sbitmap: use test_and_set_bit_lock()/clear_bit_unlock()

sbitmap_queue_get()/sbitmap_queue_clear() are used for
allocating/freeing a resource, so they should provide acquire/release
barrier semantics, respectively. sbitmap_get() currently contains a full
barrier, which is unnecessary, so use test_and_set_bit_lock() instead of
test_and_set_bit() (these are equivalent on x86_64). sbitmap_clear_bit()
does not imply any barriers, which is incorrect, as accesses of the
resource (e.g., request) could potentially get reordered to after the
clear_bit(). Introduce sbitmap_clear_bit_unlock() and use it for
sbitmap_queue_clear() (this only adds a compiler barrier on x86_64). The
other existing user of sbitmap_clear_bit() (the blk-mq software queue
pending map) is serialized through a spinlock and does not need this.
Reported-by: Tejun Heo <tj@kernel.org>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

include/linux/sbitmap.h

@@ -171,6 +171,8 @@ void sbitmap_resize(struct sbitmap *sb, unsigned int depth);
  *               starting from the last allocated bit. This is less efficient
  *               than the default behavior (false).
  *
+ * This operation provides acquire barrier semantics if it succeeds.
+ *
  * Return: Non-negative allocated bit number if successful, -1 otherwise.
  */
 int sbitmap_get(struct sbitmap *sb, unsigned int alloc_hint, bool round_robin);
@@ -300,6 +302,12 @@ static inline void sbitmap_clear_bit(struct sbitmap *sb, unsigned int bitnr)
 	clear_bit(SB_NR_TO_BIT(sb, bitnr), __sbitmap_word(sb, bitnr));
 }
 
+static inline void sbitmap_clear_bit_unlock(struct sbitmap *sb,
+					    unsigned int bitnr)
+{
+	clear_bit_unlock(SB_NR_TO_BIT(sb, bitnr), __sbitmap_word(sb, bitnr));
+}
+
 static inline int sbitmap_test_bit(struct sbitmap *sb, unsigned int bitnr)
 {
 	return test_bit(SB_NR_TO_BIT(sb, bitnr), __sbitmap_word(sb, bitnr));

lib/sbitmap.c

@@ -100,7 +100,7 @@ static int __sbitmap_get_word(unsigned long *word, unsigned long depth,
 			return -1;
 		}
 
-		if (!test_and_set_bit(nr, word))
+		if (!test_and_set_bit_lock(nr, word))
 			break;
 
 		hint = nr + 1;
@@ -434,9 +434,9 @@ static void sbq_wake_up(struct sbitmap_queue *sbq)
 	/*
 	 * Pairs with the memory barrier in set_current_state() to ensure the
 	 * proper ordering of clear_bit()/waitqueue_active() in the waker and
-	 * test_and_set_bit()/prepare_to_wait()/finish_wait() in the waiter. See
-	 * the comment on waitqueue_active(). This is __after_atomic because we
-	 * just did clear_bit() in the caller.
+	 * test_and_set_bit_lock()/prepare_to_wait()/finish_wait() in the
+	 * waiter. See the comment on waitqueue_active(). This is __after_atomic
+	 * because we just did clear_bit_unlock() in the caller.
 	 */
 	smp_mb__after_atomic();
 
@@ -469,7 +469,7 @@ static void sbq_wake_up(struct sbitmap_queue *sbq)
 void sbitmap_queue_clear(struct sbitmap_queue *sbq, unsigned int nr,
 			 unsigned int cpu)
 {
-	sbitmap_clear_bit(&sbq->sb, nr);
+	sbitmap_clear_bit_unlock(&sbq->sb, nr);
 	sbq_wake_up(sbq);
 	if (likely(!sbq->round_robin && nr < sbq->sb.depth))
 		*per_cpu_ptr(sbq->alloc_hint, cpu) = nr;