Skip to contentSkip to conversation

kvm: x86: Improve emulation of CPUID leaves 0BH and 1FH

PATCH Applied Jim Mattson 3 participants 5 messages Sep 26, 2019

Subsystems (3)

v1 of 2 Applied Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 kvm: x86: Improve emulation of CPUID leaves 0BH and 1FH

For these CPUID leaves, the EDX output is not dependent on the ECX input (i.e. the SIGNIFCANT_INDEX flag doesn't apply to EDX). Furthermore, the low byte of the ECX output is always identical to the low byte of the ECX input. KVM does not produce the correct ECX and EDX outputs for any undefined subleaves beyond the first.

Special-case these CPUID leaves in kvm_cpuid, so that the ECX and EDX outputs are properly generated for all undefined subleaves.

Fixes 0771671749b59a ("KVM: Enhance guest cpuid management")
Fixes a87f2d3a6eadab ("KVM: x86: Add Intel CPUID.1F cpuid emulation support")
Cc Sean Christopherson <[email protected]>
Cc Paolo Bonzini <[email protected]>
---
 arch/x86/kvm/cpuid.c | 83 +++++++++++++++++++++++++-------------------
 1 file changed, 47 insertions(+), 36 deletions(-)
arch/x86/kvm/cpuid.c +47 -36
index dd5985eb61b4c..35e2f930a4b79 100644
@@ -963,53 +963,64 @@ struct kvm_cpuid_entry2 *kvm_find_cpuid_entry(struct kvm_vcpu *vcpu,
 EXPORT_SYMBOL_GPL(kvm_find_cpuid_entry);
 
 /*
- * If no match is found, check whether we exceed the vCPU's limit
- * and return the content of the highest valid _standard_ leaf instead.
- * This is to satisfy the CPUID specification.
+ * If the basic or extended CPUID leaf requested is higher than the
+ * maximum supported basic or extended leaf, respectively, then it is
+ * out of range.
  */
-static struct kvm_cpuid_entry2* check_cpuid_limit(struct kvm_vcpu *vcpu,
-                                                  u32 function, u32 index)
+static bool cpuid_function_in_range(struct kvm_vcpu *vcpu, u32 function)
 {
-	struct kvm_cpuid_entry2 *maxlevel;
-
-	maxlevel = kvm_find_cpuid_entry(vcpu, function & 0x80000000, 0);
-	if (!maxlevel || maxlevel->eax >= function)
-		return NULL;
-	if (function & 0x80000000) {
-		maxlevel = kvm_find_cpuid_entry(vcpu, 0, 0);
-		if (!maxlevel)
-			return NULL;
-	}
-	return kvm_find_cpuid_entry(vcpu, maxlevel->eax, index);
+	struct kvm_cpuid_entry2 *max;
+
+	max = kvm_find_cpuid_entry(vcpu, function & 0x80000000, 0);
+	return max && function <= max->eax;
 }
 
 bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
 	       u32 *ecx, u32 *edx, bool check_limit)
 {
 	u32 function = *eax, index = *ecx;
-	struct kvm_cpuid_entry2 *best;
-	bool entry_found = true;
-
-	best = kvm_find_cpuid_entry(vcpu, function, index);
-
-	if (!best) {
-		entry_found = false;
-		if (!check_limit)
-			goto out;
+	struct kvm_cpuid_entry2 *entry;
+	struct kvm_cpuid_entry2 *max;
+	bool found;
 
-		best = check_cpuid_limit(vcpu, function, index);
+	entry = kvm_find_cpuid_entry(vcpu, function, index);
+	found = entry;
+	/*
+	 * Intel CPUID semantics treats any query for an out-of-range
+	 * leaf as if the highest basic leaf (i.e. CPUID.0H:EAX) were
+	 * requested.
+	 */
+	if (!entry && check_limit && !cpuid_function_in_range(vcpu, function)) {
+		max = kvm_find_cpuid_entry(vcpu, 0, 0);
+		if (max) {
+			function = max->eax;
+			entry = kvm_find_cpuid_entry(vcpu, function, index);
+		}
 	}
-
-out:
-	if (best) {
-		*eax = best->eax;
-		*ebx = best->ebx;
-		*ecx = best->ecx;
-		*edx = best->edx;
-	} else
+	if (entry) {
+		*eax = entry->eax;
+		*ebx = entry->ebx;
+		*ecx = entry->ecx;
+		*edx = entry->edx;
+	} else {
 		*eax = *ebx = *ecx = *edx = 0;
-	trace_kvm_cpuid(function, *eax, *ebx, *ecx, *edx, entry_found);
-	return entry_found;
+		/*
+		 * When leaf 0BH or 1FH is defined, CL is pass-through
+		 * and EDX is always the x2APIC ID, even for undefined
+		 * subleaves. Index 1 will exist iff the leaf is
+		 * implemented, so we pass through CL iff leaf 1
+		 * exists. EDX can be copied from any existing index.
+		 */
+		if (function == 0xb || function == 0x1f) {
+			entry = kvm_find_cpuid_entry(vcpu, function, 1);
+			if (entry) {
+				*ecx = index & 0xff;
+				*edx = entry->edx;
+			}
+		}
+	}
+	trace_kvm_cpuid(function, *eax, *ebx, *ecx, *edx, found);
+	return found;
 }
 EXPORT_SYMBOL_GPL(kvm_cpuid);
 
