irq_work: Fix irq_work_claim() memory ordering

When irq_work_claim() finds IRQ_WORK_PENDING flag already set, we just
return and don't raise a new IPI. We expect the destination to see
and handle our latest updades thanks to the pairing atomic_xchg()
in irq_work_run_list().

But cmpxchg() doesn't guarantee a full memory barrier upon failure. So
it's possible that the destination misses our latest updates.

So use atomic_fetch_or() instead that is unconditionally fully ordered
and also performs exactly what we want here and simplify the code.

Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul E . McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: https://lkml.kernel.org/r/20191108160858.31665-3-frederic@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>

kernel/irq_work.c

@@ -29,24 +29,16 @@ static DEFINE_PER_CPU(struct llist_head, lazy_list);
  */
 static bool irq_work_claim(struct irq_work *work)
 {
-	int flags, oflags, nflags;
+	int oflags;
 
+	oflags = atomic_fetch_or(IRQ_WORK_CLAIMED, &work->flags);
 	/*
-	 * Start with our best wish as a premise but only trust any
+	 * If the work is already pending, no need to raise the IPI.
-	 * flag value after cmpxchg() result.
+	 * The pairing atomic_xchg() in irq_work_run() makes sure
+	 * everything we did before is visible.
 	 */
-	flags = atomic_read(&work->flags) & ~IRQ_WORK_PENDING;
+	if (oflags & IRQ_WORK_PENDING)
-	for (;;) {
+		return false;
-		nflags = flags | IRQ_WORK_CLAIMED;
-		oflags = atomic_cmpxchg(&work->flags, flags, nflags);
-		if (oflags == flags)
-			break;
-		if (oflags & IRQ_WORK_PENDING)
-			return false;
-		flags = oflags;
-		cpu_relax();
-	}
-
 	return true;
 }