public inbox for linux-xfs@vger.kernel.org
 help / color / mirror / Atom feed
From: Dave Chinner <david@fromorbit.com>
To: xfs@oss.sgi.com
Subject: [PATCH 14/37] xfs: Use preallocation for inodes with extsz hints
Date: Mon, 23 Apr 2012 15:58:44 +1000	[thread overview]
Message-ID: <1335160747-17254-15-git-send-email-david@fromorbit.com> (raw)
In-Reply-To: <1335160747-17254-1-git-send-email-david@fromorbit.com>

From: Dave Chinner <dchinner@redhat.com>

xfstest 229 exposes a problem with buffered IO, delayed allocation
and extent size hints. That is when we do delayed allocation during
buffered IO, we reserve space for the extent size hint alignment and
allocate the physical space to align the extent, but we do not zero
the regions of the extent that aren't written by the write(2)
syscall. The result is that we expose stale data in unwritten
regions of the extent size hints.

There are two ways to fix this. The first is to detect that we are
doing unaligned writes, check if there is already a mapping or data
over the extent size hint range, and if not zero the page cache
first before then doing the real write. This can be very expensive
for large extent size hints, especially if the subsequent writes
fill then entire extent size before the data is written to disk.

The second, and simpler way, is simply to turn off delayed
allocation when the extent size hint is set and use preallocation
instead. This results in unwritten extents being laid down on disk
and so only the written portions will be converted. This matches the
behaviour for direct IO, and will also work for the real time
device. The disadvantage of this approach is that for small extent
size hints we can get file fragmentation, but in general extent size
hints are fairly large (e.g. stripe width sized) so this isn't a big
deal.

Implement the second approach as it is simple and effective.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
---
 fs/xfs/xfs_aops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 4588a7c..eff2ea8 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -1175,7 +1175,7 @@ __xfs_get_blocks(
 	    (!nimaps ||
 	     (imap.br_startblock == HOLESTARTBLOCK ||
 	      imap.br_startblock == DELAYSTARTBLOCK))) {
-		if (direct) {
+		if (direct || xfs_get_extsz_hint(ip)) {
 			/*
 			 * Drop the ilock in preparation for starting the block
 			 * allocation transaction.  It will be retaken
-- 
1.7.9.5

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

  parent reply	other threads:[~2012-04-23  5:59 UTC|newest]

