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.
94 #define VM_INTINFO_VECTOR(info) ((info) & 0xff)
95 #define VM_INTINFO_DEL_ERRCODE 0x800
96 #define VM_INTINFO_RSVD 0x7ffff000
97 #define VM_INTINFO_VALID 0x80000000
98 #define VM_INTINFO_TYPE 0x700
99 #define VM_INTINFO_HWINTR (0 << 8)
100 #define VM_INTINFO_NMI (2 << 8)
101 #define VM_INTINFO_HWEXCEPTION (3 << 8)
102 #define VM_INTINFO_SWINTR (4 << 8)
106 #define VM_MAX_NAMELEN 32
110 struct vm_memory_segment;
119 struct vm_guest_paging;
122 typedef int (*vmm_init_func_t)(int ipinum);
123 typedef int (*vmm_cleanup_func_t)(void);
124 typedef void (*vmm_resume_func_t)(void);
125 typedef void * (*vmi_init_func_t)(struct vm *vm, struct pmap *pmap);
126 typedef int (*vmi_run_func_t)(void *vmi, int vcpu, register_t rip,
127 struct pmap *pmap, void *rendezvous_cookie,
128 void *suspend_cookie);
129 typedef void (*vmi_cleanup_func_t)(void *vmi);
130 typedef int (*vmi_get_register_t)(void *vmi, int vcpu, int num,
132 typedef int (*vmi_set_register_t)(void *vmi, int vcpu, int num,
134 typedef int (*vmi_get_desc_t)(void *vmi, int vcpu, int num,
135 struct seg_desc *desc);
136 typedef int (*vmi_set_desc_t)(void *vmi, int vcpu, int num,
137 struct seg_desc *desc);
138 typedef int (*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval);
139 typedef int (*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val);
140 typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max);
141 typedef void (*vmi_vmspace_free)(struct vmspace *vmspace);
142 typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu);
143 typedef void (*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic);
146 vmm_init_func_t init; /* module wide initialization */
147 vmm_cleanup_func_t cleanup;
148 vmm_resume_func_t resume;
150 vmi_init_func_t vminit; /* vm-specific initialization */
151 vmi_run_func_t vmrun;
152 vmi_cleanup_func_t vmcleanup;
153 vmi_get_register_t vmgetreg;
154 vmi_set_register_t vmsetreg;
155 vmi_get_desc_t vmgetdesc;
156 vmi_set_desc_t vmsetdesc;
157 vmi_get_cap_t vmgetcap;
158 vmi_set_cap_t vmsetcap;
159 vmi_vmspace_alloc vmspace_alloc;
160 vmi_vmspace_free vmspace_free;
161 vmi_vlapic_init vlapic_init;
162 vmi_vlapic_cleanup vlapic_cleanup;
165 extern struct vmm_ops vmm_ops_intel;
166 extern struct vmm_ops vmm_ops_amd;
168 int vm_create(const char *name, struct vm **retvm);
169 void vm_destroy(struct vm *vm);
170 int vm_reinit(struct vm *vm);
171 const char *vm_name(struct vm *vm);
172 int vm_malloc(struct vm *vm, vm_paddr_t gpa, size_t len);
173 int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa);
174 int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len);
175 void *vm_gpa_hold(struct vm *, vm_paddr_t gpa, size_t len, int prot,
177 void vm_gpa_release(void *cookie);
178 int vm_gpabase2memseg(struct vm *vm, vm_paddr_t gpabase,
179 struct vm_memory_segment *seg);
180 int vm_get_memobj(struct vm *vm, vm_paddr_t gpa, size_t len,
181 vm_offset_t *offset, struct vm_object **object);
182 boolean_t vm_mem_allocated(struct vm *vm, vm_paddr_t gpa);
183 int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval);
184 int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val);
185 int vm_get_seg_desc(struct vm *vm, int vcpu, int reg,
186 struct seg_desc *ret_desc);
187 int vm_set_seg_desc(struct vm *vm, int vcpu, int reg,
188 struct seg_desc *desc);
189 int vm_run(struct vm *vm, struct vm_run *vmrun);
190 int vm_suspend(struct vm *vm, enum vm_suspend_how how);
191 int vm_inject_nmi(struct vm *vm, int vcpu);
192 int vm_nmi_pending(struct vm *vm, int vcpuid);
193 void vm_nmi_clear(struct vm *vm, int vcpuid);
194 int vm_inject_extint(struct vm *vm, int vcpu);
195 int vm_extint_pending(struct vm *vm, int vcpuid);
196 void vm_extint_clear(struct vm *vm, int vcpuid);
197 uint64_t *vm_guest_msrs(struct vm *vm, int cpu);
198 struct vlapic *vm_lapic(struct vm *vm, int cpu);
199 struct vioapic *vm_ioapic(struct vm *vm);
200 struct vhpet *vm_hpet(struct vm *vm);
201 int vm_get_capability(struct vm *vm, int vcpu, int type, int *val);
202 int vm_set_capability(struct vm *vm, int vcpu, int type, int val);
203 int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state);
204 int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state);
205 int vm_apicid2vcpuid(struct vm *vm, int apicid);
206 int vm_activate_cpu(struct vm *vm, int vcpu);
207 cpuset_t vm_active_cpus(struct vm *vm);
208 cpuset_t vm_suspended_cpus(struct vm *vm);
209 struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid);
210 void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip);
211 void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip);
212 void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip);
215 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'.
