intel · hongyuni · Apr 18, 2022 · Apr 18, 2022 · Apr 18, 2022 · Apr 18, 2022
diff --git a/KVM-TDX.README.md b/KVM-TDX.README.md
@@ -0,0 +1,172 @@
+repositories and branches
+=========================
+Linux
+-----
+git repository:
+https://github.com/intel/tdx
+
+branchs:
+kvm-upstream-workaround branch is recommended.
+- kvm-upstream: includes patches only posted to the linux and kvm community.
+- kvm-upstream-workaround: includes kvm-upstream and further patches.
+If you tests TDX KVM, kvm-upstream-workaround branch is recommended.
+
+Configurations:
+- CONFIG_INTEL_TDX_HOST=y: enable TDX KVM support.
+- CONFIG_INTEL_TDX_HOST_DEBUG: debug support for TDX module. optional. say N.
+- CONFIG_INTEL_TDX_GUEST: guest TD support.  Optional. Not needed for TDX KVM.
+- CONFIG_X86_TDX_KVM_EXPERIMENTAL=y: Enable experimental TDX KVM support. Optional, needed for https://github.com/intel/tdx/releases/tag/kvm-upstream-2022.03.29-v5.17-rc8-workaround. TDX KVM needs many patches and the patches will be merged step by step, not at once. Set it to y to enable TDX KVM support so that developper can exercise TDX KVM code.
+
+kernel command line:
+- tdx_host=on
+  Enables TDX features in host kernel. Because it's off by default, it needs to be
+  explicitly enabled.
+- disable_mtrr_cleanup
+
+Qemu
+----
+git repository:
+https://github.com/intel/qemu-tdx
+
+branches:
+tdx-upstream-wip branch is recommended.
+- tdx: includes full features.  It's being deprecated in favor for tdx-upstream-wip.
+- tdx-upstream: includes patches only posted to the qemu community
+- tdx-upstream-wip: includes tdx-upstream and further patches.
+
+build:
+build x86_64-softmmu target.
+./configure --target-list=x86_64-softmmu
+
+kvm-unit-tests
+--------------
+git repository:
+https://github.com/intel/kvm-unit-tests-tdx
+
+branches:
+- tdx: include test cases for TDX.
+
+TDVF
+----
+git repository:
+https://github.com/tianocore/edk2-staging/tree/TDVF
+
+Please refer its documentation for further information.
+https://github.com/tianocore/edk2-staging/blob/TDVF/README.md
+
+Guest
+-----
+git repository:
+https://github.com/intel/tdx
+
+branches:
+- guest-upstream: guest TD support for upstreaming
+- guest: guest TD support
+
+Configurations:
+- CONFIG_INTEL_TDX_GUEST=y
+
+Unit Testing
+============
+selftests and kvm-unit-tests are available for TDX KVM.
+
+selftests
+---------
+linux/tools/testing/selftests/kvm/x86_64/tdx_vm_tests
+
+kvm-unit-tests
+--------------
+For running unit tests with TDX enabled, refer to
+https://github.com/intel/kvm-unit-tests-tdx/blob/tdx/README.md#unit-test-in-tdx-environment
+
+
+Running guest TD
+================
+qemu tdx-upstream-wip branch
+----------------------------
+- create tdx-guest object.
+  -object tdx-guest,id=tdx0,debug=off,sept-ve-disable=on \
+  -machine confidential-guest-support=tdx0
+- specify q35 chipset and KVM
+  -machine q35,accel=kvm
+- specify TDVF
+  -bios ${OVMF}
+- specify split irqchip, disable PIC and PIT
+  -machine kernel-irqchip=split,pic=off,pit=off
+
+command line example:
+SMP=8
+MEM=512M
+KERNEL=/path/to/guest-kernel
+INITRD=/path/to/guest-initrd
+APPEND="console=hvc0 nomce no-kvmclock no-steal-acc no_console_suspend"
+DRIVE_DISC=/path/to/disk-image
+OVMF=/path/to/OVMF.fd
+qemu-system-x86_64 \
+    -s -m ${MEM} -smp ${SMP},sockets=1 \
+    -cpu host,host-phys-bits,pmu=off,pks=on \
+    -no-hpet -nographic -vga none \
+    -nodefaults \
+    -monitor stdio \
+    -object tdx-guest,id=tdx0,debug=off,sept-ve-disable=on \
+    -machine confidential-guest-support=tdx0 \
+    -machine q35,accel=kvm \
+    -machine kernel-irqchip=split,sata=off,pic=off,pit=off \
+    -bios ${OVMF} \
+    -device virtio-serial \
+    -chardev socket,id=tcp0,port=4445,host=0.0.0.0,server=on,wait=off \
+    -device virtconsole,chardev=tcp0 \
+    -kernel ${KERNEL} \
+    -initrd ${INITRD} \
+    -append "${APPEND}" \
+    -drive file=${DRIVE_DISK},if=virtio,format=qcow2,media=disk,index=0 \
+    -device vhost-vsock-pci,guest-cid=3
+
+
+qemu tdx branch
+---------------
+- create tdx-guest object.
+  -object tdx-guest,id=tdx0,debug=off,sept-ve-disable=on \
+  -machine confidential-guest-support=tdx0
+- specify q35 chipset, KVM and tdx kvm vm type.
+  -machine q35,accel=kvm,kvm-type=tdx
+- specify TDVF
+  -device loader,file=${OVMF}
+  or
+  -device loader,file=${OVMF_CODE},config-firmware-volume=${OVMF_VARS},id=fd0
+- specify split irqchip, disable PIC and PIT
+  -machine kernel-irqchip=split,pic=off,pit=off
+- UPM (Unmapping Process Memory): create memfd-private backend
+  -object memory-backend-memfd-private,id=ram1,size=${MEM} \
+  -machine memory-backend=ram1
+
+command line example:
+SMP=8
+MEM=512M
+KERNEL=/path/to/guest-kernel
+INITRD=/path/to/guest-initrd
+APPEND="console=hvc0 nomce no-kvmclock no-steal-acc no_console_suspend"
+DRIVE_DISC=/path/to/disk-image
+OVMF_CODE=/path/to/OVMF_CODE.fd
+OVMF_VARS=/path/to/OVMF_VARS.fd
+qemu-system-x86_64 \
+     -s -m ${MEM} -smp ${SMP},sockets=1 \
+     -cpu host,host-phys-bits,pmu=off,pks=on \
+     -no-hpet -nographic -vga none \
+     -nodefaults \
+     -monitor stdio \
+     -object tdx-guest,id=tdx0,debug=off,sept-ve-disable=on \
+     -machine confidential-guest-support=tdx0 \
+     -machine q35,accel=kvm,kvm-type=tdx \
+     -machine kernel-irqchip=split,sata=off,pic=off,pit=off \
+     -object memory-backend-memfd-private,id=ram1,size=${MEM} \
+     -machine memory-backend=ram1 \
+     -device loader,file=${OVMF_CODE},config-firmware-volume=${OVMF_VARS},id=fd0 \
+     -kernel ${KERNEL} \
+     -initrd ${INITRD} \
+     -append "${APPEND}" \
+     -device virtio-serial \
+     -chardev socket,id=tcp0,port=4445,host=0.0.0.0,server=on,wait=off \
+     -device virtconsole,chardev=tcp0 \
+     -drive file=${DRIVE_DISK},if=virtio,format=qcow2,media=disk,index=0 \
+     -device vhost-vsock-pci,guest-cid=3
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
@@ -1444,7 +1444,8 @@ struct kvm_x86_ops {
 	void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
 	void (*enable_irq_window)(struct kvm_vcpu *vcpu);
 	void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
-	bool (*check_apicv_inhibit_reasons)(enum kvm_apicv_inhibit reason);
+	bool (*check_apicv_inhibit_reasons)(struct kvm *kvm,
+					    enum kvm_apicv_inhibit reason);
 	void (*refresh_apicv_exec_ctrl)(struct kvm_vcpu *vcpu);
 	void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
 	void (*hwapic_isr_update)(struct kvm_vcpu *vcpu, int isr);
@@ -1678,7 +1679,7 @@ void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned long kvm_nr_mmu_pages);
 
