x86: fix EFER handling

Introduce a per-CPU shadow of what is currently in EFER, as context
switch code must re-write this MSR so that all guests run with
appropriate EFER.SCE and EFER.NX settings. Remove EFER from the set of
MSRs that VMX deals with in a generic fashion.

Signed-off-by: Jan Beulich <jbeulich@novell.com>

diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index 8267c5043bb13455ba444c0d0d91c0b0365dd77e..121c484d47b7af55370f944b181944dd848e6f91 100644 (file)
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -47,6 +47,7 @@
 #endif
 
 DEFINE_PER_CPU(struct vcpu *, curr_vcpu);
+DEFINE_PER_CPU(__u64, efer);
 
 static void paravirt_ctxt_switch_from(struct vcpu *v);
 static void paravirt_ctxt_switch_to(struct vcpu *v);
@@ -1135,21 +1136,18 @@ void context_switch(struct vcpu *prev, struct vcpu *next)
         __context_switch();
 
 #ifdef CONFIG_COMPAT
-        if ( is_idle_vcpu(prev) ||
-             (is_pv_32on64_domain(prev->domain) !=
-              is_pv_32on64_domain(next->domain)) )
+        if ( !is_hvm_vcpu(next) &&
+             (is_idle_vcpu(prev) ||
+              is_hvm_vcpu(prev) ||
+              is_pv_32on64_vcpu(prev) != is_pv_32on64_vcpu(next)) )
         {
-            uint32_t efer_lo, efer_hi;
+            uint64_t efer = read_efer();
 
             local_flush_tlb_one(GDT_VIRT_START(next) +
                                 FIRST_RESERVED_GDT_BYTE);
 
-            rdmsr(MSR_EFER, efer_lo, efer_hi);
-            if ( !is_pv_32on64_domain(next->domain) == !(efer_lo & EFER_SCE) )
-            {
-                efer_lo ^= EFER_SCE;
-                wrmsr(MSR_EFER, efer_lo, efer_hi);
-            }
+            if ( !is_pv_32on64_vcpu(next) == !(efer & EFER_SCE) )
+                write_efer(efer ^ EFER_SCE);
         }
 #endif
 

diff --git a/xen/arch/x86/hvm/svm/svm.c b/xen/arch/x86/hvm/svm/svm.c
index 9efa866158da6921a23cc199f190f0ce84189e04..3b55a427679455ae90a690805c05054d4157c8bf 100644 (file)
--- a/xen/arch/x86/hvm/svm/svm.c
+++ b/xen/arch/x86/hvm/svm/svm.c
@@ -93,11 +93,8 @@ static inline void svm_inject_exception(struct vcpu *v, int trap,
 
 static void stop_svm(void)
 {
-    u32 eax, edx;    
     /* We turn off the EFER_SVME bit. */
-    rdmsr(MSR_EFER, eax, edx);
-    eax &= ~EFER_SVME;
-    wrmsr(MSR_EFER, eax, edx);
+    write_efer(read_efer() & ~EFER_SVME);
 }
 
 static void svm_store_cpu_guest_regs(
@@ -138,7 +135,13 @@ static inline int long_mode_do_msr_write(struct cpu_user_regs *regs)
     {
     case MSR_EFER:
         /* Offending reserved bit will cause #GP. */
-        if ( msr_content & ~(EFER_LME | EFER_LMA | EFER_NX | EFER_SCE) )
+#ifdef __x86_64__
+        if ( (msr_content & ~(EFER_LME | EFER_LMA | EFER_NX | EFER_SCE)) ||
+#else
+        if ( (msr_content & ~(EFER_NX | EFER_SCE)) ||
+#endif
+             (!cpu_has_nx && (msr_content & EFER_NX)) ||
+             (!cpu_has_syscall && (msr_content & EFER_SCE)) )
         {
             gdprintk(XENLOG_WARNING, "Trying to set reserved bit in "
                      "EFER: %"PRIx64"\n", msr_content);
@@ -495,7 +498,7 @@ int svm_vmcb_restore(struct vcpu *v, struct hvm_hw_cpu *c)
 }
 
