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 <asm/signal.h>
21 #include <linux/kvm.h>
22 #include <linux/kvm_para.h>
24 #include <linux/kvm_types.h>
26 #include <asm/kvm_host.h>
29 * vcpu->requests bit members
31 #define KVM_REQ_TLB_FLUSH 0
32 #define KVM_REQ_MIGRATE_TIMER 1
33 #define KVM_REQ_REPORT_TPR_ACCESS 2
34 #define KVM_REQ_MMU_RELOAD 3
35 #define KVM_REQ_TRIPLE_FAULT 4
36 #define KVM_REQ_PENDING_TIMER 5
37 #define KVM_REQ_UNHALT 6
38 #define KVM_REQ_MMU_SYNC 7
40 #define KVM_USERSPACE_IRQ_SOURCE_ID 0
43 extern struct kmem_cache *kvm_vcpu_cache;
46 * It would be nice to use something smarter than a linear search, TBD...
47 * Thankfully we dont expect many devices to register (famous last words :),
48 * so until then it will suffice. At least its abstracted so we can change
53 #define NR_IOBUS_DEVS 6
54 struct kvm_io_device *devs[NR_IOBUS_DEVS];
57 void kvm_io_bus_init(struct kvm_io_bus *bus);
58 void kvm_io_bus_destroy(struct kvm_io_bus *bus);
59 struct kvm_io_device *kvm_io_bus_find_dev(struct kvm_io_bus *bus,
60 gpa_t addr, int len, int is_write);
61 void kvm_io_bus_register_dev(struct kvm_io_bus *bus,
62 struct kvm_io_device *dev);
66 #ifdef CONFIG_PREEMPT_NOTIFIERS
67 struct preempt_notifier preempt_notifier;
74 unsigned long requests;
75 struct kvm_guest_debug guest_debug;
81 struct kvm_vcpu_stat stat;
83 #ifdef CONFIG_HAS_IOMEM
85 int mmio_read_completed;
88 unsigned char mmio_data[8];
92 struct kvm_vcpu_arch arch;
95 struct kvm_memory_slot {
100 unsigned long *dirty_bitmap;
102 unsigned long rmap_pde;
105 unsigned long userspace_addr;
110 struct mutex lock; /* protects the vcpus array and APIC accesses */
112 struct rw_semaphore slots_lock;
113 struct mm_struct *mm; /* userspace tied to this vm */
115 struct kvm_memory_slot memslots[KVM_MEMORY_SLOTS +
116 KVM_PRIVATE_MEM_SLOTS];
117 struct kvm_vcpu *vcpus[KVM_MAX_VCPUS];
118 struct list_head vm_list;
119 struct kvm_io_bus mmio_bus;
120 struct kvm_io_bus pio_bus;
121 struct kvm_vm_stat stat;
122 struct kvm_arch arch;
123 atomic_t users_count;
124 #ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
125 struct kvm_coalesced_mmio_dev *coalesced_mmio_dev;
126 struct kvm_coalesced_mmio_ring *coalesced_mmio_ring;
129 #ifdef KVM_ARCH_WANT_MMU_NOTIFIER
130 struct mmu_notifier mmu_notifier;
131 unsigned long mmu_notifier_seq;
132 long mmu_notifier_count;
136 /* The guest did something we don't support. */
137 #define pr_unimpl(vcpu, fmt, ...) \
139 if (printk_ratelimit()) \
140 printk(KERN_ERR "kvm: %i: cpu%i " fmt, \
141 current->tgid, (vcpu)->vcpu_id , ## __VA_ARGS__); \
144 #define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
145 #define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
147 int kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id);
148 void kvm_vcpu_uninit(struct kvm_vcpu *vcpu);
150 void vcpu_load(struct kvm_vcpu *vcpu);
151 void vcpu_put(struct kvm_vcpu *vcpu);
153 int kvm_init(void *opaque, unsigned int vcpu_size,
154 struct module *module);
157 void kvm_get_kvm(struct kvm *kvm);
158 void kvm_put_kvm(struct kvm *kvm);
160 #define HPA_MSB ((sizeof(hpa_t) * 8) - 1)
161 #define HPA_ERR_MASK ((hpa_t)1 << HPA_MSB)
162 static inline int is_error_hpa(hpa_t hpa) { return hpa >> HPA_MSB; }
163 struct page *gva_to_page(struct kvm_vcpu *vcpu, gva_t gva);
165 extern struct page *bad_page;
166 extern pfn_t bad_pfn;
168 int is_error_page(struct page *page);
169 int is_error_pfn(pfn_t pfn);
170 int kvm_is_error_hva(unsigned long addr);
171 int kvm_set_memory_region(struct kvm *kvm,
172 struct kvm_userspace_memory_region *mem,
174 int __kvm_set_memory_region(struct kvm *kvm,
175 struct kvm_userspace_memory_region *mem,
177 int kvm_arch_set_memory_region(struct kvm *kvm,
178 struct kvm_userspace_memory_region *mem,
179 struct kvm_memory_slot old,
181 void kvm_arch_flush_shadow(struct kvm *kvm);
182 gfn_t unalias_gfn(struct kvm *kvm, gfn_t gfn);
183 struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn);
184 unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn);
185 void kvm_release_page_clean(struct page *page);
