drm/amdgpu: move aca/mca init functions into ras_init() stage

adjust the function position to better match aca/mca fini code in ras_fini().
Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -712,6 +712,15 @@ void amdgpu_aca_fini(struct amdgpu_device *adev)
 	atomic_set(&aca->ue_update_flag, 0);
 }
 
+int amdgpu_aca_reset(struct amdgpu_device *adev)
+{
+	struct amdgpu_aca *aca = &adev->aca;
+
+	atomic_set(&aca->ue_update_flag, 0);
+
+	return 0;
+}
+
 void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs)
 {
 	struct amdgpu_aca *aca = &adev->aca;
@@ -885,9 +894,7 @@ DEFINE_DEBUGFS_ATTRIBUTE(aca_debug_mode_fops, NULL, amdgpu_aca_smu_debug_mode_se
 void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root)
 {
 #if defined(CONFIG_DEBUG_FS)
-	if (!root ||
-	    (adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 6) &&
-	     adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 14)))
+	if (!root)
 		return;
 
 	debugfs_create_file("aca_debug_mode", 0200, root, adev, &aca_debug_mode_fops);

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

@@ -192,6 +192,7 @@ struct aca_info {
 
 int amdgpu_aca_init(struct amdgpu_device *adev);
 void amdgpu_aca_fini(struct amdgpu_device *adev);
+int amdgpu_aca_reset(struct amdgpu_device *adev);
 void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs);
 bool amdgpu_aca_is_enabled(struct amdgpu_device *adev);
 

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

@@ -193,27 +193,26 @@ static int amdgpu_mca_bank_set_merge(struct mca_bank_set *mca_set, struct mca_ba
 	return 0;
 }
 
-static int amdgpu_mca_bank_set_remove_node(struct mca_bank_set *mca_set, struct mca_bank_node *node)
+static void amdgpu_mca_bank_set_remove_node(struct mca_bank_set *mca_set, struct mca_bank_node *node)
 {
 	if (!node)
-		return -EINVAL;
+		return;
 
 	list_del(&node->node);
 	kvfree(node);
 
 	mca_set->nr_entries--;
-
-	return 0;
 }
 
 static void amdgpu_mca_bank_set_release(struct mca_bank_set *mca_set)
 {
 	struct mca_bank_node *node, *tmp;
 
-	list_for_each_entry_safe(node, tmp, &mca_set->list, node) {
-		list_del(&node->node);
-		kvfree(node);
-	}
+	if (list_empty(&mca_set->list))
+		return;
+
+	list_for_each_entry_safe(node, tmp, &mca_set->list, node)
+		amdgpu_mca_bank_set_remove_node(mca_set, node);
 }
 
 void amdgpu_mca_smu_init_funcs(struct amdgpu_device *adev, const struct amdgpu_mca_smu_funcs *mca_funcs)
@@ -608,9 +607,7 @@ DEFINE_DEBUGFS_ATTRIBUTE(mca_debug_mode_fops, NULL, amdgpu_mca_smu_debug_mode_se
 void amdgpu_mca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root)
 {
 #if defined(CONFIG_DEBUG_FS)
-	if (!root ||
-	    (amdgpu_ip_version(adev, MP1_HWIP, 0) != IP_VERSION(13, 0, 6) &&
-	     amdgpu_ip_version(adev, MP1_HWIP, 0) != IP_VERSION(13, 0, 14)))
+	if (!root)
 		return;
 
 	debugfs_create_file("mca_debug_mode", 0200, root, adev, &mca_debug_mode_fops);

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

@@ -1911,6 +1911,23 @@ static void amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
 			    obj, &amdgpu_ras_debugfs_ops);
 }
 
+static bool amdgpu_ras_aca_is_supported(struct amdgpu_device *adev)
+{
+	bool ret;
+
+	switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+	case IP_VERSION(13, 0, 6):
+	case IP_VERSION(13, 0, 14):
+		ret = true;
+		break;
+	default:
+		ret = false;
+		break;
+	}
+
+	return ret;
+}
+
 void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
@@ -1937,10 +1954,12 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev)
 		}
 	}
 
+	if (amdgpu_ras_aca_is_supported(adev)) {
 		if (amdgpu_aca_is_enabled(adev))
 			amdgpu_aca_smu_debugfs_init(adev, dir);
 		else
 			amdgpu_mca_smu_debugfs_init(adev, dir);
+	}
 }
 
 /* debugfs end */
@@ -3428,6 +3447,15 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 		goto release_con;
 	}
 
+	if (amdgpu_ras_aca_is_supported(adev)) {
+		if (amdgpu_aca_is_enabled(adev))
+			r = amdgpu_aca_init(adev);
+		else
+			r = amdgpu_mca_init(adev);
+		if (r)
+			goto release_con;
+	}
+
 	dev_info(adev->dev, "RAS INFO: ras initialized successfully, "
 		 "hardware ability[%x] ras_mask[%x]\n",
 		 adev->ras_hw_enabled, adev->ras_enabled);
@@ -3636,26 +3664,23 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev)
 
 	amdgpu_ras_event_mgr_init(adev);
 
-	if (amdgpu_aca_is_enabled(adev)) {
-		if (!amdgpu_in_reset(adev)) {
-			r = amdgpu_aca_init(adev);
+	if (amdgpu_ras_aca_is_supported(adev)) {
+		if (amdgpu_in_reset(adev)) {
+			if (amdgpu_aca_is_enabled(adev))
+				r = amdgpu_aca_reset(adev);
+			else
+				r = amdgpu_mca_reset(adev);
 			if (r)
 				return r;
 		}
 
-		if (!amdgpu_sriov_vf(adev))
+		if (!amdgpu_sriov_vf(adev)) {
+			if (amdgpu_aca_is_enabled(adev))
 				amdgpu_ras_set_aca_debug_mode(adev, false);
-	} else {
-		if (amdgpu_in_reset(adev))
-			r = amdgpu_mca_reset(adev);
 			else
-			r = amdgpu_mca_init(adev);
-		if (r)
-			return r;
-
-		if (!amdgpu_sriov_vf(adev))
 				amdgpu_ras_set_mca_debug_mode(adev, false);
 		}
+	}
 
 	/* Guest side doesn't need init ras feature */
 	if (amdgpu_sriov_vf(adev))
@@ -3728,10 +3753,12 @@ int amdgpu_ras_fini(struct amdgpu_device *adev)
 	amdgpu_ras_fs_fini(adev);
 	amdgpu_ras_interrupt_remove_all(adev);
 
+	if (amdgpu_ras_aca_is_supported(adev)) {
 		if (amdgpu_aca_is_enabled(adev))
 			amdgpu_aca_fini(adev);
 		else
 			amdgpu_mca_fini(adev);
+	}
 
 	WARN(AMDGPU_RAS_GET_FEATURES(con->features), "Feature mask is not cleared");