[PATCH kernel 2/3] KVM: SEV: Enable DebugSwap

From: Alexey Kardashevskiy
Date: Wed Nov 30 2022 - 21:22:24 EST


AMD Milan introduces support for the swapping, as type 'B',
of DR[0-3] and DR[0-3]_ADDR_MASK registers. It requires that
SEV_FEATURES[5] be set in the VMSA.

This requires the KVM to eliminate the intercept of #DB. However,
because of the infinite #DB loop DoS that a malicious guest can do,
it can only be eliminated based if CPUID Fn80000021_EAX[0]
(NoNestedDataBp) is set in the host/HV.

This eliminates #DB intercept, DR7 intercept for SEV-ES/SEV-SNP guest.
This saves DR[0-3] / DR[0-3]_ADDR_MASK in the host save area before VMRUN.
This sets SEV_FEATURES[5] in VMSA.

Signed-off-by: Alexey Kardashevskiy <aik@xxxxxxx>
---
arch/x86/include/asm/svm.h | 1 +
arch/x86/kvm/svm/svm.h | 18 +++++++++++-----
arch/x86/kvm/svm/sev.c | 22 +++++++++++++++++++-
arch/x86/kvm/svm/svm.c | 6 ++++--
4 files changed, 39 insertions(+), 8 deletions(-)

diff --git a/arch/x86/include/asm/svm.h b/arch/x86/include/asm/svm.h
index 0361626841bc..373a0edda588 100644
--- a/arch/x86/include/asm/svm.h
+++ b/arch/x86/include/asm/svm.h
@@ -273,6 +273,7 @@ enum avic_ipi_failure_cause {
#define AVIC_HPA_MASK ~((0xFFFULL << 52) | 0xFFF)
#define VMCB_AVIC_APIC_BAR_MASK 0xFFFFFFFFFF000ULL

+#define SVM_SEV_FEAT_DEBUG_SWAP BIT(5)

struct vmcb_seg {
u16 selector;
diff --git a/arch/x86/kvm/svm/svm.h b/arch/x86/kvm/svm/svm.h
index 199a2ecef1ce..4d75b14bffab 100644
--- a/arch/x86/kvm/svm/svm.h
+++ b/arch/x86/kvm/svm/svm.h
@@ -83,6 +83,7 @@ enum {
struct kvm_sev_info {
bool active; /* SEV enabled guest */
bool es_active; /* SEV-ES enabled guest */
+ bool debug_swap; /* SEV-ES Debug swap enabled */
unsigned int asid; /* ASID used for this guest */
unsigned int handle; /* SEV firmware handle */
int fd; /* SEV device fd */
@@ -388,6 +389,7 @@ static inline bool vmcb12_is_intercept(struct vmcb_ctrl_area_cached *control, u3

static inline void set_dr_intercepts(struct vcpu_svm *svm)
{
+ struct kvm_sev_info *sev = &to_kvm_svm(svm->vcpu.kvm)->sev_info;
struct vmcb *vmcb = svm->vmcb01.ptr;

if (!sev_es_guest(svm->vcpu.kvm)) {
@@ -407,20 +409,26 @@ static inline void set_dr_intercepts(struct vcpu_svm *svm)
vmcb_set_intercept(&vmcb->control, INTERCEPT_DR6_WRITE);
}

- vmcb_set_intercept(&vmcb->control, INTERCEPT_DR7_READ);
- vmcb_set_intercept(&vmcb->control, INTERCEPT_DR7_WRITE);
+ if (!sev->debug_swap) {
+ vmcb_set_intercept(&vmcb->control, INTERCEPT_DR7_READ);
+ vmcb_set_intercept(&vmcb->control, INTERCEPT_DR7_WRITE);
+ }

recalc_intercepts(svm);
}

static inline void clr_dr_intercepts(struct vcpu_svm *svm)
{
+ struct kvm_sev_info *sev = &to_kvm_svm(svm->vcpu.kvm)->sev_info;
struct vmcb *vmcb = svm->vmcb01.ptr;

vmcb->control.intercepts[INTERCEPT_DR] = 0;

- /* DR7 access must remain intercepted for an SEV-ES guest */
- if (sev_es_guest(svm->vcpu.kvm)) {
+ /*
+ * DR7 access must remain intercepted for an SEV-ES guest unless
+ * the DebugSwap feature is set
+ */
+ if (sev_es_guest(svm->vcpu.kvm) && !sev->debug_swap) {
vmcb_set_intercept(&vmcb->control, INTERCEPT_DR7_READ);
vmcb_set_intercept(&vmcb->control, INTERCEPT_DR7_WRITE);
}
@@ -677,7 +685,7 @@ int sev_handle_vmgexit(struct kvm_vcpu *vcpu);
int sev_es_string_io(struct vcpu_svm *svm, int size, unsigned int port, int in);
void sev_es_vcpu_reset(struct vcpu_svm *svm);
void sev_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector);
-void sev_es_prepare_switch_to_guest(struct sev_es_save_area *hostsa);
+void sev_es_prepare_switch_to_guest(struct kvm_vcpu *vcpu, struct sev_es_save_area *hostsa);
void sev_es_unmap_ghcb(struct vcpu_svm *svm);