         
-void svm_save_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
+static void svm_save_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
 {
     struct vmcb_struct *vmcb = v->arch.hvm_svm.vmcb;
 
@@ -511,7 +514,7 @@ void svm_save_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
 }
 
 
-void svm_load_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
+static void svm_load_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
 {
     struct vmcb_struct *vmcb = v->arch.hvm_svm.vmcb;
 
@@ -530,13 +533,13 @@ void svm_load_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
     hvm_set_guest_time(v, data->tsc);
 }
 
-void svm_save_vmcb_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
+static void svm_save_vmcb_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
 {
     svm_save_cpu_state(v, ctxt);
     svm_vmcb_save(v, ctxt);
 }
 
-int svm_load_vmcb_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
+static int svm_load_vmcb_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
 {
     svm_load_cpu_state(v, ctxt);
     if (svm_vmcb_restore(v, ctxt)) {
@@ -911,6 +914,7 @@ static struct hvm_function_table svm_function_table = {
     .paging_enabled       = svm_paging_enabled,
     .long_mode_enabled    = svm_long_mode_enabled,
     .pae_enabled          = svm_pae_enabled,
+    .nx_enabled           = svm_nx_enabled,
     .interrupts_enabled   = svm_interrupts_enabled,
     .guest_x86_mode       = svm_guest_x86_mode,
     .get_guest_ctrl_reg   = svm_get_ctrl_reg,
@@ -967,9 +971,7 @@ int start_svm(void)
          ((root_vmcb[cpu] = alloc_vmcb()) == NULL) )
         return 0;
 
-    rdmsr(MSR_EFER, eax, edx);
-    eax |= EFER_SVME;
-    wrmsr(MSR_EFER, eax, edx);
+    write_efer(read_efer() | EFER_SVME);
 
     svm_npt_detect();
 

diff --git a/xen/arch/x86/hvm/vmx/vmcs.c b/xen/arch/x86/hvm/vmx/vmcs.c
index 93c2842e06faee612f4f70f0da9bde41faa37d8e..2f4c906e30e1262cee59866aa891aa8439145408 100644 (file)
--- a/xen/arch/x86/hvm/vmx/vmcs.c
+++ b/xen/arch/x86/hvm/vmx/vmcs.c
@@ -285,11 +285,6 @@ static void construct_vmcs(struct vcpu *v)
 
     vmx_vmcs_enter(v);
 
-    v->arch.hvm_vmx.cpu_cr2 = 0;
-    v->arch.hvm_vmx.cpu_cr3 = 0;
-    memset(&v->arch.hvm_vmx.msr_state, 0, sizeof(v->arch.hvm_vmx.msr_state));
-    v->arch.hvm_vmx.vmxassist_enabled = 0;
-
     /* VMCS controls. */
     __vmwrite(PIN_BASED_VM_EXEC_CONTROL, vmx_pin_based_exec_control);
     __vmwrite(VM_EXIT_CONTROLS, vmx_vmexit_control);

diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
index e245cc79a5b58dc591567183522328447f4e950d..a0d64aa41f724dc836d352a5979168573501fd51 100644 (file)
--- a/xen/arch/x86/hvm/vmx/vmx.c
+++ b/xen/arch/x86/hvm/vmx/vmx.c
@@ -89,7 +89,7 @@ static DEFINE_PER_CPU(struct vmx_msr_state, host_msr_state);
 static u32 msr_index[VMX_MSR_COUNT] =
 {
     MSR_LSTAR, MSR_STAR, MSR_CSTAR,
-    MSR_SYSCALL_MASK, MSR_EFER,
+    MSR_SYSCALL_MASK
 };
 
 static void vmx_save_host_msrs(void)
@@ -116,8 +116,7 @@ static inline int long_mode_do_msr_read(struct cpu_user_regs *regs)
 
