Quoting Belgaumkar, Vinay (2018-02-12 18:20:31)
>
>
> On 2/10/2018 1:00 AM, Chris Wilson wrote:
> > We run the per-engine scheduling smoketests across all engines, the
> > opposite of what was intended!
> >
> > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> > ---
> > tests/gem_exec_schedule.c | 12 ++++++++----
> > 1 file changed, 8 insertions(+), 4 deletions(-)
> >
> > diff --git a/tests/gem_exec_schedule.c b/tests/gem_exec_schedule.c
> > index b45ba1670..ec02d9943 100644
> > --- a/tests/gem_exec_schedule.c
> > +++ b/tests/gem_exec_schedule.c
> > @@ -208,11 +208,15 @@ static void smoketest(int fd, unsigned ring, unsigned timeout)
> > uint32_t *ptr;
> >
> > nengine = 0;
> > - for_each_engine(fd, engine) {
> > - if (ignore_engine(fd, engine))
> > - continue;
> > + if (ring == -1) {
> > + for_each_engine(fd, engine) {
> > + if (ignore_engine(fd, engine))
> > + continue;
> >
> > - engines[nengine++] = engine;
> > + engines[nengine++] = engine;
> > + }
> > + } else {
> > + engines[nengine++] = ring;
> > }
> > igt_require(nengine);
>
> LGTM. However, do we need the random number generation(from 0 to
> nengine) in the following code if a single engine is selected?
It's just a few instructions, if the bottleneck was in userspace then
yes, removing them would be useful. As it is, the test is an exercise in
struct_mutex contention.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
