linux-kernel - Re: [PATCH 04/24] VFS: Add LSM hooks for filesystem context [ver #7]

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: <1bb08eac-517a-a07e-5e64-5771cc0c0afc@tycho.nsa.gov>
Date:   Wed, 25 Apr 2018 10:07:15 -0400
From:   Stephen Smalley <sds@...ho.nsa.gov>
To:     David Howells <dhowells@...hat.com>
Cc:     Paul Moore <paul@...l-moore.com>, linux-nfs@...r.kernel.org,
        linux-kernel@...r.kernel.org,
        linux-security-module@...r.kernel.org, viro@...iv.linux.org.uk,
        selinux@...ho.nsa.gov, linux-fsdevel@...r.kernel.org,
        linux-afs@...ts.infradead.org
Subject: Re: [PATCH 04/24] VFS: Add LSM hooks for filesystem context [ver #7]

On 04/24/2018 11:22 AM, David Howells wrote:
> Stephen Smalley <sds@...ho.nsa.gov> wrote:
> 
>> Neither fsopen() nor fscontext_fs_write() appear to perform any kind of
>> up-front permission checking (DAC or MAC), although some security hooks may
>> be ultimately called to allocate structures, parse security options, etc.
>> Is there a reason not apply a may_mount() or similar check up front?
> 
> may_mount() is called by fsmount() at the moment.  It may make sense to move
> this earlier to fsopen().  Note that there's also going to be something that
> looks like:
> 
> 	fd = fspick("/mnt");
> 	fsmount(fd, "/a", MNT_NOEXEC); // ie. bind mount
> 
> or:
> 
> 	fd = fspick("/mnt");
> 	write(fd, "o intr");
> 	write(fd, "x reconfigure"); // ie. something like remount
> 	close(fd);
> 
> I guess we'd want to call may_mount() in fspick() too.  But there's also the
> possibility of using this to create a query interfact too:
> 
> 	fd = fspick("/mnt");
> 	write(fd, "q intr");
> 	read(fd, value_buffer);

My concern was that fsopen()/fscontext_fs_write() may expose attack surface (e.g. mount option parsing code) that might not be normally accessible to unprivileged userspace (i.e. gated by may_mount() and security_sb_mount()) prior to your changes.