Skip to contentSkip to conversation

KVM: x86: vcpu_idx related cleanups

PATCH In review Sean Christopherson 5 participants 10 messages Sep 10, 2021 – Sep 22, 2021

Subsystems (4)
lore
Share & tools
lore

Sean Christopherson

author maintainer
#0 lore raw

cover KVM: x86: vcpu_idx related cleanups

Drop KVM's obsolete kvm_vcpu_get_idx() wrapper, and use vcpu_idx to identify vCPU0 when updating guest time.

Sean Christopherson (2):
KVM: x86: Query vcpu->vcpu_idx directly and drop its accessor
KVM: x86: Identify vCPU0 by its vcpu_idx instead of walking vCPUs
array

 arch/x86/kvm/hyperv.c    | 7 +++----
 arch/x86/kvm/hyperv.h    | 2 +-
 arch/x86/kvm/x86.c       | 2 +-
 include/linux/kvm_host.h | 5 -----
 4 files changed, 5 insertions(+), 11 deletions(-)
Signature · 1 line
2.33.0.309.g3052b89438-goog
author maintainer
#1 lore raw

PATCH 1/2 KVM: x86: Query vcpu->vcpu_idx directly and drop its accessor

Read vcpu->vcpu_idx directly instead of bouncing through the one-line wrapper, kvm_vcpu_get_idx(), and drop the wrapper. The wrapper is a remnant of the original implementation and serves no purpose; remove it before it gains more users.

Back when kvm_vcpu_get_idx() was added by commit 497d72d80a78 ("KVM: Add kvm_vcpu_get_idx to get vcpu index in kvm->vcpus"), the implementation was more than just a simple wrapper as vcpu->vcpu_idx did not exist and retrieving the index meant walking over the vCPU array to find the given vCPU.

When vcpu_idx was introduced by commit 8750e72a79dd ("KVM: remember position in kvm->vcpus array"), the helper was left behind, likely to avoid extra thrash (but even then there were only two users, the original arm usage having been removed at some point in the past).

No functional change intended.

---
 arch/x86/kvm/hyperv.c    | 7 +++----
 arch/x86/kvm/hyperv.h    | 2 +-
 include/linux/kvm_host.h | 5 -----
 3 files changed, 4 insertions(+), 10 deletions(-)
arch/x86/kvm/hyperv.c +3 -4
index fe4a02715266..04dbc001f4fc 100644
@@ -939,7 +939,7 @@ static int kvm_hv_vcpu_init(struct kvm_vcpu *vcpu)
 	for (i = 0; i < ARRAY_SIZE(hv_vcpu->stimer); i++)
 		stimer_init(&hv_vcpu->stimer[i], i);
 
-	hv_vcpu->vp_index = kvm_vcpu_get_idx(vcpu);
+	hv_vcpu->vp_index = vcpu->vcpu_idx;
 
 	return 0;
 }
@@ -1444,7 +1444,6 @@ static int kvm_hv_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host)
 	switch (msr) {
 	case HV_X64_MSR_VP_INDEX: {
 		struct kvm_hv *hv = to_kvm_hv(vcpu->kvm);
-		int vcpu_idx = kvm_vcpu_get_idx(vcpu);
 		u32 new_vp_index = (u32)data;
 
 		if (!host || new_vp_index >= KVM_MAX_VCPUS)
@@ -1459,9 +1458,9 @@ static int kvm_hv_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host)
 		 * VP index is changing, adjust num_mismatched_vp_indexes if
 		 * it now matches or no longer matches vcpu_idx.
 		 */
-		if (hv_vcpu->vp_index == vcpu_idx)
+		if (hv_vcpu->vp_index == vcpu->vcpu_idx)
 			atomic_inc(&hv->num_mismatched_vp_indexes);
-		else if (new_vp_index == vcpu_idx)
+		else if (new_vp_index == vcpu->vcpu_idx)
 			atomic_dec(&hv->num_mismatched_vp_indexes);
 
 		hv_vcpu->vp_index = new_vp_index;
arch/x86/kvm/hyperv.h +1 -1
index 730da8537d05..ed1c4e546d04 100644
@@ -83,7 +83,7 @@ static inline u32 kvm_hv_get_vpindex(struct kvm_vcpu *vcpu)
 {
 	struct kvm_vcpu_hv *hv_vcpu = to_hv_vcpu(vcpu);
 
