[PATCH 0/4] Memory type support for EPT

[PATCH 0/4] Memory type support for EPT

[Sheng Yang]

Hi, Avi

This patchset add memory type support for EPT, including MTRR and PAT.

The patch restruct the MTRR structure to make it easy to use, also take host
kernel MTRR as a reference.

The function guest_mtrr_type_lookup() is copied from host kernel rather than
reuse, for I think it may be much more complicate if we try to reuse the
function.

The patchset have been tested with VT-d and EPT.

The MTRR support for shadow page table is also under development, based on this
patchset.

Thanks!

--
regards
Yang, Sheng

[PATCH] KVM: VMX: Allocate MSR Bitmap for each vcpu

[Sheng Yang]

Separate msr_bitmap for each vcpu, prepared for guest PAT support.

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
---
 arch/x86/kvm/vmx.c |   53 +++++++++++++++++++++++++++++----------------------
 1 files changed, 30 insertions(+), 23 deletions(-)

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 14671f4..7f53fb7 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -90,6 +90,7 @@ struct vcpu_vmx {
 	} rmode;
 	int vpid;
 	bool emulation_required;
+	struct page *msr_bitmap;
 };
 
 static inline struct vcpu_vmx *to_vmx(struct kvm_vcpu *vcpu)
@@ -106,7 +107,6 @@ static DEFINE_PER_CPU(struct list_head, vcpus_on_cpu);
 
 static struct page *vmx_io_bitmap_a;
 static struct page *vmx_io_bitmap_b;
-static struct page *vmx_msr_bitmap;
 
 static DECLARE_BITMAP(vmx_vpid_bitmap, VMX_NR_VPIDS);
 static DEFINE_SPINLOCK(vmx_vpid_lock);
@@ -2083,6 +2083,25 @@ static void vmx_disable_intercept_for_msr(struct page *msr_bitmap, u32 msr)
 	kunmap(msr_bitmap);
 }
 
+static int setup_msr_bitmap(struct page *msr_bitmap)
+{
+	void *va;
+
+	va = kmap(msr_bitmap);
+	if (!va)
+		return -EINVAL;
+	memset(va, 0xff, PAGE_SIZE);
+	kunmap(msr_bitmap);
+
+	vmx_disable_intercept_for_msr(msr_bitmap, MSR_FS_BASE);
+	vmx_disable_intercept_for_msr(msr_bitmap, MSR_GS_BASE);
+	vmx_disable_intercept_for_msr(msr_bitmap, MSR_IA32_SYSENTER_CS);
+	vmx_disable_intercept_for_msr(msr_bitmap, MSR_IA32_SYSENTER_ESP);
+	vmx_disable_intercept_for_msr(msr_bitmap, MSR_IA32_SYSENTER_EIP);
+
+	return 0;
+}
+
 /*
  * Sets up the vmcs for emulated real mode.
  */
@@ -2100,8 +2119,10 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
 	vmcs_write64(IO_BITMAP_A, page_to_phys(vmx_io_bitmap_a));
 	vmcs_write64(IO_BITMAP_B, page_to_phys(vmx_io_bitmap_b));
 
-	if (cpu_has_vmx_msr_bitmap())
-		vmcs_write64(MSR_BITMAP, page_to_phys(vmx_msr_bitmap));
+	if (cpu_has_vmx_msr_bitmap()) {
+		setup_msr_bitmap(vmx->msr_bitmap);
+		vmcs_write64(MSR_BITMAP, page_to_phys(vmx->msr_bitmap));
+	}
 
 	vmcs_write64(VMCS_LINK_POINTER, -1ull); /* 22.3.1.5 */
 
@@ -3362,6 +3383,7 @@ static void vmx_free_vcpu(struct kvm_vcpu *vcpu)
 	vmx_free_vmcs(vcpu);
 	kfree(vmx->host_msrs);
 	kfree(vmx->guest_msrs);
+	__free_page(vmx->msr_bitmap);
 	kvm_vcpu_uninit(vcpu);
 	kmem_cache_free(kvm_vcpu_cache, vmx);
 }
@@ -3397,6 +3419,10 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
 
 	vmcs_clear(vmx->vmcs);
 
