Am 17.07.2017 um 12:41 schrieb Igor Mammedov: > On Sat, 15 Jul 2017 08:08:58 -1000 > Richard Henderson <r...@twiddle.net> wrote: > >> On 07/14/2017 03:52 AM, Igor Mammedov wrote: >>> @@ -230,6 +230,8 @@ static void m68k_cpu_realizefn(DeviceState *dev, Error >>> **errp) >>> M68kCPUClass *mcc = M68K_CPU_GET_CLASS(dev); >>> Error *local_err = NULL; >>> >>> + register_m68k_insns(&cpu->env); >>> + >> >> I think it would make more sense to do this during m68k_tcg_init. >> > it seems that m68k_cpu_initfn accesses 'env' via some global, > while cpu_mk68k_init() used to access concrete pointer of just created cpu, > > how about moving register_m68k_insns() to m68k_cpu_initfn(), instead? > it should be equivalent to what cpu_mk68k_init() used to do.
As a general note, realize should be re-entrant. Can't tell from the above diff whether that is the case here. Regards, Andreas -- SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg)
