hw/xen: Avoid crash when backend watch fires too early

The xen-block code ends up calling aio_poll() through blkconf_geometry(),
which means we see watch events during the indirect call to
xendev_class->realize() in xen_device_realize(). Unfortunately this call
is made before populating the initial frontend and backend device nodes
in xenstore and hence xen_block_frontend_changed() (which is called from
a watch event) fails to read the frontend's 'state' node, and hence
believes the device is being torn down. This in-turn sets the backend
state to XenbusStateClosed and causes the device to be deleted before it
is fully set up, leading to the crash.
By simply moving the call to xendev_class->realize() after the initial
xenstore nodes are populated, this sorry state of affairs is avoided.

Reported-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Paul Durrant <pdurrant@amazon.com>

diff --git a/hw/xen/xen-bus.c b/hw/xen/xen-bus.c
index e8d9adeb3e65018be26dc80ff2e60329c89cedfe..83121c5a43d155d71fe911096d1906a2a28ec39d 100644 (file)
--- a/hw/xen/xen-bus.c
+++ b/hw/xen/xen-bus.c
@@ -1031,13 +1031,6 @@ static void xen_device_realize(DeviceState *dev, Error **errp)
         goto unrealize;
     }
 
-    if (xendev_class->realize) {
-        xendev_class->realize(xendev, errp);
-        if (*errp) {
-            goto unrealize;
-        }
-    }
-
     xen_device_backend_printf(xendev, "frontend", "%s",
                               xendev->frontend_path);
     xen_device_backend_printf(xendev, "frontend-id", "%u",
@@ -1056,6 +1049,13 @@ static void xen_device_realize(DeviceState *dev, Error **errp)
         xen_device_frontend_set_state(xendev, XenbusStateInitialising, true);
     }
 
+    if (xendev_class->realize) {
+        xendev_class->realize(xendev, errp);
+        if (*errp) {
+            goto unrealize;
+        }
+    }
+
     xendev->exit.notify = xen_device_exit;
     qemu_add_exit_notifier(&xendev->exit);
     return;