int status = 0; + int ret; + + ret = qat_uclo_check_image(handle, addr_ptr, mem_size, CSS_MMP_FIRMWARE); + if (ret) + return ret; + + if (handle->chip_info->fw_auth) { + status = qat_uclo_map_auth_fw(handle, addr_ptr, mem_size, &desc); + if (!status) + status = qat_uclo_auth_fw(handle, desc); + qat_uclo_ummap_auth_fw(handle, &desc); + } else { + if (handle->chip_info->mmp_sram_size < mem_size) { + pr_err("MMP size is too large: 0x%x\n", mem_size); + return -EFBIG; + } + qat_uclo_wr_sram_by_words(handle, 0, addr_ptr, mem_size); + } + return status; +} + +static int qat_uclo_map_uof_obj(struct icp_qat_fw_loader_handle *handle, + void *addr_ptr, int mem_size) +{ + struct icp_qat_uof_filehdr *filehdr; + struct icp_qat_uclo_objhandle *objhdl; + + objhdl = kzalloc(sizeof(*objhdl), GFP_KERNEL); + if (!objhdl) + return -ENOMEM; + objhdl->obj_buf = kmemdup(addr_ptr, mem_size, GFP_KERNEL); + if (!objhdl->obj_buf) + goto out_objbuf_err; + filehdr = (struct icp_qat_uof_filehdr *)objhdl->obj_buf; + if (qat_uclo_check_uof_format(filehdr)) + goto out_objhdr_err; + objhdl->obj_hdr = qat_uclo_map_chunk((char *)objhdl->obj_buf, filehdr, + ICP_QAT_UOF_OBJS); + if (!objhdl->obj_hdr) { + pr_err("object file chunk is null\n"); + goto out_objhdr_err; + } + handle->obj_handle = objhdl; + if (qat_uclo_parse_uof_obj(handle)) + goto out_overlay_obj_err; + return 0; + +out_overlay_obj_err: + handle->obj_handle = NULL; + kfree(objhdl->obj_hdr); +out_objhdr_err: + kfree(objhdl->obj_buf); +out_objbuf_err: + kfree(objhdl); + return -ENOMEM; +} + +static int qat_uclo_map_mof_file_hdr(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_mof_file_hdr *mof_ptr, + u32 mof_size) +{ + struct icp_qat_mof_handle *mobj_handle = handle->mobj_handle; + unsigned int min_ver_offset; + unsigned int checksum; + + mobj_handle->file_id = ICP_QAT_MOF_FID; + mobj_handle->mof_buf = (char *)mof_ptr; + mobj_handle->mof_size = mof_size; + + min_ver_offset = mof_size - offsetof(struct icp_qat_mof_file_hdr, + min_ver); + checksum = qat_uclo_calc_str_checksum(&mof_ptr->min_ver, + min_ver_offset); + if (checksum != mof_ptr->checksum) { + pr_err("incorrect MOF checksum\n"); + return -EINVAL; + } + + mobj_handle->checksum = mof_ptr->checksum; + mobj_handle->min_ver = mof_ptr->min_ver; + mobj_handle->maj_ver = mof_ptr->maj_ver; + return 0; +} + +static void qat_uclo_del_mof(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_mof_handle *mobj_handle = handle->mobj_handle; + + kfree(mobj_handle->obj_table.obj_hdr); + mobj_handle->obj_table.obj_hdr = NULL; + kfree(handle->mobj_handle); + handle->mobj_handle = NULL; +} + +static int qat_uclo_seek_obj_inside_mof(struct icp_qat_mof_handle *mobj_handle, + const char *obj_name, char **obj_ptr, + unsigned int *obj_size) +{ + struct icp_qat_mof_objhdr *obj_hdr = mobj_handle->obj_table.obj_hdr; + unsigned int i; + + for (i = 0; i < mobj_handle->obj_table.num_objs; i++) { + if (!strncmp(obj_hdr[i].obj_name, obj_name, + ICP_QAT_SUOF_OBJ_NAME_LEN)) { + *obj_ptr = obj_hdr[i].obj_buf; + *obj_size = obj_hdr[i].obj_size; + return 0; + } + } + + pr_err("object %s is not found inside MOF\n", obj_name); + return -EINVAL; +} + +static int qat_uclo_map_obj_from_mof(struct icp_qat_mof_handle *mobj_handle, + struct icp_qat_mof_objhdr *mobj_hdr, + struct icp_qat_mof_obj_chunkhdr *obj_chunkhdr) +{ + u8 *obj; + + if (!strncmp(obj_chunkhdr->chunk_id, ICP_QAT_UOF_IMAG, + ICP_QAT_MOF_OBJ_CHUNKID_LEN)) { + obj = mobj_handle->uobjs_hdr + obj_chunkhdr->offset; + } else if (!strncmp(obj_chunkhdr->chunk_id, ICP_QAT_SUOF_IMAG, + ICP_QAT_MOF_OBJ_CHUNKID_LEN)) { + obj = mobj_handle->sobjs_hdr + obj_chunkhdr->offset; + } else { + pr_err("unsupported chunk id\n"); + return -EINVAL; + } + mobj_hdr->obj_buf = obj; + mobj_hdr->obj_size = (unsigned int)obj_chunkhdr->size; + mobj_hdr->obj_name = obj_chunkhdr->name + mobj_handle->sym_str; + return 0; +} + +static int qat_uclo_map_objs_from_mof(struct icp_qat_mof_handle *mobj_handle) +{ + struct icp_qat_mof_obj_chunkhdr *uobj_chunkhdr; + struct icp_qat_mof_obj_chunkhdr *sobj_chunkhdr; + struct icp_qat_mof_obj_hdr *uobj_hdr; + struct icp_qat_mof_obj_hdr *sobj_hdr; + struct icp_qat_mof_objhdr *mobj_hdr; + unsigned int uobj_chunk_num = 0; + unsigned int sobj_chunk_num = 0; + unsigned int *valid_chunk; + int ret, i; + + uobj_hdr = (struct icp_qat_mof_obj_hdr *)mobj_handle->uobjs_hdr; + sobj_hdr = (struct icp_qat_mof_obj_hdr *)mobj_handle->sobjs_hdr; + if (uobj_hdr) + uobj_chunk_num = uobj_hdr->num_chunks; + if (sobj_hdr) + sobj_chunk_num = sobj_hdr->num_chunks; + + mobj_hdr = kcalloc(size_add(uobj_chunk_num, sobj_chunk_num), + sizeof(*mobj_hdr), GFP_KERNEL); + if (!mobj_hdr) + return -ENOMEM; + + mobj_handle->obj_table.obj_hdr = mobj_hdr; + valid_chunk = &mobj_handle->obj_table.num_objs; + uobj_chunkhdr = (struct icp_qat_mof_obj_chunkhdr *) + ((uintptr_t)uobj_hdr + sizeof(*uobj_hdr)); + sobj_chunkhdr = (struct icp_qat_mof_obj_chunkhdr *) + ((uintptr_t)sobj_hdr + sizeof(*sobj_hdr)); + + /* map uof objects */ + for (i = 0; i < uobj_chunk_num; i++) { + ret = qat_uclo_map_obj_from_mof(mobj_handle, + &mobj_hdr[*valid_chunk], + &uobj_chunkhdr[i]); + if (ret) + return ret; + (*valid_chunk)++; + } + + /* map suof objects */ + for (i = 0; i < sobj_chunk_num; i++) { + ret = qat_uclo_map_obj_from_mof(mobj_handle, + &mobj_hdr[*valid_chunk], + &sobj_chunkhdr[i]); + if (ret) + return ret; + (*valid_chunk)++; + } + + if ((uobj_chunk_num + sobj_chunk_num) != *valid_chunk) { + pr_err("inconsistent