call_function_many: fix list delete vs add race

Peter pointed out there was nothing preventing the list_del_rcu in smp_call_function_interrupt from running before the list_add_rcu in smp_call_function_many. Fix this by not setting refs until we have gotten the lock for the list. Take advantage of the wmb in list_add_rcu to save an explicit additional one. I tried to force this race with a udelay before the lock & list_add and by mixing all 64 online cpus with just 3 random cpus in the mask, but was unsuccessful. Still, inspection shows a valid race, and the fix is a extension of the existing protection window in the current code. Cc: stable@kernel.org (v2.6.32 and later) Reported-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Milton Miller <miltonm@bga.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

call_function_many: fix list delete vs add race
Peter pointed out there was nothing preventing the list_del_rcu in smp_call_function_interrupt from running before the list_add_rcu in smp_call_function_many. Fix this by not setting refs until we have gotten the lock for the list. Take advantage of the wmb in list_add_rcu to save an explicit additional one. I tried to force this race with a udelay before the lock & list_add and by mixing all 64 online cpus with just 3 random cpus in the mask, but was unsuccessful. Still, inspection shows a valid race, and the fix is a extension of the existing protection window in the current code. Cc: stable@kernel.org (v2.6.32 and later) Reported-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Milton Miller <miltonm@bga.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Milton Miller · Linus Torvalds
1 parent ef2b4b95a6
Showing 1 changed file with 13 additions and 7 deletions Side-by-side Diff
kernel/smp.c
@@ -491,15 +491,16 @@
 	cpumask_clear_cpu(this_cpu, data->cpumask);
  
 	/*
-	 * To ensure the interrupt handler gets an complete view
-	 * we order the cpumask and refs writes and order the read
-	 * of them in the interrupt handler.  In addition we may
-	 * only clear our own cpu bit from the mask.
+	 * We reuse the call function data without waiting for any grace
+	 * period after some other cpu removes it from the global queue.
+	 * This means a cpu might find our data block as it is writen.
+	 * The interrupt handler waits until it sees refs filled out
+	 * while its cpu mask bit is set; here we may only clear our
+	 * own cpu mask bit, and must wait to set refs until we are sure
+	 * previous writes are complete and we have obtained the lock to
+	 * add the element to the queue.
 	 */
-	smp_wmb();
  
-	atomic_set(&data->refs, cpumask_weight(data->cpumask));
-
 	raw_spin_lock_irqsave(&call_function.lock, flags);
 	/*
 	 * Place entry at the _HEAD_ of the list, so that any cpu still
@@ -507,6 +508,11 @@
 	 * will not miss any other list entries:
 	 */
 	list_add_rcu(&data->csd.list, &call_function.queue);
+	/*
+	 * We rely on the wmb() in list_add_rcu to order the writes
+	 * to func, data, and cpumask before this write to refs.
+	 */
+	atomic_set(&data->refs, cpumask_weight(data->cpumask));
 	raw_spin_unlock_irqrestore(&call_function.lock, flags);
  
 	/*
...	...	@@ -491,15 +491,16 @@
491	491	cpumask_clear_cpu(this_cpu, data->cpumask);
492	492
493	493	/*
494		- * To ensure the interrupt handler gets an complete view
495		- * we order the cpumask and refs writes and order the read
496		- * of them in the interrupt handler. In addition we may
497		- * only clear our own cpu bit from the mask.
	494	+ * We reuse the call function data without waiting for any grace
	495	+ * period after some other cpu removes it from the global queue.
	496	+ * This means a cpu might find our data block as it is writen.
	497	+ * The interrupt handler waits until it sees refs filled out
	498	+ * while its cpu mask bit is set; here we may only clear our
	499	+ * own cpu mask bit, and must wait to set refs until we are sure
	500	+ * previous writes are complete and we have obtained the lock to
	501	+ * add the element to the queue.
498	502	*/
499		- smp_wmb();
500	503
501		- atomic_set(&data->refs, cpumask_weight(data->cpumask));
502		-
503	504	raw_spin_lock_irqsave(&call_function.lock, flags);
504	505	/*
505	506	* Place entry at the _HEAD_ of the list, so that any cpu still
...	...	@@ -507,6 +508,11 @@
507	508	* will not miss any other list entries:
508	509	*/
509	510	list_add_rcu(&data->csd.list, &call_function.queue);
	511	+ /*
	512	+ * We rely on the wmb() in list_add_rcu to order the writes
	513	+ * to func, data, and cpumask before this write to refs.
	514	+ */
	515	+ atomic_set(&data->refs, cpumask_weight(data->cpumask));
510	516	raw_spin_unlock_irqrestore(&call_function.lock, flags);
511	517
512	518	/*