On Wed, May 08, 2024 at 07:33:38AM -0700, Paul E. McKenney wrote:
> On Wed, May 08, 2024 at 03:29:12PM +0300, Dan Carpenter wrote:
> > Hello David Woodhouse,
> >
> > Commit 82980b1622d9 ("rcu: Kill rnp->ofl_seq and use only
> > rcu_state.ofl_lock for exclusion") from Feb 16, 2021 (linux-next),
> > leads to the following Smatch static checker warning:
> >
> > kernel/rcu/tree.c:1844 rcu_gp_init()
> > warn: mixing irq and irqsave
>
> There are actually cases where this does make sense, one example being
> where some called function (for example, rcu_report_qs_rnp() below)
> needs a flags argument.
>
I only found one false positive which was kind of related to that in
__run_hrtimer().
1643
1644 static void __run_hrtimer(struct hrtimer_cpu_base *cpu_base,
1645 struct hrtimer_clock_base *base,
1646 struct hrtimer *timer, ktime_t *now,
1647 unsigned long flags) __must_hold(&cpu_base->lock)
^^^^^
1648 {
....
1678 /*
1679 * The timer is marked as running in the CPU base, so it is
1680 * protected against migration to a different CPU even if the lock
1681 * is dropped.
1682 */
1683 raw_spin_unlock_irqrestore(&cpu_base->lock, flags);
^^^^^
We potentially enable IRQs.
1684 trace_hrtimer_expire_entry(timer, now);
1685 expires_in_hardirq = lockdep_hrtimer_enter(timer);
1686
1687 restart = fn(timer);
1688
1689 lockdep_hrtimer_exit(expires_in_hardirq);
1690 trace_hrtimer_expire_exit(timer);
1691 raw_spin_lock_irq(&cpu_base->lock);
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Then disable them again.
1692
Of course the other warnings are mostly not "bugs" because the callers
haven't disabled IRQs. They're just in need of clean up.
regards,
dan carpenter
