[igt-dev] [PATCH i-g-t] intel-ci: Broaden core_hotunplug blacklist

[Chris Wilson]

Quoting Petri Latvala (2020-11-17 15:18:24)
> On Tue, Nov 17, 2020 at 03:05:18PM +0000, Chris Wilson wrote:
> > We're close, but iommu fails upon rebinding leading to dead GPUs and
> > random failures in later tests. Random failures are a major nuisance.
> > 
> > We will have to see if we can find the fix and try again in the future.
> > 
> > Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2644
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Petri Latvala <petri.latvala at intel.com>
> > ---
> >  tests/intel-ci/blacklist.txt | 7 ++++---
> >  1 file changed, 4 insertions(+), 3 deletions(-)
> > 
> > diff --git a/tests/intel-ci/blacklist.txt b/tests/intel-ci/blacklist.txt
> > index 25b567038..37a049335 100644
> > --- a/tests/intel-ci/blacklist.txt
> > +++ b/tests/intel-ci/blacklist.txt
> > @@ -119,8 +119,9 @@ igt at .*@.*pipe-f($|-.*)
> >  igt at perf_pmu@cpu-hotplug
> >  
> >  # Currently fails and leaves the machine in a very bad state, and
> > -# causes coverage loss for other tests.
> > -igt at core_hotunplug@.*plug.*
> > +# causes coverage loss for other tests. IOMMU related.
> > +# https://gitlab.freedesktop.org/drm/intel/-/issues/2644
> > +igt at core_hotunplug@.*(hot|plug).*
> 
> The only subtest left unblocked will be unbind-rebind. That correct?

Yes. I don't recall seeing it before the mystery GPU hangs and since it
is run during BAT and we beat upon it afterwards, it has the benefit of
doubt. Because hotrebind-lateclose was the earlier troublemaker and still
causes problem, confirmation bias says I've seen it indicted for most
mystery hangs.

Speaking of confirmation bias, I'm having good results with hooking up
iommu_register_device_fault_handler().

Let's see how that works...
-Chris