Xen devel - Dec 2013 - [PATCH v4 1/7] X86: Fix xsave bug for nonlazy xstates

From f6f5b564f3f71ec92026e338321c3e7d19c4f81d Mon Sep 17 00:00:00 2001
From: Liu Jinsong <jinsong.liu@intel.com>
Date: Sun, 24 Nov 2013 23:42:28 +0800
Subject: [PATCH v4 1/7] X86: Fix xsave bug for nonlazy xstates

Nonlazy xstates should be xsaved each time when vcpu_save_fpu.
Operation to nonlazy xstates will not trigger #NM exception, so
whenever vcpu scheduled in it got restored and whenever scheduled
out it should get saved.

Currently this bug affects AMD LWP feature, and later Intel MPX
feature. With the bugfix both LWP and MPX will work fine.

Signed-off-by: Liu Jinsong <jinsong.liu@intel.com>

Furthermore, during restore we also need to set nonlazy_xstate_used
according to the incoming accumulated XCR0.

Also adjust the changes to i387.c such that there won''t be a pointless
clts()/stts() pair.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
---
 xen/arch/x86/domctl.c  |    2 ++
 xen/arch/x86/hvm/hvm.c |    2 ++
 xen/arch/x86/i387.c    |   15 +++++++++++++--
 3 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
index f7e4586..8644218 100644
--- a/xen/arch/x86/domctl.c
+++ b/xen/arch/x86/domctl.c
@@ -1146,6 +1146,8 @@ long arch_do_domctl(
             {
                 v->arch.xcr0 = _xcr0;
                 v->arch.xcr0_accum = _xcr0_accum;
+                if ( _xcr0_accum & XSTATE_NONLAZY )
+                    v->arch.nonlazy_xstate_used = 1;
                 memcpy(v->arch.xsave_area, _xsave_area,
                        evc->size - 2 * sizeof(uint64_t));
             }
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 3b353ec..9c88c73 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -1119,6 +1119,8 @@ static int hvm_load_cpu_xsave_states(struct domain *d,
hvm_domain_context_t *h)
 
     v->arch.xcr0 = ctxt->xcr0;
     v->arch.xcr0_accum = ctxt->xcr0_accum;
+    if ( ctxt->xcr0_accum & XSTATE_NONLAZY )
+        v->arch.nonlazy_xstate_used = 1;
     memcpy(v->arch.xsave_area, &ctxt->save_area,
            desc->length - offsetof(struct hvm_hw_cpu_xsave, save_area));
 
diff --git a/xen/arch/x86/i387.c b/xen/arch/x86/i387.c
index 7649274..bd72138 100644
--- a/xen/arch/x86/i387.c
+++ b/xen/arch/x86/i387.c
@@ -133,11 +133,22 @@ static inline void fpu_frstor(struct vcpu *v)
 /*******************************/
 /*      FPU Save Functions     */
 /*******************************/
+
+static inline uint64_t vcpu_xsave_mask(const struct vcpu *v)
+{
+    if ( v->fpu_dirtied )
+        return v->arch.nonlazy_xstate_used ? XSTATE_ALL : XSTATE_LAZY;
+
+    return v->arch.nonlazy_xstate_used ? XSTATE_NONLAZY : 0;
+}
+
 /* Save x87 extended state */
 static inline void fpu_xsave(struct vcpu *v)
 {
     bool_t ok;
+    uint64_t mask = vcpu_xsave_mask(v);
 
+    ASSERT(mask);
     ASSERT(v->arch.xsave_area);
     /*
      * XCR0 normally represents what guest OS set. In case of Xen itself,
@@ -145,7 +156,7 @@ static inline void fpu_xsave(struct vcpu *v)
      */
     ok = set_xcr0(v->arch.xcr0_accum | XSTATE_FP_SSE);
     ASSERT(ok);
-    xsave(v, v->arch.nonlazy_xstate_used ? XSTATE_ALL : XSTATE_LAZY);
+    xsave(v, mask);
     ok = set_xcr0(v->arch.xcr0 ?: XSTATE_FP_SSE);
     ASSERT(ok);
 }
@@ -257,7 +268,7 @@ void vcpu_restore_fpu_lazy(struct vcpu *v)
  */
 void vcpu_save_fpu(struct vcpu *v)
 {
-    if ( !v->fpu_dirtied )
+    if ( !v->fpu_dirtied && !v->arch.nonlazy_xstate_used )
         return;
 
     ASSERT(!is_idle_vcpu(v));
-- 
1.7.1


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

>>> On 02.12.13 at 09:48, "Liu, Jinsong"
<jinsong.liu@intel.com> wrote:
> From f6f5b564f3f71ec92026e338321c3e7d19c4f81d Mon Sep 17 00:00:00 2001
> From: Liu Jinsong <jinsong.liu@intel.com>
> Date: Sun, 24 Nov 2013 23:42:28 +0800
> Subject: [PATCH v4 1/7] X86: Fix xsave bug for nonlazy xstates
What''s the reason you keep sending this? It got applied almost a
week ago.

Jan

Jan Beulich wrote:
>>>> On 02.12.13 at 09:48, "Liu, Jinsong"
<jinsong.liu@intel.com> wrote:
>> From f6f5b564f3f71ec92026e338321c3e7d19c4f81d Mon Sep 17 00:00:00
>> 2001 From: Liu Jinsong <jinsong.liu@intel.com>
>> Date: Sun, 24 Nov 2013 23:42:28 +0800
>> Subject: [PATCH v4 1/7] X86: Fix xsave bug for nonlazy xstates
> 
> What''s the reason you keep sending this? It got applied almost a
> week ago.
> 
> Jan
Ah, sorry, forget this point when send patches v4.

Thanks,
Jinsong