-	return hv_vcpu ? hv_vcpu->vp_index : kvm_vcpu_get_idx(vcpu);
+	return hv_vcpu ? hv_vcpu->vp_index : vcpu->vcpu_idx;
 }
 
 int kvm_hv_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host);
include/linux/kvm_host.h +0 -5
index e4d712e9f760..31071ad821e2 100644
@@ -721,11 +721,6 @@ static inline struct kvm_vcpu *kvm_get_vcpu_by_id(struct kvm *kvm, int id)
 	return NULL;
 }
 
-static inline int kvm_vcpu_get_idx(struct kvm_vcpu *vcpu)
-{
-	return vcpu->vcpu_idx;
-}
-
 #define kvm_for_each_memslot(memslot, slots)				\
 	for (memslot = &slots->memslots[0];				\
 	     memslot < slots->memslots + slots->used_slots; memslot++)	\
Signature · 1 line
2.33.0.309.g3052b89438-goog
Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] KVM: x86: Query vcpu->vcpu_idx directly and drop its accessor

Quoted 78 lines · Sean Christopherson

Read vcpu->vcpu_idx directly instead of bouncing through the one-line wrapper, kvm_vcpu_get_idx(), and drop the wrapper. The wrapper is a remnant of the original implementation and serves no purpose; remove it before it gains more users.

Back when kvm_vcpu_get_idx() was added by commit 497d72d80a78 ("KVM: Add kvm_vcpu_get_idx to get vcpu index in kvm->vcpus"), the implementation was more than just a simple wrapper as vcpu->vcpu_idx did not exist and retrieving the index meant walking over the vCPU array to find the given vCPU.

When vcpu_idx was introduced by commit 8750e72a79dd ("KVM: remember position in kvm->vcpus array"), the helper was left behind, likely to avoid extra thrash (but even then there were only two users, the original arm usage having been removed at some point in the past).

No functional change intended.

---
 arch/x86/kvm/hyperv.c    | 7 +++----
 arch/x86/kvm/hyperv.h    | 2 +-
 include/linux/kvm_host.h | 5 -----
 3 files changed, 4 insertions(+), 10 deletions(-)
arch/x86/kvm/hyperv.c +3 -4
index fe4a02715266..04dbc001f4fc 100644
@@ -939,7 +939,7 @@ static int kvm_hv_vcpu_init(struct kvm_vcpu *vcpu)
 	for (i = 0; i < ARRAY_SIZE(hv_vcpu->stimer); i++)
 		stimer_init(&hv_vcpu->stimer[i], i);
 
-	hv_vcpu->vp_index = kvm_vcpu_get_idx(vcpu);
+	hv_vcpu->vp_index = vcpu->vcpu_idx;
 
 	return 0;
 }
@@ -1444,7 +1444,6 @@ static int kvm_hv_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host)
 	switch (msr) {
 	case HV_X64_MSR_VP_INDEX: {
 		struct kvm_hv *hv = to_kvm_hv(vcpu->kvm);
-		int vcpu_idx = kvm_vcpu_get_idx(vcpu);
 		u32 new_vp_index = (u32)data;
 
 		if (!host || new_vp_index >= KVM_MAX_VCPUS)
@@ -1459,9 +1458,9 @@ static int kvm_hv_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host)
 		 * VP index is changing, adjust num_mismatched_vp_indexes if
 		 * it now matches or no longer matches vcpu_idx.
 		 */
-		if (hv_vcpu->vp_index == vcpu_idx)
+		if (hv_vcpu->vp_index == vcpu->vcpu_idx)
 			atomic_inc(&hv->num_mismatched_vp_indexes);
-		else if (new_vp_index == vcpu_idx)
+		else if (new_vp_index == vcpu->vcpu_idx)
 			atomic_dec(&hv->num_mismatched_vp_indexes);
 
 		hv_vcpu->vp_index = new_vp_index;
arch/x86/kvm/hyperv.h +1 -1
index 730da8537d05..ed1c4e546d04 100644
@@ -83,7 +83,7 @@ static inline u32 kvm_hv_get_vpindex(struct kvm_vcpu *vcpu)
 {
 	struct kvm_vcpu_hv *hv_vcpu = to_hv_vcpu(vcpu);
 
