[Qemu-devel] [PATCH 27/31] target-i386: add support for SMBASE MSR and SMIs

[Paolo Bonzini <pbonzini@redhat.com>]

Apart from the MSR, the smi field of struct kvm_vcpu_events has to be
translated into the corresponding CPUX86State fields.  Also,
memory transaction flags depend on SMM state, so pull it from struct
kvm_run on every exit from KVM to userspace.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 linux-headers/asm-x86/kvm.h | 11 ++++++-
 linux-headers/linux/kvm.h   |  5 +++-
 target-i386/cpu.h           |  1 +
 target-i386/kvm.c           | 73 ++++++++++++++++++++++++++++++++++++++++-----
 4 files changed, 81 insertions(+), 9 deletions(-)

diff --git a/linux-headers/asm-x86/kvm.h b/linux-headers/asm-x86/kvm.h
index d7dcef5..cf84866 100644
--- a/linux-headers/asm-x86/kvm.h
+++ b/linux-headers/asm-x86/kvm.h
@@ -48,6 +48,8 @@
 /* Architectural interrupt line count. */
 #define KVM_NR_INTERRUPTS 256
 
+#define KVM_RUN_X86_SMM (1 << 0)
+
 struct kvm_memory_alias {
 	__u32 slot;  /* this has a different namespace than memory slots */
 	__u32 flags;
@@ -281,6 +283,7 @@ struct kvm_reinject_control {
 #define KVM_VCPUEVENT_VALID_NMI_PENDING	0x00000001
 #define KVM_VCPUEVENT_VALID_SIPI_VECTOR	0x00000002
 #define KVM_VCPUEVENT_VALID_SHADOW	0x00000004
+#define KVM_VCPUEVENT_VALID_SMM		0x00000008
 
 /* Interrupt shadow states */
 #define KVM_X86_SHADOW_INT_MOV_SS	0x01
@@ -309,7 +312,13 @@ struct kvm_vcpu_events {
 	} nmi;
 	__u32 sipi_vector;
 	__u32 flags;
-	__u32 reserved[10];
+	struct {
+		__u8 smm;
+		__u8 pending;
+		__u8 smm_inside_nmi;
+		__u8 pad;
+	} smi;
+	__u32 reserved[9];
 };
 
 /* for KVM_GET/SET_DEBUGREGS */
diff --git a/linux-headers/linux/kvm.h b/linux-headers/linux/kvm.h
index b96d978..15beebb 100644
--- a/linux-headers/linux/kvm.h
+++ b/linux-headers/linux/kvm.h
@@ -202,7 +202,7 @@ struct kvm_run {
 	__u32 exit_reason;
 	__u8 ready_for_interrupt_injection;
 	__u8 if_flag;
-	__u8 padding2[2];
+	__u16 flags;
 
 	/* in (pre_kvm_run), out (post_kvm_run) */
 	__u64 cr8;
@@ -814,6 +814,7 @@ struct kvm_ppc_smmu_info {
 #define KVM_CAP_S390_INJECT_IRQ 113
 #define KVM_CAP_S390_IRQ_STATE 114
 #define KVM_CAP_PPC_HWRNG 115
+#define KVM_CAP_X86_SMM 116
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -1199,6 +1200,8 @@ struct kvm_s390_ucas_mapping {
 /* Available with KVM_CAP_S390_IRQ_STATE */
 #define KVM_S390_SET_IRQ_STATE	  _IOW(KVMIO, 0xb5, struct kvm_s390_irq_state)
 #define KVM_S390_GET_IRQ_STATE	  _IOW(KVMIO, 0xb6, struct kvm_s390_irq_state)
+/* Available with KVM_CAP_X86_SMM */
+#define KVM_SMI                   _IO(KVMIO,   0xb7)
 
 #define KVM_DEV_ASSIGN_ENABLE_IOMMU	(1 << 0)
 #define KVM_DEV_ASSIGN_PCI_2_3		(1 << 1)
diff --git a/target-i386/cpu.h b/target-i386/cpu.h
index df6e885..684e3a0 100644
--- a/target-i386/cpu.h
+++ b/target-i386/cpu.h
@@ -314,6 +314,7 @@
 