Signature · 1 line
2.23.0.351.gc4317032e6-goog
#1 lore raw

PATCH 2/2 kvm: x86: Use AMD CPUID semantics for AMD vCPUs

When the guest CPUID information represents an AMD vCPU, return all zeroes for queries of undefined CPUID leaves, whether or not they are in range.

Fixes bd22f5cfcfe8f6 ("KVM: move and fix substitue search for missing CPUID entries")
Cc Sean Christopherson <[email protected]>
Cc Paolo Bonzini <[email protected]>
---
 arch/x86/kvm/cpuid.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kvm/cpuid.c +4 -2
index 35e2f930a4b79..0377d2820a7aa 100644
@@ -988,9 +988,11 @@ bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
 	/*
 	 * Intel CPUID semantics treats any query for an out-of-range
 	 * leaf as if the highest basic leaf (i.e. CPUID.0H:EAX) were
-	 * requested.
+	 * requested. AMD CPUID semantics returns all zeroes for any
+	 * undefined leaf, whether or not the leaf is in range.
 	 */
-	if (!entry && check_limit && !cpuid_function_in_range(vcpu, function)) {
+	if (!entry && check_limit && !guest_cpuid_is_amd(vcpu) &&
+	    !cpuid_function_in_range(vcpu, function)) {
 		max = kvm_find_cpuid_entry(vcpu, 0, 0);
 		if (max) {
 			function = max->eax;
Signature · 1 line
2.23.0.351.gc4317032e6-goog
#2 lore raw

Re: [PATCH 2/2] kvm: x86: Use AMD CPUID semantics for AMD vCPUs

Quoted 20 lines · Jim Mattson

When the guest CPUID information represents an AMD vCPU, return all zeroes for queries of undefined CPUID leaves, whether or not they are in range.

Fixes bd22f5cfcfe8f6 ("KVM: move and fix substitue search for missing CPUID entries")
Cc Sean Christopherson <[email protected]>
Cc Paolo Bonzini <[email protected]>
---
  arch/x86/kvm/cpuid.c | 6 ++++--
  1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kvm/cpuid.c +0 -0
index 35e2f930a4b79..0377d2820a7aa 100644

arch/x86/kvm/cpuid.c bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx, · line 995

Quoted 10 lines
arch/x86/kvm/cpuid.c +4 -2
@@ -988,9 +988,11 @@ bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
  	/*
  	 * Intel CPUID semantics treats any query for an out-of-range
  	 * leaf as if the highest basic leaf (i.e. CPUID.0H:EAX) were
-	 * requested.
+	 * requested. AMD CPUID semantics returns all zeroes for any
+	 * undefined leaf, whether or not the leaf is in range.
  	 */
-	if (!entry && check_limit && !cpuid_function_in_range(vcpu, function)) {
+	if (!entry && check_limit && !guest_cpuid_is_amd(vcpu) &&
+	    !cpuid_function_in_range(vcpu, function)) {

IIUC, the parameter check_limit is to indicate whether return highest basic leaf when out-of-range. Here you just makes check_limit meaningless.

Maybe we can do like this to use check_limit reasonably:

arch/x86/kvm/cpuid.c +2 -1
index 0377d2820a7a..e6a61f3f6c0c 100644
@@ -1035,7 +1035,8 @@ int kvm_emulate_cpuid(struct kvm_vcpu *vcpu)
 
        eax = kvm_rax_read(vcpu);
        ecx = kvm_rcx_read(vcpu);
-       kvm_cpuid(vcpu, &eax, &ebx, &ecx, &edx, true);
+       kvm_cpuid(vcpu, &eax, &ebx, &ecx, &edx,
+                       guest_cpuid_is_amd(vcpu) ? false: true);
        kvm_rax_write(vcpu, eax);
        kvm_rbx_write(vcpu, ebx);
        kvm_rcx_write(vcpu, ecx);
Quoted 4 lines
arch/x86/kvm/cpuid.c +0 -0
  		max = kvm_find_cpuid_entry(vcpu, 0, 0);
  		if (max) {
  			function = max->eax;
#3 lore raw

Re: [PATCH 2/2] kvm: x86: Use AMD CPUID semantics for AMD vCPUs

Quoted 27 lines · Xiaoyao Li
Quoted 25 lines · Jim Mattson

When the guest CPUID information represents an AMD vCPU, return all zeroes for queries of undefined CPUID leaves, whether or not they are in range.

