[igt-dev] [Intel-gfx] [PATCH i-g-t] igt/gem_eio: Make reset-stress safe

[Chris Wilson]

Quoting Antonio Argenziano (2018-06-15 21:23:51)
> 
> 
> On 15/06/18 11:56, Chris Wilson wrote:
> > As we hang ctx0 quite frequently, it needs to be harden against being
> > banned.
> > 
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > Cc: Mika Kuoppala <mika.kuoppala at intel.com>
> 
> For some reason I thought we had more hang stress tests.

We do lots of resets, but not a lot of hangcheck testing (because
hangcheck requires waiting for a long time). gem_exec_whisper is a good
example if you want to wait a day for each hang subtest.
 
> Reviewed-by: Antonio Argenziano <antonio.argenziano at intel.com>
> 
> > ---
> >   tests/gem_eio.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/tests/gem_eio.c b/tests/gem_eio.c
> > index e1aff639d..5faf7502b 100644
> > --- a/tests/gem_eio.c
> > +++ b/tests/gem_eio.c
> > @@ -642,7 +642,7 @@ static void test_inflight_internal(int fd, unsigned int wait)
> >    */
> >   static void test_reset_stress(int fd, unsigned int flags)
> >   {
> > -     uint32_t ctx0 = gem_context_create(fd);
> > +     uint32_t ctx0 = context_create_safe(fd);
> 
> I guess until (before Mika's patch) now we were covered by 'ctx'
> being created and destroyed all the time.

Before it was just that we only counted *banned* contexts against the
client, and ctx0 has a good mix of valid batches to offset the few
hangs so it didn't get banned. It's a fair question that maybe we have
the heuristics wrong if the client gets banned before the context.
-Chris