Patchwork always set EXCLUSIVE flag for modifying e2fsck

login
register
mail settings
Submitter Andreas Dilger
Date Sept. 23, 2009, 9:56 a.m.
Message ID <20090923095640.GX10562@webber.adilger.int>
Download mbox | patch
Permalink /patch/34123/
State Superseded
Headers show

Comments

Andreas Dilger - Sept. 23, 2009, 9:56 a.m.
The checks done by e2fsck for mounted vs. read-only runs is confusing.

On the one hand, if e2fsck is NOT run with the "-n" flag (i.e. it might
modify the filesystem, there is no guarantee that it will open the
filesystem with the EXCLUSIVE flag (i.e. O_EXCL) to prevent the block
device from being used (in most cases = mounted).

On the other hand, if the filesystem IS mounted it also does NOT set
the EXCLUSIVE flag to prevent it from clobbering an in-use filesystem.
That seems like a bad choice also.

On the gripping hand, if e2fsck IS run with "-n" (i.e. read-only),
and the /etc/fstab or /proc/mounts does not report the same block
device to match the mountpoint (which happens for Lustre, and can
also happen if there is an overlay mount) then the e2fsck thinks
the filesystem is unmounted, but fails because the EXCLUSIVE flag
is set even though it is running read-only.

Change the logic here so that EXCLUSIVE is ALWAYS set when e2fsck
might modify the filesystem, regardless of whether the filesystem
is mounted or not.


Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Andreas Dilger - Feb. 23, 2010, 7:56 p.m.
The checks done by e2fsck for mounted vs. read-only runs is confusing.

On the one hand, if e2fsck is NOT run with the "-n" flag (i.e. it might
modify the filesystem), there is no guarantee that it will open the
filesystem with the EXCLUSIVE flag (i.e. O_EXCL) to prevent the block
device from being used later (i.e. mounted while e2fsck is running).

On the other hand, if e2fsck IS run with "-n" (i.e. read-only),
and the /etc/fstab or /proc/mounts does not report the same block
device to match the mountpoint (which happens for Lustre, and can
also happen if there is an overlay mount) then the e2fsck incorrectly
thinks the filesystem is unmounted, but fails because the EXCLUSIVE
flag is set even though it is running read-only.

Fix this so that "e2fsck -n" never sets EXCLUSIVE, and when run without
"-n" it will always set EXCLUSIVE unless the user answered "yes" to:

         /dev/sda1 is mounted.

         WARNING!!!  Running e2fsck on a mounted filesystem may cause
         SEVERE filesystem damage.

         Do you really want to continue (y/n)?


Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.

Patch

Index: e2fsprogs-1.41.6/e2fsck/unix.c
===================================================================
--- e2fsprogs-1.41.6.orig/e2fsck/unix.c
+++ e2fsprogs-1.41.6/e2fsck/unix.c
@@ -1230,9 +1230,7 @@  restart:
 		io_ptr = unix_io_manager;
 	flags = EXT2_FLAG_NOFREE_ON_ERROR;
        if ((ctx->options & E2F_OPT_READONLY) == 0)
-		flags |= EXT2_FLAG_RW;
-	if ((ctx->mount_flags & EXT2_MF_MOUNTED) == 0)
-		flags |= EXT2_FLAG_EXCLUSIVE;
+		flags |= EXT2_FLAG_RW | EXT2_FLAG_EXCLUSIVE;
 
 	retval = try_open_fs(ctx, flags, io_ptr, &fs);
 

If we want to assume that check_mount() will abort if the filesystem
is mounted, unless the user wants to shoot themselves in the foot, then 
the above patch could instead be modified to clear EXT2_FLAG_EXCLUSIVE
if the MF_MOUNTED flag IS set, per below.

Index: e2fsprogs-1.41.6/e2fsck/unix.c
===================================================================
--- e2fsprogs-1.41.6.orig/e2fsck/unix.c
+++ e2fsprogs-1.41.6/e2fsck/unix.c
@@ -1230,9 +1230,10 @@  restart:
 		io_ptr = unix_io_manager;
 	flags = EXT2_FLAG_NOFREE_ON_ERROR;
        if ((ctx->options & E2F_OPT_READONLY) == 0)
-		flags |= EXT2_FLAG_RW;
-	if ((ctx->mount_flags & EXT2_MF_MOUNTED) != 0)
-		flags |= EXT2_FLAG_EXCLUSIVE;
+		flags |= EXT2_FLAG_RW | EXT2_FLAG_EXCLUSIVE;
+	/* we will have aborted in check_mount() unless user asks for this */
+	if ((ctx->mount_flags & EXT2_MF_MOUNTED) != 0)
+		flags &= ~EXT2_FLAG_EXCLUSIVE;
 
 	retval = try_open_fs(ctx, flags, io_ptr, &fs);