KVM: VMX: EPT Feature Detection

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

2 files changed, 83 insertions, 5 deletions

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 8e5d6645b90d..d93250d11caa 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -42,6 +42,9 @@ module_param(enable_vpid, bool, 0);
 static int flexpriority_enabled = 1;
 module_param(flexpriority_enabled, bool, 0);
 
+static int enable_ept;
+module_param(enable_ept, bool, 0);
+
 struct vmcs {
         u32 revision_id;
         u32 abort;
@@ -107,6 +110,11 @@ static struct vmcs_config {
         u32 vmentry_ctrl;
 } vmcs_config;
 
+struct vmx_capability {
+        u32 ept;
+        u32 vpid;
+} vmx_capability;
+
 #define VMX_SEGMENT_FIELD(seg)                                  \
         [VCPU_SREG_##seg] = {                                   \
                 .selector = GUEST_##seg##_SELECTOR,             \
@@ -214,6 +222,32 @@ static inline bool cpu_has_vmx_virtualize_apic_accesses(void)
                     SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES);
 }
 
+static inline int cpu_has_vmx_invept_individual_addr(void)
+{
+        return (!!(vmx_capability.ept & VMX_EPT_EXTENT_INDIVIDUAL_BIT));
+}
+
+static inline int cpu_has_vmx_invept_context(void)
+{
+        return (!!(vmx_capability.ept & VMX_EPT_EXTENT_CONTEXT_BIT));
+}
+
+static inline int cpu_has_vmx_invept_global(void)
+{
+        return (!!(vmx_capability.ept & VMX_EPT_EXTENT_GLOBAL_BIT));
+}
+
+static inline int cpu_has_vmx_ept(void)
+{
+        return (vmcs_config.cpu_based_2nd_exec_ctrl &
+                SECONDARY_EXEC_ENABLE_EPT);
+}
+
+static inline int vm_need_ept(void)
+{
+        return (cpu_has_vmx_ept() && enable_ept);
+}
+
 static inline int vm_need_virtualize_apic_accesses(struct kvm *kvm)
 {
         return ((cpu_has_vmx_virtualize_apic_accesses()) &&
@@ -985,7 +1019,7 @@ static __init int adjust_vmx_controls(u32 ctl_min, u32 ctl_opt,
 static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 {
         u32 vmx_msr_low, vmx_msr_high;
-        u32 min, opt;
+        u32 min, opt, min2, opt2;
         u32 _pin_based_exec_control = 0;
         u32 _cpu_based_exec_control = 0;
         u32 _cpu_based_2nd_exec_control = 0;
@@ -1003,6 +1037,8 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
               CPU_BASED_CR8_LOAD_EXITING |
               CPU_BASED_CR8_STORE_EXITING |
 #endif
+              CPU_BASED_CR3_LOAD_EXITING |
+              CPU_BASED_CR3_STORE_EXITING |
               CPU_BASED_USE_IO_BITMAPS |
               CPU_BASED_MOV_DR_EXITING |
               CPU_BASED_USE_TSC_OFFSETING;
@@ -1018,11 +1054,13 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
                                            ~CPU_BASED_CR8_STORE_EXITING;
 #endif
         if (_cpu_based_exec_control & CPU_BASED_ACTIVATE_SECONDARY_CONTROLS) {
-                min = 0;
-                opt = SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES |
+                min2 = 0;
+                opt2 = SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES |
                         SECONDARY_EXEC_WBINVD_EXITING |
-                        SECONDARY_EXEC_ENABLE_VPID;
-                if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PROCBASED_CTLS2,
+                        SECONDARY_EXEC_ENABLE_VPID |
+                        SECONDARY_EXEC_ENABLE_EPT;
+                if (adjust_vmx_controls(min2, opt2,
+                                        MSR_IA32_VMX_PROCBASED_CTLS2,
                                         &_cpu_based_2nd_exec_control) < 0)
                         return -EIO;
         }
@@ -1031,6 +1069,16 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
                                 SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES))
                 _cpu_based_exec_control &= ~CPU_BASED_TPR_SHADOW;
 #endif
+        if (_cpu_based_2nd_exec_control & SECONDARY_EXEC_ENABLE_EPT) {
+                /* CR3 accesses don't need to cause VM Exits when EPT enabled */
+                min &= ~(CPU_BASED_CR3_LOAD_EXITING |
+                         CPU_BASED_CR3_STORE_EXITING);
+                if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_PROCBASED_CTLS,
+                                        &_cpu_based_exec_control) < 0)
+                        return -EIO;
+                rdmsr(MSR_IA32_VMX_EPT_VPID_CAP,
+                      vmx_capability.ept, vmx_capability.vpid);
+        }
 
         min = 0;
 #ifdef CONFIG_X86_64
@@ -1638,6 +1686,9 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
                                 CPU_BASED_CR8_LOAD_EXITING;
 #endif
         }
+        if (!vm_need_ept())
+                exec_control |= CPU_BASED_CR3_STORE_EXITING |
+                                CPU_BASED_CR3_LOAD_EXITING;
         vmcs_write32(CPU_BASED_VM_EXEC_CONTROL, exec_control);
 
         if (cpu_has_secondary_exec_ctrls()) {
@@ -1647,6 +1698,8 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
                                 ~SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES;
                 if (vmx->vpid == 0)
                         exec_control &= ~SECONDARY_EXEC_ENABLE_VPID;
+                if (!vm_need_ept())
+                        exec_control &= ~SECONDARY_EXEC_ENABLE_EPT;
                 vmcs_write32(SECONDARY_VM_EXEC_CONTROL, exec_control);
         }
 
diff --git a/arch/x86/kvm/vmx.h b/arch/x86/kvm/vmx.h
index 5dff4606b988..5f7fdc965d39 100644
--- a/arch/x86/kvm/vmx.h
+++ b/arch/x86/kvm/vmx.h
@@ -35,6 +35,8 @@
 #define CPU_BASED_MWAIT_EXITING                 0x00000400
 #define CPU_BASED_RDPMC_EXITING                 0x00000800
 #define CPU_BASED_RDTSC_EXITING                 0x00001000
+#define CPU_BASED_CR3_LOAD_EXITING              0x00008000
+#define CPU_BASED_CR3_STORE_EXITING             0x00010000
 #define CPU_BASED_CR8_LOAD_EXITING              0x00080000
 #define CPU_BASED_CR8_STORE_EXITING             0x00100000
 #define CPU_BASED_TPR_SHADOW                    0x00200000
@@ -49,6 +51,7 @@
  * Definitions of Secondary Processor-Based VM-Execution Controls.
  */
 #define SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES 0x00000001
+#define SECONDARY_EXEC_ENABLE_EPT               0x00000002
 #define SECONDARY_EXEC_ENABLE_VPID              0x00000020
 #define SECONDARY_EXEC_WBINVD_EXITING           0x00000040
 
@@ -100,10 +103,22 @@ enum vmcs_field {
         VIRTUAL_APIC_PAGE_ADDR_HIGH     = 0x00002013,
         APIC_ACCESS_ADDR                = 0x00002014,
         APIC_ACCESS_ADDR_HIGH           = 0x00002015,
+        EPT_POINTER                     = 0x0000201a,
+        EPT_POINTER_HIGH                = 0x0000201b,
+        GUEST_PHYSICAL_ADDRESS          = 0x00002400,
+        GUEST_PHYSICAL_ADDRESS_HIGH     = 0x00002401,
         VMCS_LINK_POINTER               = 0x00002800,
         VMCS_LINK_POINTER_HIGH          = 0x00002801,
         GUEST_IA32_DEBUGCTL             = 0x00002802,
         GUEST_IA32_DEBUGCTL_HIGH        = 0x00002803,
+        GUEST_PDPTR0                    = 0x0000280a,
+        GUEST_PDPTR0_HIGH               = 0x0000280b,
+        GUEST_PDPTR1                    = 0x0000280c,
+        GUEST_PDPTR1_HIGH               = 0x0000280d,
+        GUEST_PDPTR2                    = 0x0000280e,
+        GUEST_PDPTR2_HIGH               = 0x0000280f,
+        GUEST_PDPTR3                    = 0x00002810,
+        GUEST_PDPTR3_HIGH               = 0x00002811,
         PIN_BASED_VM_EXEC_CONTROL       = 0x00004000,
         CPU_BASED_VM_EXEC_CONTROL       = 0x00004002,
         EXCEPTION_BITMAP                = 0x00004004,
@@ -226,6 +241,8 @@ enum vmcs_field {
 #define EXIT_REASON_MWAIT_INSTRUCTION   36
 #define EXIT_REASON_TPR_BELOW_THRESHOLD 43
 #define EXIT_REASON_APIC_ACCESS         44
+#define EXIT_REASON_EPT_VIOLATION       48
+#define EXIT_REASON_EPT_MISCONFIG       49
 #define EXIT_REASON_WBINVD              54
 
 /*
@@ -316,6 +333,7 @@ enum vmcs_field {
 #define MSR_IA32_VMX_CR4_FIXED1                 0x489
 #define MSR_IA32_VMX_VMCS_ENUM                  0x48a
 #define MSR_IA32_VMX_PROCBASED_CTLS2            0x48b
+#define MSR_IA32_VMX_EPT_VPID_CAP               0x48c
 
 #define MSR_IA32_FEATURE_CONTROL                0x3a
 #define MSR_IA32_FEATURE_CONTROL_LOCKED         0x1
@@ -327,4 +345,11 @@ enum vmcs_field {
 #define VMX_VPID_EXTENT_SINGLE_CONTEXT          1
 #define VMX_VPID_EXTENT_ALL_CONTEXT             2
 
+#define VMX_EPT_EXTENT_INDIVIDUAL_ADDR          0
+#define VMX_EPT_EXTENT_CONTEXT                  1
+#define VMX_EPT_EXTENT_GLOBAL                   2
+#define VMX_EPT_EXTENT_INDIVIDUAL_BIT           (1ull << 24)
+#define VMX_EPT_EXTENT_CONTEXT_BIT              (1ull << 25)
+#define VMX_EPT_EXTENT_GLOBAL_BIT               (1ull << 26)
+
 #endif