Extremely slow remounts with concurrent I/O

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

Quoted 2 lines · Lucas Nussbaum

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

Hi Lukas,

Thanks for this report. Are you willing to try a kernel patch? If so, could you try and see if this fixes your issue. From looking at your block trace, I saw a large number of suspicious 4k writes from the jbd2 layer.

Hmm... OK, let me make sure I understand what is going on. So you have a single file system which is mounted read/write,

and you are
doing a huge number of copies into the file system, which is keeping it busy.

You are then running a huge number of "mount -o remount" on that same file system,

which should effectively be no-op's, since the
remount isn't actually change the read/only or read/write or any other mount options. Is that right?

Why were you doing the remount in in your actual production workload, anyway?

Quoted 10 lines · Theodore Ts'o

Quoted 2 lines · Lucas Nussbaum

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

Hi Lukas,

Thanks for this report. Are you willing to try a kernel patch? If so, could you try and see if this fixes your issue. From looking at your block trace, I saw a large number of suspicious 4k writes from the jbd2 layer.

Hi Ted,

This patch doesn't solve the problem. It seems that on average, it makes the situation worse, but I'm not sure if this is just anecdotical evidence or statistically valid.

Anything else I could try to gather more information?

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

When starting hundreds of LXC containers simultaneously on a system, the boot of some containers was hanging. We tracked this down to an initscript's use of mount -o remount, which was hanging in D state.

We reproduced the problem outside of LXC, with the script available at [0]. That script initiates 1000 mount -o remount, and performs some writes using a big cp to the same filesystem during the remounts....

Quoted 10 lines · Lucas Nussbaum

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

When starting hundreds of LXC containers simultaneously on a system, the boot of some containers was hanging. We tracked this down to an initscript's use of mount -o remount, which was hanging in D state.

We reproduced the problem outside of LXC, with the script available at [0]. That script initiates 1000 mount -o remount, and performs some writes using a big cp to the same filesystem during the remounts....

+linux-fsdevel since the patch modifies fs/super.c

Lukas, can you try this patch? I'm pretty sure this is what's going on. It turns out each "mount -o remount" is implying an fsync(), so your test case is identical to copying a large file while having thousand of processes calling syncfs() on the file system, with the predictable results.

Quoted 5 lines

Lukas, can you try this patch? I'm pretty sure this is what's going on. It turns out each "mount -o remount" is implying an fsync(), so your test case is identical to copying a large file while having thousand of processes calling syncfs() on the file system, with the predictable results.

Hi Ted,

I can confirm that:
1) the patch solves my problem
2) issuing 'sync' instead of 'mount -o remount' indeed exhibits the problem again

However, I'm curious: why would such a workload (multiple syncfs() initiated during a write) block for several minutes on an ext4 filesystem? I've just tried again on ext3, and it's not a problem in that case.

Quoted 10 lines · Lucas Nussbaum

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

When starting hundreds of LXC containers simultaneously on a system, the boot of some containers was hanging. We tracked this down to an initscript's use of mount -o remount, which was hanging in D state.

We reproduced the problem outside of LXC, with the script available at [0]. That script initiates 1000 mount -o remount, and performs some writes using a big cp to the same filesystem during the remounts....

+linux-fsdevel since the patch modifies fs/super.c

Lukas, can you try this patch? I'm pretty sure this is what's going on. It turns out each "mount -o remount" is implying an fsync(), so your test case is identical to copying a large file while having thousand of processes calling syncfs() on the file system, with the predictable results.

Folks on linux-fsdevel, any objections if I carry this patch in the ext4 tree? I don't think it should cause problems for other file systems, since any file system that tries to rely on the implied syncfs() is going to be subject to races, but it might make such a race condition bug much more visible...

- Ted

commit 8862c3c69acc205b59b00baed67e50446e2fd093

Author: Theodore Ts'o <[email protected]>
Date:   Sat Mar 8 11:05:35 2014 -0500

fs: only call sync_filesystem() when remounting read-only

Currently "mount -o remount" always implies an syncfs() on the file system. This can cause a problem if a workload calls "mount -o remount" many, many times while concurrent I/O is happening:

http://article.gmane.org/gmane.comp.file-systems.ext4/42876

Whether it would ever be sane for a workload to call "mount -o remount" gazillions of times when they are effectively no-ops, it seems stupid for a remount to imply an fsync().

It's possible that there is some file system which is relying on the implied fsync(), but that's arguably broken, since aside for the remount read-only case, there's nothing that will prevent other writes from sneaking in between the sync_filesystem() and the call to sb->s_op->remount_fs().

Signed-off-by "Theodore Ts'o" <[email protected]>

fs/super.c +1 -2

index 80d5cf2..0fc87ac 100644
@@ -717,10 +717,9 @@ int do_remount_sb(struct super_block *sb, int flags, void *data, int force)
 			if (retval)
 				return retval;
 		}
+		sync_filesystem(sb);
 	}
 
-	sync_filesystem(sb);
-
 	if (sb->s_op->remount_fs) {
 		retval = sb->s_op->remount_fs(sb, &flags, data);
 		if (retval) {

Quoted 10 lines · Lucas Nussbaum

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

When starting hundreds of LXC containers simultaneously on a system, the boot of some containers was hanging. We tracked this down to an initscript's use of mount -o remount, which was hanging in D state.

We reproduced the problem outside of LXC, with the script available at [0]. That script initiates 1000 mount -o remount, and performs some writes using a big cp to the same filesystem during the remounts....

+linux-fsdevel since the patch modifies fs/super.c

Lukas, can you try this patch? I'm pretty sure this is what's going on. It turns out each "mount -o remount" is implying an fsync(), so your test case is identical to copying a large file while having thousand of processes calling syncfs() on the file system, with the predictable results.

Folks on linux-fsdevel, any objections if I carry this patch in the ext4 tree? I don't think it should cause problems for other file systems, since any file system that tries to rely on the implied syncfs() is going to be subject to races, but it might make such a race condition bug much more visible...

Quoted 5 lines

Folks on linux-fsdevel, any objections if I carry this patch in the ext4 tree? I don't think it should cause problems for other file systems, since any file system that tries to rely on the implied syncfs() is going to be subject to races, but it might make such a race condition bug much more visible...

IMO, I think that you should be looking to fix ext4 syncfs issues, not changing the VFS behaviour that might cause subtle and unnoticed problems for other filesystems. We should not be moving data inegrity operations without first auditing of all the filesystem remount operations for issues.

Quoted 5 lines

IMO, I think that you should be looking to fix ext4 syncfs issues, not changing the VFS behaviour that might cause subtle and unnoticed problems for other filesystems. We should not be moving data inegrity operations without first auditing of all the filesystem remount operations for issues.

The issue is that it's forcing a CACHE FLUSH if we don't need to force a journal commit, since it's possible that data writes could have been sent to the disk without modifying fs metadata that would require a commit. So arguably what we're doing with ext4 is _correct_, where as with ext3 we would simply not calling blkdev_issue_barrier() in that situation.

