Re: [PATCH v4 11/20] KVM:x86: Save and reload GUEST_SSP to/from SMRAM

[Chao Gao]

On Thu, Jul 20, 2023 at 11:03:43PM -0400, Yang Weijiang wrote:
>Save GUEST_SSP to SMRAM on SMI and reload it on RSM.
>KVM emulates architectural behavior when guest enters/leaves SMM
>mode, i.e., save registers to SMRAM at the entry of SMM and reload
>them at the exit of SMM. Per SDM, GUEST_SSP is defined as one of

To me, GUEST_SSP is confusing here. From QEMU's perspective, it reads/writes
the SSP register. People may confuse it with the GUEST_SSP in nVMCS field.
I prefer to rename it to MSR_KVM_SSP.

>the fields in SMRAM for 64-bit mode, so handle the state accordingly.
>
>Check HF_SMM_MASK to determine whether kvm_cet_is_msr_accessible()
>is called in SMM mode so that kvm_{set,get}_msr() works in SMM mode.
>
>Signed-off-by: Yang Weijiang <weijiang.yang@xxxxxxxxx>
>---
> arch/x86/kvm/smm.c | 17 +++++++++++++++++
> arch/x86/kvm/smm.h |  2 +-
> arch/x86/kvm/x86.c | 12 +++++++++++-
> 3 files changed, 29 insertions(+), 2 deletions(-)
>
>diff --git a/arch/x86/kvm/smm.c b/arch/x86/kvm/smm.c
>index b42111a24cc2..a4e19d72224f 100644
>--- a/arch/x86/kvm/smm.c
>+++ b/arch/x86/kvm/smm.c
>@@ -309,6 +309,15 @@ void enter_smm(struct kvm_vcpu *vcpu)
> 
> 	kvm_smm_changed(vcpu, true);
> 
>+#ifdef CONFIG_X86_64
>+	if (guest_can_use(vcpu, X86_FEATURE_SHSTK)) {
>+		u64 data;
>+
>+		if (!kvm_get_msr(vcpu, MSR_KVM_GUEST_SSP, &data))
>+			smram.smram64.ssp = data;

I don't think it is correct to continue if kvm fails to read the MSR.

how about:
		if (kvm_get_msr(vcpu, MSR_KVM_GUEST_SSP, &smram.smram64.ssp))
			goto error;
>+	}
>+#endif
>+
> 	if (kvm_vcpu_write_guest(vcpu, vcpu->arch.smbase + 0xfe00, &smram, sizeof(smram)))
> 		goto error;
> 
>@@ -586,6 +595,14 @@ int emulator_leave_smm(struct x86_emulate_ctxt *ctxt)
> 	if ((vcpu->arch.hflags & HF_SMM_INSIDE_NMI_MASK) == 0)
> 		static_call(kvm_x86_set_nmi_mask)(vcpu, false);
> 
>+#ifdef CONFIG_X86_64
>+	if (guest_can_use(vcpu, X86_FEATURE_SHSTK)) {
>+		u64 data = smram.smram64.ssp;
>+

>+		if (is_noncanonical_address(data, vcpu) && IS_ALIGNED(data, 4))

shouldn't the checks be already done inside kvm_set_msr()?

>+			kvm_set_msr(vcpu, MSR_KVM_GUEST_SSP, data);

please handle the failure case. Probably just return X86EMUL_UNHANDLEABLE like other
failure paths in this function.

>+	}
>+#endif
> 	kvm_smm_changed(vcpu, false);
> 
> 	/*
>diff --git a/arch/x86/kvm/smm.h b/arch/x86/kvm/smm.h
>index a1cf2ac5bd78..b3efef7cb1dc 100644
>--- a/arch/x86/kvm/smm.h
>+++ b/arch/x86/kvm/smm.h
>@@ -116,7 +116,7 @@ struct kvm_smram_state_64 {
> 	u32 smbase;
> 	u32 reserved4[5];
> 
>-	/* ssp and svm_* fields below are not implemented by KVM */
>+	/* svm_* fields below are not implemented by KVM */

move this comment one line downward

> 	u64 ssp;
> 	u64 svm_guest_pat;
> 	u64 svm_host_efer;
>diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>index f7558f0f6fc0..70d7c80889d6 100644
>--- a/arch/x86/kvm/x86.c
>+++ b/arch/x86/kvm/x86.c
>@@ -3653,8 +3653,18 @@ static bool kvm_cet_is_msr_accessible(struct kvm_vcpu *vcpu,
> 		if (!kvm_cpu_cap_has(X86_FEATURE_SHSTK))
> 			return false;
> 
>-		if (msr->index == MSR_KVM_GUEST_SSP)
>+		/*
>+		 * This MSR is synthesized mainly for userspace access during
>+		 * Live Migration, it also can be accessed in SMM mode by VMM.
>+		 * Guest is not allowed to access this MSR.
>+		 */
>+		if (msr->index == MSR_KVM_GUEST_SSP) {
>+			if (IS_ENABLED(CONFIG_X86_64) &&
>+			    !!(vcpu->arch.hflags & HF_SMM_MASK))

use is_smm() instead.

>+				return true;
>+
> 			return msr->host_initiated;
>+		}
> 
> 		return msr->host_initiated ||
> 			guest_cpuid_has(vcpu, X86_FEATURE_SHSTK);
>-- 
>2.27.0
>