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/list.h>
11 #include <linux/mutex.h>
12 #include <linux/spinlock.h>
16 #include <linux/kvm.h>
18 #define CR0_PE_MASK (1ULL << 0)
19 #define CR0_TS_MASK (1ULL << 3)
20 #define CR0_NE_MASK (1ULL << 5)
21 #define CR0_WP_MASK (1ULL << 16)
22 #define CR0_NW_MASK (1ULL << 29)
23 #define CR0_CD_MASK (1ULL << 30)
24 #define CR0_PG_MASK (1ULL << 31)
26 #define CR3_WPT_MASK (1ULL << 3)
27 #define CR3_PCD_MASK (1ULL << 4)
29 #define CR3_RESEVED_BITS 0x07ULL
30 #define CR3_L_MODE_RESEVED_BITS (~((1ULL << 40) - 1) | 0x0fe7ULL)
31 #define CR3_FLAGS_MASK ((1ULL << 5) - 1)
33 #define CR4_VME_MASK (1ULL << 0)
34 #define CR4_PSE_MASK (1ULL << 4)
35 #define CR4_PAE_MASK (1ULL << 5)
36 #define CR4_PGE_MASK (1ULL << 7)
37 #define CR4_VMXE_MASK (1ULL << 13)
39 #define KVM_GUEST_CR0_MASK \
40 (CR0_PG_MASK | CR0_PE_MASK | CR0_WP_MASK | CR0_NE_MASK \
41 | CR0_NW_MASK | CR0_CD_MASK)
42 #define KVM_VM_CR0_ALWAYS_ON \
43 (CR0_PG_MASK | CR0_PE_MASK | CR0_WP_MASK | CR0_NE_MASK)
44 #define KVM_GUEST_CR4_MASK \
45 (CR4_PSE_MASK | CR4_PAE_MASK | CR4_PGE_MASK | CR4_VMXE_MASK | CR4_VME_MASK)
46 #define KVM_PMODE_VM_CR4_ALWAYS_ON (CR4_VMXE_MASK | CR4_PAE_MASK)
47 #define KVM_RMODE_VM_CR4_ALWAYS_ON (CR4_VMXE_MASK | CR4_PAE_MASK | CR4_VME_MASK)
49 #define INVALID_PAGE (~(hpa_t)0)
50 #define UNMAPPED_GVA (~(gpa_t)0)
52 #define KVM_MAX_VCPUS 1
53 #define KVM_MEMORY_SLOTS 4
54 #define KVM_NUM_MMU_PAGES 256
56 #define FX_IMAGE_SIZE 512
57 #define FX_IMAGE_ALIGN 16
58 #define FX_BUF_SIZE (2 * FX_IMAGE_SIZE + FX_IMAGE_ALIGN)
68 #define SELECTOR_TI_MASK (1 << 2)
69 #define SELECTOR_RPL_MASK 0x03
76 * gva - guest virtual address
77 * gpa - guest physical address
78 * gfn - guest frame number
79 * hva - host virtual address
80 * hpa - host physical address
81 * hfn - host frame number
84 typedef unsigned long gva_t;
86 typedef unsigned long gfn_t;
88 typedef unsigned long hva_t;
90 typedef unsigned long hfn_t;
93 struct list_head link;
95 unsigned long slot_bitmap; /* One bit set per slot which has memory
96 * in this shadow page.
98 int global; /* Set if all ptes in this page are global */
108 #define vmx_msr_entry kvm_msr_entry
113 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
114 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
118 void (*new_cr3)(struct kvm_vcpu *vcpu);
119 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err);
120 void (*inval_page)(struct kvm_vcpu *vcpu, gva_t gva);
121 void (*free)(struct kvm_vcpu *vcpu);
122 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva);
125 int shadow_root_level;
130 struct kvm_guest_debug {
173 struct vcpu_svm *svm;
178 int interrupt_window_open;
179 unsigned long irq_summary; /* bit vector: 1 per word in irq_pending */
180 #define NR_IRQ_WORDS KVM_IRQ_BITMAP_SIZE(unsigned long)
181 unsigned long irq_pending[NR_IRQ_WORDS];
182 unsigned long regs[NR_VCPU_REGS]; /* for rsp: vcpu_load_rsp_rip() */
183 unsigned long rip; /* needs vcpu_load_rsp_rip() */
190 u64 pdptrs[4]; /* pae */
194 struct vmx_msr_entry *guest_msrs;
195 struct vmx_msr_entry *host_msrs;
197 struct list_head free_pages;
198 struct kvm_mmu_page page_header_buf[KVM_NUM_MMU_PAGES];
201 struct kvm_guest_debug guest_debug;
203 char fx_buf[FX_BUF_SIZE];
205 char *guest_fx_image;
208 int mmio_read_completed;
211 unsigned char mmio_data[8];
212 gpa_t mmio_phys_addr;
217 struct kvm_save_segment {
222 } tr, es, ds, fs, gs;
226 struct kvm_memory_slot {
228 unsigned long npages;
230 struct page **phys_mem;
231 unsigned long *dirty_bitmap;
235 spinlock_t lock; /* protects everything except vcpus */
237 struct kvm_memory_slot memslots[KVM_MEMORY_SLOTS];
238 struct list_head active_mmu_pages;
239 struct kvm_vcpu vcpus[KVM_MAX_VCPUS];
240 int memory_config_version;
242 unsigned long rmap_overflow;
255 u32 irq_window_exits;
257 u32 request_irq_exits;
261 struct descriptor_table {
264 } __attribute__((packed));
266 struct kvm_arch_ops {
267 int (*cpu_has_kvm_support)(void); /* __init */
268 int (*disabled_by_bios)(void); /* __init */