Quoted 6 lines · Theodore Ts'o

Quoted 5 lines

IMO, I think that you should be looking to fix ext4 syncfs issues, not changing the VFS behaviour that might cause subtle and unnoticed problems for other filesystems. We should not be moving data inegrity operations without first auditing of all the filesystem remount operations for issues.

Quoted 6 lines

The issue is that it's forcing a CACHE FLUSH if we don't need to force a journal commit, since it's possible that data writes could have been sent to the disk without modifying fs metadata that would require a commit. So arguably what we're doing with ext4 is _correct_, where as with ext3 we would simply not calling blkdev_issue_barrier() in that situation.

Doing some more digging, ext4 is currently interpreting syncfs() as requiring a data integrity sync. So we go through some extra work to guarantee that we call blkdev_issue_barrier(), even if a journal commit is not required.

This change was made by Dmitry last June:

commit 06a407f13daf9e48f0ef7189c7e54082b53940c7

Author: Dmitry Monakhov <[email protected]>
Date:   Wed Jun 12 22:25:07 2013 -0400

ext4: fix data integrity for ext4_sync_fs

Inode's data or non journaled quota may be written w/o jounral so we _must_ send a barrier at the end of ext4_sync_fs. But it can be skipped if journal commit will do it for us.

Also fix data integrity for nojournal mode.

Signed-off-by Dmitry Monakhov <[email protected]>

Signed-off-by "Theodore Ts'o" <[email protected]>

Both ext3 and xfs do *not* do this.

Looking more closely at the syncfs(2) manpage, it's not clear it requires this:

sync() causes all buffered modifications to file metadata and data to be written to the underlying filesystems.

syncfs() is like sync(), but synchronizes just the filesystem containing file referred to by the open file descriptor fd.

Unlike the fsync(2) system call, it does *not* state that the data flushed to the disk is guaranteed to be there after a crash, which I suppose justifies ext3 and xfs's current behavior.

1) Nowhere in the remount system call is it stated that it has
***any*** data integrity implifications. If you are making the rw->ro
transition, sure, you'll need to flush out any pending changes. But there doesn't seem to be any justification for requiring this this if the
remount is a no-op. So I think changing the remount code path as I
suggested is a valid option.

2) We could revert Dmitry's change from last June. This would make ext4 work the same way as ext3 and xfs. Which I think is also valid, since the syncfs(2) system call says nothing about guaranteeing data being preserved after a crash, unlike fsync(2).

3) We could say that a workload that calls thousands of no-op remounts to be stupid/busted/silly, and not do anything at all.

#1 requires core VFS changes, and Dave seems unhappy with it.

XFS most definitely considered sync_filesystem(sb) to be a data integrity operation. xfs_fs_sync_fs() calls this:

	xfs_log_force(mp, XFS_LOG_SYNC);

Which will issue a blocking journal commit which will uses REQ_FLUSH|REQ_FUA for the journal writes. Hence if there was anything dirty in the filesystem that sync_filesystem wrote to disk, it will issue a cache flush just like ext4 does.

Quoted 6 lines

1) Nowhere in the remount system call is it stated that it has
***any*** data integrity implications. If you are making the rw->ro
transition, sure, you'll need to flush out any pending changes. But there doesn't seem to be any justification for requiring this this if the
remount is a no-op. So I think changing the remount code path as I
suggested is a valid option.

What the man page says doesn't change the fact we need to audit all the existing filesystems before such a change is made.

IMO, I think that you should be looking to fix ext4 syncfs issues, not changing the VFS behaviour that might cause subtle and unnoticed problems for other filesystems. We should not be moving data inegrity operations without first auditing of all the filesystem remount operations for issues.

Previously, the no-op "mount -o mount /dev/xxx" operation when the file system is already mounted read-write causes an implied, unconditional syncfs(). This seems pretty stupid, and it's certainly documented or guaraunteed to do this, nor is it particularly useful, except in the case where the file system was mounted rw and is getting remounted read-only.

However, it's possible that there might be some file systems that are actually depending on this behavior. In most file systems, it's probably fine to only call sync_filesystem() when transitioning from read-write to read-only, and there are some file systems where this is not needed at all (for example, for a pseudo-filesystem or something like romfs).

Previously, the no-op "mount -o mount /dev/xxx" operation when the

file system is already mounted read-write causes an implied, unconditional syncfs(). This seems pretty stupid, and it's certainly documented or guaraunteed to do this, nor is it particularly useful, except in the case where the file system was mounted rw and is getting remounted read-only.

However, it's possible that there might be some file systems that are actually depending on this behavior. In most file systems, it's probably fine to only call sync_filesystem() when transitioning from read-write to read-only, and there are some file systems where this is not needed at all (for example, for a pseudo-filesystem or something like romfs).

Signed-off-by "Theodore Ts'o" <[email protected]>

Cc linux-fsdevel at vger.kernel.org

Cc Christoph Hellwig <[email protected]>

Cc Artem Bityutskiy <[email protected]>

Cc Adrian Hunter <[email protected]>

Cc Evgeniy Dushistov <[email protected]>

Cc Jan Kara <[email protected]>

Cc OGAWA Hirofumi <[email protected]>

Cc Anders Larsen <[email protected]>

Cc Phillip Lougher <[email protected]>

Cc Kees Cook <[email protected]>

Cc Mikulas Patocka <[email protected]>

Cc Petr Vandrovec <[email protected]>

Cc xfs at oss.sgi.com

Cc linux-btrfs at vger.kernel.org

Cc linux-cifs at vger.kernel.org

Cc samba-technical at lists.samba.org

Cc codalist at coda.cs.cmu.edu

Cc linux-ext4 at vger.kernel.org

Cc linux-f2fs-devel at lists.sourceforge.net

Cc fuse-devel at lists.sourceforge.net

Cc cluster-devel at redhat.com

Cc linux-mtd at lists.infradead.org

Cc jfs-discussion at lists.sourceforge.net

Cc linux-nfs at vger.kernel.org

Cc linux-nilfs at vger.kernel.org

Cc linux-ntfs-dev at lists.sourceforge.net

Cc ocfs2-devel at oss.oracle.com

Cc reiserfs-devel at vger.kernel.org

---
 fs/adfs/super.c          | 1 +
 fs/affs/super.c          | 1 +
 fs/befs/linuxvfs.c       | 1 +
 fs/btrfs/super.c         | 1 +
 fs/cifs/cifsfs.c         | 1 +
 fs/coda/inode.c          | 1 +
 fs/cramfs/inode.c        | 1 +
 fs/debugfs/inode.c       | 1 +
 fs/devpts/inode.c        | 1 +
 fs/efs/super.c           | 1 +
 fs/ext2/super.c          | 1 +
 fs/ext3/super.c          | 2 ++
 fs/ext4/super.c          | 2 ++
 fs/f2fs/super.c          | 2 ++
 fs/fat/inode.c           | 2 ++
 fs/freevxfs/vxfs_super.c | 1 +
 fs/fuse/inode.c          | 1 +
 fs/gfs2/super.c          | 2 ++
 fs/hfs/super.c           | 1 +
 fs/hfsplus/super.c       | 1 +
 fs/hpfs/super.c          | 2 ++
 fs/isofs/inode.c         | 1 +
 fs/jffs2/super.c         | 1 +
 fs/jfs/super.c           | 1 +
 fs/minix/inode.c         | 1 +
 fs/ncpfs/inode.c         | 1 +
 fs/nfs/super.c           | 2 ++
 fs/nilfs2/super.c        | 1 +
 fs/ntfs/super.c          | 2 ++
 fs/ocfs2/super.c         | 2 ++
 fs/openpromfs/inode.c    | 1 +
 fs/proc/root.c           | 2 ++
 fs/pstore/inode.c        | 1 +
 fs/qnx4/inode.c          | 1 +
 fs/qnx6/inode.c          | 1 +
 fs/reiserfs/super.c      | 1 +
 fs/romfs/super.c         | 1 +
 fs/squashfs/super.c      | 1 +
 fs/super.c               | 2 --
 fs/sysv/inode.c          | 1 +
 fs/ubifs/super.c         | 1 +
 fs/udf/super.c           | 1 +
 fs/ufs/super.c           | 1 +
 fs/xfs/xfs_super.c       | 1 +
 44 files changed, 53 insertions(+), 2 deletions(-)

fs/adfs/super.c +1 -0

index 7b3003c..952aeb0 100644
@@ -212,6 +212,7 @@ static int parse_options(struct super_block *sb, char *options)
 
 static int adfs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_NODIRATIME;
 	return parse_options(sb, data);
 }

fs/affs/super.c +1 -0

index d098731..3074530 100644
@@ -530,6 +530,7 @@ affs_remount(struct super_block *sb, int *flags, char *data)
 
 	pr_debug("AFFS: remount(flags=0x%x,opts=\"%s\")\n",*flags,data);
 
+	sync_filesystem(sb);
 	*flags |= MS_NODIRATIME;
 
 	memcpy(volume, sbi->s_volume, 32);

fs/befs/linuxvfs.c +1 -0

index 845d2d6..56d70c8 100644
@@ -913,6 +913,7 @@ befs_fill_super(struct super_block *sb, void *data, int silent)
 static int
 befs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	if (!(*flags & MS_RDONLY))
 		return -EINVAL;
 	return 0;

fs/btrfs/super.c +1 -0

index 97cc241..00cd0c5 100644
@@ -1381,6 +1381,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 	unsigned int old_metadata_ratio = fs_info->metadata_ratio;
 	int ret;
 
+	sync_filesystem(sb);
 	btrfs_remount_prepare(fs_info);
 
 	ret = btrfs_parse_options(root, data);

fs/cifs/cifsfs.c +1 -0

index 849f613..4942c94 100644
@@ -541,6 +541,7 @@ static int cifs_show_stats(struct seq_file *s, struct dentry *root)
 
 static int cifs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_NODIRATIME;
 	return 0;
 }

fs/coda/inode.c +1 -0

index 506de34..3f48000 100644
@@ -96,6 +96,7 @@ void coda_destroy_inodecache(void)
 
 static int coda_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_NOATIME;
 	return 0;
 }

fs/cramfs/inode.c +1 -0

index 06610cf..a275911 100644
@@ -244,6 +244,7 @@ static void cramfs_kill_sb(struct super_block *sb)
 
 static int cramfs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_RDONLY;
 	return 0;
 }

fs/debugfs/inode.c +1 -0

index 9c0444c..02928a9 100644
@@ -218,6 +218,7 @@ static int debugfs_remount(struct super_block *sb, int *flags, char *data)
 	int err;
 	struct debugfs_fs_info *fsi = sb->s_fs_info;
 
+	sync_filesystem(sb);
 	err = debugfs_parse_options(data, &fsi->mount_opts);
 	if (err)
 		goto fail;

fs/devpts/inode.c +1 -0

index a726b9f..c710380 100644
@@ -313,6 +313,7 @@ static int devpts_remount(struct super_block *sb, int *flags, char *data)
 	struct pts_fs_info *fsi = DEVPTS_SB(sb);
 	struct pts_mount_opts *opts = &fsi->mount_opts;
 
+	sync_filesystem(sb);
 	err = parse_mount_options(data, PARSE_REMOUNT, opts);
 
 	/*

fs/efs/super.c +1 -0

index 50215bb..103bbd8 100644
@@ -114,6 +114,7 @@ static void destroy_inodecache(void)
 
 static int efs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_RDONLY;
 	return 0;
 }

fs/ext2/super.c +1 -0

index 20d6697..d260115 100644
@@ -1254,6 +1254,7 @@ static int ext2_remount (struct super_block * sb, int * flags, char * data)
 	unsigned long old_sb_flags;
 	int err;
 
+	sync_filesystem(sb);
 	spin_lock(&sbi->s_lock);
 
 	/* Store the old options */

fs/ext3/super.c +2 -0

index 37fd31e..95c6c5a 100644
@@ -2649,6 +2649,8 @@ static int ext3_remount (struct super_block * sb, int * flags, char * data)
 	int i;
 #endif
 
+	sync_filesystem(sb);
+
 	/* Store the original options */
 	old_sb_flags = sb->s_flags;
 	old_opts.s_mount_opt = sbi->s_mount_opt;

fs/ext4/super.c +2 -0

index f5c13b8..aa3842f 100644
@@ -4767,6 +4767,8 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
 #endif
 	char *orig_data = kstrdup(data, GFP_KERNEL);
 
+	sync_filesystem(sb);
+
 	/* Store the original options */
 	old_sb_flags = sb->s_flags;
 	old_opts.s_mount_opt = sbi->s_mount_opt;

fs/f2fs/super.c +2 -0

index 1a85f83..856bdf9 100644
@@ -568,6 +568,8 @@ static int f2fs_remount(struct super_block *sb, int *flags, char *data)
 	struct f2fs_mount_info org_mount_opt;
 	int err, active_logs;
 
+	sync_filesystem(sb);
+
 	/*
 	 * Save the old mount options in case we
 	 * need to restore them.

fs/fat/inode.c +2 -0

index 854b578..343e477 100644
@@ -635,6 +635,8 @@ static int fat_remount(struct super_block *sb, int *flags, char *data)
 	struct msdos_sb_info *sbi = MSDOS_SB(sb);
 	*flags |= MS_NODIRATIME | (sbi->options.isvfat ? 0 : MS_NOATIME);
 