+	vmx->msr_bitmap = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
+	if (!vmx->msr_bitmap)
+		goto free_vmcs;
+
 	cpu = get_cpu();
 	vmx_vcpu_load(&vmx->vcpu, cpu);
 	err = vmx_vcpu_setup(vmx);
@@ -3518,12 +3544,6 @@ static int __init vmx_init(void)
 		goto out;
 	}
 
-	vmx_msr_bitmap = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
-	if (!vmx_msr_bitmap) {
-		r = -ENOMEM;
-		goto out1;
-	}
-
 	/*
 	 * Allow direct access to the PC debug port (it is often used for I/O
 	 * delays, but the vmexits simply slow things down).
@@ -3537,21 +3557,11 @@ static int __init vmx_init(void)
 	memset(va, 0xff, PAGE_SIZE);
 	kunmap(vmx_io_bitmap_b);
 
-	va = kmap(vmx_msr_bitmap);
-	memset(va, 0xff, PAGE_SIZE);
-	kunmap(vmx_msr_bitmap);
-
 	set_bit(0, vmx_vpid_bitmap); /* 0 is reserved for host */
 
 	r = kvm_init(&vmx_x86_ops, sizeof(struct vcpu_vmx), THIS_MODULE);
 	if (r)
-		goto out2;
-
-	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_FS_BASE);
-	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_GS_BASE);
-	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_CS);
-	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP);
-	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP);
+		goto out1;
 
 	if (vm_need_ept()) {
 		bypass_guest_pf = 0;
@@ -3572,8 +3582,6 @@ static int __init vmx_init(void)
 
 	return 0;
 
-out2:
-	__free_page(vmx_msr_bitmap);
 out1:
 	__free_page(vmx_io_bitmap_b);
 out:
@@ -3583,7 +3591,6 @@ out:
 
 static void __exit vmx_exit(void)
 {
-	__free_page(vmx_msr_bitmap);
 	__free_page(vmx_io_bitmap_b);
 	__free_page(vmx_io_bitmap_a);
 
-- 
1.5.4.5

[PATCH] KVM: VMX: Add PAT support for EPT

[Sheng Yang]

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
---
 arch/x86/kvm/vmx.c |   15 ++++++++++++---
 arch/x86/kvm/vmx.h |    7 +++++++
 2 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 7f53fb7..b74b43f 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -1173,12 +1173,13 @@ static __init int setup_vmcs_config(struct vmcs_config *vmcs_conf)
 #ifdef CONFIG_X86_64
 	min |= VM_EXIT_HOST_ADDR_SPACE_SIZE;
 #endif
-	opt = 0;
+	opt = VM_EXIT_LOAD_GUEST_PAT;
 	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_EXIT_CTLS,
 				&_vmexit_control) < 0)
 		return -EIO;
 
-	min = opt = 0;
+	min = 0;
+	opt = VM_ENTRY_SAVE_GUEST_PAT | VM_ENTRY_LOAD_HOST_PAT;
 	if (adjust_vmx_controls(min, opt, MSR_IA32_VMX_ENTRY_CTLS,
 				&_vmentry_control) < 0)
 		return -EIO;
@@ -2107,8 +2108,9 @@ static int setup_msr_bitmap(struct page *msr_bitmap)
  */
 static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
 {
-	u32 host_sysenter_cs;
+	u32 host_sysenter_cs, msr_low, msr_high;
 	u32 junk;
+	u64 host_pat;
 	unsigned long a;
 	struct descriptor_table dt;
 	int i;
@@ -2196,6 +2198,13 @@ static int vmx_vcpu_setup(struct vcpu_vmx *vmx)
 	vmcs_writel(HOST_IA32_SYSENTER_ESP, a);   /* 22.2.3 */
 	rdmsrl(MSR_IA32_SYSENTER_EIP, a);
 	vmcs_writel(HOST_IA32_SYSENTER_EIP, a);   /* 22.2.3 */
+	if (vmcs_config.vmentry_ctrl & VM_ENTRY_LOAD_HOST_PAT) {
+		rdmsr(MSR_IA32_CR_PAT, msr_low, msr_high);
+		host_pat = msr_low | ((u64) msr_high << 32);
+		vmcs_write64(HOST_PAT, host_pat);
+		vmx_disable_intercept_for_msr(vmx->msr_bitmap,
+					      MSR_IA32_CR_PAT);
+	}
 
 	for (i = 0; i < NR_VMX_MSR; ++i) {
 		u32 index = vmx_msr_index[i];
diff --git a/arch/x86/kvm/vmx.h b/arch/x86/kvm/vmx.h
index 0c22e5f..16b3cfb 100644
--- a/arch/x86/kvm/vmx.h
+++ b/arch/x86/kvm/vmx.h
@@ -62,11 +62,14 @@
 #define PIN_BASED_VIRTUAL_NMIS                  0x00000020
 