216 * The rendezvous 'func(arg)' is not allowed to do anything that will
217 * cause the thread to be put to sleep.
219 * If the rendezvous is being initiated from a vcpu context then the
220 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1.
222 * The caller cannot hold any locks when initiating the rendezvous.
224 * The implementation of this API may cause vcpus other than those specified
225 * by 'dest' to be stalled. The caller should not rely on any vcpus making
226 * forward progress when the rendezvous is in progress.
228 typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg);
229 void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest,
230 vm_rendezvous_func_t func, void *arg);
233 vcpu_rendezvous_pending(void *rendezvous_cookie)
236 return (*(uintptr_t *)rendezvous_cookie != 0);
240 vcpu_suspended(void *suspend_cookie)
243 return (*(int *)suspend_cookie);
247 * Return 1 if device indicated by bus/slot/func is supposed to be a
248 * pci passthrough device.
250 * Return 0 otherwise.
252 int vmm_is_pptdev(int bus, int slot, int func);
254 void *vm_iommu_domain(struct vm *vm);
263 int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state,
265 enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu);
268 vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu)
270 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING);
275 vcpu_should_yield(struct vm *vm, int vcpu)
277 return (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED));
281 void *vcpu_stats(struct vm *vm, int vcpu);
282 void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr);
283 struct vmspace *vm_get_vmspace(struct vm *vm);
284 int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
285 int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
286 struct vatpic *vm_atpic(struct vm *vm);
287 struct vatpit *vm_atpit(struct vm *vm);
290 * Inject exception 'vme' into the guest vcpu. This function returns 0 on
291 * success and non-zero on failure.
293 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling
294 * this function directly because they enforce the trap-like or fault-like
295 * behavior of an exception.
297 * This function should only be called in the context of the thread that is
298 * executing this vcpu.
300 int vm_inject_exception(struct vm *vm, int vcpuid, struct vm_exception *vme);
303 * This function is called after a VM-exit that occurred during exception or
304 * interrupt delivery through the IDT. The format of 'intinfo' is described
305 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2.
307 * If a VM-exit handler completes the event delivery successfully then it
308 * should call vm_exit_intinfo() to extinguish the pending event. For e.g.,
309 * if the task switch emulation is triggered via a task gate then it should
310 * call this function with 'intinfo=0' to indicate that the external event
311 * is not pending anymore.
313 * Return value is 0 on success and non-zero on failure.
315 int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo);
318 * This function is called before every VM-entry to retrieve a pending
319 * event that should be injected into the guest. This function combines
320 * nested events into a double or triple fault.
322 * Returns 0 if there are no events that need to be injected into the guest
323 * and non-zero otherwise.
325 int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info);
327 int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2);
329 enum vm_reg_name vm_segment_name(int seg_encoding);
339 * Set up 'copyinfo[]' to copy to/from guest linear address space starting
340 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for
341 * a copyin or PROT_WRITE for a copyout.
343 * Returns 0 on success.
344 * Returns 1 if an exception was injected into the guest.
345 * Returns -1 otherwise.
347 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if
348 * the return value is 0. The 'copyinfo[]' resources should be freed by calling
349 * 'vm_copy_teardown()' after the copy is done.
