diff options
Diffstat (limited to 'drivers/net/ethernet/amazon/ena/ena_netdev.c')
| -rw-r--r-- | drivers/net/ethernet/amazon/ena/ena_netdev.c | 2470 |
1 files changed, 1578 insertions, 892 deletions
diff --git a/drivers/net/ethernet/amazon/ena/ena_netdev.c b/drivers/net/ethernet/amazon/ena/ena_netdev.c index a70bb1bb90e7..92d149d4f091 100644 --- a/drivers/net/ethernet/amazon/ena/ena_netdev.c +++ b/drivers/net/ethernet/amazon/ena/ena_netdev.c @@ -1,42 +1,11 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB /* - * Copyright 2015 Amazon.com, Inc. or its affiliates. - * - * This software is available to you under a choice of one of two - * licenses. You may choose to be licensed under the terms of the GNU - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. + * Copyright 2015-2020 Amazon.com, Inc. or its affiliates. All rights reserved. */ #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt -#ifdef CONFIG_RFS_ACCEL -#include <linux/cpu_rmap.h> -#endif /* CONFIG_RFS_ACCEL */ #include <linux/ethtool.h> -#include <linux/if_vlan.h> #include <linux/kernel.h> #include <linux/module.h> #include <linux/numa.h> @@ -48,24 +17,25 @@ #include "ena_netdev.h" #include "ena_pci_id_tbl.h" +#include "ena_xdp.h" + +#include "ena_phc.h" -static char version[] = DEVICE_NAME " v" DRV_MODULE_VERSION "\n"; +#include "ena_devlink.h" + +#include "ena_debugfs.h" MODULE_AUTHOR("Amazon.com, Inc. or its affiliates"); MODULE_DESCRIPTION(DEVICE_NAME); MODULE_LICENSE("GPL"); -MODULE_VERSION(DRV_MODULE_VERSION); /* Time in jiffies before concluding the transmitter is hung. */ #define TX_TIMEOUT (5 * HZ) -#define ENA_NAPI_BUDGET 64 +#define ENA_MAX_RINGS min_t(unsigned int, ENA_MAX_NUM_IO_QUEUES, num_possible_cpus()) #define DEFAULT_MSG_ENABLE (NETIF_MSG_DRV | NETIF_MSG_PROBE | NETIF_MSG_IFUP | \ - NETIF_MSG_TX_DONE | NETIF_MSG_TX_ERR | NETIF_MSG_RX_ERR) -static int debug = -1; -module_param(debug, int, 0); -MODULE_PARM_DESC(debug, "Debug level (0=none,...,16=all)"); + NETIF_MSG_IFDOWN | NETIF_MSG_TX_ERR | NETIF_MSG_RX_ERR) static struct ena_aenq_handlers aenq_handlers; @@ -75,33 +45,54 @@ MODULE_DEVICE_TABLE(pci, ena_pci_tbl); static int ena_rss_init_default(struct ena_adapter *adapter); static void check_for_admin_com_state(struct ena_adapter *adapter); -static void ena_destroy_device(struct ena_adapter *adapter, bool graceful); -static int ena_restore_device(struct ena_adapter *adapter); -static void ena_tx_timeout(struct net_device *dev) +static void ena_tx_timeout(struct net_device *dev, unsigned int txqueue) { + enum ena_regs_reset_reason_types reset_reason = ENA_REGS_RESET_OS_NETDEV_WD; struct ena_adapter *adapter = netdev_priv(dev); + unsigned int time_since_last_napi, threshold; + struct ena_ring *tx_ring; + int napi_scheduled; + + if (txqueue >= adapter->num_io_queues) { + netdev_err(dev, "TX timeout on invalid queue %u\n", txqueue); + goto schedule_reset; + } + + threshold = jiffies_to_usecs(dev->watchdog_timeo); + tx_ring = &adapter->tx_ring[txqueue]; + + time_since_last_napi = jiffies_to_usecs(jiffies - tx_ring->tx_stats.last_napi_jiffies); + napi_scheduled = !!(tx_ring->napi->state & NAPIF_STATE_SCHED); + netdev_err(dev, + "TX q %d is paused for too long (threshold %u). Time since last napi %u usec. napi scheduled: %d\n", + txqueue, + threshold, + time_since_last_napi, + napi_scheduled); + + if (threshold < time_since_last_napi && napi_scheduled) { + netdev_err(dev, + "napi handler hasn't been called for a long time but is scheduled\n"); + reset_reason = ENA_REGS_RESET_SUSPECTED_POLL_STARVATION; + } +schedule_reset: /* Change the state of the device to trigger reset * Check that we are not in the middle or a trigger already */ - if (test_and_set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags)) return; - adapter->reset_reason = ENA_REGS_RESET_OS_NETDEV_WD; - u64_stats_update_begin(&adapter->syncp); - adapter->dev_stats.tx_timeout++; - u64_stats_update_end(&adapter->syncp); - - netif_err(adapter, tx_err, dev, "Transmit time out\n"); + ena_reset_device(adapter, reset_reason); + ena_increase_stat(&adapter->dev_stats.tx_timeout, 1, &adapter->syncp); } static void update_rx_ring_mtu(struct ena_adapter *adapter, int mtu) { int i; - for (i = 0; i < adapter->num_queues; i++) + for (i = 0; i < adapter->num_io_queues; i++) adapter->rx_ring[i].mtu = mtu; } @@ -112,9 +103,9 @@ static int ena_change_mtu(struct net_device *dev, int new_mtu) ret = ena_com_set_dev_mtu(adapter->ena_dev, new_mtu); if (!ret) { - netif_dbg(adapter, drv, dev, "set MTU to %d\n", new_mtu); + netif_dbg(adapter, drv, dev, "Set MTU to %d\n", new_mtu); update_rx_ring_mtu(adapter, new_mtu); - dev->mtu = new_mtu; + WRITE_ONCE(dev->mtu, new_mtu); } else { netif_err(adapter, drv, dev, "Failed to set MTU to %d\n", new_mtu); @@ -123,27 +114,52 @@ static int ena_change_mtu(struct net_device *dev, int new_mtu) return ret; } -static int ena_init_rx_cpu_rmap(struct ena_adapter *adapter) +int ena_xmit_common(struct ena_adapter *adapter, + struct ena_ring *ring, + struct ena_tx_buffer *tx_info, + struct ena_com_tx_ctx *ena_tx_ctx, + u16 next_to_use, + u32 bytes) { -#ifdef CONFIG_RFS_ACCEL - u32 i; - int rc; + int rc, nb_hw_desc; - adapter->netdev->rx_cpu_rmap = alloc_irq_cpu_rmap(adapter->num_queues); - if (!adapter->netdev->rx_cpu_rmap) - return -ENOMEM; - for (i = 0; i < adapter->num_queues; i++) { - int irq_idx = ENA_IO_IRQ_IDX(i); + if (unlikely(ena_com_is_doorbell_needed(ring->ena_com_io_sq, + ena_tx_ctx))) { + netif_dbg(adapter, tx_queued, adapter->netdev, + "llq tx max burst size of queue %d achieved, writing doorbell to send burst\n", + ring->qid); + ena_ring_tx_doorbell(ring); + } - rc = irq_cpu_rmap_add(adapter->netdev->rx_cpu_rmap, - pci_irq_vector(adapter->pdev, irq_idx)); - if (rc) { - free_irq_cpu_rmap(adapter->netdev->rx_cpu_rmap); - adapter->netdev->rx_cpu_rmap = NULL; - return rc; - } + /* prepare the packet's descriptors to dma engine */ + rc = ena_com_prepare_tx(ring->ena_com_io_sq, ena_tx_ctx, + &nb_hw_desc); + + /* In case there isn't enough space in the queue for the packet, + * we simply drop it. All other failure reasons of + * ena_com_prepare_tx() are fatal and therefore require a device reset. + */ + if (unlikely(rc)) { + netif_err(adapter, tx_queued, adapter->netdev, + "Failed to prepare tx bufs\n"); + ena_increase_stat(&ring->tx_stats.prepare_ctx_err, 1, &ring->syncp); + if (rc != -ENOMEM) + ena_reset_device(adapter, ENA_REGS_RESET_DRIVER_INVALID_STATE); + return rc; } -#endif /* CONFIG_RFS_ACCEL */ + + u64_stats_update_begin(&ring->syncp); + ring->tx_stats.cnt++; + ring->tx_stats.bytes += bytes; + u64_stats_update_end(&ring->syncp); + + tx_info->tx_descs = nb_hw_desc; + tx_info->total_tx_size = bytes; + tx_info->last_jiffies = jiffies; + tx_info->print_once = 0; + + ring->next_to_use = ENA_TX_RING_IDX_NEXT(next_to_use, + ring->ring_size); return 0; } @@ -158,14 +174,14 @@ static void ena_init_io_rings_common(struct ena_adapter *adapter, ring->adapter = adapter; ring->ena_dev = adapter->ena_dev; ring->per_napi_packets = 0; - ring->per_napi_bytes = 0; ring->cpu = 0; - ring->first_interrupt = false; + ring->numa_node = 0; ring->no_interrupt_event_cnt = 0; u64_stats_init(&ring->syncp); } -static void ena_init_io_rings(struct ena_adapter *adapter) +void ena_init_io_rings(struct ena_adapter *adapter, + int first_index, int count) { struct ena_com_dev *ena_dev; struct ena_ring *txr, *rxr; @@ -173,29 +189,39 @@ static void ena_init_io_rings(struct ena_adapter *adapter) ena_dev = adapter->ena_dev; - for (i = 0; i < adapter->num_queues; i++) { + for (i = first_index; i < first_index + count; i++) { txr = &adapter->tx_ring[i]; rxr = &adapter->rx_ring[i]; - /* TX/RX common ring state */ + /* TX common ring state */ ena_init_io_rings_common(adapter, txr, i); - ena_init_io_rings_common(adapter, rxr, i); /* TX specific ring state */ - txr->ring_size = adapter->tx_ring_size; + txr->ring_size = adapter->requested_tx_ring_size; txr->tx_max_header_size = ena_dev->tx_max_header_size; txr->tx_mem_queue_type = ena_dev->tx_mem_queue_type; txr->sgl_size = adapter->max_tx_sgl_size; txr->smoothed_interval = ena_com_get_nonadaptive_moderation_interval_tx(ena_dev); - - /* RX specific ring state */ - rxr->ring_size = adapter->rx_ring_size; - rxr->rx_copybreak = adapter->rx_copybreak; - rxr->sgl_size = adapter->max_rx_sgl_size; - rxr->smoothed_interval = - ena_com_get_nonadaptive_moderation_interval_rx(ena_dev); - rxr->empty_rx_queue = 0; + txr->disable_meta_caching = adapter->disable_meta_caching; + spin_lock_init(&txr->xdp_tx_lock); + + /* Don't init RX queues for xdp queues */ + if (!ENA_IS_XDP_INDEX(adapter, i)) { + /* RX common ring state */ + ena_init_io_rings_common(adapter, rxr, i); + + /* RX specific ring state */ + rxr->ring_size = adapter->requested_rx_ring_size; + rxr->rx_copybreak = adapter->rx_copybreak; + rxr->sgl_size = adapter->max_rx_sgl_size; + rxr->smoothed_interval = + ena_com_get_nonadaptive_moderation_interval_rx(ena_dev); + rxr->empty_rx_queue = 0; + rxr->rx_headroom = NET_SKB_PAD; + adapter->ena_napi[i].dim.mode = DIM_CQ_PERIOD_MODE_START_FROM_EQE; + rxr->xdp_ring = &adapter->tx_ring[i + adapter->num_io_queues]; + } } } @@ -224,33 +250,28 @@ static int ena_setup_tx_resources(struct ena_adapter *adapter, int qid) if (!tx_ring->tx_buffer_info) { tx_ring->tx_buffer_info = vzalloc(size); if (!tx_ring->tx_buffer_info) - return -ENOMEM; + goto err_tx_buffer_info; } size = sizeof(u16) * tx_ring->ring_size; - tx_ring->free_tx_ids = vzalloc_node(size, node); - if (!tx_ring->free_tx_ids) { - tx_ring->free_tx_ids = vzalloc(size); - if (!tx_ring->free_tx_ids) { - vfree(tx_ring->tx_buffer_info); - return -ENOMEM; - } + tx_ring->free_ids = vzalloc_node(size, node); + if (!tx_ring->free_ids) { + tx_ring->free_ids = vzalloc(size); + if (!tx_ring->free_ids) + goto err_tx_free_ids; } size = tx_ring->tx_max_header_size; tx_ring->push_buf_intermediate_buf = vzalloc_node(size, node); if (!tx_ring->push_buf_intermediate_buf) { tx_ring->push_buf_intermediate_buf = vzalloc(size); - if (!tx_ring->push_buf_intermediate_buf) { - vfree(tx_ring->tx_buffer_info); - vfree(tx_ring->free_tx_ids); - return -ENOMEM; - } + if (!tx_ring->push_buf_intermediate_buf) + goto err_push_buf_intermediate_buf; } /* Req id ring for TX out of order completions */ for (i = 0; i < tx_ring->ring_size; i++) - tx_ring->free_tx_ids[i] = i; + tx_ring->free_ids[i] = i; /* Reset tx statistics */ memset(&tx_ring->tx_stats, 0x0, sizeof(tx_ring->tx_stats)); @@ -258,7 +279,17 @@ static int ena_setup_tx_resources(struct ena_adapter *adapter, int qid) tx_ring->next_to_use = 0; tx_ring->next_to_clean = 0; tx_ring->cpu = ena_irq->cpu; + tx_ring->numa_node = node; return 0; + +err_push_buf_intermediate_buf: + vfree(tx_ring->free_ids); + tx_ring->free_ids = NULL; +err_tx_free_ids: + vfree(tx_ring->tx_buffer_info); + tx_ring->tx_buffer_info = NULL; +err_tx_buffer_info: + return -ENOMEM; } /* ena_free_tx_resources - Free I/O Tx Resources per Queue @@ -274,23 +305,19 @@ static void ena_free_tx_resources(struct ena_adapter *adapter, int qid) vfree(tx_ring->tx_buffer_info); tx_ring->tx_buffer_info = NULL; - vfree(tx_ring->free_tx_ids); - tx_ring->free_tx_ids = NULL; + vfree(tx_ring->free_ids); + tx_ring->free_ids = NULL; vfree(tx_ring->push_buf_intermediate_buf); tx_ring->push_buf_intermediate_buf = NULL; } -/* ena_setup_all_tx_resources - allocate I/O Tx queues resources for All queues - * @adapter: private structure - * - * Return 0 on success, negative on failure - */ -static int ena_setup_all_tx_resources(struct ena_adapter *adapter) +int ena_setup_tx_resources_in_range(struct ena_adapter *adapter, + int first_index, int count) { int i, rc = 0; - for (i = 0; i < adapter->num_queues; i++) { + for (i = first_index; i < first_index + count; i++) { rc = ena_setup_tx_resources(adapter, i); if (rc) goto err_setup_tx; @@ -304,40 +331,31 @@ err_setup_tx: "Tx queue %d: allocation failed\n", i); /* rewind the index freeing the rings as we go */ - while (i--) + while (first_index < i--) ena_free_tx_resources(adapter, i); return rc; } -/* ena_free_all_io_tx_resources - Free I/O Tx Resources for All Queues - * @adapter: board private structure - * - * Free all transmit software resources - */ -static void ena_free_all_io_tx_resources(struct ena_adapter *adapter) +void ena_free_all_io_tx_resources_in_range(struct ena_adapter *adapter, + int first_index, int count) { int i; - for (i = 0; i < adapter->num_queues; i++) + for (i = first_index; i < first_index + count; i++) ena_free_tx_resources(adapter, i); } -static inline int validate_rx_req_id(struct ena_ring *rx_ring, u16 req_id) +/* ena_free_all_io_tx_resources - Free I/O Tx Resources for All Queues + * @adapter: board private structure + * + * Free all transmit software resources + */ +void ena_free_all_io_tx_resources(struct ena_adapter *adapter) { - if (likely(req_id < rx_ring->ring_size)) - return 0; - - netif_err(rx_ring->adapter, rx_err, rx_ring->netdev, - "Invalid rx req_id: %hu\n", req_id); - - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.bad_req_id++; - u64_stats_update_end(&rx_ring->syncp); - - /* Trigger device reset */ - rx_ring->adapter->reset_reason = ENA_REGS_RESET_INV_RX_REQ_ID; - set_bit(ENA_FLAG_TRIGGER_RESET, &rx_ring->adapter->flags); - return -EFAULT; + ena_free_all_io_tx_resources_in_range(adapter, + 0, + adapter->xdp_num_queues + + adapter->num_io_queues); } /* ena_setup_rx_resources - allocate I/O Rx resources (Descriptors) @@ -373,18 +391,19 @@ static int ena_setup_rx_resources(struct ena_adapter *adapter, } size = sizeof(u16) * rx_ring->ring_size; - rx_ring->free_rx_ids = vzalloc_node(size, node); - if (!rx_ring->free_rx_ids) { - rx_ring->free_rx_ids = vzalloc(size); - if (!rx_ring->free_rx_ids) { + rx_ring->free_ids = vzalloc_node(size, node); + if (!rx_ring->free_ids) { + rx_ring->free_ids = vzalloc(size); + if (!rx_ring->free_ids) { vfree(rx_ring->rx_buffer_info); + rx_ring->rx_buffer_info = NULL; return -ENOMEM; } } /* Req id ring for receiving RX pkts out of order */ for (i = 0; i < rx_ring->ring_size; i++) - rx_ring->free_rx_ids[i] = i; + rx_ring->free_ids[i] = i; /* Reset rx statistics */ memset(&rx_ring->rx_stats, 0x0, sizeof(rx_ring->rx_stats)); @@ -392,6 +411,7 @@ static int ena_setup_rx_resources(struct ena_adapter *adapter, rx_ring->next_to_clean = 0; rx_ring->next_to_use = 0; rx_ring->cpu = ena_irq->cpu; + rx_ring->numa_node = node; return 0; } @@ -410,8 +430,8 @@ static void ena_free_rx_resources(struct ena_adapter *adapter, vfree(rx_ring->rx_buffer_info); rx_ring->rx_buffer_info = NULL; - vfree(rx_ring->free_rx_ids); - rx_ring->free_rx_ids = NULL; + vfree(rx_ring->free_ids); + rx_ring->free_ids = NULL; } /* ena_setup_all_rx_resources - allocate I/O Rx queues resources for all queues @@ -423,7 +443,7 @@ static int ena_setup_all_rx_resources(struct ena_adapter *adapter) { int i, rc = 0; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < adapter->num_io_queues; i++) { rc = ena_setup_rx_resources(adapter, i); if (rc) goto err_setup_rx; @@ -451,56 +471,87 @@ static void ena_free_all_io_rx_resources(struct ena_adapter *adapter) { int i; - for (i = 0; i < adapter->num_queues; i++) + for (i = 0; i < adapter->num_io_queues; i++) ena_free_rx_resources(adapter, i); } -static inline int ena_alloc_rx_page(struct ena_ring *rx_ring, - struct ena_rx_buffer *rx_info, gfp_t gfp) +static struct page *ena_alloc_map_page(struct ena_ring *rx_ring, + dma_addr_t *dma) { + struct page *page; + + /* This would allocate the page on the same NUMA node the executing code + * is running on. + */ + page = dev_alloc_page(); + if (!page) { + ena_increase_stat(&rx_ring->rx_stats.page_alloc_fail, 1, &rx_ring->syncp); + return ERR_PTR(-ENOSPC); + } + + /* To enable NIC-side port-mirroring, AKA SPAN port, + * we make the buffer readable