 #ifdef CONFIG_INTEL_TDX_HOST
 int kvm_mmu_is_page_private(struct kvm *kvm, struct kvm_memory_slot *memslot,
-			    gfn_t gfn, bool *is_private);
+			    gfn_t gfn, bool *is_private, kvm_pfn_t *pfn);
 #endif
 
 int load_pdptrs(struct kvm_vcpu *vcpu, unsigned long cr3);

diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
@@ -270,10 +270,6 @@ static inline u8 permission_fault(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
 }
 
 void kvm_zap_gfn_range(struct kvm *kvm, gfn_t gfn_start, gfn_t gfn_end);
-#ifdef CONFIG_INTEL_TDX_HOST
-int kvm_tdp_mmu_is_page_private(struct kvm *kvm, struct kvm_memory_slot *memslot,
-				gfn_t gfn, bool *is_private);
-#endif
 
 int kvm_arch_write_log_dirty(struct kvm_vcpu *vcpu);
 

diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
@@ -7251,13 +7251,25 @@ void kvm_mmu_pre_destroy_vm(struct kvm *kvm)
  */
 int kvm_mmu_is_page_private(struct kvm *kvm,
 			    struct kvm_memory_slot *memslot,
-			    gfn_t gfn, bool *is_private)
+			    gfn_t gfn, bool *is_private, kvm_pfn_t *pfn)
 {
 	struct slot_rmap_walk_iterator s_iter;
 	struct rmap_iterator r_iter;
 	u64 *sptep;
 	int ret = -EINVAL;
 
