Skip to contentSkip to conversation

KVM: arm64: Reserve SMC64 arch range per SMCCC filter documentation

PATCH Applied Oliver Upton 2 participants 4 messages Apr 8, 2023

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

cover KVM: arm64: Reserve SMC64 arch range per SMCCC filter documentation

The intention of the SMCCC filter series was that the 'Arm architecture calls' range is reserved, meaning userspace cannot apply a filter to it. Though the documentation calls out both the SMC32 and presently unused SMC64 views, only the SMC32 view of this range was actually reserved.

Small series to align UAPI behavior with the documentation and adding a test case for the missed condition. Applies to kvmarm/next.

Oliver Upton (2):
KVM: arm64: Prevent userspace from handling SMC64 arch range
KVM: arm64: Test that SMC64 arch calls are reserved

 arch/arm64/kvm/hypercalls.c                   | 25 ++++++++++++++-----
 .../selftests/kvm/aarch64/smccc_filter.c      |  8 ++++++
 2 files changed, 27 insertions(+), 6 deletions(-)

base-commit: b87b85f845976e8d23e5f9fe7e399f6fc7360d26

Signature · 1 line
2.40.0.577.gac1e443424-goog
#1 lore raw

PATCH 1/2 KVM: arm64: Prevent userspace from handling SMC64 arch range

Though presently unused, there is an SMC64 view of the Arm architecture calls defined by the SMCCC. The documentation of the SMCCC filter states that the SMC64 range is reserved, but nothing actually prevents userspace from applying a filter to the range.

Insert a range with the HANDLE action for the SMC64 arch range, thereby preventing userspace from imposing filtering/forwarding on it.

Fixes fb88707dd39b ("KVM: arm64: Use a maple tree to represent the SMCCC filter")
---
 arch/arm64/kvm/hypercalls.c | 25 +++++++++++++++++++------
 1 file changed, 19 insertions(+), 6 deletions(-)
arch/arm64/kvm/hypercalls.c +19 -6
index baf1c4e4a4dc..2e16fc7b31bf 100644
@@ -121,11 +121,17 @@ static bool kvm_smccc_test_fw_bmap(struct kvm_vcpu *vcpu, u32 func_id)
 	}
 }
 
-#define SMCCC_ARCH_RANGE_BEGIN	ARM_SMCCC_VERSION_FUNC_ID
-#define SMCCC_ARCH_RANGE_END				\
-	ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL,		\
-			   ARM_SMCCC_SMC_32,		\
-			   0, ARM_SMCCC_FUNC_MASK)
+#define SMC32_ARCH_RANGE_BEGIN	ARM_SMCCC_VERSION_FUNC_ID
+#define SMC32_ARCH_RANGE_END	ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL,		\
+						   ARM_SMCCC_SMC_32,		\
+						   0, ARM_SMCCC_FUNC_MASK)
+
+#define SMC64_ARCH_RANGE_BEGIN	ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL,		\
+						   ARM_SMCCC_SMC_64,		\
+						   0, 0)
+#define SMC64_ARCH_RANGE_END	ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL,		\
+						   ARM_SMCCC_SMC_64,		\
+						   0, ARM_SMCCC_FUNC_MASK)
 
 static void init_smccc_filter(struct kvm *kvm)
 {
@@ -139,10 +145,17 @@ static void init_smccc_filter(struct kvm *kvm)
 	 * to the guest.
 	 */
 	r = mtree_insert_range(&kvm->arch.smccc_filter,
-			       SMCCC_ARCH_RANGE_BEGIN, SMCCC_ARCH_RANGE_END,
+			       SMC32_ARCH_RANGE_BEGIN, SMC32_ARCH_RANGE_END,
 			       xa_mk_value(KVM_SMCCC_FILTER_HANDLE),
 			       GFP_KERNEL_ACCOUNT);
 	WARN_ON_ONCE(r);
+
+	r = mtree_insert_range(&kvm->arch.smccc_filter,
+			       SMC64_ARCH_RANGE_BEGIN, SMC64_ARCH_RANGE_END,
+			       xa_mk_value(KVM_SMCCC_FILTER_HANDLE),
+			       GFP_KERNEL_ACCOUNT);
+	WARN_ON_ONCE(r);
+
 }
 
 static int kvm_smccc_set_filter(struct kvm *kvm, struct kvm_smccc_filter __user *uaddr)
Signature · 1 line
2.40.0.577.gac1e443424-goog
#2 lore raw

PATCH 2/2 KVM: arm64: Test that SMC64 arch calls are reserved

In reply to Oliver Upton

Assert that the SMC64 view of the Arm architecture range is reserved by KVM and cannot be filtered by userspace.

---
 tools/testing/selftests/kvm/aarch64/smccc_filter.c | 8 ++++++++
 1 file changed, 8 insertions(+)
tools/testing/selftests/kvm/aarch64/smccc_filter.c +8 -0
index 0f9db0641847..dab671fdf239 100644
@@ -99,6 +99,7 @@ static void test_filter_reserved_range(void)
 {
 	struct kvm_vcpu *vcpu;
 	struct kvm_vm *vm = setup_vm(&vcpu);
+	uint32_t smc64_fn;
 	int r;
 
 	r = __set_smccc_filter(vm, ARM_SMCCC_ARCH_WORKAROUND_1,
@@ -106,6 +107,13 @@ static void test_filter_reserved_range(void)
 	TEST_ASSERT(r < 0 && errno == EEXIST,
 		    "Attempt to filter reserved range should return EEXIST");
 
+	smc64_fn = ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, ARM_SMCCC_SMC_64,
+				      0, 0);
+
+	r = __set_smccc_filter(vm, smc64_fn, 1, KVM_SMCCC_FILTER_DENY);
+	TEST_ASSERT(r < 0 && errno == EEXIST,
+		    "Attempt to filter reserved range should return EEXIST");
+
 	kvm_vm_free(vm);
 }
 
Signature · 1 line
2.40.0.577.gac1e443424-goog
maintainer Applied
#3 lore raw
In reply to Oliver Upton
Quoted 9 lines · Oliver Upton

The intention of the SMCCC filter series was that the 'Arm architecture calls' range is reserved, meaning userspace cannot apply a filter to it. Though the documentation calls out both the SMC32 and presently unused SMC64 views, only the SMC32 view of this range was actually reserved.

Small series to align UAPI behavior with the documentation and adding a test case for the missed condition. Applies to kvmarm/next.

[...]

Applied to next, thanks!

[1/2] KVM: arm64: Prevent userspace from handling SMC64 arch range

      commit: 5a23ad6510c82049f5ab3795841c30e8f3ca324d
[2/2] KVM: arm64: Test that SMC64 arch calls are reserved
      commit: 00e0c947118f456b622c1f2ca316c116dfb4e12c

Cheers,

	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.