KVM: VMX: Enable EPT feature for KVM

Signed-off-by: Sheng Yang <sheng.yang@intel.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

arch/x86/kvm/mmu.c

@@ -1177,8 +1177,9 @@ static int __direct_map(struct kvm_vcpu *vcpu, gpa_t v, int write,
 				return -ENOMEM;
 			}
 
-			table[index] = __pa(new_table->spt) | PT_PRESENT_MASK
-				| PT_WRITABLE_MASK | shadow_user_mask;
+			table[index] = __pa(new_table->spt)
+				| PT_PRESENT_MASK | PT_WRITABLE_MASK
+				| shadow_user_mask | shadow_x_mask;
 		}
 		table_addr = table[index] & PT64_BASE_ADDR_MASK;
 	}

arch/x86/kvm/vmx.c

@@ -42,7 +42,7 @@ module_param(enable_vpid, bool, 0);
 static int flexpriority_enabled = 1;
 module_param(flexpriority_enabled, bool, 0);
 
-static int enable_ept;
+static int enable_ept = 1;
 module_param(enable_ept, bool, 0);
 
 struct vmcs {
@@ -284,6 +284,18 @@ static inline void __invvpid(int ext, u16 vpid, gva_t gva)
 		  : : "a"(&operand), "c"(ext) : "cc", "memory");
 }
 
+static inline void __invept(int ext, u64 eptp, gpa_t gpa)
+{
+	struct {
+		u64 eptp, gpa;
+	} operand = {eptp, gpa};
+
+	asm volatile (ASM_VMX_INVEPT
+			/* CF==1 or ZF==1 --> rc = -1 */
+			"; ja 1f ; ud2 ; 1:\n"
+			: : "a" (&operand), "c" (ext) : "cc", "memory");
+}
+
 static struct kvm_msr_entry *find_msr_entry(struct vcpu_vmx *vmx, u32 msr)
 {
 	int i;
@@ -335,6 +347,33 @@ static inline void vpid_sync_vcpu_all(struct vcpu_vmx *vmx)
 	__invvpid(VMX_VPID_EXTENT_SINGLE_CONTEXT, vmx->vpid, 0);
 }
 
+static inline void ept_sync_global(void)
+{
+	if (cpu_has_vmx_invept_global())
+		__invept(VMX_EPT_EXTENT_GLOBAL, 0, 0);
+}
+
+static inline void ept_sync_context(u64 eptp)
+{
+	if (vm_need_ept()) {
+		if (cpu_has_vmx_invept_context())
+			__invept(VMX_EPT_EXTENT_CONTEXT, eptp, 0);
+		else
+			ept_sync_global();
+	}
+}
+
+static inline void ept_sync_individual_addr(u64 eptp, gpa_t gpa)
+{
+	if (vm_need_ept()) {
+		if (cpu_has_vmx_invept_individual_addr())
+			__invept(VMX_EPT_EXTENT_INDIVIDUAL_ADDR,
+					eptp, gpa);
+		else
+			ept_sync_context(eptp);
+	}
+}
+
 static unsigned long vmcs_readl(unsigned long field)
 {
 	unsigned long value;
@@ -422,6 +461,8 @@ static void update_exception_bitmap(struct kvm_vcpu *vcpu)
 		eb |= 1u << 1;
 	if (vcpu->arch.rmode.active)
 		eb = ~0;
+	if (vm_need_ept())
+		eb &= ~(1u << PF_VECTOR); /* bypass_guest_pf = 0 */
 	vmcs_write32(EXCEPTION_BITMAP, eb);
 }
 
@@ -1352,8 +1393,64 @@ static void vmx_decache_cr4_guest_bits(struct kvm_vcpu *vcpu)
 	vcpu->arch.cr4 |= vmcs_readl(GUEST_CR4) & ~KVM_GUEST_CR4_MASK;
 }
 