 #define VM_EXIT_HOST_ADDR_SPACE_SIZE            0x00000200
+#define VM_EXIT_LOAD_GUEST_PAT			0x00004000
 #define VM_EXIT_ACK_INTR_ON_EXIT                0x00008000
 
 #define VM_ENTRY_IA32E_MODE                     0x00000200
 #define VM_ENTRY_SMM                            0x00000400
 #define VM_ENTRY_DEACT_DUAL_MONITOR             0x00000800
+#define VM_ENTRY_SAVE_GUEST_PAT			0x00040000
+#define VM_ENTRY_LOAD_HOST_PAT			0x00080000
 
 /* VMCS Encodings */
 enum vmcs_field {
@@ -112,6 +115,8 @@ enum vmcs_field {
 	VMCS_LINK_POINTER_HIGH          = 0x00002801,
 	GUEST_IA32_DEBUGCTL             = 0x00002802,
 	GUEST_IA32_DEBUGCTL_HIGH        = 0x00002803,
+	GUEST_PAT			= 0x00002804,
+	GUEST_PAT_HIGH			= 0x00002805,
 	GUEST_PDPTR0                    = 0x0000280a,
 	GUEST_PDPTR0_HIGH               = 0x0000280b,
 	GUEST_PDPTR1                    = 0x0000280c,
@@ -120,6 +125,8 @@ enum vmcs_field {
 	GUEST_PDPTR2_HIGH               = 0x0000280f,
 	GUEST_PDPTR3                    = 0x00002810,
 	GUEST_PDPTR3_HIGH               = 0x00002811,
+	HOST_PAT			= 0x00002c00,
+	HOST_PAT_HIGH			= 0x00002c01,
 	PIN_BASED_VM_EXEC_CONTROL       = 0x00004000,
 	CPU_BASED_VM_EXEC_CONTROL       = 0x00004002,
 	EXCEPTION_BITMAP                = 0x00004004,
-- 
1.5.4.5

Re: [PATCH] KVM: VMX: Add PAT support for EPT

[Avi Kivity]

Sheng Yang wrote:


(some text wanted here)

> diff --git a/arch/x86/kvm/vmx.h b/arch/x86/kvm/vmx.h
> index 0c22e5f..16b3cfb 100644
> --- a/arch/x86/kvm/vmx.h
> +++ b/arch/x86/kvm/vmx.h
> @@ -62,11 +62,14 @@
>  #define PIN_BASED_VIRTUAL_NMIS                  0x00000020
>  
>  #define VM_EXIT_HOST_ADDR_SPACE_SIZE            0x00000200
> +#define VM_EXIT_LOAD_GUEST_PAT			0x00004000
>  #define VM_EXIT_ACK_INTR_ON_EXIT                0x00008000
>  
>  #define VM_ENTRY_IA32E_MODE                     0x00000200
>  #define VM_ENTRY_SMM                            0x00000400
>  #define VM_ENTRY_DEACT_DUAL_MONITOR             0x00000800
> +#define VM_ENTRY_SAVE_GUEST_PAT			0x00040000
> +#define VM_ENTRY_LOAD_HOST_PAT			0x00080000
>  
>  /* VMCS Encodings */
>  enum vmcs_field {
> @@ -112,6 +115,8 @@ enum vmcs_field {
>  	VMCS_LINK_POINTER_HIGH          = 0x00002801,
>  	GUEST_IA32_DEBUGCTL             = 0x00002802,
>  	GUEST_IA32_DEBUGCTL_HIGH        = 0x00002803,
> +	GUEST_PAT			= 0x00002804,
> +	GUEST_PAT_HIGH			= 0x00002805,
>  	GUEST_PDPTR0                    = 0x0000280a,
>  	GUEST_PDPTR0_HIGH               = 0x0000280b,
>  	GUEST_PDPTR1                    = 0x0000280c,
> @@ -120,6 +125,8 @@ enum vmcs_field {
>  	GUEST_PDPTR2_HIGH               = 0x0000280f,
>  	GUEST_PDPTR3                    = 0x00002810,
>  	GUEST_PDPTR3_HIGH               = 0x00002811,
> +	HOST_PAT			= 0x00002c00,
> +	HOST_PAT_HIGH			= 0x00002c01,
>  	PIN_BASED_VM_EXEC_CONTROL       = 0x00004000,
>  	CPU_BASED_VM_EXEC_CONTROL       = 0x00004002,
>  	EXCEPTION_BITMAP                = 0x00004004,
>   

