xref: /freebsd/sys/amd64/include/vmm.h (revision d8b88105c2ccf7686552516877f541efb54fb6c8)
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)(int ipinum);
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, void *rendezvous_cookie);
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  * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'.
140  * The rendezvous 'func(arg)' is not allowed to do anything that will
141  * cause the thread to be put to sleep.
142  *
143  * If the rendezvous is being initiated from a vcpu context then the
144  * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1.
145  *
146  * The caller cannot hold any locks when initiating the rendezvous.
147  *
148  * The implementation of this API may cause vcpus other than those specified
149  * by 'dest' to be stalled. The caller should not rely on any vcpus making
150  * forward progress when the rendezvous is in progress.
151  */
152 typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg);
153 void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest,
154     vm_rendezvous_func_t func, void *arg);
155 
156 static __inline int
157 vcpu_rendezvous_pending(void *rendezvous_cookie)
158 {
159 
160 	return (*(uintptr_t *)rendezvous_cookie != 0);
161 }
162 
163 /*
164  * Return 1 if device indicated by bus/slot/func is supposed to be a
165  * pci passthrough device.
166  *
167  * Return 0 otherwise.
168  */
169 int vmm_is_pptdev(int bus, int slot, int func);
170 
171 void *vm_iommu_domain(struct vm *vm);
172 
173 enum vcpu_state {
174 	VCPU_IDLE,
175 	VCPU_FROZEN,
176 	VCPU_RUNNING,
177 	VCPU_SLEEPING,
178 };
179 
180 int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state,
181     bool from_idle);
182 enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu);
183 
184 static int __inline
185 vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu)
186 {
187 	return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING);
188 }
189 
190 void *vcpu_stats(struct vm *vm, int vcpu);
191 void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr);
192 struct vmspace *vm_get_vmspace(struct vm *vm);
193 int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
194 int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
195 #endif	/* KERNEL */
196 
197 #include <machine/vmm_instruction_emul.h>
198 
199 #define	VM_MAXCPU	16			/* maximum virtual cpus */
200 
201 /*
202  * Identifiers for events that can be injected into the VM
203  */
204 enum vm_event_type {
205 	VM_EVENT_NONE,
206 	VM_HW_INTR,
207 	VM_NMI,
208 	VM_HW_EXCEPTION,
209 	VM_SW_INTR,
210 	VM_PRIV_SW_EXCEPTION,
211 	VM_SW_EXCEPTION,
212 	VM_EVENT_MAX
213 };
214 
215 /*
216  * Identifiers for architecturally defined registers.
217  */
218 enum vm_reg_name {
219 	VM_REG_GUEST_RAX,
220 	VM_REG_GUEST_RBX,
221 	VM_REG_GUEST_RCX,
222 	VM_REG_GUEST_RDX,
223 	VM_REG_GUEST_RSI,
224 	VM_REG_GUEST_RDI,
225 	VM_REG_GUEST_RBP,
226 	VM_REG_GUEST_R8,
227 	VM_REG_GUEST_R9,
228 	VM_REG_GUEST_R10,
229 	VM_REG_GUEST_R11,
230 	VM_REG_GUEST_R12,
231 	VM_REG_GUEST_R13,
232 	VM_REG_GUEST_R14,
233 	VM_REG_GUEST_R15,
234 	VM_REG_GUEST_CR0,
235 	VM_REG_GUEST_CR3,
236 	VM_REG_GUEST_CR4,
237 	VM_REG_GUEST_DR7,
238 	VM_REG_GUEST_RSP,
239 	VM_REG_GUEST_RIP,
240 	VM_REG_GUEST_RFLAGS,
241 	VM_REG_GUEST_ES,
242 	VM_REG_GUEST_CS,
243 	VM_REG_GUEST_SS,
244 	VM_REG_GUEST_DS,
245 	VM_REG_GUEST_FS,
246 	VM_REG_GUEST_GS,
247 	VM_REG_GUEST_LDTR,
248 	VM_REG_GUEST_TR,
249 	VM_REG_GUEST_IDTR,
250 	VM_REG_GUEST_GDTR,
251 	VM_REG_GUEST_EFER,
252 	VM_REG_LAST
253 };
254 
255 /*
256  * Identifiers for optional vmm capabilities
257  */
258 enum vm_cap_type {
259 	VM_CAP_HALT_EXIT,
260 	VM_CAP_MTRAP_EXIT,
261 	VM_CAP_PAUSE_EXIT,
262 	VM_CAP_UNRESTRICTED_GUEST,
263 	VM_CAP_ENABLE_INVPCID,
264 	VM_CAP_MAX
265 };
266 
267 enum x2apic_state {
268 	X2APIC_ENABLED,
269 	X2APIC_AVAILABLE,
270 	X2APIC_DISABLED,
271 	X2APIC_STATE_LAST
272 };
273 
274 /*
275  * The 'access' field has the format specified in Table 21-2 of the Intel
276  * Architecture Manual vol 3b.
