Am 18.03.26 um 10:46 schrieb Fiona Ebner:
> Am 17.03.26 um 5:58 PM schrieb Thomas Lamprecht:
>> Am 12.03.26 um 13:13 schrieb Fiona Ebner:
>>> @@ -715,7 +715,12 @@ index 0000000000..56e0fa6c69
>>> +
>>> + migration_incoming_state_destroy();
>>> + if (ret < 0) {
>>> -+ error_setg_errno(errp, -ret, "Error while loading VM state");
>>> ++ if (local_err) {
>>> ++ error_setg_errno(errp, -ret, "Error while loading VM state -
>>> %s",
>>> ++ error_get_pretty(local_err));
>>
>> Would we need to free the local_err here? But in any case a tiny nit
>> and not really relevant in terms of leaking memory due to being in
>> a rather unlikely error branch.
>
> Yes, you are right. If error_propagate() is not used, a local error must
> be freed. There are quite a few more instances where errors are not
> freed in the surrounding code. There is an error_propagate_prepend()
> helper that could be used.
I mean, if QEMU exits due to the error anyway it won't really matter.
> Should I send a v2 or a follow-up?
Can be addressed in a follow-up.
