[PATCH 07/10] drm/amdgpu/ttm: handle tt moves properly.

[Christian König]

Am 23.09.20 um 05:04 schrieb Dave Airlie:
> From: Dave Airlie <airlied at redhat.com>
>
> The core move code currently handles use_tt moves, for amdgpu
> this was being handled also in the driver, but not using the same
> paths.
>
> If moving between TT/SYSTEM (all the use_tt paths on amdgpu) use
> the core move function.
>
> Eventually the core will be flipped over to calling the driver.
>
> Signed-off-by: Dave Airlie <airlied at redhat.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 12 +++++++-----
>   1 file changed, 7 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> index db5f761f37ec..d3bd2fd448be 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> @@ -671,14 +671,16 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict,
>   		ttm_bo_move_null(bo, new_mem);
>   		return 0;
>   	}
> -	if ((old_mem->mem_type == TTM_PL_TT &&
> -	     new_mem->mem_type == TTM_PL_SYSTEM) ||
> -	    (old_mem->mem_type == TTM_PL_SYSTEM &&
> -	     new_mem->mem_type == TTM_PL_TT)) {
> -		/* bind is enough */
> +	if (old_mem->mem_type == TTM_PL_SYSTEM &&
> +	    new_mem->mem_type == TTM_PL_TT) {
>   		ttm_bo_move_null(bo, new_mem);

I would feel better if we nuke ttm_bo_move_null() and always use 
ttm_bo_move_ttm().

Christian.

>   		return 0;
>   	}
> +
> +	if (old_mem->mem_type == TTM_PL_TT &&
> +	    new_mem->mem_type == TTM_PL_SYSTEM)
> +		return ttm_bo_move_ttm(bo, ctx, new_mem);
> +
>   	if (old_mem->mem_type == AMDGPU_PL_GDS ||
>   	    old_mem->mem_type == AMDGPU_PL_GWS ||
>   	    old_mem->mem_type == AMDGPU_PL_OA ||