UOF/SUOF chunk amount\n"); + return -EINVAL; + } + return 0; +} + +static void qat_uclo_map_mof_symobjs(struct icp_qat_mof_handle *mobj_handle, + struct icp_qat_mof_chunkhdr *mof_chunkhdr) +{ + char **sym_str = (char **)&mobj_handle->sym_str; + unsigned int *sym_size = &mobj_handle->sym_size; + struct icp_qat_mof_str_table *str_table_obj; + + *sym_size = *(unsigned int *)(uintptr_t) + (mof_chunkhdr->offset + mobj_handle->mof_buf); + *sym_str = (char *)(uintptr_t) + (mobj_handle->mof_buf + mof_chunkhdr->offset + + sizeof(str_table_obj->tab_len)); +} + +static void qat_uclo_map_mof_chunk(struct icp_qat_mof_handle *mobj_handle, + struct icp_qat_mof_chunkhdr *mof_chunkhdr) +{ + char *chunk_id = mof_chunkhdr->chunk_id; + + if (!strncmp(chunk_id, ICP_QAT_MOF_SYM_OBJS, ICP_QAT_MOF_OBJ_ID_LEN)) + qat_uclo_map_mof_symobjs(mobj_handle, mof_chunkhdr); + else if (!strncmp(chunk_id, ICP_QAT_UOF_OBJS, ICP_QAT_MOF_OBJ_ID_LEN)) + mobj_handle->uobjs_hdr = mobj_handle->mof_buf + + mof_chunkhdr->offset; + else if (!strncmp(chunk_id, ICP_QAT_SUOF_OBJS, ICP_QAT_MOF_OBJ_ID_LEN)) + mobj_handle->sobjs_hdr = mobj_handle->mof_buf + + mof_chunkhdr->offset; +} + +static int qat_uclo_check_mof_format(struct icp_qat_mof_file_hdr *mof_hdr) +{ + int maj = mof_hdr->maj_ver & 0xff; + int min = mof_hdr->min_ver & 0xff; + + if (mof_hdr->file_id != ICP_QAT_MOF_FID) { + pr_err("invalid header 0x%x\n", mof_hdr->file_id); + return -EINVAL; + } + + if (mof_hdr->num_chunks <= 0x1) { + pr_err("MOF chunk amount is incorrect\n"); + return -EINVAL; + } + if (maj != ICP_QAT_MOF_MAJVER || min != ICP_QAT_MOF_MINVER) { + pr_err("bad MOF version, major 0x%x, minor 0x%x\n", maj, min); + return -EINVAL; + } + return 0; +} + +static int qat_uclo_map_mof_obj(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_mof_file_hdr *mof_ptr, + u32 mof_size, const char *obj_name, + char **obj_ptr, unsigned int *obj_size) +{ + struct icp_qat_mof_chunkhdr *mof_chunkhdr; + unsigned int file_id = mof_ptr->file_id; + struct icp_qat_mof_handle *mobj_handle; + unsigned short chunks_num; + unsigned int i; + int ret; + + if (file_id == ICP_QAT_UOF_FID || file_id == ICP_QAT_SUOF_FID) { + if (obj_ptr) + *obj_ptr = (char *)mof_ptr; + if (obj_size) + *obj_size = mof_size; + return 0; + } + if (qat_uclo_check_mof_format(mof_ptr)) + return -EINVAL; + + mobj_handle = kzalloc(sizeof(*mobj_handle), GFP_KERNEL); + if (!mobj_handle) + return -ENOMEM; + + handle->mobj_handle = mobj_handle; + ret = qat_uclo_map_mof_file_hdr(handle, mof_ptr, mof_size); + if (ret) + return ret; + + mof_chunkhdr = (void *)mof_ptr + sizeof(*mof_ptr); + chunks_num = mof_ptr->num_chunks; + + /* Parse MOF file chunks */ + for (i = 0; i < chunks_num; i++) + qat_uclo_map_mof_chunk(mobj_handle, &mof_chunkhdr[i]); + + /* All sym_objs uobjs and sobjs should be available */ + if (!mobj_handle->sym_str || + (!mobj_handle->uobjs_hdr && !mobj_handle->sobjs_hdr)) + return -EINVAL; + + ret = qat_uclo_map_objs_from_mof(mobj_handle); + if (ret) + return ret; + + /* Seek specified uof object in MOF */ + return qat_uclo_seek_obj_inside_mof(mobj_handle, obj_name, + obj_ptr, obj_size); +} + +int qat_uclo_map_obj(struct icp_qat_fw_loader_handle *handle, + void *addr_ptr, u32 mem_size, const char *obj_name) +{ + char *obj_addr; + u32 obj_size; + int ret; + + BUILD_BUG_ON(ICP_QAT_UCLO_MAX_AE >= + (sizeof(handle->hal_handle->ae_mask) * 8)); + + if (!handle || !addr_ptr || mem_size < 24) + return -EINVAL; + + if (obj_name) { + ret = qat_uclo_map_mof_obj(handle, addr_ptr, mem_size, obj_name, + &obj_addr, &obj_size); + if (ret) + return ret; + } else { + obj_addr = addr_ptr; + obj_size = mem_size; + } + + return (handle->chip_info->fw_auth) ? + qat_uclo_map_suof_obj(handle, obj_addr, obj_size) : + qat_uclo_map_uof_obj(handle, obj_addr, obj_size); +} + +void qat_uclo_del_obj(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned int a; + + if (handle->mobj_handle) + qat_uclo_del_mof(handle); + if (handle->sobj_handle) + qat_uclo_del_suof(handle); + if (!obj_handle) + return; + + kfree(obj_handle->uword_buf); + for (a = 0; a < obj_handle->uimage_num; a++) + kfree(obj_handle->ae_uimage[a].page); + + for (a = 0; a < handle->hal_handle->ae_max_num; a++) + qat_uclo_free_ae_data(&obj_handle->ae_data[a]); + + kfree(obj_handle->obj_hdr); + kfree(obj_handle->obj_buf); + kfree(obj_handle); + handle->obj_handle = NULL; +} + +static void qat_uclo_fill_uwords(struct icp_qat_uclo_objhandle *obj_handle, + struct icp_qat_uclo_encap_page *encap_page, + u64 *uword, unsigned int addr_p, + unsigned int raddr, u64 fill) +{ + unsigned int i, addr; + u64 uwrd = 0; + + if (!encap_page) { + *uword = fill; + return; + } + addr = (encap_page->page_region) ? raddr : addr_p; + for (i = 0; i < encap_page->uwblock_num; i++) { + if (addr >= encap_page->uwblock[i].start_addr && + addr <= encap_page->uwblock[i].start_addr + + encap_page->uwblock[i].words_num - 1) { + addr -= encap_page->uwblock[i].start_addr; + addr *= obj_handle->uword_in_bytes; + memcpy(&uwrd, (void *)(((uintptr_t) + encap_page->uwblock[i].micro_words) + addr), + obj_handle->uword_in_bytes); + uwrd = uwrd & GENMASK_ULL(43, 0); + } + } + *uword = uwrd; + if (*uword == INVLD_UWORD) + *uword = fill; +} + +static void qat_uclo_wr_uimage_raw_page(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uclo_encap_page + *encap_page, unsigned int ae) +{ + unsigned int uw_physical_addr, uw_relative_addr, i, words_num, cpylen; + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + u64 fill_pat; + + /* load the page starting at appropriate ustore address */ + /* get fill-pattern from an image -- they are all the same */ + memcpy(&fill_pat, obj_handle->ae_uimage[0].img_ptr->fill_pattern, + sizeof(u64)); + uw_physical_addr = encap_page->beg_addr_p; + uw_relative_addr = 0; + words_num = encap_page->micro_words_num; + while (words_num) { + cpylen = min(words_num, UWORD_CPYBUF_SIZE); + + /* load the buffer */ + for (i = 0; i < cpylen; i++) + qat_uclo_fill_uwords(obj_handle, encap_page, + &obj_handle->uword_buf[i], + uw_physical_addr + i, + uw_relative_addr + i, fill_pat); + + /* copy the buffer to ustore */ + qat_hal_wr_uwords(handle, (unsigned char)ae, + uw_physical_addr, cpylen, + obj_handle->uword_buf); + + uw_physical_addr += cpylen; + uw_relative_addr += cpylen; + words_num -= cpylen; + } +} + +static void qat_uclo_wr_uimage_page(struct icp_qat_fw_loader_handle *handle, + struct icp_qat_uof_image *image) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned long ae_mask = handle->hal_handle->ae_mask; + unsigned long cfg_ae_mask = handle->cfg_ae_mask; + unsigned long ae_assigned = image->ae_assigned; + struct icp_qat_uclo_aedata *aed; + unsigned int ctx_mask, s; + struct icp_qat_uclo_page *page; + unsigned char ae; + int ctx; + + if (ICP_QAT_CTX_MODE(image->ae_mode) == ICP_QAT_UCLO_MAX_CTX) + ctx_mask = 0xff; + else + ctx_mask = 0x55; + /* load the default page and set assigned CTX PC + * to the entrypoint address */ + for_each_set_bit(ae, &ae_mask, handle->hal_handle->ae_max_num) { + if (!test_bit(ae, &cfg_ae_mask)) + continue; + + if (!test_bit(ae, &ae_assigned)) + continue; + + aed = &obj_handle->ae_data[ae]; + /* find the slice to which this image is assigned */ + for (s = 0; s < aed->slice_num; s++) { + if (image->ctx_assigned & + aed->ae_slices[s].ctx_mask_assigned) + break; + } + if (s >= aed->slice_num) + continue; + page = aed->ae_slices[s].page; + if (!page->encap_page->def_page) + continue; + qat_uclo_wr_uimage_raw_page(handle, page->encap_page, ae); + + page = aed->ae_slices[s].page; + for (ctx = 0; ctx < ICP_QAT_UCLO_MAX_CTX; ctx++) + aed->ae_slices[s].cur_page[ctx] = + (ctx_mask & (1 << ctx)) ? page : NULL; + qat_hal_set_live_ctx(handle, (unsigned char)ae, + image->ctx_assigned); + qat_hal_set_pc(handle, (unsigned char)ae, image->ctx_assigned, + image->entry_address); + } +} + +static int qat_uclo_wr_suof_img(struct icp_qat_fw_loader_handle *handle) +{ + unsigned int i; + struct icp_qat_fw_auth_desc *desc = NULL; + struct icp_qat_suof_handle *sobj_handle = handle->sobj_handle; + struct icp_qat_suof_img_hdr *simg_hdr = sobj_handle->img_table.simg_hdr; + int ret; + + for (i = 0; i < sobj_handle->img_table.num_simgs; i++) { + ret = qat_uclo_check_image(handle, simg_hdr[i].simg_buf, + simg_hdr[i].simg_len, + CSS_AE_FIRMWARE); + if (ret) + return ret; + + if (qat_uclo_map_auth_fw(handle, + (char *)simg_hdr[i].simg_buf, + (unsigned int) + simg_hdr[i].simg_len, + &desc)) + goto wr_err; + if (qat_uclo_auth_fw(handle, desc)) + goto wr_err; + if (qat_uclo_is_broadcast(handle, i)) { + if (qat_uclo_broadcast_load_fw(handle, desc)) + goto wr_err; + } else { + if (qat_uclo_load_fw(handle, desc)) + goto wr_err; + } + qat_uclo_ummap_auth_fw(handle, &desc); + } + return 0; +wr_err: + qat_uclo_ummap_auth_fw(handle, &desc); + return -EINVAL; +} + +static int qat_uclo_wr_uof_img(struct icp_qat_fw_loader_handle *handle) +{ + struct icp_qat_uclo_objhandle *obj_handle = handle->obj_handle; + unsigned int i; + + if (qat_uclo_init_globals(handle)) + return -EINVAL; + for (i = 0; i < obj_handle->uimage_num; i++) { + if (!obj_handle->ae_uimage[i].img_ptr) + return -EINVAL; + if (qat_uclo_init_ustore(handle, &obj_handle->ae_uimage[i])) + return -EINVAL; + qat_uclo_wr_uimage_page(handle, + obj_handle->ae_uimage[i].img_ptr); + } + return 0; +} + +int qat_uclo_wr_all_uimage(struct icp_qat_fw_loader_handle *handle) +{ + return (handle->chip_info->fw_auth) ? qat_uclo_wr_suof_img(handle) : + qat_uclo_wr_uof_img(handle); +} + +int qat_uclo_set_cfg_ae_mask(struct icp_qat_fw_loader_handle *handle, + unsigned int cfg_ae_mask) +{ + if (!cfg_ae_mask) + return -EINVAL; + + handle->cfg_ae_mask = cfg_ae_mask; + return 0; +} diff --git a/drivers/crypto/intel/qat/qat_dh895xcc/Makefile b/drivers/crypto/intel/qat/qat_dh895xcc/Makefile new file mode 100644 index 000000000000..1427fe76f171 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xcc/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_DH895xCC) += qat_dh895xcc.o +qat_dh895xcc-y := adf_drv.o adf_dh895xcc_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_dh895xcc/adf_dh895xcc_hw_data.c b/drivers/crypto/intel/qat/qat_dh895xcc/adf_dh895xcc_hw_data.c new file mode 100644 index 000000000000..5b4bd0ba1ccb --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xcc/adf_dh895xcc_hw_data.c @@ -0,0 +1,269 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2021 Intel Corporation */ +#include <adf_accel_devices.h> +#include <adf_admin.h> +#include <adf_common_drv.h> +#include <adf_gen2_config.h> +#include <adf_gen2_hw_csr_data.h> +#include <adf_gen2_hw_data.h> +#include <adf_gen2_pfvf.h> +#include "adf_dh895xcc_hw_data.h" +#include "adf_heartbeat.h" +#include "icp_qat_hw.h" + +#define ADF_DH895XCC_VF_MSK 0xFFFFFFFF + +/* Worker thread to service arbiter mappings */ +static const u32 thrd_to_arb_map[ADF_DH895XCC_MAX_ACCELENGINES] = { + 0x12222AAA, 0x11666666, 0x12222AAA, 0x11666666, + 0x12222AAA, 0x11222222, 0x12222AAA, 0x11222222, + 0x12222AAA, 0x11222222, 0x12222AAA, 0x11222222 +}; + +static struct adf_hw_device_class dh895xcc_class = { + .name = ADF_DH895XCC_DEVICE_NAME, + .type = DEV_DH895XCC, +}; + +static u32 get_accel_mask(struct adf_hw_device_data *self) +{ + u32 fuses = self->fuses[ADF_FUSECTL0]; + + return ~fuses >> ADF_DH895XCC_ACCELERATORS_REG_OFFSET & + ADF_DH895XCC_ACCELERATORS_MASK; +} + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + u32 fuses = self->fuses[ADF_FUSECTL0]; + + return ~fuses & ADF_DH895XCC_ACCELENGINES_MASK; +} + +static u32 get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_DH895XCC_PMISC_BAR; +} + +static u32 get_ts_clock(struct adf_hw_device_data *self) +{ + /* + * Timestamp update interval is 16 AE clock ticks for dh895xcc. + */ + return self->clock_frequency / 16; +} + +static u32 get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_DH895XCC_ETR_BAR; +} + +static u32 get_sram_bar_id(struct adf_hw_device_data *self) +{ + return ADF_DH895XCC_SRAM_BAR; +} + +static u32 get_accel_cap(struct adf_accel_dev *accel_dev) +{ + struct pci_dev *pdev = accel_dev->accel_pci_dev.pci_dev; + u32 capabilities; + u32 legfuses; + + capabilities = ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC | + ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC | + ICP_ACCEL_CAPABILITIES_AUTHENTICATION | + ICP_ACCEL_CAPABILITIES_CIPHER | + ICP_ACCEL_CAPABILITIES_COMPRESSION; + + /* Read accelerator capabilities mask */ + pci_read_config_dword(pdev, ADF_DEVICE_LEGFUSE_OFFSET, &legfuses); + + /* A set bit in legfuses means the feature is OFF in this SKU */ + if (legfuses & ICP_ACCEL_MASK_CIPHER_SLICE) { + capabilities &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_SYMMETRIC; + capabilities &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + if (legfuses & ICP_ACCEL_MASK_PKE_SLICE) + capabilities &= ~ICP_ACCEL_CAPABILITIES_CRYPTO_ASYMMETRIC; + if (legfuses & ICP_ACCEL_MASK_AUTH_SLICE) { + capabilities &= ~ICP_ACCEL_CAPABILITIES_AUTHENTICATION; + capabilities &= ~ICP_ACCEL_CAPABILITIES_CIPHER; + } + if (legfuses & ICP_ACCEL_MASK_COMPRESS_SLICE) + capabilities &= ~ICP_ACCEL_CAPABILITIES_COMPRESSION; + + return capabilities; +} + +static enum dev_sku_info get_sku(struct adf_hw_device_data *self) +{ + int sku = (self->fuses[ADF_FUSECTL0] & ADF_DH895XCC_FUSECTL_SKU_MASK) + >> ADF_DH895XCC_FUSECTL_SKU_SHIFT; + + switch (sku) { + case ADF_DH895XCC_FUSECTL_SKU_1: + return DEV_SKU_1; + case ADF_DH895XCC_FUSECTL_SKU_2: + return DEV_SKU_2; + case ADF_DH895XCC_FUSECTL_SKU_3: + return DEV_SKU_3; + case ADF_DH895XCC_FUSECTL_SKU_4: + return DEV_SKU_4; + default: + return DEV_SKU_UNKNOWN; + } + return DEV_SKU_UNKNOWN; +} + +static const u32 *adf_get_arbiter_mapping(struct adf_accel_dev *accel_dev) +{ + return thrd_to_arb_map; +} + +static void enable_vf2pf_interrupts(void __iomem *pmisc_addr, u32 vf_mask) +{ + /* Enable VF2PF Messaging Ints - VFs 0 through 15 per vf_mask[15:0] */ + if (vf_mask & 0xFFFF) { + u32 val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK3) + & ~ADF_DH895XCC_ERR_MSK_VF2PF_L(vf_mask); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, val); + } + + /* Enable VF2PF Messaging Ints - VFs 16 through 31 per vf_mask[31:16] */ + if (vf_mask >> 16) { + u32 val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK5) + & ~ADF_DH895XCC_ERR_MSK_VF2PF_U(vf_mask); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK5, val); + } +} + +static void disable_all_vf2pf_interrupts(void __iomem *pmisc_addr) +{ + u32 val; + + /* Disable VF2PF interrupts for VFs 0 through 15 per vf_mask[15:0] */ + val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK3) + | ADF_DH895XCC_ERR_MSK_VF2PF_L(ADF_DH895XCC_VF_MSK); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, val); + + /* Disable VF2PF interrupts for VFs 16 through 31 per vf_mask[31:16] */ + val = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK5) + | ADF_DH895XCC_ERR_MSK_VF2PF_U(ADF_DH895XCC_VF_MSK); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK5, val); +} + +static u32 disable_pending_vf2pf_interrupts(void __iomem *pmisc_addr) +{ + u32 sources, pending, disabled; + u32 errsou3, errmsk3; + u32 errsou5, errmsk5; + + /* Get the interrupt sources triggered by VFs */ + errsou3 = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRSOU3); + errsou5 = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRSOU5); + sources = ADF_DH895XCC_ERR_REG_VF2PF_L(errsou3) + | ADF_DH895XCC_ERR_REG_VF2PF_U(errsou5); + + if (!sources) + return 0; + + /* Get the already disabled interrupts */ + errmsk3 = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK3); + errmsk5 = ADF_CSR_RD(pmisc_addr, ADF_GEN2_ERRMSK5); + disabled = ADF_DH895XCC_ERR_REG_VF2PF_L(errmsk3) + | ADF_DH895XCC_ERR_REG_VF2PF_U(errmsk5); + + pending = sources & ~disabled; + if (!pending) + return 0; + + /* Due to HW limitations, when disabling the interrupts, we can't + * just disable the requested sources, as this would lead to missed + * interrupts if sources changes just before writing to ERRMSK3 and + * ERRMSK5. + * To work around it, disable all and re-enable only the sources that + * are not in vf_mask and were not already disabled. Re-enabling will + * trigger a new interrupt for the sources that have changed in the + * meantime, if any. + */ + errmsk3 |= ADF_DH895XCC_ERR_MSK_VF2PF_L(ADF_DH895XCC_VF_MSK); + errmsk5 |= ADF_DH895XCC_ERR_MSK_VF2PF_U(ADF_DH895XCC_VF_MSK); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, errmsk3); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK5, errmsk5); + + /* Update