Am 16.09.2012 18:56, schrieb Richard Henderson: > On 2012-09-16 06:01, Andreas Färber wrote: >> This is lacking a proper description. I'd be very ashamed if we lost >> something so obvious during the QOM conversion. So what's the symptoms here? > > The most important thing here is saving cpu_model_str. > The symptom being a SEGV on clone, when cpu_model_str is null.
Ack on that. My understanding is it's not a regression but a bugfix. As for the reentrancy of the translation initialization, I would prefer that being handled in initfn, but we can first introduce it here and move it later. So, if you suggest a wording for the commit message I can apply this as part of my next CPU pull (assuming that'll be earlier than linux-user). I also have an include path fix cpu-qom.h -> cpu.h queued, and cpu_alpha_init() is not yet returning AlphaCPU. Andreas -- SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg