On Wed, Nov 08, 2017 at 07:33:05PM +0000, Chris Wilson wrote:
> Quoting Ville Syrjälä (2017-11-08 19:27:41)
> > On Wed, Nov 08, 2017 at 07:14:58PM +0000, Chris Wilson wrote:
> > > +out_unlock:
> > > if (ret == -EIO) {
> > > /* Allow engine initialisation to fail by marking the GPU as
> > > * wedged. But we only want to do this where the GPU is angry,
> > > @@ -5035,8 +5049,6 @@ int i915_gem_init(struct drm_i915_private *dev_priv)
> > > }
> > > ret = 0;
> > > }
> > > -
> > > -out_unlock:
> >
> > Is the movement of the label going to adversely affect error handling for
> > the other functions which already use the label? Or none of them can
> > return -EIO?
>
> None of those can return -EIO, but if they did the same principle
> applies to them. We would much prefer to disable GEM submission than
> disable the driver; the theory being as always if the user can see the
> display and the error messages, they can report a bug. If the driver
> dies in the middle of loading, it's likely they won't be able to see
> anything at all (since we've already kicked out the VGA console and it
> is not coming back).
Ack.
Reviewed-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
--
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
