On 04.01.2013, at 20:07, Scott Wood wrote: > On 01/04/2013 12:54:51 PM, Alexander Graf wrote: >> On 04.01.2013, at 19:50, Scott Wood wrote: >> > On 01/04/2013 04:24:59 AM, Alexander Graf wrote: >> >> msi_supported = true; >> >> list = list_be; >> >> diff --git a/target-ppc/Makefile.objs b/target-ppc/Makefile.objs >> >> index 237a0ed..6c11ef8 100644 >> >> --- a/target-ppc/Makefile.objs >> >> +++ b/target-ppc/Makefile.objs >> >> @@ -9,4 +9,3 @@ obj-y += mmu_helper.o >> >> obj-y += timebase_helper.o >> >> obj-y += misc_helper.o >> >> obj-y += mem_helper.o >> >> -obj-y += mpic_helper.o >> >> diff --git a/target-ppc/cpu.h b/target-ppc/cpu.h >> >> index e88ebe0..0db06d6 100644 >> >> --- a/target-ppc/cpu.h >> >> +++ b/target-ppc/cpu.h >> >> @@ -1068,6 +1068,8 @@ struct CPUPPCState { >> >> target_ulong ivpr_mask; >> >> target_ulong hreset_vector; >> >> hwaddr mpic_cpu_base; >> >> + /* true when the external proxy facility mode is enabled */ >> >> + int mpic_proxy; >> >> #endif >> > >> > bool? >> IIRC we can't save/restore bools easily using savevm, and this one should be >> saved/restored. > > I see bool handlers in savevm.c.
I mean vmstate of course. But you're right - there are bool handlers. Awesome :). Alex > I'm not sure how a non-fixed-size "int" is better. > > BTW, docs/migration.txt says that savevm is the "legacy" approach. > > -Scott