only section of errmsk3 and errmsk5 related to VF2PF */ + errmsk3 &= ~ADF_DH895XCC_ERR_MSK_VF2PF_L(ADF_DH895XCC_VF_MSK); + errmsk5 &= ~ADF_DH895XCC_ERR_MSK_VF2PF_U(ADF_DH895XCC_VF_MSK); + + errmsk3 |= ADF_DH895XCC_ERR_MSK_VF2PF_L(sources | disabled); + errmsk5 |= ADF_DH895XCC_ERR_MSK_VF2PF_U(sources | disabled); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK3, errmsk3); + ADF_CSR_WR(pmisc_addr, ADF_GEN2_ERRMSK5, errmsk5); + + /* Return the sources of the (new) interrupt(s) */ + return pending; +} + +static void configure_iov_threads(struct adf_accel_dev *accel_dev, bool enable) +{ + adf_gen2_cfg_iov_thds(accel_dev, enable, + ADF_DH895XCC_AE2FUNC_MAP_GRP_A_NUM_REGS, + ADF_DH895XCC_AE2FUNC_MAP_GRP_B_NUM_REGS); +} + +void adf_init_hw_data_dh895xcc(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class = &dh895xcc_class; + hw_data->instance_id = dh895xcc_class.instances++; + hw_data->num_banks = ADF_DH895XCC_ETR_MAX_BANKS; + hw_data->num_rings_per_bank = ADF_ETR_MAX_RINGS_PER_BANK; + hw_data->num_accel = ADF_DH895XCC_MAX_ACCELERATORS; + hw_data->num_logical_accel = 1; + hw_data->num_engines = ADF_DH895XCC_MAX_ACCELENGINES; + hw_data->tx_rx_gap = ADF_GEN2_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_GEN2_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN2_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_isr_resource_alloc; + hw_data->free_irq = adf_isr_resource_free; + hw_data->enable_error_correction = adf_gen2_enable_error_correction; + hw_data->get_accel_mask = get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_accel_cap = get_accel_cap; + hw_data->get_num_accels = adf_gen2_get_num_accels; + hw_data->get_num_aes = adf_gen2_get_num_aes; + hw_data->get_etr_bar_id = get_etr_bar_id; + hw_data->get_misc_bar_id = get_misc_bar_id; + hw_data->get_admin_info = adf_gen2_get_admin_info; + hw_data->get_arb_info = adf_gen2_get_arb_info; + hw_data->get_sram_bar_id = get_sram_bar_id; + hw_data->get_sku = get_sku; + hw_data->fw_name = ADF_DH895XCC_FW; + hw_data->fw_mmp_name = ADF_DH895XCC_MMP; + hw_data->init_admin_comms = adf_init_admin_comms; + hw_data->exit_admin_comms = adf_exit_admin_comms; + hw_data->configure_iov_threads = configure_iov_threads; + hw_data->send_admin_init = adf_send_admin_init; + hw_data->init_arb = adf_init_arb; + hw_data->exit_arb = adf_exit_arb; + hw_data->get_arb_mapping = adf_get_arbiter_mapping; + hw_data->enable_ints = adf_gen2_enable_ints; + hw_data->reset_device = adf_reset_sbr; + hw_data->disable_iov = adf_disable_sriov; + hw_data->dev_config = adf_gen2_dev_config; + hw_data->clock_frequency = ADF_DH895X_AE_FREQ; + hw_data->get_hb_clock = get_ts_clock; + hw_data->num_hb_ctrs = ADF_NUM_HB_CNT_PER_AE; + hw_data->check_hb_ctrs = adf_heartbeat_check_ctrs; + + adf_gen2_init_pf_pfvf_ops(&hw_data->pfvf_ops); + hw_data->pfvf_ops.enable_vf2pf_interrupts = enable_vf2pf_interrupts; + hw_data->pfvf_ops.disable_all_vf2pf_interrupts = disable_all_vf2pf_interrupts; + hw_data->pfvf_ops.disable_pending_vf2pf_interrupts = disable_pending_vf2pf_interrupts; + adf_gen2_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen2_init_dc_ops(&hw_data->dc_ops); +} + +void adf_clean_hw_data_dh895xcc(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; +} diff --git a/drivers/crypto/intel/qat/qat_dh895xcc/adf_dh895xcc_hw_data.h b/drivers/crypto/intel/qat/qat_dh895xcc/adf_dh895xcc_hw_data.h new file mode 100644 index 000000000000..cd3a21985455 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xcc/adf_dh895xcc_hw_data.h @@ -0,0 +1,44 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#ifndef ADF_DH895x_HW_DATA_H_ +#define ADF_DH895x_HW_DATA_H_ + +#include <linux/units.h> + +/* PCIe configuration space */ +#define ADF_DH895XCC_SRAM_BAR 0 +#define ADF_DH895XCC_PMISC_BAR 1 +#define ADF_DH895XCC_ETR_BAR 2 +#define ADF_DH895XCC_FUSECTL_SKU_MASK 0x300000 +#define ADF_DH895XCC_FUSECTL_SKU_SHIFT 20 +#define ADF_DH895XCC_FUSECTL_SKU_1 0x0 +#define ADF_DH895XCC_FUSECTL_SKU_2 0x1 +#define ADF_DH895XCC_FUSECTL_SKU_3 0x2 +#define ADF_DH895XCC_FUSECTL_SKU_4 0x3 +#define ADF_DH895XCC_MAX_ACCELERATORS 6 +#define ADF_DH895XCC_MAX_ACCELENGINES 12 +#define ADF_DH895XCC_ACCELERATORS_REG_OFFSET 13 +#define ADF_DH895XCC_ACCELERATORS_MASK 0x3F +#define ADF_DH895XCC_ACCELENGINES_MASK 0xFFF +#define ADF_DH895XCC_ETR_MAX_BANKS 32 + +/* Masks for VF2PF interrupts */ +#define ADF_DH895XCC_ERR_REG_VF2PF_L(vf_src) (((vf_src) & 0x01FFFE00) >> 9) +#define ADF_DH895XCC_ERR_MSK_VF2PF_L(vf_mask) (((vf_mask) & 0xFFFF) << 9) +#define ADF_DH895XCC_ERR_REG_VF2PF_U(vf_src) (((vf_src) & 0x0000FFFF) << 16) +#define ADF_DH895XCC_ERR_MSK_VF2PF_U(vf_mask) ((vf_mask) >> 16) + +/* AE to function mapping */ +#define ADF_DH895XCC_AE2FUNC_MAP_GRP_A_NUM_REGS 96 +#define ADF_DH895XCC_AE2FUNC_MAP_GRP_B_NUM_REGS 12 + +/* Clocks frequency */ +#define ADF_DH895X_AE_FREQ (933 * HZ_PER_MHZ) + +/* FW names */ +#define ADF_DH895XCC_FW "qat_895xcc.bin" +#define ADF_DH895XCC_MMP "qat_895xcc_mmp.bin" + +void adf_init_hw_data_dh895xcc(struct adf_hw_device_data *hw_data); +void adf_clean_hw_data_dh895xcc(struct adf_hw_device_data *hw_data); +#endif diff --git a/drivers/crypto/intel/qat/qat_dh895xcc/adf_drv.c b/drivers/crypto/intel/qat/qat_dh895xcc/adf_drv.c new file mode 100644 index 000000000000..b59e0cc49e52 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xcc/adf_drv.c @@ -0,0 +1,260 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/fs.