Quoting Chris Wilson (2019-11-26 17:22:23)
> Quoting Tvrtko Ursulin (2019-11-26 17:04:43)
> >
> > On 26/11/2019 15:26, Chris Wilson wrote:
> > > diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
> > > index e5512f26e20a..f6e661428b02 100644
> > > --- a/drivers/gpu/drm/i915/i915_vma.c
> > > +++ b/drivers/gpu/drm/i915/i915_vma.c
> > > @@ -905,6 +905,12 @@ int i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
> > > __i915_vma_set_map_and_fenceable(vma);
> > > }
> > >
> > > + /* Somebody else managed to gazump our placement? */
> > > + if (i915_vma_misplaced(vma, size, alignment, flags)) {
> > > + err = -EAGAIN;
> > > + goto err_active;
> > > + }
> > > +
> >
> > What about other callers? They will not be expecting this.
>
> The other paths should be quite happy with -EAGAIN from vma_pin, it's
> already part of their retry procedure. If not, there's always more duct
> tape. Hopefully the replacement is much simpler (stop laughing back
> there).
The alternative here is to pull in __i915_vma_unbind(), which might be
plausible.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
