5 * This work is licensed under the terms of the GNU GPL, version 2. See
6 * the COPYING file in the top-level directory.
9 #include <linux/types.h>
10 #include <linux/hardirq.h>
11 #include <linux/list.h>
12 #include <linux/mutex.h>
13 #include <linux/spinlock.h>
14 #include <linux/signal.h>
15 #include <linux/sched.h>
17 #include <linux/preempt.h>
18 #include <linux/marker.h>
19 #include <linux/msi.h>
20 #include <asm/signal.h>
22 #include <linux/kvm.h>
23 #include <linux/kvm_para.h>
25 #include <linux/kvm_types.h>
27 #include <asm/kvm_host.h>
30 * vcpu->requests bit members
32 #define KVM_REQ_TLB_FLUSH 0
33 #define KVM_REQ_MIGRATE_TIMER 1
34 #define KVM_REQ_REPORT_TPR_ACCESS 2
35 #define KVM_REQ_MMU_RELOAD 3
36 #define KVM_REQ_TRIPLE_FAULT 4
37 #define KVM_REQ_PENDING_TIMER 5
38 #define KVM_REQ_UNHALT 6
39 #define KVM_REQ_MMU_SYNC 7
40 #define KVM_REQ_KVMCLOCK_UPDATE 8
41 #define KVM_REQ_KICK 9
43 #define KVM_USERSPACE_IRQ_SOURCE_ID 0
46 extern struct kmem_cache
*kvm_vcpu_cache
;
49 * It would be nice to use something smarter than a linear search, TBD...
50 * Thankfully we dont expect many devices to register (famous last words :),
51 * so until then it will suffice. At least its abstracted so we can change
56 #define NR_IOBUS_DEVS 6
57 struct kvm_io_device
*devs
[NR_IOBUS_DEVS
];
60 void kvm_io_bus_init(struct kvm_io_bus
*bus
);
61 void kvm_io_bus_destroy(struct kvm_io_bus
*bus
);
62 struct kvm_io_device
*kvm_io_bus_find_dev(struct kvm_io_bus
*bus
,
63 gpa_t addr
, int len
, int is_write
);
64 void kvm_io_bus_register_dev(struct kvm_io_bus
*bus
,
65 struct kvm_io_device
*dev
);
69 #ifdef CONFIG_PREEMPT_NOTIFIERS
70 struct preempt_notifier preempt_notifier
;
76 unsigned long requests
;
77 unsigned long guest_debug
;
83 struct kvm_vcpu_stat stat
;
85 #ifdef CONFIG_HAS_IOMEM
87 int mmio_read_completed
;
90 unsigned char mmio_data
[8];
94 struct kvm_vcpu_arch arch
;
97 struct kvm_memory_slot
{
102 unsigned long *dirty_bitmap
;
104 unsigned long rmap_pde
;
107 unsigned long userspace_addr
;
111 struct kvm_kernel_irq_routing_entry
{
114 int (*set
)(struct kvm_kernel_irq_routing_entry
*e
,
115 struct kvm
*kvm
, int level
);
123 struct list_head link
;
127 struct mutex lock
; /* protects the vcpus array and APIC accesses */
129 spinlock_t requests_lock
;
130 struct rw_semaphore slots_lock
;
131 struct mm_struct
*mm
; /* userspace tied to this vm */
133 struct kvm_memory_slot memslots
[KVM_MEMORY_SLOTS
+
134 KVM_PRIVATE_MEM_SLOTS
];
135 struct kvm_vcpu
*vcpus
[KVM_MAX_VCPUS
];
136 struct list_head vm_list
;
137 struct kvm_io_bus mmio_bus
;
138 struct kvm_io_bus pio_bus
;
139 struct kvm_vm_stat stat
;
140 struct kvm_arch arch
;
141 atomic_t users_count
;
142 #ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
143 struct kvm_coalesced_mmio_dev
*coalesced_mmio_dev
;
144 struct kvm_coalesced_mmio_ring
*coalesced_mmio_ring
;
147 #ifdef CONFIG_HAVE_KVM_IRQCHIP
148 struct list_head irq_routing
; /* of kvm_kernel_irq_routing_entry */
