[v4,10/12] KVM: x86/cpuid: Add X86_FEATURE_PERFMON_V2 as a scattered flag

Message ID 20230214050757.9623-11-likexu@tencent.com
State New
Headers
Series KVM: x86: Add AMD Guest PerfMonV2 PMU support |

Commit Message

Like Xu Feb. 14, 2023, 5:07 a.m. UTC
  From: Like Xu <likexu@tencent.com>

Considering that the core kernel may also want to know this flag, to avoid
confusion this needs to be a scattered flag rather than a pure KVM flag so
that KVM can redirect it to the hardware-defined bit position, which is the
role of __feature_translate() and KVM_X86_FEATURE_PERFMON_V2.

Suggested-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Like Xu <likexu@tencent.com>
---
 arch/x86/kvm/reverse_cpuid.h | 7 +++++++
 1 file changed, 7 insertions(+)
  

Comments

Sean Christopherson April 7, 2023, 12:41 a.m. UTC | #1
This shortlog is flat out wrong.  This patch does not add X86_FEATURE_PERFMON_V2,
it adds a KVM-only leaf to redirect X86_FEATURE_PERFMON_V2 to its architectural
location when KVM queries the leaf, e.g. via guest_cpuid_has().

On Tue, Feb 14, 2023, Like Xu wrote:
> From: Like Xu <likexu@tencent.com>
> 
> Considering that the core kernel may also want to know this flag, to avoid

Please write changelogs so that they do not depend on the shortlog for context.
I know the tip maintainers in particular like to make the changelog a continuation
of the shortlog, but I find it really annoying, especially when reviewing in mutt,
e.g. as I'm typing this, I don't even have the shortlog visible.

In the vast majority of cases, being more explicit rarely adds more than a few
chars, e.g. you'll save more by reducing the fluff in this changelog.

> confusion this needs to be a scattered flag rather than a pure KVM flag so
> that KVM can redirect it to the hardware-defined bit position, which is the
> role of __feature_translate() and KVM_X86_FEATURE_PERFMON_V2.

Like the shortlog, the changelog is best misleading, X86_FEATURE_PERFMON_V2 is
already a scattered flag.

  Define a KVM-only leaf for AMD's PerfMonV2 feature flag to redirect the
  kernel's scattered version to its architectural location, e.g. so that
  KVM can query guest support via guest_cpuid_has().
  

Patch

diff --git a/arch/x86/kvm/reverse_cpuid.h b/arch/x86/kvm/reverse_cpuid.h
index 4945456fd646..333e28b0a13c 100644
--- a/arch/x86/kvm/reverse_cpuid.h
+++ b/arch/x86/kvm/reverse_cpuid.h
@@ -15,6 +15,7 @@  enum kvm_only_cpuid_leafs {
 	CPUID_12_EAX	 = NCAPINTS,
 	CPUID_7_1_EDX,
 	CPUID_8000_0007_EDX,
+	CPUID_8000_0022_EAX,
 	NR_KVM_CPU_CAPS,
 
 	NKVMCAPINTS = NR_KVM_CPU_CAPS - NCAPINTS,
@@ -47,6 +48,9 @@  enum kvm_only_cpuid_leafs {
 /* CPUID level 0x80000007 (EDX). */
 #define KVM_X86_FEATURE_CONSTANT_TSC	KVM_X86_FEATURE(CPUID_8000_0007_EDX, 8)
 
+/* CPUID level 0x80000022 (EAX) */
+#define KVM_X86_FEATURE_PERFMON_V2	KVM_X86_FEATURE(CPUID_8000_0022_EAX, 0)
+
 struct cpuid_reg {
 	u32 function;
 	u32 index;
@@ -73,6 +77,7 @@  static const struct cpuid_reg reverse_cpuid[] = {
 	[CPUID_8000_001F_EAX] = {0x8000001f, 0, CPUID_EAX},
 	[CPUID_7_1_EDX]       = {         7, 1, CPUID_EDX},
 	[CPUID_8000_0007_EDX] = {0x80000007, 0, CPUID_EDX},
+	[CPUID_8000_0022_EAX] = {0x80000022, 0, CPUID_EAX},
 };
 
 /*
@@ -107,6 +112,8 @@  static __always_inline u32 __feature_translate(int x86_feature)
 		return KVM_X86_FEATURE_SGX_EDECCSSA;
 	else if (x86_feature == X86_FEATURE_CONSTANT_TSC)
 		return KVM_X86_FEATURE_CONSTANT_TSC;
+	else if (x86_feature == X86_FEATURE_PERFMON_V2)
+		return KVM_X86_FEATURE_PERFMON_V2;
 
 	return x86_feature;
 }