Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
Message-ID: <a41a1f47-ad05-3245-8ac8-7d8e95ebde44@kernel.dk>
Date:   Sat, 7 May 2022 09:23:59 -0600
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux aarch64; rv:91.0) Gecko/20100101
 Thunderbird/91.8.1
Subject: Re: [PATCH 4/4] io_uring: implement multishot mode for accept
Content-Language: en-US
To:     Hao Xu <haoxu.linux@gmail.com>, io-uring@vger.kernel.org
Cc:     Pavel Begunkov <asml.silence@gmail.com>,
        linux-kernel@vger.kernel.org
References: <20220507152009.87367-1-haoxu.linux@gmail.com>
From:   Jens Axboe <axboe@kernel.dk>
In-Reply-To: <20220507152009.87367-1-haoxu.linux@gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 5/7/22 9:20 AM, Hao Xu wrote:
> @@ -5757,8 +5771,26 @@ static int io_accept(struct io_kiocb *req, unsigned int issue_flags)
>  		ret = io_install_fixed_file(req, file, issue_flags,
>  					    accept->file_slot - 1);
>  	}
> -	__io_req_complete(req, issue_flags, ret, 0);
> -	return 0;
> +
> +	if (!(req->flags & REQ_F_APOLL_MULTISHOT)) {
> +		__io_req_complete(req, issue_flags, ret, 0);
> +		return 0;
> +	}
> +	if (ret >= 0) {
> +		bool filled;
> +
> +		spin_lock(&ctx->completion_lock);
> +		filled = io_fill_cqe_aux(ctx, req->cqe.user_data, ret,
> +					 IORING_CQE_F_MORE);
> +		io_commit_cqring(ctx);
> +		spin_unlock(&ctx->completion_lock);
> +		if (!filled)
> +			return -ECANCELED;
> +		io_cqring_ev_posted(ctx);
> +		goto retry;
> +	}
> +
> +	return ret;

I'd still make that:

	if (filled) {
		io_cqring_ev_posted(ctx);
		goto retry;
	}
	ret = -ECANCELED;

as it flows better and shows what the likely outcome is.

>  static int io_connect_prep_async(struct io_kiocb *req)
> diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
> index f4d9ca62a5a6..7c3d70d12428 100644
> --- a/include/uapi/linux/io_uring.h
> +++ b/include/uapi/linux/io_uring.h
> @@ -224,9 +224,9 @@ enum {
>  #define IORING_RECVSEND_POLL_FIRST	(1U << 0)
>  
>  /*
> - * accept flags stored in accept_flags
> + * accept flags stored in sqe->ioprio
>   */
> -#define IORING_ACCEPT_MULTISHOT	(1U << 15)
> +#define IORING_ACCEPT_MULTISHOT	(1U << 0)
>  
>  /*
>   * IO completion data structure (Completion Queue Entry)

This hunk needs to get folded in to the patch that adds the flag in the
first place.

-- 
Jens Axboe