from the nic as well + */ + *dma = dma_map_page(rx_ring->dev, page, 0, ENA_PAGE_SIZE, + DMA_BIDIRECTIONAL); + if (unlikely(dma_mapping_error(rx_ring->dev, *dma))) { + ena_increase_stat(&rx_ring->rx_stats.dma_mapping_err, 1, + &rx_ring->syncp); + __free_page(page); + return ERR_PTR(-EIO); + } + + return page; +} + +static int ena_alloc_rx_buffer(struct ena_ring *rx_ring, + struct ena_rx_buffer *rx_info) +{ + int headroom = rx_ring->rx_headroom; struct ena_com_buf *ena_buf; struct page *page; dma_addr_t dma; + int tailroom; + + /* restore page offset value in case it has been changed by device */ + rx_info->buf_offset = headroom; /* if previous allocated page is not used */ if (unlikely(rx_info->page)) return 0; - page = alloc_page(gfp); - if (unlikely(!page)) { - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.page_alloc_fail++; - u64_stats_update_end(&rx_ring->syncp); - return -ENOMEM; - } + /* We handle DMA here */ + page = ena_alloc_map_page(rx_ring, &dma); + if (IS_ERR(page)) + return PTR_ERR(page); - dma = dma_map_page(rx_ring->dev, page, 0, ENA_PAGE_SIZE, - DMA_FROM_DEVICE); - if (unlikely(dma_mapping_error(rx_ring->dev, dma))) { - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.dma_mapping_err++; - u64_stats_update_end(&rx_ring->syncp); - - __free_page(page); - return -EIO; - } netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev, - "alloc page %p, rx_info %p\n", page, rx_info); + "Allocate page %p, rx_info %p\n", page, rx_info); + + tailroom = SKB_DATA_ALIGN(sizeof(struct skb_shared_info)); rx_info->page = page; + rx_info->dma_addr = dma; rx_info->page_offset = 0; ena_buf = &rx_info->ena_buf; - ena_buf->paddr = dma; - ena_buf->len = ENA_PAGE_SIZE; + ena_buf->paddr = dma + headroom; + ena_buf->len = ENA_PAGE_SIZE - headroom - tailroom; return 0; } +static void ena_unmap_rx_buff_attrs(struct ena_ring *rx_ring, + struct ena_rx_buffer *rx_info, + unsigned long attrs) +{ + dma_unmap_page_attrs(rx_ring->dev, rx_info->dma_addr, ENA_PAGE_SIZE, DMA_BIDIRECTIONAL, + attrs); +} + static void ena_free_rx_page(struct ena_ring *rx_ring, struct ena_rx_buffer *rx_info) { struct page *page = rx_info->page; - struct ena_com_buf *ena_buf = &rx_info->ena_buf; if (unlikely(!page)) { netif_warn(rx_ring->adapter, rx_err, rx_ring->netdev, @@ -508,8 +559,7 @@ static void ena_free_rx_page(struct ena_ring *rx_ring, return; } - dma_unmap_page(rx_ring->dev, ena_buf->paddr, ENA_PAGE_SIZE, - DMA_FROM_DEVICE); + ena_unmap_rx_buff_attrs(rx_ring, rx_info, 0); __free_page(page); rx_info->page = NULL; @@ -526,19 +576,14 @@ static int ena_refill_rx_bufs(struct ena_ring *rx_ring, u32 num) for (i = 0; i < num; i++) { struct ena_rx_buffer *rx_info; - req_id = rx_ring->free_rx_ids[next_to_use]; - rc = validate_rx_req_id(rx_ring, req_id); - if (unlikely(rc < 0)) - break; + req_id = rx_ring->free_ids[next_to_use]; rx_info = &rx_ring->rx_buffer_info[req_id]; - - rc = ena_alloc_rx_page(rx_ring, rx_info, - GFP_ATOMIC | __GFP_COMP); + rc = ena_alloc_rx_buffer(rx_ring, rx_info); if (unlikely(rc < 0)) { netif_warn(rx_ring->adapter, rx_err, rx_ring->netdev, - "failed to alloc buffer for rx queue %d\n", + "Failed to allocate buffer for rx queue %d\n", rx_ring->qid); break; } @@ -547,7 +592,7 @@ static int ena_refill_rx_bufs(struct ena_ring *rx_ring, u32 num) req_id); if (unlikely(rc)) { netif_warn(rx_ring->adapter, rx_status, rx_ring->netdev, - "failed to add buffer for rx queue %d\n", + "Failed to add buffer for rx queue %d\n", rx_ring->qid); break; } @@ -556,12 +601,11 @@ static int ena_refill_rx_bufs(struct ena_ring *rx_ring, u32 num) } if (unlikely(i < num)) { - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.refil_partial++; - u64_stats_update_end(&rx_ring->syncp); - netdev_warn(rx_ring->netdev, - "refilled rx qid %d with only %d buffers (from %d)\n", - rx_ring->qid, i, num); + ena_increase_stat(&rx_ring->rx_stats.refil_partial, 1, + &rx_ring->syncp); + netif_warn(rx_ring->adapter, rx_err, rx_ring->netdev, + "Refilled rx qid %d with only %d buffers (from %d)\n", + rx_ring->qid, i, num); } /* ena_com_write_sq_doorbell issues a wmb() */ @@ -589,21 +633,20 @@ static void ena_free_rx_bufs(struct ena_adapter *adapter, /* ena_refill_all_rx_bufs - allocate all queues Rx buffers * @adapter: board private structure - * */ static void ena_refill_all_rx_bufs(struct ena_adapter *adapter) { struct ena_ring *rx_ring; int i, rc, bufs_num; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < adapter->num_io_queues; i++) { rx_ring = &adapter->rx_ring[i]; bufs_num = rx_ring->ring_size - 1; rc = ena_refill_rx_bufs(rx_ring, bufs_num); if (unlikely(rc != bufs_num)) netif_warn(rx_ring->adapter, rx_status, rx_ring->netdev, - "refilling Queue %d failed. allocated %d buffers from: %d\n", + "Refilling Queue %d failed. allocated %d buffers from: %d\n", i, rc, bufs_num); } } @@ -612,12 +655,12 @@ static void ena_free_all_rx_bufs(struct ena_adapter *adapter) { int i; - for (i = 0; i < adapter->num_queues; i++) + for (i = 0; i < adapter->num_io_queues; i++) ena_free_rx_bufs(adapter, i); } -static inline void ena_unmap_tx_skb(struct ena_ring *tx_ring, - struct ena_tx_buffer *tx_info) +void ena_unmap_tx_buff(struct ena_ring *tx_ring, + struct ena_tx_buffer *tx_info) { struct ena_com_buf *ena_buf; u32 cnt; @@ -652,8 +695,11 @@ static inline void ena_unmap_tx_skb(struct ena_ring *tx_ring, static void ena_free_tx_bufs(struct ena_ring *tx_ring) { bool print_once = true; + bool is_xdp_ring; u32 i; + is_xdp_ring = ENA_IS_XDP_INDEX(tx_ring->adapter, tx_ring->qid); + for (i = 0; i < tx_ring->ring_size; i++) { struct ena_tx_buffer *tx_info = &tx_ring->tx_buffer_info[i]; @@ -661,22 +707,27 @@ static void ena_free_tx_bufs(struct ena_ring *tx_ring) continue; if (print_once) { - netdev_notice(tx_ring->netdev, - "free uncompleted tx skb qid %d idx 0x%x\n", - tx_ring->qid, i); + netif_notice(tx_ring->adapter, ifdown, tx_ring->netdev, + "Free uncompleted tx skb qid %d idx 0x%x\n", + tx_ring->qid, i); print_once = false; } else { - netdev_dbg(tx_ring->netdev, - "free uncompleted tx skb qid %d idx 0x%x\n", - tx_ring->qid, i); + netif_dbg(tx_ring->adapter, ifdown, tx_ring->netdev, + "Free uncompleted tx skb qid %d idx 0x%x\n", + tx_ring->qid, i); } - ena_unmap_tx_skb(tx_ring, tx_info); + ena_unmap_tx_buff(tx_ring, tx_info); - dev_kfree_skb_any(tx_info->skb); + if (is_xdp_ring) + xdp_return_frame(tx_info->xdpf); + else + dev_kfree_skb_any(tx_info->skb); } - netdev_tx_reset_queue(netdev_get_tx_queue(tx_ring->netdev, - tx_ring->qid)); + + if (!is_xdp_ring) + netdev_tx_reset_queue(netdev_get_tx_queue(tx_ring->netdev, + tx_ring->qid)); } static void ena_free_all_tx_bufs(struct ena_adapter *adapter) @@ -684,7 +735,7 @@ static void ena_free_all_tx_bufs(struct ena_adapter *adapter) struct ena_ring *tx_ring; int i; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < adapter->num_io_queues + adapter->xdp_num_queues; i++) { tx_ring = &adapter->tx_ring[i]; ena_free_tx_bufs(tx_ring); } @@ -695,7 +746,7 @@ static void ena_destroy_all_tx_queues(struct ena_adapter *adapter) u16 ena_qid; int i; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < adapter->num_io_queues + adapter->xdp_num_queues; i++) { ena_qid = ENA_IO_TXQ_IDX(i); ena_com_destroy_io_queue(adapter->ena_dev, ena_qid); } @@ -706,8 +757,10 @@ static void ena_destroy_all_rx_queues(struct ena_adapter *adapter) u16 ena_qid; int i; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < adapter->num_io_queues; i++) { ena_qid = ENA_IO_RXQ_IDX(i); + cancel_work_sync(&adapter->ena_napi[i].dim.work); + ena_xdp_unregister_rxq_info(&adapter->rx_ring[i]); ena_com_destroy_io_queue(adapter->ena_dev, ena_qid); } } @@ -718,33 +771,39 @@ static void ena_destroy_all_io_queues(struct ena_adapter *adapter) ena_destroy_all_rx_queues(adapter); } -static int validate_tx_req_id(struct ena_ring *tx_ring, u16 req_id) +int handle_invalid_req_id(struct ena_ring *ring, u16 req_id, + struct ena_tx_buffer *tx_info, bool is_xdp) { - struct ena_tx_buffer *tx_info = NULL; - - if (likely(req_id < tx_ring->ring_size)) { - tx_info = &tx_ring->tx_buffer_info[req_id]; - if (likely(tx_info->skb)) - return 0; - } - if (tx_info) - netif_err(tx_ring->adapter, tx_done, tx_ring->netdev, - "tx_info doesn't have valid skb\n"); + netif_err(ring->adapter, + tx_done, + ring->netdev, + "tx_info doesn't have valid %s. qid %u req_id %u", + is_xdp ? "xdp frame" : "skb", ring->qid, req_id); else - netif_err(tx_ring->adapter, tx_done, tx_ring->netdev, - "Invalid req_id: %hu\n", req_id); + netif_err(ring->adapter, + tx_done, + ring->netdev, + "Invalid req_id %u in qid %u\n", + req_id, ring->qid); - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.bad_req_id++; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&ring->tx_stats.bad_req_id, 1, &ring->syncp); + ena_reset_device(ring->adapter, ENA_REGS_RESET_INV_TX_REQ_ID); - /* Trigger device reset */ - tx_ring->adapter->reset_reason = ENA_REGS_RESET_INV_TX_REQ_ID; - set_bit(ENA_FLAG_TRIGGER_RESET, &tx_ring->adapter->flags); return -EFAULT; } +static int validate_tx_req_id(struct ena_ring *tx_ring, u16 req_id) +{ + struct ena_tx_buffer *tx_info; + + tx_info = &tx_ring->tx_buffer_info[req_id]; + if (likely(tx_info->skb)) + return 0; + + return handle_invalid_req_id(tx_ring, req_id, tx_info, false); +} + static int ena_clean_tx_irq(struct ena_ring *tx_ring, u32 budget) { struct netdev_queue *txq; @@ -765,9 +824,13 @@ static int ena_clean_tx_irq(struct ena_ring *tx_ring, u32 budget) rc = ena_com_tx_comp_req_id_get(tx_ring->ena_com_io_cq, &req_id); - if (rc) + if (rc) { + if (unlikely(rc == -EINVAL)) + handle_invalid_req_id(tx_ring, req_id, NULL, false); break; + } + /* validate that the request id points to a valid skb */ rc = validate_tx_req_id(tx_ring, req_id); if (rc) break; @@ -781,25 +844,24 @@ static int ena_clean_tx_irq(struct ena_ring *tx_ring, u32 budget) tx_info->skb = NULL; tx_info->last_jiffies = 0; - ena_unmap_tx_skb(tx_ring, tx_info); + ena_unmap_tx_buff(tx_ring, tx_info); netif_dbg(tx_ring->adapter, tx_done, tx_ring->netdev, "tx_poll: q %d skb %p completed\n", tx_ring->qid, skb); - tx_bytes += skb->len; + tx_bytes += tx_info->total_tx_size; dev_kfree_skb(skb); tx_pkts++; total_done += tx_info->tx_descs; - tx_ring->free_tx_ids[next_to_clean] = req_id; + tx_ring->free_ids[next_to_clean] = req_id; next_to_clean = ENA_TX_RING_IDX_NEXT(next_to_clean, tx_ring->ring_size); } tx_ring->next_to_clean = next_to_clean; ena_com_comp_ack(tx_ring->ena_com_io_sq, total_done); - ena_com_update_dev_comp_head(tx_ring->ena_com_io_cq); netdev_tx_completed_queue(txq, tx_pkts, tx_bytes); @@ -819,60 +881,87 @@ static int ena_clean_tx_irq(struct ena_ring *tx_ring, u32 budget) above_thresh = ena_com_sq_have_enough_space(tx_ring->ena_com_io_sq, ENA_TX_WAKEUP_THRESH); - if (netif_tx_queue_stopped(txq) && above_thresh) { + if (netif_tx_queue_stopped(txq) && above_thresh && + test_bit(ENA_FLAG_DEV_UP, &tx_ring->adapter->flags)) { netif_tx_wake_queue(txq); - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.queue_wakeup++; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&tx_ring->tx_stats.queue_wakeup, 1, + &tx_ring->syncp); } __netif_tx_unlock(txq); } - tx_ring->per_napi_bytes += tx_bytes; - tx_ring->per_napi_packets += tx_pkts; - return tx_pkts; } -static struct sk_buff *ena_alloc_skb(struct ena_ring *rx_ring, bool frags) +static struct sk_buff *ena_alloc_skb(struct ena_ring *rx_ring, void *first_frag, u16 len) { struct sk_buff *skb; - if (frags) - skb = napi_get_frags(rx_ring->napi); + if (!first_frag) + skb = napi_alloc_skb(rx_ring->napi, len); else - skb = netdev_alloc_skb_ip_align(rx_ring->netdev, - rx_ring->rx_copybreak); + skb = napi_build_skb(first_frag, len); if (unlikely(!skb)) { - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.skb_alloc_fail++; - u64_stats_update_end(&rx_ring->syncp); + ena_increase_stat(&rx_ring->rx_stats.skb_alloc_fail, 1, + &rx_ring->syncp); + netif_dbg(rx_ring->adapter, rx_err, rx_ring->netdev, - "Failed to allocate skb. frags: %d\n", frags); - return NULL; + "Failed to allocate skb. first_frag %s\n", + first_frag ? "provided" : "not provided"); } return skb; } +static bool ena_try_rx_buf_page_reuse(struct ena_rx_buffer *rx_info, u16 buf_len, + u16 len, int pkt_offset) +{ + struct ena_com_buf *ena_buf = &rx_info->ena_buf; + + /* More than ENA_MIN_RX_BUF_SIZE left in the reused buffer + * for data + headroom + tailroom. + */ + if (SKB_DATA_ALIGN(len + pkt_offset) + ENA_MIN_RX_BUF_SIZE <= ena_buf->len) { + page_ref_inc(rx_info->page); + rx_info->page_offset += buf_len; + ena_buf->paddr += buf_len; + ena_buf->len -= buf_len; + return true; + } + + return false; +} + static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring, struct ena_com_rx_buf_info *ena_bufs, u32 descs, u16 *next_to_clean) { - struct sk_buff *skb; + int tailroom = SKB_DATA_ALIGN(sizeof(struct skb_shared_info)); + bool is_xdp_loaded = ena_xdp_present_ring(rx_ring); struct ena_rx_buffer *rx_info; + struct ena_adapter *adapter; + int page_offset, pkt_offset; + dma_addr_t pre_reuse_paddr; u16 len, req_id, buf = 0; - void *va; + bool reuse_rx_buf_page; + struct sk_buff *skb; + void *buf_addr; + int buf_offset; + u16 buf_len; len = ena_bufs[buf].len; req_id = ena_bufs[buf].req_id; + rx_info = &rx_ring->rx_buffer_info[req_id]; if (unlikely(!rx_info->page)) { - netif_err(rx_ring->adapter, rx_err, rx_ring->netdev, - "Page is NULL\n"); + adapter = rx_ring->adapter; + netif_err(adapter, rx_err, rx_ring->netdev, + "Page is NULL. qid %u req_id %u\n", rx_ring->qid, req_id); + ena_increase_stat(&rx_ring->rx_stats.bad_req_id, 1, &rx_ring->syncp); + ena_reset_device(adapter, ENA_REGS_RESET_INV_RX_REQ_ID); return NULL; } @@ -880,57 +969,59 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring, "rx_info %p page %p\n", rx_info, rx_info->page); - /* save virt address of first buffer */ - va = page_address(rx_info->page) + rx_info->page_offset; - prefetch(va + NET_IP_ALIGN); + buf_offset = rx_info->buf_offset; + pkt_offset = buf_offset - rx_ring->rx_headroom; + page_offset = rx_info->page_offset; + buf_addr = page_address(rx_info->page) + page_offset; if (len <= rx_ring->rx_copybreak) { - skb = ena_alloc_skb(rx_ring, false); + skb = ena_alloc_skb(rx_ring, NULL, len); if (unlikely(!skb)) return NULL; - netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev, - "rx allocated small packet. len %d. data_len %d\n", - skb->len, skb->data_len); - - /* sync this buffer for CPU use */ - dma_sync_single_for_cpu(rx_ring->dev, - dma_unmap_addr(&rx_info->ena_buf, paddr), - len, - DMA_FROM_DEVICE); - skb_copy_to_linear_data(skb, va, len); + skb_copy_to_linear_data(skb, buf_addr + buf_offset, len); dma_sync_single_for_device(rx_ring->dev, - dma_unmap_addr(&rx_info->ena_buf, paddr), + dma_unmap_addr(&rx_info->ena_buf, paddr) + pkt_offset, len, DMA_FROM_DEVICE); skb_put(skb, len); + netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev, + "RX allocated small packet. len %d.