linux-ext4 - Re: EXT4-fs (mmcblk0p1): couldn't mount as ext3 due to feature incompatibilities

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20180306133803.lhgyvhwonveekcfc@ws.net.home>
Date:   Tue, 6 Mar 2018 14:38:03 +0100
From:   Karel Zak <kzak@...hat.com>
To:     Lukas Czerner <lczerner@...hat.com>
Cc:     Joakim Tjernlund <Joakim.Tjernlund@...inera.com>,
        "tytso@....edu" <tytso@....edu>,
        "sandeen@...hat.com" <sandeen@...hat.com>,
        "linux-ext4@...r.kernel.org" <linux-ext4@...r.kernel.org>
Subject: Re: EXT4-fs (mmcblk0p1): couldn't mount as ext3 due to feature
 incompatibilities

On Tue, Mar 06, 2018 at 02:11:10PM +0100, Lukas Czerner wrote:
> > > > Seems like it only hits when mounting from fstab, I have
> > > > 
> > > > LABEL=BOOT            /boot           auto            noatime         1 2
> > > > 
> > > > Anyhow, Eric's patch solved the problem.
> > > 
> > > BTW which distro was this happening on?
> > 
> > This is on Gentoo
> > 
> >  Jocke
> 
> It's good that Eric's patch fixes the complainig, however the question
> is why is it trying the other fs when you have specified auto and blkid
> is able to recognize it. Weird.
> 
> Do you know how the mount is called from your init script ?
> 
> Can you reproduce it by running mount -a (probably using a different
> file system since it's /boot we're talking about here).
> 
> If so then please can you use these debug environment variables and show
> us the output ?
> 
> LIBMOUNT_DEBUG=all LIBBLKID_DEBUG=all mount -a

I guess the problem is "mount -a -t <pattern>", and it's libmount bug.

The option -t is used in two ways by mount(8), for -a it's pattern
to filter fstab, and for standard mount operation it FS type.

Unfortunately the current -a implementation does not zeroize the
pattern when try to mount the selected FS and "auto" is not translated
to FS type (because -t <type> is expected). 

And there is fallback code in libmount to use /{proc,etc}/filesystems if FS is 
unknown... in this file ext3 is before ext4. That's the story :-)

I'll fix it.

Anyway, it would be nice to have LIBMOUNT_DEBUG=all (or =0xffff for
old versions) output to be sure it's really the -a issue.

You can try to reproduce:

 # echo 'LABEL="TESTBABY" /mnt/test auto noatime 0 0' >> /etc/fstab
 # mount -av -t nosomething

    Karel

-- 
 Karel Zak  <kzak@...hat.com>
 http://karelzak.blogspot.com