ttm_tt_set_placement_caching on vram->ram transfers

[Christian König]

Am 28.07.20 um 05:23 schrieb Dave Airlie:
> Hi Christian + Ben,
>
> Just been reviewing around driver TTM code, and found an inconsistency,
>
> amdgpu + radeon both call the above before binding the ttm and going
> gpu vram->ram copies, but I don't see nouveau doing it Not sure if it
> could cause any issues, but it does look inconsistent.

What amdgpu and radeon do here sounds superfluous to me. The tt 
structure should have been allocated with the right caching attributes 
in the first place.

BTW: Changing the caching attributes of pages on the fly is utterly 
nonsense to begin with. This only works 100% correctly on x86, and there 
only by coincident :)

Christian.

>
> Dave.