+	sync_filesystem(sb);
+
 	/* make sure we update state on remount. */
 	new_rdonly = *flags & MS_RDONLY;
 	if (new_rdonly != (sb->s_flags & MS_RDONLY)) {

fs/freevxfs/vxfs_super.c +1 -0

index e37eb27..7ca8c75 100644
@@ -124,6 +124,7 @@ vxfs_statfs(struct dentry *dentry, struct kstatfs *bufp)
 
 static int vxfs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_RDONLY;
 	return 0;
 }

fs/fuse/inode.c +1 -0

index d468643..ecdb255d 100644
@@ -135,6 +135,7 @@ static void fuse_evict_inode(struct inode *inode)
 
 static int fuse_remount_fs(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	if (*flags & MS_MANDLOCK)
 		return -EINVAL;
 

fs/gfs2/super.c +2 -0

index 60f60f6..4c6dd50 100644
@@ -1175,6 +1175,8 @@ static int gfs2_remount_fs(struct super_block *sb, int *flags, char *data)
 	struct gfs2_tune *gt = &sdp->sd_tune;
 	int error;
 
+	sync_filesystem(sb);
+
 	spin_lock(&gt->gt_spin);
 	args.ar_commit = gt->gt_logd_secs;
 	args.ar_quota_quantum = gt->gt_quota_quantum;

fs/hfs/super.c +1 -0

index 2d2039e..eee7206 100644
@@ -112,6 +112,7 @@ static int hfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int hfs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_NODIRATIME;
 	if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
 		return 0;

fs/hfsplus/super.c +1 -0

index 80875aa..8eb787b 100644
@@ -323,6 +323,7 @@ static int hfsplus_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int hfsplus_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
 		return 0;
 	if (!(*flags & MS_RDONLY)) {

fs/hpfs/super.c +2 -0

index 4534ff6..fe3463a 100644
@@ -421,6 +421,8 @@ static int hpfs_remount_fs(struct super_block *s, int *flags, char *data)
 	struct hpfs_sb_info *sbi = hpfs_sb(s);
 	char *new_opts = kstrdup(data, GFP_KERNEL);
 
+	sync_filesystem(s);
+
 	*flags |= MS_NOATIME;
 
 	hpfs_lock(s);

fs/isofs/inode.c +1 -0

index 4a9e10e..6af66ee 100644
@@ -117,6 +117,7 @@ static void destroy_inodecache(void)
 
 static int isofs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	if (!(*flags & MS_RDONLY))
 		return -EROFS;
 	return 0;

fs/jffs2/super.c +1 -0

index 0defb1c..0918f0e 100644
@@ -243,6 +243,7 @@ static int jffs2_remount_fs(struct super_block *sb, int *flags, char *data)
 	struct jffs2_sb_info *c = JFFS2_SB_INFO(sb);
 	int err;
 
+	sync_filesystem(sb);
 	err = jffs2_parse_options(c, data);
 	if (err)
 		return -EINVAL;

fs/jfs/super.c +1 -0

index e2b7483..97f7fda 100644
@@ -418,6 +418,7 @@ static int jfs_remount(struct super_block *sb, int *flags, char *data)
 	int flag = JFS_SBI(sb)->flag;
 	int ret;
 
+	sync_filesystem(sb);
 	if (!parse_options(data, sb, &newLVSize, &flag)) {
 		return -EINVAL;
 	}

fs/minix/inode.c +1 -0

index 0332109..dcdc298 100644
@@ -123,6 +123,7 @@ static int minix_remount (struct super_block * sb, int * flags, char * data)
 	struct minix_sb_info * sbi = minix_sb(sb);
 	struct minix_super_block * ms;
 
+	sync_filesystem(sb);
 	ms = sbi->s_ms;
 	if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
 		return 0;

fs/ncpfs/inode.c +1 -0

index 2cf2ebe..5f86e80 100644
@@ -99,6 +99,7 @@ static void destroy_inodecache(void)
 
 static int ncp_remount(struct super_block *sb, int *flags, char* data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_NODIRATIME;
 	return 0;
 }

fs/nfs/super.c +2 -0

index 910ed90..2cb5694 100644
@@ -2215,6 +2215,8 @@ nfs_remount(struct super_block *sb, int *flags, char *raw_data)
 	struct nfs4_mount_data *options4 = (struct nfs4_mount_data *)raw_data;
 	u32 nfsvers = nfss->nfs_client->rpc_ops->version;
 
+	sync_filesystem(sb);
+
 	/*
 	 * Userspace mount programs that send binary options generally send
 	 * them populated with default values. We have no way to know which

fs/nilfs2/super.c +1 -0

index 7ac2a12..8c532b2 100644
@@ -1129,6 +1129,7 @@ static int nilfs_remount(struct super_block *sb, int *flags, char *data)
 	unsigned long old_mount_opt;
 	int err;
 
+	sync_filesystem(sb);
 	old_sb_flags = sb->s_flags;
 	old_mount_opt = nilfs->ns_mount_opt;
 

fs/ntfs/super.c +2 -0

index 82650d5..bd5610d 100644
@@ -468,6 +468,8 @@ static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
 
 	ntfs_debug("Entering with remount options string: %s", opt);
 
+	sync_filesystem(sb);
+
 #ifndef NTFS_RW
 	/* For read-only compiled driver, enforce read-only flag. */
 	*flags |= MS_RDONLY;

fs/ocfs2/super.c +2 -0

index 49d84f8..5f9bf8f 100644
@@ -631,6 +631,8 @@ static int ocfs2_remount(struct super_block *sb, int *flags, char *data)
 	struct ocfs2_super *osb = OCFS2_SB(sb);
 	u32 tmp;
 
+	sync_filesystem(sb);
+
 	if (!ocfs2_parse_options(sb, data, &parsed_options, 1) ||
 	    !ocfs2_check_set_options(sb, &parsed_options)) {
 		ret = -EINVAL;

fs/openpromfs/inode.c +1 -0

index 8c0ceb8..15e4500 100644
@@ -368,6 +368,7 @@ static struct inode *openprom_iget(struct super_block *sb, ino_t ino)
 
 static int openprom_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_NOATIME;
 	return 0;
 }

fs/proc/root.c +2 -0

index 87dbcbe..ac823a8 100644
@@ -92,6 +92,8 @@ static int proc_parse_options(char *options, struct pid_namespace *pid)
 int proc_remount(struct super_block *sb, int *flags, char *data)
 {
 	struct pid_namespace *pid = sb->s_fs_info;
+
+	sync_filesystem(sb);
 	return !proc_parse_options(data, pid);
 }
 

fs/pstore/inode.c +1 -0

index 1282384..192297b 100644
@@ -249,6 +249,7 @@ static void parse_options(char *options)
 
 static int pstore_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	parse_options(data);
 
 	return 0;

fs/qnx4/inode.c +1 -0

index 8955881..c4bcb77 100644
@@ -44,6 +44,7 @@ static int qnx4_remount(struct super_block *sb, int *flags, char *data)
 {
 	struct qnx4_sb_info *qs;
 
+	sync_filesystem(sb);
 	qs = qnx4_sb(sb);
 	qs->Version = QNX4_VERSION;
 	*flags |= MS_RDONLY;

fs/qnx6/inode.c +1 -0

index 8d941ed..65cdaab 100644
@@ -55,6 +55,7 @@ static int qnx6_show_options(struct seq_file *seq, struct dentry *root)
 
 static int qnx6_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_RDONLY;
 	return 0;
 }

fs/reiserfs/super.c +1 -0

index 2c80335..abf2b76 100644
@@ -1319,6 +1319,7 @@ static int reiserfs_remount(struct super_block *s, int *mount_flags, char *arg)
 	int i;
 #endif
 