277  *
278  * XXX The contents of the 'access' field are architecturally defined except
279  * bit 16 - Segment Unusable.
280  */
281 struct seg_desc {
282 	uint64_t	base;
283 	uint32_t	limit;
284 	uint32_t	access;
285 };
286 
287 enum vm_exitcode {
288 	VM_EXITCODE_INOUT,
289 	VM_EXITCODE_VMX,
290 	VM_EXITCODE_BOGUS,
291 	VM_EXITCODE_RDMSR,
292 	VM_EXITCODE_WRMSR,
293 	VM_EXITCODE_HLT,
294 	VM_EXITCODE_MTRAP,
295 	VM_EXITCODE_PAUSE,
296 	VM_EXITCODE_PAGING,
297 	VM_EXITCODE_INST_EMUL,
298 	VM_EXITCODE_SPINUP_AP,
299 	VM_EXITCODE_SPINDOWN_CPU,
300 	VM_EXITCODE_RENDEZVOUS,
301 	VM_EXITCODE_IOAPIC_EOI,
302 	VM_EXITCODE_MAX
303 };
304 
305 struct vm_exit {
306 	enum vm_exitcode	exitcode;
307 	int			inst_length;	/* 0 means unknown */
308 	uint64_t		rip;
309 	union {
310 		struct {
311 			uint16_t	bytes:3;	/* 1 or 2 or 4 */
312 			uint16_t	in:1;		/* out is 0, in is 1 */
313 			uint16_t	string:1;
314 			uint16_t	rep:1;
315 			uint16_t	port;
316 			uint32_t	eax;		/* valid for out */
317 		} inout;
318 		struct {
319 			uint64_t	gpa;
320 			int		fault_type;
321 		} paging;
322 		struct {
323 			uint64_t	gpa;
324 			uint64_t	gla;
325 			uint64_t	cr3;
326 			struct vie	vie;
327 		} inst_emul;
328 		/*
329 		 * VMX specific payload. Used when there is no "better"
330 		 * exitcode to represent the VM-exit.
331 		 */
332 		struct {
333 			int		status;		/* vmx inst status */
334 			/*
335 			 * 'exit_reason' and 'exit_qualification' are valid
336 			 * only if 'status' is zero.
337 			 */
338 			uint32_t	exit_reason;
339 			uint64_t	exit_qualification;
340 			/*
341 			 * 'inst_error' and 'inst_type' are valid
342 			 * only if 'status' is non-zero.
343 			 */
344 			int		inst_type;
345 			int		inst_error;
346 		} vmx;
347 		struct {
348 			uint32_t	code;		/* ecx value */
349 			uint64_t	wval;
350 		} msr;
351 		struct {
352 			int		vcpu;
353 			uint64_t	rip;
354 		} spinup_ap;
355 		struct {
356 			uint64_t	rflags;
357 		} hlt;
358 		struct {
359 			int		vector;
360 		} ioapic_eoi;
361 	} u;
362 };
363 
364 #endif	/* _VMM_H_ */
365