linux-kernel - Re: patch to make Linux capabilities into something useful (v 0.3.1)

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 8 Sep 2006 12:52:39 +0200
From:	Pavel Machek <pavel@....cz>
To:	David Madore <david.madore@....fr>
Cc:	Linux Kernel mailing-list <linux-kernel@...r.kernel.org>
Subject: Re: patch to make Linux capabilities into something useful (v 0.3.1)

Hi!

> > You contradict yourself.
> 
> I don't see how that is.  I understand that you could be unconvinced
> by my reasoning and by my arguments, but I don't see how they are
> contradictory.

Well, you claim it is as safe as possible, and it is not quite. 

> The bottom line is that, whereas for root making syscalls fail (or,
> worse, in the case of setuid(), behave subtly diffently) is a radical
> change, for non-root it is something which should always be expected
> (fork() can fail for lack of resources, write() can fail for quota
> exhaution, etc.), and not something an attacker should be able to
> exploit.

I can bet someone will get the fork() case wrong:

f = fork();
kill(f);

fork will return -1, and kill will kill _all_ the processes.

> >			   Yes, you are decreasing security of suid
> > non-root programs, and yes, someone will take advantage of that. Plus,
> > you can easily do away without this risk.
> 
> I wish I could offer more assurance, but unfortunately the solutions
> which do away with the risk come with a great cost:
> 
> > Just add all "usual" capabilities when execing
> > suid/sgid-anything.
> 
> This makes it trivial to regain capabilities: just create a program
> suid yourself and exec it.  OK, we can say that "yourself" won't work,
> but you still only need to find another uid to hijack...  Not too

If you can find another uid to hijack, that other uid has bad
problems. And I do not think you'll commonly find another uid to
hijack.

And there are easier ways to get out of jail with your proposed
capabilities: you do not restrict ptrace, so you can just ptrace any
other process with same uid, and hijack it.

(You probably want to introduce CAP_REG_PTRACE).

Or just remove CAP_REG_XUID_EXEC when removing any other CAP_REG...?

> >		      Alternatively disallow suid/sgid-anything exec
> > when all "usual" capabilities are not present.
> 
> This is probably too stringent: remove any trivial capability
> whatsoever and you lose a rather important ability.

It is not too bad; you'll usually not want restricted programs to exec
anything setuid... (Do you have example where
restricted-but-should-be-able-to-setuid-exec makes sense?)
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/