summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorYang Wang <kevinyang.wang@amd.com>2023-09-27 11:31:16 +0800
committerAlex Deucher <alexander.deucher@amd.com>2023-10-13 11:36:11 -0400
commit3bba4bc6a0ebf424648476290750a49dc8032730 (patch)
tree8281a245615cbf3c990ff4e15cc47059a911dd6b
parent8736d17a7f98be9de7b08eb94f08578576b6b043 (diff)
drm/amdgpu: add RAS error info support for umc_v12_0
add RAS error info support for umc_v12_0. Signed-off-by: Yang Wang <kevinyang.wang@amd.com> Reviewed-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
-rw-r--r--drivers/gpu/drm/amd/amdgpu/umc_v12_0.c20
1 files changed, 14 insertions, 6 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
index 7714c2ef2cdc..b664ee3ee92d 100644
--- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
@@ -137,15 +137,23 @@ static int umc_v12_0_query_error_count(struct amdgpu_device *adev,
uint32_t ch_inst, void *data)
{
struct ras_err_data *err_data = (struct ras_err_data *)data;
+ unsigned long ue_count = 0, ce_count = 0;
+
+ /* NOTE: node_inst is converted by adev->umc.active_mask and the range is [0-3],
+ * which can be used as die ID directly */
+ struct amdgpu_smuio_mcm_config_info mcm_info = {
+ .socket_id = adev->smuio.funcs->get_socket_id(adev),
+ .die_id = node_inst,
+ };
+
uint64_t umc_reg_offset =
get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst);
- umc_v12_0_query_correctable_error_count(adev,
- umc_reg_offset,
- &(err_data->ce_count));
- umc_v12_0_query_uncorrectable_error_count(adev,
- umc_reg_offset,
- &(err_data->ue_count));
+ umc_v12_0_query_correctable_error_count(adev, umc_reg_offset, &ce_count);
+ umc_v12_0_query_uncorrectable_error_count(adev, umc_reg_offset, &ue_count);
+
+ amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count);
+ amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, ce_count);
return 0;
}