[Intel-gfx] [PATCH] drm/i915/gem: Remove gpu reloc workaround

[Matthew Auld]

On Wed, 27 Oct 2021 at 09:36, Thomas Hellström
<thomas.hellstrom at linux.intel.com> wrote:
>
> GPU relocs are gone. There should be no need for this workaround anymore.
> Remove it.
>
> Signed-off-by: Thomas Hellström <thomas.hellstrom at linux.intel.com>

I was completely wrong here, sorry. Digging through the git history it
looks like this came from:

Commit 149c84077fe717af883bae459623ef1cebd86388
Author: Daniel Vetter <daniel.vetter at ffwll.ch>
Date:   Wed Feb 15 23:50:23 2012 +0100

    drm/i915: implement SNB workaround for lazy global gtt

    PIPE_CONTROL on snb needs global gtt mappings in place to workaround a
    hw gotcha. No other commands need such a workaround. Luckily we can
    detect a PIPE_CONTROL commands easily because they have a write_domain
    = I915_GEM_DOMAIN_INSTRUCTION (and nothing else has that).

so it looks to be unrelated to GPU relocs, which AFAIK didn't exist at
the time. I just saw the MI comment and assumed it was talking about
the MI_STORE_DWORD...

> ---
>  .../gpu/drm/i915/gem/i915_gem_execbuffer.c    | 19 +------------------
>  1 file changed, 1 insertion(+), 18 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
> index 1231224728e4..04af88e6d453 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
> @@ -1327,7 +1327,6 @@ eb_relocate_entry(struct i915_execbuffer *eb,
>  {
>         struct drm_i915_private *i915 = eb->i915;
>         struct eb_vma *target;
> -       int err;
>
>         /* we've already hold a reference to all valid objects */
>         target = eb_get_vma(eb, reloc->target_handle);
> @@ -1357,25 +1356,9 @@ eb_relocate_entry(struct i915_execbuffer *eb,
>                 return -EINVAL;
>         }
>
> -       if (reloc->write_domain) {
> +       if (reloc->write_domain)
>                 target->flags |= EXEC_OBJECT_WRITE;
>
> -               /*
> -                * Sandybridge PPGTT errata: We need a global gtt mapping
> -                * for MI and pipe_control writes because the gpu doesn't
> -                * properly redirect them through the ppgtt for non_secure
> -                * batchbuffers.
> -                */
> -               if (reloc->write_domain == I915_GEM_DOMAIN_INSTRUCTION &&
> -                   GRAPHICS_VER(eb->i915) == 6) {
> -                       err = i915_vma_bind(target->vma,
> -                                           target->vma->obj->cache_level,
> -                                           PIN_GLOBAL, NULL);
> -                       if (err)
> -                               return err;
> -               }
> -       }
> -
>         /*
>          * If the relocation already has the right value in it, no
>          * more work needs to be done.
> --
> 2.31.1
>