Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
drm/amdgpu: support to find RAS bad pages via old TA
Old version of RAS TA doesn't support to convert MCA address stored on
eeprom to physical address (PA), support to find all bad pages in one
memory row by PA with old RAS TA. This approach is only suitable for
nps1 mode.
Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
authored by
Tao Zhou
and committed by
Alex Deucher
07dd49e1
b02ef407
+25
-3
+25
-3
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+25
-3
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
···
2765
2765
struct ras_err_handler_data *data;
2766
2766
struct ras_err_data err_data;
2767
2767
struct eeprom_table_record *err_rec;
2768
+
enum amdgpu_memory_partition nps = AMDGPU_NPS1_PARTITION_MODE;
2768
2769
int ret = 0;
2769
2770
uint32_t i, j, loop_cnt = 1;
2770
-
bool is_mca_add = true;
2771
+
bool is_mca_add = true, find_pages_per_pa = false;
2771
2772
2772
2773
if (!con || !con->eh_data || !bps || pages <= 0)
2773
2774
return 0;
···
2798
2797
}
2799
2798
2800
2799
loop_cnt = adev->umc.retire_unit;
2800
+
if (adev->gmc.gmc_funcs->query_mem_partition_mode)
2801
+
nps = adev->gmc.gmc_funcs->query_mem_partition_mode(adev);
2801
2802
}
2802
2803
2803
2804
for (i = 0; i < pages; i++) {
2804
2805
if (is_mca_add) {
2805
-
if (amdgpu_ras_mca2pa(adev, &bps[i], &err_data))
2806
-
goto free;
2806
+
if (!find_pages_per_pa) {
2807
+
if (amdgpu_ras_mca2pa(adev, &bps[i], &err_data)) {
2808
+
if (!i && nps == AMDGPU_NPS1_PARTITION_MODE) {
2809
+
/* may use old RAS TA, use PA to find pages in
2810
+
* one row
2811
+
*/
2812
+
if (amdgpu_umc_pages_in_a_row(adev, &err_data,
2813
+
bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT))
2814
+
goto free;
2815
+
else
2816
+
find_pages_per_pa = true;
2817
+
} else {
2818
+
/* unsupported cases */
2819
+
goto free;
2820
+
}
2821
+
}
2822
+
} else {
2823
+
if (amdgpu_umc_pages_in_a_row(adev, &err_data,
2824
+
bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT))
2825
+
goto free;
2826
+
}
2807
2827
2808
2828
err_rec = err_data.err_addr;
2809
2829
} else {