From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-2.3 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_1 autolearn=no autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id E6202C38BE0 for ; Mon, 24 Feb 2020 20:19:01 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id C4F8321556 for ; Mon, 24 Feb 2020 20:19:01 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727483AbgBXUTB (ORCPT ); Mon, 24 Feb 2020 15:19:01 -0500 Received: from verein.lst.de ([213.95.11.211]:40160 "EHLO verein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727219AbgBXUTB (ORCPT ); Mon, 24 Feb 2020 15:19:01 -0500 Received: by verein.lst.de (Postfix, from userid 2407) id A13D968B05; Mon, 24 Feb 2020 21:18:58 +0100 (CET) Date: Mon, 24 Feb 2020 21:18:58 +0100 From: Christoph Hellwig To: Damien Le Moal Cc: Christoph Hellwig , Naohiro Aota , "linux-fsdevel@vger.kernel.org" Subject: Re: [PATCH] zonefs: fix IOCB_NOWAIT handling Message-ID: <20200224201858.GA10880@lst.de> References: <20200221143723.482323-1-hch@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.17 (2007-11-01) Sender: linux-fsdevel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-fsdevel@vger.kernel.org On Mon, Feb 24, 2020 at 01:48:48PM +0000, Damien Le Moal wrote: > > The main problem with allowing IOCB_NOWAIT is that for an application that sends > multiple write AIOs with a single io_submit(), all write AIOs after one that is > failed with -EAGAIN will be failed with -EINVAL (not -EIO !) since they will > have an unaligned write offset. I am wondering if that is really a problem at > all since the application can catch the -EAGAIN and -EINVAL, indicating that the > write stream was stopped. So maybe it is reasonable to simply allow IOCB_NOWAIT ? I don't think supporting IOCB_NOWAIT with current zonefs is very useful. It will be very useful once Zone Append is supported. But more importantly my patch fixes a bug in the current zonefs implementation. We need to fix that before 5.6 is released. Any additional functionaity can still come later if we think that it is useful.