Re: enforcing RLIMIT_NPROC in setuid() ?

[David Holland <dholland-security%netbsd.org@localhost>]

On Thu, Jan 10, 2008 at 04:23:47PM -0500, Christos Zoulas wrote:
 > The biggest problem I see with the change is that
 > a process that did not exceed the quota can be penalized about it.
 > Consider the case where a root daemon forks, runs setuid and sleeps
 > bringing the user above the NPROC resource limit. Then if a different
 > shell process tries to exec, it will fail.

One could mostly work around this by only checking at exec time in
processes that have been previously marked PK_SUGID (that covers
processes that shift down from root, right?) or are about to be.

-- 
David A. Holland
dholland%netbsd.org@localhost