Skip to contentSkip to conversation

O_DIRECT: fix the splitting up of contiguous I/O

PATCH In review Jeff Moyer 4 participants 5 messages Aug 12, 2010 – Sep 6, 2010

Subsystems (2)
lore
Share & tools
lore

Jeff Moyer

author
#0 lore raw

PATCH 1/1 O_DIRECT: fix the splitting up of contiguous I/O

Hi,

commit c2c6ca4 (direct-io: do not merge logically non-contiguous requests) introduced a bug whereby all O_DIRECT I/Os were submitted a page at a time to the block layer. The problem is that the code expected dio->block_in_file to correspond to the current page in the dio. In fact, it corresponds to the previous page submitted via submit_page_section. This was purely an oversight, as the dio->cur_page_fs_offset field was introduced for just this purpose. This patch simply uses the correct variable when calculating whether there is a mismatch between contiguous logical blocks and contiguous physical blocks (as described in the comments).

I also switched the if conditional following this check to an else if, to ensure that we never call dio_bio_submit twice for the same dio (in theory, this should not happen, anyway).

I've tested this by running blktrace and verifying that a 64KB I/O was submitted as a single I/O. I also ran the patched kernel through xfstests' aio tests using xfs, ext4 (with 1k and 4k block sizes) and btrfs and verified that there were no regressions as compared to an unpatched kernel.

Comments, as always, are welcome.

Cheers,
Jeff

fs/direct-io.c +2 -2
index 7600aac..445901c 100644
@@ -632,7 +632,7 @@ static int dio_send_cur_page(struct dio *dio)
 	int ret = 0;
 
 	if (dio->bio) {
-		loff_t cur_offset = dio->block_in_file << dio->blkbits;
+		loff_t cur_offset = dio->cur_page_fs_offset;
 		loff_t bio_next_offset = dio->logical_offset_in_bio +
 			dio->bio->bi_size;
 
@@ -657,7 +657,7 @@ static int dio_send_cur_page(struct dio *dio)
 		 * Submit now if the underlying fs is about to perform a
 		 * metadata read
 		 */
-		if (dio->boundary)
+		else if (dio->boundary)
 			dio_bio_submit(dio);
 	}
 
#1 lore raw

Andrew, can you please send this on to Linus and -stable ASAP? It's causing massive problems for our users.

Quoted 55 lines · Jeff Moyer

Hi,

commit c2c6ca4 (direct-io: do not merge logically non-contiguous requests) introduced a bug whereby all O_DIRECT I/Os were submitted a page at a time to the block layer. The problem is that the code expected dio->block_in_file to correspond to the current page in the dio. In fact, it corresponds to the previous page submitted via submit_page_section. This was purely an oversight, as the dio->cur_page_fs_offset field was introduced for just this purpose. This patch simply uses the correct variable when calculating whether there is a mismatch between contiguous logical blocks and contiguous physical blocks (as described in the comments).

I also switched the if conditional following this check to an else if, to ensure that we never call dio_bio_submit twice for the same dio (in theory, this should not happen, anyway).

I've tested this by running blktrace and verifying that a 64KB I/O was submitted as a single I/O. I also ran the patched kernel through xfstests' aio tests using xfs, ext4 (with 1k and 4k block sizes) and btrfs and verified that there were no regressions as compared to an unpatched kernel.

Comments, as always, are welcome.

Cheers,
Jeff

fs/direct-io.c +2 -3
index 7600aac..445901c 100644
@@ -632,7 +632,7 @@ static int dio_send_cur_page(struct dio *dio)
 	int ret = 0;
 
 	if (dio->bio) {
-		loff_t cur_offset = dio->block_in_file << dio->blkbits;
+		loff_t cur_offset = dio->cur_page_fs_offset;
 		loff_t bio_next_offset = dio->logical_offset_in_bio +
 			dio->bio->bi_size;
 
@@ -657,7 +657,7 @@ static int dio_send_cur_page(struct dio *dio)
 		 * Submit now if the underlying fs is about to perform a
 		 * metadata read
 		 */
-		if (dio->boundary)
+		else if (dio->boundary)
 			dio_bio_submit(dio);
 	}
 
 
--
Quoted 4 lines

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

---end quoted text---

#2 lore raw
Quoted 2 lines · Christoph Hellwig

Andrew, can you please send this on to Linus and -stable ASAP? It's causing massive problems for our users.

Did this patch get dropped ? I don't see it in Linus's git..

https://patchwork.kernel.org/patch/119333/

cheers,
Chris

Signature · 4 lines
 Chris Samuel  :  http://www.csamuel.org/  :  Melbourne, VIC

This email may come with a PGP signature as a file. Do not panic.
For more info see: http://en.wikipedia.org/wiki/OpenPGP
#3 lore raw
Quoted 6 lines · On Sun, 5 Sep 2010 22:56:08 +1000 Chris Samuel <[email protected]> wrote:
Quoted 2 lines · Christoph Hellwig

Andrew, can you please send this on to Linus and -stable ASAP? It's causing massive problems for our users.

Did this patch get dropped ?

Nope. I have it queued for 2.6.36 and -stable, will send it in soon. Everything is a bit lagged at present due to conferences, summer, encroaching senility, etc.

#4 lore raw
Quoted 7 lines · Andrew Morton
Quoted 1 line · On Sun, 5 Sep 2010 22:56:08 +1000 Chris Samuel <[email protected]> wrote:
Quoted 1 line

Did this patch get dropped ?

Nope. I have it queued for 2.6.36 and -stable, will send it in soon. Everything is a bit lagged at present due to conferences, summer, encroaching senility, etc.

Mea culpa - I mixed up the date of the original email with the date of Christoph's email and so thought he'd sent that back in August. Looks like I'm the one with senility issues.. ;-)

cheers,
Chris

Signature · 1 line
 Chris Samuel  :  http://www.csamuel.org/  :  Melbourne, VIC