On Fri, Aug 2, 2019 at 5:06 PM Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote:
>
> Quoting Matthew Auld (2019-08-02 16:00:18)
> > On Tue, 30 Jul 2019 at 21:58, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > Since commit 64d6c500a384 ("drm/i915: Generalise GPU activity
> > > tracking"), we have been prepared for i915_vma_move_to_active() to fail.
> > > We can take advantage of this to report the failure for allocating the
> > > shared-fence slot in the reservation_object.
> > >
> > > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> > Reviewed-by: Matthew Auld <matthew.auld@xxxxxxxxx>
>
> Thanks. One day, we should lift even the reservation allocation out of
> the reservation lock. (That will make it move convenient to use that
> lock from inside fs-reclaim paths.)
Being able to allocate memory while holding the reservation_obj lock
is fairly fundamental assumption in all ttm based drivers. So I dont
think that'll work out, since even for our own objects dma-buf can
share them widely.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