186 void kvm_release_page_dirty(struct page *page);
187 void kvm_set_page_dirty(struct page *page);
188 void kvm_set_page_accessed(struct page *page);
190 pfn_t gfn_to_pfn(struct kvm *kvm, gfn_t gfn);
191 void kvm_release_pfn_dirty(pfn_t);
192 void kvm_release_pfn_clean(pfn_t pfn);
193 void kvm_set_pfn_dirty(pfn_t pfn);
194 void kvm_set_pfn_accessed(pfn_t pfn);
195 void kvm_get_pfn(pfn_t pfn);
197 int kvm_read_guest_page(struct kvm *kvm, gfn_t gfn, void *data, int offset,
199 int kvm_read_guest_atomic(struct kvm *kvm, gpa_t gpa, void *data,
201 int kvm_read_guest(struct kvm *kvm, gpa_t gpa, void *data, unsigned long len);
202 int kvm_write_guest_page(struct kvm *kvm, gfn_t gfn, const void *data,
203 int offset, int len);
204 int kvm_write_guest(struct kvm *kvm, gpa_t gpa, const void *data,
206 int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len);
207 int kvm_clear_guest(struct kvm *kvm, gpa_t gpa, unsigned long len);
208 struct kvm_memory_slot *gfn_to_memslot(struct kvm *kvm, gfn_t gfn);
209 int kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn);
210 void mark_page_dirty(struct kvm *kvm, gfn_t gfn);
212 void kvm_vcpu_block(struct kvm_vcpu *vcpu);
213 void kvm_resched(struct kvm_vcpu *vcpu);
214 void kvm_load_guest_fpu(struct kvm_vcpu *vcpu);
215 void kvm_put_guest_fpu(struct kvm_vcpu *vcpu);
216 void kvm_flush_remote_tlbs(struct kvm *kvm);
217 void kvm_reload_remote_mmus(struct kvm *kvm);
219 long kvm_arch_dev_ioctl(struct file *filp,
220 unsigned int ioctl, unsigned long arg);
221 long kvm_arch_vcpu_ioctl(struct file *filp,
222 unsigned int ioctl, unsigned long arg);
223 void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
224 void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
226 int kvm_dev_ioctl_check_extension(long ext);
228 int kvm_get_dirty_log(struct kvm *kvm,
229 struct kvm_dirty_log *log, int *is_dirty);
230 int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
231 struct kvm_dirty_log *log);
233 int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
235 kvm_userspace_memory_region *mem,
237 long kvm_arch_vm_ioctl(struct file *filp,
238 unsigned int ioctl, unsigned long arg);
239 void kvm_arch_destroy_vm(struct kvm *kvm);
241 int kvm_arch_vcpu_ioctl_get_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu);
242 int kvm_arch_vcpu_ioctl_set_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu);
244 int kvm_arch_vcpu_ioctl_translate(struct kvm_vcpu *vcpu,
245 struct kvm_translation *tr);
247 int kvm_arch_vcpu_ioctl_get_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs);
248 int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs);
249 int kvm_arch_vcpu_ioctl_get_sregs(struct kvm_vcpu *vcpu,
250 struct kvm_sregs *sregs);
251 int kvm_arch_vcpu_ioctl_set_sregs(struct kvm_vcpu *vcpu,
252 struct kvm_sregs *sregs);
253 int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
254 struct kvm_mp_state *mp_state);
255 int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
256 struct kvm_mp_state *mp_state);
257 int kvm_arch_vcpu_ioctl_debug_guest(struct kvm_vcpu *vcpu,
258 struct kvm_debug_guest *dbg);
259 int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run);
261 int kvm_arch_init(void *opaque);
262 void kvm_arch_exit(void);
264 int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu);
265 void kvm_arch_vcpu_uninit(struct kvm_vcpu *vcpu);
267 void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu);
268 void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
269 void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
270 struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id);
271 int kvm_arch_vcpu_setup(struct kvm_vcpu *vcpu);
272 void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu);
274 int kvm_arch_vcpu_reset(struct kvm_vcpu *vcpu);
275 void kvm_arch_hardware_enable(void *garbage);
276 void kvm_arch_hardware_disable(void *garbage);
277 int kvm_arch_hardware_setup(void);
278 void kvm_arch_hardware_unsetup(void);
279 void kvm_arch_check_processor_compat(void *rtn);
280 int kvm_arch_vcpu_runnable(struct kvm_vcpu *vcpu);
282 void kvm_free_physmem(struct kvm *kvm);
