[PATCH v8 1/2] KVM: VMX: Enable XSAVE/XRSTOR for guest

[PATCH v8 1/2] KVM: VMX: Enable XSAVE/XRSTOR for guest

[Sheng Yang]

From: Dexuan Cui <dexuan.cui@intel.com>

This patch enable guest to use XSAVE/XRSTOR instructions.

We assume that host_xcr0 would use all possible bits that OS supported.

And we loaded xcr0 in the same way we handled fpu - do it as late as we can.

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Sheng Yang <sheng@linux.intel.com>
---
 arch/x86/include/asm/kvm_host.h |    2 +
 arch/x86/include/asm/vmx.h      |    1 +
 arch/x86/kvm/kvm_cache_regs.h   |    6 ++
 arch/x86/kvm/vmx.c              |   13 ++++
 arch/x86/kvm/x86.c              |  130 ++++++++++++++++++++++++++++++++++++--
 include/linux/kvm_host.h        |    2 +-
 6 files changed, 146 insertions(+), 8 deletions(-)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 0cd0f29..91631b8 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -302,6 +302,7 @@ struct kvm_vcpu_arch {
 	} update_pte;
 
 	struct fpu guest_fpu;
+	u64 xcr0;
 
 	gva_t mmio_fault_cr2;
 	struct kvm_pio_request pio;
@@ -605,6 +606,7 @@ int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
+int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
 
 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
diff --git a/arch/x86/include/asm/vmx.h b/arch/x86/include/asm/vmx.h
index 96a5886..9f0cbd9 100644
--- a/arch/x86/include/asm/vmx.h
+++ b/arch/x86/include/asm/vmx.h
@@ -267,6 +267,7 @@ enum vmcs_field {
 #define EXIT_REASON_EPT_VIOLATION       48
 #define EXIT_REASON_EPT_MISCONFIG       49
 #define EXIT_REASON_WBINVD		54
+#define EXIT_REASON_XSETBV		55
 
 /*
  * Interruption-information format
diff --git a/arch/x86/kvm/kvm_cache_regs.h b/arch/x86/kvm/kvm_cache_regs.h
index d2a98f8..6491ac8 100644
--- a/arch/x86/kvm/kvm_cache_regs.h
+++ b/arch/x86/kvm/kvm_cache_regs.h
@@ -71,4 +71,10 @@ static inline ulong kvm_read_cr4(struct kvm_vcpu *vcpu)
 	return kvm_read_cr4_bits(vcpu, ~0UL);
 }
 
+static inline u64 kvm_read_edx_eax(struct kvm_vcpu *vcpu)
+{
+	return (kvm_register_read(vcpu, VCPU_REGS_RAX) & -1u)
+		| ((u64)(kvm_register_read(vcpu, VCPU_REGS_RDX) & -1u) << 32);
+}
+
 #endif
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 7fa6ea7..d5d6a89 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -37,6 +37,8 @@
 #include <asm/vmx.h>
 #include <asm/virtext.h>
 #include <asm/mce.h>
+#include <asm/i387.h>
+#include <asm/xcr.h>
 
 #include "trace.h"
 
@@ -3393,6 +3395,16 @@ static int handle_wbinvd(struct kvm_vcpu *vcpu)
 	return 1;
 }
 
+static int handle_xsetbv(struct kvm_vcpu *vcpu)
+{
+	u64 new_bv = kvm_read_edx_eax(vcpu);
+	u32 index = kvm_register_read(vcpu, VCPU_REGS_RCX);
+
+	if (kvm_set_xcr(vcpu, index, new_bv) == 0)
+		skip_emulated_instruction(vcpu);
+	return 1;
+}
+
 static int handle_apic_access(struct kvm_vcpu *vcpu)
 {
 	return emulate_instruction(vcpu, 0, 0, 0) == EMULATE_DONE;
@@ -3671,6 +3683,7 @@ static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu) = {
 	[EXIT_REASON_TPR_BELOW_THRESHOLD]     = handle_tpr_below_threshold,
 	[EXIT_REASON_APIC_ACCESS]             = handle_apic_access,
 	[EXIT_REASON_WBINVD]                  = handle_wbinvd,
+	[EXIT_REASON_XSETBV]                  = handle_xsetbv,
 	[EXIT_REASON_TASK_SWITCH]             = handle_task_switch,
 	[EXIT_REASON_MCE_DURING_VMENTRY]      = handle_machine_check,
 	[EXIT_REASON_EPT_VIOLATION]	      = handle_ept_violation,
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 5fa8684..7a4073b 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -65,6 +65,7 @@
 	(~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
 			  | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE	\
 			  | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR	\
+			  | X86_CR4_OSXSAVE \
 			  | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE))
 
 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
@@ -150,6 +151,13 @@ struct kvm_stats_debugfs_item debugfs_entries[] = {
 	{ NULL }
 };
 
+u64 __read_mostly host_xcr0;
+
+static inline u32 bit(int bitno)
+{
+	return 1 << (bitno & 31);
+}
+
 static void kvm_on_user_return(struct user_return_notifier *urn)
 {
 	unsigned slot;
@@ -474,6 +482,61 @@ void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw)
 }
 EXPORT_SYMBOL_GPL(kvm_lmsw);
 
+int __kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr)
+{
+	u64 xcr0;
+
+	/* Only support XCR_XFEATURE_ENABLED_MASK(xcr0) now  */
+	if (index != XCR_XFEATURE_ENABLED_MASK)
+		return 1;
+	xcr0 = xcr;
+	if (kvm_x86_ops->get_cpl(vcpu) != 0)
+		return 1;
+	if (!(xcr0 & XSTATE_FP))
+		return 1;
+	if ((xcr0 & XSTATE_YMM) && !(xcr0 & XSTATE_SSE))
+		return 1;
+	if (xcr0 & ~host_xcr0)
+		return 1;
+	vcpu->arch.xcr0 = xcr0;
+	vcpu->guest_xcr0_loaded = 0;
+	return 0;
+}
+
+int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr)
+{
+	if (__kvm_set_xcr(vcpu, index, xcr)) {
+		kvm_inject_gp(vcpu, 0);
+		return 1;
+	}
+	return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_set_xcr);
+
+static bool guest_cpuid_has_xsave(struct kvm_vcpu *vcpu)
+{
+	struct kvm_cpuid_entry2 *best;
+
+	best = kvm_find_cpuid_entry(vcpu, 1, 0);
+	return best && (best->ecx & bit(X86_FEATURE_XSAVE));
+}
+
+static void update_cpuid(struct kvm_vcpu *vcpu)
+{
+	struct kvm_cpuid_entry2 *best;
+
+	best = kvm_find_cpuid_entry(vcpu, 1, 0);
+	if (!best)
+		return;
+
+	/* Update OSXSAVE bit */
+	if (cpu_has_xsave && best->function == 0x1) {
+		best->ecx &= ~(bit(X86_FEATURE_OSXSAVE));
+		if (kvm_read_cr4_bits(vcpu, X86_CR4_OSXSAVE))
+			best->ecx |= bit(X86_FEATURE_OSXSAVE);
+	}
+}
+
 int __kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
 {
 	unsigned long old_cr4 = kvm_read_cr4(vcpu);
@@ -482,6 +545,9 @@ int __kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
 	if (cr4 & CR4_RESERVED_BITS)
 		return 1;
 
+	if (!guest_cpuid_has_xsave(vcpu) && (cr4 & X86_CR4_OSXSAVE))
+		return 1;
+
 	if (is_long_mode(vcpu)) {
 		if (!(cr4 & X86_CR4_PAE))
 			return 1;
@@ -498,6 +564,9 @@ int __kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
 	if ((cr4 ^ old_cr4) & pdptr_bits)
 		kvm_mmu_reset_context(vcpu);
 
+	if ((cr4 ^ old_cr4) & X86_CR4_OSXSAVE)
+		update_cpuid(vcpu);
+
 	return 0;
 }
 
@@ -666,11 +735,6 @@ int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val)
 }
 EXPORT_SYMBOL_GPL(kvm_get_dr);
 
