linux-kernel - Re: [RFC] vfs: skip extra attributes check on removal for symlinks

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: <20180511024234.GA30522@ZenIV.linux.org.uk>
Date:   Fri, 11 May 2018 03:42:35 +0100
From:   Al Viro <viro@...IV.linux.org.uk>
To:     "Luis R. Rodriguez" <mcgrof@...nel.org>
Cc:     linux-fsdevel@...r.kernel.org, linux-api@...r.kernel.org,
        sandeen@...deen.net, darrick.wong@...cle.com, dhowells@...hat.com,
        tytso@....edu, fliu@...e.com, jack@...e.cz, jeffm@...e.com,
        nborisov@...e.com, jake.norris@...e.com, mtk.manpages@...il.com,
        linux-xfs@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC] vfs: skip extra attributes check on removal for symlinks

On Thu, May 10, 2018 at 11:05:51PM +0000, Luis R. Rodriguez wrote:
> On Thu, May 10, 2018 at 09:48:07PM +0100, Al Viro wrote:
> > On Thu, Apr 26, 2018 at 04:46:39PM -0700, Luis R. Rodriguez wrote:
> > 
> > > Since we cannot set these attributes we should special-case the
> > > immutable/append on delete for symlinks, this would be consistent with
> > > what we *do* allow on Linux for all filesystems.
> > 
> > Er...  So why not simply sanity-check it in places that set it on
> > inodes? 
> 
> The patch is not about sanity-checks on setters though as *that* is in place
> already. Its about the case where the filesystem gets corrupted and the VFS
> *still* does process these attributes for symlinks and still prevents
> deletion because of these attributes.

... and this corrupted fs ends up setting those flags on in-core inodes.
Which is where we ought to block that.  Seriously, let's make sure that
->i_flags manipulations are done by inode_set_flags() (e.g. btrfs open-codes
that, apparently) and let's make _that_ check and reject those.