This appears to be a new feature?  My documentation (a bit old) doesn't 
show it.  If so, we need a check to see that it is available.

An alternative to switching PAT is to translate a guest PAT index into a 
host PAT index.  But I guess that doesn't work with EPT?

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

Re: [PATCH] KVM: VMX: Add PAT support for EPT

[Avi Kivity]

Avi Kivity wrote:
>
> This appears to be a new feature?  My documentation (a bit old) 
> doesn't show it.  If so, we need a check to see that it is available.

The check is actually there.

If the feature is present, we need to expose it via 
KVM_GET_SUPPORTED_CPUID, and add save/restore support for the msr via 
msrs_to_save.

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

Re: [PATCH] KVM: VMX: Add PAT support for EPT

[Yang, Sheng]

On Tuesday 09 September 2008 22:36:22 Avi Kivity wrote:
> Avi Kivity wrote:
> > This appears to be a new feature?  My documentation (a bit old)
> > doesn't show it.  If so, we need a check to see that it is available.
>
> The check is actually there.
>
> If the feature is present, we need to expose it via
> KVM_GET_SUPPORTED_CPUID, and add save/restore support for the msr via
> msrs_to_save.

Yeah, it's a feature come with EPT. Thanks for reminder! Would update the 
patch soon. 

PS: The latest spec available at 
http://www.intel.com/products/processor/manuals/

It contains EPT and VPID and other new things on Nehalem. I would work on 
clearing up the code according to the latest spec soon (yeah, we also only 
got it for days...)

--
regards
Yang, Sheng

[PATCH] KVM: Improve MTRR structure

[Sheng Yang]

Most structure copied from x86 MTRR.

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
---
 arch/x86/kvm/x86.c         |   57 ++++++++++++++++++++++++++++++++++++++++++-
 include/asm-x86/kvm_host.h |   18 +++++++++++++-
 2 files changed, 72 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 3f3cb71..5d7ea05 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -65,6 +65,9 @@ static u64 __read_mostly efer_reserved_bits = 0xfffffffffffffffeULL;
 #define VM_STAT(x) offsetof(struct kvm, stat.x), KVM_STAT_VM
 #define VCPU_STAT(x) offsetof(struct kvm_vcpu, stat.x), KVM_STAT_VCPU
 
+#define MTRRphysBase_MSR(reg) (0x200 + 2 * (reg))
+#define MTRRphysMask_MSR(reg) (0x200 + 2 * (reg) + 1)
+
 static int kvm_dev_ioctl_get_supported_cpuid(struct kvm_cpuid2 *cpuid,
 				    struct kvm_cpuid_entry2 __user *entries);
 
@@ -865,10 +868,34 @@ static bool msr_mtrr_valid(unsigned msr)
 
 static int set_msr_mtrr(struct kvm_vcpu *vcpu, u32 msr, u64 data)
 {
+	u64 *p = (u64 *)&vcpu->arch.mtrr_state.fixed_ranges;
+
 	if (!msr_mtrr_valid(msr))
 		return 1;
 
-	vcpu->arch.mtrr[msr - 0x200] = data;
+	if (msr == MSR_MTRRdefType) {
+		vcpu->arch.mtrr_state.def_type = data;
+		vcpu->arch.mtrr_state.enabled = (data & 0xc00) >> 10;
+	} else if (msr == MSR_MTRRfix64K_00000)
+		p[0] = data;
+	else if (msr == MSR_MTRRfix16K_80000 || msr == MSR_MTRRfix16K_A0000)
+		p[1 + msr - MSR_MTRRfix16K_80000] = data;
+	else if (msr >= MSR_MTRRfix4K_C0000 && msr <= MSR_MTRRfix4K_F8000)
+		p[3 + msr - MSR_MTRRfix4K_C0000] = data;
+	else {	/* Variable MTRRs */
+		int idx, is_mtrr_mask;
+		u64 *pt;
+
+		idx = (msr - 0x200) / 2;
+		is_mtrr_mask = msr - 0x200 - 2 * idx;
+		if (!is_mtrr_mask)
+			pt =
+			  (u64 *)&vcpu->arch.mtrr_state.var_ranges[idx].base_lo;
+		else
+			pt =
+			  (u64 *)&vcpu->arch.mtrr_state.var_ranges[idx].mask_lo;
+		*pt = data;
+	}
 	return 0;
 }
 