269 void (*hardware_enable)(void *dummy); /* __init */
270 void (*hardware_disable)(void *dummy);
271 int (*hardware_setup)(void); /* __init */
272 void (*hardware_unsetup)(void); /* __exit */
274 int (*vcpu_create)(struct kvm_vcpu *vcpu);
275 void (*vcpu_free)(struct kvm_vcpu *vcpu);
277 struct kvm_vcpu *(*vcpu_load)(struct kvm_vcpu *vcpu);
278 void (*vcpu_put)(struct kvm_vcpu *vcpu);
280 int (*set_guest_debug)(struct kvm_vcpu *vcpu,
281 struct kvm_debug_guest *dbg);
282 int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
283 int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
284 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
285 void (*get_segment)(struct kvm_vcpu *vcpu,
286 struct kvm_segment *var, int seg);
287 void (*set_segment)(struct kvm_vcpu *vcpu,
288 struct kvm_segment *var, int seg);
289 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
290 void (*decache_cr0_cr4_guest_bits)(struct kvm_vcpu *vcpu);
291 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
292 void (*set_cr0_no_modeswitch)(struct kvm_vcpu *vcpu,
294 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
295 void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
296 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
297 void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
298 void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
299 void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
300 void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
301 unsigned long (*get_dr)(struct kvm_vcpu *vcpu, int dr);
302 void (*set_dr)(struct kvm_vcpu *vcpu, int dr, unsigned long value,
304 void (*cache_regs)(struct kvm_vcpu *vcpu);
305 void (*decache_regs)(struct kvm_vcpu *vcpu);
306 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
307 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
309 void (*invlpg)(struct kvm_vcpu *vcpu, gva_t addr);
310 void (*tlb_flush)(struct kvm_vcpu *vcpu);
311 void (*inject_page_fault)(struct kvm_vcpu *vcpu,
312 unsigned long addr, u32 err_code);
314 void (*inject_gp)(struct kvm_vcpu *vcpu, unsigned err_code);
316 int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
317 int (*vcpu_setup)(struct kvm_vcpu *vcpu);
318 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
321 extern struct kvm_stat kvm_stat;
322 extern struct kvm_arch_ops *kvm_arch_ops;
324 #define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
325 #define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
327 int kvm_init_arch(struct kvm_arch_ops *ops, struct module *module);
328 void kvm_exit_arch(void);
330 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
331 int kvm_mmu_create(struct kvm_vcpu *vcpu);
332 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
334 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
335 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
337 hpa_t gpa_to_hpa(struct kvm_vcpu *vcpu, gpa_t gpa);
338 #define HPA_MSB ((sizeof(hpa_t) * 8) - 1)
339 #define HPA_ERR_MASK ((hpa_t)1 << HPA_MSB)
340 static inline int is_error_hpa(hpa_t hpa) { return hpa >> HPA_MSB; }
341 hpa_t gva_to_hpa(struct kvm_vcpu *vcpu, gva_t gva);
343 void kvm_emulator_want_group7_invlpg(void);
345 extern hpa_t bad_page_address;
347 static inline struct page *gfn_to_page(struct kvm_memory_slot *slot, gfn_t gfn)
349 return slot->phys_mem[gfn - slot->base_gfn];
352 struct kvm_memory_slot *gfn_to_memslot(struct kvm *kvm, gfn_t gfn);
353 void mark_page_dirty(struct kvm *kvm, gfn_t gfn);
355 enum emulation_result {
356 EMULATE_DONE, /* no further processing */
357 EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
358 EMULATE_FAIL, /* can't emulate this instruction */
361 int emulate_instruction(struct kvm_vcpu *vcpu, struct kvm_run *run,
362 unsigned long cr2, u16 error_code);
363 void realmode_lgdt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
364 void realmode_lidt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
365 void realmode_lmsw(struct kvm_vcpu *vcpu, unsigned long msw,
