Quoting Eric Blake (2016-11-30 20:33:56)
> On 11/30/2016 05:06 PM, Michael Roth wrote:
> > Currently objects specified on the command-line are only partially
> > cleaned up when 'object_del' is issued in either HMP or QMP: the
> > object itself is fully finalized, but the QemuOpts are not removed.
> > This results in the following behavior:
> >
> > x86_64-softmmu/qemu-system-x86_64 -monitor stdio \
> > -object memory-backend-ram,id=ram1,size=256M
> >
> > QEMU 2.7.91 monitor - type 'help' for more information
> > (qemu) object_del ram1
> > (qemu) object_del ram1
> > object 'ram1' not found
> > (qemu) object_add memory-backend-ram,id=ram1,size=256M
> > Duplicate ID 'ram1' for object
> > Try "help object_add" for more information
> >
> > which can be an issue for use-cases like memory hotplug.
>
> Nice analysis.
>
> >
> > This happens on the HMP side because hmp_object_add() attempts to
> > create a temporary QemuOpts entry with ID 'ram1', which ends up
> > conflicting with the command-line-created entry, since it was never
> > cleaned up during the previous hmp_object_del() call.
> >
> > We address this by adding checks in {qmp,hmp}_object_del to determine
> > whether an option group entry matching the object's ID is present,
> > and removing it if it is.
> >
> > Note that qmp_object_add() never attempts to create a temporary
> > QemuOpts entry, so it does not encounter the duplicate ID error,
> > which is why this isn't generally visible in libvirt.
>
> Phew. Yeah, libvirt avoiding HMP where possible saves a lot of hassles.
>
> > +++ b/hmp.c
> > @@ -2072,6 +2072,16 @@ void hmp_object_del(Monitor *mon, const QDict *qdict)
> >
> > user_creatable_del(id, &err);
> > hmp_handle_error(mon, &err);
> > +
> > + /* if object was defined on the command-line, remove its corresponding
> > + * option group entry
> > + */
> > + if (err == NULL) {
> > + QemuOptsList *opt_group = qemu_find_opts_err("object", &err);
> > + if (opt_group) {
> > + qemu_opts_del(qemu_opts_find(opt_group, id));
> > + }
> > + }
>
> This one looks okay.
>
> > }
> >
> > void hmp_info_memdev(Monitor *mon, const QDict *qdict)
> > diff --git a/qmp.c b/qmp.c
> > index 0028f0b..87c545d 100644
> > --- a/qmp.c
> > +++ b/qmp.c
> > @@ -686,6 +686,16 @@ void qmp_object_add(const char *type, const char *id,
> > void qmp_object_del(const char *id, Error **errp)
> > {
> > user_creatable_del(id, errp);
> > +
> > + /* if object was defined on the command-line, remove its corresponding
> > + * option group entry
> > + */
> > + if (!(errp && *errp)) {
>
> But this is wrong. Please spin a v2 that uses a local Error object, then
> use error_propagate(err, errp). Making anything conditional on whether
> an error occurred requires a local object, since the caller can pass in
> NULL, but you want your cleanup to happen even when the caller doesn't
> care about errors.
Ugh, I should know better by now. Thanks for the catch, will make sure
this is done properly in v2.
>
> > + QemuOptsList *opt_group = qemu_find_opts_err("object", errp);
> > + if (opt_group) {
> > + qemu_opts_del(qemu_opts_find(opt_group, id));
> > + }
> > + }
> > }
> >
> > MemoryDeviceInfoList *qmp_query_memory_devices(Error **errp)
> >
>
> --
> Eric Blake eblake redhat com +1-919-301-3266
> Libvirt virtualization library http://libvirt.org
>
