Received: from eggs.gnu.org ([209.51.188.92]:41896) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1goZzA-0007P0-G8 for qemu-devel@nongnu.org; Tue, 29 Jan 2019 15:27:57 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1goZz9-00024B-MV for qemu-devel@nongnu.org; Tue, 29 Jan 2019 15:27:56 -0500 References: <20190121134249.16615-1-david@redhat.com> <20190121134249.16615-2-david@redhat.com> <7f7d8582-b6d2-b120-c768-8cc8350b416c@linux.ibm.com> <86377327-ab6d-df3a-b914-688206273bd5@redhat.com> <14055713-cd8e-7fc5-d819-3724838662a3@linux.ibm.com> From: David Hildenbrand Message-ID: <376e6094-eacf-782a-ebfe-42838d235c78@redhat.com> Date: Tue, 29 Jan 2019 21:27:51 +0100 MIME-Version: 1.0 In-Reply-To: <14055713-cd8e-7fc5-d819-3724838662a3@linux.ibm.com> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH v3 1/2] s390x/pci: Introduce unplug requests and split unplug handler List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: pmorel@linux.ibm.com, qemu-devel@nongnu.org Cc: Thomas Huth , Cornelia Huck , Collin Walling , Christian Borntraeger , qemu-s390x@nongnu.org, Richard Henderson On 29.01.19 17:54, Pierre Morel wrote: > On 29/01/2019 16:14, David Hildenbrand wrote: >> On 29.01.19 14:31, Pierre Morel wrote: >>> On 21/01/2019 14:42, David Hildenbrand wrote: >>>> PCI on s390x is really weird and how it was modeled in QEMU might no= t have >>>> been the right choice. Anyhow, right now it is the case that: >>>> - Hotplugging a PCI device will silently create a zPCI device >>>> (if none is provided) >>>> - Hotunplugging a zPCI device will unplug the PCI device (if any) >>>> - Hotunplugging a PCI device will unplug also the zPCI device >>>> As far as I can see, we can no longer change this behavior. But we >>>> should fix it. >=20 >=20 >>> So unplugging PCI first will deny the guest any possibility to smooth= ly >>> relinquish a device. >>> >>> >>> Is it possible the other way around? >> >> Maybe, but it does not really matter. We unplug both devices >> synchronously, without the guest recognizing the order. We always have >> the unplug request first that notifies the guest. When we get an ACK >> from the guest, we can unpplug both devices in any order. >> >> (and if we want to change the order, we should do it in a separate >> patch, this patch does not change the order, just refactors the code) >=20 >=20 > If it is done atomically, then I have no objection. Yes, this is atomically. It is two separate steps, but only logically. The guest cannot observe it. Thanks! >=20 > Regards, > Pierre >=20 >=20 >=20 --=20 Thanks, David / dhildenb