linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jan Blunck <jblunck@suse.de>
To: Kirill Korotaev <dev@sw.ru>
Cc: viro@zeniv.linux.org.uk, linux-kernel@vger.kernel.org,
	Andrew Morton <akpm@osdl.org>,
	olh@suse.de, balbir@in.ibm.com
Subject: Re: [PATCH] shrink_dcache_parent() races against shrink_dcache_memory()
Date: Mon, 30 Jan 2006 13:03:18 +0100	[thread overview]
Message-ID: <20060130120318.GB9181@hasse.suse.de> (raw)
In-Reply-To: <43D48ED4.3010306@sw.ru>

[-- Attachment #1: Type: text/plain, Size: 701 bytes --]

On Mon, Jan 23, Kirill Korotaev wrote:

> 1. this patch doesn't fix the whole problem. iput() after sb free is 
> still possible. So busy inodes after umount too.
> 2. it has big problems with locking...
> 

Uh yeah! I fixed the second issue but since the patch doesnt helped and only
gots the reference counting a little bit cleaner I don't post it.

> comments below inside.
> 

New patch attached below. Comments are welcome.

Regards,
	Jan

-- 
Jan Blunck                                               jblunck@suse.de
SuSE LINUX AG - A Novell company
Maxfeldstr. 5                                          +49-911-74053-608
D-90409 Nürnberg                                      http://www.suse.de

[-- Attachment #2: umount-prune_one_dentry-fix.diff --]
[-- Type: text/plain, Size: 3770 bytes --]

From: Jan Blunck <jblunck@suse.de>
Subject: Fix shrink_dcache_parent() against shrink_dcache_memory() race
References: 136310

Kirill Korotaev <dev@sw.ru> discovered a race between shrink_dcache_parent()
and shrink_dcache_memory() which leads to "Busy inodes after unmount".
When unmounting a file system shrink_dcache_parent() is racing against a
possible shrink_dcache_memory(). This might lead to the situation that
shrink_dcache_parent() is returning too early. In this situation the
super_block is destroyed before shrink_dcache_memory() could put the inode.

This patch fixes the problem through introducing a prunes counter which is
incremented when a dentry is pruned but the corresponding inoded isn't put yet.
When the prunes counter is not null, shrink_dcache_parent() is waiting and
restarting its work.

Signed-off-by: Jan Blunck <jblunck@suse.de>

---

 fs/dcache.c        |   36 ++++++++++++++++++++++++++++++++++++
 fs/super.c         |    4 +++-
 include/linux/fs.h |    3 +++
 3 files changed, 42 insertions(+), 1 deletion(-)

Index: linux-2.6/fs/dcache.c
===================================================================
--- linux-2.6.orig/fs/dcache.c
+++ linux-2.6/fs/dcache.c
@@ -364,17 +364,21 @@ restart:
  */
 static inline void prune_one_dentry(struct dentry * dentry)
 {
+	struct super_block *sb = dentry->d_sb;
 	struct dentry * parent;
 
 	__d_drop(dentry);
 	list_del(&dentry->d_u.d_child);
 	dentry_stat.nr_dentry--;	/* For d_free, below */
+	sb->s_prunes++;
 	dentry_iput(dentry);
 	parent = dentry->d_parent;
 	d_free(dentry);
 	if (parent != dentry)
 		dput(parent);
 	spin_lock(&dcache_lock);
+	sb->s_prunes--;
+	wake_up(&sb->s_wait_prunes);
 }
 
