Received-SPF: pass (google.com: domain of linux-kernel+bounces-94353-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Message-ID: <ae0475d7-fcfa-4d6f-9bb0-03479e6bf83b@joelfernandes.org>
Date: Wed, 6 Mar 2024 12:57:25 -0500
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH 2/2] rcu: Allocate WQ with WQ_MEM_RECLAIM bit set
Content-Language: en-US
To: Uladzislau Rezki <urezki@gmail.com>, Z qiang <qiang.zhang1211@gmail.com>
Cc: "Paul E . McKenney" <paulmck@kernel.org>, RCU <rcu@vger.kernel.org>,
 Neeraj upadhyay <Neeraj.Upadhyay@amd.com>, Boqun Feng
 <boqun.feng@gmail.com>, Hillf Danton <hdanton@sina.com>,
 LKML <linux-kernel@vger.kernel.org>,
 Oleksiy Avramchenko <oleksiy.avramchenko@sony.com>,
 Frederic Weisbecker <frederic@kernel.org>
References: <20240305195720.42687-1-urezki@gmail.com>
 <20240305195720.42687-2-urezki@gmail.com>
 <CALm+0cWiOfKR=Gci0dj=z4gT4vSbZ=ZzMfo+CxLZCFQzL1bjfQ@mail.gmail.com>
 <ZehZ7Ef3DW2mT9fc@pc636>
From: Joel Fernandes <joel@joelfernandes.org>
In-Reply-To: <ZehZ7Ef3DW2mT9fc@pc636>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit


On 3/6/2024 6:56 AM, Uladzislau Rezki wrote:
> On Wed, Mar 06, 2024 at 10:15:44AM +0800, Z qiang wrote:
>>>
>>> synchronize_rcu() users have to be processed regardless
>>> of memory pressure so our private WQ needs to have at least
>>> one execution context what WQ_MEM_RECLAIM flag guarantees.
>>>
>>> Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
>>> ---
>>>  kernel/rcu/tree.c | 6 +++++-
>>>  1 file changed, 5 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
>>> index 475647620b12..59881a68dd26 100644
>>> --- a/kernel/rcu/tree.c
>>> +++ b/kernel/rcu/tree.c
>>> @@ -1581,6 +1581,7 @@ static void rcu_sr_put_wait_head(struct llist_node *node)
>>>  /* Disabled by default. */
>>>  static int rcu_normal_wake_from_gp;
>>>  module_param(rcu_normal_wake_from_gp, int, 0644);
>>> +static struct workqueue_struct *sync_wq;
>>>
>>>  static void rcu_sr_normal_complete(struct llist_node *node)
>>>  {
>>> @@ -1679,7 +1680,7 @@ static void rcu_sr_normal_gp_cleanup(void)
>>>          * of outstanding users(if still left) and releasing wait-heads
>>>          * added by rcu_sr_normal_gp_init() call.
>>>          */
>>> -       queue_work(system_highpri_wq, &rcu_state.srs_cleanup_work);
>>> +       queue_work(sync_wq, &rcu_state.srs_cleanup_work);
>>>  }
>>>
>>>  /*
>>> @@ -5584,6 +5585,9 @@ void __init rcu_init(void)
>>>         rcu_gp_wq = alloc_workqueue("rcu_gp", WQ_MEM_RECLAIM, 0);
>>>         WARN_ON(!rcu_gp_wq);
>>>
>>> +       sync_wq = alloc_workqueue("sync_wq", WQ_MEM_RECLAIM, 0);
>>
>> Why was WQ_HIGHPRI removed?
>>
> I would like to check perf. figures with it and send out it as a
> separate patch if it is worth it.

I guess one thing to note is that there are also other RCU-related WQ which have
WQ_MEM_RECLAIM but not WQ_HIGHPRI (such as for expedited RCU, at least some
configs). So for consistency, this makes sense to me.

Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org).

thanks,

 - Joel