[PATCH] drm/amdkfd: change SDMA MQD memory type

[PATCH] drm/amdkfd: change SDMA MQD memory type

[Eric Huang]

SDMA MQD memory type is NC that causes MQD data overwritten
accidentally by an old stable cache line. Changing it to UC
default for GART will fix the issue.

Change-Id: If609f47c78cb97e2c8dc930df2ab5c10c29dfe56
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 692abfd..77ea0f0 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1734,7 +1734,7 @@ static int allocate_hiq_sdma_mqd(struct device_queue_manager *dqm)
 
 	retval = amdgpu_amdkfd_alloc_gtt_mem(dev->kgd, size,
 		&(mem_obj->gtt_mem), &(mem_obj->gpu_addr),
-		(void *)&(mem_obj->cpu_ptr), true);
+		(void *)&(mem_obj->cpu_ptr), false);
 
 	return retval;
 }
-- 
2.7.4

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

Re: [PATCH] drm/amdkfd: change SDMA MQD memory type

[Zhao, Yong]

[-- Attachment #1.1: Type: text/plain, Size: 2201 bytes --]

[AMD Official Use Only - Internal Distribution Only]

It looks good to me. I was thinking maybe we should go one step further, adding more explanation comments around the MQD control stack workaround, so that people have a clearer idea of what's that MQD control stack workaround is about. We can do that in a following commit.

Acked-by: Yong Zhao <Yong.Zhao@amd.com>
________________________________
From: amd-gfx <amd-gfx-bounces@lists.freedesktop.org> on behalf of Eric Huang <JinhuiEric.Huang@amd.com>
Sent: Wednesday, February 26, 2020 2:37 PM
To: amd-gfx@lists.freedesktop.org <amd-gfx@lists.freedesktop.org>
Cc: Huang, JinHuiEric <JinHuiEric.Huang@amd.com>
Subject: [PATCH] drm/amdkfd: change SDMA MQD memory type

SDMA MQD memory type is NC that causes MQD data overwritten
accidentally by an old stable cache line. Changing it to UC
default for GART will fix the issue.

Change-Id: If609f47c78cb97e2c8dc930df2ab5c10c29dfe56
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 692abfd..77ea0f0 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -1734,7 +1734,7 @@ static int allocate_hiq_sdma_mqd(struct device_queue_manager *dqm)

         retval = amdgpu_amdkfd_alloc_gtt_mem(dev->kgd, size,
                 &(mem_obj->gtt_mem), &(mem_obj->gpu_addr),
-               (void *)&(mem_obj->cpu_ptr), true);
+               (void *)&(mem_obj->cpu_ptr), false);

         return retval;
 }
--
2.7.4

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&amp;data=02%7C01%7Cyong.zhao%40amd.com%7C2cbbb3322952475590f508d7baf34ce6%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637183426468344453&amp;sdata=2zCqxZJFXSbBlezU7UIZxc%2FQC0PBTum90MmIjCioHGw%3D&amp;reserved=0

[-- Attachment #1.2: Type: text/html, Size: 5402 bytes --]

[-- Attachment #2: Type: text/plain, Size: 154 bytes --]

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

Re: [PATCH] drm/amdkfd: change SDMA MQD memory type

[Felix Kuehling]

On 2020-02-26 14:37, Eric Huang wrote:
> SDMA MQD memory type is NC that causes MQD data overwritten
> accidentally by an old stable cache line. Changing it to UC
> default for GART will fix the issue.

Maybe add a statement here, that the mqd_gfx9 parameter is meant for 
control stacks that are allocated together with user mode queue MQDs. 
Setting mqd_gfx9 to true maps the control stack pages as NC. Here it was 
accidentally applied to SDMA MQDs, which are allocated together with the 
HIQ MQD. Setting the mqd_gfx9 to false avoids that.


>
> Change-Id: If609f47c78cb97e2c8dc930df2ab5c10c29dfe56
> Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>

Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>


> ---
>   drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> index 692abfd..77ea0f0 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> @@ -1734,7 +1734,7 @@ static int allocate_hiq_sdma_mqd(struct device_queue_manager *dqm)
>   
>   	retval = amdgpu_amdkfd_alloc_gtt_mem(dev->kgd, size,
>   		&(mem_obj->gtt_mem), &(mem_obj->gpu_addr),
> -		(void *)&(mem_obj->cpu_ptr), true);
> +		(void *)&(mem_obj->cpu_ptr), false);
>   
>   	return retval;
>   }
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx