Skip to contentSkip to conversation

KVM: arm64: selftests: Skip unwritable ID fields in the failed-run test

PATCH Applied Sasha Levin 4 participants 4 messages Oct 26

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

PATCH v2 1/1 KVM: arm64: selftests: Skip unwritable ID fields in the failed-run test

test_idreg_frozen_after_failed_run() checks that changing a feature ID register after a failed first KVM_RUN is rejected with -EBUSY, and does so for every field of every register in test_regs.

On a host without AArch32 support at EL0, KVM treats the AArch32 ID registers as RAZ/WI: aa32_id_visibility() returns REG_RAZ | REG_USER_WI, and kvm_sys_reg_set_user() ignores the write and returns 0 before set_id_reg() checks whether the ID registers are final. The test then fails on the first such register:

arm64/set_id_regs.c:870: r < 0 && errno == EBUSY ID_DFR0_EL1_PerfMon write after failed first run: r=0 errno=16

KernelCI hits this on every run on an AArch64 only GB300 host.

Only a field that userspace can change can be frozen, and KVM reports these registers as not writable in KVM_ARM_GET_REG_WRITABLE_MASKS. Skip the fields that the writable mask does not cover. test_vm_ftr_id_regs() already requires the mask to cover every other field in test_regs, so this only drops the fields KVM never lets userspace change. Whether KVM accepts writes to fields it reports as not writable is a separate question from finalization.

Fixes 00e309933879 ("KVM: arm64: selftests: Check ID regs are immutable after a failed run")

---
Changes in v2:
- Skip every field that KVM_ARM_GET_REG_WRITABLE_MASKS does not report as writable, instead of special casing the AArch32 ID registers (Mark Brown).

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

 tools/testing/selftests/kvm/arm64/set_id_regs.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)
tools/testing/selftests/kvm/arm64/set_id_regs.c +12 -1
index 10849d21c0dd2..1d2cc25cb3f4c 100644
@@ -815,6 +815,10 @@ static void test_idreg_frozen_after_failed_run(void)
 		SYS_REVIDR_EL1,
 		SYS_AIDR_EL1,
 	};
+	u64 masks[KVM_ARM_FEATURE_ID_RANGE_SIZE];
+	struct reg_mask_range range = {
+		.addr = (__u64)masks,
+	};
 	struct kvm_vcpu_init init;
 	struct kvm_vcpu *vcpu;
 	struct kvm_vm *vm;
@@ -831,6 +835,7 @@ static void test_idreg_frozen_after_failed_run(void)
 
 	vm = vm_create(1);
 	vm_enable_cap(vm, KVM_CAP_ARM_WRITABLE_IMP_ID_REGS, 0);
+	vm_ioctl(vm, KVM_ARM_GET_REG_WRITABLE_MASKS, &range);
 	kvm_get_default_vcpu_target(vm, &init);
 	init.features[0] |= (1 << KVM_ARM_VCPU_PMU_V3);
 	vcpu = aarch64_vcpu_add(vm, 0, &init, guest_code);
@@ -852,7 +857,9 @@ static void test_idreg_frozen_after_failed_run(void)
 	 */
 	for (int i = 0; i < ARRAY_SIZE(test_regs); i++) {
 		const struct reg_ftr_bits *ftr_bits = test_regs[i].ftr_bits;
-		u64 reg = KVM_ARM64_SYS_REG(test_regs[i].reg);
+		u32 reg_id = test_regs[i].reg;
+		u64 reg = KVM_ARM64_SYS_REG(reg_id);
+		u64 mask = masks[encoding_to_range_idx(reg_id)];
 		u64 val = vcpu_get_reg(vcpu, reg);
 
 		for (int j = 0; ftr_bits[j].type != FTR_END; j++) {
@@ -860,6 +867,10 @@ static void test_idreg_frozen_after_failed_run(void)
 			u64 safe = get_safe_value(&ftr_bits[j], ftr);
 			u64 new_val;
 
+			/* Only a field userspace could change can be frozen. */
+			if ((mask & ftr_bits[j].mask) != ftr_bits[j].mask)
+				continue;
+
 			if (safe == ftr)
 				continue;
 

base-commit: 21a7f44fd0765fcc5194130b28767786eba1d1e7

Signature · 1 line
2.53.0
reviewer Reviewed-by Tested-by on v2 1/1
#1 lore raw

Hi Sasha,

Quoted 3 lines · Sasha Levin

test_idreg_frozen_after_failed_run() checks that changing a feature ID register after a failed first KVM_RUN is rejected with -EBUSY, and does so for every field of every register in test_regs.

[...]

I tested v2 on the FVP AEM model configured with no AArch32 at any EL and with PMUv3, under nVHE: set_id_regs fails on ID_DFR0_EL1_PerfMon without this patch and passes with it. It also passes on QEMU (which supports AArch32 at EL0) under VHE, nVHE and pKVM.

Cheers,
/fuad

maintainer Applied
#2 lore raw
In reply to Sasha Levin
Quoted 11 lines · Sasha Levin

test_idreg_frozen_after_failed_run() checks that changing a feature ID register after a failed first KVM_RUN is rejected with -EBUSY, and does so for every field of every register in test_regs.

On a host without AArch32 support at EL0, KVM treats the AArch32 ID registers as RAZ/WI: aa32_id_visibility() returns REG_RAZ | REG_USER_WI, and kvm_sys_reg_set_user() ignores the write and returns 0 before set_id_reg() checks whether the ID registers are final. The test then fails on the first such register:

[...]

Applied to fixes, thanks!

[1/1] KVM: arm64: selftests: Skip unwritable ID fields in the failed-run test

      commit: a4be8d69698b11d976c122be2d32bb72b8a52809

Cheers,

	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
Reviewed-by on v2 1/1
#3 lore raw
In reply to Sasha Levin
Quoted 7 lines · Sasha Levin

Only a field that userspace can change can be frozen, and KVM reports these registers as not writable in KVM_ARM_GET_REG_WRITABLE_MASKS. Skip the fields that the writable mask does not cover. test_vm_ftr_id_regs() already requires the mask to cover every other field in test_regs, so this only drops the fields KVM never lets userspace change. Whether KVM accepts writes to fields it reports as not writable is a separate question from finalization.

It would be nice to apply the same thing to the NV instances as well but in any case:

I didn't redo the testing that Fuad already did since it'd just be the same thing.