 /**
@@ -623,6 +627,34 @@ out:
 	return found;
 }
 
+static int wait_on_prunes(struct super_block *sb)
+{
+	DEFINE_WAIT(wait);
+
+	spin_lock(&dcache_lock);
+	if (!sb->s_prunes) {
+		spin_unlock(&dcache_lock);
+		return 0;
+	}
+
+	printk(KERN_DEBUG "%s: waiting for %d prunes\n", __FUNCTION__,
+	       sb->s_prunes);
+
+	while (1) {
+		prepare_to_wait(&sb->s_wait_prunes, &wait,
+				TASK_UNINTERRUPTIBLE);
+		if (!sb->s_prunes)
+			break;
+		spin_unlock(&dcache_lock);
+		schedule();
+		spin_lock(&dcache_lock);
+	}
+
+	finish_wait(&sb->s_wait_prunes, &wait);
+	spin_unlock(&dcache_lock);
+	return 1;
+}
+
 /**
  * shrink_dcache_parent - prune dcache
  * @parent: parent of entries to prune
@@ -634,8 +666,12 @@ void shrink_dcache_parent(struct dentry 
 {
 	int found;
 
+ again:
 	while ((found = select_parent(parent)) != 0)
 		prune_dcache(found);
+
+	if (wait_on_prunes(parent->d_sb))
+		goto again;
 }
 
 /**
Index: linux-2.6/fs/super.c
===================================================================
--- linux-2.6.orig/fs/super.c
+++ linux-2.6/fs/super.c
@@ -80,6 +80,8 @@ static struct super_block *alloc_super(v
 		sema_init(&s->s_dquot.dqio_sem, 1);
 		sema_init(&s->s_dquot.dqonoff_sem, 1);
 		init_rwsem(&s->s_dquot.dqptr_sem);
+		s->s_prunes = 0;
+		init_waitqueue_head(&s->s_wait_prunes);
 		init_waitqueue_head(&s->s_wait_unfrozen);
 		s->s_maxbytes = MAX_NON_LFS;
 		s->dq_op = sb_dquot_ops;
@@ -230,8 +232,8 @@ void generic_shutdown_super(struct super
 
 	if (root) {
 		sb->s_root = NULL;
-		shrink_dcache_parent(root);
 		shrink_dcache_anon(&sb->s_anon);
+		shrink_dcache_parent(root);
 		dput(root);
 		fsync_super(sb);
 		lock_super(sb);
Index: linux-2.6/include/linux/fs.h
===================================================================
--- linux-2.6.orig/include/linux/fs.h
+++ linux-2.6/include/linux/fs.h
@@ -833,6 +833,9 @@ struct super_block {
 	struct list_head	s_instances;
 	struct quota_info	s_dquot;	/* Diskquota specific options */
 
+	int			s_prunes;
+	wait_queue_head_t	s_wait_prunes;
+
 	int			s_frozen;
 	wait_queue_head_t	s_wait_unfrozen;
 

  parent reply	other threads:[~2006-01-30 12:03 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-01-20 20:36 [PATCH] shrink_dcache_parent() races against shrink_dcache_memory() Jan Blunck
2006-01-23  5:22 ` Andrew Morton
2006-01-23  8:12   ` Kirill Korotaev
2006-01-23 15:13   ` Jan Blunck
2006-01-23  8:07 ` Kirill Korotaev
2006-01-23 15:57   ` Jan Blunck
2006-01-24  5:54     ` Balbir Singh
2006-01-24  9:48       ` Kirill Korotaev
2006-01-24 11:10         ` Balbir Singh
2006-01-24 17:18           ` Kirill Korotaev
2006-01-25  7:03             ` Balbir Singh
2006-01-30 12:03   ` Jan Blunck [this message]
2006-01-30 14:38     ` Balbir Singh
2006-01-30 14:54       ` Jan Blunck
2006-01-30 15:02         ` Kirill Korotaev
2006-01-30 15:25           ` Jan Blunck
2006-01-30 15:31             ` Kirill Korotaev
2006-01-30 14:42     ` Kirill Korotaev
2006-01-30 14:58       ` Jan Blunck
2006-01-30 15:59         ` Kirill Korotaev

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=20060130120318.GB9181@hasse.suse.de \
    --to=jblunck@suse.de \
    --cc=akpm@osdl.org \
    --cc=balbir@in.ibm.com \
    --cc=dev@sw.ru \
    --cc=linux-kernel@vger.kernel.org \
    --cc=olh@suse.de \
    --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).