[Intel-gfx] [PATCH igt 1/5] igt/gem_eio: inflight wedged requires long plugging

[Chris Wilson]

Quoting Michał Winiarski (2017-09-11 13:03:28)
> On Mon, Sep 11, 2017 at 09:55:58AM +0100, Chris Wilson wrote:
> > -     igt_post_hang_ring(fd, hang);
> > -     unplug(&cork); /* only now submit our batches */
> > -     igt_assert_eq(__gem_wait(fd, obj[1].handle, -1), 0);
> > +     fence = execbuf.rsvd2 >> 32;
> > +     igt_assert(fence != -1);
> > +
> > +     igt_post_hang_ring(fd, hang); /* wedged, with an unready batch */
> > +     sw_sync_timeline_inc(timeline, 1); /* only now submit our batches */
> > +
> > +     igt_assert_eq(__gem_wait(fd, obj.handle, -1), 0);
> 
> Is there any reason for rolling our own __gem_wait here? :>
> (not a part of this patch... just curious)

A modern invention.
-Chris