On Tue, 2013-11-12 at 17:41 +0100, Andreas Färber wrote: > Am 12.11.2013 04:02, schrieb Chen Fan: > > On Tue, 2013-11-12 at 09:54 +0800, 赵小强 wrote: > >> He asked me to drop the parent_realize. so in v2, I just replace the > >> 'init' with 'realize'. > >> > > Hmm,I'm confused, Go through the entire QEMU source code, there was no > > DeviceRealize realize field in leaf child, I think Andreas's purpose for > > you was not simple to replace the 'init' by 'realize'. Maybe I'm > > mistaken, I hope Andreas could give some comments about this. > > I was serious about that. It appears that my earlier reply didn't arrive > on the list, so I resent it today, hope that sheds some light. > > My virtio series had been reworked in the same fashion but is *still* in > need of sending to the list... > > > on the other hand, I feel good regarding refactor QOM'ify apic, I'm now > > making adding 'cpu-del' supported patches. which need unrealize the apic > > device. so please rebase your work on Andreas's git source: > > git://github.com/afaerber/qemu-cpu.git qom-cpu. > > then we can do for them together. > > Actually the only pending work is Igor's x86 properties on qom-cpu-next > as of now. Both qom-cpu and qom-next have recently been merged so should > be no different from upstream qemu.git. > > The QOM device refactorings are usually taking place on qom-next branch > at the same repository. We'll have to see how that works out with the > CPU work - ideally I'd merge qom-next first, then continue on qom-cpu. > Thanks for your clarification !
Chen > Regards, > Andreas >