Received-SPF: pass (google.com: domain of linux-kernel+bounces-171434-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Message-ID: <ccd9b28f-8e5e-4dac-920f-746439c50d6a@kernel.dk>
Date: Tue, 7 May 2024 07:28:43 -0600
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH] io_uring/io-wq: Use set_bit() and test_bit() at
 worker->flags
To: Breno Leitao <leitao@debian.org>
Cc: Pavel Begunkov <asml.silence@gmail.com>, leit@meta.com,
 "open list:IO_URING" <io-uring@vger.kernel.org>,
 open list <linux-kernel@vger.kernel.org>
References: <20240503173711.2211911-1-leitao@debian.org>
 <d05aa530-f0f5-4ec2-91ae-b193ae644395@kernel.dk> <ZjoGJH1CEk+f+U7n@gmail.com>
 <ZjoKM7ro0wDqsdWP@gmail.com>
Content-Language: en-US
From: Jens Axboe <axboe@kernel.dk>
In-Reply-To: <ZjoKM7ro0wDqsdWP@gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit

On 5/7/24 5:02 AM, Breno Leitao wrote:
> On Tue, May 07, 2024 at 03:44:54AM -0700, Breno Leitao wrote:
>> Since we are now using WRITE_ONCE() in io_wq_worker, I am wondering if
>> this is what we want to do?
>>
>> 	WRITE_ONCE(worker->flags, (IO_WORKER_F_UP| IO_WORKER_F_RUNNING) << 1);
> 
> In fact, we can't clear flags here, so, more correct approach will be:
> 
> 	WRITE_ONCE(worker->flags, READ_ONCE(worker->flags) | (IO_WORKER_F_UP | IO_WORKER_F_RUNNING) << 1);
> 
> Does it sound reasonable?

Either that, or since we aren't just assigning the startup bits, maybe
just use set_mask_bits() like Christophe suggested and not worry about
it?

-- 
Jens Axboe