On Wed, Oct 07, 2020 at 10:34:36AM +0100, Mel Gorman wrote: > On Mon, Oct 05, 2020 at 02:26:48PM +0200, Frederic Weisbecker wrote: > > On Mon, Oct 05, 2020 at 01:23:53PM +0200, Peter Zijlstra wrote: > > > On Mon, Oct 05, 2020 at 12:49:17PM +0200, Frederic Weisbecker wrote: > > > > Detect calls to schedule() between user_enter() and user_exit(). Those > > > > are symptoms of early entry code that either forgot to protect a call > > > > to schedule() inside exception_enter()/exception_exit() or, in the case > > > > of HAVE_CONTEXT_TRACKING_OFFSTACK, enabled interrupts or preemption in > > > > a wrong spot. > > > > > > > > Signed-off-by: Frederic Weisbecker <frede...@kernel.org> > > > > Cc: Marcelo Tosatti <mtosa...@redhat.com> > > > > Cc: Paul E. McKenney <paul...@kernel.org> > > > > Cc: Peter Zijlstra <pet...@infradead.org> > > > > Cc: Phil Auld <pa...@redhat.com> > > > > Cc: Thomas Gleixner <t...@linutronix.de> > > > > --- > > > > kernel/sched/core.c | 1 + > > > > 1 file changed, 1 insertion(+) > > > > > > > > diff --git a/kernel/sched/core.c b/kernel/sched/core.c > > > > index 2d95dc3f4644..d31a79e073e3 100644 > > > > --- a/kernel/sched/core.c > > > > +++ b/kernel/sched/core.c > > > > @@ -4295,6 +4295,7 @@ static inline void schedule_debug(struct > > > > task_struct *prev, bool preempt) > > > > preempt_count_set(PREEMPT_DISABLED); > > > > } > > > > rcu_sleep_check(); > > > > + WARN_ON_ONCE(ct_state() == CONTEXT_USER); > > > > > > SCHED_WARN_ON() ? > > > > Bah! That's exactly what I was looking for. > > > > > No point in unconditionally polluting that path. Although, per MeL, we > > > should probably invest in CONFIG_SCHED_DEBUG_I_MEANS_IT :/ > > > > Because CONFIG_SCHED_DEBUG is often used by default on distros? > > > > SCHED_DEBUG is generally useful (e.g. figuring out weird topology problems > on new hardware). The overhead isn't too bad when schedstats are > disabled so it would be nice to avoid adding too much overhead via > SCHED_DEBUG. > > Other debugging options -- not so much. A lot of them are useful for > development but there are people who request them be enabled anyway > thinking that they improve security somehow when in reality they might, > at best, detect a hardware issue that happens to hit a specific structure.
So we are good with SCHED_WARN_ON(), right? I'll reissue with that. Thanks.
