From patchwork Sat Sep 10 08:23:37 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maxime Coquelin X-Patchwork-Id: 9324781 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 9AB8C60780 for ; Sat, 10 Sep 2016 08:24:19 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 8B16D29710 for ; Sat, 10 Sep 2016 08:24:19 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 7F24029B01; Sat, 10 Sep 2016 08:24:19 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.9 required=2.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from lists.gnu.org (lists.gnu.org [208.118.235.17]) (using TLSv1 with cipher AES256-SHA (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id 9D00D29710 for ; Sat, 10 Sep 2016 08:24:18 +0000 (UTC) Received: from localhost ([::1]:33607 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bidaD-0001nD-PD for patchwork-qemu-devel@patchwork.kernel.org; Sat, 10 Sep 2016 04:24:17 -0400 Received: from eggs.gnu.org ([2001:4830:134:3::10]:36109) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bidZj-0001kD-CQ for qemu-devel@nongnu.org; Sat, 10 Sep 2016 04:23:49 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bidZf-0003Mb-30 for qemu-devel@nongnu.org; Sat, 10 Sep 2016 04:23:46 -0400 Received: from mx1.redhat.com ([209.132.183.28]:54038) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bidZe-0003MV-Qs; Sat, 10 Sep 2016 04:23:43 -0400 Received: from int-mx13.intmail.prod.int.phx2.redhat.com (int-mx13.intmail.prod.int.phx2.redhat.com [10.5.11.26]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 75F194E4C3; Sat, 10 Sep 2016 08:23:42 +0000 (UTC) Received: from max-t460s.redhat.com (vpn1-7-222.ams2.redhat.com [10.36.7.222]) by int-mx13.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id u8A8Ndat003565; Sat, 10 Sep 2016 04:23:40 -0400 From: Maxime Coquelin To: mst@redhat.com, qemu-devel@nongnu.org, cornelia.huck@de.ibm.com Date: Sat, 10 Sep 2016 10:23:37 +0200 Message-Id: <1473495817-29952-1-git-send-email-maxime.coquelin@redhat.com> X-Scanned-By: MIMEDefang 2.68 on 10.5.11.26 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.38]); Sat, 10 Sep 2016 08:23:42 +0000 (UTC) X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.2.x-3.x [generic] X-Received-From: 209.132.183.28 Subject: [Qemu-devel] [PATCH] virtio-bus: Plug devices after features are negotiated X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: marcel@redhat.com, vkaplans@redhat.com, Maxime Coquelin , qemu-stable@nongnu.org Errors-To: qemu-devel-bounces+patchwork-qemu-devel=patchwork.kernel.org@nongnu.org Sender: "Qemu-devel" X-Virus-Scanned: ClamAV using ClamSMTP Currently, devices are plugged before features are negotiated. If the backend doesn't support VIRTIO_F_VERSION_1, the transport need to rewind some settings. This is the case for CCW, for which a post_plugged callback had been introduced, where max_rev field is just updated if VIRTIO_F_VERSION_1 is not supported by the backend. For PCI, implementing the post_plugged would be much more complicated, so it needs to know whether the backend supports VIRTIO_F_VERSION_1 at plug time. Currently, nothing is done for PCI. Modern capabilitities get exposed to the guest even if VIRTIO_F_VERSION_1 is not supported by the backend, which confuses the guest. This patch proposes to replace existing post_plugged solution with an approach that fits with both transports. Features negociation is performed before ->device_plugged() call. A pre_plugged callback is introduced so that the transports can set their supported features. Cc: Cornelia Huck Cc: Marcel Apfelbaum Cc: qemu-stable@nongnu.org Acked-by: Michael S. Tsirkin Signed-off-by: Maxime Coquelin Reviewed-by: Marcel Apfelbaum Tested-by: Cornelia Huck [ccw] Reviewed-by: Cornelia Huck --- Hi, This patch replaces series "virtio-pci: Improve device plugging whith legacy backends", as fixes the problem in a cleaner and more generic