On 03/16/2012 09:55 AM, Benjamin Herrenschmidt wrote: > This is necessary for qemu to be able to pass the right information > to the guest, as the supported sizes and encodings can vary depending > on the machine, the type of KVM used (PR vs HV) and the version of KVM > > Signed-off-by: Benjamin Herrenschmidt <benh@xxxxxxxxxxxxxxxxxxx> > --- > > Please comment ASAP. I'm tired of the qemu side never working properly > because of that and our out-of-tree nasty patches we've been carrying > internally, so I'd like to get something like that in real quick :-) > > I have the qemu side patches that use this to generate the appropriate > device-tree when available, and use heuristics for the fallback. I'll > post them later, let's agree on the kernel interfaces first. > > The heuristics work as long as we have a reasonable guarantee that this > kernel patch will get in -before- any patch that enables the PVINFO > ioctl on "HV" KVM, that way I can rely on the later not working as > a way to differenciate PR and HV KVM if this new ioctl is not supported. > > Note: We probably want an other ioctl for getting other type of > MMU info, such as whether we support 1T segments etc... but I didn't > want to try to kill to many birds at once and end up in bike shed > painting on the mailing list for the next 6 month... > > Cheers, > Ben. > > arch/powerpc/include/asm/kvm_ppc.h | 3 ++- > arch/powerpc/kvm/book3s_hv.c | 35 +++++++++++++++++++++++++++++++++++ > arch/powerpc/kvm/book3s_pr.c | 22 ++++++++++++++++++++++ > arch/powerpc/kvm/powerpc.c | 18 +++++++++++++++++- > include/linux/kvm.h | 29 +++++++++++++++++++++++++++++ > 5 files changed, 105 insertions(+), 2 deletions(-) > -ENODOCS -- error compiling committee.c: too many arguments to function -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html