Commit 1265057fa02c7bed3b6d9ddc8a2048065a370364

Authored by Tejun Heo
1 parent 41f63c5359

workqueue: fix CPU binding of flush_delayed_work[_sync]()

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.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>

Showing 2 changed files with 5 additions and 3 deletions Side-by-side Diff

include/linux/workqueue.h
... ... @@ -102,6 +102,7 @@
102 102 struct delayed_work {
103 103 struct work_struct work;
104 104 struct timer_list timer;
  105 + int cpu;
105 106 };
106 107  
107 108 static inline struct delayed_work *to_delayed_work(struct work_struct *work)
... ... @@ -1319,7 +1319,7 @@
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 @@
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 @@
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 @@
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);