+static void ept_load_pdptrs(struct kvm_vcpu *vcpu)
+{
+	if (is_paging(vcpu) && is_pae(vcpu) && !is_long_mode(vcpu)) {
+		if (!load_pdptrs(vcpu, vcpu->arch.cr3)) {
+			printk(KERN_ERR "EPT: Fail to load pdptrs!\n");
+			return;
+		}
+		vmcs_write64(GUEST_PDPTR0, vcpu->arch.pdptrs[0]);
+		vmcs_write64(GUEST_PDPTR1, vcpu->arch.pdptrs[1]);
+		vmcs_write64(GUEST_PDPTR2, vcpu->arch.pdptrs[2]);
+		vmcs_write64(GUEST_PDPTR3, vcpu->arch.pdptrs[3]);
+	}
+}
+
+static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
+
+static void ept_update_paging_mode_cr0(unsigned long *hw_cr0,
+					unsigned long cr0,
+					struct kvm_vcpu *vcpu)
+{
+	if (!(cr0 & X86_CR0_PG)) {
+		/* From paging/starting to nonpaging */
+		vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
+			     vmcs_config.cpu_based_exec_ctrl |
+			     (CPU_BASED_CR3_LOAD_EXITING |
+			      CPU_BASED_CR3_STORE_EXITING));
+		vcpu->arch.cr0 = cr0;
+		vmx_set_cr4(vcpu, vcpu->arch.cr4);
+		*hw_cr0 |= X86_CR0_PE | X86_CR0_PG;
+		*hw_cr0 &= ~X86_CR0_WP;
+	} else if (!is_paging(vcpu)) {
+		/* From nonpaging to paging */
+		vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
+			     vmcs_config.cpu_based_exec_ctrl &
+			     ~(CPU_BASED_CR3_LOAD_EXITING |
+			       CPU_BASED_CR3_STORE_EXITING));
+		vcpu->arch.cr0 = cr0;
+		vmx_set_cr4(vcpu, vcpu->arch.cr4);
+		if (!(vcpu->arch.cr0 & X86_CR0_WP))
+			*hw_cr0 &= ~X86_CR0_WP;
+	}
+}
+
+static void ept_update_paging_mode_cr4(unsigned long *hw_cr4,
+					struct kvm_vcpu *vcpu)
+{
+	if (!is_paging(vcpu)) {
+		*hw_cr4 &= ~X86_CR4_PAE;
+		*hw_cr4 |= X86_CR4_PSE;
+	} else if (!(vcpu->arch.cr4 & X86_CR4_PAE))
+		*hw_cr4 &= ~X86_CR4_PAE;
+}
+
 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;
+
 	vmx_fpu_deactivate(vcpu);
 
 	if (vcpu->arch.rmode.active && (cr0 & X86_CR0_PE))
@@ -1371,29 +1468,61 @@ static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
 	}
 #endif
 
+	if (vm_need_ept())
+		ept_update_paging_mode_cr0(&hw_cr0, cr0, vcpu);
+
 	vmcs_writel(CR0_READ_SHADOW, cr0);
-	vmcs_writel(GUEST_CR0,
-		    (cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
+	vmcs_writel(GUEST_CR0, hw_cr0);
 	vcpu->arch.cr0 = cr0;
 
 	if (!(cr0 & X86_CR0_TS) || !(cr0 & X86_CR0_PE))
 		vmx_fpu_activate(vcpu);
 }
 
+static u64 construct_eptp(unsigned long root_hpa)
+{
+	u64 eptp;
+
+	/* TODO write the value reading from MSR */
+	eptp = VMX_EPT_DEFAULT_MT |
+		VMX_EPT_DEFAULT_GAW << VMX_EPT_GAW_EPTP_SHIFT;
+	eptp |= (root_hpa & PAGE_MASK);
+
+	return eptp;
+}
+
 static void vmx_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
 {
+	unsigned long guest_cr3;
+	u64 eptp;
+
+	guest_cr3 = cr3;
+	if (vm_need_ept()) {
+		eptp = construct_eptp(cr3);
+		vmcs_write64(EPT_POINTER, eptp);
+		ept_sync_context(eptp);
+		ept_load_pdptrs(vcpu);
+		guest_cr3 = is_paging(vcpu) ? vcpu->arch.cr3 :
+			VMX_EPT_IDENTITY_PAGETABLE_ADDR;
+	}
+
 	vmx_flush_tlb(vcpu);
-	vmcs_writel(GUEST_CR3, cr3);
+	vmcs_writel(GUEST_CR3, guest_cr3);
 	if (vcpu->arch.cr0 & X86_CR0_PE)
 		vmx_fpu_deactivate(vcpu);
 }
 
 static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
 {
-	vmcs_writel(CR4_READ_SHADOW, cr4);
-	vmcs_writel(GUEST_CR4, cr4 | (vcpu->arch.rmode.active ?
-		    KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON));
+	unsigned long hw_cr4 = cr4 | (vcpu->arch.rmode.active ?
+		    KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON);
+
 	vcpu->arch.cr4 = cr4;
+	if (vm_need_ept())
+		ept_update_paging_mode_cr4(&hw_cr4, vcpu);
+
+	vmcs_writel(CR4_READ_SHADOW, cr4);
+	vmcs_writel(GUEST_CR4, hw_cr4);
 }
 
 static void vmx_set_efer(struct kvm_vcpu *vcpu, u64 efer)
@@ -2116,6 +2245,9 @@ static int handle_exception(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 	if (intr_info & INTR_INFO_DELIVER_CODE_MASK)
 		error_code = vmcs_read32(VM_EXIT_INTR_ERROR_CODE);
 	if (is_page_fault(intr_info)) {
+		/* EPT won't cause page fault directly */
+		if (vm_need_ept())
+			BUG();
 		cr2 = vmcs_readl(EXIT_QUALIFICATION);
 		KVMTRACE_3D(PAGE_FAULT, vcpu, error_code, (u32)cr2,
 			    (u32)((u64)cr2 >> 32), handler);
@@ -2445,6 +2577,64 @@ static int handle_task_switch(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 	return kvm_task_switch(vcpu, tss_selector, reason);
 }
 
+static int handle_ept_violation(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+	u64 exit_qualification;
+	enum emulation_result er;
+	gpa_t gpa;
+	unsigned long hva;
+	int gla_validity;
+	int r;
+
+	exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
+
+	if (exit_qualification & (1 << 6)) {
+		printk(KERN_ERR "EPT: GPA exceeds GAW!\n");
+		return -ENOTSUPP;
+	}
+
+	gla_validity = (exit_qualification >> 7) & 0x3;
+	if (gla_validity != 0x3 && gla_validity != 0x1 && gla_validity != 0) {
+		printk(KERN_ERR "EPT: Handling EPT violation failed!\n");
+		printk(KERN_ERR "EPT: GPA: 0x%lx, GVA: 0x%lx\n",
+			(long unsigned int)vmcs_read64(GUEST_PHYSICAL_ADDRESS),
+			(long unsigned int)vmcs_read64(GUEST_LINEAR_ADDRESS));
+		printk(KERN_ERR "EPT: Exit qualification is 0x%lx\n",
+			(long unsigned int)exit_qualification);
+		kvm_run->exit_reason = KVM_EXIT_UNKNOWN;
+		kvm_run->hw.hardware_exit_reason = 0;
+		return -ENOTSUPP;
+	}
+
+	gpa = vmcs_read64(GUEST_PHYSICAL_ADDRESS);
+	hva = gfn_to_hva(vcpu->kvm, gpa >> PAGE_SHIFT);
+	if (!kvm_is_error_hva(hva)) {
+		r = kvm_mmu_page_fault(vcpu, gpa & PAGE_MASK, 0);
+		if (r < 0) {
+			printk(KERN_ERR "EPT: Not enough memory!\n");
+			return -ENOMEM;
+		}
+		return 1;
+	} else {
+		/* must be MMIO */
+		er = emulate_instruction(vcpu, kvm_run, 0, 0, 0);
+
+		if (er == EMULATE_FAIL) {
+			printk(KERN_ERR
+			 "EPT: Fail to handle EPT violation vmexit!er is %d\n",
+			 er);
+			printk(KERN_ERR "EPT: GPA: 0x%lx, GVA: 0x%lx\n",
+			 (long unsigned int)vmcs_read64(GUEST_PHYSICAL_ADDRESS),
+			 (long unsigned int)vmcs_read64(GUEST_LINEAR_ADDRESS));
+			printk(KERN_ERR "EPT: Exit qualification is 0x%lx\n",
+				(long unsigned int)exit_qualification);
+			return -ENOTSUPP;
+		} else if (er == EMULATE_DO_MMIO)
+			return 0;
+	}
+	return 1;
+}
+
 /*
  * The exit handlers return 1 if the exit was handled fully and guest execution
  * may resume.  Otherwise they set the kvm_run parameter to indicate what needs
@@ -2468,6 +2658,7 @@ static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu,
 	[EXIT_REASON_APIC_ACCESS]             = handle_apic_access,
 	[EXIT_REASON_WBINVD]                  = handle_wbinvd,
 	[EXIT_REASON_TASK_SWITCH]             = handle_task_switch,
+	[EXIT_REASON_EPT_VIOLATION]	      = handle_ept_violation,
 };
 
 static const int kvm_vmx_max_exit_handlers =
@@ -2486,6 +2677,13 @@ static int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
 	KVMTRACE_3D(VMEXIT, vcpu, exit_reason, (u32)vmcs_readl(GUEST_RIP),
 		    (u32)((u64)vmcs_readl(GUEST_RIP) >> 32), entryexit);
 
