KVM: X86: Introduce KVM_HC_MAP_GPA_RANGE hypercall

This hypercall is used by the SEV guest to notify a change in the page encryption status to the hypervisor. The hypercall should be invoked only when the encryption attribute is changed from encrypted -> decrypted and vice versa. By default all guest pages are considered encrypted. The hypercall exits to userspace to manage the guest shared regions and integrate with the userspace VMM's migration code.

Ashish Kalra committed 5 years ago

Commit 0dbb1121 parent e3cb6fa9 files · +123 −0Browse files

Unified · Split

9 files

Documentation/virt/kvm/api.rst +19 −0
Documentation/virt/kvm/cpuid.rst +7 −0
Documentation/virt/kvm/hypercalls.rst +21 −0
Documentation/virt/kvm/msr.rst +13 −0
arch/x86/include/asm/kvm_host.h +2 −0
arch/x86/include/uapi/asm/kvm_para.h +13 −0
arch/x86/kvm/x86.c +46 −0
include/uapi/linux/kvm.h +1 −0
include/uapi/linux/kvm_para.h +1 −0

Documentation/virt/kvm/api.rst+19−0

@@ -6950,3 +6950,22 @@ guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all · 6949 unmodified lines
6950	6950	currently implmented Hyper-V features are provided unconditionally when
6951	6951	Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
6952	6952	leaf.
	6953	+
	6954	+8.34 KVM_CAP_EXIT_HYPERCALL
	6955	+---------------------------
	6956	+
	6957	+:Capability: KVM_CAP_EXIT_HYPERCALL
	6958	+:Architectures: x86
	6959	+:Type: vm
	6960	+
	6961	+This capability, if enabled, will cause KVM to exit to userspace
	6962	+with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
	6963	+
	6964	+Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
	6965	+of hypercalls that can be configured to exit to userspace.
	6966	+Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
	6967	+
	6968	+The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
	6969	+of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace
	6970	+the hypercalls whose corresponding bit is in the argument, and return
	6971	+ENOSYS for the others.

Documentation/virt/kvm/cpuid.rst+7−0

@@ -96,6 +96,13 @@ KVM_FEATURE_MSI_EXT_DEST_ID 15 guest checks this feature bit · 95 unmodified lines
96	96	before using extended destination
97	97	ID bits in MSI address bits 11-5.
98	98
	99	+KVM_FEATURE_HC_MAP_GPA_RANGE 16 guest checks this feature bit before
	100	+ using the map gpa range hypercall
	101	+ to notify the page state change
	102	+
	103	+KVM_FEATURE_MIGRATION_CONTROL 17 guest checks this feature bit before
	104	+ using MSR_KVM_MIGRATION_CONTROL
	105	+
99	106	KVM_FEATURE_CLOCKSOURCE_STABLE_BIT 24 host will warn if no guest-side
100	107	per-cpu warps are expected in
101	108	kvmclock

Documentation/virt/kvm/hypercalls.rst+21−0

@@ -169,3 +169,24 @@ a0: destination APIC ID · 168 unmodified lines
169	169
170	170	:Usage example: When sending a call-function IPI-many to vCPUs, yield if
171	171	any of the IPI target vCPUs was preempted.
	172	+
	173	+8. KVM_HC_MAP_GPA_RANGE
	174	+-------------------------
	175	+:Architecture: x86
	176	+:Status: active
	177	+:Purpose: Request KVM to map a GPA range with the specified attributes.
	178	+
	179	+a0: the guest physical address of the start page
	180	+a1: the number of (4kb) pages (must be contiguous in GPA space)
	181	+a2: attributes
	182	+
	183	+ Where 'attributes' :
	184	+ * bits 3:0 - preferred page size encoding 0 = 4kb, 1 = 2mb, 2 = 1gb, etc...
	185	+ * bit 4 - plaintext = 0, encrypted = 1
	186	+ * bits 63:5 - reserved (must be zero)
	187	+
	188	+Implementation note: this hypercall is implemented in userspace via
	189	+the KVM_CAP_EXIT_HYPERCALL capability. Userspace must enable that capability
	190	+before advertising KVM_FEATURE_HC_MAP_GPA_RANGE in the guest CPUID. In
	191	+addition, if the guest supports KVM_FEATURE_MIGRATION_CONTROL, userspace
	192	+must also set up an MSR filter to process writes to MSR_KVM_MIGRATION_CONTROL.

