2008-03-07 10:38:24

by Oleg Nesterov

[permalink] [raw]
Subject: [PATCH] signals: send_sig_info: don't take tasklist_lock

The comment in send_sig_info() is wrong, tasklist_lock can't help.

The caller must ensure the task can't go away, otherwise ->sighand can be NULL
even before we take the lock.

p->sighand could be changed by exec(), but I can't imagine how it is possible
to prevent exit(), but not exec().

Since the things seem to work, I assume all callers are correct. However,
drm_vbl_send_signals() looks broken. block_all_signals() which is solely used
by drm is definitely broken.

Signed-off-by: Oleg Nesterov <[email protected]>

--- 25/kernel/signal.c~3_SSI_NO_TASKLIST 2008-03-07 12:20:34.000000000 +0300
+++ 25/kernel/signal.c 2008-03-07 13:06:17.000000000 +0300
@@ -1138,8 +1138,7 @@ static int kill_something_info(int sig,
*/

/*
- * These two are the most common entry points. They send a signal
- * just to the specific thread.
+ * The caller must ensure the task can't exit.
*/
int
send_sig_info(int sig, struct siginfo *info, struct task_struct *p)
@@ -1154,17 +1153,9 @@ send_sig_info(int sig, struct siginfo *i
if (!valid_signal(sig))
return -EINVAL;

- /*
- * We need the tasklist lock even for the specific
- * thread case (when we don't need to follow the group
- * lists) in order to avoid races with "p->sighand"
- * going away or changing from under us.
- */
- read_lock(&tasklist_lock);
spin_lock_irqsave(&p->sighand->siglock, flags);
ret = specific_send_sig_info(sig, info, p);
spin_unlock_irqrestore(&p->sighand->siglock, flags);
- read_unlock(&tasklist_lock);
return ret;
}


2008-03-07 19:37:13

by Roland McGrath

[permalink] [raw]
Subject: Re: [PATCH] signals: send_sig_info: don't take tasklist_lock

Looks like a good change to me.

Thanks,
Roland