2 * Copyright (c) 2011 NetApp, Inc.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
32 #include <x86/segments.h>
39 VM_SUSPEND_TRIPLEFAULT,
44 * Identifiers for architecturally defined registers.
85 VM_REG_GUEST_INTR_SHADOW,
95 #define VM_INTINFO_VECTOR(info) ((info) & 0xff)
96 #define VM_INTINFO_DEL_ERRCODE 0x800
97 #define VM_INTINFO_RSVD 0x7ffff000
98 #define VM_INTINFO_VALID 0x80000000
99 #define VM_INTINFO_TYPE 0x700
100 #define VM_INTINFO_HWINTR (0 << 8)
101 #define VM_INTINFO_NMI (2 << 8)
102 #define VM_INTINFO_HWEXCEPTION (3 << 8)
103 #define VM_INTINFO_SWINTR (4 << 8)
107 #define VM_MAX_NAMELEN 32
111 struct vm_memory_segment;
120 struct vm_guest_paging;
123 struct vm_eventinfo {
124 void *rptr; /* rendezvous cookie */
125 int *sptr; /* suspend cookie */
126 int *iptr; /* reqidle cookie */
129 typedef int (*vmm_init_func_t)(int ipinum);
130 typedef int (*vmm_cleanup_func_t)(void);
131 typedef void (*vmm_resume_func_t)(void);
132 typedef void * (*vmi_init_func_t)(struct vm *vm, struct pmap *pmap);
133 typedef int (*vmi_run_func_t)(void *vmi, int vcpu, register_t rip,
134 struct pmap *pmap, struct vm_eventinfo *info);
135 typedef void (*vmi_cleanup_func_t)(void *vmi);
136 typedef int (*vmi_get_register_t)(void *vmi, int vcpu, int num,
138 typedef int (*vmi_set_register_t)(void *vmi, int vcpu, int num,
140 typedef int (*vmi_get_desc_t)(void *vmi, int vcpu, int num,
141 struct seg_desc *desc);
142 typedef int (*vmi_set_desc_t)(void *vmi, int vcpu, int num,
143 struct seg_desc *desc);
144 typedef int (*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval);
145 typedef int (*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val);
146 typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max);
147 typedef void (*vmi_vmspace_free)(struct vmspace *vmspace);
148 typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu);
149 typedef void (*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic);
152 vmm_init_func_t init; /* module wide initialization */
153 vmm_cleanup_func_t cleanup;
154 vmm_resume_func_t resume;
156 vmi_init_func_t vminit; /* vm-specific initialization */
157 vmi_run_func_t vmrun;
158 vmi_cleanup_func_t vmcleanup;
159 vmi_get_register_t vmgetreg;
160 vmi_set_register_t vmsetreg;
161 vmi_get_desc_t vmgetdesc;
162 vmi_set_desc_t vmsetdesc;
163 vmi_get_cap_t vmgetcap;
164 vmi_set_cap_t vmsetcap;
165 vmi_vmspace_alloc vmspace_alloc;
166 vmi_vmspace_free vmspace_free;
167 vmi_vlapic_init vlapic_init;
168 vmi_vlapic_cleanup vlapic_cleanup;
171 extern struct vmm_ops vmm_ops_intel;
172 extern struct vmm_ops vmm_ops_amd;
174 int vm_create(const char *name, struct vm **retvm);
175 void vm_destroy(struct vm *vm);
176 int vm_reinit(struct vm *vm);
177 const char *vm_name(struct vm *vm);
178 int vm_malloc(struct vm *vm, vm_paddr_t gpa, size_t len);
179 int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa);
180 int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len);
181 void *vm_gpa_hold(struct vm *, vm_paddr_t gpa, size_t len, int prot,
183 void vm_gpa_release(void *cookie);
184 int vm_gpabase2memseg(struct vm *vm, vm_paddr_t gpabase,
185 struct vm_memory_segment *seg);
186 int vm_get_memobj(struct vm *vm, vm_paddr_t gpa, size_t len,
187 vm_offset_t *offset, struct vm_object **object);
188 boolean_t vm_mem_allocated(struct vm *vm, vm_paddr_t gpa);
189 int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval);
190 int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val);
191 int vm_get_seg_desc(struct vm *vm, int vcpu, int reg,
192 struct seg_desc *ret_desc);
193 int vm_set_seg_desc(struct vm *vm, int vcpu, int reg,
194 struct seg_desc *desc);
195 int vm_run(struct vm *vm, struct vm_run *vmrun);
196 int vm_suspend(struct vm *vm, enum vm_suspend_how how);
197 int vm_inject_nmi(struct vm *vm, int vcpu);
198 int vm_nmi_pending(struct vm *vm, int vcpuid);
199 void vm_nmi_clear(struct vm *vm, int vcpuid);
200 int vm_inject_extint(struct vm *vm, int vcpu);
201 int vm_extint_pending(struct vm *vm, int vcpuid);
202 void vm_extint_clear(struct vm *vm, int vcpuid);
203 struct vlapic *vm_lapic(struct vm *vm, int cpu);
204 struct vioapic *vm_ioapic(struct vm *vm);
205 struct vhpet *vm_hpet(struct vm *vm);
206 int vm_get_capability(struct vm *vm, int vcpu, int type, int *val);
207 int vm_set_capability(struct vm *vm, int vcpu, int type, int val);
208 int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state);
209 int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state);
210 int vm_apicid2vcpuid(struct vm *vm, int apicid);
211 int vm_activate_cpu(struct vm *vm, int vcpu);
212 struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid);
213 void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip);
214 void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip);
215 void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip);
216 void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip);
218 #ifdef _SYS__CPUSET_H_
220 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'.