+	sync_filesystem(s);
 	reiserfs_write_lock(s);
 
 #ifdef CONFIG_QUOTA

fs/romfs/super.c +1 -0

index d841878..ef90e8b 100644
@@ -432,6 +432,7 @@ static int romfs_statfs(struct dentry *dentry, struct kstatfs *buf)
  */
 static int romfs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_RDONLY;
 	return 0;
 }

fs/squashfs/super.c +1 -0

index 202df63..031c8d67 100644
@@ -371,6 +371,7 @@ static int squashfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int squashfs_remount(struct super_block *sb, int *flags, char *data)
 {
+	sync_filesystem(sb);
 	*flags |= MS_RDONLY;
 	return 0;
 }

fs/super.c +0 -2

index 80d5cf2..e9dc3c3 100644
@@ -719,8 +719,6 @@ int do_remount_sb(struct super_block *sb, int flags, void *data, int force)
 		}
 	}
 
-	sync_filesystem(sb);
-
 	if (sb->s_op->remount_fs) {
 		retval = sb->s_op->remount_fs(sb, &flags, data);
 		if (retval) {

fs/sysv/inode.c +1 -0

index c327d4e..4742e58 100644
@@ -60,6 +60,7 @@ static int sysv_remount(struct super_block *sb, int *flags, char *data)
 {
 	struct sysv_sb_info *sbi = SYSV_SB(sb);
 
+	sync_filesystem(sb);
 	if (sbi->s_forced_ro)
 		*flags |= MS_RDONLY;
 	return 0;

fs/ubifs/super.c +1 -0

index 5ded849..e1598ab 100644
@@ -1827,6 +1827,7 @@ static int ubifs_remount_fs(struct super_block *sb, int *flags, char *data)
 	int err;
 	struct ubifs_info *c = sb->s_fs_info;
 
+	sync_filesystem(sb);
 	dbg_gen("old flags %#lx, new flags %#x", sb->s_flags, *flags);
 
 	err = ubifs_parse_options(c, data, 1);

fs/udf/super.c +1 -0

index 3306b9f..64f2b73 100644
@@ -646,6 +646,7 @@ static int udf_remount_fs(struct super_block *sb, int *flags, char *options)
 	int error = 0;
 	struct logicalVolIntegrityDescImpUse *lvidiu = udf_sb_lvidiu(sb);
 
+	sync_filesystem(sb);
 	if (lvidiu) {
 		int write_rev = le16_to_cpu(lvidiu->minUDFWriteRev);
 		if (write_rev > UDF_MAX_WRITE_VERSION && !(*flags & MS_RDONLY))

fs/ufs/super.c +1 -0

index 329f2f5..b8c6791 100644
@@ -1280,6 +1280,7 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 	unsigned new_mount_opt, ufstype;
 	unsigned flags;
 
+	sync_filesystem(sb);
 	lock_ufs(sb);
 	mutex_lock(&UFS_SB(sb)->s_lock);
 	uspi = UFS_SB(sb)->s_uspi;

fs/xfs/xfs_super.c +1 -0

index f317488..aaa3eca 100644
@@ -1197,6 +1197,7 @@ xfs_fs_remount(
 	char			*p;
 	int			error;
 
+	sync_filesystem(sb);
 	while ((p = strsep(&options, ",")) != NULL) {
 		int token;
 

Signature · 1 line

1.9.0

Quoted 1 line · Ted Tso

Previously, the no-op "mount -o mount /dev/xxx" operation when the

^^remount

Quoted 6 lines

file system is already mounted read-write causes an implied, unconditional syncfs(). This seems pretty stupid, and it's certainly documented or guaraunteed to do this, nor is it particularly useful, except in the case where the file system was mounted rw and is getting remounted read-only.

Quoted 6 lines

However, it's possible that there might be some file systems that are actually depending on this behavior. In most file systems, it's probably fine to only call sync_filesystem() when transitioning from read-write to read-only, and there are some file systems where this is not needed at all (for example, for a pseudo-filesystem or something like romfs).

Hum, I'd avoid this excercise at least for filesystem where
sync_filesystem() is obviously useless - proc, debugfs, pstore, devpts, also always read-only filesystems such as isofs, qnx4, qnx6, befs, cramfs, efs, freevxfs, romfs, squashfs. I think you can find a couple more which clearly don't care about sync_filesystem() if you look a bit closer.

Honza

I guess the same is true for other file systems which are mounted ro too. So maybe a check for MS_RDONLY before doing the sync in those cases?

Quoted 4 lines · Steven Whitehouse

I guess the same is true for other file systems which are mounted ro too. So maybe a check for MS_RDONLY before doing the sync in those cases?

My original patch moved the sync_filesystem into the check for MS_RDONLY in the core VFS code. The objection was raised that there might be some file system out there that might depend on this behaviour. I can't imagine why, but I suppose it's at least theoretically possible.

So the idea is that this particular patch is *guaranteed* not to make any difference. That way there can be no question about the patch'es correctness.

I'm going to follow up with a patch for ext4 that does exactly that, but the idea is to allow each file system maintainer to do that for their own file system.

I could do that as well for file systems that are "obviously" read-only, but then I'll find out that there's some wierd case where the file system can be used in a read-write fashion. (Example: UDF is normally used for DVD's, but at least in theory it can be used read/write --- I'm told that Windows supports read-write UDF file systems on USB sticks, and at least in theory it could be used as a inter-OS exchange format in situations where VFAT and exFAT might not be appropriate for various reasons.)

Previously, the no-op "mount -o mount /dev/xxx" operation when the file system is already mounted read-write causes an implied, unconditional syncfs(). This seems pretty stupid, and it's certainly documented or guaraunteed to do this, nor is it particularly useful, except in the case where the file system was mounted rw and is getting remounted read-only.

However, it's possible that there might be some file systems that are actually depending on this behavior. In most file systems, it's probably fine to only call sync_filesystem() when transitioning from read-write to read-only, and there are some file systems where this is not needed at all (for example, for a pseudo-filesystem or something like romfs).

