Messages in this thread Patch in this message | | | Date | Sat, 17 May 2008 19:14:22 +0400 | From | Oleg Nesterov <> | Subject | [PATCH 3/3] posix timers: use SIGQUEUE_CANCELLED when the timer is destroyed |
| |
This change goes as a separate patch for documentation purposes.
Suggested by Linus Torvalds.
Fixes the problem pointed out by Austin Clements. Currently, when the task execs it could be killed by the fatal signal sent by the posix timer, because exec flushes the signal handlers. See http://bugzilla.kernel.org/show_bug.cgi?id=10460
This is a user visible change. With this patch sys_timer_delete() discards the pending signal which was generated by the timer.
This change goes as a separate patch for documentation purposes. We have many options how to set SIGQUEUE_CANCELLED while detroying the timer. We could set this flag in release_posix_timer() before calling sigqueue_free(), or add the new "int cancel" argument to sigqueue_free(), but since sigqueue_free() plays with q->flags anyway and nobody else uses this function, this patch changes sigqueue_free() to set SIGQUEUE_CANCELLED unconditionally.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
--- 25/kernel/signal.c~6_USE_CANCELLED 2008-05-17 17:40:09.000000000 +0400 +++ 25/kernel/signal.c 2008-05-17 18:07:10.000000000 +0400 @@ -1246,7 +1246,7 @@ void sigqueue_free(struct sigqueue *q) * __exit_signal()->flush_sigqueue(). */ spin_lock_irqsave(lock, flags); - q->flags &= ~SIGQUEUE_PREALLOC; + q->flags = SIGQUEUE_CANCELLED; /* clears SIGQUEUE_PREALLOC */ /* * If it is queued it will be freed when dequeued, * like the "regular" sigqueue.
| |