summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c68
1 files changed, 48 insertions, 20 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index dc07936d2fcb..c508697a3412 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -1107,6 +1107,9 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev,
err_info->de_count, blk_name);
}
} else {
+ if (adev->debug_disable_ce_logs)
+ return;
+
for_each_ras_error(err_node, err_data) {
err_info = &err_node->err_info;
mcm_info = &err_info->mcm_info;
@@ -2859,6 +2862,15 @@ static int __amdgpu_ras_convert_rec_array_from_rom(struct amdgpu_device *adev,
return -EINVAL;
}
} else {
+ if (bps[0].address == 0) {
+ /* for specific old eeprom data, mca address is not stored,
+ * calc it from pa
+ */
+ if (amdgpu_umc_pa2mca(adev, bps[0].retired_page << AMDGPU_GPU_PAGE_SHIFT,
+ &(bps[0].address), AMDGPU_NPS1_PARTITION_MODE))
+ return -EINVAL;
+ }
+
if (amdgpu_ras_mca2pa(adev, &bps[0], err_data)) {
if (nps == AMDGPU_NPS1_PARTITION_MODE)
memcpy(err_data->err_addr, bps,
@@ -2886,8 +2898,20 @@ static int __amdgpu_ras_convert_rec_from_rom(struct amdgpu_device *adev,
bps->retired_page << AMDGPU_GPU_PAGE_SHIFT))
return -EINVAL;
} else {
- if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data))
- return -EINVAL;
+ if (bps->address) {
+ if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data))
+ return -EINVAL;
+ } else {
+ /* for specific old eeprom data, mca address is not stored,
+ * calc it from pa
+ */
+ if (amdgpu_umc_pa2mca(adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT,
+ &(bps->address), AMDGPU_NPS1_PARTITION_MODE))
+ return -EINVAL;
+
+ if (amdgpu_ras_mca2pa(adev, bps, err_data))
+ return -EOPNOTSUPP;
+ }
}
return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr,
@@ -2982,6 +3006,15 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev,
return 0;
}
+ if (!con->eeprom_control.is_eeprom_valid) {
+ dev_warn(adev->dev,
+ "Failed to save EEPROM table data because of EEPROM data corruption!");
+ if (new_cnt)
+ *new_cnt = 0;
+
+ return 0;
+ }
+
mutex_lock(&con->recovery_lock);
control = &con->eeprom_control;
data = con->eh_data;
@@ -3273,7 +3306,6 @@ static int amdgpu_ras_poison_creation_handler(struct amdgpu_device *adev,
uint64_t de_queried_count;
uint32_t new_detect_count, total_detect_count;
uint32_t need_query_count = poison_creation_count;
- bool query_data_timeout = false;
enum ras_event_type type = RAS_EVENT_TYPE_POISON_CREATION;
memset(&info, 0, sizeof(info));
@@ -3302,21 +3334,13 @@ static int amdgpu_ras_poison_creation_handler(struct amdgpu_device *adev,
timeout = MAX_UMC_POISON_POLLING_TIME_ASYNC;
if (timeout) {
- if (!--timeout) {
- query_data_timeout = true;
+ if (!--timeout)
break;
- }
msleep(1);
}
}
} while (total_detect_count < need_query_count);
- if (query_data_timeout) {
- dev_warn(adev->dev, "Can't find deferred error! count: %u\n",
- (need_query_count - total_detect_count));
- return -ENOENT;
- }
-
if (total_detect_count)
schedule_delayed_work(&ras->page_retirement_dwork, 0);
@@ -3467,8 +3491,7 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev)
control = &con->eeprom_control;
ret = amdgpu_ras_eeprom_init(control);
- if (ret)
- return ret;
+ control->is_eeprom_valid = !ret;
if (!adev->umc.ras || !adev->umc.ras->convert_ras_err_addr)
control->ras_num_pa_recs = control->ras_num_recs;
@@ -3477,10 +3500,12 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev)
adev->umc.ras->get_retire_flip_bits)
adev->umc.ras->get_retire_flip_bits(adev);
- if (control->ras_num_recs) {
+ if (control->ras_num_recs && control->is_eeprom_valid) {
ret = amdgpu_ras_load_bad_pages(adev);
- if (ret)
- return ret;
+ if (ret) {
+ control->is_eeprom_valid = false;
+ return 0;
+ }
amdgpu_dpm_send_hbm_bad_pages_num(
adev, control->ras_num_bad_pages);
@@ -3499,7 +3524,7 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev)
dev_warn(adev->dev, "Failed to format RAS EEPROM data in V3 version!\n");
}
- return ret;
+ return 0;
}
int amdgpu_ras_recovery_init(struct amdgpu_device *adev, bool init_bp_info)
@@ -3708,7 +3733,8 @@ static void amdgpu_ras_query_ras_capablity_from_vbios(struct amdgpu_device *adev
*/
if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) ||
amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 0) ||
- amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3))
+ amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3) ||
+ amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(5, 0, 1))
adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN |
1 << AMDGPU_RAS_BLOCK__JPEG);
else
@@ -4392,8 +4418,10 @@ void amdgpu_ras_clear_err_state(struct amdgpu_device *adev)
struct amdgpu_ras *ras;
ras = amdgpu_ras_get_context(adev);
- if (ras)
+ if (ras) {
ras->ras_err_state = 0;
+ ras->gpu_reset_flags = 0;
+ }
}
void amdgpu_ras_set_err_poison(struct amdgpu_device *adev,