Signed-off-by "Theodore Ts'o" <[email protected]>

Cc linux-fsdevel at vger.kernel.org

Cc Christoph Hellwig <[email protected]>

Cc Artem Bityutskiy <[email protected]>

Cc Adrian Hunter <[email protected]>

Cc Evgeniy Dushistov <[email protected]>

Cc Jan Kara <[email protected]>

Cc OGAWA Hirofumi <[email protected]>

Cc Anders Larsen <[email protected]>

Cc Phillip Lougher <[email protected]>

Cc Kees Cook <[email protected]>

Cc Mikulas Patocka <[email protected]>

Cc Petr Vandrovec <[email protected]>

Cc xfs at oss.sgi.com

Cc linux-btrfs at vger.kernel.org

Cc linux-cifs at vger.kernel.org

Cc samba-technical at lists.samba.org

Cc codalist at coda.cs.cmu.edu

Cc linux-ext4 at vger.kernel.org

Cc linux-f2fs-devel at lists.sourceforge.net

Cc fuse-devel at lists.sourceforge.net

Cc cluster-devel at redhat.com

Cc linux-mtd at lists.infradead.org

Cc jfs-discussion at lists.sourceforge.net

Cc linux-nfs at vger.kernel.org

Cc linux-nilfs at vger.kernel.org

Cc linux-ntfs-dev at lists.sourceforge.net

Cc ocfs2-devel at oss.oracle.com

Cc reiserfs-devel at vger.kernel.org

---
 fs/adfs/super.c          | 1 +
 fs/affs/super.c          | 1 +
 fs/befs/linuxvfs.c       | 1 +
 fs/btrfs/super.c         | 1 +
 fs/cifs/cifsfs.c         | 1 +
 fs/coda/inode.c          | 1 +
 fs/cramfs/inode.c        | 1 +
 fs/debugfs/inode.c       | 1 +
 fs/devpts/inode.c        | 1 +
 fs/efs/super.c           | 1 +
 fs/ext2/super.c          | 1 +
 fs/ext3/super.c          | 2 ++
 fs/ext4/super.c          | 2 ++
 fs/f2fs/super.c          | 2 ++
 fs/fat/inode.c           | 2 ++
 fs/freevxfs/vxfs_super.c | 1 +
 fs/fuse/inode.c          | 1 +
 fs/gfs2/super.c          | 2 ++
 fs/hfs/super.c           | 1 +
 fs/hfsplus/super.c       | 1 +
 fs/hpfs/super.c          | 2 ++
 fs/isofs/inode.c         | 1 +
 fs/jffs2/super.c         | 1 +
 fs/jfs/super.c           | 1 +
 fs/minix/inode.c         | 1 +
 fs/ncpfs/inode.c         | 1 +
 fs/nfs/super.c           | 2 ++
 fs/nilfs2/super.c        | 1 +
 fs/ntfs/super.c          | 2 ++
 fs/ocfs2/super.c         | 2 ++
 fs/openpromfs/inode.c    | 1 +
 fs/proc/root.c           | 2 ++
 fs/pstore/inode.c        | 1 +
 fs/qnx4/inode.c          | 1 +
 fs/qnx6/inode.c          | 1 +
 fs/reiserfs/super.c      | 1 +
 fs/romfs/super.c         | 1 +
 fs/squashfs/super.c      | 1 +
 fs/super.c               | 2 --
 fs/sysv/inode.c          | 1 +
 fs/ubifs/super.c         | 1 +
 fs/udf/super.c           | 1 +
 fs/ufs/super.c           | 1 +
 fs/xfs/xfs_super.c       | 1 +
 44 files changed, 53 insertions(+), 2 deletions(-)

fs/adfs/super.c +1 -0

index 7b3003c..952aeb0 100644
@@ -212,6 +212,7 @@ static int parse_options(struct super_block *sb, char *options)
 
 static int adfs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_NODIRATIME;
        return parse_options(sb, data);
 }

fs/affs/super.c +1 -0

index d098731..3074530 100644
@@ -530,6 +530,7 @@ affs_remount(struct super_block *sb, int *flags, char *data)
 
        pr_debug("AFFS: remount(flags=0x%x,opts=\"%s\")\n",*flags,data);
 
+       sync_filesystem(sb);
        *flags |= MS_NODIRATIME;
 
        memcpy(volume, sbi->s_volume, 32);

fs/befs/linuxvfs.c +1 -0

index 845d2d6..56d70c8 100644
@@ -913,6 +913,7 @@ befs_fill_super(struct super_block *sb, void *data, int silent)
 static int
 befs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        if (!(*flags & MS_RDONLY))
                return -EINVAL;
        return 0;

fs/btrfs/super.c +1 -0

index 97cc241..00cd0c5 100644
@@ -1381,6 +1381,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
        unsigned int old_metadata_ratio = fs_info->metadata_ratio;
        int ret;
 
+       sync_filesystem(sb);
        btrfs_remount_prepare(fs_info);
 
        ret = btrfs_parse_options(root, data);

fs/cifs/cifsfs.c +1 -0

index 849f613..4942c94 100644
@@ -541,6 +541,7 @@ static int cifs_show_stats(struct seq_file *s, struct dentry *root)
 
 static int cifs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_NODIRATIME;
        return 0;
 }

fs/coda/inode.c +1 -0

index 506de34..3f48000 100644
@@ -96,6 +96,7 @@ void coda_destroy_inodecache(void)
 
 static int coda_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_NOATIME;
        return 0;
 }

fs/cramfs/inode.c +1 -0

index 06610cf..a275911 100644
@@ -244,6 +244,7 @@ static void cramfs_kill_sb(struct super_block *sb)
 
 static int cramfs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_RDONLY;
        return 0;
 }

fs/debugfs/inode.c +1 -0

index 9c0444c..02928a9 100644
@@ -218,6 +218,7 @@ static int debugfs_remount(struct super_block *sb, int *flags, char *data)
        int err;
        struct debugfs_fs_info *fsi = sb->s_fs_info;
 
+       sync_filesystem(sb);
        err = debugfs_parse_options(data, &fsi->mount_opts);
        if (err)
                goto fail;

fs/devpts/inode.c +1 -0

index a726b9f..c710380 100644
@@ -313,6 +313,7 @@ static int devpts_remount(struct super_block *sb, int *flags, char *data)
        struct pts_fs_info *fsi = DEVPTS_SB(sb);
        struct pts_mount_opts *opts = &fsi->mount_opts;
 
+       sync_filesystem(sb);
        err = parse_mount_options(data, PARSE_REMOUNT, opts);
 
        /*

fs/efs/super.c +1 -0

index 50215bb..103bbd8 100644
@@ -114,6 +114,7 @@ static void destroy_inodecache(void)
 
 static int efs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_RDONLY;
        return 0;
 }

fs/ext2/super.c +1 -0

index 20d6697..d260115 100644
@@ -1254,6 +1254,7 @@ static int ext2_remount (struct super_block * sb, int * flags, char * data)
        unsigned long old_sb_flags;
        int err;
 
+       sync_filesystem(sb);
        spin_lock(&sbi->s_lock);
 
        /* Store the old options */

fs/ext3/super.c +2 -0

index 37fd31e..95c6c5a 100644
@@ -2649,6 +2649,8 @@ static int ext3_remount (struct super_block * sb, int * flags, char * data)
        int i;
 #endif
 
+       sync_filesystem(sb);
+
        /* Store the original options */
        old_sb_flags = sb->s_flags;
        old_opts.s_mount_opt = sbi->s_mount_opt;

fs/ext4/super.c +2 -0

index f5c13b8..aa3842f 100644
@@ -4767,6 +4767,8 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
 #endif
        char *orig_data = kstrdup(data, GFP_KERNEL);
 
