diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 806 |
1 files changed, 633 insertions, 173 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index 0e31bdb4b7cb..fa2a22dfa048 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -1,5 +1,5 @@ /* - * Copyright 2018 Advanced Micro Devices, Inc. + * Copyright 2018-2024 Advanced Micro Devices, Inc. All rights reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -37,6 +37,7 @@ #include "df_v3_6.h" #include "df_v4_3.h" #include "df_v4_6_2.h" +#include "df_v4_15.h" #include "nbio_v6_1.h" #include "nbio_v7_0.h" #include "nbio_v7_4.h" @@ -55,6 +56,7 @@ #include "smuio_v9_0.h" #include "gmc_v10_0.h" #include "gmc_v11_0.h" +#include "gmc_v12_0.h" #include "gfxhub_v2_0.h" #include "mmhub_v2_0.h" #include "nbio_v2_3.h" @@ -68,15 +70,18 @@ #include "hdp_v7_0.h" #include "nv.h" #include "soc21.h" +#include "soc24.h" #include "navi10_ih.h" #include "ih_v6_0.h" #include "ih_v6_1.h" #include "ih_v7_0.h" #include "gfx_v10_0.h" #include "gfx_v11_0.h" +#include "gfx_v12_0.h" #include "sdma_v5_0.h" #include "sdma_v5_2.h" #include "sdma_v6_0.h" +#include "sdma_v7_0.h" #include "lsdma_v6_0.h" #include "lsdma_v7_0.h" #include "vcn_v2_0.h" @@ -90,8 +95,8 @@ #include "vcn_v4_0_5.h" #include "jpeg_v4_0_5.h" #include "amdgpu_vkms.h" -#include "mes_v10_1.h" #include "mes_v11_0.h" +#include "mes_v12_0.h" #include "smuio_v11_0.h" #include "smuio_v11_0_6.h" #include "smuio_v13_0.h" @@ -99,12 +104,25 @@ #include "smuio_v13_0_6.h" #include "smuio_v14_0_2.h" #include "vcn_v5_0_0.h" +#include "vcn_v5_0_1.h" #include "jpeg_v5_0_0.h" +#include "jpeg_v5_0_1.h" +#include "amdgpu_ras_mgr.h" #include "amdgpu_vpe.h" +#if defined(CONFIG_DRM_AMD_ISP) +#include "amdgpu_isp.h" +#endif -#define FIRMWARE_IP_DISCOVERY "amdgpu/ip_discovery.bin" -MODULE_FIRMWARE(FIRMWARE_IP_DISCOVERY); +MODULE_FIRMWARE("amdgpu/ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/vega10_ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/vega12_ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/vega20_ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/raven_ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/raven2_ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/picasso_ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/arcturus_ip_discovery.bin"); +MODULE_FIRMWARE("amdgpu/aldebaran_ip_discovery.bin"); #define mmIP_DISCOVERY_VERSION 0x16A00 #define mmRCC_CONFIG_MEMSIZE 0xde3 @@ -221,6 +239,7 @@ static int hw_id_map[MAX_HWIP] = { [DCI_HWIP] = DCI_HWID, [PCIE_HWIP] = PCIE_HWID, [VPE_HWIP] = VPE_HWID, + [ISP_HWIP] = ISP_HWID, }; static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev, uint8_t *binary) @@ -236,9 +255,9 @@ static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev, pos = tmr_offset + tmr_size - DISCOVERY_TMR_OFFSET; /* This region is read-only and reserved from system use */ - discv_regn = memremap(pos, adev->mman.discovery_tmr_size, MEMREMAP_WC); + discv_regn = memremap(pos, adev->discovery.size, MEMREMAP_WC); if (discv_regn) { - memcpy(binary, discv_regn, adev->mman.discovery_tmr_size); + memcpy(binary, discv_regn, adev->discovery.size); memunmap(discv_regn); return 0; } @@ -252,57 +271,84 @@ static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev, static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, uint8_t *binary) { + bool sz_valid = true; uint64_t vram_size; - u32 msg; int i, ret = 0; + u32 msg; - /* It can take up to a second for IFWI init to complete on some dGPUs, - * but generally it should be in the 60-100ms range. Normally this starts - * as soon as the device gets power so by the time the OS loads this has long - * completed. However, when a card is hotplugged via e.g., USB4, we need to - * wait for this to complete. Once the C2PMSG is updated, we can - * continue. - */ + if (!amdgpu_sriov_vf(adev)) { + /* It can take up to two second for IFWI init to complete on some dGPUs, + * but generally it should be in the 60-100ms range. Normally this starts + * as soon as the device gets power so by the time the OS loads this has long + * completed. However, when a card is hotplugged via e.g., USB4, we need to + * wait for this to complete. Once the C2PMSG is updated, we can + * continue. + */ - for (i = 0; i < 1000; i++) { - msg = RREG32(mmMP0_SMN_C2PMSG_33); - if (msg & 0x80000000) - break; - usleep_range(1000, 1100); + for (i = 0; i < 2000; i++) { + msg = RREG32(mmMP0_SMN_C2PMSG_33); + if (msg & 0x80000000) + break; + msleep(1); + } } - vram_size = (uint64_t)RREG32(mmRCC_CONFIG_MEMSIZE) << 20; + vram_size = RREG32(mmRCC_CONFIG_MEMSIZE); + if (!vram_size || vram_size == U32_MAX) + sz_valid = false; + else + vram_size <<= 20; + + /* + * If in VRAM, discovery TMR is marked for reservation. If it is in system mem, + * then it is not required to be reserved. + */ + if (sz_valid) { + if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) { + /* For SRIOV VFs with dynamic critical region enabled, + * we will get the IPD binary via below call. + * If dynamic critical is disabled, fall through to normal seq. + */ + if (amdgpu_virt_get_dynamic_data_info(adev, + AMD_SRIOV_MSG_IPD_TABLE_ID, binary, + &adev->discovery.size)) { + dev_err(adev->dev, + "failed to read discovery info from dynamic critical region."); + ret = -EINVAL; + goto exit; + } + } else { + uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET; - if (vram_size) { - uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET; - amdgpu_device_vram_access(adev, pos, (uint32_t *)binary, - adev->mman.discovery_tmr_size, false); + amdgpu_device_vram_access(adev, pos, (uint32_t *)binary, + adev->discovery.size, false); + adev->discovery.reserve_tmr = true; + } } else { ret = amdgpu_discovery_read_binary_from_sysmem(adev, binary); } + if (ret) + dev_err(adev->dev, + "failed to read discovery info from memory, vram size read: %llx", + vram_size); +exit: return ret; } -static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev, uint8_t *binary) +static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev, + uint8_t *binary, + const char *fw_name) { const struct firmware *fw; - const char *fw_name; int r; - switch (amdgpu_discovery) { - case 2: - fw_name = FIRMWARE_IP_DISCOVERY; - break; - default: - dev_warn(adev->dev, "amdgpu_discovery is not set properly\n"); - return -EINVAL; - } - - r = request_firmware(&fw, fw_name, adev->dev); + r = firmware_request_nowarn(&fw, fw_name, adev->dev); if (r) { - dev_err(adev->dev, "can't load firmware \"%s\"\n", - fw_name); + if (amdgpu_discovery == 2) + dev_err(adev->dev, "can't load firmware \"%s\"\n", fw_name); + else + drm_info(&adev->ddev, "Optional firmware \"%s\" was not found\n", fw_name); return r; } @@ -362,55 +408,118 @@ static void amdgpu_discovery_harvest_config_quirk(struct amdgpu_device *adev) } } +static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev, + struct binary_header *bhdr) +{ + uint8_t *discovery_bin = adev->discovery.bin; + struct table_info *info; + uint16_t checksum; + uint16_t offset; + + info = &bhdr->table_list[NPS_INFO]; + offset = le16_to_cpu(info->offset); + checksum = le16_to_cpu(info->checksum); + + struct nps_info_header *nhdr = + (struct nps_info_header *)(discovery_bin + offset); + + if (le32_to_cpu(nhdr->table_id) != NPS_INFO_TABLE_ID) { + dev_dbg(adev->dev, "invalid ip discovery nps info table id\n"); + return -EINVAL; + } + + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, + le32_to_cpu(nhdr->size_bytes), + checksum)) { + dev_dbg(adev->dev, "invalid nps info data table checksum\n"); + return -EINVAL; + } + + return 0; +} + +static const char *amdgpu_discovery_get_fw_name(struct amdgpu_device *adev) +{ + if (amdgpu_discovery == 2) { + /* Assume there is valid discovery TMR in VRAM even if binary is sideloaded */ + adev->discovery.reserve_tmr = true; + return "amdgpu/ip_discovery.bin"; + } + + switch (adev->asic_type) { + case CHIP_VEGA10: + return "amdgpu/vega10_ip_discovery.bin"; + case CHIP_VEGA12: + return "amdgpu/vega12_ip_discovery.bin"; + case CHIP_RAVEN: + if (adev->apu_flags & AMD_APU_IS_RAVEN2) + return "amdgpu/raven2_ip_discovery.bin"; + else if (adev->apu_flags & AMD_APU_IS_PICASSO) + return "amdgpu/picasso_ip_discovery.bin"; + else + return "amdgpu/raven_ip_discovery.bin"; + case CHIP_VEGA20: + return "amdgpu/vega20_ip_discovery.bin"; + case CHIP_ARCTURUS: + return "amdgpu/arcturus_ip_discovery.bin"; + case CHIP_ALDEBARAN: + return "amdgpu/aldebaran_ip_discovery.bin"; + default: + return NULL; + } +} + static int amdgpu_discovery_init(struct amdgpu_device *adev) { struct table_info *info; struct binary_header *bhdr; + uint8_t *discovery_bin; + const char *fw_name; uint16_t offset; uint16_t size; uint16_t checksum; int r; - adev->mman.discovery_tmr_size = DISCOVERY_TMR_SIZE; - adev->mman.discovery_bin = kzalloc(adev->mman.discovery_tmr_size, GFP_KERNEL); - if (!adev->mman.discovery_bin) + adev->discovery.bin = kzalloc(DISCOVERY_TMR_SIZE, GFP_KERNEL); + if (!adev->discovery.bin) return -ENOMEM; + adev->discovery.size = DISCOVERY_TMR_SIZE; + adev->discovery.debugfs_blob.data = adev->discovery.bin; + adev->discovery.debugfs_blob.size = adev->discovery.size; + discovery_bin = adev->discovery.bin; /* Read from file if it is the preferred option */ - if (amdgpu_discovery == 2) { - dev_info(adev->dev, "use ip discovery information from file"); - r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin); - - if (r) { - dev_err(adev->dev, "failed to read ip discovery binary from file\n"); - r = -EINVAL; + fw_name = amdgpu_discovery_get_fw_name(adev); + if (fw_name != NULL) { + drm_dbg(&adev->ddev, "use ip discovery information from file"); + r = amdgpu_discovery_read_binary_from_file(adev, discovery_bin, + fw_name); + if (r) goto out; - } - } else { - r = amdgpu_discovery_read_binary_from_mem( - adev, adev->mman.discovery_bin); + drm_dbg(&adev->ddev, "use ip discovery information from memory"); + r = amdgpu_discovery_read_binary_from_mem(adev, discovery_bin); if (r) goto out; } /* check the ip discovery binary signature */ - if (!amdgpu_discovery_verify_binary_signature(adev->mman.discovery_bin)) { + if (!amdgpu_discovery_verify_binary_signature(discovery_bin)) { dev_err(adev->dev, "get invalid ip discovery binary signature\n"); r = -EINVAL; goto out; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = offsetof(struct binary_header, binary_checksum) + sizeof(bhdr->binary_checksum); size = le16_to_cpu(bhdr->binary_size) - offset; checksum = le16_to_cpu(bhdr->binary_checksum); - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - size, checksum)) { + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, size, + checksum)) { dev_err(adev->dev, "invalid ip discovery binary checksum\n"); r = -EINVAL; goto out; @@ -422,15 +531,16 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct ip_discovery_header *ihdr = - (struct ip_discovery_header *)(adev->mman.discovery_bin + offset); + (struct ip_discovery_header *)(discovery_bin + offset); if (le32_to_cpu(ihdr->signature) != DISCOVERY_TABLE_SIGNATURE) { dev_err(adev->dev, "invalid ip discovery data table signature\n"); r = -EINVAL; goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le16_to_cpu(ihdr->size), checksum)) { + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, + le16_to_cpu(ihdr->size), + checksum)) { dev_err(adev->dev, "invalid ip discovery data table checksum\n"); r = -EINVAL; goto out; @@ -443,7 +553,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct gpu_info_header *ghdr = - (struct gpu_info_header *)(adev->mman.discovery_bin + offset); + (struct gpu_info_header *)(discovery_bin + offset); if (le32_to_cpu(ghdr->table_id) != GC_TABLE_ID) { dev_err(adev->dev, "invalid ip discovery gc table id\n"); @@ -451,8 +561,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le32_to_cpu(ghdr->size), checksum)) { + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, + le32_to_cpu(ghdr->size), + checksum)) { dev_err(adev->dev, "invalid gc data table checksum\n"); r = -EINVAL; goto out; @@ -465,7 +576,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct harvest_info_header *hhdr = - (struct harvest_info_header *)(adev->mman.discovery_bin + offset); + (struct harvest_info_header *)(discovery_bin + offset); if (le32_to_cpu(hhdr->signature) != HARVEST_TABLE_SIGNATURE) { dev_err(adev->dev, "invalid ip discovery harvest table signature\n"); @@ -473,8 +584,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - sizeof(struct harvest_table), checksum)) { + if (!amdgpu_discovery_verify_checksum( + discovery_bin + offset, + sizeof(struct harvest_table), checksum)) { dev_err(adev->dev, "invalid harvest data table checksum\n"); r = -EINVAL; goto out; @@ -487,7 +599,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct vcn_info_header *vhdr = - (struct vcn_info_header *)(adev->mman.discovery_bin + offset); + (struct vcn_info_header *)(discovery_bin + offset); if (le32_to_cpu(vhdr->table_id) != VCN_INFO_TABLE_ID) { dev_err(adev->dev, "invalid ip discovery vcn table id\n"); @@ -495,8 +607,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le32_to_cpu(vhdr->size_bytes), checksum)) { + if (!amdgpu_discovery_verify_checksum( + discovery_bin + offset, + le32_to_cpu(vhdr->size_bytes), checksum)) { dev_err(adev->dev, "invalid vcn data table checksum\n"); r = -EINVAL; goto out; @@ -509,7 +622,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (0 && offset) { struct mall_info_header *mhdr = - (struct mall_info_header *)(adev->mman.discovery_bin + offset); + (struct mall_info_header *)(discovery_bin + offset); if (le32_to_cpu(mhdr->table_id) != MALL_INFO_TABLE_ID) { dev_err(adev->dev, "invalid ip discovery mall table id\n"); @@ -517,8 +630,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le32_to_cpu(mhdr->size_bytes), checksum)) { + if (!amdgpu_discovery_verify_checksum( + discovery_bin + offset, + le32_to_cpu(mhdr->size_bytes), checksum)) { dev_err(adev->dev, "invalid mall data table checksum\n"); r = -EINVAL; goto out; @@ -528,8 +642,8 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) return 0; out: - kfree(adev->mman.discovery_bin); - adev->mman.discovery_bin = NULL; + kfree(adev->discovery.bin); + adev->discovery.bin = NULL; if ((amdgpu_discovery != 2) && (RREG32(mmIP_DISCOVERY_VERSION) == 4)) amdgpu_ras_query_boot_status(adev, 4); @@ -541,20 +655,23 @@ static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev); void amdgpu_discovery_fini(struct amdgpu_device *adev) { amdgpu_discovery_sysfs_fini(adev); - kfree(adev->mman.discovery_bin); - adev->mman.discovery_bin = NULL; + kfree(adev->discovery.bin); + adev->discovery.bin = NULL; } -static int amdgpu_discovery_validate_ip(const struct ip_v4 *ip) +static int amdgpu_discovery_validate_ip(struct amdgpu_device *adev, + uint8_t instance, uint16_t hw_id) { - if (ip->instance_number >= HWIP_MAX_INSTANCE) { - DRM_ERROR("Unexpected instance_number (%d) from ip discovery blob\n", - ip->instance_number); + if (instance >= HWIP_MAX_INSTANCE) { + dev_err(adev->dev, + "Unexpected instance_number (%d) from ip discovery blob\n", + instance); return -EINVAL; } - if (le16_to_cpu(ip->hw_id) >= HW_ID_MAX) { - DRM_ERROR("Unexpected hw_id (%d) from ip discovery blob\n", - le16_to_cpu(ip->hw_id)); + if (hw_id >= HW_ID_MAX) { + dev_err(adev->dev, + "Unexpected hw_id (%d) from ip discovery blob\n", + hw_id); return -EINVAL; } @@ -564,36 +681,41 @@ static int amdgpu_discovery_validate_ip(const struct ip_v4 *ip) static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, uint32_t *vcn_harvest_count) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; struct ip_discovery_header *ihdr; struct die_header *dhdr; - struct ip_v4 *ip; + struct ip *ip; uint16_t die_offset, ip_offset, num_dies, num_ips; + uint16_t hw_id; + uint8_t inst; int i, j; - bhdr = (struct binary_header *)adev->mman.discovery_bin; - ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + - le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + bhdr = (struct binary_header *)discovery_bin; + ihdr = (struct ip_discovery_header + *)(discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); num_dies = le16_to_cpu(ihdr->num_dies); /* scan harvest bit of all IP data structures */ for (i = 0; i < num_dies; i++) { die_offset = le16_to_cpu(ihdr->die_info[i].die_offset); - dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips = le16_to_cpu(dhdr->num_ips); ip_offset = die_offset + sizeof(*dhdr); for (j = 0; j < num_ips; j++) { - ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset); - - if (amdgpu_discovery_validate_ip(ip)) + ip = (struct ip *)(discovery_bin + ip_offset); + inst = ip->number_instance; + hw_id = le16_to_cpu(ip->hw_id); + if (amdgpu_discovery_validate_ip(adev, inst, hw_id)) goto next_ip; - if (le16_to_cpu(ip->variant) == 1) { - switch (le16_to_cpu(ip->hw_id)) { + if (ip->harvest == 1) { + switch (hw_id) { case VCN_HWID: (*vcn_harvest_count)++; - if (ip->instance_number == 0) { + if (inst == 0) { adev->vcn.harvest_config |= AMDGPU_VCN_HARVEST_VCN0; adev->vcn.inst_mask &= ~AMDGPU_VCN_HARVEST_VCN0; @@ -615,10 +737,8 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, } } next_ip: - if (ihdr->base_addr_64_bit) - ip_offset += struct_size(ip, base_address_64, ip->num_base_address); - else - ip_offset += struct_size(ip, base_address, ip->num_base_address); + ip_offset += struct_size(ip, base_address, + ip->num_base_address); } } } @@ -627,13 +747,14 @@ static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev, uint32_t *vcn_harvest_count, uint32_t *umc_harvest_count) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; struct harvest_table *harvest_info; u16 offset; int i; uint32_t umc_harvest_config = 0; - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[HARVEST_INFO].offset); if (!offset) { @@ -641,7 +762,7 @@ static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev, return; } - harvest_info = (struct harvest_table *)(adev->mman.discovery_bin + offset); + harvest_info = (struct harvest_table *)(discovery_bin + offset); for (i = 0; i < 32; i++) { if (le16_to_cpu(harvest_info->list[i].hw_id) == 0) @@ -676,6 +797,12 @@ static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev, adev->sdma.sdma_mask &= ~(1U << harvest_info->list[i].number_instance); break; +#if defined(CONFIG_DRM_AMD_ISP) + case ISP_HWID: + adev->isp.harvest_config |= + ~(1U << harvest_info->list[i].number_instance); + break; +#endif default: break; } @@ -931,8 +1058,8 @@ static void ip_disc_release(struct kobject *kobj) kobj); struct amdgpu_device *adev = ip_top->adev; - adev->ip_top = NULL; kfree(ip_top); + adev->discovery.ip_top = NULL; } static uint8_t amdgpu_discovery_get_harvest_info(struct amdgpu_device *adev, @@ -943,7 +1070,9 @@ static uint8_t amdgpu_discovery_get_harvest_info(struct amdgpu_device *adev, /* Until a uniform way is figured, get mask based on hwid */ switch (hw_id) { case VCN_HWID: - harvest = ((1 << inst) & adev->vcn.inst_mask) == 0; + /* VCN vs UVD+VCE */ + if (!amdgpu_ip_version(adev, VCE_HWIP, 0)) + harvest = ((1 << inst) & adev->vcn.inst_mask) == 0; break; case DMU_HWID: if (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK) @@ -970,7 +1099,10 @@ static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev, const size_t _ip_offset, const int num_ips, bool reg_base_64) { + uint8_t *discovery_bin = adev->discovery.bin; int ii, jj, kk, res; + uint16_t hw_id; + uint8_t inst; DRM_DEBUG("num_ips:%d", num_ips); @@ -985,9 +1117,11 @@ static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev, struct ip_v4 *ip; struct ip_hw_instance *ip_hw_instance; - ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset); - if (amdgpu_discovery_validate_ip(ip) || - le16_to_cpu(ip->hw_id) != ii) + ip = (struct ip_v4 *)(discovery_bin + ip_offset); + inst = ip->instance_number; + hw_id = le16_to_cpu(ip->hw_id); + if (amdgpu_discovery_validate_ip(adev, inst, hw_id) || + hw_id != ii) goto next_ip; DRM_DEBUG("match:%d @ ip_offset:%zu", ii, ip_offset); @@ -1070,17 +1204,20 @@ next_ip: static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev) { + struct ip_discovery_top *ip_top = adev->discovery.ip_top; + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; struct ip_discovery_header *ihdr; struct die_header *dhdr; - struct kset *die_kset = &adev->ip_top->die_kset; + struct kset *die_kset = &ip_top->die_kset; u16 num_dies, die_offset, num_ips; size_t ip_offset; int ii, res; - bhdr = (struct binary_header *)adev->mman.discovery_bin; - ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + - le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + bhdr = (struct binary_header *)discovery_bin; + ihdr = (struct ip_discovery_header + *)(discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); num_dies = le16_to_cpu(ihdr->num_dies); DRM_DEBUG("number of dies: %d\n", num_dies); @@ -1089,7 +1226,7 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev) struct ip_die_entry *ip_die_entry; die_offset = le16_to_cpu(ihdr->die_info[ii].die_offset); - dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips = le16_to_cpu(dhdr->num_ips); ip_offset = die_offset + sizeof(*dhdr); @@ -1123,30 +1260,32 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev) static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; + struct ip_discovery_top *ip_top; struct kset *die_kset; int res, ii; - if (!adev->mman.discovery_bin) + if (!discovery_bin) return -EINVAL; - adev->ip_top = kzalloc(sizeof(*adev->ip_top), GFP_KERNEL); - if (!adev->ip_top) + ip_top = kzalloc(sizeof(*ip_top), GFP_KERNEL); + if (!ip_top) return -ENOMEM; - adev->ip_top->adev = adev; - - res = kobject_init_and_add(&adev->ip_top->kobj, &ip_discovery_ktype, + ip_top->adev = adev; + adev->discovery.ip_top = ip_top; + res = kobject_init_and_add(&ip_top->kobj, &ip_discovery_ktype, &adev->dev->kobj, "ip_discovery"); if (res) { DRM_ERROR("Couldn't init and add ip_discovery/"); goto Err; } - die_kset = &adev->ip_top->die_kset; + die_kset = &ip_top->die_kset; kobject_set_name(&die_kset->kobj, "%s", "die"); - die_kset->kobj.parent = &adev->ip_top->kobj; + die_kset->kobj.parent = &ip_top->kobj; die_kset->kobj.ktype = &die_kobj_ktype; - res = kset_register(&adev->ip_top->die_kset); + res = kset_register(&ip_top->die_kset); if (res) { DRM_ERROR("Couldn't register die_kset"); goto Err; @@ -1160,7 +1299,7 @@ static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev) return res; Err: - kobject_put(&adev->ip_top->kobj); + kobject_put(&ip_top->kobj); return res; } @@ -1205,10 +1344,11 @@ static void amdgpu_discovery_sysfs_die_free(struct ip_die_entry *ip_die_entry) static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev) { + struct ip_discovery_top *ip_top = adev->discovery.ip_top; struct list_head *el, *tmp; struct kset *die_kset; - die_kset = &adev->ip_top->die_kset; + die_kset = &ip_top->die_kset; spin_lock(&die_kset->list_lock); list_for_each_prev_safe(el, tmp, &die_kset->list) { list_del_init(el); @@ -1217,8 +1357,8 @@ static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev) spin_lock(&die_kset->list_lock); } spin_unlock(&die_kset->list_lock); - kobject_put(&adev->ip_top->die_kset.kobj); - kobject_put(&adev->ip_top->kobj); + kobject_put(&ip_top->die_kset.kobj); + kobject_put(&ip_top->kobj); } /* ================================================== */ @@ -1229,35 +1369,39 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) struct binary_header *bhdr; struct ip_discovery_header *ihdr; struct die_header *dhdr; + uint8_t *discovery_bin; struct ip_v4 *ip; uint16_t die_offset; uint16_t ip_offset; uint16_t num_dies; + uint32_t wafl_ver; uint16_t num_ips; + uint16_t hw_id; + uint8_t inst; int hw_ip; int i, j, k; int r; r = amdgpu_discovery_init(adev); - if (r) { - DRM_ERROR("amdgpu_discovery_init failed\n"); + if (r) return r; - } - + discovery_bin = adev->discovery.bin; + wafl_ver = 0; adev->gfx.xcc_mask = 0; adev->sdma.sdma_mask = 0; adev->vcn.inst_mask = 0; adev->jpeg.inst_mask = 0; - bhdr = (struct binary_header *)adev->mman.discovery_bin; - ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + - le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + bhdr = (struct binary_header *)discovery_bin; + ihdr = (struct ip_discovery_header + *)(discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); num_dies = le16_to_cpu(ihdr->num_dies); DRM_DEBUG("number of dies: %d\n", num_dies); for (i = 0; i < num_dies; i++) { die_offset = le16_to_cpu(ihdr->die_info[i].die_offset); - dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips = le16_to_cpu(dhdr->num_ips); ip_offset = die_offset + sizeof(*dhdr); @@ -1271,9 +1415,11 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) le16_to_cpu(dhdr->die_id), num_ips); for (j = 0; j < num_ips; j++) { - ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset); + ip = (struct ip_v4 *)(discovery_bin + ip_offset); - if (amdgpu_discovery_validate_ip(ip)) + inst = ip->instance_number; + hw_id = le16_to_cpu(ip->hw_id); + if (amdgpu_discovery_validate_ip(adev, inst, hw_id)) goto next_ip; num_base_address = ip->num_base_address; @@ -1294,7 +1440,7 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) */ if (adev->vcn.num_vcn_inst < AMDGPU_MAX_VCN_INSTANCES) { - adev->vcn.vcn_config[adev->vcn.num_vcn_inst] = + adev->vcn.inst[adev->vcn.num_vcn_inst].vcn_config = ip->revision & 0xc0; adev->vcn.num_vcn_inst++; adev->vcn.inst_mask |= @@ -1342,6 +1488,10 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) adev->gfx.xcc_mask |= (1U << ip->instance_number); + if (!wafl_ver && le16_to_cpu(ip->hw_id) == WAFLC_HWID) + wafl_ver = IP_VERSION_FULL(ip->major, ip->minor, + ip->revision, 0, 0); + for (k = 0; k < num_base_address; k++) { /* * convert the endianness of base addresses in place, @@ -1407,21 +1557,32 @@ next_ip: } } + if (wafl_ver && !adev->ip_versions[XGMI_HWIP][0]) + adev->ip_versions[XGMI_HWIP][0] = wafl_ver; + return 