On Tue, 2021-07-27 at 00:20 +0000, Sean Christopherson wrote:
> On Sun, Jul 25, 2021, Robert Hoo wrote:
> > PML in VM depends on "enable PML" in VM-execution control, check
> > this
> > before vmwrite to PMLADDR, because this field doesn't exist if PML
> > is
> > disabled in VM-execution control.
>
> No, the field doesn't exist if the CPU doesn't support PML. Whether
> or not PML
> is enabled in the execution controls is irrelevant.
Right, Thanks Sean.
> pml_init() checks for both
> secondary execution controls and PML support (with bad indentation).
>
> if (!(ctrl_cpu_rev[0].clr & CPU_SECONDARY) ||
> !(ctrl_cpu_rev[1].clr & CPU_PML)) {
> printf("\tPML is not supported");
> return VMX_TEST_EXIT;
> }
>
> pml_log = alloc_page();
> vmcs_write(PMLADDR, (u64)pml_log);
> vmcs_write(GUEST_PML_INDEX, PML_INDEX - 1);
>
> > Signed-off-by: Robert Hoo <robert.hu@xxxxxxxxxxxxxxx>
> > ---
> > x86/vmx_tests.c | 9 ++++++---
> > 1 file changed, 6 insertions(+), 3 deletions(-)
> >
> > diff --git a/x86/vmx_tests.c b/x86/vmx_tests.c
> > index 4f712eb..8663112 100644
> > --- a/x86/vmx_tests.c
> > +++ b/x86/vmx_tests.c
> > @@ -1502,13 +1502,16 @@ static int pml_init(struct vmcs *vmcs)
> > return VMX_TEST_EXIT;
> > }
> >
> > + ctrl_cpu = vmcs_read(CPU_EXEC_CTRL1) & CPU_PML;
> > + if (!ctrl_cpu) {
> > + printf("\tPML is not enabled\n");
> > + return VMX_TEST_EXIT;
> > + }
> > +
> > pml_log = alloc_page();
> > vmcs_write(PMLADDR, (u64)pml_log);
> > vmcs_write(GUEST_PML_INDEX, PML_INDEX - 1);
> >
> > - ctrl_cpu = vmcs_read(CPU_EXEC_CTRL1) | CPU_PML;
> > - vmcs_write(CPU_EXEC_CTRL1, ctrl_cpu);
> > -
> > return VMX_TEST_START;
> > }
> >
> > --
> > 1.8.3.1
> >
