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] [day] [month] [year] [list]
Message-ID: <Y4UlL1UH1ks1iKOz@mit.edu>
Date:   Mon, 28 Nov 2022 16:16:31 -0500
From:   "Theodore Ts'o" <tytso@....edu>
To:     Tadeusz Struk <tadeusz.struk@...aro.org>
Cc:     Andreas Dilger <adilger.kernel@...ger.ca>,
        linux-ext4@...r.kernel.org, linux-kernel@...r.kernel.org,
        stable@...r.kernel.org,
        syzbot+a22dc4b0744ac658ed9b@...kaller.appspotmail.com
Subject: Re: [PATCH] ext4: Add extend check to prevent BUG() in ext4_es_end

On Tue, Oct 11, 2022 at 11:38:37AM -0700, Tadeusz Struk wrote:
> On 9/30/22 13:25, Tadeusz Struk wrote:
> > Syzbot reported an issue with ext4 extents. The reproducer creates
> > a corrupted ext4 fs image in memory, and mounts it as a loop device.
> > It invokes the ext4_cache_extents() and ext4_find_extent(), which
> > eventually triggers a BUG() in ext4_es_end() causing a kernel crash.
> > It triggers on mainline, and every kernel version back to v4.14.
> > Add a call ext4_ext_check_inode() in ext4_find_extent() to prevent
> > the crash.
> > 
> > To: "Theodore Ts'o"<tytso@....edu>
> > Cc: "Andreas Dilger"<adilger.kernel@...ger.ca>
> > Cc:<linux-ext4@...r.kernel.org>
> > Cc:<linux-kernel@...r.kernel.org>
> > Cc:<stable@...r.kernel.org>
> > 
> > Link:https://syzkaller.appspot.com/bug?id=641e7a4b900015c5d7a729d6cc1fba7a928a88f9
> > Reported-by:syzbot+a22dc4b0744ac658ed9b@...kaller.appspotmail.com
> > Signed-off-by: Tadeusz Struk<tadeusz.struk@...aro.org>
> 
> Any comments/feedback on this one?

Hi, I can't replicate the problem using the syzkaller repro:

root@...-xfstests:/vdc# /vtmp/repro 
[   14.840406] loop0: detected capacity change from 0 to 1051
[   14.840965] EXT4-fs (loop0): ext4_check_descriptors: Checksum for group 0 failed (60935!=0)
[   14.841468] EXT4-fs error (device loop0): ext4_ext_check_inode:520: inode #2: comm repro: pblk 0 bad header/extent: invalid magic - magic 9fe4, entries 42, max 0(0), depth 0(0)
[   14.842188] EXT4-fs (loop0): get root inode failed
[   14.842401] EXT4-fs (loop0): mount failed

And by inspection, if there is a corrupted inode, it should have been
caught much sooner, in ext4_iget():

	} else if (!ext4_has_inline_data(inode)) {
		/* validate the block references in the inode */
		if (!(EXT4_SB(sb)->s_mount_state & EXT4_FC_REPLAY) &&
			(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
			(S_ISLNK(inode->i_mode) &&
			!ext4_inode_is_fast_symlink(inode)))) {
			if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
				ret = ext4_ext_check_inode(inode);
			else
				ret = ext4_ind_check_inode(inode);
		}
	}
	if (ret)
		goto bad_inode;

... and this check has been around for quite some time.

It's much more efficient to check for a corrupted inode data structure
at the time when the inode is read into memory, as opposed to every
single time we try to lookup a logical->physical block map, in
ext4_find_extent().

If you can reproduce a failure on a modern kernel, please let me know,
but it looks like syzkaller had only reported it on androd-54,
android-5-10, linux-4.14, and linux-4.19 in any case.

Cheers,

						- Ted

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