drm/amdgpu: fix send ras disable cmd when asic not support ras

    cause:
	It is necessary to send ras disable command to ras-ta during gfx
	block ras later init, because the ras capability is disable read
	from vbios for vega20 gaming, but the ras context is released
	during ras init process, this will cause send ras disable command
	to ras-to failed.
    how:
	Delay releasing ras context, the ras context
	will be released after gfx block later init done.

Changed from V1:
    move release_ras_context into ras_resume

Changed from V2:
    check BIT(UMC) is more reasonable before access eeprom table
Signed-off-by: Stanley.Yang <Stanley.Yang@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_device.c

@@ -3521,11 +3521,11 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 			adev->virt.caps &= ~AMDGPU_SRIOV_CAPS_RUNTIME;
 			adev->virt.ops = NULL;
 			r = -EAGAIN;
-			goto failed;
+			goto release_ras_con;
 		}
 		dev_err(adev->dev, "amdgpu_device_ip_init failed\n");
 		amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL, 0, 0);
-		goto failed;
+		goto release_ras_con;
 	}
 
 	dev_info(adev->dev,
@@ -3591,7 +3591,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		if (r) {
 			dev_err(adev->dev, "amdgpu_device_ip_late_init failed\n");
 			amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL, 0, r);
-			goto failed;
+			goto release_ras_con;
 		}
 		/* must succeed. */
 		amdgpu_ras_resume(adev);
@@ -3625,6 +3625,9 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 
 	return 0;
 
+release_ras_con:
+	amdgpu_release_ras_context(adev);
+
 failed:
 	amdgpu_vf_error_trans_all(adev);
 	if (atpx)

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c

@@ -463,7 +463,7 @@ static struct ras_manager *amdgpu_ras_create_obj(struct amdgpu_device *adev,
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	struct ras_manager *obj;
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return NULL;
 
 	if (head->block >= AMDGPU_RAS_BLOCK_COUNT)
@@ -490,7 +490,7 @@ struct ras_manager *amdgpu_ras_find_obj(struct amdgpu_device *adev,
 	struct ras_manager *obj;
 	int i;
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return NULL;
 
 	if (head) {
@@ -590,6 +590,10 @@ static int __amdgpu_ras_feature_enable(struct amdgpu_device *adev,
 		con->features |= BIT(head->block);
 	} else {
 		if (obj && amdgpu_ras_is_feature_enabled(adev, head)) {
+			/* skip clean gfx ras context feature for VEGA20 Gaming.
+			 * will clean later
+			 */
+			if (!(!adev->ras_features && con->features & BIT(AMDGPU_RAS_BLOCK__GFX)))
 				con->features &= ~BIT(head->block);
 			put_obj(obj);
 		}
@@ -693,6 +697,10 @@ int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev,
 			if (ret)
 				return ret;
 
+			/* gfx block ras dsiable cmd must send to ras-ta */
+			if (head->block == AMDGPU_RAS_BLOCK__GFX)
+				con->features |= BIT(head->block);
+
 			ret = amdgpu_ras_feature_enable(adev, head, 0);
 		}
 	} else
@@ -948,7 +956,7 @@ unsigned long amdgpu_ras_query_error_count(struct amdgpu_device *adev,
 	struct ras_manager *obj;
 	struct ras_err_data data = {0, 0};
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return 0;
 
 	list_for_each_entry(obj, &con->head, node) {
@@ -1469,7 +1477,7 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev)
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	struct ras_manager *obj;
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return;
 
 	list_for_each_entry(obj, &con->head, node) {
@@ -1517,7 +1525,7 @@ static void amdgpu_ras_query_err_status(struct amdgpu_device *adev)
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	struct ras_manager *obj;
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return;
 
 	list_for_each_entry(obj, &con->head, node) {
@@ -1830,7 +1838,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
 	bool exc_err_limit = false;
 	int ret;
 
-	if (con)
+	if (adev->ras_features && con)
 		data = &con->eh_data;
 	else
 		return 0;
@@ -2005,6 +2013,15 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 	amdgpu_ras_check_supported(adev, &con->hw_supported,
 			&con->supported);
 	if (!con->hw_supported || (adev->asic_type == CHIP_VEGA10)) {
+		/* set gfx block ras context feature for VEGA20 Gaming
+		 * send ras disable cmd to ras ta during ras late init.
+		 */
+		if (!adev->ras_features && adev->asic_type == CHIP_VEGA20) {
+			con->features |= BIT(AMDGPU_RAS_BLOCK__GFX);
+
+			return 0;
+		}
+
 		r = 0;
 		goto release_con;
 	}
@@ -2118,8 +2135,12 @@ void amdgpu_ras_resume(struct amdgpu_device *adev)
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	struct ras_manager *obj, *tmp;
 
-	if (!con)
+	if (!adev->ras_features || !con) {
+		/* clean ras context for VEGA20 Gaming after send ras disable cmd */
+		amdgpu_release_ras_context(adev);
+
 		return;
+	}
 
 	if (con->flags & AMDGPU_RAS_FLAG_INIT_BY_VBIOS) {
 		/* Set up all other IPs which are not implemented. There is a
@@ -2160,7 +2181,7 @@ void amdgpu_ras_suspend(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return;
 
 	amdgpu_ras_disable_all_features(adev, 0);
@@ -2174,7 +2195,7 @@ int amdgpu_ras_pre_fini(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return 0;
 
 	/* Need disable ras on all IPs here before ip [hw/sw]fini */
@@ -2187,7 +2208,7 @@ int amdgpu_ras_fini(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 
-	if (!con)
+	if (!adev->ras_features || !con)
 		return 0;
 
 	amdgpu_ras_fs_fini(adev);
@@ -2230,3 +2251,17 @@ bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev)
 
 	return false;
 }
+
+void amdgpu_release_ras_context(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+	if (!con)
+		return;
+
+	if (!adev->ras_features && con->features & BIT(AMDGPU_RAS_BLOCK__GFX)) {
+		con->features &= ~BIT(AMDGPU_RAS_BLOCK__GFX);
+		amdgpu_ras_set_context(adev, NULL);
+		kfree(con);
+	}
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h

@@ -626,4 +626,6 @@ void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev);
 void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready);
 
 bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev);
+
+void amdgpu_release_ras_context(struct amdgpu_device *adev);
 #endif

drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c

@@ -441,7 +441,14 @@ bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev)
 	if (!__is_ras_eeprom_supported(adev))
 		return false;
 
-	if (con && (con->eeprom_control.tbl_hdr.header == EEPROM_TABLE_HDR_BAD)) {
+	/* skip check eeprom table for VEGA20 Gaming */
+	if (!con)
+		return false;
+	else
+		if (!(con->features & BIT(AMDGPU_RAS_BLOCK__UMC)))
+			return false;
+
+	if (con->eeprom_control.tbl_hdr.header == EEPROM_TABLE_HDR_BAD) {
 		dev_warn(adev->dev, "This GPU is in BAD status.");
 		dev_warn(adev->dev, "Please retire it or setting one bigger "
 				"threshold value when reloading driver.\n");