drm/amdgpu: Add fatal error detected flag

For a RAS error that needs a full reset to recover, set the fatal error
status. Clear the status once the device is reset.

Signed-off-by: Lijo Lazar <[email protected]>
Reviewed-by: Asad Kamal <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

1 files changed, 32 insertions, 0 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 46f3d1013e8c..2c94de305c69 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -2439,6 +2439,18 @@ static void amdgpu_ras_do_recovery(struct work_struct *work)
 				ras->gpu_reset_flags &= ~AMDGPU_RAS_GPU_RESET_MODE1_RESET;
 				set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
 
+				/* For any RAS error that needs a full reset to
+				 * recover, set the fatal error status
+				 */
+				if (hive) {
+					list_for_each_entry(remote_adev,
+							    &hive->device_list,
+							    gmc.xgmi.head)
+						amdgpu_ras_set_fed(remote_adev,
+								   true);
+				} else {
+					amdgpu_ras_set_fed(adev, true);
+				}
 				psp_fatal_error_recovery_quirk(&adev->psp);
 			}
 		}
@@ -3440,6 +3452,26 @@ int amdgpu_ras_fini(struct amdgpu_device *adev)
 	return 0;
 }
 
+bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (!ras)
+		return false;
+
+	return atomic_read(&ras->fed);
+}
+
+void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status)
+{
+	struct amdgpu_ras *ras;
+
+	ras = amdgpu_ras_get_context(adev);
+	if (ras)
+		atomic_set(&ras->fed, !!status);
+}
+
 void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev)
 {
 	if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) {