This series identifies some low-hanging fruit for speeding up emulation of invalid guest state. I am a bit worried about patch 2, while everything else should be relatively safe. On the kvm-unit-tests microbenchmarks I get a 1.8-2.5x speedup (from 740-1100 cycles/instruction to 280-600). This is not for 3.15, of course. Even patch 5 is not too useful without the others; saving 40 clock cycles is a 10% speedup after the previous patches, but only 4% before. Please review carefully! Paolo Paolo Bonzini (5): KVM: vmx: speed up emulation of invalid guest state KVM: x86: avoid useless set of KVM_REQ_EVENT after emulation KVM: x86: move around some checks KVM: x86: protect checks on ctxt->d by a common "if (unlikely())" KVM: x86: speed up emulated moves arch/x86/include/asm/kvm_emulate.h | 2 +- arch/x86/kvm/emulate.c | 182 ++++++++++++++++++++----------------- arch/x86/kvm/vmx.c | 5 +- arch/x86/kvm/x86.c | 28 ++++-- 4 files changed, 120 insertions(+), 97 deletions(-) -- 1.8.3.1 -- 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