Documentation/virt/kvm/msr.rst+13−0

@@ -376,3 +376,16 @@ data: · 375 unmodified lines
376	376	write '1' to bit 0 of the MSR, this causes the host to re-scan its queue
377	377	and check if there are more notifications pending. The MSR is available
378	378	if KVM_FEATURE_ASYNC_PF_INT is present in CPUID.
	379	+
	380	+MSR_KVM_MIGRATION_CONTROL:
	381	+ 0x4b564d08
	382	+
	383	+data:
	384	+ This MSR is available if KVM_FEATURE_MIGRATION_CONTROL is present in
	385	+ CPUID. Bit 0 represents whether live migration of the guest is allowed.
	386	+
	387	+ When a guest is started, bit 0 will be 0 if the guest has encrypted
	388	+ memory and 1 if the guest does not have encrypted memory. If the
	389	+ guest is communicating page encryption status to the host using the
	390	+ ``KVM_HC_MAP_GPA_RANGE`` hypercall, it can set bit 0 in this MSR to
	391	+ allow live migration of the guest.

arch/x86/include/asm/kvm_host.h+2−0

@@ -1087,6 +1087,8 @@ struct kvm_arch { · 1086 unmodified lines
1087	1087	u32 user_space_msr_mask;
1088	1088	struct kvm_x86_msr_filter __rcu *msr_filter;
1089	1089
	1090	+ u32 hypercall_exit_enabled;
	1091	+
1090	1092	/* Guest can access the SGX PROVISIONKEY. */
1091	1093	bool sgx_provisioning_allowed;
1092	1094

arch/x86/include/uapi/asm/kvm_para.h+13−0

@@ -33,6 +33,8 @@ · 32 unmodified lines
33	33	#define KVM_FEATURE_PV_SCHED_YIELD 13
34	34	#define KVM_FEATURE_ASYNC_PF_INT 14
35	35	#define KVM_FEATURE_MSI_EXT_DEST_ID 15
	36	+#define KVM_FEATURE_HC_MAP_GPA_RANGE 16
	37	+#define KVM_FEATURE_MIGRATION_CONTROL 17
36	38
37	39	#define KVM_HINTS_REALTIME 0
38	40
@@ -54,6 +56,7 @@ · 15 unmodified lines
54	56	#define MSR_KVM_POLL_CONTROL 0x4b564d05
55	57	#define MSR_KVM_ASYNC_PF_INT 0x4b564d06
56	58	#define MSR_KVM_ASYNC_PF_ACK 0x4b564d07
	59	+#define MSR_KVM_MIGRATION_CONTROL 0x4b564d08
57	60
58	61	struct kvm_steal_time {
59	62	__u64 steal;
@@ -90,6 +93,16 @@ struct kvm_clock_pairing { · 30 unmodified lines
90	93	/* MSR_KVM_ASYNC_PF_INT */
91	94	#define KVM_ASYNC_PF_VEC_MASK GENMASK(7, 0)
92	95
	96	+/* MSR_KVM_MIGRATION_CONTROL */
	97	+#define KVM_MIGRATION_READY (1 << 0)
	98	+
	99	+/* KVM_HC_MAP_GPA_RANGE */
	100	+#define KVM_MAP_GPA_RANGE_PAGE_SZ_4K 0
	101	+#define KVM_MAP_GPA_RANGE_PAGE_SZ_2M (1 << 0)
	102	+#define KVM_MAP_GPA_RANGE_PAGE_SZ_1G (1 << 1)
	103	+#define KVM_MAP_GPA_RANGE_ENC_STAT(n) (n << 4)
	104	+#define KVM_MAP_GPA_RANGE_ENCRYPTED KVM_MAP_GPA_RANGE_ENC_STAT(1)
	105	+#define KVM_MAP_GPA_RANGE_DECRYPTED KVM_MAP_GPA_RANGE_ENC_STAT(0)
93	106
94	107	/* Operations for KVM_HC_MMU_OP */
95	108	#define KVM_MMU_OP_WRITE_PTE 1