284 struct kvm *kvm_arch_create_vm(void);
285 void kvm_arch_destroy_vm(struct kvm *kvm);
286 void kvm_free_all_assigned_devices(struct kvm *kvm);
288 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
289 int kvm_cpu_has_interrupt(struct kvm_vcpu *v);
290 int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu);
291 void kvm_vcpu_kick(struct kvm_vcpu *vcpu);
293 int kvm_is_mmio_pfn(pfn_t pfn);
295 struct kvm_irq_ack_notifier {
296 struct hlist_node link;
298 void (*irq_acked)(struct kvm_irq_ack_notifier *kian);
301 struct kvm_assigned_dev_kernel {
302 struct kvm_irq_ack_notifier ack_notifier;
303 struct work_struct interrupt_work;
304 struct list_head list;
315 void kvm_set_irq(struct kvm *kvm, int irq_source_id, int irq, int level);
316 void kvm_notify_acked_irq(struct kvm *kvm, unsigned gsi);
317 void kvm_register_irq_ack_notifier(struct kvm *kvm,
318 struct kvm_irq_ack_notifier *kian);
319 void kvm_unregister_irq_ack_notifier(struct kvm *kvm,
320 struct kvm_irq_ack_notifier *kian);
321 int kvm_request_irq_source_id(struct kvm *kvm);
322 void kvm_free_irq_source_id(struct kvm *kvm, int irq_source_id);
325 int kvm_iommu_map_pages(struct kvm *kvm, gfn_t base_gfn,
326 unsigned long npages);
327 int kvm_iommu_map_guest(struct kvm *kvm,
328 struct kvm_assigned_dev_kernel *assigned_dev);
329 int kvm_iommu_unmap_guest(struct kvm *kvm);
330 #else /* CONFIG_DMAR */
331 static inline int kvm_iommu_map_pages(struct kvm *kvm,
333 unsigned long npages)
338 static inline int kvm_iommu_map_guest(struct kvm *kvm,
339 struct kvm_assigned_dev_kernel
345 static inline int kvm_iommu_unmap_guest(struct kvm *kvm)
349 #endif /* CONFIG_DMAR */
351 static inline void kvm_guest_enter(void)
353 account_system_vtime(current);
354 current->flags |= PF_VCPU;
357 static inline void kvm_guest_exit(void)
359 account_system_vtime(current);
360 current->flags &= ~PF_VCPU;
363 static inline int memslot_id(struct kvm *kvm, struct kvm_memory_slot *slot)
365 return slot - kvm->memslots;
368 static inline gpa_t gfn_to_gpa(gfn_t gfn)
370 return (gpa_t)gfn << PAGE_SHIFT;
373 static inline hpa_t pfn_to_hpa(pfn_t pfn)
375 return (hpa_t)pfn << PAGE_SHIFT;
378 static inline void kvm_migrate_timers(struct kvm_vcpu *vcpu)
380 set_bit(KVM_REQ_MIGRATE_TIMER, &vcpu->requests);
388 struct kvm_stats_debugfs_item {
391 enum kvm_stat_kind kind;
392 struct dentry *dentry;
394 extern struct kvm_stats_debugfs_item debugfs_entries[];
395 extern struct dentry *kvm_debugfs_dir;
397 #define KVMTRACE_5D(evt, vcpu, d1, d2, d3, d4, d5, name) \
398 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
399 vcpu, 5, d1, d2, d3, d4, d5)
400 #define KVMTRACE_4D(evt, vcpu, d1, d2, d3, d4, name) \
401 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
402 vcpu, 4, d1, d2, d3, d4, 0)
403 #define KVMTRACE_3D(evt, vcpu, d1, d2, d3, name) \
404 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
405 vcpu, 3, d1, d2, d3, 0, 0)
406 #define KVMTRACE_2D(evt, vcpu, d1, d2, name) \
407 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
408 vcpu, 2, d1, d2, 0, 0, 0)
409 #define KVMTRACE_1D(evt, vcpu, d1, name) \
410 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
411 vcpu, 1, d1, 0, 0, 0, 0)
412 #define KVMTRACE_0D(evt, vcpu, name) \
413 trace_mark(kvm_trace_##name, "%u %p %u %u %u %u %u %u", KVM_TRC_##evt, \
414 vcpu, 0, 0, 0, 0, 0, 0)
416 #ifdef CONFIG_KVM_TRACE
417 int kvm_trace_ioctl(unsigned int ioctl, unsigned long arg);
418 void kvm_trace_cleanup(void);
421 int kvm_trace_ioctl(unsigned int ioctl, unsigned long arg)
425 #define kvm_trace_cleanup() ((void)0)
428 #ifdef KVM_ARCH_WANT_MMU_NOTIFIER
429 static inline int mmu_notifier_retry(struct kvm_vcpu *vcpu, unsigned long mmu_seq)
431 if (unlikely(vcpu->kvm->mmu_notifier_count))
434 * Both reads happen under the mmu_lock and both values are
435 * modified under mmu_lock, so there's no need of smb_rmb()
436 * here in between, otherwise mmu_notifier_count should be
437 * read before mmu_notifier_seq, see
438 * mmu_notifier_invalidate_range_end write side.
440 if (vcpu->kvm->mmu_notifier_seq != mmu_seq)