Jan Kiszka wrote: > Avi Kivity wrote: >> On 07/08/2009 04:39 PM, Glauber Costa wrote: >>> On Wed, Jul 08, 2009 at 04:27:27PM +0300, Avi Kivity wrote: >>> >>>> On 07/08/2009 04:09 PM, Glauber Costa wrote: >>>> >>>>> Signed-off-by: Glauber Costa >>>>> --- >>>>> kvm-all.c | 27 ++++++++++++++++++--------- >>>>> qemu-kvm.h | 6 +++--- >>>>> target-i386/kvm.c | 4 ++-- >>>>> 3 files changed, 23 insertions(+), 14 deletions(-) >>>>> >>>>> >>>>> >>>> Did you test the functionality to ensure we do not regress? >>>> >>> just briefly. To be honest, I'd feel much more confortable >>> if Jan could give it a try. >>> >> A Jan try (+review) will definitely help. >> > > Queued. > Deferred until v2. My first impression is that too much upstream code is moved or touched. Glauber, if you want to use some function that is currently under KVM_UPSTREAM, don't move it, just drop the #ifdef around it. And when done, have a look at the diff between upstream and qemu-kvm to avoid unneeded variations. Another question: What prevents using CONFIG_KVM also for qemu-kvm? I would rather mask out yet unused upstream code via the well-known KVM_UPSTREAM and drop the easily confusable USE_KVM defines. Jan