Skip to contentSkip to conversation

KVM: x86: Mask off unsupported and unknown bits of IA32_ARCH_CAPABILITIES

PATCH Applied Jim Mattson 3 participants 3 messages Aug 30, 2022 – Sep 1, 2022

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 KVM: x86: Mask off unsupported and unknown bits of IA32_ARCH_CAPABILITIES

KVM should not claim to virtualize unknown IA32_ARCH_CAPABILITIES bits. When kvm_get_arch_capabilities() was originally written, there were only a few bits defined in this MSR, and KVM could virtualize all of them. However, over the years, several bits have been defined that KVM cannot just blindly pass through to the guest without additional work (such as virtualizing an MSR promised by the IA32_ARCH_CAPABILITES feature bit).

Define a mask of supported IA32_ARCH_CAPABILITIES bits, and mask off any other bits that are set in the hardware MSR.

Cc Paolo Bonzini <[email protected]>
Fixes 5b76a3cff011 ("KVM: VMX: Tell the nested hypervisor to skip L1D flush on vmentry")

---

v1 -> v2: Clarified comment about unsupported bits.

 arch/x86/kvm/x86.c | 25 +++++++++++++++++++++----
 1 file changed, 21 insertions(+), 4 deletions(-)
arch/x86/kvm/x86.c +21 -4
index 205ebdc2b11b..9a18acfcfdc8 100644
@@ -1557,12 +1557,32 @@ static const u32 msr_based_features_all[] = {
 static u32 msr_based_features[ARRAY_SIZE(msr_based_features_all)];
 static unsigned int num_msr_based_features;
 
+/*
+ * Some IA32_ARCH_CAPABILITIES bits have dependencies on MSRs that KVM
+ * does not yet virtualize. These include:
+ *   10 - MISC_PACKAGE_CTRLS
+ *   11 - ENERGY_FILTERING_CTL
+ *   12 - DOITM
+ *   18 - FB_CLEAR_CTRL
+ *   21 - XAPIC_DISABLE_STATUS
+ *   23 - OVERCLOCKING_STATUS
+ */
+
+#define KVM_SUPPORTED_ARCH_CAP \
+	(ARCH_CAP_RDCL_NO | ARCH_CAP_IBRS_ALL | ARCH_CAP_RSBA | \
+	 ARCH_CAP_SKIP_VMENTRY_L1DFLUSH | ARCH_CAP_SSB_NO | ARCH_CAP_MDS_NO | \
+	 ARCH_CAP_PSCHANGE_MC_NO | ARCH_CAP_TSX_CTRL_MSR | ARCH_CAP_TAA_NO | \
+	 ARCH_CAP_SBDR_SSDP_NO | ARCH_CAP_FBSDP_NO | ARCH_CAP_PSDP_NO | \
+	 ARCH_CAP_FB_CLEAR | ARCH_CAP_RRSBA | ARCH_CAP_PBRSB_NO)
+
 static u64 kvm_get_arch_capabilities(void)
 {
 	u64 data = 0;
 
-	if (boot_cpu_has(X86_FEATURE_ARCH_CAPABILITIES))
+	if (boot_cpu_has(X86_FEATURE_ARCH_CAPABILITIES)) {
 		rdmsrl(MSR_IA32_ARCH_CAPABILITIES, data);
+		data &= KVM_SUPPORTED_ARCH_CAP;
+	}
 
 	/*
 	 * If nx_huge_pages is enabled, KVM's shadow paging will ensure that
@@ -1610,9 +1630,6 @@ static u64 kvm_get_arch_capabilities(void)
 		 */
 	}
 
-	/* Guests don't need to know "Fill buffer clear control" exists */
-	data &= ~ARCH_CAP_FB_CLEAR_CTRL;
-
 	return data;
 }
 
Signature · 1 line
2.37.2.672.g94769d06f0-goog
Reviewed-by on v2 1/1
#1 lore raw
Quoted 10 lines · Jim Mattson

KVM should not claim to virtualize unknown IA32_ARCH_CAPABILITIES bits. When kvm_get_arch_capabilities() was originally written, there were only a few bits defined in this MSR, and KVM could virtualize all of them. However, over the years, several bits have been defined that KVM cannot just blindly pass through to the guest without additional work (such as virtualizing an MSR promised by the IA32_ARCH_CAPABILITES feature bit).