Fixes bd22f5cfcfe8f6 ("KVM: move and fix substitue search for missing CPUID entries")
Cc Sean Christopherson <[email protected]>
Cc Paolo Bonzini <[email protected]>
---
  arch/x86/kvm/cpuid.c | 6 ++++--
  1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kvm/cpuid.c +1 -1
index 35e2f930a4b79..0377d2820a7aa 100644
@@ -988,9 +988,11 @@ bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
      /*
       * Intel CPUID semantics treats any query for an out-of-range
       * leaf as if the highest basic leaf (i.e. CPUID.0H:EAX) were
-      * requested.
+      * requested. AMD CPUID semantics returns all zeroes for any
Quoted 8 lines

arch/x86/kvm/cpuid.c bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx, · line 995

Quoted 5 lines
arch/x86/kvm/cpuid.c +3 -1
+      * undefined leaf, whether or not the leaf is in range.
       */
-     if (!entry && check_limit && !cpuid_function_in_range(vcpu, function)) {
+     if (!entry && check_limit && !guest_cpuid_is_amd(vcpu) &&
+         !cpuid_function_in_range(vcpu, function)) {

IIUC, the parameter check_limit is to indicate whether return highest basic leaf when out-of-range. Here you just makes check_limit meaningless.

That's right. For AMD CPUID semantics, there is no need for check_limit.

Quoted 17 lines

Maybe we can do like this to use check_limit reasonably:

arch/x86/kvm/cpuid.c +2 -1
index 0377d2820a7a..e6a61f3f6c0c 100644
@@ -1035,7 +1035,8 @@ int kvm_emulate_cpuid(struct kvm_vcpu *vcpu)
 
         eax = kvm_rax_read(vcpu);
         ecx = kvm_rcx_read(vcpu);
-       kvm_cpuid(vcpu, &eax, &ebx, &ecx, &edx, true);
+       kvm_cpuid(vcpu, &eax, &ebx, &ecx, &edx,
+                       guest_cpuid_is_amd(vcpu) ? false: true);
         kvm_rax_write(vcpu, eax);
         kvm_rbx_write(vcpu, ebx);
         kvm_rcx_write(vcpu, ecx);
Quoted 3 lines

arch/x86/kvm/cpuid.c bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx, · line 998

Quoted 3 lines
arch/x86/kvm/cpuid.c +0 -0
              max = kvm_find_cpuid_entry(vcpu, 0, 0);
              if (max) {
                      function = max->eax;

Since over-limit CPUID queries should be rare, it seems unfortunate to pay the cost of guest_cpuid_is_amd() for every emulated CPUID instruction.

maintainer Applied
#4 lore raw

Re: [PATCH 2/2] kvm: x86: Use AMD CPUID semantics for AMD vCPUs

In reply to Jim Mattson
Quoted 24 lines · Jim Mattson

When the guest CPUID information represents an AMD vCPU, return all zeroes for queries of undefined CPUID leaves, whether or not they are in range.

Fixes bd22f5cfcfe8f6 ("KVM: move and fix substitue search for missing CPUID entries")
Cc Sean Christopherson <[email protected]>
Cc Paolo Bonzini <[email protected]>
---
 arch/x86/kvm/cpuid.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kvm/cpuid.c +0 -0
index 35e2f930a4b79..0377d2820a7aa 100644
@@ -988,9 +988,11 @@ bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
 	/*
 	 * Intel CPUID semantics treats any query for an out-of-range
 	 * leaf as if the highest basic leaf (i.e. CPUID.0H:EAX) were

arch/x86/kvm/cpuid.c bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx, · line 998

Quoted 10 lines
arch/x86/kvm/cpuid.c +4 -2
-	 * requested.
+	 * requested. AMD CPUID semantics returns all zeroes for any
+	 * undefined leaf, whether or not the leaf is in range.
 	 */
-	if (!entry && check_limit && !cpuid_function_in_range(vcpu, function)) {
+	if (!entry && check_limit && !guest_cpuid_is_amd(vcpu) &&
+	    !cpuid_function_in_range(vcpu, function)) {
 		max = kvm_find_cpuid_entry(vcpu, 0, 0);
 		if (max) {
 			function = max->eax;

Queued both, thanks.

Paolo