diff options
| author | John Clements <john.clements@amd.com> | 2020-03-19 14:41:55 +0800 |
|---|---|---|
| committer | Alex Deucher <alexander.deucher@amd.com> | 2020-03-20 10:45:00 -0400 |
| commit | 43c4d57618bef018eecd769c2805ce6f4e849a0d (patch) | |
| tree | 242649a7e7e898ccfc9d70f0d64ce3ee9babfc85 /drivers/gpu/drm/amd | |
| parent | cb7adfd6ad12a11902ebe374bec7fd4efa2cec1c (diff) | |
| download | linux-next-43c4d57618bef018eecd769c2805ce6f4e849a0d.tar.gz | |
drm/amdgpu: protect RAS sysfs during GPU reset
MMHub EDC becomes dirty after BACO reset
EDC registers should be cleared early on in reset phase
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: John Clements <john.clements@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Diffstat (limited to 'drivers/gpu/drm/amd')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 12 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 9 |
2 files changed, 20 insertions, 1 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 6f469facabfb..faa3e7102156 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -2742,6 +2742,9 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work) if (adev->asic_reset_res) goto fail; + + if (adev->mmhub.funcs && adev->mmhub.funcs->reset_ras_error_count) + adev->mmhub.funcs->reset_ras_error_count(adev); } else { task_barrier_full(&hive->tb); @@ -3910,8 +3913,15 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive, } } - if (!r && amdgpu_ras_intr_triggered()) + if (!r && amdgpu_ras_intr_triggered()) { + list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { + if (tmp_adev->mmhub.funcs && + tmp_adev->mmhub.funcs->reset_ras_error_count) + tmp_adev->mmhub.funcs->reset_ras_error_count(tmp_adev); + } + amdgpu_ras_intr_cleared(); + } list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) { if (need_full_reset) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 43055a01f35e..3c32a94d2424 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -281,6 +281,11 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user * struct ras_debug_if data; int ret = 0; + if (amdgpu_ras_intr_triggered()) { + DRM_WARN("RAS WARN: error injection currently inaccessible\n"); + return size; + } + ret = amdgpu_ras_debugfs_ctrl_parse_data(f, buf, size, pos, &data); if (ret) return -EINVAL; @@ -394,6 +399,10 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, .head = obj->head, }; + if (amdgpu_ras_intr_triggered()) + return snprintf(buf, PAGE_SIZE, + "Query currently inaccessible\n"); + if (amdgpu_ras_error_query(obj->adev, &info)) return -EINVAL; |