linux-kernel - Re: [PATCH] fs: Prevent non-symlinks from entering pick link

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: <20250618052747.GQ1880847@ZenIV>
Date: Wed, 18 Jun 2025 06:27:47 +0100
From: Al Viro <viro@...iv.linux.org.uk>
To: Edward Adam Davis <eadavis@...com>
Cc: syzbot+1aa90f0eb1fc3e77d969@...kaller.appspotmail.com,
	almaz.alexandrovich@...agon-software.com, brauner@...nel.org,
	jack@...e.cz, linux-fsdevel@...r.kernel.org,
	linux-kernel@...r.kernel.org, ntfs3@...ts.linux.dev,
	syzkaller-bugs@...glegroups.com
Subject: Re: [PATCH] fs: Prevent non-symlinks from entering pick link

On Wed, Jun 18, 2025 at 06:02:00AM +0100, Al Viro wrote:
> On Wed, Jun 18, 2025 at 05:50:16AM +0100, Al Viro wrote:
> 
> > NAK.  This is not the first time that garbage is suggested and no,
> > we are not going to paper over that shite in fs/namei.c.
> > 
> > Not going to happen.
> > 
> > You ARE NOT ALLOWED to call make_bad_inode() on a live inode, period.
> > Never, ever to be done.
> > 
> > There's a lot of assertions it violates and there's no chance in
> > hell to plaster each with that kind of checks.
> > 
> > Fix NTFS.  End of story.
> 
> To elaborate a bit: if you look at the end of e.g. their attr_set_size(),
> you'll see
> out:
>         if (is_bad) {
> bad_inode:
> 		_ntfs_bad_inode(&ni->vfs_inode);
> 	}
> 	return err;
> }
> 
> This is a bug.  So are similar places all over the place there.
> You are not supposed to use make_bad_inode() as a general-purpose
> "something went wrong, don't wanna see it anymore" tool.
> 
> And as long as it stays there, any fuzzing reports of ntfs are pretty
> much worthless - any of those places (easily located by grepping for
> _ntfs_bad_inode) can fuck the kernel up.  Once ntfs folks get around
> to saner error recovery, it would make sense to start looking into
> fuzzing that thing again.  Until then - nope.  Again, this is *NOT*
> going to be papered over in a random set of places (pretty certain
> to remain incomplete) in VFS.

Note that anything that calls __d_add(dentry, inode) with is_bad_inode(inode)
(or d_add(), or d_instantiate(), or d_splice_alias() under the same conditions)
is also FUBAR.

So's anything that calls make_bad_inode() on a struct inode that might be
in process of being passed to one of those functions by another thread.

This is fundamentally wrong; bad inodes are not supposed to end up attached
to dentries.