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/platform_device.h> +#include <linux/workqueue.h> +#include <linux/io.h> +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_cfg.h> +#include <adf_dbgfs.h> +#include "adf_dh895xcc_hw_data.h" + +static void adf_cleanup_pci_dev(struct adf_accel_dev *accel_dev) +{ + pci_release_regions(accel_dev->accel_pci_dev.pci_dev); + pci_disable_device(accel_dev->accel_pci_dev.pci_dev); +} + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *accel_pci_dev = &accel_dev->accel_pci_dev; + int i; + + for (i = 0; i < ADF_PCI_MAX_BARS; i++) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i]; + + if (bar->virt_addr) + pci_iounmap(accel_pci_dev->pci_dev, bar->virt_addr); + } + + if (accel_dev->hw_device) { + switch (accel_pci_dev->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_DH895XCC: + adf_clean_hw_data_dh895xcc(accel_dev->hw_device); + break; + default: + break; + } + kfree(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + adf_devmgr_rm_dev(accel_dev, NULL); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + int ret; + + switch (ent->device) { + case PCI_DEVICE_ID_INTEL_QAT_DH895XCC: + break; + default: + dev_err(&pdev->dev, "Invalid device 0x%x.\n", ent->device); + return -ENODEV; + } + + if (num_possible_nodes() > 1 && dev_to_node(&pdev->dev) < 0) { + /* If the accelerator is connected to a node with no memory + * there is no point in using the accelerator since the remote + * memory transaction will be very slow. */ + dev_err(&pdev->dev, "Invalid NUMA configuration.\n"); + return -EINVAL; + } + + accel_dev = kzalloc_node(sizeof(*accel_dev), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!accel_dev) + return -ENOMEM; + + INIT_LIST_HEAD(&accel_dev->crypto_list); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* Add accel device to accel table. + * This should be called before adf_cleanup_accel is called */ + if (adf_devmgr_add_dev(accel_dev, NULL)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + kfree(accel_dev); + return -EFAULT; + } + + accel_dev->owner = THIS_MODULE; + /* Allocate and configure device configuration structure */ + hw_data = kzalloc_node(sizeof(*hw_data), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + + accel_dev->hw_device = hw_data; + adf_init_hw_data_dh895xcc(accel_dev->hw_device); + pci_read_config_byte(pdev, PCI_REVISION_ID, &accel_pci_dev->revid); + pci_read_config_dword(pdev, ADF_DEVICE_FUSECTL_OFFSET, + &hw_data->fuses[ADF_FUSECTL0]); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + /* If the device has no acceleration engines then ignore it. */ + if (!hw_data->accel_mask || !hw_data->ae_mask || + ((~hw_data->ae_mask) & 0x01)) { + dev_err(&pdev->dev, "No acceleration units found"); + ret = -EFAULT; + goto out_err; + } + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + pcie_set_readrq(pdev, 1024); + + /* enable PCI device */ + if (pci_enable_device(pdev)) { + ret = -EFAULT; + goto out_err; + } + + /* set dma identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(48)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration\n"); + goto out_err_disable; + } + + if (pci_request_regions(pdev, ADF_DH895XCC_DEVICE_NAME)) { + ret = -EFAULT; + goto out_err_disable; + } + + /* Get accelerator capabilities mask */ + hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev); + + /* Find and map all the device's BARS */ + i = 0; + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM); + for_each_set_bit(bar_nr, &bar_mask, ADF_PCI_MAX_BARS * 2) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i++]; + + bar->base_addr = pci_resource_start(pdev, bar_nr); + if (!bar->base_addr) + break; + bar->size = pci_resource_len(pdev, bar_nr); + bar->virt_addr = pci_iomap(accel_pci_dev->pci_dev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to map BAR %d\n", bar_nr); + ret = -EFAULT; + goto out_err_free_reg; + } + } + pci_set_master(pdev); + + if (pci_save_state(pdev)) { + dev_err(&pdev->dev, "Failed to save pci state\n"); + ret = -ENOMEM; + goto out_err_free_reg; + } + + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, true); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err_free_reg: + pci_release_regions(accel_pci_dev->pci_dev); +out_err_disable: + pci_disable_device(accel_pci_dev->pci_dev); +out_err: + adf_cleanup_accel(accel_dev); + kfree(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); + adf_cleanup_pci_dev(accel_dev); + kfree(accel_dev); +} + +static void adf_shutdown(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + adf_dev_down(accel_dev); +} + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_DH895XCC) }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_DH895XCC_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, + .shutdown = adf_shutdown, + .sriov_configure = adf_sriov_configure, + .err_handler = &adf_err_handler, +}; + +static int __init adfdrv_init(void) +{ + request_module("intel_qat"); + + if (pci_register_driver(&adf_driver)) { + pr_err("QAT: Driver initialization failed\n"); + return -EFAULT; + } + return 0; +} + +static void __exit adfdrv_release(void) +{ + pci_unregister_driver(&adf_driver); +} + +module_init(adfdrv_init); +module_exit(adfdrv_release); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_FIRMWARE(ADF_DH895XCC_FW); +MODULE_FIRMWARE(ADF_DH895XCC_MMP); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_IMPORT_NS("CRYPTO_QAT"); diff --git a/drivers/crypto/intel/qat/qat_dh895xccvf/Makefile