+       sync_filesystem(sb);
+
        /* Store the original options */
        old_sb_flags = sb->s_flags;
        old_opts.s_mount_opt = sbi->s_mount_opt;

fs/f2fs/super.c +2 -0

index 1a85f83..856bdf9 100644
@@ -568,6 +568,8 @@ static int f2fs_remount(struct super_block *sb, int *flags, char *data)
        struct f2fs_mount_info org_mount_opt;
        int err, active_logs;
 
+       sync_filesystem(sb);
+
        /*
         * Save the old mount options in case we
         * need to restore them.

fs/fat/inode.c +2 -0

index 854b578..343e477 100644
@@ -635,6 +635,8 @@ static int fat_remount(struct super_block *sb, int *flags, char *data)
        struct msdos_sb_info *sbi = MSDOS_SB(sb);
        *flags |= MS_NODIRATIME | (sbi->options.isvfat ? 0 : MS_NOATIME);
 
+       sync_filesystem(sb);
+
        /* make sure we update state on remount. */
        new_rdonly = *flags & MS_RDONLY;
        if (new_rdonly != (sb->s_flags & MS_RDONLY)) {

fs/freevxfs/vxfs_super.c +1 -0

index e37eb27..7ca8c75 100644
@@ -124,6 +124,7 @@ vxfs_statfs(struct dentry *dentry, struct kstatfs *bufp)
 
 static int vxfs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_RDONLY;
        return 0;
 }

fs/fuse/inode.c +1 -0

index d468643..ecdb255d 100644
@@ -135,6 +135,7 @@ static void fuse_evict_inode(struct inode *inode)
 
 static int fuse_remount_fs(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        if (*flags & MS_MANDLOCK)
                return -EINVAL;
 

fs/gfs2/super.c +2 -0

index 60f60f6..4c6dd50 100644
@@ -1175,6 +1175,8 @@ static int gfs2_remount_fs(struct super_block *sb, int *flags, char *data)
        struct gfs2_tune *gt = &sdp->sd_tune;
        int error;
 
+       sync_filesystem(sb);
+
        spin_lock(&gt->gt_spin);
        args.ar_commit = gt->gt_logd_secs;
        args.ar_quota_quantum = gt->gt_quota_quantum;

fs/hfs/super.c +1 -0

index 2d2039e..eee7206 100644
@@ -112,6 +112,7 @@ static int hfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int hfs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_NODIRATIME;
        if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
                return 0;

fs/hfsplus/super.c +1 -0

index 80875aa..8eb787b 100644
@@ -323,6 +323,7 @@ static int hfsplus_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int hfsplus_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
                return 0;
        if (!(*flags & MS_RDONLY)) {

fs/hpfs/super.c +2 -0

index 4534ff6..fe3463a 100644
@@ -421,6 +421,8 @@ static int hpfs_remount_fs(struct super_block *s, int *flags, char *data)
        struct hpfs_sb_info *sbi = hpfs_sb(s);
        char *new_opts = kstrdup(data, GFP_KERNEL);
 
+       sync_filesystem(s);
+
        *flags |= MS_NOATIME;
 
        hpfs_lock(s);

fs/isofs/inode.c +1 -0

index 4a9e10e..6af66ee 100644
@@ -117,6 +117,7 @@ static void destroy_inodecache(void)
 
 static int isofs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        if (!(*flags & MS_RDONLY))
                return -EROFS;
        return 0;

fs/jffs2/super.c +1 -0

index 0defb1c..0918f0e 100644
@@ -243,6 +243,7 @@ static int jffs2_remount_fs(struct super_block *sb, int *flags, char *data)
        struct jffs2_sb_info *c = JFFS2_SB_INFO(sb);
        int err;
 
+       sync_filesystem(sb);
        err = jffs2_parse_options(c, data);
        if (err)
                return -EINVAL;

fs/jfs/super.c +1 -0

index e2b7483..97f7fda 100644
@@ -418,6 +418,7 @@ static int jfs_remount(struct super_block *sb, int *flags, char *data)
        int flag = JFS_SBI(sb)->flag;
        int ret;
 
+       sync_filesystem(sb);
        if (!parse_options(data, sb, &newLVSize, &flag)) {
                return -EINVAL;
        }

fs/minix/inode.c +1 -0

index 0332109..dcdc298 100644
@@ -123,6 +123,7 @@ static int minix_remount (struct super_block * sb, int * flags, char * data)
        struct minix_sb_info * sbi = minix_sb(sb);
        struct minix_super_block * ms;
 
+       sync_filesystem(sb);
        ms = sbi->s_ms;
        if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
                return 0;

fs/ncpfs/inode.c +1 -0

index 2cf2ebe..5f86e80 100644
@@ -99,6 +99,7 @@ static void destroy_inodecache(void)
 
 static int ncp_remount(struct super_block *sb, int *flags, char* data)
 {
+       sync_filesystem(sb);
        *flags |= MS_NODIRATIME;
        return 0;
 }

fs/nfs/super.c +2 -0

index 910ed90..2cb5694 100644
@@ -2215,6 +2215,8 @@ nfs_remount(struct super_block *sb, int *flags, char *raw_data)
        struct nfs4_mount_data *options4 = (struct nfs4_mount_data *)raw_data;
        u32 nfsvers = nfss->nfs_client->rpc_ops->version;
 
+       sync_filesystem(sb);
+
        /*
         * Userspace mount programs that send binary options generally send
         * them populated with default values. We have no way to know which

fs/nilfs2/super.c +1 -0

index 7ac2a12..8c532b2 100644
@@ -1129,6 +1129,7 @@ static int nilfs_remount(struct super_block *sb, int *flags, char *data)
        unsigned long old_mount_opt;
        int err;
 
+       sync_filesystem(sb);
        old_sb_flags = sb->s_flags;
        old_mount_opt = nilfs->ns_mount_opt;
 

fs/ntfs/super.c +2 -0

index 82650d5..bd5610d 100644
@@ -468,6 +468,8 @@ static int ntfs_remount(struct super_block *sb, int *flags, char *opt)
 
        ntfs_debug("Entering with remount options string: %s", opt);
 
+       sync_filesystem(sb);
+
 #ifndef NTFS_RW
        /* For read-only compiled driver, enforce read-only flag. */
        *flags |= MS_RDONLY;

fs/ocfs2/super.c +2 -0

index 49d84f8..5f9bf8f 100644
@@ -631,6 +631,8 @@ static int ocfs2_remount(struct super_block *sb, int *flags, char *data)
        struct ocfs2_super *osb = OCFS2_SB(sb);
        u32 tmp;
 
