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 Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id A17CEC43334 for ; Tue, 12 Jul 2022 06:32:52 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229889AbiGLGcv (ORCPT ); Tue, 12 Jul 2022 02:32:51 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:51696 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229515AbiGLGcu (ORCPT ); Tue, 12 Jul 2022 02:32:50 -0400 Received: from verein.lst.de (verein.lst.de [213.95.11.211]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 569D3DE9C; Mon, 11 Jul 2022 23:32:49 -0700 (PDT) Received: by verein.lst.de (Postfix, from userid 2407) id 6A00468AA6; Tue, 12 Jul 2022 08:32:45 +0200 (CEST) Date: Tue, 12 Jul 2022 08:32:45 +0200 From: Christoph Hellwig To: Kanchan Joshi Cc: hch@lst.de, sagi@grimberg.me, kbusch@kernel.org, axboe@kernel.dk, io-uring@vger.kernel.org, linux-nvme@lists.infradead.org, linux-block@vger.kernel.org, asml.silence@gmail.com, joshiiitr@gmail.com, anuj20.g@samsung.com, gost.dev@samsung.com Subject: Re: [PATCH for-next 2/4] nvme: compact nvme_uring_cmd_pdu struct Message-ID: <20220712063245.GA5908@lst.de> References: <20220711110155.649153-1-joshi.k@samsung.com> <20220711110155.649153-3-joshi.k@samsung.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20220711110155.649153-3-joshi.k@samsung.com> User-Agent: Mutt/1.5.17 (2007-11-01) Precedence: bulk List-ID: X-Mailing-List: linux-block@vger.kernel.org On Mon, Jul 11, 2022 at 04:31:53PM +0530, Kanchan Joshi wrote: > From: Anuj Gupta > > Mark this packed so that we can create bit more space in its container > i.e. io_uring_cmd. This is in preparation to support multipathing on > uring-passthrough. > Move its definition to nvme.h as well. I do not like this. packed structures that contain pointers are inherently dangerous as that will get us into unaligned accesses very quickly. I also do not think we should expose it any more widely than absolutely required. In fact if possible I'd really like to figure out how we can remove this pdu concept entirely an just have a small number of well typed field directly in the uring cmd. This will involved some rework of the passthrough I/O completions so that we can get at the metadata biovecs and integrity data.