90 lines
2.9 KiB
Diff
90 lines
2.9 KiB
Diff
|
# Commit 7d8b5dd98463524686bdee8b973b53c00c232122
|
||
|
|
# Date 2013-11-25 11:19:04 +0100
|
||
|
|
# Author Liu Jinsong <jinsong.liu@intel.com>
|
||
|
|
# Committer Jan Beulich <jbeulich@suse.com>
|
||
|
|
x86/xsave: fix nonlazy state handling
|
||
|
|
|
||
|
|
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>
|
||
|
|
|
||
|
|
--- 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));
|
||
|
|
}
|
||
|
|
--- a/xen/arch/x86/hvm/hvm.c
|
||
|
|
+++ b/xen/arch/x86/hvm/hvm.c
|
||
|
|
@@ -1073,6 +1073,8 @@ static int hvm_load_cpu_xsave_states(str
|
||
|
|
|
||
|
|
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));
|
||
|
|
|
||
|
|
--- a/xen/arch/x86/i387.c
|
||
|
|
+++ b/xen/arch/x86/i387.c
|
||
|
|
@@ -120,11 +120,22 @@ static inline void fpu_frstor(struct vcp
|
||
|
|
/*******************************/
|
||
|
|
/* 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,
|
||
|
|
@@ -132,7 +143,7 @@ static inline void fpu_xsave(struct vcpu
|
||
|
|
*/
|
||
|
|
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);
|
||
|
|
}
|
||
|
|
@@ -263,7 +274,7 @@ void vcpu_restore_fpu_lazy(struct vcpu *
|
||
|
|
*/
|
||
|
|
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));
|