2 * Kernel-based Virtual Machine driver for Linux
4 * This header defines architecture specific interfaces, x86 version
6 * This work is licensed under the terms of the GNU GPL, version 2. See
7 * the COPYING file in the top-level directory.
11 #ifndef ASM_KVM_HOST_H
12 #define ASM_KVM_HOST_H
14 #include <linux/types.h>
17 #include <linux/kvm.h>
18 #include <linux/kvm_para.h>
19 #include <linux/kvm_types.h>
23 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
24 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
25 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS|0xFFFFFF0000000000ULL)
27 #define KVM_GUEST_CR0_MASK \
28 (X86_CR0_PG | X86_CR0_PE | X86_CR0_WP | X86_CR0_NE \
29 | X86_CR0_NW | X86_CR0_CD)
30 #define KVM_VM_CR0_ALWAYS_ON \
31 (X86_CR0_PG | X86_CR0_PE | X86_CR0_WP | X86_CR0_NE | X86_CR0_TS \
33 #define KVM_GUEST_CR4_MASK \
34 (X86_CR4_VME | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_PGE | X86_CR4_VMXE)
35 #define KVM_PMODE_VM_CR4_ALWAYS_ON (X86_CR4_PAE | X86_CR4_VMXE)
36 #define KVM_RMODE_VM_CR4_ALWAYS_ON (X86_CR4_VME | X86_CR4_PAE | X86_CR4_VMXE)
38 #define INVALID_PAGE (~(hpa_t)0)
39 #define UNMAPPED_GVA (~(gpa_t)0)
51 #define SELECTOR_TI_MASK (1 << 2)
52 #define SELECTOR_RPL_MASK 0x03
56 #define KVM_ALIAS_SLOTS 4
58 #define KVM_PERMILLE_MMU_PAGES 20
59 #define KVM_MIN_ALLOC_MMU_PAGES 64
60 #define KVM_NUM_MMU_PAGES 1024
61 #define KVM_MIN_FREE_MMU_PAGES 5
62 #define KVM_REFILL_PAGES 25
63 #define KVM_MAX_CPUID_ENTRIES 40
65 extern spinlock_t kvm_lock;
66 extern struct list_head vm_list;
104 #include <asm/kvm_x86_emulate.h>
106 #define KVM_NR_MEM_OBJS 40
109 * We don't want allocation failures within the mmu code, so we preallocate
110 * enough memory for a single page fault in a cache.
112 struct kvm_mmu_memory_cache {
114 void *objects[KVM_NR_MEM_OBJS];
117 #define NR_PTE_CHAIN_ENTRIES 5
119 struct kvm_pte_chain {
120 u64 *parent_ptes[NR_PTE_CHAIN_ENTRIES];
121 struct hlist_node link;
125 * kvm_mmu_page_role, below, is defined as:
127 * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
128 * bits 4:7 - page table level for this shadow (1-4)
129 * bits 8:9 - page table quadrant for 2-level guests
130 * bit 16 - "metaphysical" - gfn is not a real page (huge page/real mode)
131 * bits 17:19 - common access permissions for all ptes in this shadow page
133 union kvm_mmu_page_role {
136 unsigned glevels : 4;
138 unsigned quadrant : 2;
139 unsigned pad_for_nice_hex_output : 6;
140 unsigned metaphysical : 1;
145 struct kvm_mmu_page {
146 struct list_head link;
147 struct hlist_node hash_link;
150 * The following two entries are used to key the shadow page in the
154 union kvm_mmu_page_role role;
157 /* hold the gfn of each spte inside spt */
159 unsigned long slot_bitmap; /* One bit set per slot which has memory
160 * in this shadow page.
162 int multimapped; /* More than one parent_pte? */
163 int root_count; /* Currently serving as active root */
165 u64 *parent_pte; /* !multimapped */
166 struct hlist_head parent_ptes; /* multimapped, kvm_pte_chain */
171 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
172 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
176 void (*new_cr3)(struct kvm_vcpu *vcpu);
177 int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err);
178 void (*free)(struct kvm_vcpu *vcpu);
179 gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva);
180 void (*prefetch_page)(struct kvm_vcpu *vcpu,
181 struct kvm_mmu_page *page);
184 int shadow_root_level;
189 struct kvm_vcpu_arch {
191 int interrupt_window_open;
192 unsigned long irq_summary; /* bit vector: 1 per word in irq_pending */
193 DECLARE_BITMAP(irq_pending, KVM_NR_INTERRUPTS);
194 unsigned long regs[NR_VCPU_REGS]; /* for rsp: vcpu_load_rsp_rip() */
195 unsigned long rip; /* needs vcpu_load_rsp_rip() */
202 u64 pdptrs[4]; /* pae */
205 struct kvm_lapic *apic; /* kernel irqchip context */
206 #define VCPU_MP_STATE_RUNNABLE 0
207 #define VCPU_MP_STATE_UNINITIALIZED 1
208 #define VCPU_MP_STATE_INIT_RECEIVED 2
209 #define VCPU_MP_STATE_SIPI_RECEIVED 3
210 #define VCPU_MP_STATE_HALTED 4
213 u64 ia32_misc_enable_msr;
214 bool tpr_access_reporting;
218 struct kvm_mmu_memory_cache mmu_pte_chain_cache;
219 struct kvm_mmu_memory_cache mmu_rmap_desc_cache;
220 struct kvm_mmu_memory_cache mmu_page_cache;
221 struct kvm_mmu_memory_cache mmu_page_header_cache;
223 gfn_t last_pt_write_gfn;
224 int last_pt_write_count;
225 u64 *last_pte_updated;
227 struct i387_fxsave_struct host_fx_image;
228 struct i387_fxsave_struct guest_fx_image;
230 gva_t mmio_fault_cr2;
231 struct kvm_pio_request pio;
234 struct kvm_queued_exception {
244 struct kvm_save_segment {
249 } tr, es, ds, fs, gs;
251 int halt_request; /* real mode on Intel only */
254 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
255 /* emulate context */
257 struct x86_emulate_ctxt emulate_ctxt;
260 struct kvm_mem_alias {
262 unsigned long npages;
268 struct kvm_mem_alias aliases[KVM_ALIAS_SLOTS];
270 unsigned int n_free_mmu_pages;
271 unsigned int n_requested_mmu_pages;
272 unsigned int n_alloc_mmu_pages;
273 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
275 * Hash table of struct kvm_mmu_page.