Thread overview: 96+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-04-23  5:58 [PATCH 00/37] xfs: current 3.4 patch queue Dave Chinner
2012-04-23  5:58 ` [PATCH 01/37] xfs: remove log item from AIL in xfs_qm_dqflush after a shutdown Dave Chinner
2012-04-23  5:58 ` [PATCH 02/37] xfs: remove log item from AIL in xfs_iflush " Dave Chinner
2012-04-23 15:39   ` Mark Tinguely
2012-04-23  5:58 ` [PATCH 03/37] xfs: allow assigning the tail lsn with the AIL lock held Dave Chinner
2012-04-23  5:58 ` [PATCH 04/37] xfs: implement freezing by emptying the AIL Dave Chinner
2012-04-23 15:40   ` Mark Tinguely
2012-04-29 21:43   ` Christoph Hellwig
2012-04-23  5:58 ` [PATCH 05/37] xfs: don't flush inodes from background inode reclaim Dave Chinner
2012-04-23  5:58 ` [PATCH 06/37] xfs: do not write the buffer from xfs_iflush Dave Chinner
2012-04-23  5:58 ` [PATCH 07/37] xfs: do not write the buffer from xfs_qm_dqflush Dave Chinner
2012-04-23  5:58 ` [PATCH 08/37] xfs: do not add buffers to the delwri queue until pushed Dave Chinner
2012-04-23  5:58 ` [PATCH 09/37] xfs: on-stack delayed write buffer lists Dave Chinner
2012-04-25 18:34   ` Mark Tinguely
2012-04-29 21:44   ` Christoph Hellwig
2012-04-23  5:58 ` [PATCH 10/37] xfs: remove some obsolete comments in xfs_trans_ail.c Dave Chinner
2012-04-23 15:41   ` Mark Tinguely
2012-04-23  5:58 ` [PATCH 11/37] xfs: pass shutdown method into xfs_trans_ail_delete_bulk Dave Chinner
2012-04-23  5:58 ` [PATCH 12/37] xfs: Do background CIL flushes via a workqueue Dave Chinner
2012-04-23  7:54   ` [PATCH 12/37 V2] " Dave Chinner
2012-04-29 21:46     ` Christoph Hellwig
2012-04-23  5:58 ` [PATCH 13/37] xfs: page type check in writeback only checks last buffer Dave Chinner
2012-04-23  5:58 ` Dave Chinner [this message]
2012-04-29 21:47   ` [PATCH 14/37] xfs: Use preallocation for inodes with extsz hints Christoph Hellwig
2012-04-23  5:58 ` [PATCH 15/37] xfs: fix buffer lookup race on allocation failure Dave Chinner
2012-04-23  5:58 ` [PATCH 16/37] xfs: check for buffer errors before waiting Dave Chinner
2012-04-23  5:58 ` [PATCH 17/37] xfs: fix incorrect b_offset initialisation Dave Chinner
2012-04-23  5:58 ` [PATCH 18/37] xfs: use kmem_zone_zalloc for buffers Dave Chinner
2012-04-23  5:58 ` [PATCH 19/37] xfs: clean up buffer get/read call API Dave Chinner
2012-04-23  5:58 ` [PATCH 20/37] xfs: kill b_file_offset Dave Chinner
2012-04-23  5:58 ` [PATCH 21/37] xfs: use blocks for counting length of buffers Dave Chinner
2012-04-23  5:58 ` [PATCH 22/37] xfs: use blocks for storing the desired IO size Dave Chinner
2012-04-23  5:58 ` [PATCH 23/37] xfs: kill xfs_buf_btoc Dave Chinner
2012-04-23  5:58 ` [PATCH 24/37] xfs: kill XBF_LOCK Dave Chinner
2012-04-23  5:58 ` [PATCH 25/37] xfs: kill xfs_read_buf() Dave Chinner
2012-04-23  5:58 ` [PATCH 26/37] xfs: kill XBF_DONTBLOCK Dave Chinner
2012-04-23  5:58 ` [PATCH 27/37] xfs: use iolock on XFS_IOC_ALLOCSP calls Dave Chinner
2012-04-23  5:58 ` [PATCH 28/37] xfs: move xfsagino_t to xfs_types.h Dave Chinner
2012-04-23 15:43   ` Mark Tinguely
2012-04-24 15:10   ` Mark Tinguely
2012-04-29 21:49   ` Christoph Hellwig
2012-04-30  0:32     ` Dave Chinner
2012-04-23  5:58 ` [PATCH 29/37] xfs: move busy extent handling to it's own file Dave Chinner
2012-04-23 17:57   ` Ben Myers
2012-04-24  0:25     ` [PATCH 29/37 V2] " Dave Chinner
2012-04-24 15:56       ` Mark Tinguely
2012-04-24 18:10         ` Mark Tinguely
2012-04-29 10:39           ` [PATCH 29/37 V3] " Dave Chinner
2012-04-29 21:50             ` Christoph Hellwig
2012-04-30  0:36               ` Dave Chinner
2012-04-30  2:17                 ` Dave Chinner
2012-04-23  5:59 ` [PATCH 30/37] xfs: clean up busy extent naming Dave Chinner
2012-04-24 18:11   ` Mark Tinguely
2012-04-29 10:41     ` [PATCH 30/37 V2] " Dave Chinner
2012-04-29 21:50       ` Christoph Hellwig
2012-04-23  5:59 ` [PATCH 31/37] xfs: move xfs_fsb_to_db to xfs_bmap.h Dave Chinner
2012-04-24 19:24   ` Mark Tinguely
2012-04-29 21:53   ` Christoph Hellwig
2012-04-30  2:31     ` Dave Chinner
2012-04-23  5:59 ` [PATCH 32/37] xfs: move xfs_get_extsz_hint() and kill xfs_rw.h Dave Chinner
2012-04-24 19:30   ` Mark Tinguely
2012-04-29 21:53   ` Christoph Hellwig
2012-04-23  5:59 ` [PATCH 33/37] xfs: move xfs_do_force_shutdown() and kill xfs_rw.c Dave Chinner
2012-04-24 19:37   ` Mark Tinguely
2012-04-29 21:54   ` Christoph Hellwig
2012-04-30  2:38     ` Dave Chinner
2012-04-23  5:59 ` [PATCH 34/37] xfs: clean up xfs_bit.h includes Dave Chinner
2012-04-24 19:44   ` Mark Tinguely
2012-04-29 21:55   ` Christoph Hellwig
2012-04-30  2:40     ` Dave Chinner
2012-04-23  5:59 ` [PATCH 35/37] xfs: Properly exclude IO type flags from buffer flags Dave Chinner
2012-04-24 20:02   ` Mark Tinguely
2012-04-29 21:55   ` Christoph Hellwig
2012-04-23  5:59 ` [PATCH 36/37] xfs: flush outstanding buffers on log mount failure Dave Chinner
2012-04-23 15:47   ` Mark Tinguely
2012-04-29 21:55   ` Christoph Hellwig
2012-04-23  5:59 ` [PATCH 37/37] xfs: make XBF_MAPPED the default behaviour Dave Chinner
2012-04-25 18:35   ` Mark Tinguely
2012-04-25 20:09   ` Mark Tinguely
2012-04-25 22:33     ` Dave Chinner
2012-04-29 21:57   ` Christoph Hellwig
2012-04-30  2:45     ` Dave Chinner
2012-04-23 18:01 ` [PATCH 00/37] xfs: current 3.4 patch queue Ben Myers
2012-04-23 23:29   ` Dave Chinner
2012-04-30 14:24     ` Ben Myers
2012-04-28  2:15 ` Ben Myers
2012-04-28 21:28   ` Ben Myers
2012-04-29  0:21     ` Dave Chinner
2012-04-29  0:14   ` Dave Chinner
2012-04-30 14:44     ` Ben Myers
2012-04-30 23:04       ` Dave Chinner
2012-04-30 14:32 ` Assertion failed: RB_EMPTY_NODE(&bp->b_rbnode) Ben Myers
2012-04-30 23:12   ` Dave Chinner
2012-04-30 14:34 ` [PATCH 00/37] xfs: current 3.4 patch queue Ben Myers
2012-04-30 23:20   ` Dave Chinner
2012-04-30 19:25 ` Christoph Hellwig

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1335160747-17254-15-git-send-email-david@fromorbit.com \
    --to=david@fromorbit.com \
    --cc=xfs@oss.sgi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox