[PATCH] drm/amdgpu: Set GTT_USWC flag to enable freesync v2

[Christian König]

Thanks Shashank,

Alex when do we plan the next rebase?

This also depends on upstream code and can't be merged in 
amd-staging-drm-next directly.

Regards,
Christian.

Am 15.02.21 um 13:27 schrieb Sharma, Shashank:
> [AMD Official Use Only - Internal Distribution Only]
>
> Hello Christian,
> I just tested the patch with Xonotic + PRIME,
>
> I confirm that its working same as previous version, for the Freesync scenario, ie:
> - With this patch, display framebuffer creation is successful, flips are coming and VRR property is getting set.
>
> Did some negative testing also,
> - Without this patch, Fremebuffer creation fails, no flips,  and VRR property doesn't set.
>
> Regards
> Shashank
> -----Original Message-----
> From: Christian König <ckoenig.leichtzumerken at gmail.com>
> Sent: Monday, February 15, 2021 5:15 PM
> To: amd-gfx at lists.freedesktop.org; Sharma, Shashank <Shashank.Sharma at amd.com>
> Subject: Re: [PATCH] drm/amdgpu: Set GTT_USWC flag to enable freesync v2
>
> Hi Shashank,
>
> can you give this patch a test round? In theory it should work, but I'm not 100% sure.
>
> Thanks,
> Christian.
>
> Am 15.02.21 um 12:43 schrieb Christian König:
>> From: Shashank Sharma <shashank.sharma at amd.com>
>>
>> This patch sets 'AMDGPU_GEM_CREATE_CPU_GTT_USWC' as input parameter
>> flag, during object creation of an imported DMA buffer.
>>
>> In absence of this flag:
>> 1. Function amdgpu_display_supported_domains() doesn't add
>>      AMDGPU_GEM_DOMAIN_GTT as supported domain.
>> 2. Due to which, Function amdgpu_display_user_framebuffer_create()
>>      refuses to create framebuffer for imported DMA buffers.
>> 3. Due to which, AddFB() IOCTL fails.
>> 4. Due to which, amdgpu_present_check_flip() check fails in DDX 5. Due
>> to which DDX driver doesn't allow flips (goes to blitting) 6. Due to
>> which setting Freesync/VRR property fails for PRIME buffers.
>>
>> So, this patch finally enables Freesync with PRIME buffer offloading.
>>
>> v2 (chk): instead of just checking the flag we copy it over if the
>>             exporter is an amdgpu device as well.
>>
>> Signed-off-by: Shashank Sharma <shashank.sharma at amd.com>
>> Signed-off-by: Christian König <christian.koenig at amd.com>
>> ---
>>    drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c | 14 +++++++++++---
>>    1 file changed, 11 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> index d3727f3ab277..a54a870f006c 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> @@ -422,14 +422,22 @@ amdgpu_dma_buf_create_obj(struct drm_device *dev, struct dma_buf *dma_buf)
>>    {
>>    	struct dma_resv *resv = dma_buf->resv;
>>    	struct amdgpu_device *adev = drm_to_adev(dev);
>> -	struct amdgpu_bo *bo;
>>    	struct drm_gem_object *gobj;
>> +	struct amdgpu_bo *bo;
>> +	uint64_t flags = 0;
>>    	int ret;
>>    
>>    	dma_resv_lock(resv, NULL);
>> +
>> +	if (dma_buf->ops == &amdgpu_dmabuf_ops) {
>> +		struct amdgpu_bo *other = gem_to_amdgpu_bo(dma_buf->priv);
>> +
>> +		flags |= other->flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC;
>> +	}
>> +
>>    	ret = amdgpu_gem_object_create(adev, dma_buf->size, PAGE_SIZE,
>> -			AMDGPU_GEM_DOMAIN_CPU,
>> -			0, ttm_bo_type_sg, resv, &gobj);
>> +				       AMDGPU_GEM_DOMAIN_CPU, flags,
>> +				       ttm_bo_type_sg, resv, &gobj);
>>    	if (ret)
>>    		goto error;
>>