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=-2.2 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 2CCDEC43603 for ; Mon, 9 Dec 2019 14:09:48 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 0783A20726 for ; Mon, 9 Dec 2019 14:09:48 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727875AbfLIOJr (ORCPT ); Mon, 9 Dec 2019 09:09:47 -0500 Received: from mx2.suse.de ([195.135.220.15]:40530 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1726687AbfLIOJq (ORCPT ); Mon, 9 Dec 2019 09:09:46 -0500 X-Virus-Scanned: by amavisd-new at test-mx.suse.de Received: from relay2.suse.de (unknown [195.135.220.254]) by mx1.suse.de (Postfix) with ESMTP id E54F3ACD0; Mon, 9 Dec 2019 14:09:44 +0000 (UTC) Subject: Re: [Xen-devel] [PATCH 2/4] xenbus: limit when state is forced to closed To: "Durrant, Paul" , =?UTF-8?Q?Roger_Pau_Monn=c3=a9?= Cc: "linux-kernel@vger.kernel.org" , "xen-devel@lists.xenproject.org" , Stefano Stabellini , Boris Ostrovsky References: <20191205140123.3817-1-pdurrant@amazon.com> <20191205140123.3817-3-pdurrant@amazon.com> <20191209113926.GS980@Air-de-Roger> <3412e42d13224b6786613e58dc189ebf@EX13D32EUC003.ant.amazon.com> <8d66e520-3009-cde1-e24c-26d7476e5873@suse.com> <63d653a04207451e9041c89acd04f2a2@EX13D32EUC003.ant.amazon.com> <2cd2a27e-2bb8-bd9d-45d8-1710038fb97a@suse.com> From: =?UTF-8?B?SsO8cmdlbiBHcm/Dnw==?= Message-ID: <1c4c9b3e-31a5-d8b3-01de-3ad84db6390a@suse.com> Date: Mon, 9 Dec 2019 15:09:44 +0100 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.2.1 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 09.12.19 15:06, Durrant, Paul wrote: >> -----Original Message----- >> From: Jürgen Groß >> Sent: 09 December 2019 13:39 >> To: Durrant, Paul ; Roger Pau Monné >> >> Cc: linux-kernel@vger.kernel.org; xen-devel@lists.xenproject.org; Stefano >> Stabellini ; Boris Ostrovsky >> >> Subject: Re: [Xen-devel] [PATCH 2/4] xenbus: limit when state is forced to >> closed >> >> On 09.12.19 13:19, Durrant, Paul wrote: >>>> -----Original Message----- >>>> From: Jürgen Groß >>>> Sent: 09 December 2019 12:09 >>>> To: Durrant, Paul ; Roger Pau Monné >>>> >>>> Cc: linux-kernel@vger.kernel.org; xen-devel@lists.xenproject.org; >> Stefano >>>> Stabellini ; Boris Ostrovsky >>>> >>>> Subject: Re: [Xen-devel] [PATCH 2/4] xenbus: limit when state is forced >> to >>>> closed >>>> >>>> On 09.12.19 13:03, Durrant, Paul wrote: >>>>>> -----Original Message----- >>>>>> From: Jürgen Groß >>>>>> Sent: 09 December 2019 11:55 >>>>>> To: Roger Pau Monné ; Durrant, Paul >>>>>> >>>>>> Cc: linux-kernel@vger.kernel.org; xen-devel@lists.xenproject.org; >>>> Stefano >>>>>> Stabellini ; Boris Ostrovsky >>>>>> >>>>>> Subject: Re: [Xen-devel] [PATCH 2/4] xenbus: limit when state is >> forced >>>> to >>>>>> closed >>>>>> >>>>>> On 09.12.19 12:39, Roger Pau Monné wrote: >>>>>>> On Thu, Dec 05, 2019 at 02:01:21PM +0000, Paul Durrant wrote: >>>>>>>> Only force state to closed in the case when the toolstack may need >> to >>>>>>>> clean up. This can be detected by checking whether the state in >>>>>> xenstore >>>>>>>> has been set to closing prior to device removal. >>>>>>> >>>>>>> I'm not sure I see the point of this, I would expect that a failure >> to >>>>>>> probe or the removal of the device would leave the xenbus state as >>>>>>> closed, which is consistent with the actual driver state. >>>>>>> >>>>>>> Can you explain what's the benefit of leaving a device without a >>>>>>> driver in such unknown state? >>>>>> >>>>>> And more concerning: did you check that no frontend/backend is >>>>>> relying on the closed state to be visible without closing having been >>>>>> set before? >>>>> >>>>> Blkfront doesn't seem to mind and I believe the Windows PV drivers >> cope, >>>> but I don't really understand the comment since this patch is actually >>>> removing a case where the backend transitions directly to closed. >>>> >>>> I'm not speaking of blkfront/blkback only, but of net, tpm, scsi, >> pvcall >>>> etc. frontends/backends. After all you are modifying a function common >>>> to all PV driver pairs. >>>> >>>> You are removing a state switc to "closed" in case the state was _not_ >>>> "closing" before. >>> >>> Yes, which AFAIK is against the intention of the generic PV protocol >> such that it ever existed anyway. >> >> While this might be the case we should _not_ break any guests >> running now. So this kind of reasoning is dangerous. >> >>> >>>> So any PV driver reacting to "closed" of the other end >>>> in case the previous state might not have been "closing" before is at >>>> risk to misbehave with your patch. >>> >>> Well, they will see nothing now. If the state was not closing, it gets >> left alone, so the frontend shouldn't do anything. The only risk that I >> can see is that some frontend/backend pair needed a direct 4 -> 6 >> transition to support 'unbind' before but AFAIK nothing has ever supported >> that, and blk and net crash'n'burn if you try that on upstream as it >> stands. A clean unplug would always set state to 5 first, since that's >> part of the unplug protocol. >> >> That was my question: are you sure all current and previous >> guest frontends and backends are handling unplug this way? >> >> Not "should handle", but "do handle". > > That depends on the toolstack. IIUC the only 'supported' toolstack is xl/libxl, which will set 'state' to 5 and 'online' to 0 to initiate an unplug. I guess libvirt/libxl is doing the same? At least at SUSE we still have some customers running xend based Xen installations with recent Linux or Windows guests. Juergen