0; } static void amdgpu_discovery_harvest_ip(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; + struct ip_discovery_header *ihdr; + struct binary_header *bhdr; int vcn_harvest_count = 0; int umc_harvest_count = 0; + uint16_t offset, ihdr_ver; + bhdr = (struct binary_header *)discovery_bin; + offset = le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset); + ihdr = (struct ip_discovery_header *)(discovery_bin + offset); + ihdr_ver = le16_to_cpu(ihdr->version); /* * Harvest table does not fit Navi1x and legacy GPUs, * so read harvest bit per IP data structure to set * harvest configuration. */ if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 2, 0) && - amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 3)) { + ihdr_ver <= 2) { if ((adev->pdev->device == 0x731E && (adev->pdev->revision == 0xC6 || adev->pdev->revision == 0xC7)) || @@ -1453,28 +1614,30 @@ union gc_info { struct gc_info_v1_0 v1; struct gc_info_v1_1 v1_1; struct gc_info_v1_2 v1_2; + struct gc_info_v1_3 v1_3; struct gc_info_v2_0 v2; struct gc_info_v2_1 v2_1; }; static int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; union gc_info *gc_info; u16 offset; - if (!adev->mman.discovery_bin) { + if (!discovery_bin) { DRM_ERROR("ip discovery uninitialized\n"); return -EINVAL; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[GC].offset); if (!offset) return 0; - gc_info = (union gc_info *)(adev->mman.discovery_bin + offset); + gc_info = (union gc_info *)(discovery_bin + offset); switch (le16_to_cpu(gc_info->v1.header.version_major)) { case 1: @@ -1511,6 +1674,16 @@ static int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev) adev->gfx.config.gc_gl1c_size_per_instance = le32_to_cpu(gc_info->v1_2.gc_gl1c_size_per_instance); adev->gfx.config.gc_gl2c_per_gpu = le32_to_cpu(gc_info->v1_2.gc_gl2c_per_gpu); } + if (le16_to_cpu(gc_info->v1.header.version_minor) >= 3) { + adev->gfx.config.gc_tcp_size_per_cu = le32_to_cpu(gc_info->v1_3.gc_tcp_size_per_cu); + adev->gfx.config.gc_tcp_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_tcp_cache_line_size); + adev->gfx.config.gc_instruction_cache_size_per_sqc = le32_to_cpu(gc_info->v1_3.gc_instruction_cache_size_per_sqc); + adev->gfx.config.gc_instruction_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_instruction_cache_line_size); + adev->gfx.config.gc_scalar_data_cache_size_per_sqc = le32_to_cpu(gc_info->v1_3.gc_scalar_data_cache_size_per_sqc); + adev->gfx.config.gc_scalar_data_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_scalar_data_cache_line_size); + adev->gfx.config.gc_tcc_size = le32_to_cpu(gc_info->v1_3.gc_tcc_size); + adev->gfx.config.gc_tcc_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_tcc_cache_line_size); + } break; case 2: adev->gfx.config.max_shader_engines = le32_to_cpu(gc_info->v2.gc_num_se); @@ -1557,24 +1730,25 @@ union mall_info { static int amdgpu_discovery_get_mall_info(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; union mall_info *mall_info; u32 u, mall_size_per_umc, m_s_present, half_use; u64 mall_size; u16 offset; - if (!adev->mman.discovery_bin) { + if (!discovery_bin) { DRM_ERROR("ip discovery uninitialized\n"); return -EINVAL; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[MALL_INFO].offset); if (!offset) return 0; - mall_info = (union mall_info *)(adev->mman.discovery_bin + offset); + mall_info = (union mall_info *)(discovery_bin + offset); switch (le16_to_cpu(mall_info->v1.header.version_major)) { case 1: @@ -1595,7 +1769,7 @@ static int amdgpu_discovery_get_mall_info(struct amdgpu_device *adev) break; case 2: mall_size_per_umc = le32_to_cpu(mall_info->v2.mall_size_per_umc); - adev->gmc.mall_size = mall_size_per_umc * adev->gmc.num_umc; + adev->gmc.mall_size = (uint64_t)mall_size_per_umc * adev->gmc.num_umc; break; default: dev_err(adev->dev, @@ -1613,12 +1787,13 @@ union vcn_info { static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; union vcn_info *vcn_info; u16 offset; int v; - if (!adev->mman.discovery_bin) { + if (!discovery_bin) { DRM_ERROR("ip discovery uninitialized\n"); return -EINVAL; } @@ -1633,13 +1808,13 @@ static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev) return -EINVAL; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[VCN_INFO].offset); if (!offset) return 0; - vcn_info = (union vcn_info *)(adev->mman.discovery_bin + offset); + vcn_info = (union vcn_info *)(discovery_bin + offset); switch (le16_to_cpu(vcn_info->v1.header.version_major)) { case 1: @@ -1647,7 +1822,7 @@ static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev) * so this won't overflow. */ for (v = 0; v < adev->vcn.num_vcn_inst; v++) { - adev->vcn.vcn_codec_disable_mask[v] = + adev->vcn.inst[v].vcn_codec_disable_mask = le32_to_cpu(vcn_info->v1.instance_info[v].fuse_data.all_bits); } break; @@ -1661,6 +1836,109 @@ static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev) return 0; } +union nps_info { + struct nps_info_v1_0 v1; +}; + +static int amdgpu_discovery_refresh_nps_info(struct amdgpu_device *adev, + union nps_info *nps_data) +{ + uint64_t vram_size, pos, offset; + struct nps_info_header *nhdr; + struct binary_header bhdr; + uint16_t checksum; + + vram_size = (uint64_t)RREG32(mmRCC_CONFIG_MEMSIZE) << 20; + pos = vram_size - DISCOVERY_TMR_OFFSET; + amdgpu_device_vram_access(adev, pos, &bhdr, sizeof(bhdr), false); + + offset = le16_to_cpu(bhdr.table_list[NPS_INFO].offset); + checksum = le16_to_cpu(bhdr.table_list[NPS_INFO].checksum); + + amdgpu_device_vram_access(adev, (pos + offset), nps_data, + sizeof(*nps_data), false); + + nhdr = (struct nps_info_header *)(nps_data); + if (!amdgpu_discovery_verify_checksum((uint8_t *)nps_data, + le32_to_cpu(nhdr->size_bytes), + checksum)) { + dev_err(adev->dev, "nps data refresh, checksum mismatch\n"); + return -EINVAL; + } + + return 0; +} + +int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev, + uint32_t *nps_type, + struct amdgpu_gmc_memrange **ranges, + int *range_cnt, bool refresh) +{ + uint8_t *discovery_bin = adev->discovery.bin; + struct amdgpu_gmc_memrange *mem_ranges; + struct binary_header *bhdr; + union nps_info *nps_info; + union nps_info nps_data; + u16 offset; + int i, r; + + if (!nps_type || !range_cnt || !ranges) + return -EINVAL; + + if (refresh) { + r = amdgpu_discovery_refresh_nps_info(adev, &nps_data); + if (r) + return r; + nps_info = &nps_data; + } else { + if (!discovery_bin) { + dev_err(adev->dev, + "fetch