\n", skb->len); skb->protocol = eth_type_trans(skb, rx_ring->netdev); - rx_ring->free_rx_ids[*next_to_clean] = req_id; + rx_ring->free_ids[*next_to_clean] = req_id; *next_to_clean = ENA_RX_RING_IDX_ADD(*next_to_clean, descs, rx_ring->ring_size); return skb; } - skb = ena_alloc_skb(rx_ring, true); + buf_len = SKB_DATA_ALIGN(len + buf_offset + tailroom); + + /* If XDP isn't loaded try to reuse part of the RX buffer */ + reuse_rx_buf_page = !is_xdp_loaded && + ena_try_rx_buf_page_reuse(rx_info, buf_len, len, pkt_offset); + + if (!reuse_rx_buf_page) + ena_unmap_rx_buff_attrs(rx_ring, rx_info, DMA_ATTR_SKIP_CPU_SYNC); + + skb = ena_alloc_skb(rx_ring, buf_addr, buf_len); if (unlikely(!skb)) return NULL; - do { - dma_unmap_page(rx_ring->dev, - dma_unmap_addr(&rx_info->ena_buf, paddr), - ENA_PAGE_SIZE, DMA_FROM_DEVICE); - - skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_info->page, - rx_info->page_offset, len, ENA_PAGE_SIZE); + /* Populate skb's linear part */ + skb_reserve(skb, buf_offset); + skb_put(skb, len); + skb->protocol = eth_type_trans(skb, rx_ring->netdev); + do { netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev, - "rx skb updated. len %d. data_len %d\n", + "RX skb updated. len %d. data_len %d\n", skb->len, skb->data_len); - rx_info->page = NULL; + if (!reuse_rx_buf_page) + rx_info->page = NULL; - rx_ring->free_rx_ids[*next_to_clean] = req_id; + rx_ring->free_ids[*next_to_clean] = req_id; *next_to_clean = ENA_RX_RING_IDX_NEXT(*next_to_clean, rx_ring->ring_size); @@ -940,7 +1031,31 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring, buf++; len = ena_bufs[buf].len; req_id = ena_bufs[buf].req_id; + rx_info = &rx_ring->rx_buffer_info[req_id]; + + /* rx_info->buf_offset includes rx_ring->rx_headroom */ + buf_offset = rx_info->buf_offset; + pkt_offset = buf_offset - rx_ring->rx_headroom; + buf_len = SKB_DATA_ALIGN(len + buf_offset + tailroom); + page_offset = rx_info->page_offset; + + pre_reuse_paddr = dma_unmap_addr(&rx_info->ena_buf, paddr); + + reuse_rx_buf_page = !is_xdp_loaded && + ena_try_rx_buf_page_reuse(rx_info, buf_len, len, pkt_offset); + + dma_sync_single_for_cpu(rx_ring->dev, + pre_reuse_paddr + pkt_offset, + len, + DMA_FROM_DEVICE); + + if (!reuse_rx_buf_page) + ena_unmap_rx_buff_attrs(rx_ring, rx_info, DMA_ATTR_SKIP_CPU_SYNC); + + skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_info->page, + page_offset + buf_offset, len, buf_len); + } while (1); return skb; @@ -951,7 +1066,7 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring, * @ena_rx_ctx: received packet context/metadata * @skb: skb currently being received and modified */ -static inline void ena_rx_checksum(struct ena_ring *rx_ring, +static void ena_rx_checksum(struct ena_ring *rx_ring, struct ena_com_rx_ctx *ena_rx_ctx, struct sk_buff *skb) { @@ -972,9 +1087,8 @@ static inline void ena_rx_checksum(struct ena_ring *rx_ring, (ena_rx_ctx->l3_csum_err))) { /* ipv4 checksum error */ skb->ip_summed = CHECKSUM_NONE; - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.bad_csum++; - u64_stats_update_end(&rx_ring->syncp); + ena_increase_stat(&rx_ring->rx_stats.csum_bad, 1, + &rx_ring->syncp); netif_dbg(rx_ring->adapter, rx_err, rx_ring->netdev, "RX IPv4 header checksum error\n"); return; @@ -985,9 +1099,8 @@ static inline void ena_rx_checksum(struct ena_ring *rx_ring, (ena_rx_ctx->l4_proto == ENA_ETH_IO_L4_PROTO_UDP))) { if (unlikely(ena_rx_ctx->l4_csum_err)) { /* TCP/UDP checksum error */ - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.bad_csum++; - u64_stats_update_end(&rx_ring->syncp); + ena_increase_stat(&rx_ring->rx_stats.csum_bad, 1, + &rx_ring->syncp); netif_dbg(rx_ring->adapter, rx_err, rx_ring->netdev, "RX L4 checksum error\n"); skb->ip_summed = CHECKSUM_NONE; @@ -996,10 +1109,11 @@ static inline void ena_rx_checksum(struct ena_ring *rx_ring, if (likely(ena_rx_ctx->l4_csum_checked)) { skb->ip_summed = CHECKSUM_UNNECESSARY; + ena_increase_stat(&rx_ring->rx_stats.csum_good, 1, + &rx_ring->syncp); } else { - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.csum_unchecked++; - u64_stats_update_end(&rx_ring->syncp); + ena_increase_stat(&rx_ring->rx_stats.csum_unchecked, 1, + &rx_ring->syncp); skb->ip_summed = CHECKSUM_NONE; } } else { @@ -1031,6 +1145,35 @@ static void ena_set_rx_hash(struct ena_ring *rx_ring, } } +static int ena_xdp_handle_buff(struct ena_ring *rx_ring, struct xdp_buff *xdp, u16 num_descs) +{ + struct ena_rx_buffer *rx_info; + int ret; + + /* XDP multi-buffer packets not supported */ + if (unlikely(num_descs > 1)) { + netdev_err_once(rx_ring->adapter->netdev, + "xdp: dropped unsupported multi-buffer packets\n"); + ena_increase_stat(&rx_ring->rx_stats.xdp_drop, 1, &rx_ring->syncp); + return ENA_XDP_DROP; + } + + rx_info = &rx_ring->rx_buffer_info[rx_ring->ena_bufs[0].req_id]; + xdp_prepare_buff(xdp, page_address(rx_info->page), + rx_info->buf_offset, + rx_ring->ena_bufs[0].len, false); + + ret = ena_xdp_execute(rx_ring, xdp); + + /* The xdp program might expand the headers */ + if (ret == ENA_XDP_PASS) { + rx_info->buf_offset = xdp->data - xdp->data_hard_start; + rx_ring->ena_bufs[0].len = xdp->data_end - xdp->data; + } + + return ret; +} + /* ena_clean_rx_irq - Cleanup RX irq * @rx_ring: RX ring to clean * @napi: napi handler @@ -1042,26 +1185,34 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi, u32 budget) { u16 next_to_clean = rx_ring->next_to_clean; - u32 res_budget, work_done; - struct ena_com_rx_ctx ena_rx_ctx; + struct ena_rx_buffer *rx_info; struct ena_adapter *adapter; + u32 res_budget, work_done; + int rx_copybreak_pkt = 0; + int refill_threshold; struct sk_buff *skb; int refill_required; - int refill_threshold; - int rc = 0; + struct xdp_buff xdp; + int xdp_flags = 0; int total_len = 0; - int rx_copybreak_pkt = 0; + int xdp_verdict; + u8 pkt_offset; + int rc = 0; int i; netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev, "%s qid %d\n", __func__, rx_ring->qid); res_budget = budget; + xdp_init_buff(&xdp, ENA_PAGE_SIZE, &rx_ring->xdp_rxq); do { + xdp_verdict = ENA_XDP_PASS; + skb = NULL; ena_rx_ctx.ena_bufs = rx_ring->ena_bufs; ena_rx_ctx.max_bufs = rx_ring->sgl_size; ena_rx_ctx.descs = 0; + ena_rx_ctx.pkt_offset = 0; rc = ena_com_rx_pkt(rx_ring->ena_com_io_cq, rx_ring->ena_com_io_sq, &ena_rx_ctx); @@ -1071,23 +1222,55 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi, if (unlikely(ena_rx_ctx.descs == 0)) break; + /* First descriptor might have an offset set by the device */ + rx_info = &rx_ring->rx_buffer_info[rx_ring->ena_bufs[0].req_id]; + pkt_offset = ena_rx_ctx.pkt_offset; + rx_info->buf_offset += pkt_offset; + netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev, "rx_poll: q %d got packet from ena. descs #: %d l3 proto %d l4 proto %d hash: %x\n", rx_ring->qid, ena_rx_ctx.descs, ena_rx_ctx.l3_proto, ena_rx_ctx.l4_proto, ena_rx_ctx.hash); + dma_sync_single_for_cpu(rx_ring->dev, + dma_unmap_addr(&rx_info->ena_buf, paddr) + pkt_offset, + rx_ring->ena_bufs[0].len, + DMA_FROM_DEVICE); + + if (ena_xdp_present_ring(rx_ring)) + xdp_verdict = ena_xdp_handle_buff(rx_ring, &xdp, ena_rx_ctx.descs); + /* allocate skb and fill it */ - skb = ena_rx_skb(rx_ring, rx_ring->ena_bufs, ena_rx_ctx.descs, - &next_to_clean); + if (xdp_verdict == ENA_XDP_PASS) + skb = ena_rx_skb(rx_ring, + rx_ring->ena_bufs, + ena_rx_ctx.descs, + &next_to_clean); - /* exit if we failed to retrieve a buffer */ if (unlikely(!skb)) { for (i = 0; i < ena_rx_ctx.descs; i++) { - rx_ring->free_tx_ids[next_to_clean] = - rx_ring->ena_bufs[i].req_id; + int req_id = rx_ring->ena_bufs[i].req_id; + + rx_ring->free_ids[next_to_clean] = req_id; next_to_clean = ENA_RX_RING_IDX_NEXT(next_to_clean, rx_ring->ring_size); + + /* Packets was passed for transmission, unmap it + * from RX side. + */ + if (xdp_verdict & ENA_XDP_FORWARDED) { + ena_unmap_rx_buff_attrs(rx_ring, + &rx_ring->rx_buffer_info[req_id], + DMA_ATTR_SKIP_CPU_SYNC); + rx_ring->rx_buffer_info[req_id].page = NULL; + } + } + if (xdp_verdict != ENA_XDP_PASS) { + xdp_flags |= xdp_verdict; + total_len += ena_rx_ctx.ena_bufs[0].len; + res_budget--; + continue; } break; } @@ -1098,20 +1281,17 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi, skb_record_rx_queue(skb, rx_ring->qid); - if (rx_ring->ena_bufs[0].len <= rx_ring->rx_copybreak) { - total_len += rx_ring->ena_bufs[0].len; + if (rx_ring->ena_bufs[0].len <= rx_ring->rx_copybreak) rx_copybreak_pkt++; - napi_gro_receive(napi, skb); - } else { - total_len += skb->len; - napi_gro_frags(napi); - } + + total_len += skb->len; + + napi_gro_receive(napi, skb); res_budget--; } while (likely(res_budget)); work_done = budget - res_budget; - rx_ring->per_napi_bytes += total_len; rx_ring->per_napi_packets += work_done; u64_stats_update_begin(&rx_ring->syncp); rx_ring->rx_stats.bytes += total_len; @@ -1121,74 +1301,105 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi, rx_ring->next_to_clean = next_to_clean; - refill_required = ena_com_free_desc(rx_ring->ena_com_io_sq); + refill_required = ena_com_free_q_entries(rx_ring->ena_com_io_sq); refill_threshold = min_t(int, rx_ring->ring_size / ENA_RX_REFILL_THRESH_DIVIDER, ENA_RX_REFILL_THRESH_PACKET); /* Optimization, try to batch new rx buffers */ - if (refill_required > refill_threshold) { - ena_com_update_dev_comp_head(rx_ring->ena_com_io_cq); + if (refill_required > refill_threshold) ena_refill_rx_bufs(rx_ring, refill_required); - } + + if (xdp_flags & ENA_XDP_REDIRECT) + xdp_do_flush(); return work_done; error: + if (xdp_flags & ENA_XDP_REDIRECT) + xdp_do_flush(); + adapter = netdev_priv(rx_ring->netdev); - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.bad_desc_num++; - u64_stats_update_end(&rx_ring->syncp); + if (rc == -ENOSPC) { + ena_increase_stat(&rx_ring->rx_stats.bad_desc_num, 1, &rx_ring->syncp); + ena_reset_device(adapter, ENA_REGS_RESET_TOO_MANY_RX_DESCS); + } else if (rc == -EFAULT) { + ena_reset_device(adapter, ENA_REGS_RESET_RX_DESCRIPTOR_MALFORMED); + } else { + ena_increase_stat(&rx_ring->rx_stats.bad_req_id, 1, + &rx_ring->syncp); + ena_reset_device(adapter, ENA_REGS_RESET_INV_RX_REQ_ID); + } + return 0; +} - /* Too many desc from the device. Trigger reset */ - adapter->reset_reason = ENA_REGS_RESET_TOO_MANY_RX_DESCS; - set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); +static void ena_dim_work(struct work_struct *w) +{ + struct dim *dim = container_of(w, struct dim, work); + struct dim_cq_moder cur_moder = + net_dim_get_rx_moderation(dim->mode, dim->profile_ix); + struct ena_napi *ena_napi = container_of(dim, struct ena_napi, dim); - return 0; + ena_napi->rx_ring->smoothed_interval = cur_moder.usec; + dim->state = DIM_START_MEASURE; } -inline void ena_adjust_intr_moderation(struct ena_ring *rx_ring, - struct ena_ring *tx_ring) +static void ena_adjust_adaptive_rx_intr_moderation(struct ena_napi *ena_napi) { - /* We apply adaptive moderation on Rx path only. - * Tx uses static interrupt moderation. - */ - ena_com_calculate_interrupt_delay(rx_ring->ena_dev, - rx_ring->per_napi_packets, - rx_ring->per_napi_bytes, - &rx_ring->smoothed_interval, - &rx_ring->moder_tbl_idx); - - /* Reset per napi packets/bytes */ - tx_ring->per_napi_packets = 0; - tx_ring->per_napi_bytes = 0; + struct dim_sample dim_sample; + struct ena_ring *rx_ring = ena_napi->rx_ring; + + if (!rx_ring->per_napi_packets) + return; + + rx_ring->non_empty_napi_events++; + + dim_update_sample(rx_ring->non_empty_napi_events, + rx_ring->rx_stats.cnt, + rx_ring->rx_stats.bytes, + &dim_sample); + + net_dim(&ena_napi->dim, &dim_sample); + rx_ring->per_napi_packets = 0; - rx_ring->per_napi_bytes = 0; } -static inline void ena_unmask_interrupt(struct ena_ring *tx_ring, - struct ena_ring *rx_ring) +void ena_unmask_interrupt(struct ena_ring *tx_ring, + struct ena_ring *rx_ring) { + u32 rx_interval = tx_ring->smoothed_interval; struct ena_eth_io_intr_reg intr_reg; + /* Rx ring can be NULL when for XDP tx queues which don't have an + * accompanying rx_ring pair. + */ + if (rx_ring) + rx_interval = ena_com_get_adaptive_moderation_enabled(rx_ring->ena_dev) ? + rx_ring->smoothed_interval : + ena_com_get_nonadaptive_moderation_interval_rx(rx_ring->ena_dev); + /* Update intr register: rx intr delay, * tx intr delay and interrupt unmask */ ena_com_update_intr_reg(&intr_reg, - rx_ring->smoothed_interval, + rx_interval, tx_ring->smoothed_interval, true); + ena_increase_stat(&tx_ring->tx_stats.unmask_interrupt, 1, + &tx_ring->syncp); + /* It is a shared MSI-X. * Tx and Rx CQ have pointer to it. * So we use one of them to reach the intr reg + * The Tx ring is used because the rx_ring is NULL for XDP queues */ - ena_com_unmask_intr(rx_ring->ena_com_io_cq, &intr_reg); + ena_com_unmask_intr(tx_ring->ena_com_io_cq, &intr_reg); } -static inline void ena_update_ring_numa_node(struct ena_ring *tx_ring, - struct ena_ring *rx_ring) +void ena_update_ring_numa_node(struct ena_ring *tx_ring, + struct ena_ring *rx_ring) { int cpu = get_cpu(); int numa_node; @@ -1197,17 +1408,27 @@ static inline void ena_update_ring_numa_node(struct ena_ring *tx_ring, if (likely(tx_ring->cpu == cpu)) goto out; + tx_ring->cpu = cpu; + if (rx_ring) + rx_ring->cpu = cpu; + numa_node = cpu_to_node(cpu); + + if (likely(tx_ring->numa_node == numa_node)) + goto out; + put_cpu(); if (numa_node != NUMA_NO_NODE) { ena_com_update_numa_node(tx_ring->ena_com_io_cq, numa_node); - ena_com_update_numa_node(rx_ring->ena_com_io_cq, numa_node); + tx_ring->numa_node = numa_node; + if (rx_ring) { + rx_ring->numa_node = numa_node; + ena_com_update_numa_node(rx_ring->ena_com_io_cq, + numa_node); + } } - tx_ring->cpu = cpu; - rx_ring->cpu = cpu; - return; out: put_cpu(); @@ -1217,9 +1438,8 @@ static int ena_io_poll(struct napi_struct *napi, int budget) { struct ena_napi *ena_napi = container_of(napi, struct ena_napi, napi); struct ena_ring *tx_ring, *rx_ring; - - u32 tx_work_done; - u32 rx_work_done; + int tx_work_done; + int rx_work_done = 0; int tx_budget; int napi_comp_call = 0; int ret; @@ -1236,7 +1456,11 @@ static int ena_io_poll(struct napi_struct *napi, int budget) } tx_work_done = ena_clean_tx_irq(tx_ring, tx_budget); - rx_work_done = ena_clean_rx_irq(rx_ring, napi, budget); + /* On netpoll the budget is zero and the handler should only clean the + * tx completions. + */ + if (likely(budget)) + rx_work_done = ena_clean_rx_irq(rx_ring, napi, budget); /* If the device is about to reset or down, avoid unmask * the interrupt and return 0 so NAPI won't reschedule @@ -1252,16 +1476,20 @@ static int ena_io_poll(struct napi_struct *napi, int budget) /* Update numa and unmask the interrupt only when schedule * from the interrupt context (vs from sk_busy_loop) */ - if (napi_complete_done(napi, rx_work_done)) { - /* Tx and Rx share the same interrupt vector */ + if (napi_complete_done(napi, rx_work_done) && + READ_ONCE(ena_napi->interrupts_masked)) { + smp_rmb(); /* make sure interrupts_masked is read */ + WRITE_ONCE(ena_napi->interrupts_masked, false); + /* We apply adaptive moderation on Rx path only. + * Tx uses static interrupt moderation. + */ if (ena_com_get_adaptive_moderation_enabled(rx_ring->ena_dev)) - ena_adjust_intr_moderation(rx_ring, tx_ring); + ena_adjust_adaptive_rx_intr_moderation(ena_napi); + ena_update_ring_numa_node(tx_ring, rx_ring); ena_unmask_interrupt(tx_ring, rx_ring); } - ena_update_ring_numa_node(tx_ring, rx_ring); - ret = rx_work_done; } else { ret = budget; @@ -1272,6 +1500,8 @@ static int ena_io_poll(struct napi_struct *napi, int budget) tx_ring->tx_stats.tx_poll++; u64_stats_update_end(&tx_ring->syncp); + tx_ring->tx_stats.last_napi_jiffies = jiffies; + return ret; } @@ -1296,8 +1526,11 @@ static irqreturn_t ena_intr_msix_io(int irq, void *data) { struct ena_napi *ena_napi = data; - ena_napi->tx_ring->first_interrupt = true; - ena_napi->rx_ring->first_interrupt = true; + /* Used to check HW health */ + WRITE_ONCE(ena_napi->first_interrupt, true); + + WRITE_ONCE(ena_napi->interrupts_masked, true); + smp_wmb(); /* write interrupts_masked before calling napi */ napi_schedule_irqoff(&ena_napi->napi); @@ -1309,7 +1542,7 @@ static irqreturn_t ena_intr_msix_io(int irq, void *data) * the number of potential io queues is the minimum of what the device * supports and the number of vCPUs. */ -static int ena_enable_msix(struct ena_adapter *adapter, int num_queues) +static int ena_enable_msix(struct ena_adapter *adapter) { int msix_vecs, irq_cnt; @@ -1320,9 +1553,9 @@ static int ena_enable_msix(struct ena_adapter *adapter, int num_queues) } /* Reserved the max msix vectors we might need */ - msix_vecs = ENA_MAX_MSIX_VEC(num_queues); + msix_vecs = ENA_MAX_MSIX_VEC(adapter->max_num_io_queues); netif_dbg(adapter, probe, adapter->netdev, - "trying to enable MSI-X, vectors %d\n", msix_vecs); + "Trying to enable MSI-X, vectors %d\n", msix_vecs); irq_cnt = pci_alloc_irq_vectors(adapter->pdev, ENA_MIN_MSIX_VEC, msix_vecs, PCI_IRQ_MSIX); @@ -1335,12 +1568,12 @@ static int ena_enable_msix(struct ena_adapter *adapter, int num_queues) if (irq_cnt != msix_vecs) { netif_notice(adapter, probe, adapter->netdev, - "enable only %d MSI-X (out of %d), reduce the number of queues\n", + "Enable only %d MSI-X (out of %d), reduce the number of queues\n", irq_cnt, msix_vecs); - adapter->num_queues = irq_cnt - ENA_ADMIN_MSIX_VEC; + adapter->num_io_queues = irq_cnt - ENA_ADMIN_MSIX_VEC; } - if (ena_init_rx_cpu_rmap(adapter)) + if (netif_enable_cpu_rmap(adapter->netdev, adapter->num_io_queues)) netif_warn(adapter, probe, adapter->netdev, "Failed to map IRQs to CPUs\n"); @@ -1372,10 +1605,12 @@ static void ena_setup_io_intr(struct ena_adapter *adapter) { struct net_device *netdev; int irq_idx, i, cpu; + int io_queue_count; netdev = adapter->netdev; + io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < io_queue_count; i++) { irq_idx = ENA_IO_IRQ_IDX(i); cpu = i % num_online_cpus(); @@ -1403,12 +1638,12 @@ static int ena_request_mgmnt_irq(struct ena_adapter *adapter) irq->data); if (rc) { netif_err(adapter, probe, adapter->netdev, - "failed to request admin irq\n"); + "Failed to request admin irq\n"); return rc; } netif_dbg(adapter, probe, adapter->netdev, - "set affinity hint of mgmnt irq.to 0x%lx (irq vector: %d)\n", + "Set affinity hint of mgmnt irq.to 0x%lx (irq vector: %d)\n", irq->affinity_hint_mask.bits[0], irq->vector); irq_set_affinity_hint(irq->vector, &irq->affinity_hint_mask); @@ -1418,9 +1653,10 @@ static int ena_request_mgmnt_irq(struct ena_adapter *adapter) static int ena_request_io_irq(struct ena_adapter *adapter) { + u32 io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues; + int rc = 0, i, k, irq_idx; unsigned long flags = 0; struct ena_irq *irq; - int rc = 0, i, k; if (!test_bit(ENA_FLAG_MSIX_ENABLED, &adapter->flags)) { netif_err(adapter, ifup, adapter->netdev, @@ -1428,7 +1664,7 @@ static int ena_request_io_irq(struct ena_adapter *adapter) return -EINVAL; } - for (i = ENA_IO_IRQ_FIRST_IDX; i < adapter->msix_vecs; i++) { + for (i = ENA_IO_IRQ_FIRST_IDX; i < ENA_MAX_MSIX_VEC(io_queue_count); i++) { irq = &adapter->irq_tbl[i]; rc = request_irq(irq->vector, irq->handler, flags, irq->name, irq->data); @@ -1440,12 +1676,22 @@ static int ena_request_io_irq(struct ena_adapter *adapter) } netif_dbg(adapter, ifup, adapter->netdev, - "set affinity hint of irq. index %d to 0x%lx (irq vector: %d)\n", + "Set affinity hint of irq. index %d to 0x%lx (irq vector: %d)\n", i, irq->affinity_hint_mask.bits[0], irq->vector); irq_set_affinity_hint(irq->vector, &irq->affinity_hint_mask); } + /* Now that IO IRQs have been successfully allocated map them to the + * corresponding IO NAPI instance. Note that the mgmnt IRQ does not + * have a NAPI, so care must be taken to correctly map IRQs to NAPIs. + */ + for (i = 0; i < io_queue_count; i++) { + irq_idx = ENA_IO_IRQ_IDX(i); + irq = &adapter->irq_tbl[irq_idx]; + netif_napi_set_irq(&adapter->ena_napi[i].napi, irq->vector); + } + return rc; err: @@ -1469,19 +1715,17 @@ static void ena_free_mgmnt_irq(struct ena_adapter *adapter) static void ena_free_io_irq(struct ena_adapter *adapter) { + u32 io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues; struct ena_irq *irq; int i; -#ifdef CONFIG_RFS_ACCEL - if (adapter->msix_vecs >= 1) { - free_irq_cpu_rmap(adapter->netdev->rx_cpu_rmap); - adapter->netdev->rx_cpu_rmap = NULL; - } -#endif /* CONFIG_RFS_ACCEL */ + for (i = ENA_IO_IRQ_FIRST_IDX; i < ENA_MAX_MSIX_VEC(io_queue_count); i++) { + struct ena_napi *ena_napi; - for (i = ENA_IO_IRQ_FIRST_IDX; i < adapter->msix_vecs; i++) { irq = &adapter->irq_tbl[i]; irq_set_affinity_hint(irq->vector, NULL); + ena_napi = irq->data; + netif_napi_set_irq(&ena_napi->napi, -1); free_irq(irq->vector, irq->data); } } @@ -1494,63 +1738,97 @@ static void ena_disable_msix(struct ena_adapter *adapter) static void ena_disable_io_intr_sync(struct ena_adapter *adapter) { + u32 io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues; int i; if (!netif_running(adapter->netdev)) return; - for (i = ENA_IO_IRQ_FIRST_IDX; i < adapter->msix_vecs; i++) + for (i = ENA_IO_IRQ_FIRST_IDX; i < ENA_MAX_MSIX_VEC(io_queue_count); i++) synchronize_irq(adapter->irq_tbl[i].vector); } -static void ena_del_napi(struct ena_adapter *adapter) +static void ena_del_napi_in_range(struct ena_adapter *adapter, + int first_index, + int count) { int i; - for (i = 0; i < adapter->num_queues; i++) + for (i = first_index; i < first_index + count; i++) { netif_napi_del(&adapter->ena_napi[i].napi); + + WARN_ON(ENA_IS_XDP_INDEX(adapter, i) && + adapter->ena_napi[i].rx_ring); + } } -static void ena_init_napi(struct ena_adapter *adapter) +static void ena_init_napi_in_range(struct ena_adapter *adapter, + int first_index, int count) { - struct ena_napi *napi; + int (*napi_handler)(struct napi_struct *napi, int budget); int i; - for (i = 0; i < adapter->num_queues; i++) { - napi = &adapter->ena_napi[i]; + for (i = first_index; i < first_index + count; i++) { + struct ena_napi *napi = &adapter->ena_napi[i]; + struct ena_ring *rx_ring, *tx_ring; - netif_napi_add(adapter->netdev, - &adapter->ena_napi[i].napi, - ena_io_poll, - ENA_NAPI_BUDGET); - napi->rx_ring = &adapter->rx_ring[i]; - napi->tx_ring = &adapter->tx_ring[i]; + memset(napi, 0, sizeof(*napi)); + + rx_ring = &adapter->rx_ring[i]; + tx_ring = &adapter->tx_ring[i]; + + napi_handler = ena_io_poll; + if (ENA_IS_XDP_INDEX(adapter, i)) + napi_handler = ena_xdp_io_poll; + + netif_napi_add_config(adapter->netdev, &napi->napi, napi_handler, i); + + if (!ENA_IS_XDP_INDEX(adapter, i)) + napi->rx_ring = rx_ring; + + napi->tx_ring = tx_ring; napi->qid = i; } } -static void ena_napi_disable_all(struct ena_adapter *adapter) +static void ena_napi_disable_in_range(struct ena_adapter *adapter, + int first_index, + int count) { + struct napi_struct *napi; int i; - for (i = 0; i < adapter->num_queues; i++) - napi_disable(&adapter->ena_napi[i].napi); + for (i = first_index; i < first_index + count; i++) { + napi = &adapter->ena_napi[i].napi; + if (!ENA_IS_XDP_INDEX(adapter, i)) { + /* This API is supported for non-XDP queues only */ + netif_queue_set_napi(adapter->netdev, i, + NETDEV_QUEUE_TYPE_TX, NULL); + netif_queue_set_napi(adapter->netdev, i, + NETDEV_QUEUE_TYPE_RX, NULL); + } + napi_disable(napi); + } } -static void ena_napi_enable_all(struct ena_adapter *adapter) +static void ena_napi_enable_in_range(struct ena_adapter *adapter, + int first_index, + int count) { + struct napi_struct *napi; int i; - for (i = 0; i < adapter->num_queues; i++) - napi_enable(&adapter->ena_napi[i].napi); -} - -static void ena_restore_ethtool_params(struct ena_adapter *adapter) -{ - adapter->tx_usecs = 0; - adapter->rx_usecs = 0; - adapter->tx_frames = 1; - adapter->rx_frames = 1; + for (i = first_index; i < first_index + count; i++) { + napi = &adapter->ena_napi[i].napi; + napi_enable(napi); + if (!ENA_IS_XDP_INDEX(adapter, i)) { + /* This API is supported for non-XDP queues only */ + netif_queue_set_napi(adapter->netdev, i, + NETDEV_QUEUE_TYPE_RX, napi); + netif_queue_set_napi(adapter->netdev, i, + NETDEV_QUEUE_TYPE_TX, napi); + } + } } /* Configure the Rx forwarding */ @@ -1563,8 +1841,7 @@ static int ena_rss_configure(struct ena_adapter *adapter) if (!ena_dev->rss.tbl_log_size) { rc = ena_rss_init_default(adapter); if (rc && (rc != -EOPNOTSUPP)) { - netif_err(adapter, ifup, adapter->netdev, - "Failed to init RSS rc: %d\n", rc); + netif_err(adapter, ifup, adapter->netdev, "Failed to init RSS rc: %d\n", rc); return rc; } } @@ -1602,9 +1879,9 @@ static int ena_up_complete(struct ena_adapter *adapter) /* enable transmits */ netif_tx_start_all_queues(adapter->netdev); - ena_restore_ethtool_params(adapter); - - ena_napi_enable_all(adapter); + ena_napi_enable_in_range(adapter, + 0, + adapter->xdp_num_queues + adapter->num_io_queues); return 0; } @@ -1630,8 +1907,8 @@ static int ena_create_io_tx_queue(struct ena_adapter *adapter, int qid) ctx.qid = ena_qid; ctx.mem_queue_type = ena_dev->tx_mem_queue_type; ctx.msix_vector = msix_vector; - ctx.queue_size = adapter->tx_ring_size; - ctx.numa_node = cpu_to_node(tx_ring->cpu); + ctx.queue_size = tx_ring->ring_size; + ctx.numa_node = tx_ring->numa_node; rc = ena_com_create_io_queue(ena_dev, &ctx); if (rc) { @@ -1656,12 +1933,13 @@ static int ena_create_io_tx_queue(struct ena_adapter *adapter, int qid) return rc; } -static int ena_create_all_io_tx_queues(struct ena_adapter *adapter) +int ena_create_io_tx_queues_in_range(struct ena_adapter *adapter, + int first_index, int count) { struct ena_com_dev *ena_dev = adapter->ena_dev; int rc, i; - for (i = 0; i < adapter->num_queues; i++) { + for (i = first_index; i < first_index + count; i++) { rc = ena_create_io_tx_queue(adapter, i); if (rc) goto create_err; @@ -1670,7 +1948,7 @@ static int ena_create_all_io_tx_queues(struct ena_adapter *adapter) return 0; create_err: - while (i--) + while (i-- > first_index) ena_com_destroy_io_queue(ena_dev, ENA_IO_TXQ_IDX(i)); return rc; @@ -1697,8 +1975,8 @@ static int ena_create_io_rx_queue(struct ena_adapter *adapter, int qid) ctx.direction = ENA_COM_IO_QUEUE_DIRECTION_RX; ctx.mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; ctx.msix_vector = msix_vector; - ctx.queue_size = adapter->rx_ring_size; - ctx.numa_node = cpu_to_node(rx_ring->cpu); + ctx.queue_size = rx_ring->ring_size; + ctx.numa_node = rx_ring->numa_node; rc = ena_com_create_io_queue(ena_dev, &ctx); if (rc) { @@ -1715,13 +1993,15 @@ static int ena_create_io_rx_queue(struct ena_adapter *adapter, int qid) netif_err(adapter, ifup, adapter->netdev, "Failed to get RX queue handlers. RX queue num %d rc: %d\n", qid, rc); - ena_com_destroy_io_queue(ena_dev, ena_qid); - return rc; + goto err; } ena_com_update_numa_node(rx_ring->ena_com_io_cq, ctx.numa_node); return rc; +err: + ena_com_destroy_io_queue(ena_dev, ena_qid); + return rc; } static int ena_create_all_io_rx_queues(struct ena_adapter *adapter) @@ -1729,27 +2009,151 @@ static int ena_create_all_io_rx_queues(struct ena_adapter *adapter) struct ena_com_dev *ena_dev = adapter->ena_dev; int rc, i; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < adapter->num_io_queues; i++) { rc = ena_create_io_rx_queue(adapter, i); if (rc) goto create_err; + INIT_WORK(&adapter->ena_napi[i].dim.work, ena_dim_work); + + ena_xdp_register_rxq_info(&adapter->rx_ring[i]); } return 0; create_err: - while (i--) + while (i--) { + ena_xdp_unregister_rxq_info(&adapter->rx_ring[i]); + cancel_work_sync(&adapter->ena_napi[i].dim.work); ena_com_destroy_io_queue(ena_dev, ENA_IO_RXQ_IDX(i)); + } return rc; } -static int ena_up(struct ena_adapter *adapter) +static void set_io_rings_size(struct ena_adapter *adapter, + int new_tx_size, + int new_rx_size) { - int rc, i; + int i; + + for (i = 0; i < adapter->num_io_queues; i++) { + adapter->tx_ring[i].ring_size = new_tx_size; + adapter->rx_ring[i].ring_size = new_rx_size; + } +} + +/* This function allows queue allocation to backoff when the system is + * low on memory. If there is not enough memory to allocate io queues + * the driver will try to allocate smaller queues. + * + * The backoff algorithm is as follows: + * 1. Try to allocate TX and RX and if successful. + * 1.1. return success + * + * 2. Divide by 2 the size of the larger of RX and TX queues (or both if their size is the same). + * + * 3. If TX or RX is smaller than 256 + * 3.1. return failure. + * 4. else + * 4.1. go back to 1. + */ +static int create_queues_with_size_backoff(struct ena_adapter *adapter) +{ + int rc, cur_rx_ring_size, cur_tx_ring_size; + int new_rx_ring_size, new_tx_ring_size; - netdev_dbg(adapter->netdev, "%s\n", __func__); + /* current queue sizes might be set to smaller than the requested + * ones due to past queue allocation failures. + */ + set_io_rings_size(adapter, adapter->requested_tx_ring_size, + adapter->requested_rx_ring_size); + + while (1) { + if (ena_xdp_present(adapter)) { + rc = ena_setup_and_create_all_xdp_queues(adapter); + + if (rc) + goto err_setup_tx; + } + rc = ena_setup_tx_resources_in_range(adapter, + 0, + adapter->num_io_queues); + if (rc) + goto err_setup_tx; + + rc = ena_create_io_tx_queues_in_range(adapter, + 0, + adapter->num_io_queues); + if (rc) + goto err_create_tx_queues; + + rc = ena_setup_all_rx_resources(adapter); + if (rc) + goto err_setup_rx; + rc = ena_create_all_io_rx_queues(adapter); + if (rc) + goto err_create_rx_queues; + + return 0; + +err_create_rx_queues: + ena_free_all_io_rx_resources(adapter); +err_setup_rx: + ena_destroy_all_tx_queues(adapter); +err_create_tx_queues: + ena_free_all_io_tx_resources(adapter); +err_setup_tx: + if (rc != -ENOMEM) { + netif_err(adapter, ifup, adapter->netdev, + "Queue creation failed with error code %d\n", + rc); + return rc; + } + + cur_tx_ring_size = adapter->tx_ring[0].ring_size; + cur_rx_ring_size = adapter->rx_ring[0].ring_size; + + netif_err(adapter, ifup, adapter->netdev, + "Not enough memory to create queues with sizes TX=%d, RX=%d\n", + cur_tx_ring_size, cur_rx_ring_size); + + new_tx_ring_size = cur_tx_ring_size; + new_rx_ring_size = cur_rx_ring_size; + + /* Decrease the size of the larger queue, or + * decrease both if they are the same size. + */ + if (cur_rx_ring_size <= cur_tx_ring_size) + new_tx_ring_size = cur_tx_ring_size / 2; + if (cur_rx_ring_size >= cur_tx_ring_size) + new_rx_ring_size = cur_rx_ring_size / 2; + + if (new_tx_ring_size < ENA_MIN_RING_SIZE || + new_rx_ring_size < ENA_MIN_RING_SIZE) { + netif_err(adapter, ifup, adapter->netdev, + "Queue creation failed with the smallest possible queue size of %d for both queues. Not retrying with smaller queues\n", + ENA_MIN_RING_SIZE); + return rc; + } + + netif_err(adapter, ifup, adapter->netdev, + "Retrying queue creation with sizes TX=%d, RX=%d\n", + new_tx_ring_size, + new_rx_ring_size); + + set_io_rings_size(adapter, new_tx_ring_size, + new_rx_ring_size); + } +} + +int ena_up(struct ena_adapter *adapter) +{ + int io_queue_count, rc, i; + + netif_dbg(adapter, ifup, adapter->netdev, "%s\n", __func__); + + io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues; ena_setup_io_intr(adapter); /* napi poll functions should be initialized before running @@ -1757,31 +2161,21 @@ static int ena_up(struct ena_adapter *adapter) * interrupt, causing the ISR to fire immediately while the poll * function wasn't set yet, causing a null dereference */ - ena_init_napi(adapter); + ena_init_napi_in_range(adapter, 0, io_queue_count); + + /* Enabling DIM needs to happen before enabling IRQs since DIM + * is run from napi routine + */ + if (ena_com_interrupt_moderation_supported(adapter->ena_dev)) + ena_com_enable_adaptive_moderation(adapter->ena_dev); rc = ena_request_io_irq(adapter); if (rc) goto err_req_irq; - /* allocate transmit descriptors */ - rc = ena_setup_all_tx_resources(adapter); - if (rc) - goto err_setup_tx; - - /* allocate receive descriptors */ - rc = ena_setup_all_rx_resources(adapter); - if (rc) - goto err_setup_rx; - - /* Create TX queues */ - rc = ena_create_all_io_tx_queues(adapter); + rc = create_queues_with_size_backoff(adapter); if (rc) - goto err_create_tx_queues; - - /* Create RX queues */ - rc = ena_create_all_io_rx_queues(adapter); - if (rc) - goto err_create_rx_queues; + goto err_create_queues_with_backoff; rc = ena_up_complete(adapter); if (rc) @@ -1790,64 +2184,61 @@ static int ena_up(struct ena_adapter *adapter) if (test_bit(ENA_FLAG_LINK_UP, &adapter->flags)) netif_carrier_on(adapter->netdev); - u64_stats_update_begin(&adapter->syncp); - adapter->dev_stats.interface_up++; - u64_stats_update_end(&adapter->syncp); + ena_increase_stat(&adapter->dev_stats.interface_up, 