delayed_work encodes the workqueue to use and the last CPU in delayed_work->work.data while it's on timer. The target CPU is implicitly recorded as the CPU the timer is queued on and delayed_work_timer_fn() queues delayed_work->work to the CPU it is running on. Unfortunately, this leaves flush_delayed_work[_sync]() no way to find out which CPU the delayed_work was queued for when they try to re-queue after killing the timer. Currently, it chooses the local CPU flush is running on. This can unexpectedly move a delayed_work queued on a specific CPU to another CPU and lead to subtle errors. There isn't much point in trying to save several bytes in struct delayed_work, which is already close to a hundred bytes on 64bit with all debug options turned off. This patch adds delayed_work->cpu to remember the CPU it's queued for. Note that if the timer is migrated during CPU down, the work item could be queued to the downed global_cwq after this change. As a detached global_cwq behaves like an unbound one, this doesn't change much for the delayed_work.
Showing the changes to kernel/workqueue.c (1 of 2 files). Show all
Unified · Split
| @@ -1319,7 +1319,7 @@ void delayed_work_timer_fn(unsigned long __data) · 1318 unmodified lines | ||
| 1319 | 1319 | struct cpu_workqueue_struct *cwq = get_work_cwq(&dwork->work); |
| 1320 | 1320 | |
| 1321 | 1321 | local_irq_disable(); |
| 1322 | − __queue_work(WORK_CPU_UNBOUND, cwq->wq, &dwork->work); | |
| 1322 | + __queue_work(dwork->cpu, cwq->wq, &dwork->work); | |
| 1323 | 1323 | local_irq_enable(); |
| 1324 | 1324 | } |
| 1325 | 1325 | EXPORT_SYMBOL_GPL(delayed_work_timer_fn); |
| @@ -1356,6 +1356,7 @@ static void __queue_delayed_work(int cpu, struct workqueue_struct *wq, · 30 unmodified lines | ||
| 1356 | 1356 | |
| 1357 | 1357 | set_work_cwq(work, get_cwq(lcpu, wq), 0); |
| 1358 | 1358 | |
| 1359 | + dwork->cpu = cpu; | |
| 1359 | 1360 | timer->expires = jiffies + delay; |
| 1360 | 1361 | |
| 1361 | 1362 | if (unlikely(cpu != WORK_CPU_UNBOUND)) |
| @@ -2997,7 +2998,7 @@ bool flush_delayed_work(struct delayed_work *dwork) · 1635 unmodified lines | ||
| 2997 | 2998 | { |
| 2998 | 2999 | local_irq_disable(); |
| 2999 | 3000 | if (del_timer_sync(&dwork->timer)) |
| 3000 | − __queue_work(WORK_CPU_UNBOUND, | |
| 3001 | + __queue_work(dwork->cpu, | |
| 3001 | 3002 | get_work_cwq(&dwork->work)->wq, &dwork->work); |
| 3002 | 3003 | local_irq_enable(); |
| 3003 | 3004 | return flush_work(&dwork->work); |
| @@ -3020,7 +3021,7 @@ bool flush_delayed_work_sync(struct delayed_work *dwork) · 16 unmodified lines | ||
| 3020 | 3021 | { |
| 3021 | 3022 | local_irq_disable(); |
| 3022 | 3023 | if (del_timer_sync(&dwork->timer)) |
| 3023 | − __queue_work(WORK_CPU_UNBOUND, | |
| 3024 | + __queue_work(dwork->cpu, | |
| 3024 | 3025 | get_work_cwq(&dwork->work)->wq, &dwork->work); |
| 3025 | 3026 | local_irq_enable(); |
| 3026 | 3027 | return flush_work_sync(&dwork->work); |