kernel: rerun task_work while freezing in get_signal() for-6.11/io_uring-20240711
authorPavel Begunkov <asml.silence@gmail.com>
Wed, 10 Jul 2024 17:58:18 +0000 (18:58 +0100)
committerJens Axboe <axboe@kernel.dk>
Thu, 11 Jul 2024 07:51:44 +0000 (01:51 -0600)
io_uring can asynchronously add a task_work while the task is getting
freezed. TIF_NOTIFY_SIGNAL will prevent the task from sleeping in
do_freezer_trap(), and since the get_signal()'s relock loop doesn't
retry task_work, the task will spin there not being able to sleep
until the freezing is cancelled / the task is killed / etc.

Run task_works in the freezer path. Keep the patch small and simple
so it can be easily back ported, but we might need to do some cleaning
after and look if there are other places with similar problems.

Cc: stable@vger.kernel.org
Link: https://github.com/systemd/systemd/issues/33626
Fixes: 12db8b690010c ("entry: Add support for TIF_NOTIFY_SIGNAL")
Reported-by: Julian Orth <ju.orth@gmail.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/89ed3a52933370deaaf61a0a620a6ac91f1e754d.1720634146.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
kernel/signal.c

index 1f9dd41c04be2202309acbd00c051651e739daf1..60c737e423a18827b537333ad9b954c8ca4c93d4 100644 (file)
@@ -2600,6 +2600,14 @@ static void do_freezer_trap(void)
        spin_unlock_irq(&current->sighand->siglock);
        cgroup_enter_frozen();
        schedule();
+
+       /*
+        * We could've been woken by task_work, run it to clear
+        * TIF_NOTIFY_SIGNAL. The caller will retry if necessary.
+        */
+       clear_notify_signal();
+       if (unlikely(task_work_pending(current)))
+               task_work_run();
 }
 
 static int ptrace_signal(int signr, kernel_siginfo_t *info, enum pid_type type)