-static inline u32 bit(int bitno)
-{
-	return 1 << (bitno & 31);
-}
-
 /*
  * List of msr numbers which we expose to userspace through KVM_GET_MSRS
  * and KVM_SET_MSRS, and KVM_GET_MSR_INDEX_LIST.
@@ -1814,6 +1878,7 @@ static int kvm_vcpu_ioctl_set_cpuid(struct kvm_vcpu *vcpu,
 	r = 0;
 	kvm_apic_set_version(vcpu);
 	kvm_x86_ops->cpuid_update(vcpu);
+	update_cpuid(vcpu);
 
 out_free:
 	vfree(cpuid_entries);
@@ -1837,6 +1902,7 @@ static int kvm_vcpu_ioctl_set_cpuid2(struct kvm_vcpu *vcpu,
 	vcpu->arch.cpuid_nent = cpuid->nent;
 	kvm_apic_set_version(vcpu);
 	kvm_x86_ops->cpuid_update(vcpu);
+	update_cpuid(vcpu);
 	return 0;
 
 out:
@@ -1917,7 +1983,7 @@ static void do_cpuid_ent(struct kvm_cpuid_entry2 *entry, u32 function,
 		0 /* Reserved */ | F(CX16) | 0 /* xTPR Update, PDCM */ |
 		0 /* Reserved, DCA */ | F(XMM4_1) |
 		F(XMM4_2) | F(X2APIC) | F(MOVBE) | F(POPCNT) |
-		0 /* Reserved, XSAVE, OSXSAVE */;
+		0 /* Reserved, AES */ | F(XSAVE) | 0 /* OSXSAVE */;
 	/* cpuid 0x80000001.ecx */
 	const u32 kvm_supported_word6_x86_features =
 		F(LAHF_LM) | F(CMP_LEGACY) | F(SVM) | 0 /* ExtApicSpace */ |
@@ -1932,7 +1998,7 @@ static void do_cpuid_ent(struct kvm_cpuid_entry2 *entry, u32 function,
 
 	switch (function) {
 	case 0:
-		entry->eax = min(entry->eax, (u32)0xb);
+		entry->eax = min(entry->eax, (u32)0xd);
 		break;
 	case 1:
 		entry->edx &= kvm_supported_word0_x86_features;
@@ -1990,6 +2056,20 @@ static void do_cpuid_ent(struct kvm_cpuid_entry2 *entry, u32 function,
 		}
 		break;
 	}
+	case 0xd: {
+		int i;
+
+		entry->flags |= KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
+		for (i = 1; *nent < maxnent; ++i) {
+			if (entry[i - 1].eax == 0 && i != 2)
+				break;
+			do_cpuid_1_ent(&entry[i], function, i);
+			entry[i].flags |=
+			       KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
+			++*nent;
+		}
+		break;
+	}
 	case KVM_CPUID_SIGNATURE: {
 		char signature[12] = "KVMKVMKVM\0\0";
 		u32 *sigptr = (u32 *)signature;
@@ -4125,6 +4205,9 @@ int kvm_arch_init(void *opaque)
 
 	perf_register_guest_info_callbacks(&kvm_guest_cbs);
 
+	if (cpu_has_xsave)
+		host_xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
+
 	return 0;
 
 out:
@@ -4523,6 +4606,25 @@ static void inject_pending_event(struct kvm_vcpu *vcpu)
 	}
 }
 
+static void kvm_load_guest_xcr0(struct kvm_vcpu *vcpu)
+{
+	if (kvm_read_cr4_bits(vcpu, X86_CR4_OSXSAVE) &&
+			!vcpu->guest_xcr0_loaded) {
+		/* kvm_set_xcr() also depends on this */
+		xsetbv(XCR_XFEATURE_ENABLED_MASK, vcpu->arch.xcr0);
+		vcpu->guest_xcr0_loaded = 1;
+	}
+}
+
+static void kvm_put_guest_xcr0(struct kvm_vcpu *vcpu)
+{
+	if (vcpu->guest_xcr0_loaded) {
+		if (vcpu->arch.xcr0 != host_xcr0)
+			xsetbv(XCR_XFEATURE_ENABLED_MASK, host_xcr0);
+		vcpu->guest_xcr0_loaded = 0;
+	}
+}
+
 static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 {
 	int r;
@@ -4568,6 +4670,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 	kvm_x86_ops->prepare_guest_switch(vcpu);
 	if (vcpu->fpu_active)
 		kvm_load_guest_fpu(vcpu);
+	kvm_load_guest_xcr0(vcpu);
 
