[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1253803236-20760-11-git-send-email-jens.axboe@oracle.com>
Date: Thu, 24 Sep 2009 16:40:36 +0200
From: Jens Axboe <jens.axboe@...cle.com>
To: linux-kernel@...r.kernel.org
Cc: fengguang.wu@...el.com, shaohua.li@...el.com,
chris.mason@...cle.com, jack@...e.cz,
Jens Axboe <jens.axboe@...cle.com>
Subject: [PATCH 10/10] writeback: make the super_block pinning more efficient
Currently we pin the inode->i_sb for every single inode. This
increases cache traffic on sb->s_umount sem. Lets instead
cache the inode sb pin state and keep the super_block pinned
for as long as keep writing out inodes from the same
super_block.
Signed-off-by: Jens Axboe <jens.axboe@...cle.com>
---
fs/fs-writeback.c | 46 +++++++++++++++++++++++++++++-----------------
1 files changed, 29 insertions(+), 17 deletions(-)
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 225c731..c6bf775 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -540,6 +540,17 @@ writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
return ret;
}
+static void unpin_sb_for_writeback(struct super_block **psb)
+{
+ struct super_block *sb = *psb;
+
+ if (sb) {
+ up_read(&sb->s_umount);
+ put_super(sb);
+ *psb = NULL;
+ }
+}
+
/*
* For WB_SYNC_NONE writeback, the caller does not have the sb pinned
* before calling writeback. So make sure that we do pin it, so it doesn't
@@ -549,11 +560,20 @@ writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
* 1 if we failed.
*/
static int pin_sb_for_writeback(struct writeback_control *wbc,
- struct inode *inode)
+ struct inode *inode, struct super_block **psb)
{
struct super_block *sb = inode->i_sb;
/*
+ * If this sb is already pinned, nothing more to do. If not and
+ * *psb is non-NULL, unpin the old one first
+ */
+ if (sb == *psb)
+ return 0;
+ else if (*psb)
+ unpin_sb_for_writeback(psb);
+
+ /*
* Caller must already hold the ref for this
*/
if (wbc->sync_mode == WB_SYNC_ALL) {
@@ -566,7 +586,7 @@ static int pin_sb_for_writeback(struct writeback_control *wbc,
if (down_read_trylock(&sb->s_umount)) {
if (sb->s_root) {
spin_unlock(&sb_lock);
- return 0;
+ goto pinned;
}
/*
* umounted, drop rwsem again and fall through to failure
@@ -577,24 +597,15 @@ static int pin_sb_for_writeback(struct writeback_control *wbc,
sb->s_count--;
spin_unlock(&sb_lock);
return 1;
-}
-
-static void unpin_sb_for_writeback(struct writeback_control *wbc,
- struct inode *inode)
-{
- struct super_block *sb = inode->i_sb;
-
- if (wbc->sync_mode == WB_SYNC_ALL)
- return;
-
- up_read(&sb->s_umount);
- put_super(sb);
+pinned:
+ *psb = sb;
+ return 0;
}
static void writeback_inodes_wb(struct bdi_writeback *wb,
struct writeback_control *wbc)
{
- struct super_block *sb = wbc->sb;
+ struct super_block *sb = wbc->sb, *pin_sb = NULL;
const int is_blkdev_sb = sb_is_blkdev_sb(sb);
const unsigned long start = jiffies; /* livelock avoidance */
@@ -653,7 +664,7 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
if (inode_dirtied_after(inode, start))
break;
- if (pin_sb_for_writeback(wbc, inode)) {
+ if (pin_sb_for_writeback(wbc, inode, &pin_sb)) {
requeue_io(inode);
continue;
}
@@ -662,7 +673,6 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
__iget(inode);
pages_skipped = wbc->pages_skipped;
writeback_single_inode(inode, wbc);
- unpin_sb_for_writeback(wbc, inode);
if (wbc->pages_skipped != pages_skipped) {
/*
* writeback is not making progress due to locked
@@ -682,6 +692,8 @@ static void writeback_inodes_wb(struct bdi_writeback *wb,
wbc->more_io = 1;
}
+ unpin_sb_for_writeback(&pin_sb);
+
spin_unlock(&inode_lock);
/* Leave any unwritten inodes on b_io */
}
--
1.6.4.1.207.g68ea
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists