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
;
128 spinlock_t requests_lock
;
129 struct rw_semaphore slots_lock
;
130 struct mm_struct
*mm
; /* userspace tied to this vm */
132 struct kvm_memory_slot memslots
[KVM_MEMORY_SLOTS
+
133 KVM_PRIVATE_MEM_SLOTS
];
134 struct kvm_vcpu
*bsp_vcpu
;
135 struct kvm_vcpu
*vcpus
[KVM_MAX_VCPUS
];
136 struct list_head vm_list
;
138 struct kvm_io_bus mmio_bus
;
139 struct kvm_io_bus pio_bus
;
140 #ifdef CONFIG_HAVE_KVM_EVENTFD
143 struct list_head items
;
146 struct kvm_vm_stat stat
;
147 struct kvm_arch arch
;
148 atomic_t users_count
;
149 #ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
150 struct kvm_coalesced_mmio_dev
*coalesced_mmio_dev
;
151 struct kvm_coalesced_mmio_ring
*coalesced_mmio_ring
;
154 struct mutex irq_lock
;
155 #ifdef CONFIG_HAVE_KVM_IRQCHIP
156 struct list_head irq_routing
; /* of kvm_kernel_irq_routing_entry */
157 struct hlist_head mask_notifier_list
;
160 #ifdef KVM_ARCH_WANT_MMU_NOTIFIER
161 struct mmu_notifier mmu_notifier
;
162 unsigned long mmu_notifier_seq
;
163 long mmu_notifier_count
;
167 /* The guest did something we don't support. */
168 #define pr_unimpl(vcpu, fmt, ...) \
170 if (printk_ratelimit()) \
171 printk(KERN_ERR "kvm: %i: cpu%i " fmt, \
172 current->tgid, (vcpu)->vcpu_id , ## __VA_ARGS__); \
175 #define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
176 #define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
178 int kvm_vcpu_init(struct kvm_vcpu
*vcpu
, struct kvm
*kvm
, unsigned id
);
179 void kvm_vcpu_uninit(struct kvm_vcpu
*vcpu
);
181 void vcpu_load(struct kvm_vcpu
*vcpu
);
182 void vcpu_put(struct kvm_vcpu
*vcpu
);
184 int kvm_init(void *opaque
, unsigned int vcpu_size
,
185 struct module
*module
);
188 void kvm_get_kvm(struct kvm
*kvm
);
189 void kvm_put_kvm(struct kvm
*kvm
);
191 #define HPA_MSB ((sizeof(hpa_t) * 8) - 1)
192 #define HPA_ERR_MASK ((hpa_t)1 << HPA_MSB)
193 static inline int is_error_hpa(hpa_t hpa
) { return hpa
>> HPA_MSB
; }
194 struct page
*gva_to_page(struct kvm_vcpu
*vcpu
, gva_t gva
);
196 extern struct page
*bad_page
;
197 extern pfn_t bad_pfn
;
199 int is_error_page(struct page
*page
);
200 int is_error_pfn(pfn_t pfn
);
201 int kvm_is_error_hva(unsigned long addr
);
202 int kvm_set_memory_region(struct kvm
*kvm
,
203 struct kvm_userspace_memory_region
*mem
,
205 int __kvm_set_memory_region(struct kvm
*kvm
,
206 struct kvm_userspace_memory_region
*mem
,
208 int kvm_arch_set_memory_region(struct kvm
*kvm
,
209 struct kvm_userspace_memory_region
*mem
,
210 struct kvm_memory_slot old
,
212 void kvm_arch_flush_shadow(struct kvm
*kvm
);
213 gfn_t
unalias_gfn(struct kvm
*kvm
, gfn_t gfn
);
214 struct page
*gfn_to_page(struct kvm
*kvm
, gfn_t gfn
);
215 unsigned long gfn_to_hva(struct kvm
*kvm
, gfn_t gfn
);
216 void kvm_release_page_clean(struct page
*page
);
217 void kvm_release_page_dirty(struct page
*page
);
218 void kvm_set_page_dirty(struct page
*page
);
219 void kvm_set_page_accessed(struct page
*page
);
221 pfn_t
gfn_to_pfn(struct kvm
*kvm
, gfn_t gfn
);
222 void kvm_release_pfn_dirty(pfn_t
);
223 void kvm_release_pfn_clean(pfn_t pfn
);
224 void kvm_set_pfn_dirty(pfn_t pfn
);
225 void kvm_set_pfn_accessed(pfn_t pfn
);
226 void kvm_get_pfn(pfn_t pfn
);
228 int kvm_read_guest_page(struct kvm
*kvm
, gfn_t gfn
, void *data
, int offset
,
230 int kvm_read_guest_atomic(struct kvm
*kvm
, gpa_t gpa
, void *data
,
232 int kvm_read_guest(struct kvm
*kvm
, gpa_t gpa
, void *data
, unsigned long len
);
233 int kvm_write_guest_page(struct kvm
*kvm
, gfn_t gfn
, const void *data
,
234 int offset
, int len
);
235 int kvm_write_guest(struct kvm
*kvm
, gpa_t gpa
, const void *data
,
237 int kvm_clear_guest_page(struct kvm
*kvm
, gfn_t gfn
, int offset
, int len
);
238 int kvm_clear_guest(struct kvm
*kvm
, gpa_t gpa
, unsigned long len
);
239 struct kvm_memory_slot
*gfn_to_memslot(struct kvm
*kvm
, gfn_t gfn
);
240 int kvm_is_visible_gfn(struct kvm
*kvm
, gfn_t gfn
);
241 void mark_page_dirty(struct kvm
*kvm
, gfn_t gfn
);
243 void kvm_vcpu_block(struct kvm_vcpu
*vcpu
);
244 void kvm_resched(struct kvm_vcpu
*vcpu
);
245 void kvm_load_guest_fpu(struct kvm_vcpu
*vcpu
);
246 void kvm_put_guest_fpu(struct kvm_vcpu
*vcpu
);
247 void kvm_flush_remote_tlbs(struct kvm
*kvm
);
248 void kvm_reload_remote_mmus(struct kvm
*kvm
);
250 long kvm_arch_dev_ioctl(struct file
*filp
,
251 unsigned int ioctl
, unsigned long arg
);
252 long kvm_arch_vcpu_ioctl(struct file
*filp
,
253 unsigned int ioctl
, unsigned long arg
);
255 int kvm_dev_ioctl_check_extension(long ext
);
257 int kvm_get_dirty_log(struct kvm
*kvm
,
258 struct kvm_dirty_log
*log
, int *is_dirty
);
259 int kvm_vm_ioctl_get_dirty_log(struct kvm
*kvm
,
260 struct kvm_dirty_log
*log
);
262 int kvm_vm_ioctl_set_memory_region(struct kvm
*kvm
,
264 kvm_userspace_memory_region
*mem
,
266 long kvm_arch_vm_ioctl(struct file
*filp
,
267 unsigned int ioctl
, unsigned long arg
);
269 int kvm_arch_vcpu_ioctl_get_fpu(struct kvm_vcpu
*vcpu
, struct kvm_fpu
*fpu
);
270 int kvm_arch_vcpu_ioctl_set_fpu(struct kvm_vcpu
*vcpu
, struct kvm_fpu
*fpu
);
272 int kvm_arch_vcpu_ioctl_translate(struct kvm_vcpu
*vcpu
,
273 struct kvm_translation
*tr
);
275 int kvm_arch_vcpu_ioctl_get_regs(struct kvm_vcpu
*vcpu
, struct kvm_regs
*regs
);
276 int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu
*vcpu
, struct kvm_regs
*regs
);
277 int kvm_arch_vcpu_ioctl_get_sregs(struct kvm_vcpu
*vcpu
,
278 struct kvm_sregs
*sregs
);
279 int kvm_arch_vcpu_ioctl_set_sregs(struct kvm_vcpu
*vcpu
,
280 struct kvm_sregs
*sregs
);
281 int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu
*vcpu
,
282 struct kvm_mp_state
*mp_state
);
283 int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu
*vcpu
,
284 struct kvm_mp_state
*mp_state
);
285 int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu
*vcpu
,
286 struct kvm_guest_debug
*dbg
);
287 int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu
*vcpu
, struct kvm_run
*kvm_run
);
289 int kvm_arch_init(void *opaque
);
290 void kvm_arch_exit(void);
292 int kvm_arch_vcpu_init(struct kvm_vcpu
*vcpu
);
293 void kvm_arch_vcpu_uninit(struct kvm_vcpu
*vcpu
);
295 void kvm_arch_vcpu_free(struct kvm_vcpu
*vcpu
);
296 void kvm_arch_vcpu_load(struct kvm_vcpu
*vcpu
, int cpu
);
297 void kvm_arch_vcpu_put(struct kvm_vcpu
*vcpu
);
298 struct kvm_vcpu
*kvm_arch_vcpu_create(struct kvm
*kvm
, unsigned int id
);
299 int kvm_arch_vcpu_setup(struct kvm_vcpu
*vcpu
);
300 void kvm_arch_vcpu_destroy(struct kvm_vcpu
*vcpu
);
302 int kvm_arch_vcpu_reset(struct kvm_vcpu
*vcpu
);
303 void kvm_arch_hardware_enable(void *garbage
);
304 void kvm_arch_hardware_disable(void *garbage
);
305 int kvm_arch_hardware_setup(void);
306 void kvm_arch_hardware_unsetup(void);
307 void kvm_arch_check_processor_compat(void *rtn
);
308 int kvm_arch_vcpu_runnable(struct kvm_vcpu
*vcpu
);
309 int kvm_arch_interrupt_allowed(struct kvm_vcpu
*vcpu
);
311 void kvm_free_physmem(struct kvm
*kvm
);
313 struct kvm
*kvm_arch_create_vm(void);
314 void kvm_arch_destroy_vm(struct kvm
*kvm
);
315 void kvm_free_all_assigned_devices(struct kvm
*kvm
);
316 void kvm_arch_sync_events(struct kvm
*kvm
);
318 int kvm_cpu_get_interrupt(struct kvm_vcpu
*v
);
319 int kvm_cpu_has_interrupt(struct kvm_vcpu
*v
);
320 int kvm_cpu_has_pending_timer(struct kvm_vcpu
*vcpu
);
321 void kvm_vcpu_kick(struct kvm_vcpu
*vcpu
);
323 int kvm_is_mmio_pfn(pfn_t pfn
);
325 struct kvm_irq_ack_notifier
{
326 struct hlist_node link
;
328 void (*irq_acked
)(struct kvm_irq_ack_notifier
*kian
);
331 #define KVM_ASSIGNED_MSIX_PENDING 0x1
332 struct kvm_guest_msix_entry
{
338 struct kvm_assigned_dev_kernel
{
339 struct kvm_irq_ack_notifier ack_notifier
;
340 struct work_struct interrupt_work
;
341 struct list_head list
;
345 unsigned int entries_nr
;
347 bool host_irq_disabled
;
348 struct msix_entry
*host_msix_entries
;
350 struct kvm_guest_msix_entry
*guest_msix_entries
;
351 unsigned long irq_requested_type
;
356 spinlock_t assigned_dev_lock
;
359 struct kvm_irq_mask_notifier
{
360 void (*func
)(struct kvm_irq_mask_notifier
*kimn
, bool masked
);
362 struct hlist_node link
;
365 void kvm_register_irq_mask_notifier(struct kvm
*kvm
, int irq
,
366 struct kvm_irq_mask_notifier
*kimn
);
367 void kvm_unregister_irq_mask_notifier(struct kvm
*kvm
, int irq
,
368 struct kvm_irq_mask_notifier
*kimn
);
369 void kvm_fire_mask_notifiers(struct kvm
*kvm
, int irq
, bool mask
);
371 int kvm_set_irq(struct kvm
*kvm
, int irq_source_id
, int irq
, int level
);
372 void kvm_notify_acked_irq(struct kvm
*kvm
, unsigned irqchip
, unsigned pin
);
373 void kvm_register_irq_ack_notifier(struct kvm
*kvm
,
374 struct kvm_irq_ack_notifier
*kian
);
375 void kvm_unregister_irq_ack_notifier(struct kvm
*kvm
,
376 struct kvm_irq_ack_notifier
*kian
);
377 int kvm_request_irq_source_id(struct kvm
*kvm
);
378 void kvm_free_irq_source_id(struct kvm
*kvm
, int irq_source_id
);
380 /* For vcpu->arch.iommu_flags */
381 #define KVM_IOMMU_CACHE_COHERENCY 0x1
383 #ifdef CONFIG_IOMMU_API
384 int kvm_iommu_map_pages(struct kvm
*kvm
, gfn_t base_gfn
,
385 unsigned long npages
);
386 int kvm_iommu_map_guest(struct kvm
*kvm
);
387 int kvm_iommu_unmap_guest(struct kvm
*kvm
);
388 int kvm_assign_device(struct kvm
*kvm
,
389 struct kvm_assigned_dev_kernel
*assigned_dev
);
390 int kvm_deassign_device(struct kvm
*kvm
,
391 struct kvm_assigned_dev_kernel
*assigned_dev
);
392 #else /* CONFIG_IOMMU_API */
393 static inline int kvm_iommu_map_pages(struct kvm
*kvm
,
395 unsigned long npages
)
400 static inline int kvm_iommu_map_guest(struct kvm
*kvm
)
405 static inline int kvm_iommu_unmap_guest(struct kvm
*kvm
)
410 static inline int kvm_assign_device(struct kvm
*kvm
,
411 struct kvm_assigned_dev_kernel
*assigned_dev
)
416 static inline int kvm_deassign_device(struct kvm
*kvm
,
417 struct kvm_assigned_dev_kernel
*assigned_dev
)
421 #endif /* CONFIG_IOMMU_API */