+	if (kvm_slot_is_private(memslot)) {
+		int order;
+
+		/* The caller must call kvm_memfile_put_pfn(memslot, pfn); */
+		*pfn = kvm_memfile_get_pfn(memslot, gfn, &order);
+		*is_private = VALID_PAGE(*pfn);
+		return 0;
+	}
+
+	if (is_tdp_mmu_enabled(kvm))
+		return kvm_tdp_mmu_is_page_private(kvm, memslot, gfn, is_private);
+
 	for_each_slot_rmap_range(memslot, PG_LEVEL_4K,
 				 KVM_MAX_HUGEPAGE_LEVEL,
 				 gfn, gfn, &s_iter) {

diff --git a/arch/x86/kvm/mmu/tdp_mmu.h b/arch/x86/kvm/mmu/tdp_mmu.h
@@ -99,4 +99,16 @@ static inline bool is_tdp_mmu_page(struct kvm_mmu_page *sp) { return false; }
 static inline bool is_tdp_mmu(struct kvm_mmu *mmu) { return false; }
 #endif
 
+#ifdef CONFIG_INTEL_TDX_HOST
+int kvm_tdp_mmu_is_page_private(struct kvm *kvm, struct kvm_memory_slot *memslot,
+				gfn_t gfn, bool *is_private);
+#else
+static inline int kvm_tdp_mmu_is_page_private(
+	struct kvm *kvm, struct kvm_memory_slot *memslot, gfn_t gfn,
+	bool *is_private)
+{
+	return -EOPNOTSUPP;
+}
+#endif
+
 #endif /* __KVM_X86_MMU_TDP_MMU_H */
diff --git a/arch/x86/kvm/svm/avic.c b/arch/x86/kvm/svm/avic.c
@@ -828,7 +828,8 @@ int avic_pi_update_irte(struct kvm *kvm, unsigned int host_irq,
 	return ret;
 }
 
-bool avic_check_apicv_inhibit_reasons(enum kvm_apicv_inhibit reason)
+bool avic_check_apicv_inhibit_reasons(struct kvm *kvm,
+				      enum kvm_apicv_inhibit reason)
 {
 	ulong supported = BIT(APICV_INHIBIT_REASON_DISABLE) |
 			  BIT(APICV_INHIBIT_REASON_ABSENT) |

diff --git a/arch/x86/kvm/svm/svm.h b/arch/x86/kvm/svm/svm.h
@@ -583,7 +583,8 @@ void __avic_vcpu_put(struct kvm_vcpu *vcpu);
 void avic_apicv_post_state_restore(struct kvm_vcpu *vcpu);
 void avic_set_virtual_apic_mode(struct kvm_vcpu *vcpu);
 void avic_refresh_apicv_exec_ctrl(struct kvm_vcpu *vcpu);
-bool avic_check_apicv_inhibit_reasons(enum kvm_apicv_inhibit reason);
+bool avic_check_apicv_inhibit_reasons(struct kvm *kvm,
+				      enum kvm_apicv_inhibit reason);
 void avic_hwapic_irr_update(struct kvm_vcpu *vcpu, int max_irr);
 void avic_hwapic_isr_update(struct kvm_vcpu *vcpu, int max_isr);
 bool avic_dy_apicv_has_pending_interrupt(struct kvm_vcpu *vcpu);

diff --git a/arch/x86/kvm/vmx/main.c b/arch/x86/kvm/vmx/main.c
@@ -348,7 +348,7 @@ static void vt_vcpu_after_set_cpuid(struct kvm_vcpu *vcpu)
 static void vt_update_exception_bitmap(struct kvm_vcpu *vcpu)
 {
 	if (is_td_vcpu(vcpu))
-		return;
+		return tdx_update_exception_bitmap(vcpu);
 
 	vmx_update_exception_bitmap(vcpu);
 }
@@ -454,7 +454,7 @@ static void vt_set_gdt(struct kvm_vcpu *vcpu, struct desc_ptr *dt)
 static void vt_set_dr7(struct kvm_vcpu *vcpu, unsigned long val)
 {
 	if (is_td_vcpu(vcpu))
-		return;
+		return tdx_set_dr7(vcpu, val);
 
 	vmx_set_dr7(vcpu, val);
 }
@@ -843,6 +843,15 @@ static int vt_skip_emulated_instruction(struct kvm_vcpu *vcpu)
 	return vmx_skip_emulated_instruction(vcpu);
 }
 
+static bool vt_check_apicv_inhibit_reasons(struct kvm *kvm,
+					   enum kvm_apicv_inhibit reason)
+{
+	if (is_td(kvm))
+		return tdx_check_apicv_inhibit_reasons(kvm, reason);
+
+	return vmx_check_apicv_inhibit_reasons(kvm, reason);
+}
+
 struct kvm_x86_ops vt_x86_ops __initdata = {
 	.name = "kvm_intel",
 
@@ -920,7 +929,7 @@ struct kvm_x86_ops vt_x86_ops __initdata = {
 	.refresh_apicv_exec_ctrl = vt_refresh_apicv_exec_ctrl,
 	.load_eoi_exitmap = vt_load_eoi_exitmap,
 	.apicv_post_state_restore = vt_apicv_post_state_restore,
-	.check_apicv_inhibit_reasons = vmx_check_apicv_inhibit_reasons,
+	.check_apicv_inhibit_reasons = vt_check_apicv_inhibit_reasons,
 	.hwapic_irr_update = vt_hwapic_irr_update,
 	.hwapic_isr_update = vt_hwapic_isr_update,
 	.guest_apic_has_interrupt = vt_guest_apic_has_interrupt,