xref: /freebsd/sys/amd64/include/vmm.h (revision ab0b9f6b3073e6c4d1dfbf07444d7db67a189a96)
1 /*-
2  * Copyright (c) 2011 NetApp, Inc.
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  *
14  * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND
15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17  * ARE DISCLAIMED.  IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE
18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24  * SUCH DAMAGE.
25  *
26  * $FreeBSD$
27  */
28 
29 #ifndef _VMM_H_
30 #define	_VMM_H_
31 
32 #ifdef _KERNEL
33 
34 #define	VM_MAX_NAMELEN	32
35 
36 struct vm;
37 struct vm_memory_segment;
38 struct seg_desc;
39 struct vm_exit;
40 struct vm_run;
41 struct vhpet;
42 struct vioapic;
43 struct vlapic;
44 struct vmspace;
45 struct vm_object;
46 struct pmap;
47 
48 enum x2apic_state;
49 
50 typedef int	(*vmm_init_func_t)(void);
51 typedef int	(*vmm_cleanup_func_t)(void);
52 typedef void	(*vmm_resume_func_t)(void);
53 typedef void *	(*vmi_init_func_t)(struct vm *vm, struct pmap *pmap);
54 typedef int	(*vmi_run_func_t)(void *vmi, int vcpu, register_t rip,
55 				  struct pmap *pmap);
56 typedef void	(*vmi_cleanup_func_t)(void *vmi);
57 typedef int	(*vmi_get_register_t)(void *vmi, int vcpu, int num,
58 				      uint64_t *retval);
59 typedef int	(*vmi_set_register_t)(void *vmi, int vcpu, int num,
60 				      uint64_t val);
61 typedef int	(*vmi_get_desc_t)(void *vmi, int vcpu, int num,
62 				  struct seg_desc *desc);
63 typedef int	(*vmi_set_desc_t)(void *vmi, int vcpu, int num,
64 				  struct seg_desc *desc);
65 typedef int	(*vmi_inject_event_t)(void *vmi, int vcpu,
66 				      int type, int vector,
67 				      uint32_t code, int code_valid);
68 typedef int	(*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval);
69 typedef int	(*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val);
70 typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max);
71 typedef void	(*vmi_vmspace_free)(struct vmspace *vmspace);
72 typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu);
73 typedef void	(*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic);
74 
75 struct vmm_ops {
76 	vmm_init_func_t		init;		/* module wide initialization */
77 	vmm_cleanup_func_t	cleanup;
78 	vmm_resume_func_t	resume;
79 
80 	vmi_init_func_t		vminit;		/* vm-specific initialization */
81 	vmi_run_func_t		vmrun;
82 	vmi_cleanup_func_t	vmcleanup;
83 	vmi_get_register_t	vmgetreg;
84 	vmi_set_register_t	vmsetreg;
85 	vmi_get_desc_t		vmgetdesc;
86 	vmi_set_desc_t		vmsetdesc;
87 	vmi_inject_event_t	vminject;
88 	vmi_get_cap_t		vmgetcap;
89 	vmi_set_cap_t		vmsetcap;
90 	vmi_vmspace_alloc	vmspace_alloc;
91 	vmi_vmspace_free	vmspace_free;
92 	vmi_vlapic_init		vlapic_init;
93 	vmi_vlapic_cleanup	vlapic_cleanup;
94 };
95 
96 extern struct vmm_ops vmm_ops_intel;
97 extern struct vmm_ops vmm_ops_amd;
98 
99 int vm_create(const char *name, struct vm **retvm);
100 void vm_destroy(struct vm *vm);
101 const char *vm_name(struct vm *vm);
102 int vm_malloc(struct vm *vm, vm_paddr_t gpa, size_t len);
103 int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa);
104 int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len);