-	return hv_vcpu ? hv_vcpu->vp_index : kvm_vcpu_get_idx(vcpu);
+	return hv_vcpu ? hv_vcpu->vp_index : vcpu->vcpu_idx;
 }
 
 int kvm_hv_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host);
include/linux/kvm_host.h +0 -0
index e4d712e9f760..31071ad821e2 100644
@@ -721,11 +721,6 @@ static inline struct kvm_vcpu *kvm_get_vcpu_by_id(struct kvm *kvm, int id)
 	return NULL;

include/linux/kvm_host.h static inline struct kvm_vcpu *kvm_get_vcpu_by_id(struct kvm *kvm, int id) · line 726

Quoted 10 lines
include/linux/kvm_host.h +0 -5
 }
 
-static inline int kvm_vcpu_get_idx(struct kvm_vcpu *vcpu)
-{
-	return vcpu->vcpu_idx;
-}
-
 #define kvm_for_each_memslot(memslot, slots)				\
 	for (memslot = &slots->memslots[0];				\
 	     memslot < slots->memslots + slots->used_slots; memslot++)	\

Best regards,
Maxim Levitsky

maintainer Reviewed-by on v1 1/2
#3 lore raw

Re: [PATCH 1/2] KVM: x86: Query vcpu->vcpu_idx directly and drop its accessor

In reply to Sean Christopherson
Quoted 78 lines · Sean Christopherson

Read vcpu->vcpu_idx directly instead of bouncing through the one-line wrapper, kvm_vcpu_get_idx(), and drop the wrapper. The wrapper is a remnant of the original implementation and serves no purpose; remove it before it gains more users.

Back when kvm_vcpu_get_idx() was added by commit 497d72d80a78 ("KVM: Add kvm_vcpu_get_idx to get vcpu index in kvm->vcpus"), the implementation was more than just a simple wrapper as vcpu->vcpu_idx did not exist and retrieving the index meant walking over the vCPU array to find the given vCPU.

When vcpu_idx was introduced by commit 8750e72a79dd ("KVM: remember position in kvm->vcpus array"), the helper was left behind, likely to avoid extra thrash (but even then there were only two users, the original arm usage having been removed at some point in the past).

No functional change intended.

---
 arch/x86/kvm/hyperv.c    | 7 +++----
 arch/x86/kvm/hyperv.h    | 2 +-
 include/linux/kvm_host.h | 5 -----
 3 files changed, 4 insertions(+), 10 deletions(-)
arch/x86/kvm/hyperv.c +3 -4
index fe4a02715266..04dbc001f4fc 100644
@@ -939,7 +939,7 @@ static int kvm_hv_vcpu_init(struct kvm_vcpu *vcpu)
 	for (i = 0; i < ARRAY_SIZE(hv_vcpu->stimer); i++)
 		stimer_init(&hv_vcpu->stimer[i], i);
 
-	hv_vcpu->vp_index = kvm_vcpu_get_idx(vcpu);
+	hv_vcpu->vp_index = vcpu->vcpu_idx;
 
 	return 0;
 }
@@ -1444,7 +1444,6 @@ static int kvm_hv_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host)
 	switch (msr) {
 	case HV_X64_MSR_VP_INDEX: {
 		struct kvm_hv *hv = to_kvm_hv(vcpu->kvm);
-		int vcpu_idx = kvm_vcpu_get_idx(vcpu);
 		u32 new_vp_index = (u32)data;
 
 		if (!host || new_vp_index >= KVM_MAX_VCPUS)
@@ -1459,9 +1458,9 @@ static int kvm_hv_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host)
 		 * VP index is changing, adjust num_mismatched_vp_indexes if
 		 * it now matches or no longer matches vcpu_idx.
 		 */
-		if (hv_vcpu->vp_index == vcpu_idx)
+		if (hv_vcpu->vp_index == vcpu->vcpu_idx)
 			atomic_inc(&hv->num_mismatched_vp_indexes);
-		else if (new_vp_index == vcpu_idx)
+		else if (new_vp_index == vcpu->vcpu_idx)
 			atomic_dec(&hv->num_mismatched_vp_indexes);
 
 		hv_vcpu->vp_index = new_vp_index;
arch/x86/kvm/hyperv.h +1 -1
index 730da8537d05..ed1c4e546d04 100644
@@ -83,7 +83,7 @@ static inline u32 kvm_hv_get_vpindex(struct kvm_vcpu *vcpu)
 {
 	struct kvm_vcpu_hv *hv_vcpu = to_hv_vcpu(vcpu);
 
