[PATCH v6 22/43] KVM: arm64: Validate register access for a Realm VM
Gavin Shan
gshan at redhat.com
Sat Feb 1 17:22:58 PST 2025
On 12/13/24 1:55 AM, Steven Price wrote:
> The RMM only allows setting the GPRS (x0-x30) and PC for a realm
> guest. Check this in kvm_arm_set_reg() so that the VMM can receive a
> suitable error return if other registers are accessed.
>
> Signed-off-by: Steven Price <steven.price at arm.com>
> ---
> Changes since v5:
> * Upper GPRS can be set as part of a HOST_CALL return, so fix up the
> test to allow them.
> ---
> arch/arm64/kvm/guest.c | 43 ++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 43 insertions(+)
>
> diff --git a/arch/arm64/kvm/guest.c b/arch/arm64/kvm/guest.c
> index 12dad841f2a5..1ee2fe072f1a 100644
> --- a/arch/arm64/kvm/guest.c
> +++ b/arch/arm64/kvm/guest.c
> @@ -73,6 +73,24 @@ static u64 core_reg_offset_from_id(u64 id)
> return id & ~(KVM_REG_ARCH_MASK | KVM_REG_SIZE_MASK | KVM_REG_ARM_CORE);
> }
>
> +static bool kvm_realm_validate_core_reg(u64 off)
> +{
> + /*
> + * Note that GPRs can only sometimes be controlled by the VMM.
> + * For PSCI only X0-X6 are used, higher registers are ignored (restored
> + * from the REC).
> + * For HOST_CALL all of X0-X30 are copied to the RsiHostCall structure.
> + * For emulated MMIO X0 is always used.
> + */
> + switch (off) {
> + case KVM_REG_ARM_CORE_REG(regs.regs[0]) ...
> + KVM_REG_ARM_CORE_REG(regs.regs[30]):
> + case KVM_REG_ARM_CORE_REG(regs.pc):
> + return true;
> + }
> + return false;
> +}
> +
> static int core_reg_size_from_offset(const struct kvm_vcpu *vcpu, u64 off)
> {
> int size;
> @@ -115,6 +133,9 @@ static int core_reg_size_from_offset(const struct kvm_vcpu *vcpu, u64 off)
> if (vcpu_has_sve(vcpu) && core_reg_offset_is_vreg(off))
> return -EINVAL;
>
> + if (kvm_is_realm(vcpu->kvm) && !kvm_realm_validate_core_reg(off))
> + return -EPERM;
> +
> return size;
> }
>
> @@ -783,12 +804,34 @@ int kvm_arm_get_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
> return kvm_arm_sys_reg_get_reg(vcpu, reg);
> }
>
> +/*
> + * The RMI ABI only enables setting some GPRs and PC. The selection of GPRs
> + * that are available depends on the Realm state and the reason for the last
> + * exit. All other registers are reset to architectural or otherwise defined
> + * reset values by the RMM, except for a few configuration fields that
> + * correspond to Realm parameters.
> + */
> +static bool validate_realm_set_reg(struct kvm_vcpu *vcpu,
> + const struct kvm_one_reg *reg)
> +{
> + if ((reg->id & KVM_REG_ARM_COPROC_MASK) == KVM_REG_ARM_CORE) {
> + u64 off = core_reg_offset_from_id(reg->id);
> +
> + return kvm_realm_validate_core_reg(off);
> + }
> +
> + return false;
> +}
> +
> int kvm_arm_set_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg)
> {
> /* We currently use nothing arch-specific in upper 32 bits */
> if ((reg->id & ~KVM_REG_SIZE_MASK) >> 32 != KVM_REG_ARM64 >> 32)
> return -EINVAL;
>
> + if (kvm_is_realm(vcpu->kvm) && !validate_realm_set_reg(vcpu, reg))
> + return -EINVAL;
> +
> switch (reg->id & KVM_REG_ARM_COPROC_MASK) {
> case KVM_REG_ARM_CORE: return set_core_reg(vcpu, reg);
> case KVM_REG_ARM_FW:
It looks the core registers in kvm_arm_set_reg() has been validated for twice.
ioctl(KVM_SET_ONE_REG)
kvm_arm_set_reg
validate_realm_set_reg
kvm_realm_validate_core_reg // 1
set_core_reg
core_reg_offset_from_id
core_reg_addr
core_reg_offset_from_id
core_reg_size_from_offset
kvm_realm_validate_core_reg // 2
copy_from_user
Besides, there are other types of registers that can be accessed by KVM_{GET, SET}_ONE_REG:
firmware and bitmap registers, SVE registers, timer registers, system registers. Need we
to hide any of them from the user space? As I can understand, the SVE registers are owned
by RMM at least and won't be exposed to user space.
Thanks,
Gavin
More information about the linux-arm-kernel
mailing list