Re: [PATCH 1/4] sched/wakeup: Strengthen current_save_and_set_rtlock_wait_state()

[Peter Zijlstra]

On Fri, Sep 10, 2021 at 03:17:04PM +0200, Peter Zijlstra wrote:
> On Fri, Sep 10, 2021 at 01:57:26PM +0100, Will Deacon wrote:
> > On Thu, Sep 09, 2021 at 04:27:46PM +0200, Peter Zijlstra wrote:
> > > Moo yes, so the earlier changelog I wrote was something like:
> > > 
> > > 	current_save_and_set_rtlock_wait_state();
> > > 	for (;;) {
> > > 		if (try_lock())
> > > 			break;
> > > 
> > > 		raw_spin_unlock_irq(&lock->wait_lock);
> > > 		if (!cond)
> > > 			schedule();
> > > 		raw_spin_lock_irq(&lock->wait_lock);
> > > 
> > > 		set_current_state(TASK_RTLOCK_WAIT);
> > > 	}
> > > 	current_restore_rtlock_saved_state();
> > > 
> > > which is more what the code looks like before these patches, and in that
> > > case the @cond load can be lifted before __state.
> > 
> > Ah, so that makes more sense, thanks. I can't see how the try_lock() could
> > be reordered though, as it's going to have to do an atomic rmw.
> 
> OK, lemme go update the Changelog and make it __flags for bigeasy :-)

The patch now reads:

---
Subject: sched/wakeup: Strengthen current_save_and_set_rtlock_wait_state()
From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Date: Thu, 09 Sep 2021 12:59:16 +0200

While looking at current_save_and_set_rtlock_wait_state() I'm thinking
it really ought to use smp_store_mb(), because using it for a more
traditional wait loop like:

	current_save_and_set_rtlock_wait_state();
	for (;;) {
		if (cond)
			schedule();

		set_current_state(TASK_RTLOCK_WAIT);
	}
	current_restore_rtlock_saved_state();

is actually broken, since the cond load could be re-ordered against
the state store, which could lead to a missed wakeup -> BAD (tm).

While there, make them consistent with the IRQ usage in
set_special_state().

Fixes: 5f220be21418 ("sched/wakeup: Prepare for RT sleeping spin/rwlocks")
Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
Link: https://lkml.kernel.org/r/20210909110203.767330253@xxxxxxxxxxxxx
---
 include/linux/sched.h |   21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -218,7 +218,7 @@ struct task_group;
  */
 #define set_special_state(state_value)					\
 	do {								\
-		unsigned long flags; /* may shadow */			\
+		unsigned long __flags; /* may shadow */			\
 									\
 		raw_spin_lock_irqsave(&current->pi_lock, flags);	\
 		debug_special_state_change((state_value));		\
@@ -245,7 +245,8 @@ struct task_group;
  *		if (try_lock())
  *			break;
  *		raw_spin_unlock_irq(&lock->wait_lock);
- *		schedule_rtlock();
+ *		if (!cond)
+ *			schedule_rtlock();
  *		raw_spin_lock_irq(&lock->wait_lock);
  *		set_current_state(TASK_RTLOCK_WAIT);
  *	}
@@ -253,22 +254,24 @@ struct task_group;
  */
 #define current_save_and_set_rtlock_wait_state()			\
 	do {								\
-		lockdep_assert_irqs_disabled();				\
-		raw_spin_lock(&current->pi_lock);			\
+		unsigned long __flags; /* may shadow */			\
+									\
+		raw_spin_lock_irqsave(&current->pi_lock, flags);	\
 		current->saved_state = current->__state;		\
 		debug_rtlock_wait_set_state();				\
-		WRITE_ONCE(current->__state, TASK_RTLOCK_WAIT);		\
-		raw_spin_unlock(&current->pi_lock);			\
+		smp_store_mb(current->__state, TASK_RTLOCK_WAIT);	\
+		raw_spin_unlock_irqrestore(&current->pi_lock, flags);	\
 	} while (0);
 
 #define current_restore_rtlock_saved_state()				\
 	do {								\
-		lockdep_assert_irqs_disabled();				\
-		raw_spin_lock(&current->pi_lock);			\
+		unsigned long __flags; /* may shadow */			\
+									\
+		raw_spin_lock_irqsave(&current->pi_lock, flags);	\
 		debug_rtlock_wait_restore_state();			\
 		WRITE_ONCE(current->__state, current->saved_state);	\
 		current->saved_state = TASK_RUNNING;			\
-		raw_spin_unlock(&current->pi_lock);			\
+		raw_spin_unlock_irqrestore(&current->pi_lock, flags);	\
 	} while (0);
 
 #define get_current_state()	READ_ONCE(current->__state)