Re: [PATCH v8 4/7] KVM: VMX: Load CET states on vmentry/vmexit

From: Sean Christopherson
Date: Tue Dec 10 2019 - 16:23:10 EST


On Fri, Nov 01, 2019 at 04:52:19PM +0800, Yang Weijiang wrote:
> "Load {guest,host} CET state" bit controls whether guest/host
> CET states will be loaded at VM entry/exit. Before doing that,
> KVM needs to check if CET is both enabled on host and guest.
>
> Note: SHSTK and IBT features share one control MSR:
> MSR_IA32_{U,S}_CET, which means it's difficult to hide
> one feature from another in the case of SHSTK != IBT,
> after discussed in community, it's agreed to allow Guest
> control two features independently as it won't introduce
> security hole.
>
> Co-developed-by: Zhang Yi Z <yi.z.zhang@xxxxxxxxxxxxxxx>
> Signed-off-by: Zhang Yi Z <yi.z.zhang@xxxxxxxxxxxxxxx>
> Signed-off-by: Yang Weijiang <weijiang.yang@xxxxxxxxx>
> ---

...

> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index db03d9dc1297..e392e818e7eb 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -44,6 +44,7 @@
> #include <asm/spec-ctrl.h>
> #include <asm/virtext.h>
> #include <asm/vmx.h>
> +#include <asm/cet.h>
>
> #include "capabilities.h"
> #include "cpuid.h"
> @@ -2336,7 +2337,8 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf,
> VM_EXIT_LOAD_IA32_EFER |
> VM_EXIT_CLEAR_BNDCFGS |
> VM_EXIT_PT_CONCEAL_PIP |
> - VM_EXIT_CLEAR_IA32_RTIT_CTL;
> + VM_EXIT_CLEAR_IA32_RTIT_CTL |
> + VM_EXIT_LOAD_HOST_CET_STATE;
> if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_EXIT_CTLS,
> &_vmexit_control) < 0)
> return -EIO;
> @@ -2360,7 +2362,8 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf,
> VM_ENTRY_LOAD_IA32_EFER |
> VM_ENTRY_LOAD_BNDCFGS |
> VM_ENTRY_PT_CONCEAL_PIP |
> - VM_ENTRY_LOAD_IA32_RTIT_CTL;
> + VM_ENTRY_LOAD_IA32_RTIT_CTL |
> + VM_ENTRY_LOAD_GUEST_CET_STATE;
> if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_ENTRY_CTLS,
> &_vmentry_control) < 0)
> return -EIO;
> @@ -2834,6 +2837,9 @@ void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
> struct vcpu_vmx *vmx = to_vmx(vcpu);
> unsigned long hw_cr0;
>
> + if (!(cr0 & X86_CR0_WP) && kvm_read_cr4_bits(vcpu, X86_CR4_CET))
> + cr0 |= X86_CR0_WP;

Huh? What's the interaction between CR4.CET and CR0.WP? If there really
is some non-standard interaction then it needs to be documented in at least
the changelog and probably with a comment as well.

> +
> hw_cr0 = (cr0 & ~KVM_VM_CR0_ALWAYS_OFF);
> if (enable_unrestricted_guest)
> hw_cr0 |= KVM_VM_CR0_ALWAYS_ON_UNRESTRICTED_GUEST;
> @@ -2936,6 +2942,22 @@ static bool guest_cet_allowed(struct kvm_vcpu *vcpu, u32 feature, u32 mode)
> return false;
> }
>
> +bool is_cet_bit_allowed(struct kvm_vcpu *vcpu)
> +{
> + unsigned long cr0;
> + bool cet_allowed;
> +
> + cr0 = kvm_read_cr0(vcpu);
> + cet_allowed = guest_cet_allowed(vcpu, X86_FEATURE_SHSTK,
> + XFEATURE_MASK_CET_USER) ||
> + guest_cet_allowed(vcpu, X86_FEATURE_IBT,
> + XFEATURE_MASK_CET_USER);
> + if ((cr0 & X86_CR0_WP) && cet_allowed)
> + return true;

So, attempting to set CR4.CET if CR0.WP=0 takes a #GP? But attempting
to clear CR0.WP if CR4.CET=1 is ignored?

> +
> + return false;
> +}
> +
> int vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
> {
> struct vcpu_vmx *vmx = to_vmx(vcpu);
> @@ -2976,6 +2998,9 @@ int vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
> return 1;
> }
>
> + if ((cr4 & X86_CR4_CET) && !is_cet_bit_allowed(vcpu))
> + return 1;
> +
> if (vmx->nested.vmxon && !nested_cr4_valid(vcpu, cr4))
> return 1;
>
> @@ -3839,6 +3864,12 @@ void vmx_set_constant_host_state(struct vcpu_vmx *vmx)
>
> if (cpu_has_load_ia32_efer())
> vmcs_write64(HOST_IA32_EFER, host_efer);
> +
> + if (cpu_has_load_host_cet_states_ctrl()) {
> + vmcs_writel(HOST_S_CET, 0);
> + vmcs_writel(HOST_INTR_SSP_TABLE, 0);
> + vmcs_writel(HOST_SSP, 0);
> + }
> }
>
> void set_cr4_guest_host_mask(struct vcpu_vmx *vmx)
> @@ -6436,6 +6467,7 @@ static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
> {
> struct vcpu_vmx *vmx = to_vmx(vcpu);
> unsigned long cr3, cr4;
> + bool cet_allowed;
>
> /* Record the guest's net vcpu time for enforced NMI injections. */
> if (unlikely(!enable_vnmi &&
> @@ -6466,6 +6498,25 @@ static void vmx_vcpu_run(struct kvm_vcpu *vcpu)
> vmx->loaded_vmcs->host_state.cr3 = cr3;
> }
>
> + /* To be aligned with kernel code, only user mode is supported now. */
> + cet_allowed = guest_cet_allowed(vcpu, X86_FEATURE_SHSTK,
> + XFEATURE_MASK_CET_USER) ||
> + guest_cet_allowed(vcpu, X86_FEATURE_IBT,
> + XFEATURE_MASK_CET_USER);
> + if (cpu_has_load_guest_cet_states_ctrl() && cet_allowed)
> + vmcs_set_bits(VM_ENTRY_CONTROLS,
> + VM_ENTRY_LOAD_GUEST_CET_STATE);
> + else
> + vmcs_clear_bits(VM_ENTRY_CONTROLS,
> + VM_ENTRY_LOAD_GUEST_CET_STATE);
> +
> + if (cpu_has_load_host_cet_states_ctrl() && cet_allowed)
> + vmcs_set_bits(VM_EXIT_CONTROLS,
> + VM_EXIT_LOAD_HOST_CET_STATE);
> + else
> + vmcs_clear_bits(VM_EXIT_CONTROLS,
> + VM_EXIT_LOAD_HOST_CET_STATE);
> +
> cr4 = cr4_read_shadow();
> if (unlikely(cr4 != vmx->loaded_vmcs->host_state.cr4)) {
> vmcs_writel(HOST_CR4, cr4);
> --
> 2.17.2
>