drm/amdgpu: harvest edc status when connected to host via xGMI

When connected to a host via xGMI, system fatal errors may trigger
warm reset, driver has no change to query edc status before reset.
Therefore in this case, driver should harvest previous error loging
registers during boot, instead of only resetting them.

v2:
1. IP's ras_manager object is created when its ras feature is enabled,
so change to query edc status after amdgpu_ras_late_init called

2. change to enable watchdog timer after finishing gfx edc init

Signed-off-by: Dennis Li <Dennis.Li@amd.com>
Reivewed-by: Hawking Zhang <hawking.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

1 files changed, 41 insertions, 9 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index c1516d871881..ed83a32f6f30 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -109,7 +109,7 @@ static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf,
 	ssize_t s;
 	char val[128];
 
-	if (amdgpu_ras_error_query(obj->adev, &info))
+	if (amdgpu_ras_query_error_status(obj->adev, &info))
 		return -EINVAL;
 
 	s = snprintf(val, sizeof(val), "%s: %lu\n%s: %lu\n",
@@ -434,7 +434,7 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev,
 		return snprintf(buf, PAGE_SIZE,
 				"Query currently inaccessible\n");
 
-	if (amdgpu_ras_error_query(obj->adev, &info))
+	if (amdgpu_ras_query_error_status(obj->adev, &info))
 		return -EINVAL;
 
 	return snprintf(buf, PAGE_SIZE, "%s: %lu\n%s: %lu\n",
@@ -757,8 +757,8 @@ static int amdgpu_ras_enable_all_features(struct amdgpu_device *adev,
 /* feature ctl end */
 
 /* query/inject/cure begin */
-int amdgpu_ras_error_query(struct amdgpu_device *adev,
-		struct ras_query_if *info)
+int amdgpu_ras_query_error_status(struct amdgpu_device *adev,
+	struct ras_query_if *info)
 {
 	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
 	struct ras_err_data err_data = {0, 0, 0, NULL};
@@ -787,10 +787,16 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev,
 	case AMDGPU_RAS_BLOCK__GFX:
 		if (adev->gfx.funcs->query_ras_error_count)
 			adev->gfx.funcs->query_ras_error_count(adev, &err_data);
+
+		if (adev->gfx.funcs->query_ras_error_status)
+			adev->gfx.funcs->query_ras_error_status(adev);
 		break;
 	case AMDGPU_RAS_BLOCK__MMHUB:
 		if (adev->mmhub.funcs->query_ras_error_count)
 			adev->mmhub.funcs->query_ras_error_count(adev, &err_data);
+
+		if (adev->mmhub.funcs->query_ras_error_status)
+			adev->mmhub.funcs->query_ras_error_status(adev);
 		break;
 	case AMDGPU_RAS_BLOCK__PCIE_BIF:
 		if (adev->nbio.funcs->query_ras_error_count)
@@ -826,6 +832,35 @@ int amdgpu_ras_error_query(struct amdgpu_device *adev,
 	return 0;
 }
 
+int amdgpu_ras_reset_error_status(struct amdgpu_device *adev,
+		enum amdgpu_ras_block block)
+{
+	if (!amdgpu_ras_is_supported(adev, block))
+		return -EINVAL;
+
+	switch (block) {
+	case AMDGPU_RAS_BLOCK__GFX:
+		if (adev->gfx.funcs->reset_ras_error_count)
+			adev->gfx.funcs->reset_ras_error_count(adev);
+
+		if (adev->gfx.funcs->reset_ras_error_status)
+			adev->gfx.funcs->reset_ras_error_status(adev);
+		break;
+	case AMDGPU_RAS_BLOCK__MMHUB:
+		if (adev->mmhub.funcs->reset_ras_error_count)
+			adev->mmhub.funcs->reset_ras_error_count(adev);
+		break;
+	case AMDGPU_RAS_BLOCK__SDMA:
+		if (adev->sdma.funcs->reset_ras_error_count)
+			adev->sdma.funcs->reset_ras_error_count(adev);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
 /* Trigger XGMI/WAFL error */
 static int amdgpu_ras_error_inject_xgmi(struct amdgpu_device *adev,
 				 struct ta_ras_trigger_error_input *block_info)
@@ -921,7 +956,7 @@ unsigned long amdgpu_ras_query_error_count(struct amdgpu_device *adev,
 			.head = obj->head,
 		};
 
-		if (amdgpu_ras_error_query(adev, &info))
+		if (amdgpu_ras_query_error_status(adev, &info))
 			return 0;
 
 		data.ce_count += info.ce_count;
@@ -1451,7 +1486,7 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev)
 		if (info.head.block == AMDGPU_RAS_BLOCK__PCIE_BIF)
 			continue;
 
-		amdgpu_ras_error_query(adev, &info);
+		amdgpu_ras_query_error_status(adev, &info);
 	}
 }
 
@@ -1467,9 +1502,6 @@ static void amdgpu_ras_error_status_query(struct amdgpu_device *adev,
 	case AMDGPU_RAS_BLOCK__GFX:
 		if (adev->gfx.funcs->query_ras_error_status)
 			adev->gfx.funcs->query_ras_error_status(adev);
-
-		if (adev->gfx.funcs->query_sq_timeout_status)
-			adev->gfx.funcs->query_sq_timeout_status(adev);
 		break;
 	case AMDGPU_RAS_BLOCK__MMHUB:
 		if (adev->mmhub.funcs->query_ras_error_status)