diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 82 |
1 files changed, 60 insertions, 22 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 08133de21fdd..8f47c14ecbc7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -867,9 +867,9 @@ static int amdgpu_ras_enable_all_features(struct amdgpu_device *adev, /* feature ctl end */ -void amdgpu_ras_mca_query_error_status(struct amdgpu_device *adev, - struct ras_common_if *ras_block, - struct ras_err_data *err_data) +static void amdgpu_ras_mca_query_error_status(struct amdgpu_device *adev, + struct ras_common_if *ras_block, + struct ras_err_data *err_data) { switch (ras_block->sub_block_index) { case AMDGPU_RAS_MCA_BLOCK__MP0: @@ -892,6 +892,38 @@ void amdgpu_ras_mca_query_error_status(struct amdgpu_device *adev, } } +static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_data *err_data) +{ + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + int ret = 0; + + /* + * choosing right query method according to + * whether smu support query error information + */ + ret = smu_get_ecc_info(&adev->smu, (void *)&(ras->umc_ecc)); + if (ret == -EOPNOTSUPP) { + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_count) + adev->umc.ras_funcs->query_ras_error_count(adev, err_data); + + /* umc query_ras_error_address is also responsible for clearing + * error status + */ + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_address) + adev->umc.ras_funcs->query_ras_error_address(adev, err_data); + } else if (!ret) { + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->ecc_info_query_ras_error_count) + adev->umc.ras_funcs->ecc_info_query_ras_error_count(adev, err_data); + + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->ecc_info_query_ras_error_address) + adev->umc.ras_funcs->ecc_info_query_ras_error_address(adev, err_data); + } +} + /* query/inject/cure begin */ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info) @@ -905,15 +937,7 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, switch (info->head.block) { case AMDGPU_RAS_BLOCK__UMC: - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->query_ras_error_count) - adev->umc.ras_funcs->query_ras_error_count(adev, &err_data); - /* umc query_ras_error_address is also responsible for clearing - * error status - */ - if (adev->umc.ras_funcs && - adev->umc.ras_funcs->query_ras_error_address) - adev->umc.ras_funcs->query_ras_error_address(adev, &err_data); + amdgpu_ras_get_ecc_info(adev, &err_data); break; case AMDGPU_RAS_BLOCK__SDMA: if (adev->sdma.funcs->query_ras_error_count) { @@ -1137,9 +1161,9 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, /** * amdgpu_ras_query_error_count -- Get error counts of all IPs - * adev: pointer to AMD GPU device - * ce_count: pointer to an integer to be set to the count of correctible errors. - * ue_count: pointer to an integer to be set to the count of uncorrectible + * @adev: pointer to AMD GPU device + * @ce_count: pointer to an integer to be set to the count of correctible errors. + * @ue_count: pointer to an integer to be set to the count of uncorrectible * errors. * * If set, @ce_count or @ue_count, count and return the corresponding @@ -1568,6 +1592,7 @@ static void amdgpu_ras_interrupt_handler(struct ras_manager *obj) /* Let IP handle its data, maybe we need get the output * from the callback to udpate the error type/count, etc */ + memset(&err_data, 0, sizeof(err_data)); ret = data->cb(obj->adev, &err_data, &entry); /* ue will trigger an interrupt, and in that case * we need do a reset to recovery the whole system. @@ -1723,6 +1748,16 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev) if (info.head.block == AMDGPU_RAS_BLOCK__PCIE_BIF) continue; + /* + * this is a workaround for aldebaran, skip send msg to + * smu to get ecc_info table due to smu handle get ecc + * info table failed temporarily. + * should be removed until smu fix handle ecc_info table. + */ + if ((info.head.block == AMDGPU_RAS_BLOCK__UMC) && + (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2))) + continue; + amdgpu_ras_query_error_status(adev, &info); } } @@ -1804,8 +1839,7 @@ static int amdgpu_ras_badpages_read(struct amdgpu_device *adev, .size = AMDGPU_GPU_PAGE_SIZE, .flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED, }; - status = amdgpu_vram_mgr_query_page_status( - ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM), + status = amdgpu_vram_mgr_query_page_status(&adev->mman.vram_mgr, data->bps[i].retired_page); if (status == -EBUSY) (*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING; @@ -1906,8 +1940,7 @@ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, goto out; } - amdgpu_vram_mgr_reserve_range( - ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM), + amdgpu_vram_mgr_reserve_range(&adev->mman.vram_mgr, bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT, AMDGPU_GPU_PAGE_SIZE); @@ -1935,9 +1968,11 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev) if (!con || !con->eh_data) return 0; + mutex_lock(&con->recovery_lock); control = &con->eeprom_control; data = con->eh_data; save_count = data->count - control->ras_num_recs; + mutex_unlock(&con->recovery_lock); /* only new entries are saved */ if (save_count > 0) { if (amdgpu_ras_eeprom_append(control, @@ -2336,7 +2371,11 @@ int amdgpu_ras_init(struct amdgpu_device *adev) } /* Init poison supported flag, the default value is false */ - if (adev->df.funcs && + if (adev->gmc.xgmi.connected_to_cpu) { + /* enabled by default when GPU is connected to CPU */ + con->poison_supported = true; + } + else if (adev->df.funcs && adev->df.funcs->query_ras_poison_mode && adev->umc.ras_funcs && adev->umc.ras_funcs->query_ras_poison_mode) { @@ -2477,7 +2516,6 @@ void amdgpu_ras_late_fini(struct amdgpu_device *adev, amdgpu_ras_sysfs_remove(adev, ras_block); if (ih_info->cb) amdgpu_ras_interrupt_remove_handler(adev, ih_info); - amdgpu_ras_feature_enable(adev, ras_block, 0); } /* do some init work after IP late init as dependence. @@ -2647,7 +2685,7 @@ static int amdgpu_bad_page_notifier(struct notifier_block *nb, * and error occurred in DramECC (Extended error code = 0) then only * process the error, else bail out. */ - if (!m || !((smca_get_bank_type(m->bank) == SMCA_UMC_V2) && + if (!m || !((smca_get_bank_type(m->extcpu, m->bank) == SMCA_UMC_V2) && (XEC(m->status, 0x3f) == 0x0))) return NOTIFY_DONE; |