 	atomic_set(&vcpu->guest_mode, 1);
 	smp_wmb();
@@ -5124,6 +5227,11 @@ int fx_init(struct kvm_vcpu *vcpu)
 
 	fpu_finit(&vcpu->arch.guest_fpu);
 
+	/*
+	 * Ensure guest xcr0 is valid for loading
+	 */
+	vcpu->arch.xcr0 = XSTATE_FP;
+
 	vcpu->arch.cr0 |= X86_CR0_ET;
 
 	return 0;
@@ -5140,6 +5248,12 @@ void kvm_load_guest_fpu(struct kvm_vcpu *vcpu)
 	if (vcpu->guest_fpu_loaded)
 		return;
 
+	/*
+	 * Restore all possible states in the guest,
+	 * and assume host would use all available bits.
+	 * Guest xcr0 would be loaded later.
+	 */
+	kvm_put_guest_xcr0(vcpu);
 	vcpu->guest_fpu_loaded = 1;
 	unlazy_fpu(current);
 	fpu_restore_checking(&vcpu->arch.guest_fpu);
@@ -5148,6 +5262,8 @@ void kvm_load_guest_fpu(struct kvm_vcpu *vcpu)
 
 void kvm_put_guest_fpu(struct kvm_vcpu *vcpu)
 {
+	kvm_put_guest_xcr0(vcpu);
+
 	if (!vcpu->guest_fpu_loaded)
 		return;
 
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 2c62319..2d96555 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -88,7 +88,7 @@ struct kvm_vcpu {
 	int srcu_idx;
 
 	int fpu_active;
-	int guest_fpu_loaded;
+	int guest_fpu_loaded, guest_xcr0_loaded;
 	wait_queue_head_t wq;
 	int sigset_active;
 	sigset_t sigset;
-- 
1.7.0.1

[PATCH 2/2] KVM: x86: XSAVE/XRSTOR live migration support

[Sheng Yang]

This patch enable save/restore of xsave state.

Signed-off-by: Sheng Yang <sheng@linux.intel.com>
---
 Documentation/kvm/api.txt  |   76 ++++++++++++++++++++++++++++++
 arch/x86/include/asm/kvm.h |   24 ++++++++++
 arch/x86/kvm/x86.c         |  110 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/kvm.h        |   12 +++++
 4 files changed, 222 insertions(+), 0 deletions(-)

diff --git a/Documentation/kvm/api.txt b/Documentation/kvm/api.txt
index 159b4ef..cb0afa2 100644
--- a/Documentation/kvm/api.txt
+++ b/Documentation/kvm/api.txt
@@ -922,6 +922,82 @@ Define which vcpu is the Bootstrap Processor (BSP).  Values are the same
 as the vcpu id in KVM_CREATE_VCPU.  If this ioctl is not called, the default
 is vcpu 0.
 
