Commit d93febe2 authored by Oliver Upton's avatar Oliver Upton
Browse files

KVM: arm64: nv: Forward FEAT_XNX permissions to the shadow stage-2 



Add support for FEAT_XNX to shadow stage-2 MMUs, being careful to only
evaluate XN[0] when the feature is actually exposed to the VM.
Restructure the layering of permissions in the fault handler to assume
pX and uX then restricting based on the guest's stage-2 afterwards.

Reviewed-by: default avatarMarc Zyngier <maz@kernel.org>
Tested-by: default avatarMarc Zyngier <maz@kernel.org>
Link: https://msgid.link/20251124190158.177318-4-oupton@kernel.org


Signed-off-by: default avatarOliver Upton <oupton@kernel.org>
parent 2608563b

arch/arm64/include/asm/kvm_nested.h

+35 −2
Original line number Diff line number Diff line
@@ -120,9 +120,42 @@ static inline bool kvm_s2_trans_writable(struct kvm_s2_trans *trans) 
	return trans->writable;

}



static inline bool kvm_s2_trans_executable(struct kvm_s2_trans *trans)

static inline bool kvm_has_xnx(struct kvm *kvm)

{

	return !(trans->desc & BIT(54));

	return cpus_have_final_cap(ARM64_HAS_XNX) &&

		kvm_has_feat(kvm, ID_AA64MMFR1_EL1, XNX, IMP);

}



static inline bool kvm_s2_trans_exec_el0(struct kvm *kvm, struct kvm_s2_trans *trans)

{

	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);



	switch (xn) {

	case 0b00:

	case 0b01:

		return true;

	default:

		return false;

	}

}



static inline bool kvm_s2_trans_exec_el1(struct kvm *kvm, struct kvm_s2_trans *trans)

{

	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);



	switch (xn) {

	case 0b00:

	case 0b11:

		return true;

	default:

		return false;

	}

}



extern int kvm_walk_nested_s2(struct kvm_vcpu *vcpu, phys_addr_t gipa,

arch/arm64/kvm/mmu.c

+18 −5
Original line number Diff line number Diff line
@@ -1521,6 +1521,16 @@ static void adjust_nested_fault_perms(struct kvm_s2_trans *nested, 
	*prot |= kvm_encode_nested_level(nested);

}



static void adjust_nested_exec_perms(struct kvm *kvm,

				     struct kvm_s2_trans *nested,

				     enum kvm_pgtable_prot *prot)

{

	if (!kvm_s2_trans_exec_el0(kvm, nested))

		*prot &= ~KVM_PGTABLE_PROT_UX;

	if (!kvm_s2_trans_exec_el1(kvm, nested))

		*prot &= ~KVM_PGTABLE_PROT_PX;

}



#define KVM_PGTABLE_WALK_MEMABORT_FLAGS (KVM_PGTABLE_WALK_HANDLE_FAULT | KVM_PGTABLE_WALK_SHARED)



static int gmem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
@@ -1572,11 +1582,12 @@ static int gmem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa, 
	if (writable)

		prot |= KVM_PGTABLE_PROT_W;



	if (exec_fault ||

	    (cpus_have_final_cap(ARM64_HAS_CACHE_DIC) &&

	     (!nested || kvm_s2_trans_executable(nested))))

	if (exec_fault || cpus_have_final_cap(ARM64_HAS_CACHE_DIC))

		prot |= KVM_PGTABLE_PROT_X;



	if (nested)

		adjust_nested_exec_perms(kvm, nested, &prot);



	kvm_fault_lock(kvm);

	if (mmu_invalidate_retry(kvm, mmu_seq)) {

		ret = -EAGAIN;
@@ -1851,11 +1862,13 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa, 
			prot |= KVM_PGTABLE_PROT_NORMAL_NC;

		else

			prot |= KVM_PGTABLE_PROT_DEVICE;

	} else if (cpus_have_final_cap(ARM64_HAS_CACHE_DIC) &&

		   (!nested || kvm_s2_trans_executable(nested))) {

	} else if (cpus_have_final_cap(ARM64_HAS_CACHE_DIC)) {

		prot |= KVM_PGTABLE_PROT_X;

	}



	if (nested)

		adjust_nested_exec_perms(kvm, nested, &prot);



	/*

	 * Under the premise of getting a FSC_PERM fault, we just need to relax

	 * permissions only if vma_pagesize equals fault_granule. Otherwise,

arch/arm64/kvm/nested.c

+4 −1
Original line number Diff line number Diff line
@@ -788,7 +788,10 @@ int kvm_s2_handle_perm_fault(struct kvm_vcpu *vcpu, struct kvm_s2_trans *trans) 
		return 0;



	if (kvm_vcpu_trap_is_iabt(vcpu)) {

		forward_fault = !kvm_s2_trans_executable(trans);

		if (vcpu_mode_priv(vcpu))

			forward_fault = !kvm_s2_trans_exec_el1(vcpu->kvm, trans);

		else

			forward_fault = !kvm_s2_trans_exec_el0(vcpu->kvm, trans);

	} else {

		bool write_fault = kvm_is_write_fault(vcpu);