@@ -966,10 +993,35 @@ int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata)
 
 static int get_msr_mtrr(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata)
 {
+	u64 *p = (u64 *)&vcpu->arch.mtrr_state.fixed_ranges;
+
 	if (!msr_mtrr_valid(msr))
 		return 1;
 
-	*pdata = vcpu->arch.mtrr[msr - 0x200];
+	if (msr == MSR_MTRRdefType)
+		*pdata = vcpu->arch.mtrr_state.def_type +
+			 (vcpu->arch.mtrr_state.enabled << 10);
+	else if (msr == MSR_MTRRfix64K_00000)
+		*pdata = p[0];
+	else if (msr == MSR_MTRRfix16K_80000 || msr == MSR_MTRRfix16K_A0000)
+		*pdata = p[1 + msr - MSR_MTRRfix16K_80000];
+	else if (msr >= MSR_MTRRfix4K_C0000 && msr <= MSR_MTRRfix4K_F8000)
+		*pdata = p[3 + msr - MSR_MTRRfix4K_C0000];
+	else {	/* Variable MTRRs */
+		int idx, is_mtrr_mask;
+		u64 *pt;
+
+		idx = (msr - 0x200) / 2;
+		is_mtrr_mask = msr - 0x200 - 2 * idx;
+		if (!is_mtrr_mask)
+			pt =
+			  (u64 *)&vcpu->arch.mtrr_state.var_ranges[idx].base_lo;
+		else
+			pt =
+			  (u64 *)&vcpu->arch.mtrr_state.var_ranges[idx].mask_lo;
+		*pdata = *pt;
+	}
+
 	return 0;
 }
 
@@ -4121,6 +4173,7 @@ int kvm_arch_vcpu_setup(struct kvm_vcpu *vcpu)
 	/* We do fxsave: this must be aligned. */
 	BUG_ON((unsigned long)&vcpu->arch.host_fx_image & 0xF);
 
+	vcpu->arch.mtrr_state.have_fixed = 1;
 	vcpu_load(vcpu);
 	r = kvm_arch_vcpu_reset(vcpu);
 	if (r == 0)
diff --git a/include/asm-x86/kvm_host.h b/include/asm-x86/kvm_host.h
index 815efc3..2d65df2 100644
--- a/include/asm-x86/kvm_host.h
+++ b/include/asm-x86/kvm_host.h
@@ -86,6 +86,7 @@
 #define KVM_MIN_FREE_MMU_PAGES 5
 #define KVM_REFILL_PAGES 25
 #define KVM_MAX_CPUID_ENTRIES 40
+#define KVM_NR_FIXED_MTRR_REGION 88
 #define KVM_NR_VAR_MTRR 8
 
 extern spinlock_t kvm_lock;
@@ -227,6 +228,21 @@ struct kvm_mmu {
 	u64 *pae_root;
 };
 
+struct mtrr_var_range {
+	u32 base_lo;
+	u32 base_hi;
+	u32 mask_lo;
+	u32 mask_hi;
+};
+
+struct mtrr_state_type {
+	struct mtrr_var_range var_ranges[KVM_NR_VAR_MTRR];
+	u8 fixed_ranges[KVM_NR_FIXED_MTRR_REGION];
+	unsigned char enabled;
+	unsigned char have_fixed;
+	u8 def_type;
+};
+
 struct kvm_vcpu_arch {
 	u64 host_tsc;
 	int interrupt_window_open;
@@ -322,7 +338,7 @@ struct kvm_vcpu_arch {
 	bool nmi_pending;
 	bool nmi_injected;
 
