public inbox for linux-block@vger.kernel.org
 help / color / mirror / Atom feed
* [PATCH] loop: fix the the direct I/O support check when used on top of block devices
@ 2024-01-17 17:59 Christoph Hellwig
  2024-01-18  8:21 ` Ming Lei
  2024-01-18 15:21 ` Jens Axboe
  0 siblings, 2 replies; 3+ messages in thread
From: Christoph Hellwig @ 2024-01-17 17:59 UTC (permalink / raw)
  To: axboe; +Cc: ming.lei, linux-block

__loop_update_dio only checks the alignment requirement for block backed
file systems, but misses them for the case where the loop device is
created directly on top of another block device.  Due to this creating
a loop device with default option plus the direct I/O flag on a > 512 byte
sector size file system will lead to incorrect I/O being submitted to the
lower block device and a lot of error from the lock layer.  This can
be seen with xfstests generic/563.

Fix the code in __loop_update_dio by factoring the alignment check into
a helper, and calling that also for the struct block_device of a block
device inode.

Also remove the TODO comment talking about dynamically switching between
buffered and direct I/O, which is a would be a recipe for horrible
performance and occasional data loss.

Fixes: 2e5ab5f379f9 ("block: loop: prepare for supporing direct IO")
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 drivers/block/loop.c | 52 +++++++++++++++++++++-----------------------
 1 file changed, 25 insertions(+), 27 deletions(-)

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 2bd10f3bfcb296..01bb9436240484 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -165,39 +165,37 @@ static loff_t get_loop_size(struct loop_device *lo, struct file *file)
 	return get_size(lo->lo_offset, lo->lo_sizelimit, file);
 }
 
+/*
+ * We support direct I/O only if lo_offset is aligned with the logical I/O size
+ * of backing device, and the logical block size of loop is bigger than that of
+ * the backing device.
+ */
+static bool lo_bdev_can_use_dio(struct loop_device *lo,
+		struct block_device *backing_bdev)
+{
+	unsigned short sb_bsize = bdev_logical_block_size(backing_bdev);
+
+	if (queue_logical_block_size(lo->lo_queue) < sb_bsize)
+		return false;
+	if (lo->lo_offset & (sb_bsize - 1))
+		return false;
+	return true;
+}
+
 static void __loop_update_dio(struct loop_device *lo, bool dio)
 {
 	struct file *file = lo->lo_backing_file;
-	struct address_space *mapping = file->f_mapping;
-	struct inode *inode = mapping->host;
-	unsigned short sb_bsize = 0;
-	unsigned dio_align = 0;
+	struct inode *inode = file->f_mapping->host;
+	struct block_device *backing_bdev = NULL;
 	bool use_dio;
 
-	if (inode->i_sb->s_bdev) {
-		sb_bsize = bdev_logical_block_size(inode->i_sb->s_bdev);
-		dio_align = sb_bsize - 1;
-	}
+	if (S_ISBLK(inode->i_mode))
+		backing_bdev = I_BDEV(inode);
+	else if (inode->i_sb->s_bdev)
+		backing_bdev = inode->i_sb->s_bdev;
 
-	/*
-	 * We support direct I/O only if lo_offset is aligned with the
-	 * logical I/O size of backing device, and the logical block
-	 * size of loop is bigger than the backing device's.
-	 *
-	 * TODO: the above condition may be loosed in the future, and
-	 * direct I/O may be switched runtime at that time because most
-	 * of requests in sane applications should be PAGE_SIZE aligned
-	 */
-	if (dio) {
-		if (queue_logical_block_size(lo->lo_queue) >= sb_bsize &&
-		    !(lo->lo_offset & dio_align) &&
-		    (file->f_mode & FMODE_CAN_ODIRECT))
-			use_dio = true;
-		else
-			use_dio = false;
-	} else {
-		use_dio = false;
-	}
+	use_dio = dio && (file->f_mode & FMODE_CAN_ODIRECT) &&
+		(!backing_bdev || lo_bdev_can_use_dio(lo, backing_bdev));
 
 	if (lo->use_dio == use_dio)
 		return;
-- 
2.39.2


^ permalink raw reply related	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2024-01-18 15:21 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-01-17 17:59 [PATCH] loop: fix the the direct I/O support check when used on top of block devices Christoph Hellwig
2024-01-18  8:21 ` Ming Lei
2024-01-18 15:21 ` Jens Axboe

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox