Check gpu status first, if MC/VMC/DISPLAY hang, directly triger full reset.
If engine hangs, then triger engine soft reset, if soft reset fails, will
fallback to full reset.
Change-Id: I2d8a2c1ae72cbb1aefa4db2b216903664ed27876
Signed-off-by: Chunming Zhou <David1.Zhou at amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 31 ++++++++++++++++++++++++++++++
1 file changed, 31 insertions(+)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 9a7e574..f825311 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -1963,6 +1963,17 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
struct drm_atomic_state *state = NULL;
int i, r;
int resched;
+ u32 reset_mask;
+ bool need_full_reset = false;
+
+ reset_mask = amdgpu_asic_check_soft_reset(adev);
+ /* ? VCE always busy ?*/
+ reset_mask &= ~AMDGPU_RESET_VCE;
+ if (!reset_mask) {
+ DRM_INFO("No hardware hang detected. Did some blocks stall?\n");
+ return 0;
+ }
+ DRM_INFO("reset mask: 0x%x\n", reset_mask);
atomic_inc(&adev->gpu_reset_counter);
@@ -1980,9 +1991,28 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
}
/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
amdgpu_fence_driver_force_completion(adev);
+
+ if (reset_mask &
+ (AMDGPU_RESET_MC | AMDGPU_RESET_VMC | AMDGPU_RESET_DISPLAY))
+ need_full_reset = true;
+ if (!need_full_reset) {
+ amdgpu_gpu_soft_reset(adev, reset_mask);
+ reset_mask = amdgpu_asic_check_soft_reset(adev);
+ /* ? VCE always busy ?*/
+ if (reset_mask & ~AMDGPU_RESET_VCE) {
+ need_full_reset = true;
+ DRM_INFO("Couldn't recover, still busy: 0x%x\n",
+ reset_mask);
+ }
+ }
+
/* store modesetting */
if (amdgpu_device_has_dal_support(adev))
state = drm_atomic_helper_suspend(adev->ddev);
+ if(!need_full_reset) {
+ r = 0;
+ goto out;
+ }
/* save scratch */
amdgpu_atombios_scratch_regs_save(adev);
@@ -2005,6 +2035,7 @@ retry:
}
/* restore scratch */
amdgpu_atombios_scratch_regs_restore(adev);
+out:
if (!r) {
r = amdgpu_ib_ring_tests(adev);
if (r) {
--
1.9.1
