From e2f6e0931ff79fb9de94fc4c759b1e997be0af54 Mon Sep 17 00:00:00 2001
From: Tao Zhou <tao.zhou1@amd.com>
Date: Mon, 20 Dec 2021 16:36:54 +0800
Subject: drm/amdgpu: save error count in RAS poison handler
Git-commit: fec8c5244fc07b1f6a3249a8714489f594ff5c4f
Patch-mainline: v5.17-rc1
References: jsc#PED-1166 jsc#PED-1168 jsc#PED-1170 jsc#PED-1218 jsc#PED-1220 jsc#PED-1222 jsc#PED-1223 jsc#PED-1225
Otherwise the RAS error count couldn't be queried from sysfs.
Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Stanley.Yang <Stanley.Yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Patrik Jakobsson <pjakobsson@suse.de>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 2 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 168 ++++++++++++---------
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h | 3 +-
3 files changed, 97 insertions(+), 76 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 0bf09a94d944..776a947b45df 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -727,7 +727,7 @@ void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, bo
/* CPU MCA will handle page retirement if connected_to_cpu is 1 */
if (!adev->gmc.xgmi.connected_to_cpu)
- amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset);
+ amdgpu_umc_poison_handler(adev, &err_data, reset);
else if (reset)
amdgpu_amdkfd_gpu_reset(adev);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
index 0c33f367a4e5..46264a4002f7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@@ -23,79 +23,7 @@
#include "amdgpu_ras.h"
-static int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
- void *ras_error_status,
- struct amdgpu_iv_entry *entry)
-{
- return amdgpu_umc_do_page_retirement(adev, ras_error_status, entry, true);
-}
-
-int amdgpu_umc_ras_late_init(struct amdgpu_device *adev)
-{
- int r;
- struct ras_fs_if fs_info = {
- .sysfs_name = "umc_err_count",
- };
- struct ras_ih_if ih_info = {
- .cb = amdgpu_umc_process_ras_data_cb,
- };
-
- if (!adev->umc.ras_if) {
- adev->umc.ras_if =
- kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
- if (!adev->umc.ras_if)
- return -ENOMEM;
- adev->umc.ras_if->block = AMDGPU_RAS_BLOCK__UMC;
- adev->umc.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
- adev->umc.ras_if->sub_block_index = 0;
- }
- ih_info.head = fs_info.head = *adev->umc.ras_if;
-
- r = amdgpu_ras_late_init(adev, adev->umc.ras_if,
- &fs_info, &ih_info);
- if (r)
- goto free;
-
- if (amdgpu_ras_is_supported(adev, adev->umc.ras_if->block)) {
- r = amdgpu_irq_get(adev, &adev->gmc.ecc_irq, 0);
- if (r)
- goto late_fini;
- } else {
- r = 0;
- goto free;
- }
-
- /* ras init of specific umc version */
- if (adev->umc.ras_funcs &&
- adev->umc.ras_funcs->err_cnt_init)
- adev->umc.ras_funcs->err_cnt_init(adev);
-
- return 0;
-
-late_fini:
- amdgpu_ras_late_fini(adev, adev->umc.ras_if, &ih_info);
-free:
- kfree(adev->umc.ras_if);
- adev->umc.ras_if = NULL;
- return r;
-}
-
-void amdgpu_umc_ras_fini(struct amdgpu_device *adev)
-{
- if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC) &&
- adev->umc.ras_if) {
- struct ras_common_if *ras_if = adev->umc.ras_if;
- struct ras_ih_if ih_info = {
- .head = *ras_if,
- .cb = amdgpu_umc_process_ras_data_cb,
- };
-
- amdgpu_ras_late_fini(adev, ras_if, &ih_info);
- kfree(ras_if);
- }
-}
-
-int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
+static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
void *ras_error_status,
struct amdgpu_iv_entry *entry,
bool reset)
@@ -180,6 +108,100 @@ int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
return AMDGPU_RAS_SUCCESS;
}
+int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
+ void *ras_error_status,
+ bool reset)
+{
+ int ret;
+ struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+ struct ras_common_if head = {
+ .block = AMDGPU_RAS_BLOCK__UMC,
+ };
+ struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
+
+ ret =
+ amdgpu_umc_do_page_retirement(adev, ras_error_status, NULL, reset);
+
+ if (ret == AMDGPU_RAS_SUCCESS && obj) {
+ obj->err_data.ue_count += err_data->ue_count;
+ obj->err_data.ce_count += err_data->ce_count;
+ }
+
+ return ret;
+}
+
+static int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
+ void *ras_error_status,
+ struct amdgpu_iv_entry *entry)
+{
+ return amdgpu_umc_do_page_retirement(adev, ras_error_status, entry, true);
+}
+
+int amdgpu_umc_ras_late_init(struct amdgpu_device *adev)
+{
+ int r;
+ struct ras_fs_if fs_info = {
+ .sysfs_name = "umc_err_count",
+ };
+ struct ras_ih_if ih_info = {
+ .cb = amdgpu_umc_process_ras_data_cb,
+ };
+
+ if (!adev->umc.ras_if) {
+ adev->umc.ras_if =
+ kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
+ if (!adev->umc.ras_if)
+ return -ENOMEM;
+ adev->umc.ras_if->block = AMDGPU_RAS_BLOCK__UMC;
+ adev->umc.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+ adev->umc.ras_if->sub_block_index = 0;
+ }
+ ih_info.head = fs_info.head = *adev->umc.ras_if;
+
+ r = amdgpu_ras_late_init(adev, adev->umc.ras_if,
+ &fs_info, &ih_info);
+ if (r)
+ goto free;
+
+ if (amdgpu_ras_is_supported(adev, adev->umc.ras_if->block)) {
+ r = amdgpu_irq_get(adev, &adev->gmc.ecc_irq, 0);
+ if (r)
+ goto late_fini;
+ } else {
+ r = 0;
+ goto free;
+ }
+
+ /* ras init of specific umc version */
+ if (adev->umc.ras_funcs &&
+ adev->umc.ras_funcs->err_cnt_init)
+ adev->umc.ras_funcs->err_cnt_init(adev);
+
+ return 0;
+
+late_fini:
+ amdgpu_ras_late_fini(adev, adev->umc.ras_if, &ih_info);
+free:
+ kfree(adev->umc.ras_if);
+ adev->umc.ras_if = NULL;
+ return r;
+}
+
+void amdgpu_umc_ras_fini(struct amdgpu_device *adev)
+{
+ if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC) &&
+ adev->umc.ras_if) {
+ struct ras_common_if *ras_if = adev->umc.ras_if;
+ struct ras_ih_if ih_info = {
+ .head = *ras_if,
+ .cb = amdgpu_umc_process_ras_data_cb,
+ };
+
+ amdgpu_ras_late_fini(adev, ras_if, &ih_info);
+ kfree(ras_if);
+ }
+}
+
int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
index 8d18d5121f66..b72194e8bfe5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
@@ -78,9 +78,8 @@ struct amdgpu_umc {
int amdgpu_umc_ras_late_init(struct amdgpu_device *adev);
void amdgpu_umc_ras_fini(struct amdgpu_device *adev);
-int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
+int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
void *ras_error_status,
- struct amdgpu_iv_entry *entry,
bool reset);
int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
--
2.38.1