From b758667f55a09bc86ac0cb230c4e0e5ca931a3a5 Mon Sep 17 00:00:00 2001 From: Mangesh Gadre Date: Tue, 13 May 2025 13:20:20 +0800 Subject: drm/amdgpu: update ras support check update ras support check for vcn 5.0.1 Signed-off-by: Mangesh Gadre Reviewed-by: Stanley.Yang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index dc07936d2fcb..8a13a0c7b9cd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3708,7 +3708,8 @@ static void amdgpu_ras_query_ras_capablity_from_vbios(struct amdgpu_device *adev */ if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) || amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 0) || - amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3)) + amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3) || + amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(5, 0, 1)) adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN | 1 << AMDGPU_RAS_BLOCK__JPEG); else -- cgit v1.2.3 From 31e837d242cbb58afed8e0d2a722745bc50ad154 Mon Sep 17 00:00:00 2001 From: ganglxie Date: Thu, 22 May 2025 14:28:52 +0800 Subject: drm/amdgpu: handle old RAS eeprom data in non-nps1 mode Get MCA address from PA in nps1, then convert MCA address to PA in specific nps mode. Signed-off-by: ganglxie Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 16 ++++++++++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 23 +++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h | 2 ++ 3 files changed, 39 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 8a13a0c7b9cd..b0b1dbd4ae35 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2886,8 +2886,20 @@ static int __amdgpu_ras_convert_rec_from_rom(struct amdgpu_device *adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT)) return -EINVAL; } else { - if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data)) - return -EINVAL; + if (bps->address) { + if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data)) + return -EINVAL; + } else { + /* for specific old eeprom data, mca address is not stored, + * calc it from pa + */ + if (amdgpu_umc_pa2mca(adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT, + &(bps->address), AMDGPU_NPS1_PARTITION_MODE)) + return -EINVAL; + + if (amdgpu_ras_mca2pa(adev, bps, err_data)) + return -EOPNOTSUPP; + } } return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 8c6e55b5b967..c92b8794aa73 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -562,3 +562,26 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev, return 0; } + +int amdgpu_umc_pa2mca(struct amdgpu_device *adev, + uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps) +{ + struct ta_ras_query_address_input addr_in; + struct ta_ras_query_address_output addr_out; + int ret; + + /* nps: the pa belongs to */ + addr_in.pa.pa = pa | ((uint64_t)nps << 58); + addr_in.addr_type = TA_RAS_PA_TO_MCA; + ret = psp_ras_query_address(&adev->psp, &addr_in, &addr_out); + if (ret) { + dev_warn(adev->dev, "Failed to query RAS MCA address for 0x%llx", + pa); + + return ret; + } + + *mca = addr_out.ma.err_addr; + + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 29ce6b1d214a..ec203f9e5ffa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -189,4 +189,6 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev, uint64_t err_addr, uint32_t ch, uint32_t umc, uint32_t node, uint32_t socket, struct ta_ras_query_address_output *addr_out, bool dump_addr); +int amdgpu_umc_pa2mca(struct amdgpu_device *adev, + uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps); #endif -- cgit v1.2.3 From fce0afca3562ceb5386241159280590beeee2f51 Mon Sep 17 00:00:00 2001 From: ganglxie Date: Thu, 22 May 2025 15:14:28 +0800 Subject: drm/amdgpu: Get mca address for old eeprom records after getting mca address for old eeprom records with 'address==0', it can be correctly parsed under none-nps1, or it will be dropped. Signed-off-by: ganglxie Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index b0b1dbd4ae35..de0944947eaf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2859,6 +2859,15 @@ static int __amdgpu_ras_convert_rec_array_from_rom(struct amdgpu_device *adev, return -EINVAL; } } else { + if (bps[0].address == 0) { + /* for specific old eeprom data, mca address is not stored, + * calc it from pa + */ + if (amdgpu_umc_pa2mca(adev, bps[0].retired_page << AMDGPU_GPU_PAGE_SHIFT, + &(bps[0].address), AMDGPU_NPS1_PARTITION_MODE)) + return -EINVAL; + } + if (amdgpu_ras_mca2pa(adev, &bps[0], err_data)) { if (nps == AMDGPU_NPS1_PARTITION_MODE) memcpy(err_data->err_addr, bps, -- cgit v1.2.3