149 struct hlist_head mask_notifier_list
;
152 #ifdef KVM_ARCH_WANT_MMU_NOTIFIER
153 struct mmu_notifier mmu_notifier
;
154 unsigned long mmu_notifier_seq
;
155 long mmu_notifier_count
;
159 /* The guest did something we don't support. */
160 #define pr_unimpl(vcpu, fmt, ...) \
162 if (printk_ratelimit()) \
163 printk(KERN_ERR "kvm: %i: cpu%i " fmt, \
164 current->tgid, (vcpu)->vcpu_id , ## __VA_ARGS__); \
167 #define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
168 #define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
170 int kvm_vcpu_init(struct kvm_vcpu
*vcpu
, struct kvm
*kvm
, unsigned id
);
171 void kvm_vcpu_uninit(struct kvm_vcpu
*vcpu
);
173 void vcpu_load(struct kvm_vcpu
*vcpu
);
174 void vcpu_put(struct kvm_vcpu
*vcpu
);
176 int kvm_init(void *opaque
, unsigned int vcpu_size
,
177 struct module
*module
);
180 void kvm_get_kvm(struct kvm
*kvm
);
181 void kvm_put_kvm(struct kvm
*kvm
);
183 #define HPA_MSB ((sizeof(hpa_t) * 8) - 1)
184 #define HPA_ERR_MASK ((hpa_t)1 << HPA_MSB)
185 static inline int is_error_hpa(hpa_t hpa
) { return hpa
>> HPA_MSB
; }
186 struct page
*gva_to_page(struct kvm_vcpu
*vcpu
, gva_t gva
);
188 extern struct page
*bad_page
;
189 extern pfn_t bad_pfn
;
191 int is_error_page(struct page
*page
);
192 int is_error_pfn(pfn_t pfn
);
193 int kvm_is_error_hva(unsigned long addr
);
194 int kvm_set_memory_region(struct kvm
*kvm
,
195 struct kvm_userspace_memory_region
*mem
,
197 int __kvm_set_memory_region(struct kvm
*kvm
,
198 struct kvm_userspace_memory_region
*mem
,
200 int kvm_arch_set_memory_region(struct kvm
*kvm
,
201 struct kvm_userspace_memory_region
*mem
,
202 struct kvm_memory_slot old
,
204 void kvm_arch_flush_shadow(struct kvm
*kvm
);
205 gfn_t
unalias_gfn(struct kvm
*kvm
, gfn_t gfn
);
206 struct page
*gfn_to_page(struct kvm
*kvm
, gfn_t gfn
);
207 unsigned long gfn_to_hva(struct kvm
*kvm
, gfn_t gfn
);
208 void kvm_release_page_clean(struct page
*page
);
209 void kvm_release_page_dirty(struct page
*page
);
210 void kvm_set_page_dirty(struct page
*page
);
211 void kvm_set_page_accessed(struct page
*page
);
213 pfn_t
gfn_to_pfn(struct kvm
*kvm
, gfn_t gfn
);
214 void kvm_release_pfn_dirty(pfn_t
);
215 void kvm_release_pfn_clean(pfn_t pfn
);
216 void kvm_set_pfn_dirty(pfn_t pfn
);
217 void kvm_set_pfn_accessed(pfn_t pfn
);
218 void kvm_get_pfn(pfn_t pfn
);
220 int kvm_read_guest_page(struct kvm
*kvm
, gfn_t gfn
, void *data
, int offset
,
222 int kvm_read_guest_atomic(struct kvm
*kvm
, gpa_t gpa
, void *data
,
224 int kvm_read_guest(struct kvm
*kvm
, gpa_t gpa
, void *data
, unsigned long len
);
225 int kvm_write_guest_page(struct kvm
*kvm
, gfn_t gfn
, const void *data
,
226 int offset
, int len
);
227 int kvm_write_guest(struct kvm
*kvm
, gpa_t gpa
, const void *data
,
229 int kvm_clear_guest_page(struct kvm
*kvm
, gfn_t gfn
, int offset
, int len
);
230 int kvm_clear_guest(struct kvm
*kvm
, gpa_t gpa
, unsigned long len
);
231 struct kvm_memory_slot
