Skip to contentSkip to conversation

KVM: arm64: nv: Fixes for FEAT_XNX (and the lack thereof)

PATCH In review Oliver Upton 2 participants 7 messages Jun 2 – Jun 4

Subsystems (3)
lore
Share & tools
lore

Oliver Upton

author maintainer
#0 lore raw

v2 cover KVM: arm64: nv: Fixes for FEAT_XNX (and the lack thereof)

Sending a v2 to address Wei-Lin's feedback and fix an additional issue spotted by sashiko.

v1: https://lore.kernel.org/kvmarm/[email protected]/

Oliver Upton (2):
KVM: arm64: nv: Fix handling of XN[0] when !FEAT_XNX
KVM: arm64: Correctly identify executable PTEs at stage-2

 arch/arm64/include/asm/kvm_nested.h | 4 ++--
 arch/arm64/kvm/hyp/pgtable.c        | 4 +++-
 2 files changed, 5 insertions(+), 3 deletions(-)

base-commit: 5d6919055dec134de3c40167a490f33c74c12581

Signature · 1 line
2.47.3
author maintainer
#1 lore raw

PATCH v2 1/2 KVM: arm64: nv: Fix handling of XN[0] when !FEAT_XNX

XN has already been extracted from its bitfield position so using FIELD_PREP() on the mask that clears XN[0] is completely broken, having the effect of unconditionally granting execute permissions...

Fix the obvious mistake by manipulating the right bit.

Fixes d93febe2ed2e ("KVM: arm64: nv: Forward FEAT_XNX permissions to the shadow stage-2")
---
 arch/arm64/include/asm/kvm_nested.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm64/include/asm/kvm_nested.h +2 -2
index 091544e6af44..a0eb83319c2e 100644
@@ -131,7 +131,7 @@ static inline bool kvm_s2_trans_exec_el0(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 
 	switch (xn) {
 	case 0b00:
@@ -147,7 +147,7 @@ static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 
 	switch (xn) {
 	case 0b00:
Signature · 1 line
2.47.3
#2 lore raw

Re: [PATCH v2 1/2] KVM: arm64: nv: Fix handling of XN[0] when !FEAT_XNX

Hi Oliver,

Quoted 28 lines · Oliver Upton

XN has already been extracted from its bitfield position so using FIELD_PREP() on the mask that clears XN[0] is completely broken, having the effect of unconditionally granting execute permissions...

Fix the obvious mistake by manipulating the right bit.

Fixes d93febe2ed2e ("KVM: arm64: nv: Forward FEAT_XNX permissions to the shadow stage-2")
---
 arch/arm64/include/asm/kvm_nested.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm64/include/asm/kvm_nested.h +1 -1
index 091544e6af44..a0eb83319c2e 100644
@@ -131,7 +131,7 @@ static inline bool kvm_s2_trans_exec_el0(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 
 	switch (xn) {
 	case 0b00:

arch/arm64/include/asm/kvm_nested.h static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *t · line 151

Quoted 6 lines
arch/arm64/include/asm/kvm_nested.h +1 -1
@@ -147,7 +147,7 @@ static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 

Now that the other patch brings up kvm_pgtable_stage2_pte_prot(), what do you think about also using that here? It can save a little bit of duplicated decode logic.

Other than this being in a header and we'll have to move the code around for this to work, I'm curious are there any other issues with this idea?

Thanks,
Wei-Lin Chang

Quoted 5 lines
arch/arm64/include/asm/kvm_nested.h +0 -0
 	switch (xn) {
 	case 0b00:
Signature · 1 line
2.47.3
author maintainer
#3 lore raw

Re: [PATCH v2 1/2] KVM: arm64: nv: Fix handling of XN[0] when !FEAT_XNX

Hey Wei-Lin,

Thanks for the review.

Quoted 42 lines · Wei-Lin Chang

Hi Oliver,

Quoted 28 lines · Oliver Upton

XN has already been extracted from its bitfield position so using FIELD_PREP() on the mask that clears XN[0] is completely broken, having the effect of unconditionally granting execute permissions...

Fix the obvious mistake by manipulating the right bit.

Fixes d93febe2ed2e ("KVM: arm64: nv: Forward FEAT_XNX permissions to the shadow stage-2")
---
 arch/arm64/include/asm/kvm_nested.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm64/include/asm/kvm_nested.h +1 -1
index 091544e6af44..a0eb83319c2e 100644
@@ -131,7 +131,7 @@ static inline bool kvm_s2_trans_exec_el0(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 
 	switch (xn) {
 	case 0b00:

arch/arm64/include/asm/kvm_nested.h static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *t · line 151

Quoted 6 lines
arch/arm64/include/asm/kvm_nested.h +1 -1
@@ -147,7 +147,7 @@ static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 

Now that the other patch brings up kvm_pgtable_stage2_pte_prot(), what do you think about also using that here? It can save a little bit of duplicated decode logic.

Quoted 3 lines

Other than this being in a header and we'll have to move the code around for this to work, I'm curious are there any other issues with this idea?

No issues with your suggestion but I plan on nuking the kvm_s2_trans*() accessors soon :)

Ultimately kvm_s2_trans should just contain pre-computed permissions, which matters more when dealing with descriptor fields that require the MMU context to make sense of (like DBM). On top of that, getters for obviously named fields isn't adding a whole lot.

Any concerns with leaving as-is for now?

Signature · 2 lines
Thanks,
Oliver
#4 lore raw

Re: [PATCH v2 1/2] KVM: arm64: nv: Fix handling of XN[0] when !FEAT_XNX

Quoted 47 lines · Oliver Upton

Hey Wei-Lin,

Thanks for the review.

