In task_schedule(), before attempting to set the new task expiration
date, make sure it is not running by trying to set the TASK_RUNNING
flag, and waiting if it is already there. Having the flag set will
ensure that the task won't be running while we're modifying it.
There is a very rare race condition, where the expire would be set by
task_schedule(), then the running task might set it to something else,
and if it sets it to TICK_ETERNITY before task_schedule() calls
__task_queue(), then we will hit a BUG_ON() there.
This is very hard to reproduce, but has been reported a few times,
included in Github issue #3327, which should now be fixed.
This should be backported as far back as 2.8.
WIP: Make sure the task is not running before changing expire
#ifdef USE_THREAD
if (task->tid < 0) {
+ int was_running;
+ /*
+ * Make sure the task is not already running before changing
+ * its expire, otherwise it could overwrite our modification
+ */
+ if (task == th_ctx->current)
+ was_running = 1;
+ else {
+ was_running = 0;
+ while (HA_ATOMIC_FETCH_OR(&task->state, TASK_RUNNING) & TASK_RUNNING)
+ __ha_cpu_relax();
+ }
+
/* FIXME: is it really needed to lock the WQ during the check ? */
HA_RWLOCK_WRLOCK(TASK_WQ_LOCK, &wq_lock);
if (task_in_wq(task))
when = tick_first(when, task->expire);
task->expire = when;
+ if (!was_running)
+ task_drop_running(task, 0);
if (!task_in_wq(task) || tick_is_lt(task->expire, task->wq.key)) {
if (likely(caller)) {
caller = HA_ATOMIC_XCHG(&task->caller, caller);