*gfn_to_memslot(struct kvm
*kvm
, gfn_t gfn
);
232 int kvm_is_visible_gfn(struct kvm
*kvm
, gfn_t gfn
);
233 void mark_page_dirty(struct kvm
*kvm
, gfn_t gfn
);
235 void kvm_vcpu_block(struct kvm_vcpu
*vcpu
);
236 void kvm_resched(struct kvm_vcpu
*vcpu
);
237 void kvm_load_guest_fpu(struct kvm_vcpu
*vcpu
);
238 void kvm_put_guest_fpu(struct kvm_vcpu
*vcpu
);
239 void kvm_flush_remote_tlbs(struct kvm
*kvm
);
240 void kvm_reload_remote_mmus(struct kvm
*kvm
);
242 long kvm_arch_dev_ioctl(struct file
*filp
,
243 unsigned int ioctl
, unsigned long arg
);
244 long kvm_arch_vcpu_ioctl(struct file
*filp
,
245 unsigned int ioctl
, unsigned long arg
);
246 void kvm_arch_vcpu_load(struct kvm_vcpu
*vcpu
, int cpu
);
247 void kvm_arch_vcpu_put(struct kvm_vcpu
*vcpu
);
249 int kvm_dev_ioctl_check_extension(long ext
);
251 int kvm_get_dirty_log(struct kvm
*kvm
,
252 struct kvm_dirty_log
*log
, int *is_dirty
);
253 int kvm_vm_ioctl_get_dirty_log(struct kvm
*kvm
,
254 struct kvm_dirty_log
*log
);
256 int kvm_vm_ioctl_set_memory_region(struct kvm
*kvm
,
258 kvm_userspace_memory_region
*mem
,
260 long kvm_arch_vm_ioctl(struct file
*filp
,
261 unsigned int ioctl
, unsigned long arg
);
263 int kvm_arch_vcpu_ioctl_get_fpu(struct kvm_vcpu
*vcpu
, struct kvm_fpu
*fpu
);
264 int kvm_arch_vcpu_ioctl_set_fpu(struct kvm_vcpu
*vcpu
, struct kvm_fpu
*fpu
);
266 int kvm_arch_vcpu_ioctl_translate(struct kvm_vcpu
*vcpu
,
267 struct kvm_translation
*tr
);
269 int kvm_arch_vcpu_ioctl_get_regs(struct kvm_vcpu
*vcpu
, struct kvm_regs
*regs
);
270 int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu
*vcpu
, struct kvm_regs
*regs
);
271 int kvm_arch_vcpu_ioctl_get_sregs(struct kvm_vcpu
*vcpu
,
272 struct kvm_sregs
*sregs
);
273 int kvm_arch_vcpu_ioctl_set_sregs(struct kvm_vcpu
*vcpu
,
274 struct kvm_sregs
*sregs
);
275 int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu
*vcpu
,
276 struct kvm_mp_state
*mp_state
);
277 int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu
*vcpu
,
278 struct kvm_mp_state
*mp_state
);
279 int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu
*vcpu
,
280 struct kvm_guest_debug
*dbg
);
281 int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu
*vcpu
, struct kvm_run
*kvm_run
);
283 int kvm_arch_init(void *opaque
);
284 void kvm_arch_exit(void);
286 int kvm_arch_vcpu_init(struct kvm_vcpu
*vcpu
);
287 void kvm_arch_vcpu_uninit(struct kvm_vcpu
*vcpu
);
289 void kvm_arch_vcpu_free(struct kvm_vcpu
*vcpu
);
290 void kvm_arch_vcpu_load(struct kvm_vcpu
*vcpu
, int cpu
);
291 void kvm_arch_vcpu_put(struct kvm_vcpu
*vcpu
);
292 struct kvm_vcpu
*kvm_arch_vcpu_create(struct kvm
*kvm
, unsigned int id
);
293 int kvm_arch_vcpu_setup(struct kvm_vcpu
*vcpu
);
294 void kvm_arch_vcpu_destroy(struct kvm_vcpu
*vcpu
);
296 int kvm_arch_vcpu_reset(struct kvm_vcpu
*vcpu
);
297 void kvm_arch_hardware_enable(void *garbage
);
298 void kvm_arch_hardware_disable(void *garbage
);
299 int kvm_arch_hardware_setup(void);