1, + &adapter->syncp); set_bit(ENA_FLAG_DEV_UP, &adapter->flags); /* Enable completion queues interrupt */ - for (i = 0; i < adapter->num_queues; i++) + for (i = 0; i < adapter->num_io_queues; i++) ena_unmask_interrupt(&adapter->tx_ring[i], &adapter->rx_ring[i]); /* schedule napi in case we had pending packets * from the last time we disable napi */ - for (i = 0; i < adapter->num_queues; i++) + for (i = 0; i < io_queue_count; i++) napi_schedule(&adapter->ena_napi[i].napi); return rc; err_up: - ena_destroy_all_rx_queues(adapter); -err_create_rx_queues: ena_destroy_all_tx_queues(adapter); -err_create_tx_queues: - ena_free_all_io_rx_resources(adapter); -err_setup_rx: ena_free_all_io_tx_resources(adapter); -err_setup_tx: + ena_destroy_all_rx_queues(adapter); + ena_free_all_io_rx_resources(adapter); +err_create_queues_with_backoff: ena_free_io_irq(adapter); err_req_irq: + ena_del_napi_in_range(adapter, 0, io_queue_count); return rc; } -static void ena_down(struct ena_adapter *adapter) +void ena_down(struct ena_adapter *adapter) { - netif_info(adapter, ifdown, adapter->netdev, "%s\n", __func__); + int io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues; + + netif_dbg(adapter, ifdown, adapter->netdev, "%s\n", __func__); clear_bit(ENA_FLAG_DEV_UP, &adapter->flags); - u64_stats_update_begin(&adapter->syncp); - adapter->dev_stats.interface_down++; - u64_stats_update_end(&adapter->syncp); + ena_increase_stat(&adapter->dev_stats.interface_down, 1, + &adapter->syncp); netif_carrier_off(adapter->netdev); netif_tx_disable(adapter->netdev); /* After this point the napi handler won't enable the tx queue */ - ena_napi_disable_all(adapter); - - /* After destroy the queue there won't be any new interrupts */ + ena_napi_disable_in_range(adapter, 0, io_queue_count); if (test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags)) { int rc; rc = ena_com_dev_reset(adapter->ena_dev, adapter->reset_reason); if (rc) - dev_err(&adapter->pdev->dev, "Device reset failed\n"); + netif_err(adapter, ifdown, adapter->netdev, + "Device reset failed\n"); /* stop submitting admin commands on a device that was reset */ ena_com_set_admin_running_state(adapter->ena_dev, false); } @@ -1856,7 +2247,7 @@ static void ena_down(struct ena_adapter *adapter) ena_disable_io_intr_sync(adapter); ena_free_io_irq(adapter); - ena_del_napi(adapter); + ena_del_napi_in_range(adapter, 0, io_queue_count); ena_free_all_tx_bufs(adapter); ena_free_all_rx_bufs(adapter); @@ -1881,13 +2272,13 @@ static int ena_open(struct net_device *netdev) int rc; /* Notify the stack of the actual queue counts. */ - rc = netif_set_real_num_tx_queues(netdev, adapter->num_queues); + rc = netif_set_real_num_tx_queues(netdev, adapter->num_io_queues); if (rc) { netif_err(adapter, ifup, netdev, "Can't set num tx queues\n"); return rc; } - rc = netif_set_real_num_rx_queues(netdev, adapter->num_queues); + rc = netif_set_real_num_rx_queues(netdev, adapter->num_io_queues); if (rc) { netif_err(adapter, ifup, netdev, "Can't set num rx queues\n"); return rc; @@ -1936,7 +2327,97 @@ static int ena_close(struct net_device *netdev) return 0; } -static void ena_tx_csum(struct ena_com_tx_ctx *ena_tx_ctx, struct sk_buff *skb) +int ena_update_queue_params(struct ena_adapter *adapter, + u32 new_tx_size, + u32 new_rx_size, + u32 new_llq_header_len) +{ + bool dev_was_up, large_llq_changed = false; + int rc = 0; + + dev_was_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags); + ena_close(adapter->netdev); + adapter->requested_tx_ring_size = new_tx_size; + adapter->requested_rx_ring_size = new_rx_size; + ena_init_io_rings(adapter, + 0, + adapter->xdp_num_queues + + adapter->num_io_queues); + + large_llq_changed = adapter->ena_dev->tx_mem_queue_type == + ENA_ADMIN_PLACEMENT_POLICY_DEV; + large_llq_changed &= + new_llq_header_len != adapter->ena_dev->tx_max_header_size; + + /* a check that the configuration is valid is done by caller */ + if (large_llq_changed) { + adapter->large_llq_header_enabled = !adapter->large_llq_header_enabled; + + ena_destroy_device(adapter, false); + rc = ena_restore_device(adapter); + } + + return dev_was_up && !rc ? ena_up(adapter) : rc; +} + +int ena_set_rx_copybreak(struct ena_adapter *adapter, u32 rx_copybreak) +{ + struct ena_ring *rx_ring; + int i; + + if (rx_copybreak > min_t(u16, adapter->netdev->mtu, ENA_PAGE_SIZE)) + return -EINVAL; + + adapter->rx_copybreak = rx_copybreak; + + for (i = 0; i < adapter->num_io_queues; i++) { + rx_ring = &adapter->rx_ring[i]; + rx_ring->rx_copybreak = rx_copybreak; + } + + return 0; +} + +int ena_update_queue_count(struct ena_adapter *adapter, u32 new_channel_count) +{ + struct ena_com_dev *ena_dev = adapter->ena_dev; + int prev_channel_count; + bool dev_was_up; + + dev_was_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags); + ena_close(adapter->netdev); + prev_channel_count = adapter->num_io_queues; + adapter->num_io_queues = new_channel_count; + if (ena_xdp_present(adapter) && + ena_xdp_allowed(adapter) == ENA_XDP_ALLOWED) { + adapter->xdp_first_ring = new_channel_count; + adapter->xdp_num_queues = new_channel_count; + if (prev_channel_count > new_channel_count) + ena_xdp_exchange_program_rx_in_range(adapter, + NULL, + new_channel_count, + prev_channel_count); + else + ena_xdp_exchange_program_rx_in_range(adapter, + adapter->xdp_bpf_prog, + prev_channel_count, + new_channel_count); + } + + /* We need to destroy the rss table so that the indirection + * table will be reinitialized by ena_up() + */ + ena_com_rss_destroy(ena_dev); + ena_init_io_rings(adapter, + 0, + adapter->xdp_num_queues + + adapter->num_io_queues); + return dev_was_up ? ena_open(adapter->netdev) : 0; +} + +static void ena_tx_csum(struct ena_com_tx_ctx *ena_tx_ctx, + struct sk_buff *skb, + bool disable_meta_caching) { u32 mss = skb_shinfo(skb)->gso_size; struct ena_com_tx_meta *ena_meta = &ena_tx_ctx->ena_meta; @@ -1980,7 +2461,9 @@ static void ena_tx_csum(struct ena_com_tx_ctx *ena_tx_ctx, struct sk_buff *skb) ena_meta->l3_hdr_len = skb_network_header_len(skb); ena_meta->l3_hdr_offset = skb_network_offset(skb); ena_tx_ctx->meta_valid = 1; - + } else if (disable_meta_caching) { + memset(ena_meta, 0, sizeof(*ena_meta)); + ena_tx_ctx->meta_valid = 1; } else { ena_tx_ctx->meta_valid = 0; } @@ -2001,15 +2484,12 @@ static int ena_check_and_linearize_skb(struct ena_ring *tx_ring, (header_len < tx_ring->tx_max_header_size)) return 0; - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.linearize++; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&tx_ring->tx_stats.linearize, 1, &tx_ring->syncp); rc = skb_linearize(skb); if (unlikely(rc)) { - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.linearize_failed++; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&tx_ring->tx_stats.linearize_failed, 1, + &tx_ring->syncp); } return rc; @@ -2049,9 +2529,8 @@ static int ena_tx_map_skb(struct ena_ring *tx_ring, tx_ring->push_buf_intermediate_buf); *header_len = push_len; if (unlikely(skb->data != *push_hdr)) { - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.llq_buffer_copy++; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&tx_ring->tx_stats.llq_buffer_copy, 1, + &tx_ring->syncp); delta = push_len - skb_head_len; } @@ -2108,15 +2587,14 @@ static int ena_tx_map_skb(struct ena_ring *tx_ring, return 0; error_report_dma_error: - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.dma_mapping_err++; - u64_stats_update_end(&tx_ring->syncp); - netdev_warn(adapter->netdev, "failed to map skb\n"); + ena_increase_stat(&tx_ring->tx_stats.dma_mapping_err, 1, + &tx_ring->syncp); + netif_warn(adapter, tx_queued, adapter->netdev, "Failed to map skb\n"); tx_info->skb = NULL; tx_info->num_of_bufs += i; - ena_unmap_tx_skb(tx_ring, tx_info); + ena_unmap_tx_buff(tx_ring, tx_info); return -EINVAL; } @@ -2131,7 +2609,7 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) struct netdev_queue *txq; void *push_hdr; u16 next_to_use, req_id, header_len; - int qid, rc, nb_hw_desc; + int qid, rc; netif_dbg(adapter, tx_queued, dev, "%s skb %p\n", __func__, skb); /* Determine which tx ring we will be placed on */ @@ -2143,10 +2621,8 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) if (unlikely(rc)) goto error_drop_packet; - skb_tx_timestamp(skb); - next_to_use = tx_ring->next_to_use; - req_id = tx_ring->free_tx_ids[next_to_use]; + req_id = tx_ring->free_ids[next_to_use]; tx_info = &tx_ring->tx_buffer_info[req_id]; tx_info->num_of_bufs = 0; @@ -2164,45 +2640,19 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) ena_tx_ctx.header_len = header_len; /* set flags and meta data */ - ena_tx_csum(&ena_tx_ctx, skb); - - /* prepare the packet's descriptors to dma engine */ - rc = ena_com_prepare_tx(tx_ring->ena_com_io_sq, &ena_tx_ctx, - &nb_hw_desc); - - /* ena_com_prepare_tx() can't fail due to overflow of tx queue, - * since the number of free descriptors in the queue is checked - * after sending the previous packet. In case there isn't enough - * space in the queue for the next packet, it is stopped - * until there is again enough available space in the queue. - * All other failure reasons of ena_com_prepare_tx() are fatal - * and therefore require a device reset. - */ - if (unlikely(rc)) { - netif_err(adapter, tx_queued, dev, - "failed to prepare tx bufs\n"); - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.prepare_ctx_err++; - u64_stats_update_end(&tx_ring->syncp); - adapter->reset_reason = ENA_REGS_RESET_DRIVER_INVALID_STATE; - set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); + ena_tx_csum(&ena_tx_ctx, skb, tx_ring->disable_meta_caching); + + rc = ena_xmit_common(adapter, + tx_ring, + tx_info, + &ena_tx_ctx, + next_to_use, + skb->len); + if (rc) goto error_unmap_dma; - } netdev_tx_sent_queue(txq, skb->len); - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.cnt++; - tx_ring->tx_stats.bytes += skb->len; - u64_stats_update_end(&tx_ring->syncp); - - tx_info->tx_descs = nb_hw_desc; - tx_info->last_jiffies = jiffies; - tx_info->print_once = 0; - - tx_ring->next_to_use = ENA_TX_RING_IDX_NEXT(next_to_use, - tx_ring->ring_size); - /* stop the queue when no more space available, the packet can have up * to sgl_size + 2. one for the meta descriptor and one for header * (if the header is larger than tx_max_header_size). @@ -2213,9 +2663,8 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) __func__, qid); netif_tx_stop_queue(txq); - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.queue_stop++; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&tx_ring->tx_stats.queue_stop, 1, + &tx_ring->syncp); /* There is a rare condition where this function decide to * stop the queue but meanwhile clean_tx_irq updates @@ -2230,26 +2679,23 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) if (ena_com_sq_have_enough_space(tx_ring->ena_com_io_sq, ENA_TX_WAKEUP_THRESH)) { netif_tx_wake_queue(txq); - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.queue_wakeup++; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&tx_ring->tx_stats.queue_wakeup, 1, + &tx_ring->syncp); } } - if (netif_xmit_stopped(txq) || !skb->xmit_more) { - /* trigger the dma engine. ena_com_write_sq_doorbell() - * has a mb + skb_tx_timestamp(skb); + + if (netif_xmit_stopped(txq) || !netdev_xmit_more()) + /* trigger the dma engine. ena_ring_tx_doorbell() + * calls a memory barrier inside it. */ - ena_com_write_sq_doorbell(tx_ring->ena_com_io_sq); - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.doorbells++; - u64_stats_update_end(&tx_ring->syncp); - } + ena_ring_tx_doorbell(tx_ring); return NETDEV_TX_OK; error_unmap_dma: - ena_unmap_tx_skb(tx_ring, tx_info); + ena_unmap_tx_buff(tx_ring, tx_info); tx_info->skb = NULL; error_drop_packet: @@ -2257,59 +2703,59 @@ error_drop_packet: return NETDEV_TX_OK; } -static u16 ena_select_queue(struct net_device *dev, struct sk_buff *skb, - struct net_device *sb_dev, - select_queue_fallback_t fallback) -{ - u16 qid; - /* we suspect that this is good for in--kernel network services that - * want to loop incoming skb rx to tx in normal user generated traffic, - * most probably we will not get to this - */ - if (skb_rx_queue_recorded(skb)) - qid = skb_get_rx_queue(skb); - else - qid = fallback(dev, skb, NULL); - - return qid; -} - -static void ena_config_host_info(struct ena_com_dev *ena_dev, - struct pci_dev *pdev) +static void ena_config_host_info(struct ena_com_dev *ena_dev, struct pci_dev *pdev) { + struct device *dev = &pdev->dev; struct ena_admin_host_info *host_info; + ssize_t ret; int rc; /* Allocate only the host info */ rc = ena_com_allocate_host_info(ena_dev); if (rc) { - pr_err("Cannot allocate host info\n"); + dev_err(dev, "Cannot allocate host info\n"); return; } host_info = ena_dev->host_attr.host_info; - host_info->bdf = (pdev->bus->number << 8) | pdev->devfn; + host_info->bdf = pci_dev_id(pdev); host_info->os_type = ENA_ADMIN_OS_LINUX; host_info->kernel_ver = LINUX_VERSION_CODE; - strncpy(host_info->kernel_ver_str, utsname()->version, - sizeof(host_info->kernel_ver_str) - 1); + ret = strscpy(host_info->kernel_ver_str, utsname()->version, + sizeof(host_info->kernel_ver_str)); + if (ret < 0) + dev_dbg(dev, + "kernel version string will be truncated, status = %zd\n", ret); + host_info->os_dist = 0; - strncpy(host_info->os_dist_str, utsname()->release, - sizeof(host_info->os_dist_str) - 1); + ret = strscpy(host_info->os_dist_str, utsname()->release, + sizeof(host_info->os_dist_str)); + if (ret < 0) + dev_dbg(dev, + "OS distribution string will be truncated, status = %zd\n", ret); + host_info->driver_version = - (DRV_MODULE_VER_MAJOR) | - (DRV_MODULE_VER_MINOR << ENA_ADMIN_HOST_INFO_MINOR_SHIFT) | - (DRV_MODULE_VER_SUBMINOR << ENA_ADMIN_HOST_INFO_SUB_MINOR_SHIFT) | + (DRV_MODULE_GEN_MAJOR) | + (DRV_MODULE_GEN_MINOR << ENA_ADMIN_HOST_INFO_MINOR_SHIFT) | + (DRV_MODULE_GEN_SUBMINOR << ENA_ADMIN_HOST_INFO_SUB_MINOR_SHIFT) | ("K"[0] << ENA_ADMIN_HOST_INFO_MODULE_TYPE_SHIFT); host_info->num_cpus = num_online_cpus(); + host_info->driver_supported_features = + ENA_ADMIN_HOST_INFO_RX_OFFSET_MASK | + ENA_ADMIN_HOST_INFO_INTERRUPT_MODERATION_MASK | + ENA_ADMIN_HOST_INFO_RX_BUF_MIRRORING_MASK | + ENA_ADMIN_HOST_INFO_RSS_CONFIGURABLE_FUNCTION_KEY_MASK | + ENA_ADMIN_HOST_INFO_RX_PAGE_REUSE_MASK | + ENA_ADMIN_HOST_INFO_PHC_MASK; + rc = ena_com_set_host_attributes(ena_dev); if (rc) { if (rc == -EOPNOTSUPP) - pr_warn("Cannot set host attributes\n"); + dev_warn(dev, "Cannot set host attributes\n"); else - pr_err("Cannot set host attributes\n"); + dev_err(dev, "Cannot set host attributes\n"); goto err; } @@ -2337,15 +2783,15 @@ static void ena_config_debug_area(struct ena_adapter *adapter) rc = ena_com_allocate_debug_area(adapter->ena_dev, debug_area_size); if (rc) { - pr_err("Cannot allocate debug area\n"); + netif_err(adapter, drv, adapter->netdev, + "Cannot allocate debug area\n"); return; } rc = ena_com_set_host_attributes(adapter->ena_dev); if (rc) { if (rc == -EOPNOTSUPP) - netif_warn(adapter, drv, adapter->netdev, - "Cannot set host attributes\n"); + netif_warn(adapter, drv, adapter->netdev, "Cannot set host attributes\n"); else netif_err(adapter, drv, adapter->netdev, "Cannot set host attributes\n"); @@ -2362,45 +2808,55 @@ static void ena_get_stats64(struct net_device *netdev, { struct ena_adapter *adapter = netdev_priv(netdev); struct ena_ring *rx_ring, *tx_ring; + u64 total_xdp_rx_drops = 0; unsigned int start; u64 rx_drops; + u64 tx_drops; int i; if (!test_bit(ENA_FLAG_DEV_UP, &adapter->flags)) return; - for (i = 0; i < adapter->num_queues; i++) { - u64 bytes, packets; + for (i = 0; i < adapter->num_io_queues + adapter->xdp_num_queues; i++) { + u64 bytes, packets, xdp_rx_drops; tx_ring = &adapter->tx_ring[i]; do { - start = u64_stats_fetch_begin_irq(&tx_ring->syncp); + start = u64_stats_fetch_begin(&tx_ring->syncp); packets = tx_ring->tx_stats.cnt; bytes = tx_ring->tx_stats.bytes; - } while (u64_stats_fetch_retry_irq(&tx_ring->syncp, start)); + } while (u64_stats_fetch_retry(&tx_ring->syncp, start)); stats->tx_packets += packets; stats->tx_bytes += bytes; + /* In XDP there isn't an RX queue counterpart */ + if (ENA_IS_XDP_INDEX(adapter, i)) + continue; + rx_ring = &adapter->rx_ring[i]; do { - start = u64_stats_fetch_begin_irq(&rx_ring->syncp); + start = u64_stats_fetch_begin(&rx_ring->syncp); packets = rx_ring->rx_stats.cnt; bytes = rx_ring->rx_stats.bytes; - } while (u64_stats_fetch_retry_irq(&rx_ring->syncp, start)); + xdp_rx_drops = rx_ring->rx_stats.xdp_drop; + } while (u64_stats_fetch_retry(&rx_ring->syncp, start)); stats->rx_packets += packets; stats->rx_bytes += bytes; + total_xdp_rx_drops += xdp_rx_drops; } do { - start = u64_stats_fetch_begin_irq(&adapter->syncp); + start = u64_stats_fetch_begin(&adapter->syncp); rx_drops = adapter->dev_stats.rx_drops; - } while (u64_stats_fetch_retry_irq(&adapter->syncp, start)); + tx_drops = adapter->dev_stats.tx_drops; + } while (u64_stats_fetch_retry(&adapter->syncp, start)); - stats->rx_dropped = rx_drops; + stats->rx_dropped = rx_drops + total_xdp_rx_drops; + stats->tx_dropped = tx_drops; stats->multicast = 0; stats->collisions = 0; @@ -2420,14 +2876,120 @@ static const struct net_device_ops ena_netdev_ops = { .ndo_open = ena_open, .ndo_stop = ena_close, .ndo_start_xmit = ena_start_xmit, - .ndo_select_queue = ena_select_queue, .ndo_get_stats64 = ena_get_stats64, .ndo_tx_timeout = ena_tx_timeout, .ndo_change_mtu = ena_change_mtu, - .ndo_set_mac_address = NULL, .ndo_validate_addr = eth_validate_addr, + .ndo_bpf = ena_xdp, + .ndo_xdp_xmit = ena_xdp_xmit, }; +static int ena_calc_io_queue_size(struct ena_adapter *adapter, + struct ena_com_dev_get_features_ctx *get_feat_ctx) +{ + struct ena_admin_feature_llq_desc *llq = &get_feat_ctx->llq; + struct ena_com_dev *ena_dev = adapter->ena_dev; + u32 tx_queue_size = ENA_DEFAULT_RING_SIZE; + u32 rx_queue_size = ENA_DEFAULT_RING_SIZE; + u32 max_tx_queue_size; + u32 max_rx_queue_size; + + /* If this function is called after driver load, the ring sizes have already + * been configured. Take it into account when recalculating ring size. + */ + if (adapter->tx_ring->ring_size) + tx_queue_size = adapter->tx_ring->ring_size; + + if (adapter->rx_ring->ring_size) + rx_queue_size = adapter->rx_ring->ring_size; + + if (ena_dev->supported_features & BIT(ENA_ADMIN_MAX_QUEUES_EXT)) { + struct ena_admin_queue_ext_feature_fields *max_queue_ext = + &get_feat_ctx->max_queue_ext.max_queue_ext; + max_rx_queue_size = min_t(u32, max_queue_ext->max_rx_cq_depth, + max_queue_ext->max_rx_sq_depth); + max_tx_queue_size = max_queue_ext->max_tx_cq_depth; + + if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) + max_tx_queue_size = min_t(u32, max_tx_queue_size, + llq->max_llq_depth); + else + max_tx_queue_size = min_t(u32, max_tx_queue_size, + max_queue_ext->max_tx_sq_depth); + + adapter->max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queue_ext->max_per_packet_tx_descs); + adapter->max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queue_ext->max_per_packet_rx_descs); + } else { + struct ena_admin_queue_feature_desc *max_queues = + &get_feat_ctx->max_queues; + max_rx_queue_size = min_t(u32, max_queues->max_cq_depth, + max_queues->max_sq_depth); + max_tx_queue_size = max_queues->max_cq_depth; + + if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) + max_tx_queue_size = min_t(u32, max_tx_queue_size, + llq->max_llq_depth); + else + max_tx_queue_size = min_t(u32, max_tx_queue_size, + max_queues->max_sq_depth); + + adapter->max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queues->max_packet_tx_descs); + adapter->max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queues->max_packet_rx_descs); + } + + max_tx_queue_size = rounddown_pow_of_two(max_tx_queue_size); + max_rx_queue_size = rounddown_pow_of_two(max_rx_queue_size); + + if (max_tx_queue_size < ENA_MIN_RING_SIZE) { + netdev_err(adapter->netdev, "Device max TX queue size: %d < minimum: %d\n", + max_tx_queue_size, ENA_MIN_RING_SIZE); + return -EINVAL; + } + + if (max_rx_queue_size < ENA_MIN_RING_SIZE) { + netdev_err(adapter->netdev, "Device max RX queue size: %d < minimum: %d\n", + max_rx_queue_size, ENA_MIN_RING_SIZE); + return -EINVAL; + } + + /* When forcing large headers, we multiply the entry size by 2, and therefore divide + * the queue size by 2, leaving the amount of memory used by the queues unchanged. + */ + if (adapter->large_llq_header_enabled) { + if ((llq->entry_size_ctrl_supported & ENA_ADMIN_LIST_ENTRY_SIZE_256B) && + ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) { + max_tx_queue_size /= 2; + dev_info(&adapter->pdev->dev, + "Forcing large headers and decreasing maximum TX queue size to %d\n", + max_tx_queue_size); + } else { + dev_err(&adapter->pdev->dev, + "Forcing large headers failed: LLQ is disabled or device does not support large headers\n"); + + adapter->large_llq_header_enabled = false; + } + } + + tx_queue_size = clamp_val(tx_queue_size, ENA_MIN_RING_SIZE, + max_tx_queue_size); + rx_queue_size = clamp_val(rx_queue_size, ENA_MIN_RING_SIZE, + max_rx_queue_size); + + tx_queue_size = rounddown_pow_of_two(tx_queue_size); + rx_queue_size = rounddown_pow_of_two(rx_queue_size); + + adapter->max_tx_ring_size = max_tx_queue_size; + adapter->max_rx_ring_size = max_rx_queue_size; + adapter->requested_tx_ring_size = tx_queue_size; + adapter->requested_rx_ring_size = rx_queue_size; + + return 0; +} + static int ena_device_validate_params(struct ena_adapter *adapter, struct ena_com_dev_get_features_ctx *get_feat_ctx) { @@ -2442,13 +3004,6 @@ static int ena_device_validate_params(struct ena_adapter *adapter, return -EINVAL; } - if ((get_feat_ctx->max_queues.max_cq_num < adapter->num_queues) || - (get_feat_ctx->max_queues.max_sq_num < adapter->num_queues)) { - netif_err(adapter, drv, netdev, - "Error, device doesn't support enough queues\n"); - return -EINVAL; - } - if (get_feat_ctx->dev_attr.max_mtu < netdev->mtu) { netif_err(adapter, drv, netdev, "Error, device max mtu is smaller than netdev MTU\n"); @@ -2458,10 +3013,90 @@ static int ena_device_validate_params(struct ena_adapter *adapter, return 0; } -static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev, +static void set_default_llq_configurations(struct ena_adapter *adapter, + struct ena_llq_configurations *llq_config, + struct ena_admin_feature_llq_desc *llq) +{ + struct ena_com_dev *ena_dev = adapter->ena_dev; + + llq_config->llq_header_location = ENA_ADMIN_INLINE_HEADER; + llq_config->llq_stride_ctrl = ENA_ADMIN_MULTIPLE_DESCS_PER_ENTRY; + llq_config->llq_num_decs_before_header = ENA_ADMIN_LLQ_NUM_DESCS_BEFORE_HEADER_2; + + adapter->large_llq_header_supported = + !!(ena_dev->supported_features & BIT(ENA_ADMIN_LLQ)); + adapter->large_llq_header_supported &= + !!(llq->entry_size_ctrl_supported & + ENA_ADMIN_LIST_ENTRY_SIZE_256B); + + if ((llq->entry_size_ctrl_supported & ENA_ADMIN_LIST_ENTRY_SIZE_256B) && + adapter->large_llq_header_enabled) { + llq_config->llq_ring_entry_size = ENA_ADMIN_LIST_ENTRY_SIZE_256B; + llq_config->llq_ring_entry_size_value = 256; + } else { + llq_config->llq_ring_entry_size = ENA_ADMIN_LIST_ENTRY_SIZE_128B; + llq_config->llq_ring_entry_size_value = 128; + } +} + +static int ena_set_queues_placement_policy(struct pci_dev *pdev, + struct ena_com_dev *ena_dev, + struct ena_admin_feature_llq_desc *llq, + struct ena_llq_configurations *llq_default_configurations) +{ + int rc; + u32 llq_feature_mask; + + llq_feature_mask = 1 << ENA_ADMIN_LLQ; + if (!(ena_dev->supported_features & llq_feature_mask)) { + dev_warn(&pdev->dev, + "LLQ is not supported Fallback to host mode policy.\n"); + ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; + return 0; + } + + if (!ena_dev->mem_bar) { + netdev_err(ena_dev->net_device, + "LLQ is advertised as supported but device doesn't expose mem bar\n"); + ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; + return 0; + } + + rc = ena_com_config_dev_mode(ena_dev, llq, llq_default_configurations); + if (unlikely(rc)) { + dev_err(&pdev->dev, + "Failed to configure the device mode. Fallback to host mode policy.\n"); + ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; + } + + return 0; +} + +static int ena_map_llq_mem_bar(struct pci_dev *pdev, struct ena_com_dev *ena_dev, + int bars) +{ + bool has_mem_bar = !!(bars & BIT(ENA_MEM_BAR)); + + if (!has_mem_bar) + return 0; + + ena_dev->mem_bar = devm_ioremap_wc(&pdev->dev, + pci_resource_start(pdev, ENA_MEM_BAR), + pci_resource_len(pdev, ENA_MEM_BAR)); + + if (!ena_dev->mem_bar) + return -EFAULT; + + return 0; +} + +static int ena_device_init(struct ena_adapter *adapter, struct pci_dev *pdev, struct ena_com_dev_get_features_ctx *get_feat_ctx, bool *wd_state) { + struct ena_com_dev *ena_dev = adapter->ena_dev; + struct net_device *netdev = adapter->netdev; + struct ena_llq_configurations llq_config; struct device *dev = &pdev->dev; bool readless_supported; u32 aenq_groups; @@ -2470,7 +3105,7 @@ static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev, rc = ena_com_mmio_reg_read_request_init(ena_dev); if (rc) { - dev_err(dev, "failed to init mmio read less\n"); + dev_err(dev, "Failed to init mmio read less\n"); return rc; } @@ -2488,7 +3123,7 @@ static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev, rc = ena_com_validate_version(ena_dev); if (rc) { - dev_err(dev, "device version is too low\n"); + dev_err(dev, "Device version is too low\n"); goto err_mmio_read_less; } @@ -2499,18 +3134,13 @@ static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev, goto err_mmio_read_less; } - rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(dma_width)); + rc = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(dma_width)); if (rc) { - dev_err(dev, "pci_set_dma_mask failed 0x%x\n", rc); + dev_err(dev, "dma_set_mask_and_coherent failed %d\n", rc); goto err_mmio_read_less; } - rc = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(dma_width)); - if (rc) { - dev_err(dev, "err_pci_set_consistent_dma_mask failed 0x%x\n", - rc); - goto err_mmio_read_less; - } + ena_devlink_params_get(adapter->devlink); /* ENA admin level init */ rc = ena_com_admin_init(ena_dev, &aenq_handlers); @@ -2552,9 +3182,28 @@ static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev, *wd_state = !!(aenq_groups & BIT(ENA_ADMIN_KEEP_ALIVE)); + set_default_llq_configurations(adapter, &llq_config, &get_feat_ctx->llq); + + rc = ena_set_queues_placement_policy(pdev, ena_dev, &get_feat_ctx->llq, + &llq_config); + if (rc) { + netdev_err(netdev, "Cannot set queues placement policy rc= %d\n", rc); + goto err_admin_init; + } + + rc = ena_calc_io_queue_size(adapter, get_feat_ctx); + if (unlikely(rc)) + goto err_admin_init; + + rc = ena_phc_init(adapter); + if (unlikely(rc && (rc != -EOPNOTSUPP))) + netdev_err(netdev, "Failed initializing PHC, error: %d\n", rc); + return 0; err_admin_init: + ena_com_abort_admin_commands(ena_dev); + ena_com_wait_for_abort_completion(ena_dev); ena_com_delete_host_info(ena_dev); ena_com_admin_destroy(ena_dev); err_mmio_read_less: @@ -2563,14 +3212,13 @@ err_mmio_read_less: return rc; } -static int ena_enable_msix_and_set_admin_interrupts(struct ena_adapter *adapter, - int io_vectors) +static int ena_enable_msix_and_set_admin_interrupts(struct ena_adapter *adapter) { struct ena_com_dev *ena_dev = adapter->ena_dev; struct device *dev = &adapter->pdev->dev; int rc; - rc = ena_enable_msix(adapter, io_vectors); + rc = ena_enable_msix(adapter); if (rc) { dev_err(dev, "Can not reserve msix vectors\n"); return rc; @@ -2596,32 +3244,33 @@ err_disable_msix: return rc; } -static void ena_destroy_device(struct ena_adapter *adapter, bool graceful) +int ena_destroy_device(struct ena_adapter *adapter, bool graceful) { struct net_device *netdev = adapter->netdev; struct ena_com_dev *ena_dev = adapter->ena_dev; bool dev_up; + int rc = 0; if (!test_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags)) - return; + return 0; netif_carrier_off(netdev); - del_timer_sync(&adapter->timer_service); + timer_delete_sync(&adapter->timer_service); dev_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags); adapter->dev_up_before_reset = dev_up; if (!graceful) ena_com_set_admin_running_state(ena_dev, false); - if (test_bit(ENA_FLAG_DEV_UP, &adapter->flags)) + if (dev_up) ena_down(adapter); /* Stop the device from sending AENQ events (in case reset flag is set * and device is up, ena_down() already reset the device. */ if (!(test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags) && dev_up)) - ena_com_dev_reset(adapter->ena_dev, adapter->reset_reason); + rc = ena_com_dev_reset(adapter->ena_dev, adapter->reset_reason); ena_free_mgmnt_irq(adapter); @@ -2633,43 +3282,50 @@ static void ena_destroy_device(struct ena_adapter *adapter, bool graceful) ena_com_admin_destroy(ena_dev); + ena_phc_destroy(adapter); + ena_com_mmio_reg_read_request_destroy(ena_dev); + /* return reset reason to default value */ adapter->reset_reason = ENA_REGS_RESET_NORMAL; clear_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); clear_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags); + + return rc; } -static int ena_restore_device(struct ena_adapter *adapter) +int ena_restore_device(struct ena_adapter *adapter) { struct ena_com_dev_get_features_ctx get_feat_ctx; struct ena_com_dev *ena_dev = adapter->ena_dev; struct pci_dev *pdev = adapter->pdev; + struct ena_ring *txr; + int rc, count, i; bool wd_state; - int rc; set_bit(ENA_FLAG_ONGOING_RESET, &adapter->flags); - rc = ena_device_init(ena_dev, adapter->pdev, &get_feat_ctx, &wd_state); + rc = ena_device_init(adapter, adapter->pdev, &get_feat_ctx, &wd_state); if (rc) { dev_err(&pdev->dev, "Can not initialize device\n"); goto err; } adapter->wd_state = wd_state; + count = adapter->xdp_num_queues + adapter->num_io_queues; + for (i = 0 ; i < count; i++) { + txr = &adapter->tx_ring[i]; + txr->tx_mem_queue_type = ena_dev->tx_mem_queue_type; + txr->tx_max_header_size = ena_dev->tx_max_header_size; + } + rc = ena_device_validate_params(adapter, &get_feat_ctx); if (rc) { dev_err(&pdev->dev, "Validation of device parameters failed\n"); goto err_device_destroy; } - clear_bit(ENA_FLAG_ONGOING_RESET, &adapter->flags); - /* Make sure we don't have a race with AENQ Links state handler */ - if (test_bit(ENA_FLAG_LINK_UP, &adapter->flags)) - netif_carrier_on(adapter->netdev); - - rc = ena_enable_msix_and_set_admin_interrupts(adapter, - adapter->num_queues); + rc = ena_enable_msix_and_set_admin_interrupts(adapter); if (rc) { dev_err(&pdev->dev, "Enable MSI-X failed\n"); goto err_device_destroy; @@ -2684,10 +3340,13 @@ static int ena_restore_device(struct ena_adapter *adapter) } set_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags); + + clear_bit(ENA_FLAG_ONGOING_RESET, &adapter->flags); + if (test_bit(ENA_FLAG_LINK_UP, &adapter->flags)) + netif_carrier_on(adapter->netdev); + mod_timer(&adapter->timer_service, round_jiffies(jiffies + HZ)); - dev_err(&pdev->dev, - "Device reset completed successfully, Driver info: %s\n", - version); + adapter->last_keep_alive_jiffies = jiffies; return rc; err_disable_msix: @@ -2698,6 +3357,7 @@ err_device_destroy: ena_com_wait_for_abort_completion(ena_dev); ena_com_admin_destroy(ena_dev); ena_com_dev_reset(ena_dev, ENA_REGS_RESET_DRIVER_INVALID_STATE); + ena_phc_destroy(adapter); ena_com_mmio_reg_read_request_destroy(ena_dev); err: clear_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags); @@ -2710,25 +3370,30 @@ err: static void ena_fw_reset_device(struct work_struct *work) { + int rc = 0; + struct ena_adapter *adapter = container_of(work, struct ena_adapter, reset_task); - struct pci_dev *pdev = adapter->pdev; - if (unlikely(!test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags))) { - dev_err(&pdev->dev, - "device reset schedule while reset bit is off\n"); - return; - } rtnl_lock(); - ena_destroy_device(adapter, false); - ena_restore_device(adapter); + + if (likely(test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags))) { + rc |= ena_destroy_device(adapter, false); + rc |= ena_restore_device(adapter); + adapter->dev_stats.reset_fail += !!rc; + + dev_err(&adapter->pdev->dev, "Device reset completed successfully\n"); + } + rtnl_unlock(); } static int check_for_rx_interrupt_queue(struct ena_adapter *adapter, struct ena_ring *rx_ring) { - if (likely(rx_ring->first_interrupt)) + struct ena_napi *ena_napi = container_of(rx_ring->napi, struct ena_napi, napi); + + if (likely(READ_ONCE(ena_napi->first_interrupt))) return 0; if (ena_com_cq_empty(rx_ring->ena_com_io_cq)) @@ -2740,9 +3405,8 @@ static int check_for_rx_interrupt_queue(struct ena_adapter *adapter, netif_err(adapter, rx_err, adapter->netdev, "Potential MSIX issue on Rx side Queue = %d. Reset the device\n", rx_ring->qid); - adapter->reset_reason = ENA_REGS_RESET_MISS_INTERRUPT; - smp_mb__before_atomic(); - set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); + + ena_reset_device(adapter, ENA_REGS_RESET_MISS_INTERRUPT); return -EIO; } @@ -2752,11 +3416,19 @@ static int check_for_rx_interrupt_queue(struct ena_adapter *adapter, static int check_missing_comp_in_tx_queue(struct ena_adapter *adapter, struct ena_ring *tx_ring) { + struct ena_napi *ena_napi = container_of(tx_ring->napi, struct ena_napi, napi); + enum ena_regs_reset_reason_types reset_reason = ENA_REGS_RESET_MISS_TX_CMPL; + unsigned int time_since_last_napi; + unsigned int missing_tx_comp_to; + bool is_tx_comp_time_expired; struct ena_tx_buffer *tx_buf; unsigned long last_jiffies; + int napi_scheduled; u32 missed_tx = 0; int i, rc = 0; + missing_tx_comp_to = jiffies_to_msecs(adapter->missing_tx_completion_to); + for (i = 0; i < tx_ring->ring_size; i++) { tx_buf = &tx_ring->tx_buffer_info[i]; last_jiffies = tx_buf->last_jiffies; @@ -2765,46 +3437,68 @@ static int check_missing_comp_in_tx_queue(struct ena_adapter *adapter, /* no pending Tx at this location */ continue; - if (unlikely(!tx_ring->first_interrupt && time_is_before_jiffies(last_jiffies + - 2 * adapter->missing_tx_completion_to))) { + is_tx_comp_time_expired = time_is_before_jiffies(last_jiffies + + 2 * adapter->missing_tx_completion_to); + + if (unlikely(!READ_ONCE(ena_napi->first_interrupt) && is_tx_comp_time_expired)) { /* If after graceful period interrupt is still not * received, we schedule a reset */ netif_err(adapter, tx_err, adapter->netdev, "Potential MSIX issue on Tx side Queue = %d. Reset the device\n", tx_ring->qid); - adapter->reset_reason = ENA_REGS_RESET_MISS_INTERRUPT; - smp_mb__before_atomic(); - set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); + ena_reset_device(adapter, ENA_REGS_RESET_MISS_INTERRUPT); return -EIO; } - if (unlikely(time_is_before_jiffies(last_jiffies + - adapter->missing_tx_completion_to))) { - if (!tx_buf->print_once) - netif_notice(adapter, tx_err, adapter->netdev, - "Found a Tx that wasn't completed on time, qid %d, index %d.\n", - tx_ring->qid, i); + is_tx_comp_time_expired = time_is_before_jiffies(last_jiffies + + adapter->missing_tx_completion_to); + + if (unlikely(is_tx_comp_time_expired)) { + time_since_last_napi = + jiffies_to_usecs(jiffies - tx_ring->tx_stats.last_napi_jiffies); + napi_scheduled = !!(ena_napi->napi.state & NAPIF_STATE_SCHED); + + if (missing_tx_comp_to < time_since_last_napi && napi_scheduled) { + /* We suspect napi isn't called because the + * bottom half is not run. Require a bigger + * timeout for these cases + */ + if (!time_is_before_jiffies(last_jiffies + + 2 * adapter->missing_tx_completion_to)) + continue; + + reset_reason = ENA_REGS_RESET_SUSPECTED_POLL_STARVATION; + } - tx_buf->print_once = 1; missed_tx++; + + if (tx_buf->print_once) + continue; + + netif_notice(adapter, tx_err, adapter->netdev, + "TX hasn't completed, qid %d, index %d. %u usecs from last napi execution, napi scheduled: %d\n", + tx_ring->qid, i, time_since_last_napi, napi_scheduled); + + tx_buf->print_once = 1; } } if (unlikely(missed_tx > adapter->missing_tx_completion_threshold)) { netif_err(adapter, tx_err, adapter->netdev, - "The number of lost tx completions is above the threshold (%d > %d). Reset the device\n", + "Lost TX completions are above the threshold (%d > %d). Completion transmission timeout: %u.\n", missed_tx, - adapter->missing_tx_completion_threshold); - adapter->reset_reason = - ENA_REGS_RESET_MISS_TX_CMPL; - set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); + adapter->missing_tx_completion_threshold, + missing_tx_comp_to); + netif_err(adapter, tx_err, adapter->netdev, + "Resetting the device\n"); + + ena_reset_device(adapter, reset_reason); rc = -EIO; } - u64_stats_update_begin(&tx_ring->syncp); - tx_ring->tx_stats.missed_tx = missed_tx; - u64_stats_update_end(&tx_ring->syncp); + ena_increase_stat(&tx_ring->tx_stats.missed_tx, missed_tx, + &tx_ring->syncp); return rc; } @@ -2813,7 +3507,10 @@ static void check_for_missing_completions(struct ena_adapter *adapter) { struct ena_ring *tx_ring; struct ena_ring *rx_ring; - int i, budget, rc; + int qid, budget, rc; + int io_queue_count; + + io_queue_count = adapter->xdp_num_queues + adapter->num_io_queues; /* Make sure the driver doesn't turn the device in other process */ smp_rmb(); @@ -2827,26 +3524,29 @@ static void check_for_missing_completions(struct ena_adapter *adapter) if (adapter->missing_tx_completion_to == ENA_HW_HINTS_NO_TIMEOUT) return; - budget = ENA_MONITORED_TX_QUEUES; + budget = min_t(u32, io_queue_count, ENA_MONITORED_TX_QUEUES); - for (i = adapter->last_monitored_tx_qid; i < adapter->num_queues; i++) { - tx_ring = &adapter->tx_ring[i]; - rx_ring = &adapter->rx_ring[i]; + qid = adapter->last_monitored_tx_qid; + + while (budget) { + qid = (qid + 1) % io_queue_count; + + tx_ring = &adapter->tx_ring[qid]; + rx_ring = &adapter->rx_ring[qid]; rc = check_missing_comp_in_tx_queue(adapter, tx_ring); if (unlikely(rc)) return; - rc = check_for_rx_interrupt_queue(adapter, rx_ring); + rc = !ENA_IS_XDP_INDEX(adapter, qid) ? + check_for_rx_interrupt_queue(adapter, rx_ring) : 0; if (unlikely(rc)) return; budget--; - if (!budget) - break; } - adapter->last_monitored_tx_qid = i % adapter->num_queues; + adapter->last_monitored_tx_qid = qid; } /* trigger napi schedule after 2 consecutive detections */ @@ -2876,21 +3576,19 @@ static void check_for_empty_rx_ring(struct ena_adapter *adapter) if (test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags)) return; - for (i = 0; i < adapter->num_queues; i++) { + for (i = 0; i < adapter->num_io_queues; i++) { rx_ring = &adapter->rx_ring[i]; - refill_required = - ena_com_free_desc(rx_ring->ena_com_io_sq); + refill_required = ena_com_free_q_entries(rx_ring->ena_com_io_sq); if (unlikely(refill_required == (rx_ring->ring_size - 1))) { rx_ring->empty_rx_queue++; if (rx_ring->empty_rx_queue >= EMPTY_RX_REFILL) { - u64_stats_update_begin(&rx_ring->syncp); - rx_ring->rx_stats.empty_rx_ring++; - u64_stats_update_end(&rx_ring->syncp); + ena_increase_stat(&rx_ring->rx_stats.empty_rx_ring, 1, + &rx_ring->syncp); netif_err(adapter, drv, adapter->netdev, - "trigger refill for ring %d\n", i); + "Trigger refill for ring %d\n", i); napi_schedule(rx_ring->napi); rx_ring->empty_rx_queue = 0; @@ -2912,16 +3610,14 @@ static void check_for_missing_keep_alive(struct ena_adapter *adapter) if (adapter->keep_alive_timeout == ENA_HW_HINTS_NO_TIMEOUT) return; - keep_alive_expired = round_jiffies(adapter->last_keep_alive_jiffies + - adapter->keep_alive_timeout); + keep_alive_expired = adapter->last_keep_alive_jiffies + + adapter->keep_alive_timeout; if (unlikely(time_is_before_jiffies(keep_alive_expired))) { netif_err(adapter, drv, adapter->netdev, "Keep alive watchdog timeout.\n"); - u64_stats_update_begin(&adapter->syncp); - adapter->dev_stats.wd_expired++; - u64_stats_update_end(&adapter->syncp); - adapter->reset_reason = ENA_REGS_RESET_KEEP_ALIVE_TO; - set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); + ena_increase_stat(&adapter->dev_stats.wd_expired, 1, + &adapter->syncp); + ena_reset_device(adapter, ENA_REGS_RESET_KEEP_ALIVE_TO); } } @@ -2930,11 +3626,9 @@ static void check_for_admin_com_state(struct ena_adapter *adapter) if (unlikely(!ena_com_get_admin_running_state(adapter->ena_dev))) { netif_err(adapter, drv, adapter->netdev, "ENA admin queue is not in running state!\n"); - u64_stats_update_begin(&adapter->syncp); - adapter->dev_stats.admin_q_pause++; - u64_stats_update_end(&adapter->syncp); - adapter->reset_reason = ENA_REGS_RESET_ADMIN_TO; - set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); + ena_increase_stat(&adapter->dev_stats.admin_q_pause, 1, + &adapter->syncp); + ena_reset_device(adapter, ENA_REGS_RESET_ADMIN_TO); } } @@ -2987,7 +3681,8 @@ static void ena_update_host_info(struct ena_admin_host_info *host_info, static void ena_timer_service(struct timer_list *t) { - struct ena_adapter *adapter = from_timer(adapter, t, timer_service); + struct ena_adapter *adapter = timer_container_of(adapter, t, + timer_service); u8 *debug_area = adapter->ena_dev->host_attr.debug_area_virt_addr; struct ena_admin_host_info *host_info = adapter->ena_dev->host_attr.host_info; @@ -3015,81 +3710,43 @@ static void ena_timer_service(struct timer_list *t) } /* Reset the timer */ - mod_timer(&adapter->timer_service, jiffies + HZ); -} - -static int ena_calc_io_queue_num(struct pci_dev *pdev, - struct ena_com_dev *ena_dev, - struct ena_com_dev_get_features_ctx *get_feat_ctx) -{ - int io_sq_num, io_queue_num; - - /* In case of LLQ use the llq number in the get feature cmd */ - if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) - io_sq_num = get_feat_ctx->llq.max_llq_num; - else - io_sq_num = get_feat_ctx->max_queues.max_sq_num; - - io_queue_num = min_t(int, num_online_cpus(), ENA_MAX_NUM_IO_QUEUES); - io_queue_num = min_t(int, io_queue_num, io_sq_num); - io_queue_num = min_t(int, io_queue_num, - get_feat_ctx->max_queues.max_cq_num); - /* 1 IRQ for for mgmnt and 1 IRQs for each IO direction */ - io_queue_num = min_t(int, io_queue_num, pci_msix_vec_count(pdev) - 1); - if (unlikely(!io_queue_num)) { - dev_err(&pdev->dev, "The device doesn't have io queues\n"); - return -EFAULT; - } - - return io_queue_num; + mod_timer(&adapter->timer_service, round_jiffies(jiffies + HZ)); } -static int ena_set_queues_placement_policy(struct pci_dev *pdev, - struct ena_com_dev *ena_dev, - struct ena_admin_feature_llq_desc *llq, - struct ena_llq_configurations *llq_default_configurations) +static u32 ena_calc_max_io_queue_num(struct pci_dev *pdev, + struct ena_com_dev *ena_dev, + struct ena_com_dev_get_features_ctx *get_feat_ctx) { - bool has_mem_bar; - int rc; - u32 llq_feature_mask; + u32 io_tx_sq_num, io_tx_cq_num, io_rx_num, max_num_io_queues; - llq_feature_mask = 1 << ENA_ADMIN_LLQ; - if (!(ena_dev->supported_features & llq_feature_mask)) { - dev_err(&pdev->dev, - "LLQ is not supported Fallback to host mode policy.\n"); - ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; - return 0; - } - - has_mem_bar = pci_select_bars(pdev, IORESOURCE_MEM) & BIT(ENA_MEM_BAR); - - rc = ena_com_config_dev_mode(ena_dev, llq, llq_default_configurations); - if (unlikely(rc)) { - dev_err(&pdev->dev, - "Failed to configure the device mode. Fallback to host mode policy.\n"); - ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; - return 0; - } - - /* Nothing to config, exit */ - if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_HOST) - return 0; + if (ena_dev->supported_features & BIT(ENA_ADMIN_MAX_QUEUES_EXT)) { + struct ena_admin_queue_ext_feature_fields *max_queue_ext = + &get_feat_ctx->max_queue_ext.max_queue_ext; + io_rx_num = min_t(u32, max_queue_ext->max_rx_sq_num, + max_queue_ext->max_rx_cq_num); - if (!has_mem_bar) { - dev_err(&pdev->dev, - "ENA device does not expose LLQ bar. Fallback to host mode policy.\n"); - ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; - return 0; + io_tx_sq_num = max_queue_ext->max_tx_sq_num; + io_tx_cq_num = max_queue_ext->max_tx_cq_num; + } else { + struct ena_admin_queue_feature_desc *max_queues = + &get_feat_ctx->max_queues; + io_tx_sq_num = max_queues->max_sq_num; + io_tx_cq_num = max_queues->max_cq_num; + io_rx_num = min_t(u32, io_tx_sq_num, io_tx_cq_num); } - ena_dev->mem_bar = devm_ioremap_wc(&pdev->dev, - pci_resource_start(pdev, ENA_MEM_BAR), - pci_resource_len(pdev, ENA_MEM_BAR)); + /* In case of LLQ use the llq fields for the tx SQ/CQ */ + if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) + io_tx_sq_num = get_feat_ctx->llq.max_llq_num; - if (!ena_dev->mem_bar) - return -EFAULT; + max_num_io_queues = min_t(u32, num_online_cpus(), ENA_MAX_NUM_IO_QUEUES); + max_num_io_queues = min_t(u32, max_num_io_queues, io_rx_num); + max_num_io_queues = min_t(u32, max_num_io_queues, io_tx_sq_num); + max_num_io_queues = min_t(u32, max_num_io_queues, io_tx_cq_num); + /* 1 IRQ for mgmnt and 1 IRQs for each IO direction */ + max_num_io_queues = min_t(u32, max_num_io_queues, pci_msix_vec_count(pdev) - 1); - return 0; + return max_num_io_queues; } static void ena_set_dev_offloads(struct ena_com_dev_get_features_ctx *feat, @@ -3144,7 +3801,7 @@ static void ena_set_conf_feat_params(struct ena_adapter *adapter, ether_addr_copy(adapter->mac_addr, netdev->dev_addr); } else { ether_addr_copy(adapter->mac_addr, feat->dev_attr.mac_addr); - ether_addr_copy(netdev->dev_addr, adapter->mac_addr); + eth_hw_addr_set(netdev, adapter->mac_addr); } /* Set offload features */ @@ -3169,17 +3826,17 @@ static int ena_rss_init_default(struct ena_adapter *adapter) } for (i = 0; i < ENA_RX_RSS_TABLE_SIZE; i++) { - val = ethtool_rxfh_indir_default(i, adapter->num_queues); + val = ethtool_rxfh_indir_default(i, adapter->num_io_queues); rc = ena_com_indirect_table_fill_entry(ena_dev, i, ENA_IO_RXQ_IDX(val)); - if (unlikely(rc && (rc != -EOPNOTSUPP))) { + if (unlikely(rc)) { dev_err(dev, "Cannot fill indirect table\n"); goto err_fill_indir; } } - rc = ena_com_fill_hash_function(ena_dev, ENA_ADMIN_CRC32, NULL, - ENA_HASH_KEY_SIZE, 0xFFFFFFFF); + rc = ena_com_fill_hash_function(ena_dev, ENA_ADMIN_TOEPLITZ, NULL, ENA_HASH_KEY_SIZE, + 0xFFFFFFFF); if (unlikely(rc && (rc != -EOPNOTSUPP))) { dev_err(dev, "Cannot fill hash function\n"); goto err_fill_indir; @@ -3207,47 +3864,6 @@ static void ena_release_bars(struct ena_com_dev *ena_dev, struct pci_dev *pdev) pci_release_selected_regions(pdev, release_bars); } -static inline void set_default_llq_configurations(struct ena_llq_configurations *llq_config) -{ - llq_config->llq_header_location = ENA_ADMIN_INLINE_HEADER; - llq_config->llq_ring_entry_size = ENA_ADMIN_LIST_ENTRY_SIZE_128B; - llq_config->llq_stride_ctrl = ENA_ADMIN_MULTIPLE_DESCS_PER_ENTRY; - llq_config->llq_num_decs_before_header = ENA_ADMIN_LLQ_NUM_DESCS_BEFORE_HEADER_2; - llq_config->llq_ring_entry_size_value = 128; -} - -static int ena_calc_queue_size(struct pci_dev *pdev, - struct ena_com_dev *ena_dev, - u16 *max_tx_sgl_size, - u16 *max_rx_sgl_size, - struct ena_com_dev_get_features_ctx *get_feat_ctx) -{ - u32 queue_size = ENA_DEFAULT_RING_SIZE; - - queue_size = min_t(u32, queue_size, - get_feat_ctx->max_queues.max_cq_depth); - queue_size = min_t(u32, queue_size, - get_feat_ctx->max_queues.max_sq_depth); - - if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) - queue_size = min_t(u32, queue_size, - get_feat_ctx->llq.max_llq_depth); - - queue_size = rounddown_pow_of_two(queue_size); - - if (unlikely(!queue_size)) { - dev_err(&pdev->dev, "Invalid queue size\n"); - return -EFAULT; - } - - *max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, - get_feat_ctx->max_queues.max_packet_tx_descs); - *max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, - get_feat_ctx->max_queues.max_packet_rx_descs); - - return queue_size; -} - /* ena_probe - Device Initialization Routine * @pdev: PCI device information struct * @ent: entry in ena_pci_tbl @@ -3261,30 +3877,29 @@ static int ena_calc_queue_size(struct pci_dev *pdev, static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) { struct ena_com_dev_get_features_ctx get_feat_ctx; - static int version_printed; - struct net_device *netdev; - struct ena_adapter *adapter; - struct ena_llq_configurations llq_config; struct ena_com_dev *ena_dev = NULL; - char *queue_type_str; + struct ena_adapter *adapter; + struct net_device *netdev; static int adapters_found; - int io_queue_num, bars, rc; - int queue_size; - u16 tx_sgl_size = 0; - u16 rx_sgl_size = 0; + struct devlink *devlink; + u32 max_num_io_queues; bool wd_state; + int bars, rc; dev_dbg(&pdev->dev, "%s\n", __func__); - if (version_printed++ == 0) - dev_info(&pdev->dev, "%s", version); - rc = pci_enable_device_mem(pdev); if (rc) { dev_err(&pdev->dev, "pci_enable_device_mem() failed!