On Mon, Nov 26, 2018 at 8:31 AM Sean Christopherson <sean.j.christopherson@xxxxxxxxx> wrote: > > On Mon, Nov 19, 2018 at 09:50:52AM -0800, Jim Mattson wrote: > > I vaguely recall that at some time in the distant past, field 440EH of > > the VMCS was referred to as the "VMX instruction-information field." > > Very distant past :) Internal documentation has referred to the field > as VM_EXIT_INTSTRUCTION_INFO since at least 2002. Odds are this was a > SDM typo or a KVM quirk. I'm not imagining it, I swear! See Appendix C of Intel's document C97063-002 (April 2005). > > However, this field now provides instruction-information for string > > PIO VM-exits, descriptor table VM-exits, RDRAND VM-exits, and RDSEED > > VM-exits. The SDM now refers to it as the "VM-exit > > instruction-information field." Since this field still is not yet > > exposed as part of a userspace API, let's rename it to match the SDM. > > > > Signed-off-by: Jim Mattson <jmattson@xxxxxxxxxx> > > Reviewed-by: Peter Shier <pshier@xxxxxxxxxx> > > --- > > ... > > > @@ -9122,7 +9123,7 @@ static int handle_vmread(struct kvm_vcpu *vcpu) > > unsigned long field; > > u64 field_value; > > unsigned long exit_qualification = vmcs_readl(EXIT_QUALIFICATION); > > - u32 vmx_instruction_info = vmcs_read32(VMX_INSTRUCTION_INFO); > > + u32 vm_exit_instruction_info = vmcs_read32(VM_EXIT_INSTRUCTION_INFO); > > What about using "instr_info" when caching the field in a local variable? > It'd eliminate a bit of weird wrapping, and personally I always use > "instr info" when talking/thinking about the field. handle_vmptrst() > already uses the shorthand and I find that to be much more readable. Sure. V2 shortly.