     switch ( (u32)regs->ecx ) {
     case MSR_EFER:
-        HVM_DBG_LOG(DBG_LEVEL_2, "EFER msr_content 0x%"PRIx64, msr_content);
-        msr_content = guest_msr_state->msrs[VMX_INDEX_MSR_EFER];
+        msr_content = v->arch.hvm_vmx.efer;
         break;
 
     case MSR_FS_BASE:
@@ -129,7 +128,7 @@ static inline int long_mode_do_msr_read(struct cpu_user_regs *regs)
         goto check_long_mode;
 
     case MSR_SHADOW_GS_BASE:
-        msr_content = guest_msr_state->shadow_gs;
+        msr_content = v->arch.hvm_vmx.shadow_gs;
     check_long_mode:
         if ( !(vmx_long_mode_enabled(v)) )
         {
@@ -181,7 +180,9 @@ static inline int long_mode_do_msr_write(struct cpu_user_regs *regs)
     {
     case MSR_EFER:
         /* offending reserved bit will cause #GP */
-        if ( msr_content & ~(EFER_LME | EFER_LMA | EFER_NX | EFER_SCE) )
+        if ( (msr_content & ~(EFER_LME | EFER_LMA | EFER_NX | EFER_SCE)) ||
+             (!cpu_has_nx && (msr_content & EFER_NX)) ||
+             (!cpu_has_syscall && (msr_content & EFER_SCE)) )
         {
             gdprintk(XENLOG_WARNING, "Trying to set reserved bit in "
                      "EFER: %"PRIx64"\n", msr_content);
@@ -189,7 +190,7 @@ static inline int long_mode_do_msr_write(struct cpu_user_regs *regs)
         }
 
         if ( (msr_content & EFER_LME)
-             &&  !(guest_msr_state->msrs[VMX_INDEX_MSR_EFER] & EFER_LME) )
+             &&  !(v->arch.hvm_vmx.efer & EFER_LME) )
         {
             if ( unlikely(vmx_paging_enabled(v)) )
             {
@@ -199,7 +200,7 @@ static inline int long_mode_do_msr_write(struct cpu_user_regs *regs)
             }
         }
         else if ( !(msr_content & EFER_LME)
-                  && (guest_msr_state->msrs[VMX_INDEX_MSR_EFER] & EFER_LME) )
+                  && (v->arch.hvm_vmx.efer & EFER_LME) )
         {
             if ( unlikely(vmx_paging_enabled(v)) )
             {
@@ -209,7 +210,11 @@ static inline int long_mode_do_msr_write(struct cpu_user_regs *regs)
             }
         }
 
-        guest_msr_state->msrs[VMX_INDEX_MSR_EFER] = msr_content;
+        if ( (msr_content ^ v->arch.hvm_vmx.efer) & (EFER_NX|EFER_SCE) )
+            write_efer((read_efer() & ~(EFER_NX|EFER_SCE)) |
+                       (msr_content & (EFER_NX|EFER_SCE)));
+
+        v->arch.hvm_vmx.efer = msr_content;
         break;
 
     case MSR_FS_BASE:
@@ -227,7 +232,7 @@ static inline int long_mode_do_msr_write(struct cpu_user_regs *regs)
             __vmwrite(GUEST_GS_BASE, msr_content);
         else
         {
-            v->arch.hvm_vmx.msr_state.shadow_gs = msr_content;
+            v->arch.hvm_vmx.shadow_gs = msr_content;
             wrmsrl(MSR_SHADOW_GS_BASE, msr_content);
         }
 
@@ -279,12 +284,14 @@ static void vmx_restore_host_msrs(void)
         wrmsrl(msr_index[i], host_msr_state->msrs[i]);
         clear_bit(i, &host_msr_state->flags);
     }
+    if ( cpu_has_nx && !(read_efer() & EFER_NX) )
+        write_efer(read_efer() | EFER_NX);
 }
 
 static void vmx_save_guest_msrs(struct vcpu *v)
 {
     /* MSR_SHADOW_GS_BASE may have been changed by swapgs instruction. */
-    rdmsrl(MSR_SHADOW_GS_BASE, v->arch.hvm_vmx.msr_state.shadow_gs);
+    rdmsrl(MSR_SHADOW_GS_BASE, v->arch.hvm_vmx.shadow_gs);
 }
 
 static void vmx_restore_guest_msrs(struct vcpu *v)
@@ -296,11 +303,9 @@ static void vmx_restore_guest_msrs(struct vcpu *v)
     guest_msr_state = &v->arch.hvm_vmx.msr_state;
     host_msr_state = &this_cpu(host_msr_state);
 