300 void kvm_arch_hardware_unsetup(void);
301 void kvm_arch_check_processor_compat(void *rtn
);
302 int kvm_arch_vcpu_runnable(struct kvm_vcpu
*vcpu
);
303 int kvm_arch_interrupt_allowed(struct kvm_vcpu
*vcpu
);
305 void kvm_free_physmem(struct kvm
*kvm
);
307 struct kvm
*kvm_arch_create_vm(void);
308 void kvm_arch_destroy_vm(struct kvm
*kvm
);
309 void kvm_free_all_assigned_devices(struct kvm
*kvm
);
310 void kvm_arch_sync_events(struct kvm
*kvm
);
312 int kvm_cpu_get_interrupt(struct kvm_vcpu
*v
);
313 int kvm_cpu_has_interrupt(struct kvm_vcpu
*v
);
314 int kvm_cpu_has_pending_timer(struct kvm_vcpu
*vcpu
);
315 void kvm_vcpu_kick(struct kvm_vcpu
*vcpu
);
317 int kvm_is_mmio_pfn(pfn_t pfn
);
319 struct kvm_irq_ack_notifier
{
320 struct hlist_node link
;
322 void (*irq_acked
)(struct kvm_irq_ack_notifier
*kian
);
325 #define KVM_ASSIGNED_MSIX_PENDING 0x1
326 struct kvm_guest_msix_entry
{
332 struct kvm_assigned_dev_kernel
{
333 struct kvm_irq_ack_notifier ack_notifier
;
334 struct work_struct interrupt_work
;
335 struct list_head list
;
339 unsigned int entries_nr
;
341 bool host_irq_disabled
;
342 struct msix_entry
*host_msix_entries
;
344 struct kvm_guest_msix_entry
*guest_msix_entries
;
345 unsigned long irq_requested_type
;
350 spinlock_t assigned_dev_lock
;
353 struct kvm_irq_mask_notifier
{
354 void (*func
)(struct kvm_irq_mask_notifier
*kimn
, bool masked
);
356 struct hlist_node link
;
359 void kvm_register_irq_mask_notifier(struct kvm
*kvm
, int irq
,
360 struct kvm_irq_mask_notifier
*kimn
);
361 void kvm_unregister_irq_mask_notifier(struct kvm
*kvm
, int irq
,
362 struct kvm_irq_mask_notifier
*kimn
);
363 void kvm_fire_mask_notifiers(struct kvm
*kvm
, int irq
, bool mask
);
365 int kvm_set_irq(struct kvm
*kvm
, int irq_source_id
, int irq
, int level
);
366 void kvm_notify_acked_irq(struct kvm
*kvm
, unsigned irqchip
, unsigned pin
);
367 void kvm_register_irq_ack_notifier(struct kvm
*kvm
,
368 struct kvm_irq_ack_notifier
*kian
);
369 void kvm_unregister_irq_ack_notifier(struct kvm_irq_ack_notifier
*kian
);
370 int kvm_request_irq_source_id(struct kvm
*kvm
);
371 void kvm_free_irq_source_id(struct kvm
*kvm
, int irq_source_id
);
373 /* For vcpu->arch.iommu_flags */
374 #define KVM_IOMMU_CACHE_COHERENCY 0x1
376 #ifdef CONFIG_IOMMU_API
377 int kvm_iommu_map_pages(struct kvm
*kvm
, gfn_t base_gfn
,
378 unsigned long npages
);
379 int kvm_iommu_map_guest(struct kvm
*kvm
);
380 int kvm_iommu_unmap_guest(struct kvm
*kvm
);
381 int kvm_assign_device(struct kvm
*kvm
,
382 struct kvm_assigned_dev_kernel
*assigned_dev
);
383 int kvm_deassign_device(struct kvm
*kvm
,
384 struct kvm_assigned_dev_kernel
*assigned_dev
);
385 #else /* CONFIG_IOMMU_API */
386 static inline int kvm_iommu_map_pages(struct kvm
*kvm
,
388 unsigned long npages
)
393 static inline int kvm_iommu_map_guest(struct kvm
*kvm
)
398 static inline int kvm_iommu_unmap_guest(struct kvm
*kvm
)
403 static inline int kvm_assign_device(struct kvm
*kvm
,
