linux-kernel - Re: Question on io-wq

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8dd4bc4c-9d8e-fb5a-6931-3e861ad9b4bf@kernel.dk>
Date:   Thu, 22 Oct 2020 08:08:09 -0600
From:   Jens Axboe <axboe@...nel.dk>
To:     "Zhang,Qiang" <qiang.zhang@...driver.com>
Cc:     viro@...iv.linux.org.uk, io-uring@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org
Subject: Re: Question on io-wq

On 10/22/20 3:02 AM, Zhang,Qiang wrote:
> 
> Hi Jens Axboe
> 
> There are some problem in 'io_wqe_worker' thread, when the 
> 'io_wqe_worker' be create and  Setting the affinity of CPUs in NUMA 
> nodes, due to CPU hotplug, When the last CPU going down, the 
> 'io_wqe_worker' thread will run anywhere. when the CPU in the node goes 
> online again, we should restore their cpu bindings?

Something like the below should help in ensuring affinities are
always correct - trigger an affinity set for an online CPU event. We
should not need to do it for offlining. Can you test it?


diff --git a/fs/io-wq.c b/fs/io-wq.c
index 4012ff541b7b..3bf029d1170e 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -19,6 +19,7 @@
 #include <linux/task_work.h>
 #include <linux/blk-cgroup.h>
 #include <linux/audit.h>
+#include <linux/cpu.h>
 
 #include "io-wq.h"
 
@@ -123,9 +124,13 @@ struct io_wq {
 	refcount_t refs;
 	struct completion done;
 
+	struct hlist_node cpuhp_node;
+
 	refcount_t use_refs;
 };
 
+static enum cpuhp_state io_wq_online;
+
 static bool io_worker_get(struct io_worker *worker)
 {
 	return refcount_inc_not_zero(&worker->ref);
@@ -1096,6 +1101,13 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
 		return ERR_PTR(-ENOMEM);
 	}
 
+	ret = cpuhp_state_add_instance_nocalls(io_wq_online, &wq->cpuhp_node);
+	if (ret) {
+		kfree(wq->wqes);
+		kfree(wq);
+		return ERR_PTR(ret);
+	}
+
 	wq->free_work = data->free_work;
 	wq->do_work = data->do_work;
 
@@ -1145,6 +1157,7 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
 	ret = PTR_ERR(wq->manager);
 	complete(&wq->done);
 err:
+	cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
 	for_each_node(node)
 		kfree(wq->wqes[node]);
 	kfree(wq->wqes);
@@ -1164,6 +1177,8 @@ static void __io_wq_destroy(struct io_wq *wq)
 {
 	int node;
 
+	cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
+
 	set_bit(IO_WQ_BIT_EXIT, &wq->state);
 	if (wq->manager)
 		kthread_stop(wq->manager);
@@ -1191,3 +1206,40 @@ struct task_struct *io_wq_get_task(struct io_wq *wq)
 {
 	return wq->manager;
 }
+
+static bool io_wq_worker_affinity(struct io_worker *worker, void *data)
+{
+	struct task_struct *task = worker->task;
+	unsigned long flags;
+
+	raw_spin_lock_irqsave(&task->pi_lock, flags);
+	do_set_cpus_allowed(task, cpumask_of_node(worker->wqe->node));
+	task->flags |= PF_NO_SETAFFINITY;
+	raw_spin_unlock_irqrestore(&task->pi_lock, flags);
+	return false;
+}
+
+static int io_wq_cpu_online(unsigned int cpu, struct hlist_node *node)
+{
+	struct io_wq *wq = hlist_entry_safe(node, struct io_wq, cpuhp_node);
+	int i;
+
+	rcu_read_lock();
+	for_each_node(i)
+		io_wq_for_each_worker(wq->wqes[i], io_wq_worker_affinity, NULL);
+	rcu_read_unlock();
+	return 0;
+}
+
+static __init int io_wq_init(void)
+{
+	int ret;
+
+	ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, "io-wq/online",
+					io_wq_cpu_online, NULL);
+	if (ret < 0)
+		return ret;
+	io_wq_online = ret;
+	return 0;
+}
+subsys_initcall(io_wq_init);

-- 
Jens Axboe