+4.41 KVM_GET_XSAVE
+
+Capability: KVM_CAP_XSAVE
+Architectures: x86
+Type: vcpu ioctl
+Parameters: struct kvm_xsave (out)
+Returns: 0 on success, -1 on error
+
+struct kvm_xsave {
+	__u32 size;
+	__u32 region[1000];
+};
+
+This ioctl would copy current vcpu's xsave struct to the userspace.
+
+4.42 KVM_SET_XSAVE
+
+Capability: KVM_CAP_XSAVE
+Architectures: x86
+Type: vcpu ioctl
+Parameters: struct kvm_xsave (in)
+Returns: 0 on success, -1 on error
+
+struct kvm_xsave {
+	__u32 size;
+	__u32 region[1000];
+};
+
+This ioctl would copy userspace's xsave struct to the kernel.
+
+4.43 KVM_GET_XCRS
+
+Capability: KVM_CAP_XCRS
+Architectures: x86
+Type: vcpu ioctl
+Parameters: struct kvm_xcrs (out)
+Returns: 0 on success, -1 on error
+
+struct kvm_xcr {
+	__u32 xcr;
+	__u32 reserved;
+	__u64 value;
+};
+
+struct kvm_xcrs {
+	__u32 nr_xcrs;
+	__u32 flags;
+	struct kvm_xcr xcrs[KVM_MAX_XCRS];
+	__u64 padding[16];
+};
+
+This ioctl would copy current vcpu's xcrs to the userspace.
+
+4.44 KVM_SET_XCRS
+
+Capability: KVM_CAP_XCRS
+Architectures: x86
+Type: vcpu ioctl
+Parameters: struct kvm_xcrs (in)
+Returns: 0 on success, -1 on error
+
+struct kvm_xcr {
+	__u32 xcr;
+	__u32 reserved;
+	__u64 value;
+};
+
+struct kvm_xcrs {
+	__u32 nr_xcrs;
+	__u32 flags;
+	struct kvm_xcr xcrs[KVM_MAX_XCRS];
+	__u64 padding[16];
+};
+
+This ioctl would set vcpu's xcr to the value userspace specified.
+
 5. The kvm_run structure
 
 Application code obtains a pointer to the kvm_run structure by
diff --git a/arch/x86/include/asm/kvm.h b/arch/x86/include/asm/kvm.h
index ff90055..a12a24a 100644
--- a/arch/x86/include/asm/kvm.h
+++ b/arch/x86/include/asm/kvm.h
@@ -22,6 +22,8 @@
 #define __KVM_HAVE_XEN_HVM
 #define __KVM_HAVE_VCPU_EVENTS
 #define __KVM_HAVE_DEBUGREGS
+#define __KVM_HAVE_XSAVE
+#define __KVM_HAVE_XCRS
 
 /* Architectural interrupt line count. */
 #define KVM_NR_INTERRUPTS 256
@@ -299,4 +301,26 @@ struct kvm_debugregs {
 	__u64 reserved[9];
 };
 
+/* for KVM_CAP_XSAVE */
+struct kvm_xsave {
+	__u32 size;
+	__u32 reserved[3];
+	__u32 region[1020];
+};
+
+#define KVM_MAX_XCRS	1
+
+struct kvm_xcr {
+	__u32 xcr;
+	__u32 reserved;
+	__u64 value;
+};
+
+struct kvm_xcrs {
+	__u32 nr_xcrs;
+	__u32 flags;
+	struct kvm_xcr xcrs[KVM_MAX_XCRS];
+	__u64 padding[16];
+};
+
 #endif /* _ASM_X86_KVM_H */
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 7a4073b..0fed328 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -1698,6 +1698,8 @@ int kvm_dev_ioctl_check_extension(long ext)
 	case KVM_CAP_PCI_SEGMENT:
 	case KVM_CAP_DEBUGREGS:
 	case KVM_CAP_X86_ROBUST_SINGLESTEP:
+	case KVM_CAP_XSAVE:
+	case KVM_CAP_XCRS:
 		r = 1;
 		break;
 	case KVM_CAP_COALESCED_MMIO:
@@ -2373,6 +2375,60 @@ static int kvm_vcpu_ioctl_x86_set_debugregs(struct kvm_vcpu *vcpu,
 	return 0;
 }
 
