Re: [PATCH bpf-next v8 11/11] selftests: Remove fmod_ret from benchmarks and test_overhead

[Andrii Nakryiko <andrii.nakryiko@gmail.com>]

On Wed, Sep 23, 2020 at 6:08 PM Alexei Starovoitov
<alexei.starovoitov@gmail.com> wrote:
>
> On Tue, Sep 22, 2020 at 08:38:45PM +0200, Toke Høiland-Jørgensen wrote:
> > -const struct bench bench_trig_fmodret = {
> > -     .name = "trig-fmodret",
> > -     .validate = trigger_validate,
> > -     .setup = trigger_fmodret_setup,
> > -     .producer_thread = trigger_producer,
> > -     .consumer_thread = trigger_consumer,
> > -     .measure = trigger_measure,
> > -     .report_progress = hits_drops_report_progress,
> > -     .report_final = hits_drops_report_final,
> > -};
> > diff --git a/tools/testing/selftests/bpf/progs/trigger_bench.c b/tools/testing/selftests/bpf/progs/trigger_bench.c
> > index 9a4d09590b3d..1af23ac0c37c 100644
> > --- a/tools/testing/selftests/bpf/progs/trigger_bench.c
> > +++ b/tools/testing/selftests/bpf/progs/trigger_bench.c
> > @@ -45,10 +45,3 @@ int bench_trigger_fentry_sleep(void *ctx)
> >       __sync_add_and_fetch(&hits, 1);
> >       return 0;
> >  }
> > -
> > -SEC("fmod_ret/__x64_sys_getpgid")
> > -int bench_trigger_fmodret(void *ctx)
> > -{
> > -     __sync_add_and_fetch(&hits, 1);
> > -     return -22;
> > -}
>
> why are you removing this? There is no problem here.
> All syscalls are error-injectable.
> I'm surprised Andrii acked this :(

Andrii didn't know that all syscalls are error-injectable, thanks for
catching :) after fmod_ret/__set_task_comm I just assumed that I've
been abusing fmod_ret all this time...