-    wrmsrl(MSR_SHADOW_GS_BASE, guest_msr_state->shadow_gs);
+    wrmsrl(MSR_SHADOW_GS_BASE, v->arch.hvm_vmx.shadow_gs);
 
     guest_flags = guest_msr_state->flags;
-    if ( !guest_flags )
-        return;
 
     while ( guest_flags ) {
         i = find_first_set_bit(guest_flags);
@@ -312,23 +317,90 @@ static void vmx_restore_guest_msrs(struct vcpu *v)
         wrmsrl(msr_index[i], guest_msr_state->msrs[i]);
         clear_bit(i, &guest_flags);
     }
+
+    if ( (v->arch.hvm_vmx.efer ^ read_efer()) & (EFER_NX|EFER_SCE) )
+    {
+        HVM_DBG_LOG(DBG_LEVEL_2,
+                    "restore guest's EFER with value %lx",
+                    v->arch.hvm_vmx.efer);
+        write_efer((read_efer() & ~(EFER_NX|EFER_SCE)) |
+                   (v->arch.hvm_vmx.efer & (EFER_NX|EFER_SCE)));
+    }
 }
 
 #else  /* __i386__ */
 
 #define vmx_save_host_msrs()        ((void)0)
-#define vmx_restore_host_msrs()     ((void)0)
+
+static void vmx_restore_host_msrs(void)
+{
+    if ( cpu_has_nx && !(read_efer() & EFER_NX) )
+        write_efer(read_efer() | EFER_NX);
+}
+
 #define vmx_save_guest_msrs(v)      ((void)0)
-#define vmx_restore_guest_msrs(v)   ((void)0)
+
+static void vmx_restore_guest_msrs(struct vcpu *v)
+{
+    if ( (v->arch.hvm_vmx.efer ^ read_efer()) & EFER_NX )
+    {
+        HVM_DBG_LOG(DBG_LEVEL_2,
+                    "restore guest's EFER with value %lx",
+                    v->arch.hvm_vmx.efer);
+        write_efer((read_efer() & ~EFER_NX) |
+                   (v->arch.hvm_vmx.efer & EFER_NX));
+    }
+}
 
 static inline int long_mode_do_msr_read(struct cpu_user_regs *regs)
 {
-    return 0;
+    u64 msr_content = 0;
+    struct vcpu *v = current;
+
+    switch ( regs->ecx ) {
+    case MSR_EFER:
+        msr_content = v->arch.hvm_vmx.efer;
+        break;
+
+    default:
+        return 0;
+    }
+
+    regs->eax = msr_content >>  0;
+    regs->edx = msr_content >> 32;
+
+    return 1;
 }
 
 static inline int long_mode_do_msr_write(struct cpu_user_regs *regs)
 {
-    return 0;
+    u64 msr_content = regs->eax | ((u64)regs->edx << 32);
+    struct vcpu *v = current;
+
+    switch ( regs->ecx )
+    {
+    case MSR_EFER:
+        /* offending reserved bit will cause #GP */
+        if ( (msr_content & ~EFER_NX) ||
+             (!cpu_has_nx && (msr_content & EFER_NX)) )
+        {
+            gdprintk(XENLOG_WARNING, "Trying to set reserved bit in "
+                     "EFER: %"PRIx64"\n", msr_content);
+            vmx_inject_hw_exception(v, TRAP_gp_fault, 0);
+            return 0;
+        }
+
+        if ( (msr_content ^ v->arch.hvm_vmx.efer) & EFER_NX )
+            write_efer((read_efer() & ~EFER_NX) | (msr_content & EFER_NX));
+
+        v->arch.hvm_vmx.efer = msr_content;
+        break;
+
+    default:
+        return 0;
+    }
+
+    return 1;
 }
 
 #endif /* __i386__ */
@@ -636,7 +708,7 @@ int vmx_vmcs_restore(struct vcpu *v, struct hvm_hw_cpu *c)
     return -EINVAL;
 }
 
-#ifdef HVM_DEBUG_SUSPEND
+#if defined(__x86_64__) && defined(HVM_DEBUG_SUSPEND)
 static void dump_msr_state(struct vmx_msr_state *m)
 {
     int i = 0;
@@ -647,17 +719,16 @@ static void dump_msr_state(struct vmx_msr_state *m)
     printk("\n");
 }
 #else
-static void dump_msr_state(struct vmx_msr_state *m)
-{
-}
+#define dump_msr_state(m) ((void)0)
 #endif
         
-void vmx_save_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
+static void vmx_save_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
 {
+#ifdef __x86_64__
     struct vmx_msr_state *guest_state = &v->arch.hvm_vmx.msr_state;
     unsigned long guest_flags = guest_state->flags;
 
-    data->shadow_gs = guest_state->shadow_gs;
+    data->shadow_gs = v->arch.hvm_vmx.shadow_gs;
 
     /* save msrs */
     data->msr_flags        = guest_flags;
@@ -665,15 +736,18 @@ void vmx_save_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
     data->msr_star         = guest_state->msrs[VMX_INDEX_MSR_STAR];
     data->msr_cstar        = guest_state->msrs[VMX_INDEX_MSR_CSTAR];
     data->msr_syscall_mask = guest_state->msrs[VMX_INDEX_MSR_SYSCALL_MASK];
-    data->msr_efer         = guest_state->msrs[VMX_INDEX_MSR_EFER];
+#endif
+
+    data->msr_efer = v->arch.hvm_vmx.efer;
 
     data->tsc = hvm_get_guest_time(v);
     
     dump_msr_state(guest_state);
 }
 
-void vmx_load_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
+static void vmx_load_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
 {
+#ifdef __x86_64__
     struct vmx_msr_state *guest_state = &v->arch.hvm_vmx.msr_state;
 
     /* restore msrs */
@@ -682,9 +756,11 @@ void vmx_load_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
     guest_state->msrs[VMX_INDEX_MSR_STAR]         = data->msr_star;
     guest_state->msrs[VMX_INDEX_MSR_CSTAR]        = data->msr_cstar;
     guest_state->msrs[VMX_INDEX_MSR_SYSCALL_MASK] = data->msr_syscall_mask;
-    guest_state->msrs[VMX_INDEX_MSR_EFER]         = data->msr_efer;
 
-    guest_state->shadow_gs = data->shadow_gs;
+    v->arch.hvm_vmx.shadow_gs = data->shadow_gs;
+#endif
+
+    v->arch.hvm_vmx.efer = data->msr_efer;
 
     v->arch.hvm_vmx.vmxassist_enabled = !(data->cr0 & X86_CR0_PE);
 
@@ -694,7 +770,7 @@ void vmx_load_cpu_state(struct vcpu *v, struct hvm_hw_cpu *data)
 }
 
 
-void vmx_save_vmcs_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
+static void vmx_save_vmcs_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
 {
     vmx_save_cpu_state(v, ctxt);
     vmx_vmcs_enter(v);
@@ -702,7 +778,7 @@ void vmx_save_vmcs_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
     vmx_vmcs_exit(v);
 }
 
-int vmx_load_vmcs_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
+static int vmx_load_vmcs_ctxt(struct vcpu *v, struct hvm_hw_cpu *ctxt)
 {
     vmx_load_cpu_state(v, ctxt);
     if (vmx_vmcs_restore(v, ctxt)) {
@@ -1016,6 +1092,11 @@ static int vmx_pae_enabled(struct vcpu *v)
     return (vmx_paging_enabled(v) && (cr4 & X86_CR4_PAE));
 }
 
+static int vmx_nx_enabled(struct vcpu *v)
+{
+    return v->arch.hvm_vmx.efer & EFER_NX;
+}
+
 static int vmx_interrupts_enabled(struct vcpu *v) 
 {
     unsigned long eflags = __vmread(GUEST_RFLAGS); 
@@ -1096,6 +1177,7 @@ static struct hvm_function_table vmx_function_table = {
     .paging_enabled       = vmx_paging_enabled,
     .long_mode_enabled    = vmx_long_mode_enabled,
     .pae_enabled          = vmx_pae_enabled,
+    .nx_enabled           = vmx_nx_enabled,
     .interrupts_enabled   = vmx_interrupts_enabled,
     .guest_x86_mode       = vmx_guest_x86_mode,
     .get_guest_ctrl_reg   = vmx_get_ctrl_reg,
@@ -1996,8 +2078,7 @@ static int vmx_set_cr0(unsigned long value)
             else
             {
                 HVM_DBG_LOG(DBG_LEVEL_1, "Enabling long mode\n");
-                v->arch.hvm_vmx.msr_state.msrs[VMX_INDEX_MSR_EFER]
-                    |= EFER_LMA;
+                v->arch.hvm_vmx.efer |= EFER_LMA;
                 vm_entry_value = __vmread(VM_ENTRY_CONTROLS);
                 vm_entry_value |= VM_ENTRY_IA32E_MODE;
                 __vmwrite(VM_ENTRY_CONTROLS, vm_entry_value);
@@ -2046,8 +2127,7 @@ static int vmx_set_cr0(unsigned long value)
              */
             if ( vmx_long_mode_enabled(v) )
             {
-                v->arch.hvm_vmx.msr_state.msrs[VMX_INDEX_MSR_EFER]
-                    &= ~EFER_LMA;
+                v->arch.hvm_vmx.efer &= ~EFER_LMA;
                 vm_entry_value = __vmread(VM_ENTRY_CONTROLS);
                 vm_entry_value &= ~VM_ENTRY_IA32E_MODE;
                 __vmwrite(VM_ENTRY_CONTROLS, vm_entry_value);
@@ -2079,7 +2159,7 @@ static int vmx_set_cr0(unsigned long value)
     {
         if ( vmx_long_mode_enabled(v) )
         {
-            v->arch.hvm_vmx.msr_state.msrs[VMX_INDEX_MSR_EFER] &= ~EFER_LMA;
+            v->arch.hvm_vmx.efer &= ~EFER_LMA;
             vm_entry_value = __vmread(VM_ENTRY_CONTROLS);
             vm_entry_value &= ~VM_ENTRY_IA32E_MODE;
             __vmwrite(VM_ENTRY_CONTROLS, vm_entry_value);

diff --git a/xen/arch/x86/mm/shadow/multi.c b/xen/arch/x86/mm/shadow/multi.c
index 82f3756fcad6a359e6acce5d27a51774ceb5a6c9..f55382ac292bb88e4628074078127ce7680513ec 100644 (file)
--- a/xen/arch/x86/mm/shadow/multi.c
+++ b/xen/arch/x86/mm/shadow/multi.c
@@ -181,11 +181,11 @@ guest_supports_superpages(struct vcpu *v)
 static inline int
 guest_supports_nx(struct vcpu *v)
 {
+    if ( GUEST_PAGING_LEVELS == 2 || !cpu_has_nx )
+        return 0;
     if ( !is_hvm_vcpu(v) )
-        return cpu_has_nx;
-
-    // XXX - fix this!
-    return 1;
+        return 1;
+    return hvm_nx_enabled(v);
 }
 
