x86/vmx: Support remote access to the MSR lists

At the moment, all modifications of the MSR lists are in current context.
However, future changes may need to put MSR_EFER into the lists from domctl
hypercall context.

Plumb a struct vcpu parameter down through the infrastructure, and use
vmx_vmcs_{enter,exit}() for safe access to the VMCS in vmx_add_msr().  Use
assertions to ensure that access is either in current context, or while the
vcpu is paused.

Note these expectations beside the fields in arch_vmx_struct, and reorder the
fields to avoid unnecessary padding.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
(cherry picked from commit 80599f0b77)

xen/arch/x86/cpu/vpmu_intel.c

@@ -483,12 +483,12 @@ static int core2_vpmu_alloc_resource(struct vcpu *v)
     if ( has_hvm_container_vcpu(v) )
     {
         wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0);
-        if ( vmx_add_host_load_msr(MSR_CORE_PERF_GLOBAL_CTRL) )
+        if ( vmx_add_host_load_msr(v, MSR_CORE_PERF_GLOBAL_CTRL) )
             goto out_err;
 
-        if ( vmx_add_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL) )
+        if ( vmx_add_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL) )
             goto out_err;
-        vmx_write_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL, 0);
+        vmx_write_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL, 0);
     }
 
     core2_vpmu_cxt = xzalloc_bytes(sizeof(*core2_vpmu_cxt) +
@@ -642,7 +642,7 @@ static int core2_vpmu_do_wrmsr(unsigned int msr, uint64_t msr_content,
             return -EINVAL;
 
         if ( has_hvm_container_vcpu(v) )
-            vmx_read_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL,
+            vmx_read_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL,
                                &core2_vpmu_cxt->global_ctrl);
         else
             rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, core2_vpmu_cxt->global_ctrl);
@@ -711,7 +711,7 @@ static int core2_vpmu_do_wrmsr(unsigned int msr, uint64_t msr_content,
                 return -EINVAL;
 
             if ( has_hvm_container_vcpu(v) )
-                vmx_read_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL,
+                vmx_read_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL,
                                    &core2_vpmu_cxt->global_ctrl);
             else
                 rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, core2_vpmu_cxt->global_ctrl);
@@ -730,7 +730,7 @@ static int core2_vpmu_do_wrmsr(unsigned int msr, uint64_t msr_content,
     else
     {
         if ( has_hvm_container_vcpu(v) )
-            vmx_write_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
+            vmx_write_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
         else
             wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
     }
@@ -764,7 +764,7 @@ static int core2_vpmu_do_rdmsr(unsigned int msr, uint64_t *msr_content)
             break;
         case MSR_CORE_PERF_GLOBAL_CTRL:
             if ( has_hvm_container_vcpu(v) )
-                vmx_read_guest_msr(MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
+                vmx_read_guest_msr(v, MSR_CORE_PERF_GLOBAL_CTRL, msr_content);
             else
                 rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, *msr_content);
             break;

xen/arch/x86/hvm/vmx/vmcs.c

@@ -1357,13 +1357,15 @@ static struct vmx_msr_entry *locate_msr_entry(
     return start;
 }
 
-struct vmx_msr_entry *vmx_find_msr(uint32_t msr, enum vmx_msr_list_type type)
+struct vmx_msr_entry *vmx_find_msr(const struct vcpu *v, uint32_t msr,
+                                   enum vmx_msr_list_type type)
 {
-    struct vcpu *curr = current;
-    struct arch_vmx_struct *vmx = &curr->arch.hvm_vmx;
+    const struct arch_vmx_struct *vmx = &v->arch.hvm_vmx;
     struct vmx_msr_entry *start = NULL, *ent, *end;
     unsigned int total;
 
+    ASSERT(v == current || !vcpu_runnable(v));
+
     switch ( type )
     {
     case VMX_MSR_HOST:
@@ -1389,12 +1391,14 @@ struct vmx_msr_entry *vmx_find_msr(uint32_t msr, enum vmx_msr_list_type type)
     return ((ent < end) && (ent->index == msr)) ? ent : NULL;
 }
 