/* vmenter.S */
diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index efaaef2b7ae1..fac8b48e3162 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -21,6 +21,7 @@
#include <asm/pkru.h>
#include <asm/trapnr.h>
#include <asm/fpu/xcr.h>
+#include <asm/debugreg.h>

#include "mmu.h"
#include "x86.h"
@@ -253,6 +254,7 @@ static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
if (asid < 0)
goto e_no_asid;
sev->asid = asid;
+ sev->debug_swap = sev->es_active && kvm_cpu_cap_get(KVM_X86_FEATURE_NO_NESTED_DATA_BP);

ret = sev_platform_init(&argp->error);
if (ret)
@@ -564,6 +566,7 @@ static int sev_launch_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp)
static int sev_es_sync_vmsa(struct vcpu_svm *svm)
{
struct sev_es_save_area *save = svm->sev_es.vmsa;
+ struct kvm_sev_info *sev = &to_kvm_svm(svm->vcpu.kvm)->sev_info;

/* Check some debug related fields before encrypting the VMSA */
if (svm->vcpu.guest_debug || (svm->vmcb->save.dr7 & ~DR7_FIXED_1))
@@ -604,6 +607,9 @@ static int sev_es_sync_vmsa(struct vcpu_svm *svm)
save->xss = svm->vcpu.arch.ia32_xss;
save->dr6 = svm->vcpu.arch.dr6;

+ if (sev->debug_swap)
+ save->sev_features |= SVM_SEV_FEAT_DEBUG_SWAP;
+
pr_debug("Virtual Machine Save Area (VMSA):\n");
print_hex_dump_debug("", DUMP_PREFIX_NONE, 16, 1, save, sizeof(*save), false);

@@ -3010,8 +3016,10 @@ void sev_es_vcpu_reset(struct vcpu_svm *svm)
sev_enc_bit));
}

-void sev_es_prepare_switch_to_guest(struct sev_es_save_area *hostsa)
+void sev_es_prepare_switch_to_guest(struct kvm_vcpu *vcpu, struct sev_es_save_area *hostsa)
{
+ struct kvm_sev_info *sev = &to_kvm_svm(vcpu->kvm)->sev_info;
+
/*
* As an SEV-ES guest, hardware will restore the host state on VMEXIT,
* of which one step is to perform a VMLOAD. KVM performs the
@@ -3027,6 +3035,18 @@ void sev_es_prepare_switch_to_guest(struct sev_es_save_area *hostsa)

/* MSR_IA32_XSS is restored on VMEXIT, save the currnet host value */
hostsa->xss = host_xss;
+
+ /* The DebugSwap SEV feature does Type B swaps of DR[0-3] */
+ if (sev->debug_swap) {
+ hostsa->dr0 = native_get_debugreg(0);
+ hostsa->dr1 = native_get_debugreg(1);
+ hostsa->dr2 = native_get_debugreg(2);
+ hostsa->dr3 = native_get_debugreg(3);
+ hostsa->dr0_addr_mask = get_dr_addr_mask(0);
+ hostsa->dr1_addr_mask = get_dr_addr_mask(1);
+ hostsa->dr2_addr_mask = get_dr_addr_mask(2);
+ hostsa->dr3_addr_mask = get_dr_addr_mask(3);
+ }
}

void sev_vcpu_deliver_sipi_vector(struct kvm_vcpu *vcpu, u8 vector)
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index ce362e88a567..ee0e56521d26 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -1173,6 +1173,7 @@ static void init_vmcb(struct kvm_vcpu *vcpu)
struct vmcb *vmcb = svm->vmcb01.ptr;
struct vmcb_control_area *control = &vmcb->control;
struct vmcb_save_area *save = &vmcb->save;
+ struct kvm_sev_info *sev = &to_kvm_svm(vcpu->kvm)->sev_info;

svm_set_intercept(svm, INTERCEPT_CR0_READ);
svm_set_intercept(svm, INTERCEPT_CR3_READ);
@@ -1189,7 +1190,8 @@ static void init_vmcb(struct kvm_vcpu *vcpu)
set_exception_intercept(svm, UD_VECTOR);
set_exception_intercept(svm, MC_VECTOR);
set_exception_intercept(svm, AC_VECTOR);
- set_exception_intercept(svm, DB_VECTOR);
+ if (!sev->debug_swap)
+ set_exception_intercept(svm, DB_VECTOR);
/*
* Guest access to VMware backdoor ports could legitimately
* trigger #GP because of TSS I/O permission bitmap.
@@ -1461,7 +1463,7 @@ static void svm_prepare_switch_to_guest(struct kvm_vcpu *vcpu)
struct sev_es_save_area *hostsa;
hostsa = (struct sev_es_save_area *)(page_address(sd->save_area) + 0x400);

- sev_es_prepare_switch_to_guest(hostsa);
+ sev_es_prepare_switch_to_guest(vcpu, hostsa);
}

if (tsc_scaling)
--
2.38.1