When __queue_delayed_work() is called, it chooses a cpu for handling the timer interrupt. As of today, it will pick either the cpu passed as parameter or the last cpu used for this. This is not good if a system does use CPU isolation, because it can take away some valuable cpu time to: 1 - deal with the timer interrupt, 2 - schedule-out the desired task, 3 - queue work on a random workqueue, and 4 - schedule the desired task back to the cpu. So to fix this, during __queue_delayed_work(), if cpu isolation is in place, pick a random non-isolated cpu to handle the timer interrupt. As an optimization, if the current cpu is not isolated, use it instead of looking for another candidate.
Showing the changes to kernel/workqueue.c (1 of 1 file). Show all
Unified · Split
| @@ -2362,10 +2362,18 @@ static void __queue_delayed_work(int cpu, struct workqueue_struct *wq, · 2361 unmodified lines | ||
| 2362 | 2362 | dwork->cpu = cpu; |
| 2363 | 2363 | timer->expires = jiffies + delay; |
| 2364 | 2364 | |
| 2365 | − if (unlikely(cpu != WORK_CPU_UNBOUND)) | |
| 2365 | + if (housekeeping_enabled(HK_TYPE_TIMER)) { | |
| 2366 | + /* If the current cpu is a housekeeping cpu, use it. */ | |
| 2367 | + cpu = smp_processor_id(); | |
| 2368 | + if (!housekeeping_test_cpu(cpu, HK_TYPE_TIMER)) | |
| 2369 | + cpu = housekeeping_any_cpu(HK_TYPE_TIMER); | |
| 2366 | 2370 | add_timer_on(timer, cpu); |
| 2367 | − else | |
| 2368 | − add_timer(timer); | |
| 2371 | + } else { | |
| 2372 | + if (likely(cpu == WORK_CPU_UNBOUND)) | |
| 2373 | + add_timer(timer); | |
| 2374 | + else | |
| 2375 | + add_timer_on(timer, cpu); | |
| 2376 | + } | |
| 2369 | 2377 | } |
| 2370 | 2378 | |
| 2371 | 2379 | /** |