x86: Xsave support for PV guests.

Signed-off-by: Shan Haitao <haitao.shan@intel.com>
Signed-off-by: Han Weidong <weidong.han@intel.com>

3 files changed, 36 insertions, 15 deletions

diff --git a/xen/include/asm-x86/domain.h b/xen/include/asm-x86/domain.h
index 7549126504..15b050ac0e 100644
--- a/xen/include/asm-x86/domain.h
+++ b/xen/include/asm-x86/domain.h
@@ -400,6 +400,23 @@ struct arch_vcpu
     pagetable_t monitor_table;          /* (MFN) hypervisor PT (for HVM) */
     unsigned long cr3;                  /* (MA) value to install in HW CR3 */
 
+    /*
+     * The save area for Processor Extended States and the bitmask of the
+     * XSAVE/XRSTOR features. They are used by: 1) when a vcpu (which has
+     * dirtied FPU/SSE) is scheduled out we XSAVE the states here; 2) in
+     * #NM handler, we XRSTOR the states we XSAVE-ed;
+     */
+    void *xsave_area;
+    uint64_t xcr0;
+    /* Accumulated eXtended features mask for using XSAVE/XRESTORE by Xen
+     * itself, as we can never know whether guest OS depends on content
+     * preservation whenever guest OS clears one feature flag (for example,
+     * temporarily).
+     * However, processor should not be able to touch eXtended states before
+     * it explicitly enables it via xcr0.
+     */
+    uint64_t xcr0_accum;
+
     /* Current LDT details. */
     unsigned long shadow_ldt_mapcnt;
     spinlock_t shadow_ldt_lock;
@@ -435,7 +452,8 @@ unsigned long pv_guest_cr4_fixup(const struct vcpu *, unsigned long guest_cr4);
 #define pv_guest_cr4_to_real_cr4(v)                         \
     (((v)->arch.guest_context.ctrlreg[4]                    \
       | (mmu_cr4_features & (X86_CR4_PGE | X86_CR4_PSE))    \
-      | ((v)->domain->arch.vtsc ? X86_CR4_TSD : 0))         \
+      | ((v)->domain->arch.vtsc ? X86_CR4_TSD : 0)         \
+      | ((cpu_has_xsave)? X86_CR4_OSXSAVE : 0))              \
       & ~X86_CR4_DE)
 #define real_cr4_to_pv_guest_cr4(c) \
     ((c) & ~(X86_CR4_PGE | X86_CR4_PSE | X86_CR4_TSD | X86_CR4_OSXSAVE))
diff --git a/xen/include/asm-x86/hvm/vcpu.h b/xen/include/asm-x86/hvm/vcpu.h
index 53ef98320f..1d72ecfc60 100644
--- a/xen/include/asm-x86/hvm/vcpu.h
+++ b/xen/include/asm-x86/hvm/vcpu.h
@@ -49,15 +49,6 @@ struct hvm_vcpu {
      */
     unsigned long       hw_cr[5];
 
-    /*
-     * The save area for Processor Extended States and the bitmask of the
-     * XSAVE/XRSTOR features. They are used by: 1) when a vcpu (which has
-     * dirtied FPU/SSE) is scheduled out we XSAVE the states here; 2) in
-     * #NM handler, we XRSTOR the states we XSAVE-ed;
-     */
-    void *xsave_area;
-    uint64_t xcr0;
-
     struct vlapic       vlapic;
     s64                 cache_tsc_offset;
     u64                 guest_time;
diff --git a/xen/include/asm-x86/i387.h b/xen/include/asm-x86/i387.h
index ba365c0b3c..ba773e99ed 100644
--- a/xen/include/asm-x86/i387.h
+++ b/xen/include/asm-x86/i387.h
@@ -49,6 +49,8 @@ struct xsave_struct
 #define REX_PREFIX
 #endif
 
+DECLARE_PER_CPU(uint64_t, xcr0);
+
 static inline void xsetbv(u32 index, u64 xfeatures)
 {
     u32 hi = xfeatures >> 32;
@@ -60,14 +62,20 @@ static inline void xsetbv(u32 index, u64 xfeatures)
 
 static inline void set_xcr0(u64 xfeatures)
 {
+    this_cpu(xcr0) = xfeatures;
     xsetbv(XCR_XFEATURE_ENABLED_MASK, xfeatures);
 }
 
+static inline uint64_t get_xcr0(void)
+{
+    return this_cpu(xcr0);
+}
+
 static inline void xsave(struct vcpu *v)
 {
     struct xsave_struct *ptr;
 
-    ptr =(struct xsave_struct *)v->arch.hvm_vcpu.xsave_area;
+    ptr =(struct xsave_struct *)v->arch.xsave_area;
 
     asm volatile (".byte " REX_PREFIX "0x0f,0xae,0x27"
         :
@@ -79,7 +87,7 @@ static inline void xrstor(struct vcpu *v)
 {
     struct xsave_struct *ptr;
 
-    ptr =(struct xsave_struct *)v->arch.hvm_vcpu.xsave_area;
+    ptr =(struct xsave_struct *)v->arch.xsave_area;
 
     asm volatile (".byte " REX_PREFIX "0x0f,0xae,0x2f"
         :
@@ -108,14 +116,18 @@ static inline void setup_fpu(struct vcpu *v)
     if ( !v->fpu_dirtied )
     {
         v->fpu_dirtied = 1;
-        if ( cpu_has_xsave && is_hvm_vcpu(v) )
+        if ( cpu_has_xsave )
         {
             if ( !v->fpu_initialised )
                 v->fpu_initialised = 1;
 
-            set_xcr0(v->arch.hvm_vcpu.xcr0 | XSTATE_FP_SSE);
+            /* XCR0 normally represents what guest OS set. In case of Xen
+             * itself, we set all supported feature mask before doing
+             * save/restore.
+             */
+            set_xcr0(v->arch.xcr0_accum);
             xrstor(v);
-            set_xcr0(v->arch.hvm_vcpu.xcr0);
+            set_xcr0(v->arch.xcr0);
         }
         else
         {