b/drivers/crypto/intel/qat/qat_dh895xccvf/Makefile new file mode 100644 index 000000000000..c2fdb6e0f68f --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xccvf/Makefile @@ -0,0 +1,3 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_CRYPTO_DEV_QAT_DH895xCCVF) += qat_dh895xccvf.o +qat_dh895xccvf-y := adf_drv.o adf_dh895xccvf_hw_data.o diff --git a/drivers/crypto/intel/qat/qat_dh895xccvf/adf_dh895xccvf_hw_data.c b/drivers/crypto/intel/qat/qat_dh895xccvf/adf_dh895xccvf_hw_data.c new file mode 100644 index 000000000000..828456c43b76 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xccvf/adf_dh895xccvf_hw_data.c @@ -0,0 +1,101 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2015 - 2021 Intel Corporation */ +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_gen2_config.h> +#include <adf_gen2_hw_csr_data.h> +#include <adf_gen2_hw_data.h> +#include <adf_gen2_pfvf.h> +#include <adf_pfvf_vf_msg.h> +#include "adf_dh895xccvf_hw_data.h" + +static struct adf_hw_device_class dh895xcciov_class = { + .name = ADF_DH895XCCVF_DEVICE_NAME, + .type = DEV_DH895XCCVF, +}; + +static u32 get_accel_mask(struct adf_hw_device_data *self) +{ + return ADF_DH895XCCIOV_ACCELERATORS_MASK; +} + +static u32 get_ae_mask(struct adf_hw_device_data *self) +{ + return ADF_DH895XCCIOV_ACCELENGINES_MASK; +} + +static u32 get_num_accels(struct adf_hw_device_data *self) +{ + return ADF_DH895XCCIOV_MAX_ACCELERATORS; +} + +static u32 get_num_aes(struct adf_hw_device_data *self) +{ + return ADF_DH895XCCIOV_MAX_ACCELENGINES; +} + +static u32 get_misc_bar_id(struct adf_hw_device_data *self) +{ + return ADF_DH895XCCIOV_PMISC_BAR; +} + +static u32 get_etr_bar_id(struct adf_hw_device_data *self) +{ + return ADF_DH895XCCIOV_ETR_BAR; +} + +static enum dev_sku_info get_sku(struct adf_hw_device_data *self) +{ + return DEV_SKU_VF; +} + +static int adf_vf_int_noop(struct adf_accel_dev *accel_dev) +{ + return 0; +} + +static void adf_vf_void_noop(struct adf_accel_dev *accel_dev) +{ +} + +void adf_init_hw_data_dh895xcciov(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class = &dh895xcciov_class; + hw_data->num_banks = ADF_DH895XCCIOV_ETR_MAX_BANKS; + hw_data->num_rings_per_bank = ADF_ETR_MAX_RINGS_PER_BANK; + hw_data->num_accel = ADF_DH895XCCIOV_MAX_ACCELERATORS; + hw_data->num_logical_accel = 1; + hw_data->num_engines = ADF_DH895XCCIOV_MAX_ACCELENGINES; + hw_data->tx_rx_gap = ADF_DH895XCCIOV_RX_RINGS_OFFSET; + hw_data->tx_rings_mask = ADF_DH895XCCIOV_TX_RINGS_MASK; + hw_data->ring_to_svc_map = ADF_GEN2_DEFAULT_RING_TO_SRV_MAP; + hw_data->alloc_irq = adf_vf_isr_resource_alloc; + hw_data->free_irq = adf_vf_isr_resource_free; + hw_data->enable_error_correction = adf_vf_void_noop; + hw_data->init_admin_comms = adf_vf_int_noop; + hw_data->exit_admin_comms = adf_vf_void_noop; + hw_data->send_admin_init = adf_vf2pf_notify_init; + hw_data->init_arb = adf_vf_int_noop; + hw_data->exit_arb = adf_vf_void_noop; + hw_data->disable_iov = adf_vf2pf_notify_shutdown; + hw_data->get_accel_mask = get_accel_mask; + hw_data->get_ae_mask = get_ae_mask; + hw_data->get_num_accels = get_num_accels; + hw_data->get_num_aes = get_num_aes; + hw_data->get_etr_bar_id = get_etr_bar_id; + hw_data->get_misc_bar_id = get_misc_bar_id; + hw_data->get_sku = get_sku; + hw_data->enable_ints = adf_vf_void_noop; + hw_data->dev_class->instances++; + hw_data->dev_config = adf_gen2_dev_config; + adf_devmgr_update_class_index(hw_data); + adf_gen2_init_vf_pfvf_ops(&hw_data->pfvf_ops); + adf_gen2_init_hw_csr_ops(&hw_data->csr_ops); + adf_gen2_init_dc_ops(&hw_data->dc_ops); +} + +void adf_clean_hw_data_dh895xcciov(struct adf_hw_device_data *hw_data) +{ + hw_data->dev_class->instances--; + adf_devmgr_update_class_index(hw_data); +} diff --git a/drivers/crypto/intel/qat/qat_dh895xccvf/adf_dh895xccvf_hw_data.h b/drivers/crypto/intel/qat/qat_dh895xccvf/adf_dh895xccvf_hw_data.h new file mode 100644 index 000000000000..6973fa967bc8 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xccvf/adf_dh895xccvf_hw_data.h @@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) */ +/* Copyright(c) 2015 - 2020 Intel Corporation */ +#ifndef ADF_DH895XVF_HW_DATA_H_ +#define ADF_DH895XVF_HW_DATA_H_ + +#define ADF_DH895XCCIOV_PMISC_BAR 1 +#define ADF_DH895XCCIOV_ACCELERATORS_MASK 0x1 +#define ADF_DH895XCCIOV_ACCELENGINES_MASK 0x1 +#define ADF_DH895XCCIOV_MAX_ACCELERATORS 1 +#define ADF_DH895XCCIOV_MAX_ACCELENGINES 1 +#define ADF_DH895XCCIOV_RX_RINGS_OFFSET 8 +#define ADF_DH895XCCIOV_TX_RINGS_MASK 0xFF +#define ADF_DH895XCCIOV_ETR_BAR 0 +#define ADF_DH895XCCIOV_ETR_MAX_BANKS 1 + +void adf_init_hw_data_dh895xcciov(struct adf_hw_device_data *hw_data); +void adf_clean_hw_data_dh895xcciov(struct adf_hw_device_data *hw_data); +#endif diff --git a/drivers/crypto/intel/qat/qat_dh895xccvf/adf_drv.c b/drivers/crypto/intel/qat/qat_dh895xccvf/adf_drv.c new file mode 100644 index 000000000000..7cd528ee31e7 --- /dev/null +++ b/drivers/crypto/intel/qat/qat_dh895xccvf/adf_drv.c @@ -0,0 +1,229 @@ +// SPDX-License-Identifier: (BSD-3-Clause OR GPL-2.0-only) +/* Copyright(c) 2014 - 2020 Intel Corporation */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <linux/types.h> +#include <linux/fs.h> +#include <linux/slab.h> +#include <linux/errno.h> +#include <linux/device.h> +#include <linux/dma-mapping.h> +#include <linux/platform_device.h> +#include <linux/workqueue.h> +#include <linux/io.h> +#include <adf_accel_devices.h> +#include <adf_common_drv.h> +#include <adf_cfg.h> +#include <adf_dbgfs.h> +#include "adf_dh895xccvf_hw_data.h" + +static const struct pci_device_id adf_pci_tbl[] = { + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QAT_DH895XCC_VF), }, + { } +}; +MODULE_DEVICE_TABLE(pci, adf_pci_tbl); + +static int adf_probe(struct pci_dev *dev, const struct pci_device_id *ent); +static void adf_remove(struct pci_dev *dev); + +static struct pci_driver adf_driver = { + .id_table = adf_pci_tbl, + .name = ADF_DH895XCCVF_DEVICE_NAME, + .probe = adf_probe, + .remove = adf_remove, +}; + +static void adf_cleanup_pci_dev(struct adf_accel_dev *accel_dev) +{ + pci_release_regions(accel_dev->accel_pci_dev.pci_dev); + pci_disable_device(accel_dev->accel_pci_dev.pci_dev); +} + +static void adf_cleanup_accel(struct adf_accel_dev *accel_dev) +{ + struct adf_accel_pci *accel_pci_dev = &accel_dev->accel_pci_dev; + struct adf_accel_dev *pf; + int i; + + for (i = 0; i < ADF_PCI_MAX_BARS; i++) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i]; + + if (bar->virt_addr) + pci_iounmap(accel_pci_dev->pci_dev, bar->virt_addr); + } + + if (accel_dev->hw_device) { + switch (accel_pci_dev->pci_dev->device) { + case PCI_DEVICE_ID_INTEL_QAT_DH895XCC_VF: + adf_clean_hw_data_dh895xcciov(accel_dev->hw_device); + break; + default: + break; + } + kfree(accel_dev->hw_device); + accel_dev->hw_device = NULL; + } + adf_dbgfs_exit(accel_dev); + adf_cfg_dev_remove(accel_dev); + pf = adf_devmgr_pci_to_accel_dev(accel_pci_dev->pci_dev->physfn); + adf_devmgr_rm_dev(accel_dev, pf); +} + +static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + struct adf_accel_dev *accel_dev; + struct adf_accel_dev *pf; + struct adf_accel_pci *accel_pci_dev; + struct adf_hw_device_data *hw_data; + unsigned int i, bar_nr; + unsigned long bar_mask; + int ret; + + switch (ent->device) { + case PCI_DEVICE_ID_INTEL_QAT_DH895XCC_VF: + break; + default: + dev_err(&pdev->dev, "Invalid device 0x%x.\n", ent->device); + return -ENODEV; + } + + accel_dev = kzalloc_node(sizeof(*accel_dev), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!accel_dev) + return -ENOMEM; + + accel_dev->is_vf = true; + pf = adf_devmgr_pci_to_accel_dev(pdev->physfn); + accel_pci_dev = &accel_dev->accel_pci_dev; + accel_pci_dev->pci_dev = pdev; + + /* Add accel device to accel table */ + if (adf_devmgr_add_dev(accel_dev, pf)) { + dev_err(&pdev->dev, "Failed to add new accelerator device.\n"); + kfree(accel_dev); + return -EFAULT; + } + INIT_LIST_HEAD(&accel_dev->crypto_list); + + accel_dev->owner = THIS_MODULE; + /* Allocate and configure device configuration structure */ + hw_data = kzalloc_node(sizeof(*hw_data), GFP_KERNEL, + dev_to_node(&pdev->dev)); + if (!hw_data) { + ret = -ENOMEM; + goto out_err; + } + accel_dev->hw_device = hw_data; + adf_init_hw_data_dh895xcciov(accel_dev->hw_device); + + /* Get Accelerators and Accelerators Engines masks */ + hw_data->accel_mask = hw_data->get_accel_mask(hw_data); + hw_data->ae_mask = hw_data->get_ae_mask(hw_data); + accel_pci_dev->sku = hw_data->get_sku(hw_data); + + /* Create device configuration table */ + ret = adf_cfg_dev_add(accel_dev); + if (ret) + goto out_err; + + /* enable PCI device */ + if (pci_enable_device(pdev)) { + ret = -EFAULT; + goto out_err; + } + + /* set dma identifier */ + ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(48)); + if (ret) { + dev_err(&pdev->dev, "No usable DMA configuration\n"); + goto out_err_disable; + } + + if (pci_request_regions(pdev, ADF_DH895XCCVF_DEVICE_NAME)) { + ret = -EFAULT; + goto out_err_disable; + } + + /* Find and map all the device's BARS */ + i = 0; + bar_mask = pci_select_bars(pdev, IORESOURCE_MEM); + for_each_set_bit(bar_nr, &bar_mask, ADF_PCI_MAX_BARS * 2) { + struct adf_bar *bar = &accel_pci_dev->pci_bars[i++]; + + bar->base_addr = pci_resource_start(pdev, bar_nr); + if (!bar->base_addr) + break; + bar->size = pci_resource_len(pdev, bar_nr); + bar->virt_addr = pci_iomap(accel_pci_dev->pci_dev, bar_nr, 0); + if (!bar->virt_addr) { + dev_err(&pdev->dev, "Failed to map BAR %d\n", bar_nr); + ret = -EFAULT; + goto out_err_free_reg; + } + } + pci_set_master(pdev); + /* Completion for VF2PF request/response message exchange */ + init_completion(&accel_dev->vf.msg_received); + + adf_dbgfs_init(accel_dev); + + ret = adf_dev_up(accel_dev, false); + if (ret) + goto out_err_dev_stop; + + return ret; + +out_err_dev_stop: + adf_dev_down(accel_dev); +out_err_free_reg: + pci_release_regions(accel_pci_dev->pci_dev); +out_err_disable: + pci_disable_device(accel_pci_dev->pci_dev); +out_err: + adf_cleanup_accel(accel_dev); + kfree(accel_dev); + return ret; +} + +static void adf_remove(struct pci_dev *pdev) +{ + struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev); + + if (!accel_dev) { + pr_err("QAT: Driver removal failed\n"); + return; + } + adf_flush_vf_wq(accel_dev); + adf_dev_down(accel_dev); + adf_cleanup_accel(accel_dev); + adf_cleanup_pci_dev(accel_dev); + kfree(accel_dev); +} + +static int __init adfdrv_init(void) +{ + request_module("intel_qat"); + + if (pci_register_driver(&adf_driver)) { + pr_err("QAT: Driver initialization failed\n"); + return -EFAULT; + } + return 0; +} + +static void __exit adfdrv_release(void) +{ + pci_unregister_driver(&adf_driver); + adf_clean_vf_map(true); +} + +module_init(adfdrv_init); +module_exit(adfdrv_release); + +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel"); +MODULE_DESCRIPTION("Intel(R) QuickAssist Technology"); +MODULE_VERSION(ADF_DRV_VERSION); +MODULE_IMPORT_NS("CRYPTO_QAT"); |