| Message ID | 20180705145845.24005-1-chris@chris-wilson.co.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Jul 05, 2018 at 03:58:45PM +0100, Chris Wilson wrote: > If the GPU is terminally wedged we cannot submit any requests into a > context, completely unfulfilling our purpose of doing so. As this > expectedly fails, skip over the test. > > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> > --- > drivers/gpu/drm/i915/selftests/i915_gem_context.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_context.c b/drivers/gpu/drm/i915/selftests/i915_gem_context.c > index cc848ceeb3c3..0b36265a0f96 100644 > --- a/drivers/gpu/drm/i915/selftests/i915_gem_context.c > +++ b/drivers/gpu/drm/i915/selftests/i915_gem_context.c > @@ -599,6 +599,9 @@ int i915_gem_context_live_selftests(struct drm_i915_private *dev_priv) > bool fake_alias = false; > int err; > > + if (i915_terminally_wedged(&dev_priv->gpu_error)) > + return 0; > + I wonder if this could mask a real failure under the skips? but if actual error is already captured somewhere else than probably better to avoid the noise of a ENOTRECOVERABLE or whatever... > /* Install a fake aliasing gtt for exercise */ > if (USES_PPGTT(dev_priv) && !dev_priv->mm.aliasing_ppgtt) { > mutex_lock(&dev_priv->drm.struct_mutex); > -- > 2.18.0 > > _______________________________________________ > Intel-gfx mailing list > Intel-gfx@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoting Rodrigo Vivi (2018-07-05 21:52:10) > On Thu, Jul 05, 2018 at 03:58:45PM +0100, Chris Wilson wrote: > > If the GPU is terminally wedged we cannot submit any requests into a > > context, completely unfulfilling our purpose of doing so. As this > > expectedly fails, skip over the test. > > > > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> > > --- > > drivers/gpu/drm/i915/selftests/i915_gem_context.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_context.c b/drivers/gpu/drm/i915/selftests/i915_gem_context.c > > index cc848ceeb3c3..0b36265a0f96 100644 > > --- a/drivers/gpu/drm/i915/selftests/i915_gem_context.c > > +++ b/drivers/gpu/drm/i915/selftests/i915_gem_context.c > > @@ -599,6 +599,9 @@ int i915_gem_context_live_selftests(struct drm_i915_private *dev_priv) > > bool fake_alias = false; > > int err; > > > > + if (i915_terminally_wedged(&dev_priv->gpu_error)) > > + return 0; > > + > > I wonder if this could mask a real failure under the skips? The *test* can't be run, so what failure relevant to this *test* can be shown? As you notice, when we get to the reset test, we do proclaim failure as we've already demonstrated reset is bust. -Chris
On Fri, Jul 06, 2018 at 07:42:00AM +0100, Chris Wilson wrote: > Quoting Rodrigo Vivi (2018-07-05 21:52:10) > > On Thu, Jul 05, 2018 at 03:58:45PM +0100, Chris Wilson wrote: > > > If the GPU is terminally wedged we cannot submit any requests into a > > > context, completely unfulfilling our purpose of doing so. As this > > > expectedly fails, skip over the test. > > > > > > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> > > > --- > > > drivers/gpu/drm/i915/selftests/i915_gem_context.c | 3 +++ > > > 1 file changed, 3 insertions(+) > > > > > > diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_context.c b/drivers/gpu/drm/i915/selftests/i915_gem_context.c > > > index cc848ceeb3c3..0b36265a0f96 100644 > > > --- a/drivers/gpu/drm/i915/selftests/i915_gem_context.c > > > +++ b/drivers/gpu/drm/i915/selftests/i915_gem_context.c > > > @@ -599,6 +599,9 @@ int i915_gem_context_live_selftests(struct drm_i915_private *dev_priv) > > > bool fake_alias = false; > > > int err; > > > > > > + if (i915_terminally_wedged(&dev_priv->gpu_error)) > > > + return 0; > > > + > > > > I wonder if this could mask a real failure under the skips? > > The *test* can't be run, so what failure relevant to this *test* can be > shown? > > As you notice, when we get to the reset test, we do proclaim failure as > we've already demonstrated reset is bust. Makes sense... I was going to add rv-b here for this and others, but I saw you already got hem and is already pushing ;) Also thanks for the explanation on -ENOTRECOVERABLE one.. that also makes sense. > -Chris
diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_context.c b/drivers/gpu/drm/i915/selftests/i915_gem_context.c index cc848ceeb3c3..0b36265a0f96 100644 --- a/drivers/gpu/drm/i915/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/selftests/i915_gem_context.c @@ -599,6 +599,9 @@ int i915_gem_context_live_selftests(struct drm_i915_private *dev_priv) bool fake_alias = false; int err; + if (i915_terminally_wedged(&dev_priv->gpu_error)) + return 0; + /* Install a fake aliasing gtt for exercise */ if (USES_PPGTT(dev_priv) && !dev_priv->mm.aliasing_ppgtt) { mutex_lock(&dev_priv->drm.struct_mutex);
If the GPU is terminally wedged we cannot submit any requests into a context, completely unfulfilling our purpose of doing so. As this expectedly fails, skip over the test. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> --- drivers/gpu/drm/i915/selftests/i915_gem_context.c | 3 +++ 1 file changed, 3 insertions(+)