 

diff --git a/xen/include/asm-x86/cpufeature.h b/xen/include/asm-x86/cpufeature.h
index 754c6dbf5adaf7b426a651cf257af7d62bf55e18..22a72f4bd8dbc868321245482a9504e7fd7654cf 100644 (file)
--- a/xen/include/asm-x86/cpufeature.h
+++ b/xen/include/asm-x86/cpufeature.h
@@ -114,6 +114,7 @@
 #define cpu_has_xmm2           boot_cpu_has(X86_FEATURE_XMM2)
 #define cpu_has_xmm3           boot_cpu_has(X86_FEATURE_XMM3)
 #define cpu_has_ht             boot_cpu_has(X86_FEATURE_HT)
+#define cpu_has_syscall                boot_cpu_has(X86_FEATURE_SYSCALL)
 #define cpu_has_mp             boot_cpu_has(X86_FEATURE_MP)
 #define cpu_has_nx             boot_cpu_has(X86_FEATURE_NX)
 #define cpu_has_k6_mtrr                boot_cpu_has(X86_FEATURE_K6_MTRR)
@@ -136,6 +137,7 @@
 #define cpu_has_xmm2           1
 #define cpu_has_xmm3           boot_cpu_has(X86_FEATURE_XMM3)
 #define cpu_has_ht             boot_cpu_has(X86_FEATURE_HT)
+#define cpu_has_syscall                1
 #define cpu_has_mp             1
 #define cpu_has_nx             boot_cpu_has(X86_FEATURE_NX)
 #define cpu_has_k6_mtrr                0

diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index 2a3802423f57fc352fa73dd9367ae3e625138a1f..a89f242cbe69b7e5feab48e82648ae905f83e991 100644 (file)
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -93,14 +93,17 @@ struct hvm_function_table {
      * 1) determine whether paging is enabled,
      * 2) determine whether long mode is enabled,
      * 3) determine whether PAE paging is enabled,
-     * 4) determine whether interrupts are enabled or not,
-     * 5) determine the mode the guest is running in,
-     * 6) return the current guest control-register value
-     * 7) return the current guest segment descriptor base
+     * 4) determine whether NX is enabled,
+     * 5) determine whether interrupts are enabled or not,
+     * 6) determine the mode the guest is running in,
+     * 7) return the current guest control-register value
+     * 8) return the current guest segment descriptor base
+     * 9) return the current guest segment descriptor
      */
     int (*paging_enabled)(struct vcpu *v);
     int (*long_mode_enabled)(struct vcpu *v);
     int (*pae_enabled)(struct vcpu *v);
+    int (*nx_enabled)(struct vcpu *v);
     int (*interrupts_enabled)(struct vcpu *v);
     int (*guest_x86_mode)(struct vcpu *v);
     unsigned long (*get_guest_ctrl_reg)(struct vcpu *v, unsigned int num);
@@ -198,6 +201,12 @@ hvm_interrupts_enabled(struct vcpu *v)
     return hvm_funcs.interrupts_enabled(v);
 }
 
+static inline int
+hvm_nx_enabled(struct vcpu *v)
+{
+    return hvm_funcs.nx_enabled(v);
+}
+
 static inline int
 hvm_guest_x86_mode(struct vcpu *v)
 {

diff --git a/xen/include/asm-x86/hvm/svm/svm.h b/xen/include/asm-x86/hvm/svm/svm.h
index 8fb50fb0264f3c1611d51630f43a565b86a75b7e..341ccf0963bc31977ecd0be5b73deb1a1f2244a1 100644 (file)
--- a/xen/include/asm-x86/hvm/svm/svm.h
+++ b/xen/include/asm-x86/hvm/svm/svm.h
@@ -60,6 +60,11 @@ static inline int svm_pae_enabled(struct vcpu *v)
     return svm_paging_enabled(v) && (guest_cr4 & X86_CR4_PAE);
 }
 
+static inline int svm_nx_enabled(struct vcpu *v)
+{
+    return v->arch.hvm_svm.cpu_shadow_efer & EFER_NX;
+}
+
 static inline int svm_pgbit_test(struct vcpu *v)
 {
     return v->arch.hvm_svm.cpu_shadow_cr0 & X86_CR0_PG;

diff --git a/xen/include/asm-x86/hvm/vmx/vmcs.h b/xen/include/asm-x86/hvm/vmx/vmcs.h
index 1947c6686f557a31449ca616b2a0f544bbfb71a2..839f450d3984393d536e5b5be3fe812f3fe1ef8e 100644 (file)
--- a/xen/include/asm-x86/hvm/vmx/vmcs.h
+++ b/xen/include/asm-x86/hvm/vmx/vmcs.h
@@ -39,7 +39,6 @@ enum {
     VMX_INDEX_MSR_STAR,
     VMX_INDEX_MSR_CSTAR,
     VMX_INDEX_MSR_SYSCALL_MASK,
-    VMX_INDEX_MSR_EFER,
 
