[PATCH] drm/amdgpu: disable userptr if swiotlb is active

[Christian König]

Well in the worst case you could end up having a bounce buffer mapped 
instead of the real page.

Christian.

Am 20.02.19 um 18:02 schrieb Kuehling, Felix:
> I guess we'll need something similar for KFD? I don't think we've ever
> intentionally tested KFD with swiotlb. But I've seen some backtraces
> with swiotlb in them before. I wonder how badly broken it is ...
>
> Regards,
>     Felix
>
> On 2019-02-20 8:46 a.m., Christian König wrote:
>> Otherwise we can't be sure that we won't end up with a bounce buffer.
>>
>> Signed-off-by: Christian König <christian.koenig at amd.com>
>> ---
>>    drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 4 ++++
>>    1 file changed, 4 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>> index d21dd2f369da..abc65633119b 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>> @@ -289,6 +289,10 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
>>    	if (offset_in_page(args->addr | args->size))
>>    		return -EINVAL;
>>    
>> +	/* We can't do this when swiotlb is active */
>> +	if (adev->needs_swiotlb)
>> +		return -ENXIO;
>> +
>>    	/* reject unknown flag values */
>>    	if (args->flags & ~(AMDGPU_GEM_USERPTR_READONLY |
>>    	    AMDGPU_GEM_USERPTR_ANONONLY | AMDGPU_GEM_USERPTR_VALIDATE |