 #define MSR_P6_PERFCTR0                 0xc1
 
+#define MSR_IA32_SMBASE                 0x9e
 #define MSR_MTRRcap                     0xfe
 #define MSR_MTRRcap_VCNT                8
 #define MSR_MTRRcap_FIXRANGE_SUPPORT    (1 << 8)
diff --git a/target-i386/kvm.c b/target-i386/kvm.c
index 009bf74..09b4fc7 100644
--- a/target-i386/kvm.c
+++ b/target-i386/kvm.c
@@ -74,6 +74,7 @@ static bool has_msr_feature_control;
 static bool has_msr_async_pf_en;
 static bool has_msr_pv_eoi_en;
 static bool has_msr_misc_enable;
+static bool has_msr_smbase;
 static bool has_msr_bndcfgs;
 static bool has_msr_kvm_steal_time;
 static int lm_capable_kernel;
@@ -820,6 +821,10 @@ static int kvm_get_supported_msrs(KVMState *s)
                     has_msr_tsc_deadline = true;
                     continue;
                 }
+                if (kvm_msr_list->indices[i] == MSR_IA32_SMBASE) {
+                    has_msr_smbase = true;
+                    continue;
+                }
                 if (kvm_msr_list->indices[i] == MSR_IA32_MISC_ENABLE) {
                     has_msr_misc_enable = true;
                     continue;
@@ -1246,6 +1251,9 @@ static int kvm_put_msrs(X86CPU *cpu, int level)
         kvm_msr_entry_set(&msrs[n++], MSR_IA32_MISC_ENABLE,
                           env->msr_ia32_misc_enable);
     }
+    if (has_msr_smbase) {
+        kvm_msr_entry_set(&msrs[n++], MSR_IA32_SMBASE, env->smbase);
+    }
     if (has_msr_bndcfgs) {
         kvm_msr_entry_set(&msrs[n++], MSR_IA32_BNDCFGS, env->msr_bndcfgs);
     }
@@ -1607,6 +1615,9 @@ static int kvm_get_msrs(X86CPU *cpu)
     if (has_msr_misc_enable) {
         msrs[n++].index = MSR_IA32_MISC_ENABLE;
     }
+    if (has_msr_smbase) {
+        msrs[n++].index = MSR_IA32_SMBASE;
+    }
     if (has_msr_feature_control) {
         msrs[n++].index = MSR_IA32_FEATURE_CONTROL;
     }
@@ -1761,6 +1772,9 @@ static int kvm_get_msrs(X86CPU *cpu)
         case MSR_IA32_MISC_ENABLE:
             env->msr_ia32_misc_enable = msrs[i].data;
             break;
+        case MSR_IA32_SMBASE:
+            env->smbase = msrs[i].data;
+            break;
         case MSR_IA32_FEATURE_CONTROL:
             env->msr_ia32_feature_control = msrs[i].data;
             break;
@@ -1948,6 +1962,16 @@ static int kvm_put_vcpu_events(X86CPU *cpu, int level)
 
     events.sipi_vector = env->sipi_vector;
 
+    if (has_msr_smbase) {
+        events.smi.smm = !!(env->hflags & HF_SMM_MASK);
+        /* smi.pending is stored as CPU_INTERRUPT_SMI in cpu->interrupt_request.
+         * It is transferred to the kernel before the next vmentry
+         */
+        events.smi.pending = 0;
+        events.smi.smm_inside_nmi = !!(env->hflags2 & HF2_SMM_INSIDE_NMI_MASK);
+        events.flags |= KVM_VCPUEVENT_VALID_SMM;
+    }
+
     events.flags = 0;
     if (level >= KVM_PUT_RESET_STATE) {
         events.flags |=
@@ -1967,6 +1991,7 @@ static int kvm_get_vcpu_events(X86CPU *cpu)
         return 0;
     }
 
+    memset(&events, 0, sizeof(events));
     ret = kvm_vcpu_ioctl(CPU(cpu), KVM_GET_VCPU_EVENTS, &events);
     if (ret < 0) {
        return ret;
@@ -1988,6 +2013,24 @@ static int kvm_get_vcpu_events(X86CPU *cpu)
         env->hflags2 &= ~HF2_NMI_MASK;
     }
 
+    if (events.flags & KVM_VCPUEVENT_VALID_SMM) {
+        if (events.smi.smm) {
+            env->hflags |= HF_SMM_MASK;
+        } else {
+            env->hflags &= ~HF_SMM_MASK;
+        }
+        if (events.smi.pending) {
+            cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
+        } else {
+            cpu_reset_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
+        }
+        if (events.smi.smm_inside_nmi) {
+            env->hflags2 |= HF2_SMM_INSIDE_NMI_MASK;
+        } else {
+            env->hflags2 &= ~HF2_SMM_INSIDE_NMI_MASK;
+        }
+    }
+
     env->sipi_vector = events.sipi_vector;
 
     return 0;
@@ -2191,13 +2234,24 @@ void kvm_arch_pre_run(CPUState *cpu, struct kvm_run *run)
     int ret;
 
     /* Inject NMI */
-    if (cpu->interrupt_request & CPU_INTERRUPT_NMI) {
-        cpu->interrupt_request &= ~CPU_INTERRUPT_NMI;
-        DPRINTF("injected NMI\n");
-        ret = kvm_vcpu_ioctl(cpu, KVM_NMI);
-        if (ret < 0) {
-            fprintf(stderr, "KVM: injection failed, NMI lost (%s)\n",
-                    strerror(-ret));
+    if (cpu->interrupt_request & (CPU_INTERRUPT_NMI | CPU_INTERRUPT_SMI)) {
+        if (cpu->interrupt_request & CPU_INTERRUPT_NMI) {
+            cpu->interrupt_request &= ~CPU_INTERRUPT_NMI;
+            DPRINTF("injected NMI\n");
+            ret = kvm_vcpu_ioctl(cpu, KVM_NMI);
+            if (ret < 0) {
+                fprintf(stderr, "KVM: injection failed, NMI lost (%s)\n",
+                        strerror(-ret));
+            }
+        }
+        if (cpu->interrupt_request & CPU_INTERRUPT_SMI) {
+            cpu->interrupt_request &= ~CPU_INTERRUPT_SMI;
+            DPRINTF("injected SMI\n");
+            ret = kvm_vcpu_ioctl(cpu, KVM_SMI);
+            if (ret < 0) {
+                fprintf(stderr, "KVM: injection failed, SMI lost (%s)\n",
+                        strerror(-ret));
+            }
         }
     }
 
@@ -2252,6 +2306,11 @@ MemTxAttrs kvm_arch_post_run(CPUState *cpu, struct kvm_run *run)
     X86CPU *x86_cpu = X86_CPU(cpu);
     CPUX86State *env = &x86_cpu->env;
 
+    if (run->flags & KVM_RUN_X86_SMM) {
+        env->hflags |= HF_SMM_MASK;
+    } else {
+        env->hflags &= HF_SMM_MASK;
+    }
     if (run->if_flag) {
         env->eflags |= IF_MASK;
     } else {
-- 
1.8.3.1