+static void kvm_vcpu_ioctl_x86_get_xsave(struct kvm_vcpu *vcpu,
+					struct kvm_xsave *guest_xsave)
+{
+	u32 size;
+
+	if (cpu_has_xsave)
+		size = sizeof(struct xsave_struct);
+	else
+		size = sizeof(struct i387_fxsave_struct);
+
+	guest_xsave->size = size;
+	memcpy(guest_xsave->region, &vcpu->arch.guest_fpu.state->xsave, size);
+}
+
+static int kvm_vcpu_ioctl_x86_set_xsave(struct kvm_vcpu *vcpu,
+					struct kvm_xsave *guest_xsave)
+{
+	u32 size;
+
+	if (cpu_has_xsave)
+		size = sizeof(struct xsave_struct);
+	else
+		size = sizeof(struct i387_fxsave_struct);
+
+	if (guest_xsave->size > size)
+		return -EFAULT;
+	else
+		size = guest_xsave->size;
+	memcpy(&vcpu->arch.guest_fpu.state->xsave, guest_xsave->region, size);
+	return 0;
+}
+
+static void kvm_vcpu_ioctl_x86_get_xcrs(struct kvm_vcpu *vcpu,
+					struct kvm_xcrs *guest_xcrs)
+{
+	guest_xcrs->nr_xcrs = 1;
+	guest_xcrs->flags = 0;
+	guest_xcrs->xcrs[0].value = vcpu->arch.xcr0;
+}
+
+static int kvm_vcpu_ioctl_x86_set_xcrs(struct kvm_vcpu *vcpu,
+				       struct kvm_xcrs *guest_xcrs)
+{
+	if (guest_xcrs->nr_xcrs < 1)
+		return 1;
+
+	/* Userspace may override the initial value of xcr0... */
+	if (guest_xcrs->xcrs[0].value != 0) {
+		vcpu->arch.xcr0 = guest_xcrs->xcrs[0].value;
+		vcpu->guest_xcr0_loaded = 0;
+	}
+	return 0;
+}
+
 long kvm_arch_vcpu_ioctl(struct file *filp,
 			 unsigned int ioctl, unsigned long arg)
 {
@@ -2574,6 +2630,60 @@ long kvm_arch_vcpu_ioctl(struct file *filp,
 		r = kvm_vcpu_ioctl_x86_set_debugregs(vcpu, &dbgregs);
 		break;
 	}
+	case KVM_GET_XSAVE: {
+		struct kvm_xsave *xsave;
+
+		xsave = kzalloc(sizeof(struct kvm_xsave), GFP_KERNEL);
+		r = -ENOMEM;
+		if (!xsave)
+			break;
+
+		kvm_vcpu_ioctl_x86_get_xsave(vcpu, xsave);
+
+		r = -EFAULT;
+		if (copy_to_user(argp, xsave, sizeof(struct kvm_xsave)))
+			break;
+		r = 0;
+		break;
+	}
+	case KVM_SET_XSAVE: {
+		struct kvm_xsave *xsave;
+
+		xsave = kzalloc(sizeof(struct kvm_xsave), GFP_KERNEL);
+		r = -ENOMEM;
+		if (!xsave)
+			break;
+
+		r = -EFAULT;
+		if (copy_from_user(xsave, argp, sizeof(struct kvm_xsave)))
+			break;
+
+		r = kvm_vcpu_ioctl_x86_set_xsave(vcpu, xsave);
+		break;
+	}
+	case KVM_GET_XCRS: {
+		struct kvm_xcrs xcrs;
+
+		kvm_vcpu_ioctl_x86_get_xcrs(vcpu, &xcrs);
+
+		r = -EFAULT;
+		if (copy_to_user(argp, &xcrs,
+				 sizeof(struct kvm_xcrs)))
+			break;
+		r = 0;
+		break;
+	}
+	case KVM_SET_XCRS: {
+		struct kvm_xcrs xcrs;
+
+		r = -EFAULT;
+		if (copy_from_user(&xcrs, argp,
+				   sizeof(struct kvm_xcrs)))
+			break;
+
+		r = kvm_vcpu_ioctl_x86_set_xcrs(vcpu, &xcrs);
+		break;
+	}
 	default:
 		r = -EINVAL;
 	}
diff --git a/include/linux/kvm.h b/include/linux/kvm.h
index 23ea022..6fd40f5 100644
--- a/include/linux/kvm.h
+++ b/include/linux/kvm.h
@@ -524,6 +524,12 @@ struct kvm_enable_cap {
 #define KVM_CAP_PPC_OSI 52
 #define KVM_CAP_PPC_UNSET_IRQ 53
 #define KVM_CAP_ENABLE_CAP 54
+#ifdef __KVM_HAVE_XSAVE
+#define KVM_CAP_XSAVE 55
+#endif
+#ifdef __KVM_HAVE_XCRS
+#define KVM_CAP_XCRS 56
+#endif
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -714,6 +720,12 @@ struct kvm_clock_data {
 #define KVM_GET_DEBUGREGS         _IOR(KVMIO,  0xa1, struct kvm_debugregs)
 #define KVM_SET_DEBUGREGS         _IOW(KVMIO,  0xa2, struct kvm_debugregs)
 #define KVM_ENABLE_CAP            _IOW(KVMIO,  0xa3, struct kvm_enable_cap)
+/* Available with KVM_CAP_XSAVE */
+#define KVM_GET_XSAVE		  _IOR(KVMIO,  0xa4, struct kvm_xsave)
+#define KVM_SET_XSAVE		  _IOW(KVMIO,  0xa5, struct kvm_xsave)
+/* Available with KVM_CAP_XCRS */
+#define KVM_GET_XCRS		  _IOR(KVMIO,  0xa6, struct kvm_xcrs)
+#define KVM_SET_XCRS		  _IOW(KVMIO,  0xa7, struct kvm_xcrs)
 
 #define KVM_DEV_ASSIGN_ENABLE_IOMMU	(1 << 0)
 
-- 
1.7.0.1

Re: [PATCH 2/2] KVM: x86: XSAVE/XRSTOR live migration support

[Avi Kivity]

On 06/10/2010 06:27 AM, Sheng Yang wrote:
> This patch enable save/restore of xsave state.
>
> Signed-off-by: Sheng Yang<sheng@linux.intel.com>
> ---
>   Documentation/kvm/api.txt  |   76 ++++++++++++++++++++++++++++++
>   arch/x86/include/asm/kvm.h |   24 ++++++++++
>   arch/x86/kvm/x86.c         |  110 ++++++++++++++++++++++++++++++++++++++++++++
>   include/linux/kvm.h        |   12 +++++
>   4 files changed, 222 insertions(+), 0 deletions(-)
>
> diff --git a/Documentation/kvm/api.txt b/Documentation/kvm/api.txt
> index 159b4ef..cb0afa2 100644
> --- a/Documentation/kvm/api.txt
> +++ b/Documentation/kvm/api.txt
> @@ -922,6 +922,82 @@ Define which vcpu is the Bootstrap Processor (BSP).  Values are the same
>   as the vcpu id in KVM_CREATE_VCPU.  If this ioctl is not called, the default
>   is vcpu 0.
>
> +4.41 KVM_GET_XSAVE
> +
> +Capability: KVM_CAP_XSAVE
> +Architectures: x86
> +Type: vcpu ioctl
> +Parameters: struct kvm_xsave (out)
> +Returns: 0 on success, -1 on error
> +
> +struct kvm_xsave {
> +	__u32 size;
> +	__u32 region[1000];
> +};
> +
> +This ioctl would copy current vcpu's xsave struct to the userspace.
>    

How is size interpreted?

I think we can leave it out since it is implied by the xsave header.

>
> +/* for KVM_CAP_XSAVE */
> +struct kvm_xsave {
> +	__u32 size;
> +	__u32 reserved[3];
> +	__u32 region[1020];
> +};
>    

That's different from the documentation.

> +
> +#define KVM_MAX_XCRS	1
>    

Needs to be much bigger for future compatibility, since it's part of the 
ABI.

> @@ -2373,6 +2375,60 @@ static int kvm_vcpu_ioctl_x86_set_debugregs(struct kvm_vcpu *vcpu,
>   	return 0;
>   }
>
> +static void kvm_vcpu_ioctl_x86_get_xsave(struct kvm_vcpu *vcpu,
> +					struct kvm_xsave *guest_xsave)
> +{
> +	u32 size;
> +
> +	if (cpu_has_xsave)
> +		size = sizeof(struct xsave_struct);
> +	else
> +		size = sizeof(struct i387_fxsave_struct);
> +
> +	guest_xsave->size = size;
> +	memcpy(guest_xsave->region,&vcpu->arch.guest_fpu.state->xsave, size);
>    

If !cpu_has_xsave, we can create a fake xsave header that says we have 
the fpu and sse, so userspace doesn't have to look at size.

