On Thu, Mar 05, 2020 at 01:10:57PM -0800, Jim Mattson wrote: > On Thu, Mar 5, 2020 at 11:25 AM Sean Christopherson > <sean.j.christopherson@xxxxxxxxx> wrote: > > > > On Thu, Mar 05, 2020 at 10:43:51AM -0800, Jim Mattson wrote: > > > On Wed, Mar 4, 2020 at 5:34 PM Sean Christopherson > > > <sean.j.christopherson@xxxxxxxxx> wrote: > > > > > > > > Rework the masking in the out-of-range CPUID logic to handle the > > > > Hypervisor sub-classes, as well as the Centaur class if the guest > > > > virtual CPU vendor is Centaur. > > > > > > > > Masking against 0x80000000 only handles basic and extended leafs, which > > > > results in Hypervisor range checks being performed against the basic > > > > CPUID class, and Centuar range checks being performed against the > > > > Extended class. E.g. if CPUID.0x40000000.EAX returns 0x4000000A and > > > > there is no entry for CPUID.0x40000006, then function 0x40000006 would > > > > be incorrectly reported as out of bounds. > > > > > > > > While there is no official definition of what constitutes a class, the > > > > convention established for Hypervisor classes effectively uses bits 31:8 > > > > as the mask by virtue of checking for different bases in increments of > > > > 0x100, e.g. KVM advertises its CPUID functions starting at 0x40000100 > > > > when HyperV features are advertised at the default base of 0x40000000. > > > > > > > > The bad range check doesn't cause functional problems for any known VMM > > > > because out-of-range semantics only come into play if the exact entry > > > > isn't found, and VMMs either support a very limited Hypervisor range, > > > > e.g. the official KVM range is 0x40000000-0x40000001 (effectively no > > > > room for undefined leafs) or explicitly defines gaps to be zero, e.g. > > > > Qemu explicitly creates zeroed entries up to the Cenatur and Hypervisor > > > > limits (the latter comes into play when providing HyperV features). > > > > > > > > The bad behavior can be visually confirmed by dumping CPUID output in > > > > the guest when running Qemu with a stable TSC, as Qemu extends the limit > > > > of range 0x40000000 to 0x40000010 to advertise VMware's cpuid_freq, > > > > without defining zeroed entries for 0x40000002 - 0x4000000f. > > > > > > > > Note, documentation of Centaur/VIA CPUs is hard to come by. Designating > > > > 0xc0000000 - 0xcfffffff as the Centaur class is a best guess as to the > > > > behavior of a real Centaur/VIA CPU. > > > > > > Don't forget Transmeta's CPUID range at 0x80860000 through 0x8086FFFF! > > > > Hmm, is it actually needed here? KVM doesn't advertise support for that > > range in KVM_GET_SUPPORTED_CPUID. > > That's also why I limited the Centaur > > range to vendor==CENTAUR, as KVM_GET_SUPPORTED_CPUID enumerates the > > Centaur range if and only if the host CPU is Centaur. Actually, I take this part of that comment back. I limited this to vendor==CENTAUR so that it wouldn't conflict with an actual Intel CPU if Intel adds a CPUID leaf in the 0xc0000000 range. > Ah. So cross-vendor CPUID specifications are not supported? Cross-vendor CPUID is sort of allowed? E.g. this plays nice with creating a Centaur CPU on an Intel platform. My interpretation of GET_SUPPORTED... is that KVM won't prevent enumerating what you want in CPUID, but it only promises to correctly support select leafs.
