On 15/03/2016 17:31, Markus Armbruster wrote: > Paolo Bonzini <pbonz...@redhat.com> writes: > >> On 15/03/2016 16:03, Markus Armbruster wrote: >>> diff --git a/include/hw/hw.h b/include/hw/hw.h >>> index 2cb69d5..151d65e 100644 >>> --- a/include/hw/hw.h >>> +++ b/include/hw/hw.h >>> @@ -15,6 +15,7 @@ >>> #include "qemu/module.h" >>> >>> #ifdef NEED_CPU_H >>> +#include "cpu.h" >>> #if TARGET_LONG_BITS == 64 >>> #define qemu_put_betl qemu_put_be64 >>> #define qemu_get_betl qemu_get_be64 >> >> This actually makes things somewhat worse. This dependency of hw/hw.h >> on NEED_CPU_H should go away altogether. >> >> Would it be okay to stick to patches 1-13 for now? > > Yes, as long as "for now" approximately means "until 2.7 opens up". > > Proposed plan: I pursue 01-13 now, you rebase your need-cpu-h branch > onto this work when it's ready, I rebase 14-15 onto yours, and carry on. > Makes sense?
Yes, totally. I've already started rebasing and there's nothing else that sticks out. Since I have your attention, what about getting rid of fprintf-fn.h and moving it into qemu/osdep.h? Paolo
