From bf83060408fea52eccdcf695f3b4b16c71207691 Mon Sep 17 00:00:00 2001
From: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Date: Thu, 17 May 2018 11:18:34 -0400
Subject: [PATCH] Remove calls to suspend/resume atomic helpers from
 amdgpu_device_gpu_recover. (v2)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

First of all it's already being called from the display code from amd_ip_funcs.suspend/resume hooks.
Second of all, the place in amdgpu_device_gpu_recover it's being called is wrong for GPU stalls since
it is called BEFORE we cancel and force completion of all in flight jobs which were not yet processed.
So, as Bas pointed in the ticket we will try to wait for fence  in amdgpu_pm_compute_clocks but the pipe
is hanged so we end up in deadlock.

v2: remove unused variable

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106500
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 0e3f69d31b80..adeb48ec4897 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -3176,7 +3176,6 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
 int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 			      struct amdgpu_job *job, bool force)
 {
-	struct drm_atomic_state *state = NULL;
 	int i, r, resched;
 
 	if (!force && !amdgpu_device_ip_check_soft_reset(adev)) {
@@ -3199,10 +3198,6 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	/* block TTM */
 	resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
 
-	/* store modesetting */
-	if (amdgpu_device_has_dc_support(adev))
-		state = drm_atomic_helper_suspend(adev->ddev);
-
 	/* block all schedulers and reset given job's ring */
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
@@ -3242,10 +3237,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 		kthread_unpark(ring->sched.thread);
 	}
 
-	if (amdgpu_device_has_dc_support(adev)) {
-		if (drm_atomic_helper_resume(adev->ddev, state))
-			dev_info(adev->dev, "drm resume failed:%d\n", r);
-	} else {
+	if (!amdgpu_device_has_dc_support(adev)) {
 		drm_helper_resume_force_mode(adev->ddev);
 	}
 
-- 
GitLab