Thread (21 messages) 21 messages, 3 authors, 2025-02-25

[PATCHv3 7/7] block: protect read_ahead_kb using q->limits_lock

From: Nilay Shroff <hidden>
Date: 2025-02-24 13:33:31
Subsystem: block layer, the rest · Maintainers: Jens Axboe, Linus Torvalds

The bdi->ra_pages could be updated under q->limits_lock because it's
usually calculated from the queue limits by queue_limits_commit_update.
So protect reading/writing the sysfs attribute read_ahead_kb using
q->limits_lock instead of q->sysfs_lock.

Signed-off-by: Nilay Shroff <redacted>
---
 block/blk-sysfs.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 8f47d9f30fbf..228f81a9060f 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -93,9 +93,9 @@ static ssize_t queue_ra_show(struct gendisk *disk, char *page)
 {
 	ssize_t ret;
 
-	mutex_lock(&disk->queue->sysfs_lock);
+	mutex_lock(&disk->queue->limits_lock);
 	ret = queue_var_show(disk->bdi->ra_pages << (PAGE_SHIFT - 10), page);
-	mutex_unlock(&disk->queue->sysfs_lock);
+	mutex_unlock(&disk->queue->limits_lock);
 
 	return ret;
 }
@@ -111,12 +111,15 @@ queue_ra_store(struct gendisk *disk, const char *page, size_t count)
 	ret = queue_var_store(&ra_kb, page, count);
 	if (ret < 0)
 		return ret;
-
-	mutex_lock(&q->sysfs_lock);
+	/*
+	 * ->ra_pages is protected by ->limits_lock because it is usually
+	 * calculated from the queue limits by queue_limits_commit_update.
+	 */
+	mutex_lock(&q->limits_lock);
 	memflags = blk_mq_freeze_queue(q);
 	disk->bdi->ra_pages = ra_kb >> (PAGE_SHIFT - 10);
+	mutex_unlock(&q->limits_lock);
 	blk_mq_unfreeze_queue(q, memflags);
-	mutex_unlock(&q->sysfs_lock);
 
 	return ret;
 }
@@ -670,7 +673,8 @@ static struct attribute *queue_attrs[] = {
 	&queue_dma_alignment_entry.attr,
 
 	/*
-	 * attributes protected with q->sysfs_lock
+	 * attributes which require some form of locking
+	 * other than q->sysfs_lock
 	 */
 	&queue_ra_entry.attr,
 
-- 
2.47.1
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help