If a struct sigevent with SIGEV_SIGNAL is passed to timer_create(2),

check the signal number to be in the allowed range. An invalid
signal number could crash the kernel by overflowing the sigset_t
array.
More checks would be good, and SIGEV_THREAD shouldn't be dropped
silently, but this fixes at least the local DOS vulnerability.
This commit is contained in:
drochner 2009-12-10 12:39:12 +00:00
parent fb022d0910
commit a1a04dd1be
1 changed files with 6 additions and 3 deletions

View File

@ -1,4 +1,4 @@
/* $NetBSD: kern_time.c,v 1.162 2009/10/03 20:48:42 elad Exp $ */
/* $NetBSD: kern_time.c,v 1.163 2009/12/10 12:39:12 drochner Exp $ */
/*-
* Copyright (c) 2000, 2004, 2005, 2007, 2008, 2009 The NetBSD Foundation, Inc.
@ -61,7 +61,7 @@
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: kern_time.c,v 1.162 2009/10/03 20:48:42 elad Exp $");
__KERNEL_RCSID(0, "$NetBSD: kern_time.c,v 1.163 2009/12/10 12:39:12 drochner Exp $");
#include <sys/param.h>
#include <sys/resourcevar.h>
@ -531,7 +531,10 @@ timer_create1(timer_t *tid, clockid_t id, struct sigevent *evp,
if (((error =
(*fetch_event)(evp, &pt->pt_ev, sizeof(pt->pt_ev))) != 0) ||
((pt->pt_ev.sigev_notify < SIGEV_NONE) ||
(pt->pt_ev.sigev_notify > SIGEV_SA))) {
(pt->pt_ev.sigev_notify > SIGEV_SA)) ||
(pt->pt_ev.sigev_notify == SIGEV_SIGNAL &&
(pt->pt_ev.sigev_signo <= 0 ||
pt->pt_ev.sigev_signo >= NSIG))) {
pool_put(&ptimer_pool, pt);
return (error ? error : EINVAL);
}