Hi Jing,
On 8/1/23 23:19, Jing Zhang wrote:
Add a VM ioctl to allow userspace to get writable masks for feature ID
registers in below system register space:
op0 = 3, op1 = {0, 1, 3}, CRn = 0, CRm = {0 - 7}, op2 = {0 - 7}
This is used to support mix-and-match userspace and kernels for writable
ID registers, where userspace may want to know upfront whether it can
actually tweak the contents of an idreg or not.
Suggested-by: Marc Zyngier <maz@xxxxxxxxxx>
Suggested-by: Cornelia Huck <cohuck@xxxxxxxxxx>
Signed-off-by: Jing Zhang <jingzhangos@xxxxxxxxxx>
---
arch/arm64/include/asm/kvm_host.h | 2 ++
arch/arm64/include/uapi/asm/kvm.h | 25 +++++++++++++++
arch/arm64/kvm/arm.c | 3 ++
arch/arm64/kvm/sys_regs.c | 51 +++++++++++++++++++++++++++++++
include/uapi/linux/kvm.h | 2 ++
5 files changed, 83 insertions(+)
diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index d3dd05bbfe23..3996a3707f4e 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -1074,6 +1074,8 @@ int kvm_vm_ioctl_mte_copy_tags(struct kvm *kvm,
struct kvm_arm_copy_mte_tags *copy_tags);
int kvm_vm_ioctl_set_counter_offset(struct kvm *kvm,
struct kvm_arm_counter_offset *offset);
+int kvm_vm_ioctl_get_feature_id_writable_masks(struct kvm *kvm,
+ u64 __user *masks);
/* Guest/host FPSIMD coordination helpers */
int kvm_arch_vcpu_run_map_fp(struct kvm_vcpu *vcpu);
diff --git a/arch/arm64/include/uapi/asm/kvm.h b/arch/arm64/include/uapi/asm/kvm.h
index f7ddd73a8c0f..2970c0d792ee 100644
--- a/arch/arm64/include/uapi/asm/kvm.h
+++ b/arch/arm64/include/uapi/asm/kvm.h
@@ -505,6 +505,31 @@ struct kvm_smccc_filter {
#define KVM_HYPERCALL_EXIT_SMC (1U << 0)
#define KVM_HYPERCALL_EXIT_16BIT (1U << 1)
+/* Get feature ID registers userspace writable mask. */
+/*
+ * From DDI0487J.a, D19.2.66 ("ID_AA64MMFR2_EL1, AArch64 Memory Model
+ * Feature Register 2"):
+ *
+ * "The Feature ID space is defined as the System register space in
+ * AArch64 with op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7},
+ * op2=={0-7}."
+ *
+ * This covers all R/O registers that indicate anything useful feature
+ * wise, including the ID registers.
+ */
+#define ARM64_FEATURE_ID_SPACE_IDX(op0, op1, crn, crm, op2) \
+ ({ \
+ __u64 __op1 = (op1) & 3; \
+ __op1 -= (__op1 == 3); \
+ (__op1 << 6 | ((crm) & 7) << 3 | (op2)); \
+ })
+
+#define ARM64_FEATURE_ID_SPACE_SIZE (3 * 8 * 8)
+
+struct feature_id_writable_masks {
+ __u64 mask[ARM64_FEATURE_ID_SPACE_SIZE];
+};
+
#endif
#endif /* __ARM_KVM_H__ */
diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
index 72dc53a75d1c..c9cd14057c58 100644
--- a/arch/arm64/kvm/arm.c
+++ b/arch/arm64/kvm/arm.c
@@ -1630,6 +1630,9 @@ int kvm_arch_vm_ioctl(struct file *filp, unsigned int ioctl, unsigned long arg)
return kvm_vm_set_attr(kvm, &attr);
}
+ case KVM_ARM_GET_FEATURE_ID_WRITABLE_MASKS: {
+ return kvm_vm_ioctl_get_feature_id_writable_masks(kvm, argp);
+ }
default:
return -EINVAL;
}
diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index 2ca2973abe66..d9317b640ba5 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -3560,6 +3560,57 @@ int kvm_arm_copy_sys_reg_indices(struct kvm_vcpu *vcpu, u64 __user *uindices)
return write_demux_regids(uindices);
}
+#define ARM64_FEATURE_ID_SPACE_INDEX(r) \
+ ARM64_FEATURE_ID_SPACE_IDX(sys_reg_Op0(r), \
+ sys_reg_Op1(r), \
+ sys_reg_CRn(r), \
+ sys_reg_CRm(r), \
+ sys_reg_Op2(r))
+
+static bool is_feature_id_reg(u32 encoding)
+{
+ return (sys_reg_Op0(encoding) == 3 &&
+ (sys_reg_Op1(encoding) < 2 || sys_reg_Op1(encoding) == 3) &&
+ sys_reg_CRn(encoding) == 0 &&
+ sys_reg_CRm(encoding) <= 7);
+}
A fool question.
In the code base, there is a function is_id_reg() to check if it's a
id_reg, what's the difference between the is_feature_id_reg() and the
is_id_reg()?
/*
* Return true if the register's (Op0, Op1, CRn, CRm, Op2) is
* (3, 0, 0, crm, op2), where 1<=crm<8, 0<=op2<8.
*/
static inline bool is_id_reg(u32 id)
{
return (sys_reg_Op0(id) == 3 && sys_reg_Op1(id) == 0 &&
sys_reg_CRn(id) == 0 && sys_reg_CRm(id) >= 1 &&
sys_reg_CRm(id) < 8);
}
Thanks,
Shaoqin
+
+int kvm_vm_ioctl_get_feature_id_writable_masks(struct kvm *kvm, u64 __user *masks)
+{
+ /* Wipe the whole thing first */
+ for (int i = 0; i < ARM64_FEATURE_ID_SPACE_SIZE; i++)
+ if (put_user(0, masks + i))
+ return -EFAULT;
+
+ for (int i = 0; i < ARRAY_SIZE(sys_reg_descs); i++) {
+ const struct sys_reg_desc *reg = &sys_reg_descs[i];
+ u32 encoding = reg_to_encoding(reg);
+ u64 val;
+
+ 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)
+ continue;
+ val = reg->val;
+ } else {
+ val = ~0UL;
+ }
+
+ if (put_user(val, (masks + ARM64_FEATURE_ID_SPACE_INDEX(encoding))))
+ return -EFAULT;
+ }
+
+ return 0;
+}
+
int __init kvm_sys_reg_table_init(void)
{
struct sys_reg_params params;
diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
index f089ab290978..86ffdf134eb8 100644
--- a/include/uapi/linux/kvm.h
+++ b/include/uapi/linux/kvm.h
@@ -1555,6 +1555,8 @@ struct kvm_s390_ucas_mapping {
#define KVM_ARM_MTE_COPY_TAGS _IOR(KVMIO, 0xb4, struct kvm_arm_copy_mte_tags)
/* Available with KVM_CAP_COUNTER_OFFSET */
#define KVM_ARM_SET_COUNTER_OFFSET _IOW(KVMIO, 0xb5, struct kvm_arm_counter_offset)
+#define KVM_ARM_GET_FEATURE_ID_WRITABLE_MASKS \
+ _IOR(KVMIO, 0xb6, struct feature_id_writable_masks)
/* ioctl for vm fd */
#define KVM_CREATE_DEVICE _IOWR(KVMIO, 0xe0, struct kvm_create_device)
--
Shaoqin