+	/* Access CR3 don't cause VMExit in paging mode, so we need
+	 * to sync with guest real CR3. */
+	if (vm_need_ept() && is_paging(vcpu)) {
+		vcpu->arch.cr3 = vmcs_readl(GUEST_CR3);
+		ept_load_pdptrs(vcpu);
+	}
+
 	if (unlikely(vmx->fail)) {
 		kvm_run->exit_reason = KVM_EXIT_FAIL_ENTRY;
 		kvm_run->fail_entry.hardware_entry_failure_reason
@@ -2494,7 +2692,8 @@ static int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
 	}
 
 	if ((vectoring_info & VECTORING_INFO_VALID_MASK) &&
-				exit_reason != EXIT_REASON_EXCEPTION_NMI)
+			(exit_reason != EXIT_REASON_EXCEPTION_NMI &&
+			exit_reason != EXIT_REASON_EPT_VIOLATION))
 		printk(KERN_WARNING "%s: unexpected, valid vectoring info and "
 		       "exit reason is 0x%x\n", __func__, exit_reason);
 	if (exit_reason < kvm_vmx_max_exit_handlers
@@ -2796,6 +2995,15 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
 		return ERR_PTR(-ENOMEM);
 
 	allocate_vpid(vmx);
+	if (id == 0 && vm_need_ept()) {
+		kvm_mmu_set_base_ptes(VMX_EPT_READABLE_MASK |
+			VMX_EPT_WRITABLE_MASK |
+			VMX_EPT_DEFAULT_MT << VMX_EPT_MT_EPTE_SHIFT);
+		kvm_mmu_set_mask_ptes(0ull, VMX_EPT_FAKE_ACCESSED_MASK,
+				VMX_EPT_FAKE_DIRTY_MASK, 0ull,
+				VMX_EPT_EXECUTABLE_MASK);
+		kvm_enable_tdp();
+	}
 
 	err = kvm_vcpu_init(&vmx->vcpu, kvm, id);
 	if (err)
@@ -2975,9 +3183,14 @@ static int __init vmx_init(void)
 	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP);
 	vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP);
 
+	if (cpu_has_vmx_ept())
+		bypass_guest_pf = 0;
+
 	if (bypass_guest_pf)
 		kvm_mmu_set_nonpresent_ptes(~0xffeull, 0ull);
 
+	ept_sync_global();
+
 	return 0;
 
 out2:

arch/x86/kvm/vmx.h

@@ -353,6 +353,15 @@ enum vmcs_field {
 #define VMX_EPT_EXTENT_CONTEXT_BIT		(1ull << 25)
 #define VMX_EPT_EXTENT_GLOBAL_BIT		(1ull << 26)
 #define VMX_EPT_DEFAULT_GAW			3
+#define VMX_EPT_MAX_GAW				0x4
+#define VMX_EPT_MT_EPTE_SHIFT			3
+#define VMX_EPT_GAW_EPTP_SHIFT			3
+#define VMX_EPT_DEFAULT_MT			0x6ull
+#define VMX_EPT_READABLE_MASK			0x1ull
+#define VMX_EPT_WRITABLE_MASK			0x2ull
+#define VMX_EPT_EXECUTABLE_MASK			0x4ull
+#define VMX_EPT_FAKE_ACCESSED_MASK		(1ull << 62)
+#define VMX_EPT_FAKE_DIRTY_MASK			(1ull << 63)
 
 #define VMX_EPT_IDENTITY_PAGETABLE_ADDR		0xfffbc000ul
 

include/asm-x86/kvm_host.h

@@ -651,6 +651,7 @@ static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
 #define ASM_VMX_VMWRITE_RSP_RDX   ".byte 0x0f, 0x79, 0xd4"
 #define ASM_VMX_VMXOFF            ".byte 0x0f, 0x01, 0xc4"
 #define ASM_VMX_VMXON_RAX         ".byte 0xf3, 0x0f, 0xc7, 0x30"
+#define ASM_VMX_INVEPT		  ".byte 0x66, 0x0f, 0x38, 0x80, 0x08"
 #define ASM_VMX_INVVPID		  ".byte 0x66, 0x0f, 0x38, 0x81, 0x08"
 
 #define MSR_IA32_TIME_STAMP_COUNTER		0x010