\n"); return rc; } + rc = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(ENA_MAX_PHYS_ADDR_SIZE_BITS)); + if (rc) { + dev_err(&pdev->dev, "dma_set_mask_and_coherent failed %d\n", rc); + goto err_disable_device; + } + pci_set_master(pdev); ena_dev = vzalloc(sizeof(*ena_dev)); @@ -3305,79 +3920,96 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) pci_resource_start(pdev, ENA_REG_BAR), pci_resource_len(pdev, ENA_REG_BAR)); if (!ena_dev->reg_bar) { - dev_err(&pdev->dev, "failed to remap regs bar\n"); + dev_err(&pdev->dev, "Failed to remap regs bar\n"); rc = -EFAULT; goto err_free_region; } + ena_dev->ena_min_poll_delay_us = ENA_ADMIN_POLL_DELAY_US; + ena_dev->dmadev = &pdev->dev; - rc = ena_device_init(ena_dev, pdev, &get_feat_ctx, &wd_state); - if (rc) { - dev_err(&pdev->dev, "ena device init failed\n"); - if (rc == -ETIME) - rc = -EPROBE_DEFER; + netdev = alloc_etherdev_mq(sizeof(struct ena_adapter), ENA_MAX_RINGS); + if (!netdev) { + dev_err(&pdev->dev, "alloc_etherdev_mq failed\n"); + rc = -ENOMEM; goto err_free_region; } - set_default_llq_configurations(&llq_config); + SET_NETDEV_DEV(netdev, &pdev->dev); + adapter = netdev_priv(netdev); + adapter->ena_dev = ena_dev; + adapter->netdev = netdev; + adapter->pdev = pdev; + adapter->msg_enable = DEFAULT_MSG_ENABLE; - rc = ena_set_queues_placement_policy(pdev, ena_dev, &get_feat_ctx.llq, - &llq_config); + ena_dev->net_device = netdev; + + pci_set_drvdata(pdev, adapter); + + rc = ena_phc_alloc(adapter); if (rc) { - dev_err(&pdev->dev, "ena device init failed\n"); - goto err_device_destroy; + netdev_err(netdev, "ena_phc_alloc failed\n"); + goto err_netdev_destroy; } - /* initial Tx interrupt delay, Assumes 1 usec granularity. - * Updated during device initialization with the real granularity - */ - ena_dev->intr_moder_tx_interval = ENA_INTR_INITIAL_TX_INTERVAL_USECS; - io_queue_num = ena_calc_io_queue_num(pdev, ena_dev, &get_feat_ctx); - queue_size = ena_calc_queue_size(pdev, ena_dev, &tx_sgl_size, - &rx_sgl_size, &get_feat_ctx); - if ((queue_size <= 0) || (io_queue_num <= 0)) { - rc = -EFAULT; - goto err_device_destroy; + rc = ena_com_allocate_customer_metrics_buffer(ena_dev); + if (rc) { + netdev_err(netdev, "ena_com_allocate_customer_metrics_buffer failed\n"); + goto err_free_phc; } - dev_info(&pdev->dev, "creating %d io queues. queue size: %d. LLQ is %s\n", - io_queue_num, queue_size, - (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) ? - "ENABLED" : "DISABLED"); + rc = ena_map_llq_mem_bar(pdev, ena_dev, bars); + if (rc) { + dev_err(&pdev->dev, "ENA LLQ bar mapping failed\n"); + goto err_metrics_destroy; + } - /* dev zeroed in init_etherdev */ - netdev = alloc_etherdev_mq(sizeof(struct ena_adapter), io_queue_num); - if (!netdev) { - dev_err(&pdev->dev, "alloc_etherdev_mq failed\n"); + /* Need to do this before ena_device_init */ + devlink = ena_devlink_alloc(adapter); + if (!devlink) { + netdev_err(netdev, "ena_devlink_alloc failed\n"); rc = -ENOMEM; - goto err_device_destroy; + goto err_metrics_destroy; } - SET_NETDEV_DEV(netdev, &pdev->dev); - - adapter = netdev_priv(netdev); - pci_set_drvdata(pdev, adapter); + rc = ena_device_init(adapter, pdev, &get_feat_ctx, &wd_state); + if (rc) { + dev_err(&pdev->dev, "ENA device init failed\n"); + if (rc == -ETIME) + rc = -EPROBE_DEFER; + goto ena_devlink_destroy; + } - adapter->ena_dev = ena_dev; - adapter->netdev = netdev; - adapter->pdev = pdev; + /* Initial TX and RX interrupt delay. Assumes 1 usec granularity. + * Updated during device initialization with the real granularity + */ + ena_dev->intr_moder_tx_interval = ENA_INTR_INITIAL_TX_INTERVAL_USECS; + ena_dev->intr_moder_rx_interval = ENA_INTR_INITIAL_RX_INTERVAL_USECS; + ena_dev->intr_delay_resolution = ENA_DEFAULT_INTR_DELAY_RESOLUTION; + max_num_io_queues = ena_calc_max_io_queue_num(pdev, ena_dev, &get_feat_ctx); + if (unlikely(!max_num_io_queues)) { + rc = -EFAULT; + goto err_device_destroy; + } ena_set_conf_feat_params(adapter, &get_feat_ctx); - adapter->msg_enable = netif_msg_init(debug, DEFAULT_MSG_ENABLE); adapter->reset_reason = ENA_REGS_RESET_NORMAL; - adapter->tx_ring_size = queue_size; - adapter->rx_ring_size = queue_size; - - adapter->max_tx_sgl_size = tx_sgl_size; - adapter->max_rx_sgl_size = rx_sgl_size; - - adapter->num_queues = io_queue_num; + adapter->num_io_queues = max_num_io_queues; + adapter->max_num_io_queues = max_num_io_queues; adapter->last_monitored_tx_qid = 0; + adapter->xdp_first_ring = 0; + adapter->xdp_num_queues = 0; + adapter->rx_copybreak = ENA_DEFAULT_RX_COPYBREAK; + if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) + adapter->disable_meta_caching = + !!(get_feat_ctx.llq.accel_mode.u.get.supported_flags & + BIT(ENA_ADMIN_DISABLE_META_CACHING)); + adapter->wd_state = wd_state; snprintf(adapter->name, ENA_NAME_MAX_LEN, "ena_%d", adapters_found); @@ -3386,9 +4018,13 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) if (rc) { dev_err(&pdev->dev, "Failed to query interrupt moderation feature\n"); - goto err_netdev_destroy; + goto err_device_destroy; } - ena_init_io_rings(adapter); + + ena_init_io_rings(adapter, + 0, + adapter->xdp_num_queues + + adapter->num_io_queues); netdev->netdev_ops = &ena_netdev_ops; netdev->watchdog_timeo = TX_TIMEOUT; @@ -3398,7 +4034,7 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) u64_stats_init(&adapter->syncp); - rc = ena_enable_msix_and_set_admin_interrupts(adapter, io_queue_num); + rc = ena_enable_msix_and_set_admin_interrupts(adapter); if (rc) { dev_err(&pdev->dev, "Failed to enable and set the admin interrupts\n"); @@ -3412,6 +4048,10 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) ena_config_debug_area(adapter); + if (ena_xdp_legal_queue_count(adapter, adapter->num_io_queues)) + netdev->xdp_features = NETDEV_XDP_ACT_BASIC | + NETDEV_XDP_ACT_REDIRECT; + memcpy(adapter->netdev->perm_addr, adapter->mac_addr, netdev->addr_len); netif_carrier_off(netdev); @@ -3422,6 +4062,8 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) goto err_rss; } + ena_debugfs_init(netdev); + INIT_WORK(&adapter->reset_task, ena_fw_reset_device); adapter->last_keep_alive_jiffies = jiffies; @@ -3434,20 +4076,21 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) timer_setup(&adapter->timer_service, ena_timer_service, 0); mod_timer(&adapter->timer_service, round_jiffies(jiffies + HZ)); - if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_HOST) - queue_type_str = "Regular"; - else - queue_type_str = "Low Latency"; - dev_info(&pdev->dev, - "%s found at mem %lx, mac addr %pM Queues %d, Placement policy: %s\n", + "%s found at mem %lx, mac addr %pM\n", DEVICE_NAME, (long)pci_resource_start(pdev, 0), - netdev->dev_addr, io_queue_num, queue_type_str); + netdev->dev_addr); set_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags); adapters_found++; + /* From this point, the devlink device is visible to users. + * Perform the registration last to ensure that all the resources + * are available and that the netdevice is registered. + */ + ena_devlink_register(devlink, &pdev->dev); + return 0; err_rss: @@ -3460,13 +4103,18 @@ err_free_msix: ena_free_mgmnt_irq(adapter); ena_disable_msix(adapter); err_worker_destroy: - ena_com_destroy_interrupt_moderation(ena_dev); - del_timer(&adapter->timer_service); -err_netdev_destroy: - free_netdev(netdev); + timer_delete(&adapter->timer_service); err_device_destroy: ena_com_delete_host_info(ena_dev); ena_com_admin_destroy(ena_dev); +ena_devlink_destroy: + ena_devlink_free(devlink); +err_metrics_destroy: + ena_com_delete_customer_metrics_buffer(ena_dev); +err_free_phc: + ena_phc_free(adapter); +err_netdev_destroy: + free_netdev(netdev); err_free_region: ena_release_bars(ena_dev, pdev); err_free_ena_dev: @@ -3478,13 +4126,15 @@ err_disable_device: /*****************************************************************************/ -/* ena_remove - Device Removal Routine +/* __ena_shutoff - Helper used in both PCI remove/shutdown routines * @pdev: PCI device information struct + * @shutdown: Is it a shutdown operation? If false, means it is a removal * - * ena_remove is called by the PCI subsystem to alert the driver - * that it should release a PCI device. + * __ena_shutoff is a helper routine that does the real work on shutdown and + * removal paths; the difference between those paths is with regards to whether + * dettach or unregister the netdevice. */ -static void ena_remove(struct pci_dev *pdev) +static void __ena_shutoff(struct pci_dev *pdev, bool shutdown) { struct ena_adapter *adapter = pci_get_drvdata(pdev); struct ena_com_dev *ena_dev; @@ -3493,23 +4143,32 @@ static void ena_remove(struct pci_dev *pdev) ena_dev = adapter->ena_dev; netdev = adapter->netdev; -#ifdef CONFIG_RFS_ACCEL - if ((adapter->msix_vecs >= 1) && (netdev->rx_cpu_rmap)) { - free_irq_cpu_rmap(netdev->rx_cpu_rmap); - netdev->rx_cpu_rmap = NULL; - } -#endif /* CONFIG_RFS_ACCEL */ - del_timer_sync(&adapter->timer_service); + ena_debugfs_terminate(netdev); + /* Make sure timer and reset routine won't be called after + * freeing device resources. + */ + timer_delete_sync(&adapter->timer_service); cancel_work_sync(&adapter->reset_task); - rtnl_lock(); + rtnl_lock(); /* lock released inside the below if-else block */ + adapter->reset_reason = ENA_REGS_RESET_SHUTDOWN; ena_destroy_device(adapter, true); - rtnl_unlock(); - unregister_netdev(netdev); + ena_phc_free(adapter); - free_netdev(netdev); + ena_devlink_unregister(adapter->devlink); + ena_devlink_free(adapter->devlink); + + if (shutdown) { + netif_device_detach(netdev); + dev_close(netdev); + rtnl_unlock(); + } else { + rtnl_unlock(); + unregister_netdev(netdev); + free_netdev(netdev); + } ena_com_rss_destroy(ena_dev); @@ -3517,32 +4176,53 @@ static void ena_remove(struct pci_dev *pdev) ena_com_delete_host_info(ena_dev); + ena_com_delete_customer_metrics_buffer(ena_dev); + ena_release_bars(ena_dev, pdev); pci_disable_device(pdev); - ena_com_destroy_interrupt_moderation(ena_dev); - vfree(ena_dev); } -#ifdef CONFIG_PM -/* ena_suspend - PM suspend callback +/* ena_remove - Device Removal Routine * @pdev: PCI device information struct - * @state:power state + * + * ena_remove is called by the PCI subsystem to alert the driver + * that it should release a PCI device. */ -static int ena_suspend(struct pci_dev *pdev, pm_message_t state) + +static void ena_remove(struct pci_dev *pdev) { + __ena_shutoff(pdev, false); +} + +/* ena_shutdown - Device Shutdown Routine + * @pdev: PCI device information struct + * + * ena_shutdown is called by the PCI subsystem to alert the driver that + * a shutdown/reboot (or kexec) is happening and device must be disabled. + */ + +static void ena_shutdown(struct pci_dev *pdev) +{ + __ena_shutoff(pdev, true); +} + +/* ena_suspend - PM suspend callback + * @dev_d: Device information struct + */ +static int __maybe_unused ena_suspend(struct device *dev_d) +{ + struct pci_dev *pdev = to_pci_dev(dev_d); struct ena_adapter *adapter = pci_get_drvdata(pdev); - u64_stats_update_begin(&adapter->syncp); - adapter->dev_stats.suspend++; - u64_stats_update_end(&adapter->syncp); + ena_increase_stat(&adapter->dev_stats.suspend, 1, &adapter->syncp); rtnl_lock(); if (unlikely(test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags))) { dev_err(&pdev->dev, - "ignoring device reset request as the device is being suspended\n"); + "Ignoring device reset request as the device is being suspended\n"); clear_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags); } ena_destroy_device(adapter, true); @@ -3551,40 +4231,36 @@ static int ena_suspend(struct pci_dev *pdev, pm_message_t state) } /* ena_resume - PM resume callback - * @pdev: PCI device information struct - * + * @dev_d: Device information struct */ -static int ena_resume(struct pci_dev *pdev) +static int __maybe_unused ena_resume(struct device *dev_d) { - struct ena_adapter *adapter = pci_get_drvdata(pdev); + struct ena_adapter *adapter = dev_get_drvdata(dev_d); int rc; - u64_stats_update_begin(&adapter->syncp); - adapter->dev_stats.resume++; - u64_stats_update_end(&adapter->syncp); + ena_increase_stat(&adapter->dev_stats.resume, 1, &adapter->syncp); rtnl_lock(); rc = ena_restore_device(adapter); rtnl_unlock(); return rc; } -#endif + +static SIMPLE_DEV_PM_OPS(ena_pm_ops, ena_suspend, ena_resume); static struct pci_driver ena_pci_driver = { .name = DRV_MODULE_NAME, .id_table = ena_pci_tbl, .probe = ena_probe, .remove = ena_remove, -#ifdef CONFIG_PM - .suspend = ena_suspend, - .resume = ena_resume, -#endif + .shutdown = ena_shutdown, + .driver.pm = &ena_pm_ops, .sriov_configure = pci_sriov_configure_simple, }; static int __init ena_init(void) { - pr_info("%s", version); + int ret; ena_wq = create_singlethread_workqueue(DRV_MODULE_NAME); if (!ena_wq) { @@ -3592,7 +4268,11 @@ static int __init ena_init(void) return -ENOMEM; } - return pci_register_driver(&ena_pci_driver); + ret = pci_register_driver(&ena_pci_driver); + if (ret) + destroy_workqueue(ena_wq); + + return ret; } static void __exit ena_cleanup(void) @@ -3621,7 +4301,7 @@ static void ena_update_on_link_change(void *adapter_data, ENA_ADMIN_AENQ_LINK_CHANGE_DESC_LINK_STATUS_MASK; if (status) { - netdev_dbg(adapter->netdev, "%s\n", __func__); + netif_dbg(adapter, ifup, adapter->netdev, "%s\n", __func__); set_bit(ENA_FLAG_LINK_UP, &adapter->flags); if (!test_bit(ENA_FLAG_ONGOING_RESET, &adapter->flags)) netif_carrier_on(adapter->netdev); @@ -3637,14 +4317,20 @@ static void ena_keep_alive_wd(void *adapter_data, struct ena_adapter *adapter = (struct ena_adapter *)adapter_data; struct ena_admin_aenq_keep_alive_desc *desc; u64 rx_drops; + u64 tx_drops; desc = (struct ena_admin_aenq_keep_alive_desc *)aenq_e; adapter->last_keep_alive_jiffies = jiffies; rx_drops = ((u64)desc->rx_drops_high << 32) | desc->rx_drops_low; + tx_drops = ((u64)desc->tx_drops_high << 32) | desc->tx_drops_low; u64_stats_update_begin(&adapter->syncp); + /* These stats are accumulated by the device, so the counters indicate + * all drops since last reset. + */ adapter->dev_stats.rx_drops = rx_drops; + adapter->dev_stats.tx_drops = tx_drops; u64_stats_update_end(&adapter->syncp); } @@ -3659,7 +4345,7 @@ static void ena_notification(void *adapter_data, aenq_e->aenq_common_desc.group, ENA_ADMIN_NOTIFICATION); - switch (aenq_e->aenq_common_desc.syndrom) { + switch (aenq_e->aenq_common_desc.syndrome) { case ENA_ADMIN_UPDATE_HINTS: hints = (struct ena_admin_ena_hw_hints *) (&aenq_e->inline_data_w4); @@ -3668,7 +4354,7 @@ static void ena_notification(void *adapter_data, default: netif_err(adapter, drv, adapter->netdev, "Invalid aenq notification link state %d\n", - aenq_e->aenq_common_desc.syndrom); + aenq_e->aenq_common_desc.syndrome); } } |