-	u64 mtrr[0x100];
+	struct mtrr_state_type mtrr_state;
 };
 
 struct kvm_mem_alias {
-- 
1.5.4.5

[PATCH] KVM: VMX: Add MTRR support for EPT

[Sheng Yang]

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
---
 arch/x86/kvm/mmu.c         |  115 +++++++++++++++++++++++++++++++++++++++++++-
 arch/x86/kvm/svm.c         |    6 ++
 arch/x86/kvm/vmx.c         |   24 ++++++++-
 arch/x86/kvm/x86.c         |    2 +-
 include/asm-x86/kvm_host.h |    3 +-
 5 files changed, 144 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index a87a11e..2cd772a 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -33,6 +33,7 @@
 #include <asm/page.h>
 #include <asm/cmpxchg.h>
 #include <asm/io.h>
+#include <asm/mtrr.h>
 
 /*
  * When setting this variable to true it enables Two-Dimensional-Paging
@@ -159,6 +160,7 @@ static u64 __read_mostly shadow_x_mask;	/* mutual exclusive with nx_mask */
 static u64 __read_mostly shadow_user_mask;
 static u64 __read_mostly shadow_accessed_mask;
 static u64 __read_mostly shadow_dirty_mask;
+static u64 __read_mostly shadow_mt_mask;
 
 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte)
 {
@@ -174,13 +176,14 @@ void kvm_mmu_set_base_ptes(u64 base_pte)
 EXPORT_SYMBOL_GPL(kvm_mmu_set_base_ptes);
 
 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
-		u64 dirty_mask, u64 nx_mask, u64 x_mask)
+		u64 dirty_mask, u64 nx_mask, u64 x_mask, u64 mt_mask)
 {
 	shadow_user_mask = user_mask;
 	shadow_accessed_mask = accessed_mask;
 	shadow_dirty_mask = dirty_mask;
 	shadow_nx_mask = nx_mask;
 	shadow_x_mask = x_mask;
+	shadow_mt_mask = mt_mask;
 }
 EXPORT_SYMBOL_GPL(kvm_mmu_set_mask_ptes);
 
@@ -1143,6 +1146,110 @@ struct page *gva_to_page(struct kvm_vcpu *vcpu, gva_t gva)
 	return page;
 }
 
+/*
+ * The function is based on mtrr_type_lookup() in
+ * arch/x86/kernel/cpu/mtrr/generic.c
+ */
+static int guest_mtrr_type_lookup(struct kvm_vcpu *vcpu, u64 start, u64 end)
+{
+	int i;
+	u64 base, mask;
+	u8 prev_match, curr_match;
+	int num_var_ranges = KVM_NR_VAR_MTRR;
+	struct mtrr_state_type mtrr_state = vcpu->arch.mtrr_state;
+
+	if (!mtrr_state.enabled)
+		return 0xFF;
+
+	/* Make end inclusive end, instead of exclusive */
+	end--;
+
+	/* Look in fixed ranges. Just return the type as per start */
+	if (mtrr_state.have_fixed && (start < 0x100000)) {
+		int idx;
+
+		if (start < 0x80000) {
+			idx = 0;
+			idx += (start >> 16);
+			return mtrr_state.fixed_ranges[idx];
+		} else if (start < 0xC0000) {
+			idx = 1 * 8;
+			idx += ((start - 0x80000) >> 14);
+			return mtrr_state.fixed_ranges[idx];
+		} else if (start < 0x1000000) {
+			idx = 3 * 8;
+			idx += ((start - 0xC0000) >> 12);
+			return mtrr_state.fixed_ranges[idx];
+		}
+	}
+
+	/*
+	 * Look in variable ranges
+	 * Look of multiple ranges matching this address and pick type
+	 * as per MTRR precedence
+	 */
+	if (!(mtrr_state.enabled & 2))
+		return mtrr_state.def_type;
+
+	prev_match = 0xFF;
+	for (i = 0; i < num_var_ranges; ++i) {
+		unsigned short start_state, end_state;
+
+		if (!(mtrr_state.var_ranges[i].mask_lo & (1 << 11)))
+			continue;
+
+		base = (((u64)mtrr_state.var_ranges[i].base_hi) << 32) +
+		       (mtrr_state.var_ranges[i].base_lo & PAGE_MASK);
+		mask = (((u64)mtrr_state.var_ranges[i].mask_hi) << 32) +
+		       (mtrr_state.var_ranges[i].mask_lo & PAGE_MASK);
+
+		start_state = ((start & mask) == (base & mask));
+		end_state = ((end & mask) == (base & mask));
+		if (start_state != end_state)
+			return 0xFE;
+
+		if ((start & mask) != (base & mask))
+			continue;
+
+		curr_match = mtrr_state.var_ranges[i].base_lo & 0xff;
+		if (prev_match == 0xFF) {
+			prev_match = curr_match;
+			continue;
+		}
+
+		if (prev_match == MTRR_TYPE_UNCACHABLE ||
+		    curr_match == MTRR_TYPE_UNCACHABLE)
+			return MTRR_TYPE_UNCACHABLE;
+
+		if ((prev_match == MTRR_TYPE_WRBACK &&
+		     curr_match == MTRR_TYPE_WRTHROUGH) ||
+		    (prev_match == MTRR_TYPE_WRTHROUGH &&
+		     curr_match == MTRR_TYPE_WRBACK)) {
+			prev_match = MTRR_TYPE_WRTHROUGH;
+			curr_match = MTRR_TYPE_WRTHROUGH;
+		}
+
+		if (prev_match != curr_match)
+			return MTRR_TYPE_UNCACHABLE;
+	}
+
+	if (prev_match != 0xFF)
+		return prev_match;
+
+	return mtrr_state.def_type;
+}
+
+static u8 get_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn)
+{
+	u8 mtrr;
+
+	mtrr = guest_mtrr_type_lookup(vcpu, gfn << PAGE_SHIFT,
+		       (gfn << PAGE_SHIFT) + PAGE_SIZE);
+	if (mtrr == 0xfe || mtrr == 0xff)
+		mtrr = MTRR_TYPE_WRBACK;
+	return mtrr;
+}
+
 static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte,
 			 unsigned pt_access, unsigned pte_access,
 			 int user_fault, int write_fault, int dirty,
@@ -1152,6 +1259,7 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte,
 	u64 spte;
 	int was_rmapped = 0;
 	int was_writeble = is_writeble_pte(*shadow_pte);
+	u64 mt_mask = shadow_mt_mask;
 