351 int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging,
352 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo,
354 void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
356 void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
357 void *kaddr, size_t len);
358 void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr,
359 struct vm_copyinfo *copyinfo, size_t len);
362 #define VM_MAXCPU 16 /* maximum virtual cpus */
365 * Identifiers for optional vmm capabilities
371 VM_CAP_UNRESTRICTED_GUEST,
372 VM_CAP_ENABLE_INVPCID,
376 enum vm_intr_trigger {
382 * The 'access' field has the format specified in Table 21-2 of the Intel
383 * Architecture Manual vol 3b.
385 * XXX The contents of the 'access' field are architecturally defined except
386 * bit 16 - Segment Unusable.
393 #define SEG_DESC_TYPE(access) ((access) & 0x001f)
394 #define SEG_DESC_DPL(access) (((access) >> 5) & 0x3)
395 #define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0)
396 #define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0)
397 #define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0)
398 #define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0)
403 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */
404 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */
407 enum vm_paging_mode {
414 struct vm_guest_paging {
417 enum vm_cpu_mode cpu_mode;
418 enum vm_paging_mode paging_mode;
422 * The data structures 'vie' and 'vie_op' are meant to be opaque to the
423 * consumers of instruction decoding. The only reason why their contents
424 * need to be exposed is because they are part of the 'vm_exit' structure.
427 uint8_t op_byte; /* actual opcode byte */
428 uint8_t op_type; /* type of operation (e.g. MOV) */
432 #define VIE_INST_SIZE 15
434 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */
435 uint8_t num_valid; /* size of the instruction */
436 uint8_t num_processed;
438 uint8_t addrsize:4, opsize:4; /* address and operand sizes */
439 uint8_t rex_w:1, /* REX prefix */
444 opsize_override:1, /* Operand size override */
445 addrsize_override:1; /* Address size override */
447 uint8_t mod:2, /* ModRM byte */
451 uint8_t ss:2, /* SIB byte */
459 int base_register; /* VM_REG_GUEST_xyz */
460 int index_register; /* VM_REG_GUEST_xyz */
462 int64_t displacement; /* optional addr displacement */
463 int64_t immediate; /* optional immediate operand */
465 uint8_t decoded; /* set to 1 if successfully decoded */
467 struct vie_op op; /* opcode description */
480 VM_EXITCODE_INST_EMUL,
481 VM_EXITCODE_SPINUP_AP,
482 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */
483 VM_EXITCODE_RENDEZVOUS,
484 VM_EXITCODE_IOAPIC_EOI,
485 VM_EXITCODE_SUSPENDED,
486 VM_EXITCODE_INOUT_STR,
487 VM_EXITCODE_TASK_SWITCH,
492 uint16_t bytes:3; /* 1 or 2 or 4 */
497 uint32_t eax; /* valid for out */
500 struct vm_inout_str {
501 struct vm_inout inout; /* must be the first element */
502 struct vm_guest_paging paging;
506 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */
508 enum vm_reg_name seg_name;
509 struct seg_desc seg_desc;
512 enum task_switch_reason {
516 TSR_IDT_GATE, /* task gate in IDT */
519 struct vm_task_switch {
520 uint16_t tsssel; /* new TSS selector */
521 int ext; /* task switch due to external event */
523 int errcode_valid; /* push 'errcode' on the new stack */
524 enum task_switch_reason reason;
525 struct vm_guest_paging paging;
529 enum vm_exitcode exitcode;
530 int inst_length; /* 0 means unknown */
533 struct vm_inout inout;
534 struct vm_inout_str inout_str;
543 struct vm_guest_paging paging;
547 * VMX specific payload. Used when there is no "better"
548 * exitcode to represent the VM-exit.
551 int status; /* vmx inst status */
553 * 'exit_reason' and 'exit_qualification' are valid
554 * only if 'status' is zero.
556 uint32_t exit_reason;
557 uint64_t exit_qualification;
559 * 'inst_error' and 'inst_type' are valid
560 * only if 'status' is non-zero.
566 uint32_t code; /* ecx value */
580 enum vm_suspend_how how;
582 struct vm_task_switch task_switch;
586 /* APIs to inject faults into the guest */
587 void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid,
591 vm_inject_ud(void *vm, int vcpuid)
593 vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0);
597 vm_inject_gp(void *vm, int vcpuid)
599 vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0);
603 vm_inject_ac(void *vm, int vcpuid, int errcode)
605 vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode);
609 vm_inject_ss(void *vm, int vcpuid, int errcode)
611 vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode);
614 void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2);