277 struct list_head active_mmu_pages;
278 struct kvm_pic *vpic;
279 struct kvm_ioapic *vioapic;
281 int round_robin_prev_vcpu;
282 unsigned int tss_addr;
283 struct page *apic_access_page;
287 u32 mmu_shadow_zapped;
294 u32 remote_tlb_flush;
297 struct kvm_vcpu_stat {
307 u32 irq_window_exits;
310 u32 request_irq_exits;
312 u32 host_state_reload;
316 u32 insn_emulation_fail;
319 struct descriptor_table {
322 } __attribute__((packed));
325 int (*cpu_has_kvm_support)(void); /* __init */
326 int (*disabled_by_bios)(void); /* __init */
327 void (*hardware_enable)(void *dummy); /* __init */
328 void (*hardware_disable)(void *dummy);
329 void (*check_processor_compatibility)(void *rtn);
330 int (*hardware_setup)(void); /* __init */
331 void (*hardware_unsetup)(void); /* __exit */
333 /* Create, but do not attach this VCPU */
334 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
335 void (*vcpu_free)(struct kvm_vcpu *vcpu);
336 int (*vcpu_reset)(struct kvm_vcpu *vcpu);
338 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
339 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
340 void (*vcpu_put)(struct kvm_vcpu *vcpu);
341 void (*vcpu_decache)(struct kvm_vcpu *vcpu);
343 int (*set_guest_debug)(struct kvm_vcpu *vcpu,
344 struct kvm_debug_guest *dbg);
345 void (*guest_debug_pre)(struct kvm_vcpu *vcpu);
346 int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
347 int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
348 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
349 void (*get_segment)(struct kvm_vcpu *vcpu,
350 struct kvm_segment *var, int seg);
351 void (*set_segment)(struct kvm_vcpu *vcpu,
352 struct kvm_segment *var, int seg);
353 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
354 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
355 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
356 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
357 void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
358 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
359 void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
360 void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
361 void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
362 void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
363 unsigned long (*get_dr)(struct kvm_vcpu *vcpu, int dr);
364 void (*set_dr)(struct kvm_vcpu *vcpu, int dr, unsigned long value,
366 void (*cache_regs)(struct kvm_vcpu *vcpu);
367 void (*decache_regs)(struct kvm_vcpu *vcpu);
368 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
369 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
371 void (*tlb_flush)(struct kvm_vcpu *vcpu);
373 void (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
374 int (*handle_exit)(struct kvm_run *run, struct kvm_vcpu *vcpu);
375 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
376 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
377 unsigned char *hypercall_addr);
378 int (*get_irq)(struct kvm_vcpu *vcpu);
379 void (*set_irq)(struct kvm_vcpu *vcpu, int vec);
380 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
381 bool has_error_code, u32 error_code);
382 bool (*exception_injected)(struct kvm_vcpu *vcpu);
383 void (*inject_pending_irq)(struct kvm_vcpu *vcpu);
384 void (*inject_pending_vectors)(struct kvm_vcpu *vcpu,
385 struct kvm_run *run);
387 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
390 extern struct kvm_x86_ops *kvm_x86_ops;
392 int kvm_mmu_module_init(void);
393 void kvm_mmu_module_exit(void);
395 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
396 int kvm_mmu_create(struct kvm_vcpu *vcpu);
397 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
398 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte);
400 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
401 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
402 void kvm_mmu_zap_all(struct kvm *kvm);
403 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
404 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
406 enum emulation_result {
407 EMULATE_DONE, /* no further processing */
408 EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
409 EMULATE_FAIL, /* can't emulate this instruction */
412 int emulate_instruction(struct kvm_vcpu *vcpu, struct kvm_run *run,
413 unsigned long cr2, u16 error_code, int no_decode);
414 void kvm_report_emulation_failure(struct kvm_vcpu *cvpu, const char *context);
415 void realmode_lgdt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
416 void realmode_lidt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
417 void realmode_lmsw(struct kvm_vcpu *vcpu, unsigned long msw,
418 unsigned long *rflags);
420 unsigned long realmode_get_cr(struct kvm_vcpu *vcpu, int cr);
