Suggested by Pavel Emelyanov.
send_sigqueue/send_group_sigqueue are only differ in how they lock ->siglock.
Unify them. send_group_sigqueue() uses spin_lock() because it knows the task
can't exit, but in that case lock_task_sighand() can't fail and doesn't hurt.
Note that the "sig" argument is ignored, it is always equal to ->si_signo.
Signed-off-by: Pavel Emelyanov <[email protected]>
Signed-off-by: Oleg Nesterov <[email protected]>
--- 25/kernel/signal.c~5_SSQ_UNIFY 2008-03-08 16:54:57.000000000 +0300
+++ 25/kernel/signal.c 2008-03-08 17:15:59.000000000 +0300
@@ -1240,14 +1240,27 @@ void sigqueue_free(struct sigqueue *q)
__sigqueue_free(q);
}
-static int do_send_sigqueue(int sig, struct sigqueue *q, struct task_struct *t,
+static int do_send_sigqueue(struct sigqueue *q, struct task_struct *t,
int group)
{
+ int sig = q->info.si_signo;
struct sigpending *pending;
+ unsigned long flags;
+ int ret;
BUG_ON(!(q->flags & SIGQUEUE_PREALLOC));
+
+ ret = -1;
+ if (!likely(lock_task_sighand(t, &flags)))
+ goto ret;
+
handle_stop_signal(sig, t);
+ ret = 1;
+ if (sig_ignored(t, sig))
+ goto out;
+
+ ret = 0;
if (unlikely(!list_empty(&q->list))) {
/*
* If an SI_TIMER entry is already queue just increment
@@ -1256,58 +1269,29 @@ static int do_send_sigqueue(int sig, str
BUG_ON(q->info.si_code != SI_TIMER);
q->info.si_overrun++;
- return 0;
+ goto out;
}
- if (sig_ignored(t, sig))
- return 1;
-
signalfd_notify(t, sig);
pending = group ? &t->signal->shared_pending : &t->pending;
list_add_tail(&q->list, &pending->list);
sigaddset(&pending->signal, sig);
complete_signal(sig, t, group);
-
- return 0;
+out:
+ unlock_task_sighand(t, &flags);
+ret:
+ return ret;
}
int send_sigqueue(int sig, struct sigqueue *q, struct task_struct *p)
{
- unsigned long flags;
- int ret = -1;
-
- /*
- * The rcu based delayed sighand destroy makes it possible to
- * run this without tasklist lock held. The task struct itself
- * cannot go away as create_timer did get_task_struct().
- *
- * We return -1, when the task is marked exiting, so
- * posix_timer_event can redirect it to the group leader
- */
- if (!likely(lock_task_sighand(p, &flags)))
- goto out_err;
-
- ret = do_send_sigqueue(sig, q, p, 0);
-
- unlock_task_sighand(p, &flags);
-out_err:
- return ret;
+ return do_send_sigqueue(q, p, 0);
}
int
send_group_sigqueue(int sig, struct sigqueue *q, struct task_struct *p)
{
- unsigned long flags;
- int ret;
-
- /* Since it_lock is held, p->sighand cannot be NULL. */
- spin_lock_irqsave(&p->sighand->siglock, flags);
-
- ret = do_send_sigqueue(sig, q, p, 1);
-
- spin_unlock_irqrestore(&p->sighand->siglock, flags);
-
- return ret;
+ return do_send_sigqueue(q, p, 1);
}
/*
Looks fine. But I'd like a comment on the "ret = 1" case making
explicit what that magic return case is for and why the value
matters (to posix_timer_event).
Thanks,
Roland
On 03/10, Roland McGrath wrote:
>
> Looks fine. But I'd like a comment on the "ret = 1" case making
> explicit what that magic return case is for and why the value
> matters (to posix_timer_event).
Well, currently this helper is solely used by posix_timer_event(), but despite
the fact it has BUG_ON(!SI_TIMER) it is quite generic, there were patches which
used it for aio.
So I don't think we should document the interaction with posix timers, but OK,
I'll send a small comment update.
Oleg.
> So I don't think we should document the interaction with posix timers, but OK,
> I'll send a small comment update.
I didn't say the documentation has to be misleading about the general
utility of the function. I said it has to explain the subtlety of the
function's calling convention.
Thanks,
Roland