[Intel-gfx] [PATCH igt 11/24] igt/drv_hangman: Skip if resets are disallowed

[Chris Wilson]

Quoting Daniel Vetter (2017-08-15 09:52:21)
> On Mon, Aug 14, 2017 at 09:18:35PM +0100, Chris Wilson wrote:
> > If we tell the machine to reset but they are disallowed, we will leave
> > the system in a wedged state, preventing the majority of subsequent
> > tests.
> > ---
> >  tests/drv_hangman.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/tests/drv_hangman.c b/tests/drv_hangman.c
> > index 0551ec16..de57e128 100644
> > --- a/tests/drv_hangman.c
> > +++ b/tests/drv_hangman.c
> > @@ -30,6 +30,7 @@
> >  #include <limits.h>
> >  #include <sys/types.h>
> >  #include <sys/stat.h>
> > +#include <sys/ioctl.h>
> >  #include <fcntl.h>
> >  
> >  #include "igt_sysfs.h"
> > @@ -239,7 +240,9 @@ igt_main
> >               int idx;
> >  
> >               device = drm_open_driver(DRIVER_INTEL);
> > +             igt_allow_hang(device, 0, 0);
> 
> I think an igt_require_hang_ring here would be better self-documenting.
> Also, igt_allow/disallow_hang are lacking gtkdocs.

Not the same unfortunately.
 
> What I'm also wondering is where our unexpected hang detection has gone
> to.

It's "a significant but normal condition" since we recover, so it's a
KERN_NOTE. If we fail to reset, then we add a KERN_ERR. It's still
there, the logs are flooded with them.

For igt, we have unexpected hang detection which has to enabled when we
don't expect hangs and disabled when we do, because we also don't want
the side-effects for simulated hangs
-Chris