366 unsigned long *rflags);
368 unsigned long realmode_get_cr(struct kvm_vcpu *vcpu, int cr);
369 void realmode_set_cr(struct kvm_vcpu *vcpu, int cr, unsigned long value,
370 unsigned long *rflags);
372 struct x86_emulate_ctxt;
374 int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address);
375 int emulate_clts(struct kvm_vcpu *vcpu);
376 int emulator_get_dr(struct x86_emulate_ctxt* ctxt, int dr,
377 unsigned long *dest);
378 int emulator_set_dr(struct x86_emulate_ctxt *ctxt, int dr,
379 unsigned long value);
381 void set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
382 void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr0);
383 void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
384 void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
385 void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
387 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
388 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
390 void fx_init(struct kvm_vcpu *vcpu);
392 void load_msrs(struct vmx_msr_entry *e, int n);
393 void save_msrs(struct vmx_msr_entry *e, int n);
394 void kvm_resched(struct kvm_vcpu *vcpu);
396 int kvm_read_guest(struct kvm_vcpu *vcpu,
401 int kvm_write_guest(struct kvm_vcpu *vcpu,
406 unsigned long segment_base(u16 selector);
408 static inline struct page *_gfn_to_page(struct kvm *kvm, gfn_t gfn)
410 struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);
411 return (slot) ? slot->phys_mem[gfn - slot->base_gfn] : NULL;
414 static inline int is_long_mode(struct kvm_vcpu *vcpu)
417 return vcpu->shadow_efer & EFER_LME;
423 static inline int is_pae(struct kvm_vcpu *vcpu)
425 return vcpu->cr4 & CR4_PAE_MASK;
428 static inline int is_pse(struct kvm_vcpu *vcpu)
430 return vcpu->cr4 & CR4_PSE_MASK;
433 static inline int is_paging(struct kvm_vcpu *vcpu)
435 return vcpu->cr0 & CR0_PG_MASK;
438 static inline int memslot_id(struct kvm *kvm, struct kvm_memory_slot *slot)
440 return slot - kvm->memslots;
443 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
445 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
447 return (struct kvm_mmu_page *)page->private;
450 static inline u16 read_fs(void)
453 asm ("mov %%fs, %0" : "=g"(seg));
457 static inline u16 read_gs(void)
460 asm ("mov %%gs, %0" : "=g"(seg));
464 static inline u16 read_ldt(void)
467 asm ("sldt %0" : "=g"(ldt));
471 static inline void load_fs(u16 sel)
473 asm ("mov %0, %%fs" : : "rm"(sel));
476 static inline void load_gs(u16 sel)
478 asm ("mov %0, %%gs" : : "rm"(sel));
482 static inline void load_ldt(u16 sel)
484 asm ("lldt %0" : : "g"(sel));
488 static inline void get_idt(struct descriptor_table *table)
490 asm ("sidt %0" : "=m"(*table));
493 static inline void get_gdt(struct descriptor_table *table)
495 asm ("sgdt %0" : "=m"(*table));
498 static inline unsigned long read_tr_base(void)
501 asm ("str %0" : "=g"(tr));
502 return segment_base(tr);
506 static inline unsigned long read_msr(unsigned long msr)
515 static inline void fx_save(void *image)
517 asm ("fxsave (%0)":: "r" (image));
520 static inline void fx_restore(void *image)
522 asm ("fxrstor (%0)":: "r" (image));
525 static inline void fpu_init(void)
530 static inline u32 get_rdx_init_val(void)
532 return 0x600; /* P6 family */
535 #define ASM_VMX_VMCLEAR_RAX ".byte 0x66, 0x0f, 0xc7, 0x30"
536 #define ASM_VMX_VMLAUNCH ".byte 0x0f, 0x01, 0xc2"
537 #define ASM_VMX_VMRESUME ".byte 0x0f, 0x01, 0xc3"
538 #define ASM_VMX_VMPTRLD_RAX ".byte 0x0f, 0xc7, 0x30"
539 #define ASM_VMX_VMREAD_RDX_RAX ".byte 0x0f, 0x78, 0xd0"
540 #define ASM_VMX_VMWRITE_RAX_RDX ".byte 0x0f, 0x79, 0xd0"
541 #define ASM_VMX_VMWRITE_RSP_RDX ".byte 0x0f, 0x79, 0xd4"
542 #define ASM_VMX_VMXOFF ".byte 0x0f, 0x01, 0xc4"
543 #define ASM_VMX_VMXON_RAX ".byte 0xf3, 0x0f, 0xc7, 0x30"
545 #define MSR_IA32_TIME_STAMP_COUNTER 0x010
547 #define TSS_IOPB_BASE_OFFSET 0x66
548 #define TSS_BASE_SIZE 0x68
549 #define TSS_IOPB_SIZE (65536 / 8)
550 #define TSS_REDIRECTION_SIZE (256 / 8)
551 #define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)