Am 06.07.2013 14:44, schrieb Peter Maydell: > On 6 July 2013 13:40, Andreas Färber <afaer...@suse.de> wrote: >> softmmu would do it after the future QMP qom-set phase. The mess there >> is reset handler registration order: We cannot have most CPUs register a >> reset handler themselves yet because some machines (including most ARM >> ones) register reset handlers to tweak registers before the CPU would >> have reset in that future scenario. > > I'm not really a fan of that "use reset handler to simulate > bootloader firmware" code, so if you have a cleaner solution > to suggest I'd be happy to move to that.
I once suggested a secondary reset hook in CPUClass, to be invoked from CPUClass::reset, that boards could set to piggy-back initializations, but IIRC Anthony didn't like that. For PReP I am trying to avoid NIP tweaking sneaking in with the ELF loading requests since I know how hard it'll be to keep working. Andreas > (if the cleaner solution is "provide a firmware blob for > all boards" that might be too much work though :-)) > > -- PMM -- SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg
