summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
diff options
context:
space:
mode:
authorDave Airlie <airlied@redhat.com>2025-12-03 09:43:09 +1000
committerDave Airlie <airlied@redhat.com>2025-12-03 09:43:49 +1000
commit0692602defb0c273f80dec9c564ca50726404aca (patch)
tree2fea4ab096be220f377eab9934c39790cb453c83 /drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
parentb3239df349c2c2c94686674489c9629c89ca49a1 (diff)
parent3925683515e93844be204381d2d5a1df5de34f31 (diff)
Merge tag 'amd-drm-next-6.19-2025-12-02' of https://gitlab.freedesktop.org/agd5f/linux into drm-next
amd-drm-next-6.19-2025-12-02: amdgpu: - Unified MES fix - SMU 11 unbalanced irq fix - Fix for driver reloading on APUs - pp_table sysfs fix - Fix memory leak in fence handling - HDMI fix - DC cursor fixes - eDP panel parsing fix - Brightness fix - DC analog fixes - EDID retry fixes - UserQ fixes - RAS fixes - IP discovery fix - Add missing locking in amdgpu_ttm_access_memory_sdma() - Smart Power OLED fix - PRT and page fault fixes for GC 6-8 - VMID reservation fix - ACP platform device fix - Add missing vm fault handling for GC 11-12 - VPE fix amdkfd: - Partitioning fix Signed-off-by: Dave Airlie <airlied@redhat.com> From: Alex Deucher <alexander.deucher@amd.com> Link: https://patch.msgid.link/20251202220101.2039347-1-alexander.deucher@amd.com
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c28
1 files changed, 27 insertions, 1 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 9e2e098af86c..2a6cf7963dde 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -150,6 +150,8 @@ static void amdgpu_ras_critical_region_fini(struct amdgpu_device *adev);
#ifdef CONFIG_X86_MCE_AMD
static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev);
+static void
+amdgpu_unregister_bad_pages_mca_notifier(struct amdgpu_device *adev);
struct mce_notifier_adev_list {
struct amdgpu_device *devs[MAX_GPU_INSTANCE];
int num_gpu;
@@ -3954,7 +3956,9 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev)
mutex_unlock(&con->recovery_lock);
amdgpu_ras_critical_region_init(adev);
-
+#ifdef CONFIG_X86_MCE_AMD
+ amdgpu_unregister_bad_pages_mca_notifier(adev);
+#endif
return 0;
}
/* recovery end */
@@ -4988,6 +4992,28 @@ static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev)
notifier_registered = true;
}
}
+static void amdgpu_unregister_bad_pages_mca_notifier(struct amdgpu_device *adev)
+{
+ int i, j;
+
+ if (!notifier_registered && !mce_adev_list.num_gpu)
+ return;
+ for (i = 0, j = 0; i < mce_adev_list.num_gpu; i++) {
+ if (mce_adev_list.devs[i] == adev)
+ mce_adev_list.devs[i] = NULL;
+ if (!mce_adev_list.devs[i])
+ ++j;
+ }
+
+ if (j == mce_adev_list.num_gpu) {
+ mce_adev_list.num_gpu = 0;
+ /* Unregister x86 notifier with MCE subsystem. */
+ if (notifier_registered) {
+ mce_unregister_decode_chain(&amdgpu_bad_page_nb);
+ notifier_registered = false;
+ }
+ }
+}
#endif
struct amdgpu_ras *amdgpu_ras_get_context(struct amdgpu_device *adev)