-	return hv_vcpu ? hv_vcpu->vp_index : kvm_vcpu_get_idx(vcpu);
+	return hv_vcpu ? hv_vcpu->vp_index : vcpu->vcpu_idx;
 }
 
 int kvm_hv_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data, bool host);
include/linux/kvm_host.h +0 -0
index e4d712e9f760..31071ad821e2 100644
@@ -721,11 +721,6 @@ static inline struct kvm_vcpu *kvm_get_vcpu_by_id(struct kvm *kvm, int id)
 	return NULL;

include/linux/kvm_host.h static inline struct kvm_vcpu *kvm_get_vcpu_by_id(struct kvm *kvm, int id) · line 726

Quoted 10 lines
include/linux/kvm_host.h +0 -5
 }
 
-static inline int kvm_vcpu_get_idx(struct kvm_vcpu *vcpu)
-{
-	return vcpu->vcpu_idx;
-}
-
 #define kvm_for_each_memslot(memslot, slots)				\
 	for (memslot = &slots->memslots[0];				\
 	     memslot < slots->memslots + slots->used_slots; memslot++)	\
Signature · 1 line
Vitaly
author maintainer
#4 lore raw

PATCH 2/2 KVM: x86: Identify vCPU0 by its vcpu_idx instead of walking vCPUs array

In reply to Sean Christopherson

Use vcpu_idx to identify vCPU0 when updating HyperV's TSC page, which is shared by all vCPUs and "owned" by vCPU0 (because vCPU0 is the only vCPU that's guaranteed to exist). Using kvm_get_vcpu() to find vCPU works, but it's a rather odd and suboptimal method to check the index of a given vCPU.

No functional change intended.

---
 arch/x86/kvm/x86.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +1 -1
index 86539c1686fa..6ab851df08d1 100644
@@ -2969,7 +2969,7 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 				       offsetof(struct compat_vcpu_info, time));
 	if (vcpu->xen.vcpu_time_info_set)
 		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
-	if (v == kvm_get_vcpu(v->kvm, 0))
+	if (!v->vcpu_idx)
 		kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
 	return 0;
 }
Signature · 1 line
2.33.0.309.g3052b89438-goog
Reviewed-by on v1 2/2
#5 lore raw

Re: [PATCH 2/2] KVM: x86: Identify vCPU0 by its vcpu_idx instead of walking vCPUs array

Quoted 10 lines · Sean Christopherson

Use vcpu_idx to identify vCPU0 when updating HyperV's TSC page, which is shared by all vCPUs and "owned" by vCPU0 (because vCPU0 is the only vCPU that's guaranteed to exist). Using kvm_get_vcpu() to find vCPU works, but it's a rather odd and suboptimal method to check the index of a given vCPU.

No functional change intended.

Reviewed-by on v1 2/2
#6 lore raw

Re: [PATCH 2/2] KVM: x86: Identify vCPU0 by its vcpu_idx instead of walking vCPUs array

In reply to Sean Christopherson
Quoted 18 lines · Sean Christopherson

Use vcpu_idx to identify vCPU0 when updating HyperV's TSC page, which is shared by all vCPUs and "owned" by vCPU0 (because vCPU0 is the only vCPU that's guaranteed to exist). Using kvm_get_vcpu() to find vCPU works, but it's a rather odd and suboptimal method to check the index of a given vCPU.

No functional change intended.

---
 arch/x86/kvm/x86.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +0 -0
index 86539c1686fa..6ab851df08d1 100644

arch/x86/kvm/x86.c static int kvm_guest_time_update(struct kvm_vcpu *v) · line 2975

Quoted 8 lines
arch/x86/kvm/x86.c +1 -1
@@ -2969,7 +2969,7 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 				       offsetof(struct compat_vcpu_info, time));
 	if (vcpu->xen.vcpu_time_info_set)
 		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
-	if (v == kvm_get_vcpu(v->kvm, 0))
+	if (!v->vcpu_idx)
 		kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
 	return 0;
 }

Best regards,
Maxim Levitsky

maintainer Reviewed-by on v1 2/2
#7 lore raw

Re: [PATCH 2/2] KVM: x86: Identify vCPU0 by its vcpu_idx instead of walking vCPUs array

In reply to Sean Christopherson
Quoted 18 lines · Sean Christopherson

Use vcpu_idx to identify vCPU0 when updating HyperV's TSC page, which is shared by all vCPUs and "owned" by vCPU0 (because vCPU0 is the only vCPU that's guaranteed to exist). Using kvm_get_vcpu() to find vCPU works, but it's a rather odd and suboptimal method to check the index of a given vCPU.

No functional change intended.