221 * The rendezvous 'func(arg)' is not allowed to do anything that will
222 * cause the thread to be put to sleep.
224 * If the rendezvous is being initiated from a vcpu context then the
225 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1.
227 * The caller cannot hold any locks when initiating the rendezvous.
229 * The implementation of this API may cause vcpus other than those specified
230 * by 'dest' to be stalled. The caller should not rely on any vcpus making
231 * forward progress when the rendezvous is in progress.
233 typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg);
234 void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest,
235 vm_rendezvous_func_t func, void *arg);
236 cpuset_t vm_active_cpus(struct vm *vm);
237 cpuset_t vm_suspended_cpus(struct vm *vm);
238 #endif /* _SYS__CPUSET_H_ */
241 vcpu_rendezvous_pending(struct vm_eventinfo *info)
244 return (*((uintptr_t *)(info->rptr)) != 0);
248 vcpu_suspended(struct vm_eventinfo *info)
251 return (*info->sptr);
255 vcpu_reqidle(struct vm_eventinfo *info)
258 return (*info->iptr);
262 * Return 1 if device indicated by bus/slot/func is supposed to be a
263 * pci passthrough device.
265 * Return 0 otherwise.
267 int vmm_is_pptdev(int bus, int slot, int func);
269 void *vm_iommu_domain(struct vm *vm);
278 int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state,
280 enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu);
283 vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu)
285 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING);
290 vcpu_should_yield(struct vm *vm, int vcpu)
293 if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED))
295 else if (curthread->td_owepreempt)
302 void *vcpu_stats(struct vm *vm, int vcpu);
303 void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr);
304 struct vmspace *vm_get_vmspace(struct vm *vm);
305 int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
306 int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
307 struct vatpic *vm_atpic(struct vm *vm);
308 struct vatpit *vm_atpit(struct vm *vm);
309 struct vpmtmr *vm_pmtmr(struct vm *vm);
310 struct vrtc *vm_rtc(struct vm *vm);
313 * Inject exception 'vector' into the guest vcpu. This function returns 0 on
314 * success and non-zero on failure.
316 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling
317 * this function directly because they enforce the trap-like or fault-like
318 * behavior of an exception.
320 * This function should only be called in the context of the thread that is
321 * executing this vcpu.
323 int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid,
324 uint32_t errcode, int restart_instruction);
327 * This function is called after a VM-exit that occurred during exception or
328 * interrupt delivery through the IDT. The format of 'intinfo' is described
329 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2.
331 * If a VM-exit handler completes the event delivery successfully then it
332 * should call vm_exit_intinfo() to extinguish the pending event. For e.g.,
333 * if the task switch emulation is triggered via a task gate then it should
334 * call this function with 'intinfo=0' to indicate that the external event
335 * is not pending anymore.
337 * Return value is 0 on success and non-zero on failure.
339 int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo);
342 * This function is called before every VM-entry to retrieve a pending
343 * event that should be injected into the guest. This function combines
344 * nested events into a double or triple fault.
346 * Returns 0 if there are no events that need to be injected into the guest
347 * and non-zero otherwise.
349 int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info);
351 int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2);
353 enum vm_reg_name vm_segment_name(int seg_encoding);
363 * Set up 'copyinfo[]' to copy to/from guest linear address space starting
364 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for
365 * a copyin or PROT_WRITE for a copyout.
367 * retval is_fault Intepretation
369 * 0 1 An exception was injected into the guest
370 * EFAULT N/A Unrecoverable error
372 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if
373 * the return value is 0. The 'copyinfo[]' resources should be freed by calling
374 * 'vm_copy_teardown()' after the copy is done.
