drm/amdgpu: Add reset_context flag for host FLR

[ Upstream commit 25c0119 ] There are other reset sources that pass NULL as the job pointer, such as amdgpu_amdkfd_reset_work. Therefore, using the job pointer to check if the FLR comes from the host does not work. Add a flag in reset_context to explicitly mark host triggered reset, and set this flag when we receive host reset notification. Signed-off-by: Yunxiang Li <[email protected]> Reviewed-by: Emily Deng <[email protected]> Reviewed-by: Zhigang Luo <[email protected]> Signed-off-by: Alex Deucher <[email protected]> Stable-dep-of: 6e4aa08 ("drm/amdgpu: Fix amdgpu_device_reset_sriov retry logic") Signed-off-by: Sasha Levin <[email protected]>
xanmod · Sep 12, 2024 · 3adb4ae · 3adb4ae
1 parent 1f49070
commit 3adb4ae
Show file tree

Hide file tree

Showing 5 changed files with 12 additions and 5 deletions.
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -5057,13 +5057,13 @@ static int amdgpu_device_recover_vram(struct amdgpu_device *adev)
  * amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf
  *
  * @adev: amdgpu_device pointer
- * @from_hypervisor: request from hypervisor
+ * @reset_context: amdgpu reset context pointer
  *
  * do VF FLR and reinitialize Asic
  * return 0 means succeeded otherwise failed
  */
 static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
-				     bool from_hypervisor)
+				     struct amdgpu_reset_context *reset_context)
 {
 	int r;
 	struct amdgpu_hive_info *hive = NULL;
@@ -5072,12 +5072,15 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
 retry:
 	amdgpu_amdkfd_pre_reset(adev);
 
-	if (from_hypervisor)
+	if (test_bit(AMDGPU_HOST_FLR, &reset_context->flags)) {
+		clear_bit(AMDGPU_HOST_FLR, &reset_context->flags);
 		r = amdgpu_virt_request_full_gpu(adev, true);
-	else
+	} else {
 		r = amdgpu_virt_reset_gpu(adev);
+	}
 	if (r)
 		return r;
+
 	amdgpu_ras_set_fed(adev, false);
 	amdgpu_irq_gpu_reset_resume_helper(adev);
 
@@ -5831,7 +5834,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	/* Actual ASIC resets if needed.*/
 	/* Host driver will handle XGMI hive reset for SRIOV */
 	if (amdgpu_sriov_vf(adev)) {
-		r = amdgpu_device_reset_sriov(adev, job ? false : true);
+		r = amdgpu_device_reset_sriov(adev, reset_context);
 		if (r)
 			adev->asic_reset_res = r;
 

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
@@ -33,6 +33,7 @@ enum AMDGPU_RESET_FLAGS {
 	AMDGPU_NEED_FULL_RESET = 0,
 	AMDGPU_SKIP_HW_RESET = 1,
 	AMDGPU_SKIP_COREDUMP = 2,
+	AMDGPU_HOST_FLR = 3,
 };
 
 struct amdgpu_reset_context {

diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
@@ -292,6 +292,7 @@ static void xgpu_ai_mailbox_flr_work(struct work_struct *work)
 		reset_context.method = AMD_RESET_METHOD_NONE;
 		reset_context.reset_req_dev = adev;
 		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+		set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
 
 		amdgpu_device_gpu_recover(adev, NULL, &reset_context);
 	}

diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
@@ -328,6 +328,7 @@ static void xgpu_nv_mailbox_flr_work(struct work_struct *work)
 		reset_context.method = AMD_RESET_METHOD_NONE;
 		reset_context.reset_req_dev = adev;
 		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+		set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
 
 		amdgpu_device_gpu_recover(adev, NULL, &reset_context);
 	}

diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
@@ -529,6 +529,7 @@ static void xgpu_vi_mailbox_flr_work(struct work_struct *work)
 		reset_context.method = AMD_RESET_METHOD_NONE;
 		reset_context.reset_req_dev = adev;
 		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+		set_bit(AMDGPU_HOST_FLR, &reset_context.flags);
 
 		amdgpu_device_gpu_recover(adev, NULL, &reset_context);
 	}