-int vmx_add_msr(uint32_t msr, enum vmx_msr_list_type type)
+int vmx_add_msr(struct vcpu *v, uint32_t msr, enum vmx_msr_list_type type)
 {
-    struct vcpu *curr = current;
-    struct arch_vmx_struct *vmx = &curr->arch.hvm_vmx;
+    struct arch_vmx_struct *vmx = &v->arch.hvm_vmx;
     struct vmx_msr_entry **ptr, *start = NULL, *ent, *end;
     unsigned int total;
+    int rc;
+
+    ASSERT(v == current || !vcpu_runnable(v));
 
     switch ( type )
     {
@@ -1413,13 +1417,18 @@ int vmx_add_msr(uint32_t msr, enum vmx_msr_list_type type)
         return -EINVAL;
     }
 
+    vmx_vmcs_enter(v);
+
     /* Allocate memory on first use. */
     if ( unlikely(!*ptr) )
     {
         paddr_t addr;
 
         if ( (*ptr = alloc_xenheap_page()) == NULL )
-            return -ENOMEM;
+        {
+            rc = -ENOMEM;
+            goto out;
+        }
 
         addr = virt_to_maddr(*ptr);
 
@@ -1441,10 +1450,16 @@ int vmx_add_msr(uint32_t msr, enum vmx_msr_list_type type)
     ent   = locate_msr_entry(start, end, msr);
 
     if ( (ent < end) && (ent->index == msr) )
-        return 0;
+    {
+        rc = 0;
+        goto out;
+    }
 
     if ( total == (PAGE_SIZE / sizeof(*ent)) )
-        return -ENOSPC;
+    {
+        rc = -ENOSPC;
+        goto out;
+    }
 
     memmove(ent + 1, ent, sizeof(*ent) * (end - ent));
 
@@ -1465,7 +1480,12 @@ int vmx_add_msr(uint32_t msr, enum vmx_msr_list_type type)
         break;
     }
 
-    return 0;
+    rc = 0;
+
+ out:
+    vmx_vmcs_exit(v);
+
+    return rc;
 }
 
 void vmx_set_eoi_exit_bitmap(struct vcpu *v, u8 vector)

xen/arch/x86/hvm/vmx/vmx.c

@@ -2742,6 +2742,8 @@ static int is_last_branch_msr(u32 ecx)
 
 static int vmx_msr_read_intercept(unsigned int msr, uint64_t *msr_content)
 {
+    struct vcpu *curr = current;
+
     HVM_DBG_LOG(DBG_LEVEL_MSR, "ecx=%#x", msr);
 
     switch ( msr )
@@ -2804,7 +2806,7 @@ static int vmx_msr_read_intercept(unsigned int msr, uint64_t *msr_content)
                 goto done;
         }
 
-        if ( vmx_read_guest_msr(msr, msr_content) == 0 )
+        if ( vmx_read_guest_msr(curr, msr, msr_content) == 0 )
             break;
 
         if ( is_last_branch_msr(msr) )
@@ -2984,12 +2986,12 @@ static int vmx_msr_write_intercept(unsigned int msr, uint64_t msr_content)
 
             for ( ; (rc == 0) && lbr->count; lbr++ )
                 for ( i = 0; (rc == 0) && (i < lbr->count); i++ )
-                    if ( (rc = vmx_add_guest_msr(lbr->base + i)) == 0 )
+                    if ( (rc = vmx_add_guest_msr(v, lbr->base + i)) == 0 )
                         vmx_disable_intercept_for_msr(v, lbr->base + i, MSR_TYPE_R | MSR_TYPE_W);
         }
 
         if ( (rc < 0) ||
-             (msr_content && (vmx_add_host_load_msr(msr) < 0)) )
+             (msr_content && (vmx_add_host_load_msr(v, msr) < 0)) )
             hvm_inject_hw_exception(TRAP_machine_check, 0);
         else
             __vmwrite(GUEST_IA32_DEBUGCTL, msr_content);
@@ -3026,7 +3028,7 @@ static int vmx_msr_write_intercept(unsigned int msr, uint64_t msr_content)
         switch ( long_mode_do_msr_write(msr, msr_content) )
         {
             case HNDL_unhandled:
-                if ( (vmx_write_guest_msr(msr, msr_content) != 0) &&
+                if ( (vmx_write_guest_msr(v, msr, msr_content) != 0) &&
                      !is_last_branch_msr(msr) )
                     switch ( wrmsr_hypervisor_regs(msr, msr_content) )
                     {