way. Goal is to have it also in stable tree. Michael, I added your ack, as the changes compared to the RFC are minor: - Rebased on top of your pci branch - Improve error hanling when modern no supported and legacy disabled by the user I ran check-qtest, and tested PCI with vhost-user-bridge with and without VERSION_1 enabled. What is missing is testing CCW, Cornelia, can you handle that? Thanks, Maxime --- hw/s390x/virtio-ccw.c | 30 +++++++++++++++--------------- hw/virtio/virtio-bus.c | 9 +++++---- hw/virtio/virtio-pci.c | 36 ++++++++++++++++++++++++++++++++---- hw/virtio/virtio-pci.h | 5 +++++ include/hw/virtio/virtio-bus.h | 10 +++++----- 5 files changed, 62 insertions(+), 28 deletions(-) diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c index 9678956..9f3f386 100644 --- a/hw/s390x/virtio-ccw.c +++ b/hw/s390x/virtio-ccw.c @@ -1261,6 +1261,16 @@ static int virtio_ccw_load_config(DeviceState *d, QEMUFile *f) return 0; } +static void virtio_ccw_pre_plugged(DeviceState *d, Error **errp) +{ + VirtioCcwDevice *dev = VIRTIO_CCW_DEVICE(d); + VirtIODevice *vdev = virtio_bus_get_device(&dev->bus); + + if (dev->max_rev >= 1) { + virtio_add_feature(&vdev->host_features, VIRTIO_F_VERSION_1); + } +} + /* This is called by virtio-bus just after the device is plugged. */ static void virtio_ccw_device_plugged(DeviceState *d, Error **errp) { @@ -1270,6 +1280,10 @@ static void virtio_ccw_device_plugged(DeviceState *d, Error **errp) SubchDev *sch = ccw_dev->sch; int n = virtio_get_num_queues(vdev); + if (!virtio_has_feature(vdev->host_features, VIRTIO_F_VERSION_1)) { + dev->max_rev = 0; + } + if (virtio_get_num_queues(vdev) > VIRTIO_CCW_QUEUE_MAX) { error_setg(errp, "The number of virtqueues %d " "exceeds ccw limit %d", n, @@ -1283,25 +1297,11 @@ static void virtio_ccw_device_plugged(DeviceState *d, Error **errp) sch->id.cu_model = virtio_bus_get_vdev_id(&dev->bus); - if (dev->max_rev >= 1) { - virtio_add_feature(&vdev->host_features, VIRTIO_F_VERSION_1); - } css_generate_sch_crws(sch->cssid, sch->ssid, sch->schid, d->hotplugged, 1); } -static void virtio_ccw_post_plugged(DeviceState *d, Error **errp) -{ - VirtioCcwDevice *dev = VIRTIO_CCW_DEVICE(d); - VirtIODevice *vdev = virtio_bus_get_device(&dev->bus); - - if (!virtio_host_has_feature(vdev, VIRTIO_F_VERSION_1)) { - /* A backend didn't support modern virtio. */ - dev->max_rev = 0; - } -} - static void virtio_ccw_device_unplugged(DeviceState *d) { VirtioCcwDevice *dev = VIRTIO_CCW_DEVICE(d); @@ -1593,8 +1593,8 @@ static void virtio_ccw_bus_class_init(ObjectClass *klass, void *data) k->load_queue = virtio_ccw_load_queue; k->save_config = virtio_ccw_save_config; k->load_config = virtio_ccw_load_config; + k->pre_plugged = virtio_ccw_pre_plugged; k->device_plugged = virtio_ccw_device_plugged; - k->post_plugged = virtio_ccw_post_plugged; k->device_unplugged = virtio_ccw_device_unplugged; k->ioeventfd_started = virtio_ccw_ioeventfd_started; k->ioeventfd_set_started = virtio_ccw_ioeventfd_set_started; diff --git a/hw/virtio/virtio-bus.c b/hw/virtio/virtio-bus.c index 1492793..11f65bd 100644 --- a/hw/virtio/virtio-bus.c +++ b/hw/virtio/virtio-bus.c @@ -49,16 +49,17 @@ void virtio_bus_device_plugged(VirtIODevice *vdev, Error **errp) DPRINTF("%s: plug device.\n", qbus->name); - if (klass->device_plugged != NULL) { - klass->device_plugged(qbus->parent, errp); + if (klass->pre_plugged != NULL) { + klass->pre_plugged(qbus->parent, errp); } /* Get the features of the plugged device. */ assert(vdc->get_features != NULL); vdev->host_features = vdc->get_features(vdev, vdev->host_features, errp); - if (klass->post_plugged != NULL) { - klass->post_plugged(qbus->parent, errp); + + if (klass->device_plugged != NULL) { + klass->device_plugged(qbus->parent, errp); } } diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c index dde71a5..2d60a00 100644 --- a/hw/virtio/virtio-pci.c +++ b/hw/virtio/virtio-pci.c @@ -1576,18 +1576,48 @@ static void virtio_pci_modern_io_region_unmap(VirtIOPCIProxy *proxy, ®ion->mr); } +static void virtio_pci_pre_plugged(DeviceState *d, Error **errp) +{ + VirtIOPCIProxy *proxy = VIRTIO_PCI(d); + VirtIODevice *vdev = virtio_bus_get_device(&proxy->bus); + + if (virtio_pci_modern(proxy)) { + virtio_add_feature(&vdev->host_features, VIRTIO_F_VERSION_1); + } + + virtio_add_feature(&vdev->host_features, VIRTIO_F_BAD_FEATURE); +} + /* This is called by virtio-bus just after the device is plugged. */ static void virtio_pci_device_plugged(DeviceState *d, Error **errp) { VirtIOPCIProxy *proxy = VIRTIO_PCI(d); VirtioBusState *bus = &proxy->bus; bool legacy = virtio_pci_legacy(proxy); - bool modern = virtio_pci_modern(proxy); + bool modern; bool modern_pio = proxy->flags & VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY; uint8_t *config; uint32_t size; VirtIODevice *vdev = virtio_bus_get_device(&proxy->bus); + /* + * Virtio capabilities present without + * VIRTIO_F_VERSION_1 confuses guests + */ + if (!virtio_has_feature(vdev->host_features, VIRTIO_F_VERSION_1)) { + virtio_pci_disable_modern(proxy); + + if (!legacy) { + error_setg(errp, "Device doesn't support modern mode, and legacy" + " mode is disabled"); + error_append_hint(errp, "Set disable-legacy to off\n"); + + return; + } + } + + modern = virtio_pci_modern(proxy); + config = proxy->pci_dev.config; if (proxy->class_code) { pci_config_set_class(config, proxy->class_code); @@ -1629,7 +1659,6 @@ static void virtio_pci_device_plugged(DeviceState *d, Error **errp) struct virtio_pci_cfg_cap *cfg_mask; - virtio_add_feature(&vdev->host_features, VIRTIO_F_VERSION_1); virtio_pci_modern_regions_init(proxy); virtio_pci_modern_mem_region_map(proxy, &proxy->common, &cap); @@ -1694,8 +1723,6 @@ static void virtio_pci_device_plugged(DeviceState *d, Error **errp) if (!kvm_has_many_ioeventfds()) { proxy->flags &= ~VIRTIO_PCI_FLAG_USE_IOEVENTFD; } - - virtio_add_feature(&vdev->host_features, VIRTIO_F_BAD_FEATURE); } static void virtio_pci_device_unplugged(DeviceState *d) @@ -2508,6 +2535,7 @@ static void virtio_pci_bus_class_init(ObjectClass *klass, void *data) k->query_guest_notifiers = virtio_pci_query_guest_notifiers; k->set_guest_notifiers = virtio_pci_set_guest_notifiers; k->vmstate_change = virtio_pci_vmstate_change; + k->pre_plugged = virtio_pci_pre_plugged; k->device_plugged = virtio_pci_device_plugged; k->device_unplugged = virtio_pci_device_unplugged; k->query_nvectors = virtio_pci_query_nvectors; diff --git a/hw/virtio/virtio-pci.h b/hw/virtio/virtio-pci.h index 0698157..541cbdb 100644 --- a/hw/virtio/virtio-pci.h +++ b/hw/virtio/virtio-pci.h @@ -181,6 +181,11 @@ static inline void virtio_pci_force_virtio_1(VirtIOPCIProxy *proxy) proxy->disable_legacy = ON_OFF_AUTO_ON; } +static inline void virtio_pci_disable_modern(VirtIOPCIProxy *proxy) +{ + proxy->disable_modern = true; +} + /* * virtio-scsi-pci: This extends VirtioPCIProxy. */ diff --git a/include/hw/virtio/virtio-bus.h b/include/hw/virtio/virtio-bus.h index f3e5ef3..24caa0a 100644 --- a/include/hw/virtio/virtio-bus.h +++ b/include/hw/virtio/virtio-bus.h @@ -54,16 +54,16 @@ typedef struct VirtioBusClass { int (*set_guest_notifiers)(DeviceState *d, int nvqs, bool assign); void (*vmstate_change)(DeviceState *d, bool running); /* + * Expose the features the transport layer supports before + * the negotiation takes place. + */ + void (*pre_plugged)(DeviceState *d, Error **errp); + /* * transport independent init function. * This is called by virtio-bus just after the device is plugged. */ void (*device_plugged)(DeviceState *d, Error **errp); /* - * Re-evaluate setup after feature bits have been validated - * by the device backend. - */ - void (*post_plugged)(DeviceState *d, Error **errp); - /* * transport independent exit function. * This is called by virtio-bus just before the device is unplugged. */