Skip to contentSkip to conversation

KVM: x86: use array_index_nospec with indices that come from guest

PATCH In review Greg Kroah-Hartman 3 participants 11 messages Jul 24, 2025 – Aug 15, 2025

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

PATCH 1/1 KVM: x86: use array_index_nospec with indices that come from guest

From: Thijs Raymakers <[email protected]>

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)
arch/x86/kvm/lapic.c +2 -0
index 73418dc0ebb2..e10d6ad236c9 100644
@@ -852,6 +852,8 @@ static int __pv_send_ipi(unsigned long *ipi_bitmap, struct kvm_apic_map *map,
 	if (min > map->max_apic_id)
 		return 0;
 
+	min = array_index_nospec(min, map->max_apic_id);
+
 	for_each_set_bit(i, ipi_bitmap,
 		min((u32)BITS_PER_LONG, (map->max_apic_id - min + 1))) {
 		if (map->phys_map[min + i]) {
arch/x86/kvm/x86.c +5 -2
index 93636f77c42d..872e43defa67 100644
@@ -10051,8 +10051,11 @@ static void kvm_sched_yield(struct kvm_vcpu *vcpu, unsigned long dest_id)
 	rcu_read_lock();
 	map = rcu_dereference(vcpu->kvm->arch.apic_map);
 
-	if (likely(map) && dest_id <= map->max_apic_id && map->phys_map[dest_id])
-		target = map->phys_map[dest_id]->vcpu;
+	if (likely(map) && dest_id <= map->max_apic_id) {
+		dest_id = array_index_nospec(dest_id, map->max_apic_id);
+		if (map->phys_map[dest_id])
+			target = map->phys_map[dest_id]->vcpu;
+	}
 
 	rcu_read_unlock();
 
Signature · 1 line
2.50.1
#1 lore raw
Quoted 21 lines · Greg Kroah-Hartman

From: Thijs Raymakers <[email protected]>

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)
arch/x86/kvm/lapic.c +0 -0
index 73418dc0ebb2..e10d6ad236c9 100644

arch/x86/kvm/lapic.c static int __pv_send_ipi(unsigned long *ipi_bitmap, struct kvm_apic_map *map, · line 855

Quoted 4 lines
arch/x86/kvm/lapic.c +1 -0
@@ -852,6 +852,8 @@ static int __pv_send_ipi(unsigned long *ipi_bitmap, struct kvm_apic_map *map,
 	if (min > map->max_apic_id)
 		return 0;
 
+	min = array_index_nospec(min, map->max_apic_id);

This is wrong, max_apic_id is inclusive, whereas array_index_nospec() takes a size/length as the second argument. I.e. this needs to be:

	min = array_index_nospec(min, map->max_apic_id + 1);
Quoted 9 lines
arch/x86/kvm/lapic.c +1 -0
+
 	for_each_set_bit(i, ipi_bitmap,
 		min((u32)BITS_PER_LONG, (map->max_apic_id - min + 1))) {
 		if (map->phys_map[min + i]) {
arch/x86/kvm/x86.c +0 -0
index 93636f77c42d..872e43defa67 100644

arch/x86/kvm/x86.c static void kvm_sched_yield(struct kvm_vcpu *vcpu, unsigned long dest_id) · line 10055

Quoted 7 lines
arch/x86/kvm/x86.c +2 -2
@@ -10051,8 +10051,11 @@ static void kvm_sched_yield(struct kvm_vcpu *vcpu, unsigned long dest_id)
 	rcu_read_lock();
 	map = rcu_dereference(vcpu->kvm->arch.apic_map);
 
-	if (likely(map) && dest_id <= map->max_apic_id && map->phys_map[dest_id])
-		target = map->phys_map[dest_id]->vcpu;
+	if (likely(map) && dest_id <= map->max_apic_id) {
+		dest_id = array_index_nospec(dest_id, map->max_apic_id);

Same thing here.

Quoted 9 lines
arch/x86/kvm/x86.c +3 -0
+		if (map->phys_map[dest_id])
+			target = map->phys_map[dest_id]->vcpu;
+	}
 
 	rcu_read_unlock();
 
Signature · 1 line
2.50.1
#2 lore raw

PATCH v2 1/1 KVM: x86: use array_index_nospec with indices that come from guest

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)
arch/x86/kvm/lapic.c +2 -0
index 73418dc0ebb2..0725d2cae742 100644
@@ -852,6 +852,8 @@ static int __pv_send_ipi(unsigned long *ipi_bitmap, struct kvm_apic_map *map,
 	if (min > map->max_apic_id)
 		return 0;
 
+	min = array_index_nospec(min, map->max_apic_id + 1);
+
 	for_each_set_bit(i, ipi_bitmap,
 		min((u32)BITS_PER_LONG, (map->max_apic_id - min + 1))) {
 		if (map->phys_map[min + i]) {
arch/x86/kvm/x86.c +5 -2
index 93636f77c42d..43b63f1d1594 100644
@@ -10051,8 +10051,11 @@ static void kvm_sched_yield(struct kvm_vcpu *vcpu, unsigned long dest_id)
 	rcu_read_lock();
 	map = rcu_dereference(vcpu->kvm->arch.apic_map);
 
-	if (likely(map) && dest_id <= map->max_apic_id && map->phys_map[dest_id])
-		target = map->phys_map[dest_id]->vcpu;
+	if (likely(map) && dest_id <= map->max_apic_id) {
+		dest_id = array_index_nospec(dest_id, map->max_apic_id + 1);
+		if (map->phys_map[dest_id])
+			target = map->phys_map[dest_id]->vcpu;
+	}
 
 	rcu_read_unlock();
 
Signature · 1 line
2.50.1
#3 lore raw
Quoted 9 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Quoted 4 lines
---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

thanks,

greg k-h

#4 lore raw
Quoted 5 lines · Greg Kroah-Hartman
Quoted 4 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Quoted 8 lines
Quoted 5 lines

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Want me to keep your SoB when applying, or drop it?

Quoted 9 lines
Quoted 4 lines
---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

I require the same things, but I also don't mind doing fixup when applying if that's the path of least resistance (and it's not a recurring problem).

I also strongly dislike using In-Reply-To for new versions, as it tends to confuse b4, and often confuses me as well.

But for this, I don't see any reason to send a v3.

#5 lore raw
Quoted 12 lines · Sean Christopherson
Quoted 11 lines · Greg Kroah-Hartman
Quoted 9 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Quoted 4 lines
Quoted 2 lines

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Want me to keep your SoB when applying, or drop it?

Keep it please, I was just letting Thijs know.

Quoted 10 lines
Quoted 9 lines
Quoted 4 lines
---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

Quoted 7 lines

I require the same things, but I also don't mind doing fixup when applying if that's the path of least resistance (and it's not a recurring problem).

I also strongly dislike using In-Reply-To for new versions, as it tends to confuse b4, and often confuses me as well.

But for this, I don't see any reason to send a v3.

That's great, thanks.

greg k-h

#6 lore raw
Quoted 12 lines · Greg Kroah-Hartman
Quoted 11 lines · Sean Christopherson
Quoted 10 lines · Greg Kroah-Hartman
Quoted 9 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Quoted 4 lines
Quoted 3 lines
Quoted 2 lines

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Want me to keep your SoB when applying, or drop it?

Keep it please, I was just letting Thijs know.

Sorry about that. I was not entirely sure whether tags like Signed-Off should be kept or removed in a new revision. Thanks for the feedback.

Quoted 10 lines
Quoted 10 lines
Quoted 8 lines
Quoted 4 lines
---
  arch/x86/kvm/lapic.c | 2 ++
  arch/x86/kvm/x86.c   | 7 +++++--
  2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

I require the same things, but I also don't mind doing fixup when applying if that's the path of least resistance (and it's not a recurring problem).

Changes in v2:
- As noted by Sean Christopherson, max_apic_id is inclusive but array_index_nospec is not. v2 adds one to the array_index_nospec size so the bounds do include max_apic_id
- Link to v1: https://lore.kernel.org/kvm/2025072540-eggbeater-crate-50af@gregkh/T/#u

Quoted 2 lines
Quoted 2 lines

I also strongly dislike using In-Reply-To for new versions, as it tends to confuse b4, and often confuses me as well.

Noted, will not do it like that next time.

