[PATCH] drm/amdgpu: fix contiguous handling for IB parsing

[Dave Airlie]

On Wed, 24 Jul 2024 at 23:32, Alex Deucher <alexdeucher at gmail.com> wrote:
>
> On Wed, Jul 24, 2024 at 4:00 AM Christian König
> <ckoenig.leichtzumerken at gmail.com> wrote:
> >
> > Otherwise we won't get correct access to the IB.
> >
> > Signed-off-by: Christian König <christian.koenig at amd.com>
>
> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3501
> Fixes: e362b7c8f8c7 ("drm/amdgpu: Modify the contiguous flags behaviour")
>
> Reviewed-by: Alex Deucher <alexander.deucher at amd.com>

NAK (just capitialising to avoid it getting committed)

This doesn't fix it, I think you also need
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS set as well.

Dave.
>
> > ---
> >  drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 15 +++++++--------
> >  1 file changed, 7 insertions(+), 8 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> > index ec888fc6ead8..41055224930f 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
> > @@ -1763,7 +1763,7 @@ int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
> >         struct ttm_operation_ctx ctx = { false, false };
> >         struct amdgpu_vm *vm = &fpriv->vm;
> >         struct amdgpu_bo_va_mapping *mapping;
> > -       int r;
> > +       int i, r;
> >
> >         addr /= AMDGPU_GPU_PAGE_SIZE;
> >
> > @@ -1778,13 +1778,12 @@ int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
> >         if (dma_resv_locking_ctx((*bo)->tbo.base.resv) != &parser->exec.ticket)
> >                 return -EINVAL;
> >
> > -       if (!((*bo)->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)) {
> > -               (*bo)->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
> > -               amdgpu_bo_placement_from_domain(*bo, (*bo)->allowed_domains);
> > -               r = ttm_bo_validate(&(*bo)->tbo, &(*bo)->placement, &ctx);
> > -               if (r)
> > -                       return r;
> > -       }
> > +       amdgpu_bo_placement_from_domain(*bo, (*bo)->allowed_domains);
> > +       for (i = 0; i < (*bo)->placement.num_placement; i++)
> > +               (*bo)->placements[i].flags |= TTM_PL_FLAG_CONTIGUOUS;
> > +       r = ttm_bo_validate(&(*bo)->tbo, &(*bo)->placement, &ctx);
> > +       if (r)
> > +               return r;
> >
> >         return amdgpu_ttm_alloc_gart(&(*bo)->tbo);
> >  }
> > --
> > 2.34.1
> >