(on top of signals-cleanup-security_task_kill-usage-implementation.patch)
This wasn't documented, but as Atsushi Tsuji <a-tsuji@bk.jp.nec.com> pointed
out check_kill_permission() needs tasklist_lock for task_session_nr().
I missed this fact when removed tasklist from the callers.
Change check_kill_permission() to take tasklist_lock for the SIGCONT case.
Re-order security checks so that we take tasklist_lock only if/when it is
actually needed. This is a minimal fix for now, tasklist will be removed
later.
Also change the code to use task_session() instead of task_session_nr().
Also, remove the SIGCONT check from cap_task_kill(), it is bogus (and the
whole function is bogus. Serge, Eric, why it is still alive?).
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
--- 25/kernel/signal.c~CKP_TAKE_TASKLIST 2008-03-18 14:47:00.000000000 +0300
+++ 25/kernel/signal.c 2008-03-18 17:25:19.000000000 +0300
@@ -533,6 +533,7 @@ static int rm_from_queue(unsigned long m
static int check_kill_permission(int sig, struct siginfo *info,
struct task_struct *t)
{
+ struct pid *sid;
int error;
if (!valid_signal(sig))
@@ -545,11 +546,24 @@ static int check_kill_permission(int sig
if (error)
return error;
- if (((sig != SIGCONT) || (task_session_nr(current) != task_session_nr(t)))
- && (current->euid ^ t->suid) && (current->euid ^ t->uid)
- && (current->uid ^ t->suid) && (current->uid ^ t->uid)
- && !capable(CAP_KILL))
- return -EPERM;
+ if ((current->euid ^ t->suid) && (current->euid ^ t->uid) &&
+ (current->uid ^ t->suid) && (current->uid ^ t->uid) &&
+ !capable(CAP_KILL)) {
+ switch (sig) {
+ case SIGCONT:
+ read_lock(&tasklist_lock);
+ sid = task_session(t);
+ read_unlock(&tasklist_lock);
+ /*
+ * We don't return the error if sid == NULL. The
+ * task was unhashed, the caller must notice this.
+ */
+ if (!sid || sid == task_session(current))
+ break;
+ default:
+ return -EPERM;
+ }
+ }
return ...
