Re: [PATCH v3] selftests: KVM: avoid failures due to reserved HyperTransport region

From: Andrew Jones
Date: Tue Dec 14 2021 - 10:07:57 EST


On Thu, Dec 09, 2021 at 05:30:40PM -0500, Paolo Bonzini wrote:
> AMD proceessors define an address range that is reserved by HyperTransport
> and causes a failure if used for guest physical addresses. Avoid
> selftests failures by reserving those guest physical addresses; the
> rules are:
>
> - On parts with <40 bits, its fully hidden from software.
>
> - Before Fam17h, it was always 12G just below 1T, even if there was more
> RAM above this location. In this case we just not use any RAM above 1T.
>
> - On Fam17h and later, it is variable based on SME, and is either just
> below 2^48 (no encryption) or 2^43 (encryption).
>
> Fixes: ef4c9f4f6546 ("KVM: selftests: Fix 32-bit truncation of vm_get_max_gfn()")
> Cc: stable@xxxxxxxxxxxxxxx
> Cc: David Matlack <dmatlack@xxxxxxxxxx>
> Reported-by: Maxim Levitsky <mlevitsk@xxxxxxxxxx>
> Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> Message-Id: <20210805105423.412878-1-pbonzini@xxxxxxxxxx>
> Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> ---
> .../testing/selftests/kvm/include/kvm_util.h | 9 +++
> tools/testing/selftests/kvm/lib/kvm_util.c | 2 +-
> .../selftests/kvm/lib/x86_64/processor.c | 69 +++++++++++++++++++
> 3 files changed, 79 insertions(+), 1 deletion(-)
>
> diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h
> index 6a1a37f30494..da2b702da71a 100644
> --- a/tools/testing/selftests/kvm/include/kvm_util.h
> +++ b/tools/testing/selftests/kvm/include/kvm_util.h
> @@ -71,6 +71,15 @@ enum vm_guest_mode {
>
> #endif
>
> +#if defined(__x86_64__)
> +unsigned long vm_compute_max_gfn(struct kvm_vm *vm);
> +#else
> +static inline unsigned long vm_compute_max_gfn(struct kvm_vm *vm)
> +{
> + return ((1ULL << vm->pa_bits) >> vm->page_shift) - 1;
> +}

This breaks compiling on non-x86 architectures because of how we keep
the vm struct private. I'll send a patch that puts vm_compute_max_gfn
in lib/kvm_util.c but as a weak symbol.

(Maybe we should stop keeping the vm struct private...)

Thanks,
drew