Re: [PATCH 0/2] x86/kvm: Force legacy PCI hole as WB under SNP/TDX

[Sean Christopherson <seanjc@google.com>]

On Thu, Aug 21, 2025, Sean Christopherson wrote:
> On Thu, Aug 21, 2025, Binbin Wu wrote:
> > On 8/21/2025 11:30 AM, Binbin Wu wrote:
> > > Variable MTRR has requirement for range size and alignment:
> > > For ranges greater than 4 KBytes, each range must be of length 2^n and its base
> > > address must be aligned on a 2^n boundary, where n is a value equal to or
> > > greater than 12. The base-address alignment value cannot be less than its length.
> > 
> > Wait, Linux kernel converts MTRR register values to MTRR state (base and size) and
> > cache it for later lookups (refer to map_add_var()). I.e., in Linux kernel,
> > only the cached state will be used.
> > 
> > These MTRR register values are never programmed when using
> > guest_force_mtrr_state() , so even the values doesn't meet the requirement
> > from hardware perspective, Linux kernel can still get the right base and
> > size.
> 
> Yeah.  I forget what happens if the ranges don't meet the power-of-2 requirements,
> but the mask+match logic should work jus tfine.
> 
> > No bothering to force the base and size alignment.
> > But a comment would be helpful.
> > Also, BIT(11) could be replaced by MTRR_PHYSMASK_V.
> 
> Ha!  I spent a good 5 minutes looking for a #define couldn't find one.  What a
> bizarre name...
> 
> > How about:
> > diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> > index 90097df4eafd..a9582ffc3088 100644
> > --- a/arch/x86/kernel/kvm.c
> > +++ b/arch/x86/kernel/kvm.c
> > @@ -934,9 +934,15 @@ static void kvm_sev_hc_page_enc_status(unsigned long pfn, int npages, bool enc)
> >  static void __init kvm_init_platform(void)
> >  {
> >         u64 tolud = e820__end_of_low_ram_pfn() << PAGE_SHIFT;
> > +       /*
> > +        * The range's base address and size may not meet the alignment
> > +        * requirement for variable MTRR. However, Linux guest never
> > +        * programs MTRRs when forcing guest MTRR state, no bothering to
> > +        * enforce the base and range size alignment.
> > +        */
> >         struct mtrr_var_range pci_hole = {
> >                 .base_lo = tolud | X86_MEMTYPE_UC,
> > -               .mask_lo = (u32)(~(SZ_4G - tolud - 1)) | BIT(11),
> > +               .mask_lo = (u32)(~(SZ_4G - tolud - 1)) | MTRR_PHYSMASK_V,
> >                 .mask_hi = (BIT_ULL(boot_cpu_data.x86_phys_bits) - 1) >> 32,
> >         };
> > 
> > 
> > I tested it in my setup, it can fix the issue of TPM driver failure with the
> > modified ACPI table for TPM in QEMU.
> > 
> > 
> > Hi Vishal,
> > Could you test it with google's VMM?
> 
> Vishal is OOO for a few days.  I pinged our internal bug tracker, I'll find
> someone to test.

Got confirmation this fixes the vTPM woes with Google's VMM.  v2 incoming...