sched/all: Change all BUG_ON() instances in the scheduler to WARN_ON_ONCE()

There's no good reason to crash a user's system with a BUG_ON(),
chances are high that they'll never even see the crash message on
Xorg, and it won't make it into the syslog either.

By using a WARN_ON_ONCE() we at least give the user a chance to report
any bugs triggered here - instead of getting silent hangs.

None of these WARN_ON_ONCE()s are supposed to trigger, ever - so we ignore
cases where a NULL check is done via a BUG_ON() and we let a NULL
pointer through after a WARN_ON_ONCE().

There's one exception: WARN_ON_ONCE() arguments with side-effects,
such as locking - in this case we use the return value of the
WARN_ON_ONCE(), such as in:

 -       BUG_ON(!lock_task_sighand(p, &flags));
 +       if (WARN_ON_ONCE(!lock_task_sighand(p, &flags)))
 +               return;

Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/YvSsKcAXISmshtHo@gmail.com

1 files changed, 3 insertions, 3 deletions

diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index e26688d387ae..7a44dceeb50a 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -2709,8 +2709,8 @@ static inline void double_rq_lock(struct rq *rq1, struct rq *rq2)
 	__acquires(rq1->lock)
 	__acquires(rq2->lock)
 {
-	BUG_ON(!irqs_disabled());
-	BUG_ON(rq1 != rq2);
+	WARN_ON_ONCE(!irqs_disabled());
+	WARN_ON_ONCE(rq1 != rq2);
 	raw_spin_rq_lock(rq1);
 	__acquire(rq2->lock);	/* Fake it out ;) */
 	double_rq_clock_clear_update(rq1, rq2);
@@ -2726,7 +2726,7 @@ static inline void double_rq_unlock(struct rq *rq1, struct rq *rq2)
 	__releases(rq1->lock)
 	__releases(rq2->lock)
 {
-	BUG_ON(rq1 != rq2);
+	WARN_ON_ONCE(rq1 != rq2);
 	raw_spin_rq_unlock(rq1);
 	__release(rq2->lock);
 }