You're memcpy()ing state->xsave unconditionally, even if 
!cpu_has_xsave.  That works, but it's cleaner to access the ->fxsave member.

> +
> +static void kvm_vcpu_ioctl_x86_get_xcrs(struct kvm_vcpu *vcpu,
> +					struct kvm_xcrs *guest_xcrs)
> +{
> +	guest_xcrs->nr_xcrs = 1;
> +	guest_xcrs->flags = 0;
> +	guest_xcrs->xcrs[0].value = vcpu->arch.xcr0;
> +}
>    


guest_xcrs[0].xcr = ...

What if !cpu_has_xsave?  Probably should set nr_xcrs = 0.

> +
> +static int kvm_vcpu_ioctl_x86_set_xcrs(struct kvm_vcpu *vcpu,
> +				       struct kvm_xcrs *guest_xcrs)
> +{
> +	if (guest_xcrs->nr_xcrs<  1)
> +		return 1;
>    

Not really needed.

> +
> +	/* Userspace may override the initial value of xcr0... */
> +	if (guest_xcrs->xcrs[0].value != 0) {
>    

Why check the value?  What about checking the xcrs[].xcr?

> +		vcpu->arch.xcr0 = guest_xcrs->xcrs[0].value;
> +		vcpu->guest_xcr0_loaded = 0;
> +	}
>    

Need to loop over the array.  Also, call kvm_set_xcr() to get checks to 
be performed.

-- 
error compiling committee.c: too many arguments to function

Re: [PATCH v8 1/2] KVM: VMX: Enable XSAVE/XRSTOR for guest

[Avi Kivity]

On 06/10/2010 06:27 AM, Sheng Yang wrote:
> From: Dexuan Cui<dexuan.cui@intel.com>
>
> This patch enable guest to use XSAVE/XRSTOR instructions.
>
> We assume that host_xcr0 would use all possible bits that OS supported.
>
> And we loaded xcr0 in the same way we handled fpu - do it as late as we can.
>
>    

Looks good.  Will wait for Marcelo to review before applying.

-- 
error compiling committee.c: too many arguments to function

Re: [PATCH v8 1/2] KVM: VMX: Enable XSAVE/XRSTOR for guest

[Marcelo Tosatti]

On Thu, Jun 10, 2010 at 11:27:12AM +0800, Sheng Yang wrote:
> From: Dexuan Cui <dexuan.cui@intel.com>
> 
> This patch enable guest to use XSAVE/XRSTOR instructions.
> 
> We assume that host_xcr0 would use all possible bits that OS supported.
> 
> And we loaded xcr0 in the same way we handled fpu - do it as late as we can.
> 
> Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
> Signed-off-by: Sheng Yang <sheng@linux.intel.com>
> ---
>  arch/x86/include/asm/kvm_host.h |    2 +
>  arch/x86/include/asm/vmx.h      |    1 +
>  arch/x86/kvm/kvm_cache_regs.h   |    6 ++
>  arch/x86/kvm/vmx.c              |   13 ++++
>  arch/x86/kvm/x86.c              |  130 ++++++++++++++++++++++++++++++++++++--
>  include/linux/kvm_host.h        |    2 +-
>  6 files changed, 146 insertions(+), 8 deletions(-)

Reviewed-by: Marcelo Tosatti <mtosatti@redhat.com>

Re: [PATCH v8 1/2] KVM: VMX: Enable XSAVE/XRSTOR for guest

[Avi Kivity]

On 06/10/2010 06:27 AM, Sheng Yang wrote:
> From: Dexuan Cui<dexuan.cui@intel.com>
>
> This patch enable guest to use XSAVE/XRSTOR instructions.
>
> We assume that host_xcr0 would use all possible bits that OS supported.
>
> And we loaded xcr0 in the same way we handled fpu - do it as late as we can.
>
>    

Applied, thanks.

-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.