Re: RLIMIT_NPROC check in set_user()

[Linus Torvalds <torvalds@linux-foundation.org>]

On Wed, Jul 6, 2011 at 10:36 AM, Vasiliy Kulikov <segoon@openwall.com> wrote:
> On Sun, Jun 12, 2011 at 17:09 +0400, Vasiliy Kulikov wrote:
>> I'd be happy to hear opinions about improving the fixes above or
>> alternative fixes.
>
> No comments?  Even "Sigh, what a nasty problem.  But we cannot really
> fix it without significantly breaking the stuff.  Go and drink something." ?

Thanks for reminding me.

My reaction is: "let's just remote the crazy check from set_user()
entirely". If somebody has credentials to change users, they damn well
have credentials to override the RLIMIT_NPROC too, and as you say,
failure is likely a bigger security threat than success.

The whole point of RLIMIT_NPROC is to avoid fork-bombs. If we go over
the limit for some other reason that is controlled by the super-user,
who cares?

So let's keep it in kernel/fork.c where we actually create a *new*
process (and where everybody knows exactly what the limit means, and
people who don't check for error cases are just broken). And remove it
from everywhere else.

Hmm?

                     Linus