105 void *vm_gpa_hold(struct vm *, vm_paddr_t gpa, size_t len, int prot,
106 		  void **cookie);
107 void vm_gpa_release(void *cookie);
108 int vm_gpabase2memseg(struct vm *vm, vm_paddr_t gpabase,
109 	      struct vm_memory_segment *seg);
110 int vm_get_memobj(struct vm *vm, vm_paddr_t gpa, size_t len,
111 		  vm_offset_t *offset, struct vm_object **object);
112 boolean_t vm_mem_allocated(struct vm *vm, vm_paddr_t gpa);
113 int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval);
114 int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val);
115 int vm_get_seg_desc(struct vm *vm, int vcpu, int reg,
116 		    struct seg_desc *ret_desc);
117 int vm_set_seg_desc(struct vm *vm, int vcpu, int reg,
118 		    struct seg_desc *desc);
119 int vm_run(struct vm *vm, struct vm_run *vmrun);
120 int vm_inject_event(struct vm *vm, int vcpu, int type,
121 		    int vector, uint32_t error_code, int error_code_valid);
122 int vm_inject_nmi(struct vm *vm, int vcpu);
123 int vm_nmi_pending(struct vm *vm, int vcpuid);
124 void vm_nmi_clear(struct vm *vm, int vcpuid);
125 uint64_t *vm_guest_msrs(struct vm *vm, int cpu);
126 struct vlapic *vm_lapic(struct vm *vm, int cpu);
127 struct vioapic *vm_ioapic(struct vm *vm);
128 struct vhpet *vm_hpet(struct vm *vm);
129 int vm_get_capability(struct vm *vm, int vcpu, int type, int *val);
130 int vm_set_capability(struct vm *vm, int vcpu, int type, int val);
131 int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state);
132 int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state);
133 int vm_apicid2vcpuid(struct vm *vm, int apicid);
134 void vm_activate_cpu(struct vm *vm, int vcpu);
135 cpuset_t vm_active_cpus(struct vm *vm);
136 struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid);
137 
138 /*
139  * Return 1 if device indicated by bus/slot/func is supposed to be a
140  * pci passthrough device.
141  *
142  * Return 0 otherwise.
143  */
144 int vmm_is_pptdev(int bus, int slot, int func);
145 
146 void *vm_iommu_domain(struct vm *vm);
147 
148 enum vcpu_state {
149 	VCPU_IDLE,
150 	VCPU_FROZEN,
151 	VCPU_RUNNING,
152 	VCPU_SLEEPING,
153 };
154 
155 int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state,
156     bool from_idle);
157 enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu);
158 
159 static int __inline
160 vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu)
161 {
162 	return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING);
163 }
164 
165 void *vcpu_stats(struct vm *vm, int vcpu);
166 void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr);
167 struct vmspace *vm_get_vmspace(struct vm *vm);
168 int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
169 int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
170 #endif	/* KERNEL */
171 
172 #include <machine/vmm_instruction_emul.h>
173 
174 #define	VM_MAXCPU	16			/* maximum virtual cpus */
175 
176 /*
177  * Identifiers for events that can be injected into the VM
178  */
179 enum vm_event_type {
180 	VM_EVENT_NONE,
181 	VM_HW_INTR,
182 	VM_NMI,
183 	VM_HW_EXCEPTION,
184 	VM_SW_INTR,
185 	VM_PRIV_SW_EXCEPTION,
186 	VM_SW_EXCEPTION,
187 	VM_EVENT_MAX
188 };
189 
190 /*
191  * Identifiers for architecturally defined registers.
