drm/amdgpu: refine ras related message print
authorGuchun Chen <guchun.chen@amd.com>
Fri, 10 Apr 2020 07:51:14 +0000 (15:51 +0800)
committerAlex Deucher <alexander.deucher@amd.com>
Mon, 13 Apr 2020 16:01:50 +0000 (12:01 -0400)
Prefix ras related kernel message logging with PCI
device info by replacing DRM_INFO/WARN/ERROR with
dev_info/warn/err. This can clearly tell user about
GPU device information where ras is. And add some
other ras message printing to make it more clear
and friendly as well.

Suggested-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c

index b0aa4e1ed4df7a8aebf10d1d00fcf384a3693d3c..73ae913aee26e286fab08c8f1c40e969e35abaaf 100644 (file)
@@ -296,7 +296,8 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user *
        int ret = 0;
 
        if (!amdgpu_ras_get_error_query_ready(adev)) {
-               DRM_WARN("RAS WARN: error injection currently inaccessible\n");
+               dev_warn(adev->dev, "RAS WARN: error injection "
+                               "currently inaccessible\n");
                return size;
        }
 
@@ -324,7 +325,8 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user *
                /* umc ce/ue error injection for a bad page is not allowed */
                if ((data.head.block == AMDGPU_RAS_BLOCK__UMC) &&
                    amdgpu_ras_check_bad_page(adev, data.inject.address)) {
-                       DRM_WARN("RAS WARN: 0x%llx has been marked as bad before error injection!\n",
+                       dev_warn(adev->dev, "RAS WARN: 0x%llx has been marked "
+                                       "as bad before error injection!\n",
                                        data.inject.address);
                        break;
                }
@@ -590,7 +592,8 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev,
        if (!amdgpu_ras_intr_triggered()) {
                ret = psp_ras_enable_features(&adev->psp, &info, enable);
                if (ret) {
-                       DRM_ERROR("RAS ERROR: %s %s feature failed ret %d\n",
+                       dev_err(adev->dev, "RAS ERROR: %s %s feature "
+                                       "failed ret %d\n",
                                        enable ? "enable":"disable",
                                        ras_block_str(head->block),
                                        ret);
@@ -632,7 +635,8 @@ int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev,
                        if (ret == -EINVAL) {
                                ret = __amdgpu_ras_feature_enable(adev, head, 1);
                                if (!ret)
-                                       DRM_INFO("RAS INFO: %s setup object\n",
+                                       dev_info(adev->dev,
+                                               "RAS INFO: %s setup object\n",
                                                ras_block_str(head->block));
                        }
                } else {
@@ -758,12 +762,17 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev,
        info->ce_count = obj->err_data.ce_count;
 
        if (err_data.ce_count) {
-               dev_info(adev->dev, "%ld correctable errors detected in %s block\n",
-                        obj->err_data.ce_count, ras_block_str(info->head.block));
+               dev_info(adev->dev, "%ld correctable hardware errors "
+                                       "detected in %s block, no user "
+                                       "action is needed.\n",
+                                       obj->err_data.ce_count,
+                                       ras_block_str(info->head.block));
        }
        if (err_data.ue_count) {
-               dev_info(adev->dev, "%ld uncorrectable errors detected in %s block\n",
-                        obj->err_data.ue_count, ras_block_str(info->head.block));
+               dev_info(adev->dev, "%ld uncorrectable hardware errors "
+                                       "detected in %s block\n",
+                                       obj->err_data.ue_count,
+                                       ras_block_str(info->head.block));
        }
 
        return 0;
@@ -807,13 +816,13 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev,
                ret = psp_ras_trigger_error(&adev->psp, &block_info);
                break;
        default:
-               DRM_INFO("%s error injection is not supported yet\n",
+               dev_info(adev->dev, "%s error injection is not supported yet\n",
                         ras_block_str(info->head.block));
                ret = -EINVAL;
        }
 
        if (ret)
-               DRM_ERROR("RAS ERROR: inject %s error failed ret %d\n",
+               dev_err(adev->dev, "RAS ERROR: inject %s error failed ret %d\n",
                                ras_block_str(info->head.block),
                                ret);
 
@@ -1549,7 +1558,7 @@ static int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev)
                                                        &data->bps[control->num_recs],
                                                        true,
                                                        save_count)) {
-                       DRM_ERROR("Failed to save EEPROM table data!");
+                       dev_err(adev->dev, "Failed to save EEPROM table data!");
                        return -EIO;
                }
 
@@ -1577,7 +1586,7 @@ static int amdgpu_ras_load_bad_pages(struct amdgpu_device *adev)
 
        if (amdgpu_ras_eeprom_process_recods(control, bps, false,
                control->num_recs)) {
-               DRM_ERROR("Failed to load EEPROM table records!");
+               dev_err(adev->dev, "Failed to load EEPROM table records!");
                ret = -EIO;
                goto out;
        }
@@ -1651,7 +1660,8 @@ int amdgpu_ras_reserve_bad_pages(struct amdgpu_device *adev)
                                               AMDGPU_GPU_PAGE_SIZE,
                                               AMDGPU_GEM_DOMAIN_VRAM,
                                               &bo, NULL))
-                       DRM_WARN("RAS WARN: reserve vram for retired page %llx fail\n", bp);
+                       dev_warn(adev->dev, "RAS WARN: reserve vram for "
+                                       "retired page %llx fail\n", bp);
 
                data->bps_bo[i] = bo;
                data->last_reserved = i + 1;
@@ -1739,7 +1749,7 @@ free:
        kfree(*data);
        con->eh_data = NULL;
 out:
-       DRM_WARN("Failed to initialize ras recovery!\n");
+       dev_warn(adev->dev, "Failed to initialize ras recovery!\n");
 
        return ret;
 }
@@ -1801,18 +1811,18 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev,
                return;
 
        if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
-               DRM_INFO("HBM ECC is active.\n");
+               dev_info(adev->dev, "HBM ECC is active.\n");
                *hw_supported |= (1 << AMDGPU_RAS_BLOCK__UMC |
                                1 << AMDGPU_RAS_BLOCK__DF);
        } else
-               DRM_INFO("HBM ECC is not presented.\n");
+               dev_info(adev->dev, "HBM ECC is not presented.\n");
 
        if (amdgpu_atomfirmware_sram_ecc_supported(adev)) {
-               DRM_INFO("SRAM ECC is active.\n");
+               dev_info(adev->dev, "SRAM ECC is active.\n");
                *hw_supported |= ~(1 << AMDGPU_RAS_BLOCK__UMC |
                                1 << AMDGPU_RAS_BLOCK__DF);
        } else
-               DRM_INFO("SRAM ECC is not presented.\n");
+               dev_info(adev->dev, "SRAM ECC is not presented.\n");
 
        /* hw_supported needs to be aligned with RAS block mask. */
        *hw_supported &= AMDGPU_RAS_BLOCK_MASK;
@@ -1869,7 +1879,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
        if (amdgpu_ras_fs_init(adev))
                goto fs_out;
 
-       DRM_INFO("RAS INFO: ras initialized successfully, "
+       dev_info(adev->dev, "RAS INFO: ras initialized successfully, "
                        "hardware ability[%x] ras_mask[%x]\n",
                        con->hw_supported, con->supported);
        return 0;
@@ -2055,7 +2065,8 @@ void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev)
                return;
 
        if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) {
-               DRM_WARN("RAS event of type ERREVENT_ATHUB_INTERRUPT detected!\n");
+               dev_info(adev->dev, "uncorrectable hardware error"
+                       "(ERREVENT_ATHUB_INTERRUPT) detected!\n");
 
                amdgpu_ras_reset_gpu(adev);
        }
index 8ffa015bc1dc681da3430f329b3e4eda26859a5e..af1b1ccf613c98589a266e609951ada186af7ddb 100644 (file)
@@ -110,7 +110,8 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
                 * even NOMEM error is encountered
                 */
                if(!err_data->err_addr)
-                       DRM_WARN("Failed to alloc memory for umc error address record!\n");
+                       dev_warn(adev->dev, "Failed to alloc memory for "
+                                       "umc error address record!\n");
 
                /* umc query_ras_error_address is also responsible for clearing
                 * error status
@@ -120,13 +121,14 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
 
        /* only uncorrectable error needs gpu reset */
        if (err_data->ue_count) {
-               dev_info(adev->dev, "%ld uncorrectable errors detected in UMC block\n",
-                       err_data->ue_count);
+               dev_info(adev->dev, "%ld uncorrectable hardware errors "
+                               "detected in UMC block\n",
+                               err_data->ue_count);
 
                if (err_data->err_addr_cnt &&
                    amdgpu_ras_add_bad_pages(adev, err_data->err_addr,
                                                err_data->err_addr_cnt))
-                       DRM_WARN("Failed to add ras bad page!\n");
+                       dev_warn(adev->dev, "Failed to add ras bad page!\n");
 
                amdgpu_ras_reset_gpu(adev);
        }
index 72bdd06ed50841b1f611a930e4719164ec1a57c0..e629156173d31b46b6e2bada2f5ed2a19710a06a 100644 (file)
@@ -323,14 +323,20 @@ static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device
                obj->err_data.ce_count += err_data.ce_count;
 
                if (err_data.ce_count)
-                       DRM_INFO("%ld correctable errors detected in %s block\n",
-                               obj->err_data.ce_count, adev->nbio.ras_if->name);
+                       dev_info(adev->dev, "%ld correctable hardware "
+                                       "errors detected in %s block, "
+                                       "no user action is needed.\n",
+                                       obj->err_data.ce_count,
+                                       adev->nbio.ras_if->name);
 
                if (err_data.ue_count)
-                       DRM_INFO("%ld uncorrectable errors detected in %s block\n",
-                               obj->err_data.ue_count, adev->nbio.ras_if->name);
+                       dev_info(adev->dev, "%ld uncorrectable hardware "
+                                       "errors detected in %s block\n",
+                                       obj->err_data.ue_count,
+                                       adev->nbio.ras_if->name);
 
-               DRM_WARN("RAS controller interrupt triggered by NBIF error\n");
+               dev_info(adev->dev, "RAS controller interrupt triggered "
+                                       "by NBIF error\n");
 
                /* ras_controller_int is dedicated for nbif ras error,
                 * not the global interrupt for sync flood