Matthew Auld <matthew.auld@xxxxxxxxx> writes:
> If we need to mark the PDE as operating in 64K GTT mode, we should be
> paranoid and flush the extra writes, like we already do for the PTEs. On
> some platforms the clflush can apparently add the just the right amount
> of magical delay to force the GPU to see the updated entry.
>
> Signed-off-by: Matthew Auld <matthew.auld@xxxxxxxxx>
> Cc: Mika Kuoppala <mika.kuoppala@xxxxxxxxxxxxxxx>
Makes sense to follow the same pattern as the other writes.
Reviewed-by: Mika Kuoppala <mika.kuoppala@xxxxxxxxxxxxxxx>
> ---
> drivers/gpu/drm/i915/gt/gen8_ppgtt.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
> index 6e0e52eeb87a..6a5af995f5b1 100644
> --- a/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
> +++ b/drivers/gpu/drm/i915/gt/gen8_ppgtt.c
> @@ -548,6 +548,7 @@ static void gen8_ppgtt_insert_huge(struct i915_vma *vma,
> I915_GTT_PAGE_SIZE_2M)))) {
> vaddr = px_vaddr(pd);
> vaddr[maybe_64K] |= GEN8_PDE_IPS_64K;
> + clflush_cache_range(vaddr, PAGE_SIZE);
> page_size = I915_GTT_PAGE_SIZE_64K;
>
> /*
> @@ -568,6 +569,7 @@ static void gen8_ppgtt_insert_huge(struct i915_vma *vma,
> for (i = 1; i < index; i += 16)
> memset64(vaddr + i, encode, 15);
>
> + clflush_cache_range(vaddr, PAGE_SIZE);
> }
> }
>
> --
> 2.26.3