192  */
193 enum vm_reg_name {
194 	VM_REG_GUEST_RAX,
195 	VM_REG_GUEST_RBX,
196 	VM_REG_GUEST_RCX,
197 	VM_REG_GUEST_RDX,
198 	VM_REG_GUEST_RSI,
199 	VM_REG_GUEST_RDI,
200 	VM_REG_GUEST_RBP,
201 	VM_REG_GUEST_R8,
202 	VM_REG_GUEST_R9,
203 	VM_REG_GUEST_R10,
204 	VM_REG_GUEST_R11,
205 	VM_REG_GUEST_R12,
206 	VM_REG_GUEST_R13,
207 	VM_REG_GUEST_R14,
208 	VM_REG_GUEST_R15,
209 	VM_REG_GUEST_CR0,
210 	VM_REG_GUEST_CR3,
211 	VM_REG_GUEST_CR4,
212 	VM_REG_GUEST_DR7,
213 	VM_REG_GUEST_RSP,
214 	VM_REG_GUEST_RIP,
215 	VM_REG_GUEST_RFLAGS,
216 	VM_REG_GUEST_ES,
217 	VM_REG_GUEST_CS,
218 	VM_REG_GUEST_SS,
219 	VM_REG_GUEST_DS,
220 	VM_REG_GUEST_FS,
221 	VM_REG_GUEST_GS,
222 	VM_REG_GUEST_LDTR,
223 	VM_REG_GUEST_TR,
224 	VM_REG_GUEST_IDTR,
225 	VM_REG_GUEST_GDTR,
226 	VM_REG_GUEST_EFER,
227 	VM_REG_LAST
228 };
229 
230 /*
231  * Identifiers for optional vmm capabilities
232  */
233 enum vm_cap_type {
234 	VM_CAP_HALT_EXIT,
235 	VM_CAP_MTRAP_EXIT,
236 	VM_CAP_PAUSE_EXIT,
237 	VM_CAP_UNRESTRICTED_GUEST,
238 	VM_CAP_ENABLE_INVPCID,
239 	VM_CAP_MAX
240 };
241 
242 enum x2apic_state {
243 	X2APIC_ENABLED,
244 	X2APIC_AVAILABLE,
245 	X2APIC_DISABLED,
246 	X2APIC_STATE_LAST
247 };
248 
249 /*
250  * The 'access' field has the format specified in Table 21-2 of the Intel
251  * Architecture Manual vol 3b.
252  *
253  * XXX The contents of the 'access' field are architecturally defined except
254  * bit 16 - Segment Unusable.
255  */
256 struct seg_desc {
257 	uint64_t	base;
258 	uint32_t	limit;
259 	uint32_t	access;
260 };
261 
262 enum vm_exitcode {
263 	VM_EXITCODE_INOUT,
264 	VM_EXITCODE_VMX,
265 	VM_EXITCODE_BOGUS,
266 	VM_EXITCODE_RDMSR,
267 	VM_EXITCODE_WRMSR,
268 	VM_EXITCODE_HLT,
269 	VM_EXITCODE_MTRAP,
270 	VM_EXITCODE_PAUSE,
271 	VM_EXITCODE_PAGING,
272 	VM_EXITCODE_INST_EMUL,
273 	VM_EXITCODE_SPINUP_AP,
274 	VM_EXITCODE_SPINDOWN_CPU,
275 	VM_EXITCODE_MAX
276 };
277 
278 struct vm_exit {
279 	enum vm_exitcode	exitcode;
280 	int			inst_length;	/* 0 means unknown */
281 	uint64_t		rip;
282 	union {
283 		struct {
284 			uint16_t	bytes:3;	/* 1 or 2 or 4 */
285 			uint16_t	in:1;		/* out is 0, in is 1 */
286 			uint16_t	string:1;
287 			uint16_t	rep:1;
288 			uint16_t	port;
289 			uint32_t	eax;		/* valid for out */
290 		} inout;
291 		struct {
292 			uint64_t	gpa;
293 			int		fault_type;
294 		} paging;
295 		struct {
296 			uint64_t	gpa;
297 			uint64_t	gla;
298 			uint64_t	cr3;
299 			struct vie	vie;
300 		} inst_emul;
301 		/*
302 		 * VMX specific payload. Used when there is no "better"
303 		 * exitcode to represent the VM-exit.
304 		 */
305 		struct {
306 			int		status;		/* vmx inst status */
307 			/*
308 			 * 'exit_reason' and 'exit_qualification' are valid
309 			 * only if 'status' is zero.
310 			 */
311 			uint32_t	exit_reason;
312 			uint64_t	exit_qualification;
313 			/*
314 			 * 'inst_error' and 'inst_type' are valid
315 			 * only if 'status' is non-zero.
316 			 */
317 			int		inst_type;
318 			int		inst_error;
319 		} vmx;
320 		struct {
321 			uint32_t	code;		/* ecx value */
322 			uint64_t	wval;
323 		} msr;
324 		struct {
325 			int		vcpu;
326 			uint64_t	rip;
327 		} spinup_ap;
328 		struct {
329 			uint64_t	rflags;
330 		} hlt;
331 	} u;
332 };
333 
334 #endif	/* _VMM_H_ */
335