On Wed, Jun 04, 2014 at 10:08:12PM +0300, Michael S. Tsirkin wrote: > On Wed, Jun 04, 2014 at 06:34:04PM +0200, Paolo Bonzini wrote: > > Il 04/06/2014 16:44, Alexander Graf ha scritto: > > > > > > > > >>Obviously, if you really like the current behavior better you can > > >>always reject whatever patch I'll come up with, but I'd like to at > > >>least try and see what it would look like :) > > > > > >I think it's perfectly fine to leave mwait always implemented as NOP - > > >it's valid behavior. > > > > > >As for the CPUID exposure, that should be a pure QEMU thing. If > > >overriding CPUID bits the kernel mask tells us doesn't work today, we > > >should just make it possible :). > > > > That should be the purpose of KVM_GET_EMULATED_CPUID, so MWAIT could be > > added in __do_cpuid_ent_emulated. However, the corresponding QEMU patches > > were never included. Borislav, can you refresh them? > > > > Paolo > > I don't understand why would we want mwait bit set in CPUID. > The only reason we want the nop is because of broken guests which > don't check CPUID.
E.g., OS X 10.5 *does* check CPUID, and panics if it doesn't find it. It needs the MONITOR cpuid flag to be on, *and* the actual instructions to work. HOWEVER: I really do NOT want us to bend over backwards to support it, I think having 10.6 and up working is good enough. Besides, there are other problems we'd run into with 10.5 if we got the mwait situation taken care of, so even more reason to leave well enough alone. Thanks, --Gabriel PS. Thanks again for everyone's patience while I wrapped my head around the fine details :) -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