mem range failed, ip discovery uninitialized\n"); + return -EINVAL; + } + + bhdr = (struct binary_header *)discovery_bin; + offset = le16_to_cpu(bhdr->table_list[NPS_INFO].offset); + + if (!offset) + return -ENOENT; + + /* If verification fails, return as if NPS table doesn't exist */ + if (amdgpu_discovery_verify_npsinfo(adev, bhdr)) + return -ENOENT; + + nps_info = (union nps_info *)(discovery_bin + offset); + } + + switch (le16_to_cpu(nps_info->v1.header.version_major)) { + case 1: + mem_ranges = kvcalloc(nps_info->v1.count, + sizeof(*mem_ranges), + GFP_KERNEL); + if (!mem_ranges) + return -ENOMEM; + *nps_type = nps_info->v1.nps_type; + *range_cnt = nps_info->v1.count; + for (i = 0; i < *range_cnt; i++) { + mem_ranges[i].base_address = + nps_info->v1.instance_info[i].base_address; + mem_ranges[i].limit_address = + nps_info->v1.instance_info[i].limit_address; + mem_ranges[i].nid_mask = -1; + mem_ranges[i].flags = 0; + } + *ranges = mem_ranges; + break; + default: + dev_err(adev->dev, "Unhandled NPS info table %d.%d\n", + le16_to_cpu(nps_info->v1.header.version_major), + le16_to_cpu(nps_info->v1.header.version_minor)); + return -EINVAL; + } + + return 0; +} + static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev) { /* what IP to use for this? */ @@ -1674,6 +1952,8 @@ static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(9, 4, 1): case IP_VERSION(9, 4, 2): case IP_VERSION(9, 4, 3): + case IP_VERSION(9, 4, 4): + case IP_VERSION(9, 5, 0): amdgpu_device_ip_block_add(adev, &vega10_common_ip_block); break; case IP_VERSION(10, 1, 10): @@ -1698,8 +1978,14 @@ static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 4): case IP_VERSION(11, 5, 0): case IP_VERSION(11, 5, 1): + case IP_VERSION(11, 5, 2): + case IP_VERSION(11, 5, 3): amdgpu_device_ip_block_add(adev, &soc21_common_ip_block); break; + case IP_VERSION(12, 0, 0): + case IP_VERSION(12, 0, 1): + amdgpu_device_ip_block_add(adev, &soc24_common_ip_block); + break; default: dev_err(adev->dev, "Failed to add common ip block(GC_HWIP:0x%x)\n", @@ -1722,6 +2008,8 @@ static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(9, 4, 1): case IP_VERSION(9, 4, 2): case IP_VERSION(9, 4, 3): + case IP_VERSION(9, 4, 4): + case IP_VERSION(9, 5, 0): amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block); break; case IP_VERSION(10, 1, 10): @@ -1746,8 +2034,14 @@ static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 4): case IP_VERSION(11, 5, 0): case IP_VERSION(11, 5, 1): + case IP_VERSION(11, 5, 2): + case IP_VERSION(11, 5, 3): amdgpu_device_ip_block_add(adev, &gmc_v11_0_ip_block); break; + case IP_VERSION(12, 0, 0): + case IP_VERSION(12, 0, 1): + amdgpu_device_ip_block_add(adev, &gmc_v12_0_ip_block); + break; default: dev_err(adev->dev, "Failed to add gmc ip block(GC_HWIP:0x%x)\n", amdgpu_ip_version(adev, GC_HWIP, 0)); @@ -1770,6 +2064,7 @@ static int amdgpu_discovery_set_ih_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(4, 2, 1): case IP_VERSION(4, 4, 0): case IP_VERSION(4, 4, 2): + case IP_VERSION(4, 4, 5): amdgpu_device_ip_block_add(adev, &vega20_ih_ip_block); break; case IP_VERSION(5, 0, 0): @@ -1820,6 +2115,7 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 12): case IP_VERSION(11, 0, 13): case IP_VERSION(11, 5, 0): + case IP_VERSION(11, 5, 2): amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block); break; case IP_VERSION(11, 0, 8): @@ -1839,8 +2135,11 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(13, 0, 8): case IP_VERSION(13, 0, 10): case IP_VERSION(13, 0, 11): + case IP_VERSION(13, 0, 12): + case IP_VERSION(13, 0, 14): case IP_VERSION(14, 0, 0): case IP_VERSION(14, 0, 1): + case IP_VERSION(14, 0, 4): amdgpu_device_ip_block_add(adev, &psp_v13_0_ip_block); break; case IP_VERSION(13, 0, 4): @@ -1848,6 +2147,7 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev) break; case IP_VERSION(14, 0, 2): case IP_VERSION(14, 0, 3): + case IP_VERSION(14, 0, 5): amdgpu_device_ip_block_add(adev, &psp_v14_0_ip_block); break; default: @@ -1875,13 +2175,17 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 5): case IP_VERSION(11, 0, 9): case IP_VERSION(11, 0, 7): - case IP_VERSION(11, 0, 8): case IP_VERSION(11, 0, 11): case IP_VERSION(11, 0, 12): case IP_VERSION(11, 0, 13): case IP_VERSION(11, 5, 0): + case IP_VERSION(11, 5, 2): amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block); break; + case IP_VERSION(11, 0, 8): + if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) + amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block); + break; case IP_VERSION(12, 0, 0): case IP_VERSION(12, 0, 1): amdgpu_device_ip_block_add(adev, &smu_v12_0_ip_block); @@ -1897,12 +2201,16 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(13, 0, 8): case IP_VERSION(13, 0, 10): case IP_VERSION(13, 0, 11): + case IP_VERSION(13, 0, 14): + case IP_VERSION(13, 0, 12): amdgpu_device_ip_block_add(adev, &smu_v13_0_ip_block); break; case IP_VERSION(14, 0, 0): case IP_VERSION(14, 0, 1): case IP_VERSION(14, 0, 2): case IP_VERSION(14, 0, 3): + case IP_VERSION(14, 0, 4): + case IP_VERSION(14, 0, 5): amdgpu_device_ip_block_add(adev, &smu_v14_0_ip_block); break; default: @@ -1954,6 +2262,12 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(3, 2, 1): case IP_VERSION(3, 5, 0): case IP_VERSION(3, 5, 1): + case IP_VERSION(3, 6, 0): + case IP_VERSION(4, 1, 0): + /* TODO: Fix IP version. DC code expects version 4.0.1 */ + if (adev->ip_versions[DCE_HWIP][0] == IP_VERSION(4, 1, 0)) + adev->ip_versions[DCE_HWIP][0] = IP_VERSION(4, 0, 1); + if (amdgpu_sriov_vf(adev)) amdgpu_discovery_set_sriov_display(adev); else @@ -2000,6 +2314,8 @@ static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block); break; case IP_VERSION(9, 4, 3): + case IP_VERSION(9, 4, 4): + case IP_VERSION(9, 5, 0): amdgpu_device_ip_block_add(adev, &gfx_v9_4_3_ip_block); break; case IP_VERSION(10, 1, 10): @@ -2024,8 +2340,14 @@ static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 4): case IP_VERSION(11, 5, 0): case IP_VERSION(11, 5, 1): + case IP_VERSION(11, 5, 2): + case IP_VERSION(11, 5, 3): amdgpu_device_ip_block_add(adev, &gfx_v11_0_ip_block); break; + case IP_VERSION(12, 0, 0): + case IP_VERSION(12, 0, 1): + amdgpu_device_ip_block_add(adev, &gfx_v12_0_ip_block); + break; default: dev_err(adev->dev, "Failed to add gfx ip block(GC_HWIP:0x%x)\n", amdgpu_ip_version(adev, GC_HWIP, 0)); @@ -2048,6 +2370,8 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block); break; case