376 int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging,
377 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo,
378 int num_copyinfo, int *is_fault);
379 void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
381 void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
382 void *kaddr, size_t len);
383 void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr,
384 struct vm_copyinfo *copyinfo, size_t len);
386 int vcpu_trace_exceptions(struct vm *vm, int vcpuid);
389 #define VM_MAXCPU 16 /* maximum virtual cpus */
392 * Identifiers for optional vmm capabilities
398 VM_CAP_UNRESTRICTED_GUEST,
399 VM_CAP_ENABLE_INVPCID,
403 enum vm_intr_trigger {
409 * The 'access' field has the format specified in Table 21-2 of the Intel
410 * Architecture Manual vol 3b.
412 * XXX The contents of the 'access' field are architecturally defined except
413 * bit 16 - Segment Unusable.
420 #define SEG_DESC_TYPE(access) ((access) & 0x001f)
421 #define SEG_DESC_DPL(access) (((access) >> 5) & 0x3)
422 #define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0)
423 #define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0)
424 #define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0)
425 #define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0)
430 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */
431 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */
434 enum vm_paging_mode {
441 struct vm_guest_paging {
444 enum vm_cpu_mode cpu_mode;
445 enum vm_paging_mode paging_mode;
449 * The data structures 'vie' and 'vie_op' are meant to be opaque to the
450 * consumers of instruction decoding. The only reason why their contents
451 * need to be exposed is because they are part of the 'vm_exit' structure.
454 uint8_t op_byte; /* actual opcode byte */
455 uint8_t op_type; /* type of operation (e.g. MOV) */
459 #define VIE_INST_SIZE 15
461 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */
462 uint8_t num_valid; /* size of the instruction */
463 uint8_t num_processed;
465 uint8_t addrsize:4, opsize:4; /* address and operand sizes */
466 uint8_t rex_w:1, /* REX prefix */
471 repz_present:1, /* REP/REPE/REPZ prefix */
472 repnz_present:1, /* REPNE/REPNZ prefix */
473 opsize_override:1, /* Operand size override */
474 addrsize_override:1, /* Address size override */
475 segment_override:1; /* Segment override */
477 uint8_t mod:2, /* ModRM byte */
481 uint8_t ss:2, /* SIB byte */
489 int base_register; /* VM_REG_GUEST_xyz */
490 int index_register; /* VM_REG_GUEST_xyz */
491 int segment_register; /* VM_REG_GUEST_xyz */
493 int64_t displacement; /* optional addr displacement */
494 int64_t immediate; /* optional immediate operand */
496 uint8_t decoded; /* set to 1 if successfully decoded */
498 struct vie_op op; /* opcode description */
511 VM_EXITCODE_INST_EMUL,
512 VM_EXITCODE_SPINUP_AP,
513 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */
514 VM_EXITCODE_RENDEZVOUS,
515 VM_EXITCODE_IOAPIC_EOI,
516 VM_EXITCODE_SUSPENDED,
517 VM_EXITCODE_INOUT_STR,
518 VM_EXITCODE_TASK_SWITCH,
527 uint16_t bytes:3; /* 1 or 2 or 4 */
532 uint32_t eax; /* valid for out */
535 struct vm_inout_str {
536 struct vm_inout inout; /* must be the first element */
537 struct vm_guest_paging paging;
541 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */
543 enum vm_reg_name seg_name;
544 struct seg_desc seg_desc;
547 enum task_switch_reason {
551 TSR_IDT_GATE, /* task gate in IDT */
554 struct vm_task_switch {
555 uint16_t tsssel; /* new TSS selector */
556 int ext; /* task switch due to external event */
558 int errcode_valid; /* push 'errcode' on the new stack */
559 enum task_switch_reason reason;
560 struct vm_guest_paging paging;
564 enum vm_exitcode exitcode;
565 int inst_length; /* 0 means unknown */
568 struct vm_inout inout;
569 struct vm_inout_str inout_str;
579 struct vm_guest_paging paging;
583 * VMX specific payload. Used when there is no "better"
584 * exitcode to represent the VM-exit.
587 int status; /* vmx inst status */
589 * 'exit_reason' and 'exit_qualification' are valid
590 * only if 'status' is zero.
592 uint32_t exit_reason;
593 uint64_t exit_qualification;
595 * 'inst_error' and 'inst_type' are valid
596 * only if 'status' is non-zero.
602 * SVM specific payload.
610 uint32_t code; /* ecx value */
624 enum vm_suspend_how how;
626 struct vm_task_switch task_switch;
630 /* APIs to inject faults into the guest */
631 void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid,
635 vm_inject_ud(void *vm, int vcpuid)
637 vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0);
641 vm_inject_gp(void *vm, int vcpuid)
643 vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0);
647 vm_inject_ac(void *vm, int vcpuid, int errcode)
649 vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode);
653 vm_inject_ss(void *vm, int vcpuid, int errcode)
655 vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode);
658 void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2);
660 int vm_restart_instruction(void *vm, int vcpuid);