Instead of using ~0UL provide the actual writable mask for
non-id feature registers in the output of the
KVM_ARM_GET_REG_WRITABLE_MASKS ioctl.

Explicitely set the mask for CLIDR_EL1 to make sure we present
the same value to userspace than before.

This changes the mask for the CTR_EL0 register in the output
of the KVM_ARM_GET_REG_WRITABLE_MASKS ioctl.

Signed-off-by: Sebastian Ott <sebott@redhat.com>
---
 arch/arm64/kvm/sys_regs.c | 19 +++++--------------
 1 file changed, 5 insertions(+), 14 deletions(-)

diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index 2fe3492ba3c4..7d5d55e5c16a 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -2532,7 +2532,7 @@ static const struct sys_reg_desc sys_reg_descs[] = {
 
 	{ SYS_DESC(SYS_CCSIDR_EL1), access_ccsidr },
 	{ SYS_DESC(SYS_CLIDR_EL1), access_clidr, reset_clidr, CLIDR_EL1,
-	  .set_user = set_clidr },
+	  .set_user = set_clidr, .val = ~0UL },
 	{ SYS_DESC(SYS_CCSIDR2_EL1), undef_access },
 	{ SYS_DESC(SYS_SMIDR_EL1), undef_access },
 	{ SYS_DESC(SYS_CSSELR_EL1), access_csselr, reset_unknown, CSSELR_EL1 },
@@ -4087,20 +4087,11 @@ int kvm_vm_ioctl_get_reg_writable_masks(struct kvm *kvm, struct reg_mask_range *
 		if (!is_feature_id_reg(encoding) || !reg->set_user)
 			continue;
 
-		/*
-		 * For ID registers, we return the writable mask. Other feature
-		 * registers return a full 64bit mask. That's not necessary
-		 * compliant with a given revision of the architecture, but the
-		 * RES0/RES1 definitions allow us to do that.
-		 */
-		if (is_id_reg(encoding)) {
-			if (!reg->val ||
-			    (is_aa32_id_reg(encoding) && !kvm_supports_32bit_el0()))
-				continue;
-			val = reg->val;
-		} else {
-			val = ~0UL;
+		if (!reg->val ||
+		    (is_aa32_id_reg(encoding) && !kvm_supports_32bit_el0())) {
+			continue;
 		}
+		val = reg->val;
 
 		if (put_user(val, (masks + KVM_ARM_FEATURE_ID_RANGE_INDEX(encoding))))
 			return -EFAULT;
-- 
2.42.0

On Fri, Apr 26, 2024 at 12:49:49PM +0200, Sebastian Ott wrote:
> Instead of using ~0UL provide the actual writable mask for
> non-id feature registers in the output of the
> KVM_ARM_GET_REG_WRITABLE_MASKS ioctl.

Urgh, yeah in retrospect I think we should've constrained this to the
registers KVM considers "id regs" (the space occupied by known registers
or otherwise RAZ).

> Explicitely set the mask for CLIDR_EL1 to make sure we present
> the same value to userspace than before.

typo: explicitly

Also, we know the set of mutable bits for CLIDR_EL1 as it is handled in
set_clidr(). The mask really should match that, since the UAPI is
documented as "allowing userspace to know what fields can be changed for
the system register"

-- 
Thanks,
Oliver

On Wed, 1 May 2024, Oliver Upton wrote:

> On Fri, Apr 26, 2024 at 12:49:49PM +0200, Sebastian Ott wrote:
>> Instead of using ~0UL provide the actual writable mask for
>> non-id feature registers in the output of the
>> KVM_ARM_GET_REG_WRITABLE_MASKS ioctl.
>
> Urgh, yeah in retrospect I think we should've constrained this to the
> registers KVM considers "id regs" (the space occupied by known registers
> or otherwise RAZ).
>
>> Explicitely set the mask for CLIDR_EL1 to make sure we present
>> the same value to userspace than before.
>
> typo: explicitly
>
> Also, we know the set of mutable bits for CLIDR_EL1 as it is handled in
> set_clidr(). The mask really should match that, since the UAPI is
> documented as "allowing userspace to know what fields can be changed for
> the system register"

Done for the next revision.

Thanks,
Sebastian