[libvirt] [PATCH] qemu: avoid double reservation of PCI address when hotplugging interface type='hostdev'

[Laine Stump]

Commit 01ca4010d86 (libvirt v5.1.0) moved address reservation for
hotplugged interface devices up to an earlier point in
qemuDomainAttachNetDevice() because some function called when
hotplugging on aarch64 needed to know the address type, and failed
when it was "none".

This unfortunately caused a regression, because it also made PCI
address reservation happen before we noticed that the device was a
*hostdev* interface. Those interfaces are hotplugged by just calling
out to qemuDomainAttachHostdevDevice() - that function would then also
attempt to reserve the *same PCI address* that had just been reserved
in qemuDomainAttachNetDevice().

The solution is to move the bit of code that short-circuits out to
virDomainHostdevAttach() up *even earlier* so that no PCI address has
been allocated by the time it's called.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1744523
Signed-off-by: Laine Stump <laine at redhat.com>
---
 src/qemu/qemu_hotplug.c | 26 +++++++++++++++-----------
 1 file changed, 15 insertions(+), 11 deletions(-)

diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
index 55696ebe9a..80ddee6587 100644
--- a/src/qemu/qemu_hotplug.c
+++ b/src/qemu/qemu_hotplug.c
@@ -1181,6 +1181,17 @@ qemuDomainAttachNetDevice(virQEMUDriverPtr driver,
     if (qemuAssignDeviceNetAlias(vm->def, net, -1) < 0)
         goto cleanup;
 
+    if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
+        /* This is really a "smart hostdev", so it should be attached
+         * as a hostdev (the hostdev code will reach over into the
+         * netdev-specific code as appropriate), then also added to
+         * the nets list (see cleanup:) if successful.
+         */
+        ret = qemuDomainAttachHostDevice(driver, vm,
+                                         virDomainNetGetActualHostdev(net));
+        goto cleanup;
+    }
+
     if (qemuDomainIsS390CCW(vm->def) &&
         net->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI &&
         virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_CCW)) {
@@ -1260,17 +1271,6 @@ qemuDomainAttachNetDevice(virQEMUDriverPtr driver,
             goto cleanup;
         break;
 
-    case VIR_DOMAIN_NET_TYPE_HOSTDEV:
-        /* This is really a "smart hostdev", so it should be attached
-         * as a hostdev (the hostdev code will reach over into the
-         * netdev-specific code as appropriate), then also added to
-         * the nets list (see cleanup:) if successful.
-         */
-        ret = qemuDomainAttachHostDevice(driver, vm,
-                                         virDomainNetGetActualHostdev(net));
-        goto cleanup;
-        break;
-
     case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
         queueSize = net->driver.virtio.queues;
         if (!queueSize)
@@ -1313,6 +1313,10 @@ qemuDomainAttachNetDevice(virQEMUDriverPtr driver,
         }
         break;
 
+    case VIR_DOMAIN_NET_TYPE_HOSTDEV:
+        /* this switch is skipped by hostdev interfaces */
+        break;
+
     case VIR_DOMAIN_NET_TYPE_SERVER:
     case VIR_DOMAIN_NET_TYPE_CLIENT:
     case VIR_DOMAIN_NET_TYPE_MCAST:
-- 
2.21.0