+       sync_filesystem(sb);
+
        if (!ocfs2_parse_options(sb, data, &parsed_options, 1) ||
            !ocfs2_check_set_options(sb, &parsed_options)) {
                ret = -EINVAL;

fs/openpromfs/inode.c +1 -0

index 8c0ceb8..15e4500 100644
@@ -368,6 +368,7 @@ static struct inode *openprom_iget(struct super_block *sb, ino_t ino)
 
 static int openprom_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_NOATIME;
        return 0;
 }

fs/proc/root.c +2 -0

index 87dbcbe..ac823a8 100644
@@ -92,6 +92,8 @@ static int proc_parse_options(char *options, struct pid_namespace *pid)
 int proc_remount(struct super_block *sb, int *flags, char *data)
 {
        struct pid_namespace *pid = sb->s_fs_info;
+
+       sync_filesystem(sb);
        return !proc_parse_options(data, pid);
 }
 

fs/pstore/inode.c +1 -0

index 1282384..192297b 100644
@@ -249,6 +249,7 @@ static void parse_options(char *options)
 
 static int pstore_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        parse_options(data);
 
        return 0;

fs/qnx4/inode.c +1 -0

index 8955881..c4bcb77 100644
@@ -44,6 +44,7 @@ static int qnx4_remount(struct super_block *sb, int *flags, char *data)
 {
        struct qnx4_sb_info *qs;
 
+       sync_filesystem(sb);
        qs = qnx4_sb(sb);
        qs->Version = QNX4_VERSION;
        *flags |= MS_RDONLY;

fs/qnx6/inode.c +1 -0

index 8d941ed..65cdaab 100644
@@ -55,6 +55,7 @@ static int qnx6_show_options(struct seq_file *seq, struct dentry *root)
 
 static int qnx6_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_RDONLY;
        return 0;
 }

fs/reiserfs/super.c +1 -0

index 2c80335..abf2b76 100644
@@ -1319,6 +1319,7 @@ static int reiserfs_remount(struct super_block *s, int *mount_flags, char *arg)
        int i;
 #endif
 
+       sync_filesystem(s);
        reiserfs_write_lock(s);
 
 #ifdef CONFIG_QUOTA

fs/romfs/super.c +1 -0

index d841878..ef90e8b 100644
@@ -432,6 +432,7 @@ static int romfs_statfs(struct dentry *dentry, struct kstatfs *buf)
  */
 static int romfs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_RDONLY;
        return 0;
 }

fs/squashfs/super.c +1 -0

index 202df63..031c8d67 100644
@@ -371,6 +371,7 @@ static int squashfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int squashfs_remount(struct super_block *sb, int *flags, char *data)
 {
+       sync_filesystem(sb);
        *flags |= MS_RDONLY;
        return 0;
 }

fs/super.c +0 -2

index 80d5cf2..e9dc3c3 100644
@@ -719,8 +719,6 @@ int do_remount_sb(struct super_block *sb, int flags, void *data, int force)
                }
        }
 
-       sync_filesystem(sb);
-
        if (sb->s_op->remount_fs) {
                retval = sb->s_op->remount_fs(sb, &flags, data);
                if (retval) {

fs/sysv/inode.c +1 -0

index c327d4e..4742e58 100644
@@ -60,6 +60,7 @@ static int sysv_remount(struct super_block *sb, int *flags, char *data)
 {
        struct sysv_sb_info *sbi = SYSV_SB(sb);
 
+       sync_filesystem(sb);
        if (sbi->s_forced_ro)
                *flags |= MS_RDONLY;
        return 0;

fs/ubifs/super.c +1 -0

index 5ded849..e1598ab 100644
@@ -1827,6 +1827,7 @@ static int ubifs_remount_fs(struct super_block *sb, int *flags, char *data)
        int err;
        struct ubifs_info *c = sb->s_fs_info;
 
+       sync_filesystem(sb);
        dbg_gen("old flags %#lx, new flags %#x", sb->s_flags, *flags);
 
        err = ubifs_parse_options(c, data, 1);

fs/udf/super.c +1 -0

index 3306b9f..64f2b73 100644
@@ -646,6 +646,7 @@ static int udf_remount_fs(struct super_block *sb, int *flags, char *options)
        int error = 0;
        struct logicalVolIntegrityDescImpUse *lvidiu = udf_sb_lvidiu(sb);
 
+       sync_filesystem(sb);
        if (lvidiu) {
                int write_rev = le16_to_cpu(lvidiu->minUDFWriteRev);
                if (write_rev > UDF_MAX_WRITE_VERSION && !(*flags & MS_RDONLY))

fs/ufs/super.c +1 -0

index 329f2f5..b8c6791 100644
@@ -1280,6 +1280,7 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
        unsigned new_mount_opt, ufstype;
        unsigned flags;
 
+       sync_filesystem(sb);
        lock_ufs(sb);
        mutex_lock(&UFS_SB(sb)->s_lock);
        uspi = UFS_SB(sb)->s_uspi;

fs/xfs/xfs_super.c +1 -1

index f317488..aaa3eca 100644
@@ -1197,6 +1197,7 @@ xfs_fs_remount(
        char                    *p;
        int                     error;
 
+       sync_filesystem(sb);
        while ((p = strsep(&options, ",")) != NULL) {
                int token;
 
--

1.9.0

--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo at vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Quoted 6 lines · Theodore Ts'o

Previously, the no-op "mount -o mount /dev/xxx" operation when the file system is already mounted read-write causes an implied, unconditional syncfs(). This seems pretty stupid, and it's certainly documented or guaraunteed to do this, nor is it particularly useful, except in the case where the file system was mounted rw and is getting remounted read-only.

Is there a case where a file system, not mounted read-only, would want to skip the syncfs on remount? I don't know of any particular reason to do a syncfs on remount unless caching behavior is changing (or moving to read-only mount), but if as you say it is documented and guaranteed...

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

When starting hundreds of LXC containers simultaneously on a system, the boot of some containers was hanging. We tracked this down to an initscript's use of mount -o remount, which was hanging in D state.

We reproduced the problem outside of LXC, with the script available at [0]. That script initiates 1000 mount -o remount, and performs some writes using a big cp to the same filesystem during the remounts....

TL;DR: we experience long temporary hangs when doing multiple mount -o remount at the same time as other I/O on an ext4 filesystem.

Hi,

When starting hundreds of LXC containers simultaneously on a system, the boot of some containers was hanging. We tracked this down to an initscript's use of mount -o remount, which was hanging in D state.

We reproduced the problem outside of LXC, with the script available at [0]. That script initiates 1000 mount -o remount, and performs some writes using a big cp to the same filesystem during the remounts.

Some other things we tried:
1) we tried to 'sync' after removing the files, and dropping the caches (as shown in the commented lines in [0]). That makes the problem disappear (or at least makes it less frequent). The overall script execution is actually faster with the post-rm sync and dropping caches than without them!

2) We tried switching to the noop scheduler (instead of cfq). The problem could still be reproduced. A btrace dump with noop is available at [2].

3) We tried with ext3 instead of ext4. The problem could never be reproduced.

4) We tried on different machines, and we could reproduce the problem. However, on a machine with SSD drives, we were not able to reproduce the problem.

Any ideas?