Linux 6.13-rc3 fixes pricey CPUID coping with on Sapphire Rapids CPUs — an issue resulting in Skylake CPUs being 4X sooner for CPUID-related updates

The Linux crew has revealed an intermediate Linux 6.13-rc3 to restore a “hilarious” bug inside the kernel-based Digital Machine code — Phoronix experiences, the place CPUID updates are taking significantly longer time to be processed on Sapphire Rapids CPUs than they should be, resulting in loads older Skylake CPUs being as a lot as 4X sooner with regard to CPUID worth.

Phoronix reveals that Google engineers initially discovered this technical problem in how Sapphire Rapids processes CPUID information in nested VMs. Notably, Google’s Sean Christopherson states, “On Intel’s Emerald Rapids, CPUID is *wildly* pricey, to the aim the place recomputing XSAVE offsets and sizes ends in a 4x enhance in latency of nested VM-Enter and VM-Exit (nested transitions can set off xstate_required_size() quite a few situations per transition), relative to using cached values. The issue is unquestionably seen by working `perf prime` whereas triggering nested transitions: kvm_update_cpuid_runtime() reveals up at a whopping 50%…”