[PATCH 2/2] drm/amdgpu: Do bad page retirement for deferred errors

[Zhang, Hawking]

[AMD Official Use Only - General]

Let's drop the following message.

+               dev_info(adev->dev, "%ld uncorrectable hardware errors and "
+                               "%ld deferred hardware errors detected in UMC block\n",
+                               err_data->ue_count, err_data->de_count);

With that fixed, the series is

Reviewed-by: Hawking Zhang <Hawking.Zhang at amd.com>

Regards,
Hawking
-----Original Message-----
From: amd-gfx <amd-gfx-bounces at lists.freedesktop.org> On Behalf Of Candice Li
Sent: Wednesday, January 10, 2024 16:39
To: amd-gfx at lists.freedesktop.org
Cc: Li, Candice <Candice.Li at amd.com>
Subject: [PATCH 2/2] drm/amdgpu: Do bad page retirement for deferred errors

Needs to do bad page retirement for deferred errors.

Signed-off-by: Candice Li <candice.li at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
index 848df7acdd3210..df61df7e9b155f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@@ -93,6 +93,7 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
        struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
        struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
        int ret = 0;
+       unsigned long err_count;

        kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
        ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc)); @@ -147,16 +148,17 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
        }

        /* only uncorrectable error needs gpu reset */
-       if (err_data->ue_count) {
-               dev_info(adev->dev, "%ld uncorrectable hardware errors "
-                               "detected in UMC block\n",
-                               err_data->ue_count);
+       if (err_data->ue_count || err_data->de_count) {
+               dev_info(adev->dev, "%ld uncorrectable hardware errors and "
+                               "%ld deferred hardware errors detected in UMC block\n",
+                               err_data->ue_count, err_data->de_count);

+               err_count = err_data->ue_count + err_data->de_count;
                if ((amdgpu_bad_page_threshold != 0) &&
                        err_data->err_addr_cnt) {
                        amdgpu_ras_add_bad_pages(adev, err_data->err_addr,
                                                err_data->err_addr_cnt);
-                       amdgpu_ras_save_bad_pages(adev, &(err_data->ue_count));
+                       amdgpu_ras_save_bad_pages(adev, &err_count);

                        amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs);

--
2.25.1