IP_VERSION(4, 4, 2): + case IP_VERSION(4, 4, 5): + case IP_VERSION(4, 4, 4): amdgpu_device_ip_block_add(adev, &sdma_v4_4_2_ip_block); break; case IP_VERSION(5, 0, 0): @@ -2072,14 +2396,35 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(6, 0, 3): case IP_VERSION(6, 1, 0): case IP_VERSION(6, 1, 1): + case IP_VERSION(6, 1, 2): + case IP_VERSION(6, 1, 3): amdgpu_device_ip_block_add(adev, &sdma_v6_0_ip_block); break; + case IP_VERSION(7, 0, 0): + case IP_VERSION(7, 0, 1): + amdgpu_device_ip_block_add(adev, &sdma_v7_0_ip_block); + break; default: dev_err(adev->dev, "Failed to add sdma ip block(SDMA0_HWIP:0x%x)\n", amdgpu_ip_version(adev, SDMA0_HWIP, 0)); return -EINVAL; } + + return 0; +} + +static int amdgpu_discovery_set_ras_ip_blocks(struct amdgpu_device *adev) +{ + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { + case IP_VERSION(13, 0, 6): + case IP_VERSION(13, 0, 12): + case IP_VERSION(13, 0, 14): + amdgpu_device_ip_block_add(adev, &ras_v1_0_ip_block); + break; + default: + break; + } return 0; } @@ -2166,6 +2511,10 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &vcn_v5_0_0_ip_block); amdgpu_device_ip_block_add(adev, &jpeg_v5_0_0_ip_block); break; + case IP_VERSION(5, 0, 1): + amdgpu_device_ip_block_add(adev, &vcn_v5_0_1_ip_block); + amdgpu_device_ip_block_add(adev, &jpeg_v5_0_1_ip_block); + break; default: dev_err(adev->dev, "Failed to add vcn/jpeg ip block(UVD_HWIP:0x%x)\n", @@ -2179,25 +2528,6 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_mes_ip_blocks(struct amdgpu_device *adev) { switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { - case IP_VERSION(10, 1, 10): - case IP_VERSION(10, 1, 1): - case IP_VERSION(10, 1, 2): - case IP_VERSION(10, 1, 3): - case IP_VERSION(10, 1, 4): - case IP_VERSION(10, 3, 0): - case IP_VERSION(10, 3, 1): - case IP_VERSION(10, 3, 2): - case IP_VERSION(10, 3, 3): - case IP_VERSION(10, 3, 4): - case IP_VERSION(10, 3, 5): - case IP_VERSION(10, 3, 6): - if (amdgpu_mes) { - amdgpu_device_ip_block_add(adev, &mes_v10_1_ip_block); - adev->enable_mes = true; - if (amdgpu_mes_kiq) - adev->enable_mes_kiq = true; - } - break; case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 1): case IP_VERSION(11, 0, 2): @@ -2205,10 +2535,20 @@ static int amdgpu_discovery_set_mes_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 4): case IP_VERSION(11, 5, 0): case IP_VERSION(11, 5, 1): + case IP_VERSION(11, 5, 2): + case IP_VERSION(11, 5, 3): amdgpu_device_ip_block_add(adev, &mes_v11_0_ip_block); adev->enable_mes = true; adev->enable_mes_kiq = true; break; + case IP_VERSION(12, 0, 0): + case IP_VERSION(12, 0, 1): + amdgpu_device_ip_block_add(adev, &mes_v12_0_ip_block); + adev->enable_mes = true; + adev->enable_mes_kiq = true; + if (amdgpu_uni_mes) + adev->enable_uni_mes = true; + break; default: break; } @@ -2219,6 +2559,8 @@ static void amdgpu_discovery_init_soc_config(struct amdgpu_device *adev) { switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 3): + case IP_VERSION(9, 4, 4): + case IP_VERSION(9, 5, 0): aqua_vanjaram_init_soc_config(adev); break; default: @@ -2231,6 +2573,7 @@ static int amdgpu_discovery_set_vpe_ip_blocks(struct amdgpu_device *adev) switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) { case IP_VERSION(6, 1, 0): case IP_VERSION(6, 1, 1): + case IP_VERSION(6, 1, 3): amdgpu_device_ip_block_add(adev, &vpe_v6_1_ip_block); break; default: @@ -2257,15 +2600,40 @@ static int amdgpu_discovery_set_umsch_mm_ip_blocks(struct amdgpu_device *adev) return 0; } +static int amdgpu_discovery_set_isp_ip_blocks(struct amdgpu_device *adev) +{ +#if defined(CONFIG_DRM_AMD_ISP) + switch (amdgpu_ip_version(adev, ISP_HWIP, 0)) { + case IP_VERSION(4, 1, 0): + amdgpu_device_ip_block_add(adev, &isp_v4_1_0_ip_block); + break; + case IP_VERSION(4, 1, 1): + amdgpu_device_ip_block_add(adev, &isp_v4_1_1_ip_block); + break; + default: + break; + } +#endif + + return 0; +} + int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) { int r; switch (adev->asic_type) { case CHIP_VEGA10: + /* This is not fatal. We only need the discovery + * binary for sysfs. We don't need it for a + * functional system. + */ + amdgpu_discovery_init(adev); vega10_reg_base_init(adev); adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; adev->gmc.num_umc = 4; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 0, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 0, 0); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 0); @@ -2285,9 +2653,16 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 0, 0); break; case CHIP_VEGA12: + /* This is not fatal. We only need the discovery + * binary for sysfs. We don't need it for a + * functional system. + */ + amdgpu_discovery_init(adev); vega10_reg_base_init(adev); adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; adev->gmc.num_umc = 4; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 3, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 3, 0); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 1); @@ -2307,10 +2682,17 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 0, 1); break; case CHIP_RAVEN: + /* This is not fatal. We only need the discovery + * binary for sysfs. We don't need it for a + * functional system. + */ + amdgpu_discovery_init(adev); vega10_reg_base_init(adev); adev->sdma.num_instances = 1; + adev->sdma.sdma_mask = 1; adev->vcn.num_vcn_inst = 1; adev->gmc.num_umc = 2; + adev->gfx.xcc_mask = 1; if (adev->apu_flags & AMD_APU_IS_RAVEN2) { adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 2, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 2, 0); @@ -2327,6 +2709,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 2, 2); adev->ip_versions[UVD_HWIP][0] = IP_VERSION(1, 0, 1); adev->ip_versions[DCE_HWIP][0] = IP_VERSION(1, 0, 1); + adev->ip_versions[ISP_HWIP][0] = IP_VERSION(2, 0, 0); } else { adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 1, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 1, 0); @@ -2343,12 +2726,20 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[GC_HWIP][0] = IP_VERSION(9, 1, 0); adev->ip_versions[UVD_HWIP][0] = IP_VERSION(1, 0, 0); adev->ip_versions[DCE_HWIP][0] = IP_VERSION(1, 0, 0); + adev->ip_versions[ISP_HWIP][0] = IP_VERSION(2, 0, 0); } break; case CHIP_VEGA20: + /* This is not fatal. We only need the discovery + * binary for sysfs. We don't need it for a + * functional system. + */ + amdgpu_discovery_init(adev); vega20_reg_base_init(adev); adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; adev->gmc.num_umc = 8; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 0); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 0); @@ -2369,10 +2760,17 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[DCI_HWIP][0] = IP_VERSION(12, 1, 0); break; case CHIP_ARCTURUS: + /* This is not fatal. We only need the discovery + * binary for sysfs. We don't need it for a + * functional system. + */ + amdgpu_discovery_init(adev); arct_reg_base_init(adev); adev->sdma.num_instances = 8; + adev->sdma.sdma_mask = 0xff; adev->vcn.num_vcn_inst = 2; adev->gmc.num_umc = 8; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 1); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 1); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 1); @@ -2397,10 +2795,17 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[UVD_HWIP][1] = IP_VERSION(2, 5, 0); break; case CHIP_ALDEBARAN: + /* This is not fatal. We only need the discovery + * binary for sysfs. We don't need it for a + * functional system. + */ + amdgpu_discovery_init(adev); aldebaran_reg_base_init(adev); adev->sdma.num_instances = 5; + adev->sdma.sdma_mask = 0x1f; adev->vcn.num_vcn_inst = 2; adev->gmc.num_umc = 4; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 2); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 2); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 4, 0); @@ -2422,10 +2827,44 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->ip_versions[UVD_HWIP][1] = IP_VERSION(2, 6, 0); adev->ip_versions[XGMI_HWIP][0] = IP_VERSION(6, 1, 0); break; + case CHIP_CYAN_SKILLFISH: + if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) { + r = amdgpu_discovery_reg_base_init(adev); + if (r) + return -EINVAL; + + amdgpu_discovery_harvest_ip(adev); + amdgpu_discovery_get_gfx_info(adev); + amdgpu_discovery_get_mall_info(adev); + amdgpu_discovery_get_vcn_info(adev); + } else { + cyan_skillfish_reg_base_init(adev); + adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; + adev->gfx.xcc_mask = 1; + adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(2, 0, 3); + adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(2, 0, 3); + adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(5, 0, 1); + adev->ip_versions[HDP_HWIP][0] = IP_VERSION(5, 0, 1); + adev->ip_versions[SDMA0_HWIP][0] = IP_VERSION(5, 0, 1); + adev->ip_versions[SDMA1_HWIP][1] = IP_VERSION(5, 0, 1); + adev->ip_versions[DF_HWIP][0] = IP_VERSION(3, 5, 0); + adev->ip_versions[NBIO_HWIP][0] = IP_VERSION(2, 1, 1); + adev->ip_versions[UMC_HWIP][0] = IP_VERSION(8, 1, 1); + adev->ip_versions[MP0_HWIP][0] = IP_VERSION(11, 0, 8); + adev->ip_versions[MP1_HWIP][0] = IP_VERSION(11, 0, 8); + adev->ip_versions[THM_HWIP][0] = IP_VERSION(11, 0, 1); + adev->ip_versions[SMUIO_HWIP][0] = IP_VERSION(11, 0, 8); + adev->ip_versions[GC_HWIP][0] = IP_VERSION(10, 1, 3); + adev->ip_versions[UVD_HWIP][0] = IP_VERSION(2, 0, 3); + } + break; default: r = amdgpu_discovery_reg_base_init(adev); - if (r) - return -EINVAL; + if (r) { + drm_err(&adev->ddev, "discovery failed: %d\n", r); + return r; + } amdgpu_discovery_harvest_ip(adev); amdgpu_discovery_get_gfx_info(adev); @@ -2444,6 +2883,8 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(9, 4, 1): case IP_VERSION(9, 4, 2): case IP_VERSION(9, 4, 3): + case IP_VERSION(9, 4, 4): + case IP_VERSION(9, 5, 0): adev->family = AMDGPU_FAMILY_AI; break; case IP_VERSION(9, 1, 0): @@ -2486,8 +2927,14 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; case IP_VERSION(11, 5, 0): case IP_VERSION(11, 5, 1): + case IP_VERSION(11, 5, 2): + case IP_VERSION(11, 5, 3): adev->family = AMDGPU_FAMILY_GC_11_5_0; break; + case IP_VERSION(12, 0, 0): + case IP_VERSION(12, 0, 1): + adev->family = AMDGPU_FAMILY_GC_12_0_0; + break; default: return -EINVAL; } @@ -2506,18 +2953,14 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 4): case IP_VERSION(11, 5, 0): case IP_VERSION(11, 5, 1): + case IP_VERSION(11, 5, 2): + case IP_VERSION(11, 5, 3): adev->flags |= AMD_IS_APU; break; default: break; } - if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == IP_VERSION(4, 8, 0)) - adev->gmc.xgmi.supported = true; - - if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) - adev->ip_versions[XGMI_HWIP][0] = IP_VERSION(6, 4, 0); - /* set NBIO version */ switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(6, 1, 0): @@ -2538,11 +2981,14 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->nbio.hdp_flush_reg = &nbio_v7_4_hdp_flush_reg; break; case IP_VERSION(7, 9, 0): + case IP_VERSION(7, 9, 1): adev->nbio.funcs = &nbio_v7_9_funcs; adev->nbio.hdp_flush_reg = &nbio_v7_9_hdp_flush_reg; break; case IP_VERSION(7, 11, 0): case IP_VERSION(7, 11, 1): + case IP_VERSION(7, 11, 2): + case IP_VERSION(7, 11, 3): adev->nbio.funcs = &nbio_v7_11_funcs; adev->nbio.hdp_flush_reg = &nbio_v7_11_hdp_flush_reg; break; @@ -2596,6 +3042,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(4, 2, 1): case IP_VERSION(4, 4, 0): case IP_VERSION(4, 4, 2): + case IP_VERSION(4, 4, 5): adev->hdp.funcs = &hdp_v4_0_funcs; break; case IP_VERSION(5, 0, 0): @@ -2640,6 +3087,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(4, 6, 2): adev->df.funcs = &df_v4_6_2_funcs; break; + case IP_VERSION(4, 15, 0): + case IP_VERSION(4, 15, 1): + adev->df.funcs = &df_v4_15_funcs; + break; default: break; } @@ -2664,6 +3115,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(11, 0, 10): case IP_VERSION(11, 0, 11): case IP_VERSION(11, 5, 0): + case IP_VERSION(11, 5, 2): case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 9): case IP_VERSION(13, 0, 10): @@ -2673,6 +3125,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) adev->smuio.funcs = &smuio_v13_0_funcs; break; case IP_VERSION(13, 0, 3): + case IP_VERSION(13, 0, 11): adev->smuio.funcs = &smuio_v13_0_3_funcs; if (adev->smuio.funcs->get_pkg_type(adev) == AMDGPU_PKG_TYPE_APU) { adev->flags |= AMD_IS_APU; @@ -2752,6 +3205,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) if (r) return r; + r = amdgpu_discovery_set_ras_ip_blocks(adev); + if (r) + return r; + if ((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT && !amdgpu_sriov_vf(adev)) || (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO && amdgpu_dpm == 1)) { @@ -2776,6 +3233,9 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) if (r) return r; + r = amdgpu_discovery_set_isp_ip_blocks(adev); + if (r) + return r; return 0; } |
