Patchwork [v5,3/3] KVM: x86: Enable Intel MPX for guest

login
register
mail settings
Submitter Liu, Jinsong
Date Feb. 24, 2014, 10:58 a.m.
Message ID <DE8DF0795D48FD4CA783C40EC82923350150E43C@SHSMSX101.ccr.corp.intel.com>
Download mbox | patch
Permalink /patch/323562/
State New
Headers show

Comments

Liu, Jinsong - Feb. 24, 2014, 10:58 a.m.
From 44c2abca2c2eadc6f2f752b66de4acc8131880c4 Mon Sep 17 00:00:00 2001
From: Liu Jinsong <jinsong.liu@intel.com>
Date: Mon, 24 Feb 2014 18:12:31 +0800
Subject: [PATCH v5 3/3] KVM: x86: Enable Intel MPX for guest

This patch enable Intel MPX feature to guest.

Signed-off-by: Xudong Hao <xudong.hao@intel.com>
Signed-off-by: Liu Jinsong <jinsong.liu@intel.com>
---
 arch/x86/kvm/cpuid.c |    4 +++-
 arch/x86/kvm/x86.c   |    6 ++++++
 arch/x86/kvm/x86.h   |    3 ++-
 3 files changed, 11 insertions(+), 2 deletions(-)
Paolo Bonzini - Feb. 24, 2014, 11:13 a.m.
Il 24/02/2014 11:58, Liu, Jinsong ha scritto:
> @@ -599,6 +599,9 @@ int __kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr)
>  	u64 old_xcr0 = vcpu->arch.xcr0;
>  	u64 valid_bits;
>
> +	if (!kvm_x86_ops->mpx_supported || !kvm_x86_ops->mpx_supported())
> +		xcr0 &= ~(XSTATE_BNDREGS | XSTATE_BNDCSR);
> +
>  	/* Only support XCR_XFEATURE_ENABLED_MASK(xcr0) now  */
>  	if (index != XCR_XFEATURE_ENABLED_MASK)
>  		return 1;

This hunk is incorrect, and I can simply drop it when applying.  If MPX 
is not supported, it should not be in the 0Dh CPUID leaf and thus in 
vcpu->arch.guest_supported_xcr0.

This however relies on userspace passing a "sensible" value of CPUID. 
I'll send a patch to strengthen the computation of guest_supported_xcr0.

Thanks!

Paolo
Liu, Jinsong - Feb. 24, 2014, 3:37 p.m.
Paolo Bonzini wrote:
> Il 24/02/2014 11:58, Liu, Jinsong ha scritto:
>> @@ -599,6 +599,9 @@ int __kvm_set_xcr(struct kvm_vcpu *vcpu, u32
>>  	index, u64 xcr) u64 old_xcr0 = vcpu->arch.xcr0;
>>  	u64 valid_bits;
>> 
>> +	if (!kvm_x86_ops->mpx_supported || !kvm_x86_ops->mpx_supported())
>> +		xcr0 &= ~(XSTATE_BNDREGS | XSTATE_BNDCSR);
>> +
>>  	/* Only support XCR_XFEATURE_ENABLED_MASK(xcr0) now  */
>>  	if (index != XCR_XFEATURE_ENABLED_MASK)
>>  		return 1;
> 
> This hunk is incorrect, and I can simply drop it when applying.  If
> MPX is not supported, it should not be in the 0Dh CPUID leaf and thus
> in vcpu->arch.guest_supported_xcr0.
> 
> This however relies on userspace passing a "sensible" value of CPUID.
> I'll send a patch to strengthen the computation of
> guest_supported_xcr0. 
> 
> Thanks!
> 
> Paolo

So patch v5 would be applied except you will remove the incorrect hunk, and you will send a patch strengthenning guest_supported_xcr0?

Thanks,
Jinsong
Paolo Bonzini - Feb. 24, 2014, 4:40 p.m.
Il 24/02/2014 16:37, Liu, Jinsong ha scritto:
> So patch v5 would be applied except you will remove the incorrect
> hunk, and you will send a patch strengthenning guest_supported_xcr0?

Yes.

Paolo

Patch

diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
index b241325..ddc8a7e 100644
--- a/arch/x86/kvm/cpuid.c
+++ b/arch/x86/kvm/cpuid.c
@@ -256,6 +256,8 @@  static inline int __do_cpuid_ent(struct kvm_cpuid_entry2 *entry, u32 function,
 #endif
 	unsigned f_rdtscp = kvm_x86_ops->rdtscp_supported() ? F(RDTSCP) : 0;
 	unsigned f_invpcid = kvm_x86_ops->invpcid_supported() ? F(INVPCID) : 0;
+	unsigned f_mpx = kvm_x86_ops->mpx_supported ?
+			 (kvm_x86_ops->mpx_supported() ? F(MPX) : 0) : 0;
 
 	/* cpuid 1.edx */
 	const u32 kvm_supported_word0_x86_features =
@@ -303,7 +305,7 @@  static inline int __do_cpuid_ent(struct kvm_cpuid_entry2 *entry, u32 function,
 	/* cpuid 7.0.ebx */
 	const u32 kvm_supported_word9_x86_features =
 		F(FSGSBASE) | F(BMI1) | F(HLE) | F(AVX2) | F(SMEP) |
-		F(BMI2) | F(ERMS) | f_invpcid | F(RTM) | F(RDSEED) |
+		F(BMI2) | F(ERMS) | f_invpcid | F(RTM) | f_mpx | F(RDSEED) |
 		F(ADX);
 
 	/* all calls to cpuid_count() should be made on the same cpu */
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 89e4e27..3570e71 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -599,6 +599,9 @@  int __kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr)
 	u64 old_xcr0 = vcpu->arch.xcr0;
 	u64 valid_bits;
 
+	if (!kvm_x86_ops->mpx_supported || !kvm_x86_ops->mpx_supported())
+		xcr0 &= ~(XSTATE_BNDREGS | XSTATE_BNDCSR);
+
 	/* Only support XCR_XFEATURE_ENABLED_MASK(xcr0) now  */
 	if (index != XCR_XFEATURE_ENABLED_MASK)
 		return 1;
@@ -616,6 +619,9 @@  int __kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr)
 	if (xcr0 & ~valid_bits)
 		return 1;
 
+	if ((!(xcr0 & XSTATE_BNDREGS)) != (!(xcr0 & XSTATE_BNDCSR)))
+		return 1;
+
 	kvm_put_guest_xcr0(vcpu);
 	vcpu->arch.xcr0 = xcr0;
 
diff --git a/arch/x86/kvm/x86.h b/arch/x86/kvm/x86.h
index 8da5823..392ecbf 100644
--- a/arch/x86/kvm/x86.h
+++ b/arch/x86/kvm/x86.h
@@ -122,7 +122,8 @@  int kvm_write_guest_virt_system(struct x86_emulate_ctxt *ctxt,
 	gva_t addr, void *val, unsigned int bytes,
 	struct x86_exception *exception);
 
-#define KVM_SUPPORTED_XCR0	(XSTATE_FP | XSTATE_SSE | XSTATE_YMM)
+#define KVM_SUPPORTED_XCR0     (XSTATE_FP | XSTATE_SSE | XSTATE_YMM \
+				| XSTATE_BNDREGS | XSTATE_BNDCSR)
 extern u64 host_xcr0;
 
 extern unsigned int min_timer_period_us;