In cases when we are adding a <transient/> disk with sharing backend
(and thus hotplugging it) we need to re-initialize ACPI tables so that
the VM boots from the correct device.

This has a side-effect of emitting the RESET event and handling it which
in case when the 'on_reset' policy is set to 'destroy' can even kill the
VM.

Fix this by ignoring RESET events during startup of the VM.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
---
 src/qemu/qemu_process.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c
index 521fda57da..d6ed4a3943 100644
--- a/src/qemu/qemu_process.c
+++ b/src/qemu/qemu_process.c
@@ -429,12 +429,20 @@ qemuProcessHandleReset(qemuMonitor *mon G_GNUC_UNUSED,
                        void *opaque)
 {
     virQEMUDriver *driver = opaque;
-    virObjectEvent *event;
+    virObjectEvent *event = NULL;
     qemuDomainObjPrivate *priv;
     g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);

     virObjectLock(vm);

+    /* ignore reset events on VM startup. Libvirt in certain instances does a
+     * reset during startup so that the ACPI tables are re-generated */
+    if (vm->state.state == VIR_DOMAIN_PAUSED &&
+        vm->state.reason == VIR_DOMAIN_PAUSED_STARTING_UP) {
+        VIR_DEBUG("ignoring reset event during startup");
+        goto cleanup;
+    }
+
     event = virDomainEventRebootNewFromObj(vm);
     priv = vm->privateData;
     if (priv->agent)
-- 
2.31.1

On Fri, Jul 23, 2021 at 16:46:14 +0200, Peter Krempa wrote:
> In cases when we are adding a <transient/> disk with sharing backend
> (and thus hotplugging it) we need to re-initialize ACPI tables so that
> the VM boots from the correct device.
> 
> This has a side-effect of emitting the RESET event and handling it which
> in case when the 'on_reset' policy is set to 'destroy' can even kill the
> VM.
> 
> Fix this by ignoring RESET events during startup of the VM.
> 
> Signed-off-by: Peter Krempa <pkrempa@redhat.com>
> ---
>  src/qemu/qemu_process.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)

I'll be posting another version of this as QEMU on aarch64 uses a
SHUTDOWN event instead of RESET.