421 void realmode_set_cr(struct kvm_vcpu *vcpu, int cr, unsigned long value,
422 unsigned long *rflags);
423 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
424 int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
426 struct x86_emulate_ctxt;
428 int kvm_emulate_pio(struct kvm_vcpu *vcpu, struct kvm_run *run, int in,
429 int size, unsigned port);
430 int kvm_emulate_pio_string(struct kvm_vcpu *vcpu, struct kvm_run *run, int in,
431 int size, unsigned long count, int down,
432 gva_t address, int rep, unsigned port);
433 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
434 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
435 int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address);
436 int emulate_clts(struct kvm_vcpu *vcpu);
437 int emulator_get_dr(struct x86_emulate_ctxt *ctxt, int dr,
438 unsigned long *dest);
439 int emulator_set_dr(struct x86_emulate_ctxt *ctxt, int dr,
440 unsigned long value);
442 void set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
443 void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr0);
444 void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
445 void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
446 unsigned long get_cr8(struct kvm_vcpu *vcpu);
447 void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
448 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
450 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
451 int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
453 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
454 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
455 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, unsigned long cr2,
458 void fx_init(struct kvm_vcpu *vcpu);
460 int emulator_read_std(unsigned long addr,
463 struct kvm_vcpu *vcpu);
464 int emulator_write_emulated(unsigned long addr,
467 struct kvm_vcpu *vcpu);
469 unsigned long segment_base(u16 selector);
471 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
472 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
473 const u8 *new, int bytes);
474 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
475 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
476 int kvm_mmu_load(struct kvm_vcpu *vcpu);
477 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
479 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
481 int kvm_fix_hypercall(struct kvm_vcpu *vcpu);
483 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code);
485 int load_pdptrs(struct kvm_vcpu *vcpu, unsigned long cr3);
486 int complete_pio(struct kvm_vcpu *vcpu);
488 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
490 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
492 return (struct kvm_mmu_page *)page_private(page);
495 static inline u16 read_fs(void)
498 asm("mov %%fs, %0" : "=g"(seg));
502 static inline u16 read_gs(void)
505 asm("mov %%gs, %0" : "=g"(seg));
509 static inline u16 read_ldt(void)
512 asm("sldt %0" : "=g"(ldt));
516 static inline void load_fs(u16 sel)
518 asm("mov %0, %%fs" : : "rm"(sel));
521 static inline void load_gs(u16 sel)
523 asm("mov %0, %%gs" : : "rm"(sel));
527 static inline void load_ldt(u16 sel)
529 asm("lldt %0" : : "rm"(sel));
533 static inline void get_idt(struct descriptor_table *table)
535 asm("sidt %0" : "=m"(*table));
538 static inline void get_gdt(struct descriptor_table *table)
540 asm("sgdt %0" : "=m"(*table));
543 static inline unsigned long read_tr_base(void)
546 asm("str %0" : "=g"(tr));
547 return segment_base(tr);
551 static inline unsigned long read_msr(unsigned long msr)
560 static inline void fx_save(struct i387_fxsave_struct *image)
562 asm("fxsave (%0)":: "r" (image));
565 static inline void fx_restore(struct i387_fxsave_struct *image)
567 asm("fxrstor (%0)":: "r" (image));
570 static inline void fpu_init(void)
575 static inline u32 get_rdx_init_val(void)
577 return 0x600; /* P6 family */
580 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
582 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
585 #define ASM_VMX_VMCLEAR_RAX ".byte 0x66, 0x0f, 0xc7, 0x30"
586 #define ASM_VMX_VMLAUNCH ".byte 0x0f, 0x01, 0xc2"
587 #define ASM_VMX_VMRESUME ".byte 0x0f, 0x01, 0xc3"
588 #define ASM_VMX_VMPTRLD_RAX ".byte 0x0f, 0xc7, 0x30"
589 #define ASM_VMX_VMREAD_RDX_RAX ".byte 0x0f, 0x78, 0xd0"
590 #define ASM_VMX_VMWRITE_RAX_RDX ".byte 0x0f, 0x79, 0xd0"
591 #define ASM_VMX_VMWRITE_RSP_RDX ".byte 0x0f, 0x79, 0xd4"
592 #define ASM_VMX_VMXOFF ".byte 0x0f, 0x01, 0xc4"
593 #define ASM_VMX_VMXON_RAX ".byte 0xf3, 0x0f, 0xc7, 0x30"
595 #define MSR_IA32_TIME_STAMP_COUNTER 0x010
597 #define TSS_IOPB_BASE_OFFSET 0x66
598 #define TSS_BASE_SIZE 0x68
599 #define TSS_IOPB_SIZE (65536 / 8)
600 #define TSS_REDIRECTION_SIZE (256 / 8)
601 #define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)