Define a mask of supported IA32_ARCH_CAPABILITIES bits, and mask off any other bits that are set in the hardware MSR.

Quoted 59 lines
Cc Paolo Bonzini <[email protected]>
Fixes 5b76a3cff011 ("KVM: VMX: Tell the nested hypervisor to skip L1D flush on vmentry")

---

v1 -> v2: Clarified comment about unsupported bits.

  arch/x86/kvm/x86.c | 25 +++++++++++++++++++++----
  1 file changed, 21 insertions(+), 4 deletions(-)
arch/x86/kvm/x86.c +21 -4
index 205ebdc2b11b..9a18acfcfdc8 100644
@@ -1557,12 +1557,32 @@ static const u32 msr_based_features_all[] = {
  static u32 msr_based_features[ARRAY_SIZE(msr_based_features_all)];
  static unsigned int num_msr_based_features;
 
+/*
+ * Some IA32_ARCH_CAPABILITIES bits have dependencies on MSRs that KVM
+ * does not yet virtualize. These include:
+ *   10 - MISC_PACKAGE_CTRLS
+ *   11 - ENERGY_FILTERING_CTL
+ *   12 - DOITM
+ *   18 - FB_CLEAR_CTRL
+ *   21 - XAPIC_DISABLE_STATUS
+ *   23 - OVERCLOCKING_STATUS
+ */
+
+#define KVM_SUPPORTED_ARCH_CAP \
+	(ARCH_CAP_RDCL_NO | ARCH_CAP_IBRS_ALL | ARCH_CAP_RSBA | \
+	 ARCH_CAP_SKIP_VMENTRY_L1DFLUSH | ARCH_CAP_SSB_NO | ARCH_CAP_MDS_NO | \
+	 ARCH_CAP_PSCHANGE_MC_NO | ARCH_CAP_TSX_CTRL_MSR | ARCH_CAP_TAA_NO | \
+	 ARCH_CAP_SBDR_SSDP_NO | ARCH_CAP_FBSDP_NO | ARCH_CAP_PSDP_NO | \
+	 ARCH_CAP_FB_CLEAR | ARCH_CAP_RRSBA | ARCH_CAP_PBRSB_NO)
+
  static u64 kvm_get_arch_capabilities(void)
  {
  	u64 data = 0;
 
-	if (boot_cpu_has(X86_FEATURE_ARCH_CAPABILITIES))
+	if (boot_cpu_has(X86_FEATURE_ARCH_CAPABILITIES)) {
  		rdmsrl(MSR_IA32_ARCH_CAPABILITIES, data);
+		data &= KVM_SUPPORTED_ARCH_CAP;
+	}
 
  	/*
  	 * If nx_huge_pages is enabled, KVM's shadow paging will ensure that
@@ -1610,9 +1630,6 @@ static u64 kvm_get_arch_capabilities(void)
  		 */
  	}
 
-	/* Guests don't need to know "Fill buffer clear control" exists */
-	data &= ~ARCH_CAP_FB_CLEAR_CTRL;
-
  	return data;
  }
 
maintainer Applied
#2 lore raw
In reply to Jim Mattson
Quoted 11 lines · Jim Mattson

KVM should not claim to virtualize unknown IA32_ARCH_CAPABILITIES bits. When kvm_get_arch_capabilities() was originally written, there were only a few bits defined in this MSR, and KVM could virtualize all of them. However, over the years, several bits have been defined that KVM cannot just blindly pass through to the guest without additional work (such as virtualizing an MSR promised by the IA32_ARCH_CAPABILITES feature bit).

Define a mask of supported IA32_ARCH_CAPABILITIES bits, and mask off any other bits that are set in the hardware MSR.

Quoted 7 lines
Cc Paolo Bonzini <[email protected]>
Fixes 5b76a3cff011 ("KVM: VMX: Tell the nested hypervisor to skip L1D flush on vmentry")

---

v1 -> v2: Clarified comment about unsupported bits.

Queued, thanks.

Paolo