404 struct kvm_assigned_dev_kernel
*assigned_dev
)
409 static inline int kvm_deassign_device(struct kvm
*kvm
,
410 struct kvm_assigned_dev_kernel
*assigned_dev
)
414 #endif /* CONFIG_IOMMU_API */
416 static inline void kvm_guest_enter(void)
418 account_system_vtime(current
);
419 current
->flags
|= PF_VCPU
;
422 static inline void kvm_guest_exit(void)
424 account_system_vtime(current
);
425 current
->flags
&= ~PF_VCPU
;
428 static inline int memslot_id(struct kvm
*kvm
, struct kvm_memory_slot
*slot
)
430 return slot
- kvm
->memslots
;
433 static inline gpa_t
gfn_to_gpa(gfn_t gfn
)
435 return (gpa_t
)gfn
<< PAGE_SHIFT
;
438 static inline hpa_t
pfn_to_hpa(pfn_t pfn
)
440 return (hpa_t
)pfn
<< PAGE_SHIFT
;
443 static inline void kvm_migrate_timers(struct kvm_vcpu
*vcpu
)
445 set_bit(KVM_REQ_MIGRATE_TIMER
, &vcpu
->requests
);
453 struct kvm_stats_debugfs_item
{
456 enum kvm_stat_kind kind
;
457 struct dentry
*dentry
;
459 extern struct kvm_stats_debugfs_item debugfs_entries
[];
460 extern struct dentry
*kvm_debugfs_dir
;
462 #define KVMTRACE_5D(evt, vcpu, d1, d2, d3, d4, d5, name) \
463 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
464 vcpu, 5, d1, d2, d3, d4, d5)
465 #define KVMTRACE_4D(evt, vcpu, d1, d2, d3, d4, name) \
466 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
467 vcpu, 4, d1, d2, d3, d4, 0)
468 #define KVMTRACE_3D(evt, vcpu, d1, d2, d3, name) \
469 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
470 vcpu, 3, d1, d2, d3, 0, 0)
471 #define KVMTRACE_2D(evt, vcpu, d1, d2, name) \
472 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
473 vcpu, 2, d1, d2, 0, 0, 0)
474 #define KVMTRACE_1D(evt, vcpu, d1, name) \
475 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
476 vcpu, 1, d1, 0, 0, 0, 0)
477 #define KVMTRACE_0D(evt, vcpu, name) \
478 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
479 vcpu, 0, 0, 0, 0, 0, 0)
481 #ifdef CONFIG_KVM_TRACE
482 int kvm_trace_ioctl(unsigned int ioctl
, unsigned long arg
);
483 void kvm_trace_cleanup(void);
486 int kvm_trace_ioctl(unsigned int ioctl
, unsigned long arg
)
490 #define kvm_trace_cleanup() ((void)0)
493 #ifdef KVM_ARCH_WANT_MMU_NOTIFIER
494 static inline int mmu_notifier_retry(struct kvm_vcpu
*vcpu
, unsigned long mmu_seq
)
496 if (unlikely(vcpu
->kvm
->mmu_notifier_count
))
499 * Both reads happen under the mmu_lock and both values are
500 * modified under mmu_lock, so there's no need of smb_rmb()
501 * here in between, otherwise mmu_notifier_count should be
502 * read before mmu_notifier_seq, see
503 * mmu_notifier_invalidate_range_end write side.
505 if (vcpu
->kvm
->mmu_notifier_seq
!= mmu_seq
)
511 #ifdef CONFIG_HAVE_KVM_IRQCHIP
513 #define KVM_MAX_IRQ_ROUTES 1024
515 int kvm_setup_default_irq_routing(struct kvm
*kvm
);
516 int kvm_set_irq_routing(struct kvm
*kvm
,
517 const struct kvm_irq_routing_entry
*entries
,
520 void kvm_free_irq_routing(struct kvm
*kvm
);
524 static inline void kvm_free_irq_routing(struct kvm
*kvm
) {}