yeahuh / rpms / qemu-kvm

Forked from rpms/qemu-kvm 2 years ago
Clone

Blame SOURCES/kvm-virtio-scsi-fix-hotplug-reset-vs-event-race.patch

ae23c9
From b872baa52f5e8e8c84b3a19bb6cf5b6e92106fb9 Mon Sep 17 00:00:00 2001
ae23c9
From: Stefan Hajnoczi <stefanha@redhat.com>
ae23c9
Date: Tue, 24 Jul 2018 15:13:08 +0200
ae23c9
Subject: [PATCH 264/268] virtio-scsi: fix hotplug ->reset() vs event race
ae23c9
ae23c9
RH-Author: Stefan Hajnoczi <stefanha@redhat.com>
ae23c9
Message-id: <20180724151308.20500-3-stefanha@redhat.com>
ae23c9
Patchwork-id: 81486
ae23c9
O-Subject: [RHEL-7.6 qemu-kvm-rhev PATCH 2/2] virtio-scsi: fix hotplug ->reset() vs event race
ae23c9
Bugzilla: 1607891
ae23c9
RH-Acked-by: Igor Mammedov <imammedo@redhat.com>
ae23c9
RH-Acked-by: Pankaj Gupta <pagupta@redhat.com>
ae23c9
RH-Acked-by: Cornelia Huck <cohuck@redhat.com>
ae23c9
ae23c9
There is a race condition during hotplug when iothread is used.  It
ae23c9
occurs because virtio-scsi may be processing command queues in the
ae23c9
iothread while the monitor performs SCSI device hotplug.
ae23c9
ae23c9
When a SCSI device is hotplugged the HotplugHandler->plug() callback is
ae23c9
invoked and virtio-scsi emits a rescan event to the guest.
ae23c9
ae23c9
If the guest submits a SCSI command at this point then it may be
ae23c9
cancelled before hotplug completes.  This happens because ->reset() is
ae23c9
called by hw/core/qdev.c:device_set_realized() after
ae23c9
HotplugHandler->plug() has been called and
ae23c9
hw/scsi/scsi-disk.c:scsi_disk_reset() purges all requests.
ae23c9
ae23c9
This patch uses the new HotplugHandler->post_plug() callback to emit the
ae23c9
rescan event after ->reset().  This eliminates the race conditions where
ae23c9
requests could be cancelled.
ae23c9
ae23c9
Reported-by: l00284672 <lizhengui@huawei.com>
ae23c9
Cc: Paolo Bonzini <pbonzini@redhat.com>
ae23c9
Cc: Fam Zheng <famz@redhat.com>
ae23c9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
ae23c9
Message-Id: <20180716083732.3347-3-stefanha@redhat.com>
ae23c9
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
ae23c9
(cherry picked from commit 8449bcf94986156a1476d6647c75ec1ce3db64d0)
ae23c9
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
ae23c9
Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>
ae23c9
---
ae23c9
 hw/scsi/virtio-scsi.c | 11 ++++++++++-
ae23c9
 1 file changed, 10 insertions(+), 1 deletion(-)
ae23c9
ae23c9
diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
ae23c9
index 9f754c4..52a3c1d 100644
ae23c9
--- a/hw/scsi/virtio-scsi.c
ae23c9
+++ b/hw/scsi/virtio-scsi.c
ae23c9
@@ -806,8 +806,16 @@ static void virtio_scsi_hotplug(HotplugHandler *hotplug_dev, DeviceState *dev,
ae23c9
         virtio_scsi_acquire(s);
ae23c9
         blk_set_aio_context(sd->conf.blk, s->ctx);
ae23c9
         virtio_scsi_release(s);
ae23c9
-
ae23c9
     }
ae23c9
+}
ae23c9
+
ae23c9
+/* Announce the new device after it has been plugged */
ae23c9
+static void virtio_scsi_post_hotplug(HotplugHandler *hotplug_dev,
ae23c9
+                                     DeviceState *dev)
ae23c9
+{
ae23c9
+    VirtIODevice *vdev = VIRTIO_DEVICE(hotplug_dev);
ae23c9
+    VirtIOSCSI *s = VIRTIO_SCSI(vdev);
ae23c9
+    SCSIDevice *sd = SCSI_DEVICE(dev);
ae23c9
 
ae23c9
     if (virtio_vdev_has_feature(vdev, VIRTIO_SCSI_F_HOTPLUG)) {
ae23c9
         virtio_scsi_acquire(s);
ae23c9
@@ -977,6 +985,7 @@ static void virtio_scsi_class_init(ObjectClass *klass, void *data)
ae23c9
     vdc->start_ioeventfd = virtio_scsi_dataplane_start;
ae23c9
     vdc->stop_ioeventfd = virtio_scsi_dataplane_stop;
ae23c9
     hc->plug = virtio_scsi_hotplug;
ae23c9
+    hc->post_plug = virtio_scsi_post_hotplug;
ae23c9
     hc->unplug = virtio_scsi_hotunplug;
ae23c9
 }
ae23c9
 
ae23c9
-- 
ae23c9
1.8.3.1
ae23c9