Quoted 42 lines · Wei-Lin Chang

Hi Oliver,

Quoted 28 lines · Oliver Upton

XN has already been extracted from its bitfield position so using FIELD_PREP() on the mask that clears XN[0] is completely broken, having the effect of unconditionally granting execute permissions...

Fix the obvious mistake by manipulating the right bit.

Fixes d93febe2ed2e ("KVM: arm64: nv: Forward FEAT_XNX permissions to the shadow stage-2")
---
 arch/arm64/include/asm/kvm_nested.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm64/include/asm/kvm_nested.h +1 -1
index 091544e6af44..a0eb83319c2e 100644
@@ -131,7 +131,7 @@ static inline bool kvm_s2_trans_exec_el0(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 
 	switch (xn) {
 	case 0b00:

arch/arm64/include/asm/kvm_nested.h static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *t · line 151

Quoted 6 lines
arch/arm64/include/asm/kvm_nested.h +1 -1
@@ -147,7 +147,7 @@ static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *t
 	u8 xn = FIELD_GET(KVM_PTE_LEAF_ATTR_HI_S2_XN, trans->desc);
 
 	if (!kvm_has_xnx(kvm))
-		xn &= FIELD_PREP(KVM_PTE_LEAF_ATTR_HI_S2_XN, 0b10);
+		xn &= 0b10;
 

Now that the other patch brings up kvm_pgtable_stage2_pte_prot(), what do you think about also using that here? It can save a little bit of duplicated decode logic.

Quoted 6 lines
Quoted 3 lines

Other than this being in a header and we'll have to move the code around for this to work, I'm curious are there any other issues with this idea?

No issues with your suggestion but I plan on nuking the kvm_s2_trans*() accessors soon :)

Ah, cool cool.

Quoted 5 lines

Ultimately kvm_s2_trans should just contain pre-computed permissions, which matters more when dealing with descriptor fields that require the MMU context to make sense of (like DBM). On top of that, getters for obviously named fields isn't adding a whole lot.

I see, makes sense to me.

Quoted 2 lines

Any concerns with leaving as-is for now?

No, and thanks for sharing!

Thanks,
Wei-Lin Chang

Quoted 4 lines
Signature · 3 lines
-- 
Thanks,
Oliver
author maintainer
#5 lore raw

PATCH v2 2/2 KVM: arm64: Correctly identify executable PTEs at stage-2

In reply to Oliver Upton

KVM invalidates the I-cache before installing an executable PTE on implementations without DIC. Unfortunately, support for FEAT_XNX broke this check as KVM_PTE_LEAF_ATTR_HI_S2_XN was expanded to a bitfield.

Fix it by reusing kvm_pgtable_stage2_pte_prot() and testing the abstract permission bits instead.

Fixes 2608563b466b ("KVM: arm64: Add support for FEAT_XNX stage-2 permissions")
Reported-by Sashiko (gemini/gemini-3.1-pro-preview)
---
 arch/arm64/kvm/hyp/pgtable.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kvm/hyp/pgtable.c +3 -1
index 0c1defa5fb0f..91a7dfad6686 100644
@@ -925,7 +925,9 @@ static bool stage2_pte_cacheable(struct kvm_pgtable *pgt, kvm_pte_t pte)
 
 static bool stage2_pte_executable(kvm_pte_t pte)
 {
-	return kvm_pte_valid(pte) && !(pte & KVM_PTE_LEAF_ATTR_HI_S2_XN);
+	enum kvm_pgtable_prot prot = kvm_pgtable_stage2_pte_prot(pte);
+
+	return prot & (KVM_PGTABLE_PROT_UX | KVM_PGTABLE_PROT_PX);
 }
 
 static u64 stage2_map_walker_phys_addr(const struct kvm_pgtable_visit_ctx *ctx,
Signature · 1 line
2.47.3
Reviewed-by on v2 2/2
#6 lore raw

Re: [PATCH v2 2/2] KVM: arm64: Correctly identify executable PTEs at stage-2

Quoted 33 lines · Oliver Upton

KVM invalidates the I-cache before installing an executable PTE on implementations without DIC. Unfortunately, support for FEAT_XNX broke this check as KVM_PTE_LEAF_ATTR_HI_S2_XN was expanded to a bitfield.

Fix it by reusing kvm_pgtable_stage2_pte_prot() and testing the abstract permission bits instead.

Fixes 2608563b466b ("KVM: arm64: Add support for FEAT_XNX stage-2 permissions")
Reported-by Sashiko (gemini/gemini-3.1-pro-preview)
---
 arch/arm64/kvm/hyp/pgtable.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kvm/hyp/pgtable.c +3 -1
index 0c1defa5fb0f..91a7dfad6686 100644
@@ -925,7 +925,9 @@ static bool stage2_pte_cacheable(struct kvm_pgtable *pgt, kvm_pte_t pte)
 
 static bool stage2_pte_executable(kvm_pte_t pte)
 {
-	return kvm_pte_valid(pte) && !(pte & KVM_PTE_LEAF_ATTR_HI_S2_XN);
+	enum kvm_pgtable_prot prot = kvm_pgtable_stage2_pte_prot(pte);
+
+	return prot & (KVM_PGTABLE_PROT_UX | KVM_PGTABLE_PROT_PX);
 }
 
 static u64 stage2_map_walker_phys_addr(const struct kvm_pgtable_visit_ctx *ctx,
Signature · 1 line
2.47.3

Thanks,
Wei-Lin Chang