[Intel-gfx] [igt-dev] [PATCH i-g-t] igt/gem_render_copy: Check for GEM before running

[Chris Wilson]

Quoting Ville Syrjälä (2018-07-09 14:19:40)
> On Mon, Jul 09, 2018 at 11:28:47AM +0100, Chris Wilson wrote:
> > gem_render_copy requires a working GPU so check first.
> > 
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > ---
> >  tests/gem_render_copy.c | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/tests/gem_render_copy.c b/tests/gem_render_copy.c
> > index 8373cd738..238e70e97 100644
> > --- a/tests/gem_render_copy.c
> > +++ b/tests/gem_render_copy.c
> > @@ -529,6 +529,7 @@ int main(int argc, char **argv)
> >       igt_fixture {
> >               data.drm_fd = drm_open_driver_render(DRIVER_INTEL);
> >               data.devid = intel_get_drm_devid(data.drm_fd);
> > +             igt_require_gem(data.drm_fd);
> 
> Is there ever a case where we'd be happy with open_driver() failing?
> Just wondering about putting the require into the function itself.

I suppose we could add require_gem to open_driver_render(), but it
doesn't seem that generic. We definitely can do KMS operations even if
execbuf() returns -EIO.

At some point, we'll just have to bite the bullet and start rewriting
GEM tests to apply to amdgpu as well -- there are quite a few that I
think aren't i915 centric, e.g. gem_sync, that might help define what we
can and what we can't generalise.
-Chris