KVM: nVMX: Don't "put" vCPU or host state when switching VMCS

Browse source

When switching between vmcs01 and vmcs02, KVM isn't actually switching
between guest and host.  If guest state is already loaded (the likely,
if not guaranteed, case), keep the guest state loaded and manually swap
the loaded_cpu_state pointer after propagating saved host state to the
new vmcs0{1,2}.

Avoiding the switch between guest and host reduces the latency of
switching between vmcs01 and vmcs02 by several hundred cycles, and
reduces the roundtrip time of a nested VM by upwards of 1000 cycles.

Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

...

This commit is contained in:

Sean Christopherson 2019-05-07 09:06:31 -07:00 • committed by Paolo Bonzini

parent b464f57e13

commit 13b964a29d

3 changed files with 53 additions and 26 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

23

arch/x86/kvm/vmx/nested.c

View file

53

arch/x86/kvm/vmx/vmx.c

View file

3

arch/x86/kvm/vmx/vmx.h

View file