On Wed, Jul 28, 2021 at 11:11:30AM -1000, Richard Henderson wrote: > On 7/28/21 10:37 AM, Peter Xu wrote: > > A quick fix attached; would that work for us? > > Looks plausible, though perhaps just as easy to list the 5 platforms as just > the one: > > #if defined(__linux__) && \ > (defined(HOST_X86_64) || \ > defined(HOST_S390X) || \ > ...) > # define HAVE_KVM > #endif
That looks good to me, especially for the long term to identify whether kvm is with us, but for the short-term I hope I can still use the (literally :) simpler patch as attached so hopefully that'll be more welcomed as rc2+ material.. Note again that the kvm.h inclusion is only for kvm dirty ring test in migration-test so far, meanwhile that's only supported on x86_64, so we won't lose anything on the rest 4 archs. Thanks! -- Peter Xu