 	pgprintk("%s: spte %llx access %x write_fault %d"
 		 " user_fault %d gfn %lx\n",
@@ -1199,6 +1307,11 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte,
 		spte |= shadow_user_mask;
 	if (largepage)
 		spte |= PT_PAGE_SIZE_MASK;
+	if (mt_mask) {
+		mt_mask = get_memory_type(vcpu, gfn) <<
+			  kvm_x86_ops->get_mt_mask_shift();
+		spte |= mt_mask;
+	}
 
 	spte |= (u64)pfn << PAGE_SHIFT;
 
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 6f7f316..341f47e 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -1899,6 +1899,11 @@ static int get_npt_level(void)
 #endif
 }
 
+static int svm_get_mt_mask_shift(void)
+{
+	return 0;
+}
+
 static struct kvm_x86_ops svm_x86_ops = {
 	.cpu_has_kvm_support = has_svm,
 	.disabled_by_bios = is_disabled,
@@ -1954,6 +1959,7 @@ static struct kvm_x86_ops svm_x86_ops = {
 
 	.set_tss_addr = svm_set_tss_addr,
 	.get_tdp_level = get_npt_level,
+	.get_mt_mask_shift = svm_get_mt_mask_shift,
 };
 
 static int __init svm_init(void)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index b74b43f..0a86ca3 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -1508,10 +1508,17 @@ static void ept_update_paging_mode_cr4(unsigned long *hw_cr4,
 		*hw_cr4 &= ~X86_CR4_PAE;
 }
 
+static void vmx_mtrr_update(struct kvm_vcpu *vcpu)
+{
+	if (vm_need_ept())
+		kvm_mmu_reset_context(vcpu);
+}
+
 static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
 {
 	unsigned long hw_cr0 = (cr0 & ~KVM_GUEST_CR0_MASK) |
 				KVM_VM_CR0_ALWAYS_ON;
+	unsigned long last_cr0 = vmcs_readl(CR0_READ_SHADOW);
 
 	vmx_fpu_deactivate(vcpu);
 