---
 arch/x86/kvm/x86.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +0 -0
index 86539c1686fa..6ab851df08d1 100644

arch/x86/kvm/x86.c static int kvm_guest_time_update(struct kvm_vcpu *v) · line 2975

Quoted 8 lines
arch/x86/kvm/x86.c +1 -1
@@ -2969,7 +2969,7 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 				       offsetof(struct compat_vcpu_info, time));
 	if (vcpu->xen.vcpu_time_info_set)
 		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
-	if (v == kvm_get_vcpu(v->kvm, 0))
+	if (!v->vcpu_idx)
 		kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
 	return 0;
 }

" ... instead of walking vCPUs array" in the Subject line is a bit confusing because kvm_get_vcpu() doesn't actually walk anything, it just returns 'kvm->vcpus[i]' after checking that we actually have that many vCPUs. The patch itself is OK, so

Signature · 1 line
Vitaly
author maintainer
#8 lore raw

Re: [PATCH 2/2] KVM: x86: Identify vCPU0 by its vcpu_idx instead of walking vCPUs array

Quoted 25 lines · Vitaly Kuznetsov
Quoted 23 lines · Sean Christopherson

Use vcpu_idx to identify vCPU0 when updating HyperV's TSC page, which is shared by all vCPUs and "owned" by vCPU0 (because vCPU0 is the only vCPU that's guaranteed to exist). Using kvm_get_vcpu() to find vCPU works, but it's a rather odd and suboptimal method to check the index of a given vCPU.

No functional change intended.

---
 arch/x86/kvm/x86.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +1 -1
index 86539c1686fa..6ab851df08d1 100644
@@ -2969,7 +2969,7 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
 				       offsetof(struct compat_vcpu_info, time));
 	if (vcpu->xen.vcpu_time_info_set)
 		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
-	if (v == kvm_get_vcpu(v->kvm, 0))
+	if (!v->vcpu_idx)
Quoted 8 lines

arch/x86/kvm/x86.c static int kvm_guest_time_update(struct kvm_vcpu *v) · line 2975

Quoted 3 lines
arch/x86/kvm/x86.c +0 -0
 		kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
 	return 0;
 }

" ... instead of walking vCPUs array" in the Subject line is a bit confusing because kvm_get_vcpu() doesn't actually walk anything, it just returns 'kvm->vcpus[i]' after checking that we actually have that many vCPUs. The patch itself is OK, so

Argh, yes, I have a feeling I wrote the changelog after digging into the history of kvm_get_vcpu().

Paolo, can you tweak the shortlog to:

KVM: x86: Identify vCPU0 by its vcpu_idx instead of its vCPUs array entry

maintainer
#9 lore raw

Re: [PATCH 2/2] KVM: x86: Identify vCPU0 by its vcpu_idx instead of walking vCPUs array

Quoted 35 lines · Sean Christopherson
Quoted 25 lines · Vitaly Kuznetsov
Quoted 23 lines · Sean Christopherson

Use vcpu_idx to identify vCPU0 when updating HyperV's TSC page, which is shared by all vCPUs and "owned" by vCPU0 (because vCPU0 is the only vCPU that's guaranteed to exist). Using kvm_get_vcpu() to find vCPU works, but it's a rather odd and suboptimal method to check the index of a given vCPU.

No functional change intended.

---
  arch/x86/kvm/x86.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +1 -1
index 86539c1686fa..6ab851df08d1 100644
@@ -2969,7 +2969,7 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
  				       offsetof(struct compat_vcpu_info, time));
  	if (vcpu->xen.vcpu_time_info_set)
  		kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
-	if (v == kvm_get_vcpu(v->kvm, 0))
+	if (!v->vcpu_idx)
Quoted 8 lines

arch/x86/kvm/x86.c static int kvm_guest_time_update(struct kvm_vcpu *v) · line 2975

Quoted 3 lines
arch/x86/kvm/x86.c +0 -0
  		kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
  	return 0;
  }

" ... instead of walking vCPUs array" in the Subject line is a bit confusing because kvm_get_vcpu() doesn't actually walk anything, it just returns 'kvm->vcpus[i]' after checking that we actually have that many vCPUs. The patch itself is OK, so

Quoted 7 lines

Argh, yes, I have a feeling I wrote the changelog after digging into the history of kvm_get_vcpu().

Paolo, can you tweak the shortlog to:

KVM: x86: Identify vCPU0 by its vcpu_idx instead of its vCPUs array entry

Done and queued. Patch 1 required some further s390 changes.

Paolo