arch/x86/kvm/x86.c+46−0

@@ -103,6 +103,8 @@ static u64 __read_mostly efer_reserved_bits = ~((u64)EFER_SCE); · 102 unmodified lines
103	103
104	104	static u64 __read_mostly cr4_reserved_bits = CR4_RESERVED_BITS;
105	105
	106	+#define KVM_EXIT_HYPERCALL_VALID_MASK (1 << KVM_HC_MAP_GPA_RANGE)
	107	+
106	108	#define KVM_X2APIC_API_VALID_FLAGS (KVM_X2APIC_API_USE_32BIT_IDS \| \
107	109	KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK)
108	110
@@ -3996,6 +3998,9 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) · 3887 unmodified lines
3996	3998	case KVM_CAP_SREGS2:
3997	3999	r = 1;
3998	4000	break;
	4001	+ case KVM_CAP_EXIT_HYPERCALL:
	4002	+ r = KVM_EXIT_HYPERCALL_VALID_MASK;
	4003	+ break;
3999	4004	case KVM_CAP_SET_GUEST_DEBUG2:
4000	4005	return KVM_GUESTDBG_VALID_MASK;
4001	4006	#ifdef CONFIG_KVM_XEN
@@ -5622,6 +5627,14 @@ split_irqchip_unlock: · 1620 unmodified lines
5622	5627	if (kvm_x86_ops.vm_copy_enc_context_from)
5623	5628	r = kvm_x86_ops.vm_copy_enc_context_from(kvm, cap->args[0]);
5624	5629	return r;
	5630	+ case KVM_CAP_EXIT_HYPERCALL:
	5631	+ if (cap->args[0] & ~KVM_EXIT_HYPERCALL_VALID_MASK) {
	5632	+ r = -EINVAL;
	5633	+ break;
	5634	+ }
	5635	+ kvm->arch.hypercall_exit_enabled = cap->args[0];
	5636	+ r = 0;
	5637	+ break;
5625	5638	default:
5626	5639	r = -EINVAL;
5627	5640	break;
@@ -8548,6 +8561,17 @@ no_yield: · 2920 unmodified lines
8548	8561	return;
8549	8562	}
8550	8563
	8564	+static int complete_hypercall_exit(struct kvm_vcpu *vcpu)
	8565	+{
	8566	+ u64 ret = vcpu->run->hypercall.ret;
	8567	+
	8568	+ if (!is_64_bit_mode(vcpu))
	8569	+ ret = (u32)ret;
	8570	+ kvm_rax_write(vcpu, ret);
	8571	+ ++vcpu->stat.hypercalls;
	8572	+ return kvm_skip_emulated_instruction(vcpu);
	8573	+}
	8574	+
8551	8575	int kvm_emulate_hypercall(struct kvm_vcpu *vcpu)
8552	8576	{
8553	8577	unsigned long nr, a0, a1, a2, a3, ret;
@@ -8613,6 +8637,28 @@ int kvm_emulate_hypercall(struct kvm_vcpu *vcpu) · 59 unmodified lines
8613	8637	kvm_sched_yield(vcpu, a0);
8614	8638	ret = 0;
8615	8639	break;
	8640	+ case KVM_HC_MAP_GPA_RANGE: {
	8641	+ u64 gpa = a0, npages = a1, attrs = a2;
	8642	+
	8643	+ ret = -KVM_ENOSYS;
	8644	+ if (!(vcpu->kvm->arch.hypercall_exit_enabled & (1 << KVM_HC_MAP_GPA_RANGE)))
	8645	+ break;
	8646	+
	8647	+ if (!PAGE_ALIGNED(gpa) \|\| !npages \|\|
	8648	+ gpa_to_gfn(gpa) + npages <= gpa_to_gfn(gpa)) {
	8649	+ ret = -KVM_EINVAL;
	8650	+ break;
	8651	+ }
	8652	+
	8653	+ vcpu->run->exit_reason = KVM_EXIT_HYPERCALL;
	8654	+ vcpu->run->hypercall.nr = KVM_HC_MAP_GPA_RANGE;
	8655	+ vcpu->run->hypercall.args[0] = gpa;
	8656	+ vcpu->run->hypercall.args[1] = npages;
	8657	+ vcpu->run->hypercall.args[2] = attrs;
	8658	+ vcpu->run->hypercall.longmode = op_64_bit;
	8659	+ vcpu->arch.complete_userspace_io = complete_hypercall_exit;
	8660	+ return 0;
	8661	+ }
8616	8662	default:
8617	8663	ret = -KVM_ENOSYS;
8618	8664	break;

include/uapi/linux/kvm.h+1−0

@@ -1085,6 +1085,7 @@ struct kvm_ppc_resize_hpt { · 1084 unmodified lines
1085	1085	#define KVM_CAP_PTP_KVM 198
1086	1086	#define KVM_CAP_HYPERV_ENFORCE_CPUID 199
1087	1087	#define KVM_CAP_SREGS2 200
	1088	+#define KVM_CAP_EXIT_HYPERCALL 201
1088	1089
1089	1090	#ifdef KVM_CAP_IRQ_ROUTING
1090	1091

include/uapi/linux/kvm_para.h+1−0

@@ -29,6 +29,7 @@ · 28 unmodified lines
29	29	#define KVM_HC_CLOCK_PAIRING 9
30	30	#define KVM_HC_SEND_IPI 10
31	31	#define KVM_HC_SCHED_YIELD 11
	32	+#define KVM_HC_MAP_GPA_RANGE 12
32	33
33	34	/*
34	35	* hypercalls use architecture specific