423 static inline void kvm_guest_enter(void)
425 account_system_vtime(current
);
426 current
->flags
|= PF_VCPU
;
429 static inline void kvm_guest_exit(void)
431 account_system_vtime(current
);
432 current
->flags
&= ~PF_VCPU
;
435 static inline int memslot_id(struct kvm
*kvm
, struct kvm_memory_slot
*slot
)
437 return slot
- kvm
->memslots
;
440 static inline gpa_t
gfn_to_gpa(gfn_t gfn
)
442 return (gpa_t
)gfn
<< PAGE_SHIFT
;
445 static inline hpa_t
pfn_to_hpa(pfn_t pfn
)
447 return (hpa_t
)pfn
<< PAGE_SHIFT
;
450 static inline void kvm_migrate_timers(struct kvm_vcpu
*vcpu
)
452 set_bit(KVM_REQ_MIGRATE_TIMER
, &vcpu
->requests
);
460 struct kvm_stats_debugfs_item
{
463 enum kvm_stat_kind kind
;
464 struct dentry
*dentry
;
466 extern struct kvm_stats_debugfs_item debugfs_entries
[];
467 extern struct dentry
*kvm_debugfs_dir
;
469 #define KVMTRACE_5D(evt, vcpu, d1, d2, d3, d4, d5, name) \
470 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
471 vcpu, 5, d1, d2, d3, d4, d5)
472 #define KVMTRACE_4D(evt, vcpu, d1, d2, d3, d4, name) \
473 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
474 vcpu, 4, d1, d2, d3, d4, 0)
475 #define KVMTRACE_3D(evt, vcpu, d1, d2, d3, name) \
476 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
477 vcpu, 3, d1, d2, d3, 0, 0)
478 #define KVMTRACE_2D(evt, vcpu, d1, d2, name) \
479 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
480 vcpu, 2, d1, d2, 0, 0, 0)
481 #define KVMTRACE_1D(evt, vcpu, d1, name) \
482 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
483 vcpu, 1, d1, 0, 0, 0, 0)
484 #define KVMTRACE_0D(evt, vcpu, name) \
485 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
486 vcpu, 0, 0, 0, 0, 0, 0)
488 #ifdef CONFIG_KVM_TRACE
489 int kvm_trace_ioctl(unsigned int ioctl
, unsigned long arg
);
490 void kvm_trace_cleanup(void);
493 int kvm_trace_ioctl(unsigned int ioctl
, unsigned long arg
)
497 #define kvm_trace_cleanup() ((void)0)
500 #ifdef KVM_ARCH_WANT_MMU_NOTIFIER
501 static inline int mmu_notifier_retry(struct kvm_vcpu
*vcpu
, unsigned long mmu_seq
)
503 if (unlikely(vcpu
->kvm
->mmu_notifier_count
))
506 * Both reads happen under the mmu_lock and both values are
507 * modified under mmu_lock, so there's no need of smb_rmb()
508 * here in between, otherwise mmu_notifier_count should be
509 * read before mmu_notifier_seq, see
510 * mmu_notifier_invalidate_range_end write side.
512 if (vcpu
->kvm
->mmu_notifier_seq
!= mmu_seq
)
518 #ifdef CONFIG_HAVE_KVM_IRQCHIP
520 #define KVM_MAX_IRQ_ROUTES 1024
522 int kvm_setup_default_irq_routing(struct kvm
*kvm
);
523 int kvm_set_irq_routing(struct kvm
*kvm
,
524 const struct kvm_irq_routing_entry
*entries
,
527 void kvm_free_irq_routing(struct kvm
*kvm
);
531 static inline void kvm_free_irq_routing(struct kvm
*kvm
) {}
535 #ifdef CONFIG_HAVE_KVM_EVENTFD
537 void kvm_irqfd_init(struct kvm
*kvm
);
538 int kvm_irqfd(struct kvm
*kvm
, int fd
, int gsi
, int flags
);
539 void kvm_irqfd_release(struct kvm
*kvm
);
543 static inline void kvm_irqfd_init(struct kvm
*kvm
) {}
544 static inline int kvm_irqfd(struct kvm
*kvm
, int fd
, int gsi
, int flags
)
549 static inline void kvm_irqfd_release(struct kvm
*kvm
) {}
551 #endif /* CONFIG_HAVE_KVM_EVENTFD */
553 static inline bool kvm_vcpu_is_bsp(struct kvm_vcpu
*vcpu
)
555 return vcpu
->kvm
->bsp_vcpu
== vcpu
;