(In reply to Ryan Crews from comment #0)
> Description of problem: The mkfs.ext4 utility can create a filesystem with a
> revision number that breaks support in e2fsprogs utilities such as dumpe2fs
> and e2fsck. The filesystem can be mounted fine, but eventually an automatic
> fsck will fail (if enabled).
> 
> Version-Release number of selected component (if applicable):
>  e2fsprogs-1.41.12-18.el6.x86_64
>  2.6.32-431.11.2.el6.x86_64
> 
> How reproducible: 100%
> 
> Steps to Reproduce:
> 1. Add storage. (dd if=/dev/zero of=~/test.fs bs=1M count=100)
> 2. Make filesystem with non 0/1 revision number (mkfs.ext4 -r 999 ~/test.fs)

Ok, but - why on earth would you do that?

there are all sorts of administrative missteps that can be taken; this looks like one of them.

<looks at case ...>

Oh, I see, they were trying to set reserved space.  Hohum.

So, there is a test in the kernel at mount time:

        if (le32_to_cpu(es->s_rev_level) > EXT4_MAX_SUPP_REV) {
                ext4_msg(sb, KERN_ERR, "revision level too high, "
                         "forcing read-only mode");
                res = MS_RDONLY;
        }

but it's not working, apparently.

Agreed that userspace shouldn't be able to set a level higher than it knows about, I'll send a patch upstream, and see why the kernel test is failing.

Thanks,
-Eric

Well, the kernel test is working to some degree:

EXT4-fs (sdc): revision level too high, forcing read-only mode

but it's ... not actually making the fs read-only.

Ah:

commit 7e84b6216467b84cd332c8e567bf5aa113fd2f38
Author: Eric Sandeen <sandeen>
Date:   Mon May 28 14:17:25 2012 -0400

    ext4: force ro mount if ext4_setup_super() fails
    
    If ext4_setup_super() fails i.e. due to a too-high revision,
    the error is logged in dmesg but the fs is not mounted RO as
    indicated.

Which is only present in kernels v3.5 and higher.

-Eric

[PATCH] mke2fs: don't accept too-high revision levels

It's a bit strange to accept revision levels higher than
the code creating the filesystem can understand, so don't
allow it.

At least the kernel will mount the fs readonly if it's too
high, but no other utility will touch it, so you can't
fix the error.

Just reject anything > EXT2_MAX_SUPP_REV at mkfs time.

Signed-off-by: Eric Sandeen <sandeen>
---

diff --git a/misc/mke2fs.c b/misc/mke2fs.c
index aecd5d5..82019dc 100644
--- a/misc/mke2fs.c
+++ b/misc/mke2fs.c
@@ -1679,9 +1679,10 @@ profile_error:
 			break;
 		case 'r':
 			r_opt = strtoul(optarg, &tmp, 0);
-			if (*tmp) {
+			if (*tmp || r_opt > EXT2_MAX_SUPP_REV ) {
 				com_err(program_name, 0,
-					_("bad revision level - %s"), optarg);
+					_("bad revision level - %s (max %d)"),
+					   optarg, EXT2_MAX_SUPP_REV);
 				exit(1);
 			}
 			fs_param.s_rev_level = r_opt;


Setting request for RHEL6.6, no reason to wait on a fix, IMHO.

Created attachment 891628 [details]
patch to limit filesystem revision number

Frank, I took the liberty of sending your patch as a V2 for mine; yours is cleaner, thanks!  (It was worth the extra 20 mins) :)

-Eric

Built in e2fsprogs-1.41.12-19.el6

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHBA-2014-1566.html