[PATCH] 2.5.68 fs/ext3/super.c fix for orphan recovery error path

[PATCH] 2.5.68 fs/ext3/super.c fix for orphan recovery error path

[Ernie Petrides]

Stephen/Andrew, please consider applying this patch to reverse the order
of checks in ext3_orphan_cleanup() for read-only mounts and prior errors.

The problem resolved by this patch is that if a root file system has an
error recorded from a previous mount, and then (when rebooting) the orphan
recovery procedure is initiated, the recovery is correctly skipped but the
file system is incorrectly left in a writable state.

This causes the subsequent fsck to fail due to the root file system
being dirty, and then requires manual intervention to get the system
fully booted.

Thanks in advance.  -ernie



diff -urpN linux-2.5.68/fs/ext3/super.c{.orig,}
--- linux-2.5.68/fs/ext3/super.c.orig	2003-04-19 22:50:47.000000000 -0400
+++ linux-2.5.68/fs/ext3/super.c	2003-04-21 15:18:08.000000000 -0400
@@ -982,12 +982,6 @@ static void ext3_orphan_cleanup (struct 
 		return;
 	}
 
-	if (s_flags & MS_RDONLY) {
-		printk(KERN_INFO "EXT3-fs: %s: orphan cleanup on readonly fs\n",
-		       sb->s_id);
-		sb->s_flags &= ~MS_RDONLY;
-	}
-
 	if (EXT3_SB(sb)->s_mount_state & EXT3_ERROR_FS) {
 		if (es->s_last_orphan)
 			jbd_debug(1, "Errors on filesystem, "
@@ -997,6 +991,12 @@ static void ext3_orphan_cleanup (struct 
 		return;
 	}
 
+	if (s_flags & MS_RDONLY) {
+		printk(KERN_INFO "EXT3-fs: %s: orphan cleanup on readonly fs\n",
+		       sb->s_id);
+		sb->s_flags &= ~MS_RDONLY;
+	}
+
 	while (es->s_last_orphan) {
 		struct inode *inode;
 

Re: [PATCH] 2.5.68 fs/ext3/super.c fix for orphan recovery error path

[Andrew Morton]

Ernie Petrides <petrides@redhat.com> wrote:
>
> Stephen/Andrew, please consider applying this patch to reverse the order
> of checks in ext3_orphan_cleanup() for read-only mounts and prior errors.
> 

Thanks, that is definitely needed.  We should do this in 2.4 as well.

Re: [PATCH] 2.5.68 fs/ext3/super.c fix for orphan recovery error path

[Stephen C. Tweedie]

Hi,

On Mon, 2003-04-21 at 23:08, Andrew Morton wrote:
> Ernie Petrides <petrides@redhat.com> wrote:
> >
> > Stephen/Andrew, please consider applying this patch to reverse the order
> > of checks in ext3_orphan_cleanup() for read-only mounts and prior errors.
> > 
> 
> Thanks, that is definitely needed.  We should do this in 2.4 as well.

Yes, I'll push it to marcelo.

--Stephen