07.09.2015 18:46, Andreas Färber wrote: > Am 04.09.2015 um 23:01 schrieb marcandre.lur...@redhat.com: >> From: Marc-André Lureau <marcandre.lur...@redhat.com> >> >> It looks like this documentation is obsolete: a child object may lookup >> its parent stored in the Object struct. >> >> Signed-off-by: Marc-André Lureau <marcandre.lur...@redhat.com> >> --- >> include/qom/object.h | 3 --- >> 1 file changed, 3 deletions(-) > > Either once again you are trying to do stuff behind my back, or your > setup is really broken: I double-checked that include/qom/ is listed in > MAINTAINERS, so I should've been CC'ed rather than just -trivial. > > It's been a valid rule not to mess with these internal fields, therefore > this is not trivial at all, and that's one reason why my x86 CPU series > using it was an RFC. We should either come up with a proper wrapper > function object_get_parent(), or with a wrapper function adding a link<> > property (where we would need to be careful with ref counts) - long time > only the composition tree needed to mess with an object's parent. > > If you have a concrete use case of parent access, please point to it.
So, should the current documentation remain, or should the patch be applied? If the latter, Andreas, please take care of it. Thanks, /mjt