Quoted 3 lines
Quoted 2 lines

But for this, I don't see any reason to send a v3.

That's great, thanks.

Thanks. I'm fairly new to the process of submitting patches over email, so apologies for my mistakes. Thank you for your patience and feedback, it is much appreciated. If you do prefer a v3 that does include the change log, just let me know.

- Thijs

#7 lore raw
In reply to Sean Christopherson
Quoted 27 lines · Sean Christopherson
Quoted 5 lines · Greg Kroah-Hartman
Quoted 4 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Quoted 8 lines
Quoted 5 lines

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Want me to keep your SoB when applying, or drop it?

Quoted 9 lines
Quoted 4 lines
---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

Quoted 7 lines

I require the same things, but I also don't mind doing fixup when applying if that's the path of least resistance (and it's not a recurring problem).

I also strongly dislike using In-Reply-To for new versions, as it tends to confuse b4, and often confuses me as well.

But for this, I don't see any reason to send a v3.

Any status on this? I don't see it in linux-next at all, nor in 6.17-rc1

thanks,

greg k-h

#8 lore raw
Quoted 31 lines · Greg Kroah-Hartman
Quoted 30 lines · Sean Christopherson
Quoted 5 lines · Greg Kroah-Hartman
Quoted 4 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Quoted 8 lines
Quoted 5 lines

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Want me to keep your SoB when applying, or drop it?

Quoted 9 lines
Quoted 4 lines
---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

I require the same things, but I also don't mind doing fixup when applying if that's the path of least resistance (and it's not a recurring problem).

Quoted 7 lines
Quoted 4 lines

I also strongly dislike using In-Reply-To for new versions, as it tends to confuse b4, and often confuses me as well.

But for this, I don't see any reason to send a v3.

Any status on this? I don't see it in linux-next at all, nor in 6.17-rc1

I'll get it applied and sent along to Paolo/Linus this week.

#9 lore raw
Quoted 35 lines · Sean Christopherson
Quoted 34 lines · Greg Kroah-Hartman
Quoted 33 lines · Sean Christopherson
Quoted 5 lines · Greg Kroah-Hartman
Quoted 4 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Quoted 8 lines
Quoted 5 lines

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Want me to keep your SoB when applying, or drop it?

Quoted 9 lines
Quoted 4 lines
---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

I require the same things, but I also don't mind doing fixup when applying if that's the path of least resistance (and it's not a recurring problem).

I also strongly dislike using In-Reply-To for new versions, as it tends to confuse b4, and often confuses me as well.

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

But for this, I don't see any reason to send a v3.

Any status on this? I don't see it in linux-next at all, nor in 6.17-rc1

I'll get it applied and sent along to Paolo/Linus this week.

Thanks!

#10 lore raw
In reply to Sean Christopherson
Quoted 35 lines · Sean Christopherson
Quoted 34 lines · Greg Kroah-Hartman
Quoted 33 lines · Sean Christopherson
Quoted 5 lines · Greg Kroah-Hartman
Quoted 4 lines · Thijs Raymakers

min and dest_id are guest-controlled indices. Using array_index_nospec() after the bounds checks clamps these values to mitigate speculative execution side-channels.

Quoted 8 lines
Quoted 5 lines

Nit, you shouldn't have added my signed off on a new version, but that's ok, I'm fine with it.

Want me to keep your SoB when applying, or drop it?

Quoted 9 lines
Quoted 4 lines
---
 arch/x86/kvm/lapic.c | 2 ++
 arch/x86/kvm/x86.c   | 7 +++++--
 2 files changed, 7 insertions(+), 2 deletions(-)

You also forgot to say what changed down here.

Don't know how strict the KVM maintainers are, I know I require these things fixed up...

I require the same things, but I also don't mind doing fixup when applying if that's the path of least resistance (and it's not a recurring problem).

I also strongly dislike using In-Reply-To for new versions, as it tends to confuse b4, and often confuses me as well.

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

But for this, I don't see any reason to send a v3.

Any status on this? I don't see it in linux-next at all, nor in 6.17-rc1

I'll get it applied and sent along to Paolo/Linus this week.

I haven't forgotten about this, but I was out sick most of this week and v6.17-rc1 is crashing on my test systems, so I won't get this sent along until next week. Sorry for the delay.