The code just does not match the comment above which says we should
claim success.  And it makes sense since a removal from qemu was
requested and qemu could not find the device.  Without this patch such
codepath would lead to libvirt not removing the device from the XML and
no error being set, but the API would still return an error value.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
---
 src/qemu/qemu_hotplug.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
index f856e26c1877..d23da0a553c0 100644
--- a/src/qemu/qemu_hotplug.c
+++ b/src/qemu/qemu_hotplug.c
@@ -136,6 +136,7 @@ qemuDomainDeleteDevice(virDomainObj *vm,
              * domain XML is queried right after detach API the
              * device would still be there.  */
             VIR_DEBUG("Detaching of device %s failed and no event arrived", alias);
+            rc = 0;
         }
     }
 
-- 
2.47.0

On Mon, Nov 11, 2024 at 09:53:09 +0100, Martin Kletzander wrote:
> The code just does not match the comment above which says we should
> claim success.  And it makes sense since a removal from qemu was
> requested and qemu could not find the device.  Without this patch such
> codepath would lead to libvirt not removing the device from the XML and
> no error being set, but the API would still return an error value.
> 
> Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
> ---
>  src/qemu/qemu_hotplug.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
> index f856e26c1877..d23da0a553c0 100644
> --- a/src/qemu/qemu_hotplug.c
> +++ b/src/qemu/qemu_hotplug.c
> @@ -136,6 +136,7 @@ qemuDomainDeleteDevice(virDomainObj *vm,
>               * domain XML is queried right after detach API the
>               * device would still be there.  */
>              VIR_DEBUG("Detaching of device %s failed and no event arrived", alias);
> +            rc = 0;

After doing this the function can't return -2, but it will still be
documented as such. There is also a caller which explicitly checks for
-2.

I guess a better course of actions would be to add the '-2' check to
qemuDomainDetachDeviceChr instead. Possibly also to
'qemuDomainHotplugDelVcpu' but there that shouldn't be possible to
happen.