linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Fernando Luis Vázquez Cao" <fernando_b1@lab.ntt.co.jp>
To: Al Viro <viro@zeniv.linux.org.uk>
Cc: Josef Bacik <jbacik@fusionio.com>,
	Eric Sandeen <sandeen@redhat.com>,
	Dave Chinner <dchinner@redhat.com>,
	Christoph Hellwig <hch@infradead.org>, Jan Kara <jack@suse.cz>,
	Luiz Capitulino <lcapitulino@redhat.com>,
	linux-fsdevel@vger.kernel.org
Subject: [PATCH 4/17] fsfreeze: emergency thaw will deadlock on s_umount
Date: Mon, 07 Jan 2013 20:26:12 +0900	[thread overview]
Message-ID: <1357557972.8183.7.camel@nexus.lab.ntt.co.jp> (raw)
In-Reply-To: <1357557492.8183.1.camel@nexus.lab.ntt.co.jp>

The emergency thaw process uses iterate_super() which holds the
sb->s_umount lock in read mode. The current thaw_super() code takes
the sb->s_umount lock in write mode, hence leading to an instant
deadlock.
 
Use the unlocked version of thaw_super() to do the thawing and replace
iterate_supers() with iterate_supers_write() so that the unfreeze operation can
be performed with s_umount held as the locking rules for fsfreeze indicate.

As a bonus, by using thaw_super(), which does not nest, instead of thaw_bdev()
when can get rid of the ugly while loop.

Jan Kara pointed out that with this approach we will leave the block devices
frozen, but this is a problem we have had since the introduction of the
superblock level API: if we thaw the filesystem using the superblock level API
(be it through the thaw ioctl or emergency thaw) the bdev level freeze
reference counter (bd_fsfreeze_count) will not be updated and even though
subsequent calls to thaw_bdev() will decrease it it will never get back to 0
(if thaw_super() returns an error, and it will when the superblock is unfrozen,
thaw_bdev() will return without decreasing the counter). The solution I propose
(and will be implementing in the followup patch "fsfreeze: freeze_super and
thaw_bdev don't play well together") is letting bd_fsfreeze_count
become zero when the superblock sitting on top of it is unfrozen, so that
future calls to freeze_bdev() actually try to freeze the superblock.

Cc: linux-fsdevel@vger.kernel.org
Cc: Josef Bacik <jbacik@fusionio.com>
Cc: Eric Sandeen <sandeen@redhat.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Dave Chinner <dchinner@redhat.com>
Cc: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Fernando Luis Vazquez Cao <fernando@oss.ntt.co.jp>
---

diff -urNp linux-3.8-rc1-orig/drivers/tty/sysrq.c linux-3.8-rc1/drivers/tty/sysrq.c
--- linux-3.8-rc1-orig/drivers/tty/sysrq.c	2012-12-25 10:27:40.614737000 +0900
+++ linux-3.8-rc1/drivers/tty/sysrq.c	2012-12-25 11:40:06.128018000 +0900
@@ -363,7 +363,6 @@ static struct sysrq_key_op sysrq_moom_op
 	.enable_mask	= SYSRQ_ENABLE_SIGNAL,
 };
 
-#ifdef CONFIG_BLOCK
 static void sysrq_handle_thaw(int key)
 {
 	emergency_thaw_all();
@@ -374,7 +373,6 @@ static struct sysrq_key_op sysrq_thaw_op
 	.action_msg	= "Emergency Thaw of all frozen filesystems",
 	.enable_mask	= SYSRQ_ENABLE_SIGNAL,
 };
-#endif
 
 static void sysrq_handle_kill(int key)
 {
diff -urNp linux-3.8-rc1-orig/fs/buffer.c linux-3.8-rc1/fs/buffer.c
--- linux-3.8-rc1-orig/fs/buffer.c	2012-12-25 11:30:38.208018000 +0900
+++ linux-3.8-rc1/fs/buffer.c	2012-12-25 11:40:06.128018000 +0900
@@ -512,15 +512,33 @@ repeat:
 
 static void do_thaw_one(struct super_block *sb, void *unused)
 {
-	char b[BDEVNAME_SIZE];
-	while (sb->s_bdev && !thaw_bdev(sb->s_bdev, sb))
-		printk(KERN_WARNING "Emergency Thaw on %s\n",
+	int res;
+
+	if (sb->s_bdev) {
+		char b[BDEVNAME_SIZE];
+		printk(KERN_WARNING "Emergency Thaw on %s.\n",
 		       bdevname(sb->s_bdev, b));
+	}
+
+	/*
+	 * We got here from __iterate_supers with the superblock lock taken
+	 * so we can call the lockless version of thaw_super() safely.
+	 */
+	res = __thaw_super(sb);
+	if (!res) {
+		deactivate_locked_super(sb);
+		/*
+		 * We have to re-acquire s_umount because
+		 * iterate_supers_write() will unlock it. It still holds
+		 * passive reference so sb cannot be freed under us.
+		 */
+		down_write(&sb->s_umount);
+	}
 }
 