     VMX_MSR_COUNT
 };
@@ -47,7 +46,6 @@ enum {
 struct vmx_msr_state {
     unsigned long flags;
     unsigned long msrs[VMX_MSR_COUNT];
-    unsigned long shadow_gs;
 };
 
 struct arch_vmx_struct {
@@ -76,7 +74,11 @@ struct arch_vmx_struct {
     unsigned long        cpu_shadow_cr4; /* copy of guest read shadow CR4 */
     unsigned long        cpu_cr2; /* save CR2 */
     unsigned long        cpu_cr3;
+#ifdef __x86_64__
     struct vmx_msr_state msr_state;
+    unsigned long        shadow_gs;
+#endif
+    unsigned long        efer;
     unsigned long        vmxassist_enabled:1;
 };
 

diff --git a/xen/include/asm-x86/hvm/vmx/vmx.h b/xen/include/asm-x86/hvm/vmx/vmx.h
index 38ad5e25bd0c5c3f751990941cf83ba8c9193905..e2a5b59cea7a8cd9ec97e45f4e60adc190f460e9 100644 (file)
--- a/xen/include/asm-x86/hvm/vmx/vmx.h
+++ b/xen/include/asm-x86/hvm/vmx/vmx.h
@@ -261,14 +261,12 @@ static inline int vmx_paging_enabled(struct vcpu *v)
 
 static inline int vmx_long_mode_enabled(struct vcpu *v)
 {
-    u64 efer = v->arch.hvm_vmx.msr_state.msrs[VMX_INDEX_MSR_EFER];
-    return efer & EFER_LMA;
+    return v->arch.hvm_vmx.efer & EFER_LMA;
 }
 
 static inline int vmx_lme_is_set(struct vcpu *v)
 {
-    u64 efer = v->arch.hvm_vmx.msr_state.msrs[VMX_INDEX_MSR_EFER];
-    return efer & EFER_LME;
+    return v->arch.hvm_vmx.efer & EFER_LME;
 }
 
 static inline int vmx_pgbit_test(struct vcpu *v)

diff --git a/xen/include/asm-x86/msr.h b/xen/include/asm-x86/msr.h
index bd7f27b6c52388d3c0963222d2f17e9b445c8c94..d1ea050b6826386376b4b22aa170b76fc59c50c9 100644 (file)
--- a/xen/include/asm-x86/msr.h
+++ b/xen/include/asm-x86/msr.h
@@ -3,6 +3,9 @@
 
 #ifndef __ASSEMBLY__
 
+#include <xen/smp.h>
+#include <xen/percpu.h>
+
 #define rdmsr(msr,val1,val2) \
      __asm__ __volatile__("rdmsr" \
                          : "=a" (val1), "=d" (val2) \
@@ -142,6 +145,25 @@ static inline void wrmsrl(unsigned int msr, __u64 val)
 #define EFER_NX (1<<_EFER_NX)
 #define EFER_SVME (1<<_EFER_SVME)
 
+#ifndef __ASSEMBLY__
+
+DECLARE_PER_CPU(__u64, efer);
+
+static inline __u64 read_efer(void)
+{
+    if (!this_cpu(efer))
+        rdmsrl(MSR_EFER, this_cpu(efer));
+    return this_cpu(efer);
+}
+
+static inline void write_efer(__u64 val)
+{
+    this_cpu(efer) = val;
+    wrmsrl(MSR_EFER, val);
+}
+
+#endif
+
 /* Intel MSRs. Some also available on other CPUs */
 #define MSR_IA32_PLATFORM_ID   0x17