On 2011-09-07 15:23, Anthony Liguori wrote: > On 09/07/2011 08:17 AM, Michael S. Tsirkin wrote: >> On Wed, Sep 07, 2011 at 03:13:00PM +0200, Jan Kiszka wrote: >>> On 2011-09-07 15:06, Michael S. Tsirkin wrote: >>>> But if the command is not for users at all, if it's >>>> for qemu debugging, then exposing internals is a very >>>> logical thing. Only problem is - we must make it very very clear >>>> which commands are for qemu debugging. >>> >>> This command it also for users, to debug guests. >>> >>> Jan >> >> Hmm, guest visible state must be stable by definition. >> So why can't we make the interfaces stable then? > > Because right now how you reference devices cannot be stable. > > Going back to a previous thread, what if the command took a qdev class > type as a filter? > > That would be a stable name, not require anonymous IDs, and would have > the property of usually being a single device.
It's surely not usual that there is only a single instance of a device type. That's why my old series introduced (unstable) instance numbering. I refrained for warming those patches up again. Jan -- Siemens AG, Corporate Technology, CT T DE IT 1 Corporate Competence Center Embedded Linux