xen/include/asm-x86/atomic.h

@@ -94,7 +94,7 @@ typedef struct { int counter; } atomic_t;
  *
  * Atomically reads the value of @v.
  */
-static inline int atomic_read(atomic_t *v)
+static inline int atomic_read(const atomic_t *v)
 {
     return read_atomic(&v->counter);
 }

xen/include/asm-x86/hvm/vmx/vmcs.h

@@ -193,10 +193,17 @@ struct arch_vmx_struct {
     unsigned long        cstar;
 
     unsigned long       *msr_bitmap;
-    unsigned int         msr_count;
+
+    /*
+     * Most accesses to the MSR host/guest load/save lists are in current
+     * context.  However, the data can be modified by toolstack/migration
+     * actions.  Remote access is only permitted for paused vcpus, and is
+     * protected under the domctl lock.
+     */
     struct vmx_msr_entry *msr_area;
-    unsigned int         host_msr_count;
     struct vmx_msr_entry *host_msr_area;
+    unsigned int         msr_count;
+    unsigned int         host_msr_count;
 
     unsigned long        eoi_exitmap_changed;
     DECLARE_BITMAP(eoi_exit_bitmap, NR_VECTORS);
@@ -585,23 +592,25 @@ enum vmx_msr_list_type {
     VMX_MSR_GUEST,          /* MSRs saved on VMExit, loaded on VMEntry. */
 };
 
-int vmx_add_msr(uint32_t msr, enum vmx_msr_list_type type);
+int vmx_add_msr(struct vcpu *v, uint32_t msr, enum vmx_msr_list_type type);
 
-static inline int vmx_add_host_load_msr(uint32_t msr)
+static inline int vmx_add_guest_msr(struct vcpu *v, uint32_t msr)
 {
-    return vmx_add_msr(msr, VMX_MSR_HOST);
+    return vmx_add_msr(v, msr, VMX_MSR_GUEST);
 }
 
-static inline int vmx_add_guest_msr(uint32_t msr)
+static inline int vmx_add_host_load_msr(struct vcpu *v, uint32_t msr)
 {
-    return vmx_add_msr(msr, VMX_MSR_GUEST);
+    return vmx_add_msr(v, msr, VMX_MSR_HOST);
 }
 
-struct vmx_msr_entry *vmx_find_msr(uint32_t msr, enum vmx_msr_list_type type);
+struct vmx_msr_entry *vmx_find_msr(const struct vcpu *v, uint32_t msr,
+                                   enum vmx_msr_list_type type);
 
-static inline int vmx_read_guest_msr(uint32_t msr, uint64_t *val)
+static inline int vmx_read_guest_msr(const struct vcpu *v, uint32_t msr,
+                                     uint64_t *val)
 {
-    const struct vmx_msr_entry *ent = vmx_find_msr(msr, VMX_MSR_GUEST);
+    const struct vmx_msr_entry *ent = vmx_find_msr(v, msr, VMX_MSR_GUEST);
 
     if ( !ent )
         return -ESRCH;
@@ -611,9 +620,10 @@ static inline int vmx_read_guest_msr(uint32_t msr, uint64_t *val)
     return 0;
 }
 
-static inline int vmx_write_guest_msr(uint32_t msr, uint64_t val)
+static inline int vmx_write_guest_msr(struct vcpu *v, uint32_t msr,
+                                      uint64_t val)
 {
-    struct vmx_msr_entry *ent = vmx_find_msr(msr, VMX_MSR_GUEST);
+    struct vmx_msr_entry *ent = vmx_find_msr(v, msr, VMX_MSR_GUEST);
 
     if ( !ent )
         return -ESRCH;

xen/include/xen/sched.h

@@ -768,7 +768,7 @@ static inline struct domain *next_domain_in_cpupool(
 #define _VPF_in_reset        7
 #define VPF_in_reset         (1UL<<_VPF_in_reset)
 
-static inline int vcpu_runnable(struct vcpu *v)
+static inline int vcpu_runnable(const struct vcpu *v)
 {
     return !(v->pause_flags |
              atomic_read(&v->pause_count) |