[PATCH 1/3] drm/amdgpu: delete duplicated printk info.

[PATCH 1/3] drm/amdgpu: delete duplicated printk info.

[Rex Zhu]

Change-Id: I5f336cc41b1868b98f263874fe5c6ea20e430744
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 57ccac4..eaf44fb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2288,9 +2288,6 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
 	if (r)
 		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
 
-	if (r)
-		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
-
 	amdgpu_fence_driver_resume(adev);
 
 	if (resume) {
-- 
1.9.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 2/3] drm/amdgpu: when resume failed, return error to avoid system hang.

[Rex Zhu]

Change-Id: I9e908edbb3f0c767b084ffaac7de5f97047eaa16
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index eaf44fb..18af88d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2285,9 +2285,10 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
 	}
 
 	r = amdgpu_resume(adev);
-	if (r)
+	if (r) {
 		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
-
+		return r;
+	}
 	amdgpu_fence_driver_resume(adev);
 
 	if (resume) {
-- 
1.9.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 3/3] drm/amdgpu: do not block gpu resume when cp ring test failed.

[Rex Zhu]

As kiq enabled, ring test may randomly failed because of timing.
it was not critial error, gpu can continue to resume.

Change-Id: I5d85c7d5b5989dd6892a13b47a5322b33e8a48ac
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 31b1312..cdda93b 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -4975,8 +4975,7 @@ static int gfx_v8_0_kiq_resume(struct amdgpu_device *adev)
 			goto done;
 
 		ring->ready = true;
-		r = amdgpu_ring_test_ring(ring);
-		if (r)
+		if (amdgpu_ring_test_ring(ring))
 			ring->ready = false;
 	}
 
-- 
1.9.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 3/3] drm/amdgpu: do not block gpu resume when cp ring test failed.

[Michel Dänzer]

On 30/03/17 02:42 PM, Rex Zhu wrote:
> As kiq enabled, ring test may randomly failed because of timing.
> it was not critial error, gpu can continue to resume.
> 
> Change-Id: I5d85c7d5b5989dd6892a13b47a5322b33e8a48ac
> Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> index 31b1312..cdda93b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> @@ -4975,8 +4975,7 @@ static int gfx_v8_0_kiq_resume(struct amdgpu_device *adev)
>  			goto done;
>  
>  		ring->ready = true;
> -		r = amdgpu_ring_test_ring(ring);
> -		if (r)
> +		if (amdgpu_ring_test_ring(ring))
>  			ring->ready = false;
>  	}
>  
> 

The random failures in amdgpu_ring_test_ring need to be fixed instead.
Failure in there normally does mean we cannot continue.


-- 
Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 1/3] drm/amdgpu: delete duplicated printk info.

[Michel Dänzer]

On 30/03/17 02:42 PM, Rex Zhu wrote:
> Change-Id: I5f336cc41b1868b98f263874fe5c6ea20e430744
> Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 ---
>  1 file changed, 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 57ccac4..eaf44fb 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -2288,9 +2288,6 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
>  	if (r)
>  		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
>  
> -	if (r)
> -		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
> -
>  	amdgpu_fence_driver_resume(adev);
>  
>  	if (resume) {
> 

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>


-- 
Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 1/3] drm/amdgpu: delete duplicated printk info.

[Christian König]

Patch 1 and 2 are Reviewed-by: Christian König <christian.koenig@amd.com>.

Patch 3 is a clear NAK. The ring test is there to make sure that the 
hardware is working as expected before proceeding.

If it fails randomly with KIQ enabled we need to figure out why and not 
just disable the test.

Christian.

Am 30.03.2017 um 07:42 schrieb Rex Zhu:
> Change-Id: I5f336cc41b1868b98f263874fe5c6ea20e430744
> Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 ---
>   1 file changed, 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 57ccac4..eaf44fb 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -2288,9 +2288,6 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
>   	if (r)
>   		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
>   
> -	if (r)
> -		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
> -
>   	amdgpu_fence_driver_resume(adev);
>   
>   	if (resume) {


_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

RE: [PATCH 3/3] drm/amdgpu: do not block gpu resume when cp ring test failed.

[Zhu, Rex]

Yes, we tried to fix it. it is a timing issue.
But it was not a critical error. So no need to block gpu resume.

In fact, We only return error when the last ring test failed in the original logic 

Best Regards
Rex
 
-----Original Message-----
From: Michel Dänzer [mailto:michel@daenzer.net] 
Sent: Thursday, March 30, 2017 2:14 PM
To: Zhu, Rex
Cc: amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH 3/3] drm/amdgpu: do not block gpu resume when cp ring test failed.

On 30/03/17 02:42 PM, Rex Zhu wrote:
> As kiq enabled, ring test may randomly failed because of timing.
> it was not critial error, gpu can continue to resume.
> 
> Change-Id: I5d85c7d5b5989dd6892a13b47a5322b33e8a48ac
> Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> index 31b1312..cdda93b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
> @@ -4975,8 +4975,7 @@ static int gfx_v8_0_kiq_resume(struct amdgpu_device *adev)
>  			goto done;
>  
>  		ring->ready = true;
> -		r = amdgpu_ring_test_ring(ring);
> -		if (r)
> +		if (amdgpu_ring_test_ring(ring))
>  			ring->ready = false;
>  	}
>  
> 

The random failures in amdgpu_ring_test_ring need to be fixed instead.
Failure in there normally does mean we cannot continue.


-- 
Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 3/3] drm/amdgpu: do not block gpu resume when cp ring test failed.

[Michel Dänzer]

On 30/03/17 03:45 PM, Zhu, Rex wrote:
> Yes, we tried to fix it. it is a timing issue.
> But it was not a critical error. So no need to block gpu resume.
> 
> In fact, We only return error when the last ring test failed in the original logic 

That makes some sense to me, but at least the commit log would need to
be changed not to talk about the KIQ related random failures but just
about that gfx_v8_0_kiq_resume would only return failure if the last
ring test failed anyway.


-- 
Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx