It is said in the document that the timer which is being deleted by del_timer_sync() should not be restarted: Synchronization rules: Callers must prevent restarting of the timer, otherwise this function is meaningless.
Repeating timer may cause the del_timer_sync() spin longer, or even spin forever in very very very very extreme condition. It is not considered a bug for me, but it disobeys the document. To fix it, we need: 1) don't requeue the mayday timer when !need_to_create_worker() it is a preparation/cleaup step for the fix. Otherwise the timer will repeat forever if we only do the 2). 2) remove the del_timer_sync()s in maybe_create_worker() Fix the problem. It is not enough if we only do the 1), need_to_create_worker() would be probably true when the time del_timer_sync() is called, the timer is still repeating. Signed-off-by: Lai Jiangshan <la...@cn.fujitsu.com> --- kernel/workqueue.c | 6 ++---- 1 files changed, 2 insertions(+), 4 deletions(-) diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 35974ac..593b9bc 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -1866,12 +1866,12 @@ static void pool_mayday_timeout(unsigned long __pool) */ list_for_each_entry(work, &pool->worklist, entry) send_mayday(work); + + mod_timer(&pool->mayday_timer, jiffies + MAYDAY_INTERVAL); } spin_unlock(&pool->lock); spin_unlock_irq(&wq_mayday_lock); - - mod_timer(&pool->mayday_timer, jiffies + MAYDAY_INTERVAL); } /** @@ -1913,7 +1913,6 @@ restart: worker = create_worker(pool); if (worker) { - del_timer_sync(&pool->mayday_timer); spin_lock_irq(&pool->lock); start_worker(worker); if (WARN_ON_ONCE(need_to_create_worker(pool))) @@ -1931,7 +1930,6 @@ restart: break; } - del_timer_sync(&pool->mayday_timer); spin_lock_irq(&pool->lock); if (need_to_create_worker(pool)) goto restart; -- 1.7.7.6 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/