@@ -1539,6 +1546,11 @@ static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
 
 	if (!(cr0 & X86_CR0_TS) || !(cr0 & X86_CR0_PE))
 		vmx_fpu_activate(vcpu);
+
+	/* According to SDM 10.11.8, check if need update MTRR */
+	if ((last_cr0 & X86_CR0_CD) && !(last_cr0 & X86_CR0_NW) &&
+	    !(cr0 & X86_CR0_CD) && !(cr0 & X86_CR0_NW))
+		vmx_mtrr_update(vcpu);
 }
 
 static u64 construct_eptp(unsigned long root_hpa)
@@ -3481,6 +3493,11 @@ static int get_ept_level(void)
 	return VMX_EPT_DEFAULT_GAW + 1;
 }
 
+static int vmx_get_mt_mask_shift(void)
+{
+	return VMX_EPT_MT_EPTE_SHIFT;
+}
+
 static struct kvm_x86_ops vmx_x86_ops = {
 	.cpu_has_kvm_support = cpu_has_kvm_support,
 	.disabled_by_bios = vmx_disabled_by_bios,
@@ -3536,6 +3553,7 @@ static struct kvm_x86_ops vmx_x86_ops = {
 
 	.set_tss_addr = vmx_set_tss_addr,
 	.get_tdp_level = get_ept_level,
+	.get_mt_mask_shift = vmx_get_mt_mask_shift,
 };
 
 static int __init vmx_init(void)
@@ -3575,11 +3593,11 @@ static int __init vmx_init(void)
 	if (vm_need_ept()) {
 		bypass_guest_pf = 0;
 		kvm_mmu_set_base_ptes(VMX_EPT_READABLE_MASK |
-			VMX_EPT_WRITABLE_MASK |
-			VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT);
+			VMX_EPT_WRITABLE_MASK);
 		kvm_mmu_set_mask_ptes(0ull, VMX_EPT_FAKE_ACCESSED_MASK,
 				VMX_EPT_FAKE_DIRTY_MASK, 0ull,
-				VMX_EPT_EXECUTABLE_MASK);
+				VMX_EPT_EXECUTABLE_MASK,
+				VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT);
 		kvm_enable_tdp();
 	} else
 		kvm_disable_tdp();
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 5d7ea05..231d6fe 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2831,7 +2831,7 @@ int kvm_arch_init(void *opaque)
 	kvm_mmu_set_nonpresent_ptes(0ull, 0ull);
 	kvm_mmu_set_base_ptes(PT_PRESENT_MASK);
 	kvm_mmu_set_mask_ptes(PT_USER_MASK, PT_ACCESSED_MASK,
-			PT_DIRTY_MASK, PT64_NX_MASK, 0);
+			PT_DIRTY_MASK, PT64_NX_MASK, 0, 0);
 	return 0;
 
 out:
diff --git a/include/asm-x86/kvm_host.h b/include/asm-x86/kvm_host.h
index 2d65df2..6369f52 100644
--- a/include/asm-x86/kvm_host.h
+++ b/include/asm-x86/kvm_host.h
@@ -502,6 +502,7 @@ struct kvm_x86_ops {
 
 	int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
 	int (*get_tdp_level)(void);
+	int (*get_mt_mask_shift)(void);
 };
 
 extern struct kvm_x86_ops *kvm_x86_ops;
@@ -515,7 +516,7 @@ int kvm_mmu_setup(struct kvm_vcpu *vcpu);
 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte);
 void kvm_mmu_set_base_ptes(u64 base_pte);
 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
-		u64 dirty_mask, u64 nx_mask, u64 x_mask);
+		u64 dirty_mask, u64 nx_mask, u64 x_mask, u64 mt_mask);
 
 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
-- 
1.5.4.5

Re: [PATCH 0/4] Memory type support for EPT

[Avi Kivity]

Sheng Yang wrote:
> Hi, Avi
>
> This patchset add memory type support for EPT, including MTRR and PAT.
>
> The patch restruct the MTRR structure to make it easy to use, also take host
> kernel MTRR as a reference.
>
> The function guest_mtrr_type_lookup() is copied from host kernel rather than
> reuse, for I think it may be much more complicate if we try to reuse the
> function.
>
> The patchset have been tested with VT-d and EPT.
>
> The MTRR support for shadow page table is also under development, based on this
> patchset.
>   

Please send with git send-email -n -- that preserves the patch order for 
me to apply.

(and thanks for getting git send-email to work, it's much more pleasant 
to review and apply those emails).

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