 static void do_thaw_all(struct work_struct *work)
 {
-	iterate_supers_read(do_thaw_one, NULL);
+	iterate_supers_write(do_thaw_one, NULL);
 	kfree(work);
 	printk(KERN_WARNING "Emergency Thaw complete\n");
 }
diff -urNp linux-3.8-rc1-orig/include/linux/fs.h linux-3.8-rc1/include/linux/fs.h
--- linux-3.8-rc1-orig/include/linux/fs.h	2012-12-25 11:35:55.488018000 +0900
+++ linux-3.8-rc1/include/linux/fs.h	2012-12-25 11:40:06.132018000 +0900
@@ -1881,6 +1881,7 @@ extern int vfs_ustat(dev_t, struct kstat
 extern int freeze_super(struct super_block *super);
 extern int __thaw_super(struct super_block *super);
 extern int thaw_super(struct super_block *super);
+extern void emergency_thaw_all(void);
 extern bool our_mnt(struct vfsmount *mnt);
 
 extern int current_umask(void);
@@ -2053,7 +2054,6 @@ extern void iterate_bdevs(void (*)(struc
 extern int sync_blockdev(struct block_device *bdev);
 extern void kill_bdev(struct block_device *);
 extern struct super_block *freeze_bdev(struct block_device *);
-extern void emergency_thaw_all(void);
 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
 extern int fsync_bdev(struct block_device *);
 #else



  parent reply	other threads:[~2013-01-07 11:27 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-01-07 11:18 [PATCH v6 0/17] fsfreeze: miscellaneous fixes and cleanups Fernando Luis Vázquez Cao
2013-01-07 11:21 ` [PATCH 1/17] vfs: add __iterate_supers() and helpers around it Fernando Luis Vázquez Cao
2013-01-07 11:22 ` [PATCH 2/17] fsfreeze: add unlocked version of thaw_super Fernando Luis Vázquez Cao
2013-01-07 11:23 ` [PATCH 3/17] fsfreeze: fix emergency thaw infinite loop Fernando Luis Vázquez Cao
2013-01-07 11:26 ` Fernando Luis Vázquez Cao [this message]
2013-01-09 16:12   ` [PATCH 4/17] fsfreeze: emergency thaw will deadlock on s_umount Jan Kara
2013-01-07 11:27 ` [PATCH 5/17] xfs: switch to using super methods for fsfreeze Fernando Luis Vázquez Cao
2013-01-07 11:29 ` [PATCH 6/17] fsfreeze: move emergency thaw code to fs/super.c Fernando Luis Vázquez Cao
2013-01-07 11:30 ` [PATCH 7/17] fsfreeze: fix nested freezing of sb-less bdevs Fernando Luis Vázquez Cao
2013-01-09 16:24   ` Jan Kara
2013-01-07 11:32 ` [PATCH 8/17] fsfreeze: allow bdev level thaws when the sb is unfrozen Fernando Luis Vázquez Cao
2013-01-09 16:26   ` Jan Kara
2013-01-07 11:34 ` [PATCH 9/17] fsfreeze: freeze_super and thaw_bdev don't play well together Fernando Luis Vázquez Cao
2013-01-07 11:35 ` [PATCH 10/17] fsfreeze: automatically thaw on umount Fernando Luis Vázquez Cao
2013-01-09 17:20   ` Jan Kara
2013-01-10  9:14     ` Fernando Luis Vazquez Cao
2013-01-07 11:36 ` [PATCH 11/17] fsfreeze: add thaw_super_force Fernando Luis Vázquez Cao
2013-01-07 11:38 ` [PATCH 12/17] fsfreeze: sb-level/bdev-level fsfreeze integration Fernando Luis Vázquez Cao
2013-01-09 16:37   ` Jan Kara
2013-01-10  9:57     ` Fernando Luis Vazquez Cao
2013-01-07 11:39 ` [PATCH 13/17] fsfreeze: unfreeze bdevs in addition to filesystems during emergency thaw Fernando Luis Vázquez Cao
2013-01-09 16:41   ` Jan Kara
2013-01-07 11:41 ` [PATCH 14/17] vfs: leverage bd_super in get_super and get_active_super Fernando Luis Vázquez Cao
2013-01-09 16:44   ` Jan Kara
2013-01-07 11:42 ` [PATCH 15/17] btrfs: store pointer to superblock in bd_super Fernando Luis Vázquez Cao
2013-01-07 11:43 ` [PATCH 16/17] fsfreeze: allow freeze counter lock nesting Fernando Luis Vázquez Cao
2013-01-07 11:44 ` [PATCH 17/17] fsfreeze: export freeze_count through mountinfo Fernando Luis Vázquez Cao

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=1357557972.8183.7.camel@nexus.lab.ntt.co.jp \
    --to=fernando_b1@lab.ntt.co.jp \
    --cc=dchinner@redhat.com \
    --cc=hch@infradead.org \
    --cc=jack@suse.cz \
    --cc=jbacik@fusionio.com \
    --cc=lcapitulino@redhat.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=sandeen@redhat.com \
    --cc=viro@zeniv.linux.org.uk \
    /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;
as well as URLs for NNTP newsgroup(s).