[RFC PATCH] workqueue: Unbind workers before sending them to exit()

[Valentin Schneider <vschneid@redhat.com>]

It has been reported that isolated CPUs can suffer from interference due to
per-CPU kworkers waking up just to die.

A surge of workqueue activity (sleeping workfn's exacerbate this) during
initial setup can cause extra per-CPU kworkers to be spawned. Then, a
latency-sensitive task can be running merrily on an isolated CPU only to be
interrupted sometime later by a kworker marked for death (cf.
IDLE_WORKER_TIMEOUT, 5 minutes after last kworker activity).

Affine kworkers to the wq_unbound_cpumask (which doesn't contain isolated
CPUs, cf. HK_TYPE_WQ) before waking them up after marking them with
WORKER_DIE.

This follows the logic of CPU hot-unplug, which has been packaged into
helpers for the occasion.

Signed-off-by: Valentin Schneider <vschneid@redhat.com>
---
 kernel/workqueue.c | 35 ++++++++++++++++++++++++++---------
 1 file changed, 26 insertions(+), 9 deletions(-)

diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 1ea50f6be843..0f1a25ea4924 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -1972,6 +1972,18 @@ static struct worker *create_worker(struct worker_pool *pool)
 	return NULL;
 }
 
+static void unbind_worker(struct worker *worker)
+{
+	kthread_set_per_cpu(worker->task, -1);
+	WARN_ON_ONCE(set_cpus_allowed_ptr(worker->task, wq_unbound_cpumask) < 0);
+}
+
+static void rebind_worker(struct worker *worker, struct worker_pool *pool)
+{
+	kthread_set_per_cpu(worker->task, pool->cpu);
+	WARN_ON_ONCE(set_cpus_allowed_ptr(worker->task, pool->attrs->cpumask) < 0);
+}
+
 /**
  * destroy_worker - destroy a workqueue worker
  * @worker: worker to be destroyed
@@ -1999,6 +2011,16 @@ static void destroy_worker(struct worker *worker)
 
 	list_del_init(&worker->entry);
 	worker->flags |= WORKER_DIE;
+
+	/*
+	 * We're sending that thread off to die, so any CPU would do. This is
+	 * especially relevant for pcpu kworkers affined to an isolated CPU:
+	 * we'd rather not interrupt an isolated CPU just for a kworker to
+	 * do_exit().
+	 */
+	if (!(worker->flags & WORKER_UNBOUND))
+		unbind_worker(worker);
+
 	wake_up_process(worker->task);
 }
 
@@ -4999,10 +5021,8 @@ static void unbind_workers(int cpu)
 
 		raw_spin_unlock_irq(&pool->lock);
 
-		for_each_pool_worker(worker, pool) {
-			kthread_set_per_cpu(worker->task, -1);
-			WARN_ON_ONCE(set_cpus_allowed_ptr(worker->task, wq_unbound_cpumask) < 0);
-		}
+		for_each_pool_worker(worker, pool)
+			unbind_worker(worker);
 
 		mutex_unlock(&wq_pool_attach_mutex);
 	}
@@ -5027,11 +5047,8 @@ static void rebind_workers(struct worker_pool *pool)
 	 * of all workers first and then clear UNBOUND.  As we're called
 	 * from CPU_ONLINE, the following shouldn't fail.
 	 */
-	for_each_pool_worker(worker, pool) {
-		kthread_set_per_cpu(worker->task, pool->cpu);
-		WARN_ON_ONCE(set_cpus_allowed_ptr(worker->task,
-						  pool->attrs->cpumask) < 0);
-	}
+	for_each_pool_worker(worker, pool)
+		rebind_worker(worker, pool);
 
 	raw_spin_lock_irq(&pool->lock);
 
-- 
2.31.1