[PATCH] KVM: arm64: selftests: Cope with arch silliness in EL2 selftest

[PATCH] KVM: arm64: selftests: Cope with arch silliness in EL2 selftest

[Oliver Upton]

Implementations without FEAT_FGT aren't required to trap the entire ID
register space when HCR_EL2.TID3 is set. This is a terrible idea, as the
hypervisor may need to advertise the absence of a feature to the VM
using a negative value in a signed field, FEAT_E2H0 being a great
example of this.

Cope with uncooperative implementations in the EL2 selftest by accepting
a zero value when FEAT_FGT is absent and otherwise only tolerating the
expected nonzero value.

Signed-off-by: Oliver Upton <oliver.upton@linux.dev>
---

Applies on top of the EL2 selftests series I shared earlier.

https://lore.kernel.org/kvmarm/20250917212044.294760-1-oliver.upton@linux.dev/#t

 tools/testing/selftests/kvm/arm64/hello_el2.c | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/kvm/arm64/hello_el2.c b/tools/testing/selftests/kvm/arm64/hello_el2.c
index 3c1f44e43f65..ff77c0ba4bb1 100644
--- a/tools/testing/selftests/kvm/arm64/hello_el2.c
+++ b/tools/testing/selftests/kvm/arm64/hello_el2.c
@@ -13,15 +13,28 @@
 
 static void guest_code(void)
 {
+	u64 mmfr0 = read_sysreg_s(SYS_ID_AA64MMFR0_EL1);
 	u64 mmfr1 = read_sysreg_s(SYS_ID_AA64MMFR1_EL1);
 	u64 mmfr4 = read_sysreg_s(SYS_ID_AA64MMFR4_EL1);
+	u8 e2h0 = SYS_FIELD_GET(ID_AA64MMFR4_EL1, E2H0, mmfr4);
 
 	GUEST_ASSERT_EQ(get_current_el(), 2);
 	GUEST_ASSERT(read_sysreg(hcr_el2) & HCR_EL2_E2H);
 	GUEST_ASSERT_EQ(SYS_FIELD_GET(ID_AA64MMFR1_EL1, VH, mmfr1),
 			ID_AA64MMFR1_EL1_VH_IMP);
-	GUEST_ASSERT_EQ(SYS_FIELD_GET(ID_AA64MMFR4_EL1, E2H0, mmfr4),
-			ID_AA64MMFR4_EL1_E2H0_NI_NV1);
+
+	/*
+	 * Traps of the complete ID register space are IMPDEF without FEAT_FGT,
+	 * which is really annoying to deal with in KVM describing E2H as RES1.
+	 *
+	 * If the implementation doesn't honor the trap then expect the register
+	 * to return all zeros.
+	 */
+	if (e2h0 == ID_AA64MMFR4_EL1_E2H0_IMP)
+		GUEST_ASSERT_EQ(SYS_FIELD_GET(ID_AA64MMFR0_EL1, FGT, mmfr0),
+				ID_AA64MMFR0_EL1_FGT_NI);
+	else
+		GUEST_ASSERT_EQ(e2h0, ID_AA64MMFR4_EL1_E2H0_NI_NV1);
 
 	GUEST_DONE();
 }
-- 
2.47.3

Re: [PATCH] KVM: arm64: selftests: Cope with arch silliness in EL2 selftest

[Marc Zyngier]

On Tue, 23 Sep 2025 10:30:06 -0700, Oliver Upton wrote:
> Implementations without FEAT_FGT aren't required to trap the entire ID
> register space when HCR_EL2.TID3 is set. This is a terrible idea, as the
> hypervisor may need to advertise the absence of a feature to the VM
> using a negative value in a signed field, FEAT_E2H0 being a great
> example of this.
> 
> Cope with uncooperative implementations in the EL2 selftest by accepting
> a zero value when FEAT_FGT is absent and otherwise only tolerating the
> expected nonzero value.
> 
> [...]

Applied to next, thanks!

[1/1] KVM: arm64: selftests: Cope with arch silliness in EL2 selftest
      commit: 75b2fdc1a82195997cab4836f9e511cbe8c8eb52

Cheers,

	M.
-- 
Without deviation from the norm, progress is not possible.