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.
14 #include <linux/types.h>
17 #include <linux/kvm.h>
18 #include <linux/kvm_para.h>
24 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
25 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
26 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS|0xFFFFFF0000000000ULL)
28 #define KVM_GUEST_CR0_MASK \
29 (X86_CR0_PG | X86_CR0_PE | X86_CR0_WP | X86_CR0_NE \
30 | X86_CR0_NW | X86_CR0_CD)
31 #define KVM_VM_CR0_ALWAYS_ON \
32 (X86_CR0_PG | X86_CR0_PE | X86_CR0_WP | X86_CR0_NE | X86_CR0_TS \
34 #define KVM_GUEST_CR4_MASK \
35 (X86_CR4_VME | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_PGE | X86_CR4_VMXE)
36 #define KVM_PMODE_VM_CR4_ALWAYS_ON (X86_CR4_PAE | X86_CR4_VMXE)
37 #define KVM_RMODE_VM_CR4_ALWAYS_ON (X86_CR4_VME | X86_CR4_PAE | X86_CR4_VMXE)
39 #define INVALID_PAGE (~(hpa_t)0)
40 #define UNMAPPED_GVA (~(gpa_t)0)
52 #define SELECTOR_TI_MASK (1 << 2)
53 #define SELECTOR_RPL_MASK 0x03
57 #define KVM_ALIAS_SLOTS 4
59 #define KVM_PERMILLE_MMU_PAGES 20
60 #define KVM_MIN_ALLOC_MMU_PAGES 64
61 #define KVM_NUM_MMU_PAGES 1024
62 #define KVM_MIN_FREE_MMU_PAGES 5
63 #define KVM_REFILL_PAGES 25
64 #define KVM_MAX_CPUID_ENTRIES 40
66 extern spinlock_t kvm_lock
;
67 extern struct list_head vm_list
;
105 #include "x86_emulate.h"
107 #define KVM_NR_MEM_OBJS 40
110 * We don't want allocation failures within the mmu code, so we preallocate
111 * enough memory for a single page fault in a cache.
113 struct kvm_mmu_memory_cache
{
115 void *objects
[KVM_NR_MEM_OBJS
];
118 #define NR_PTE_CHAIN_ENTRIES 5
120 struct kvm_pte_chain
{
121 u64
*parent_ptes
[NR_PTE_CHAIN_ENTRIES
];
122 struct hlist_node link
;
126 * kvm_mmu_page_role, below, is defined as:
128 * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
129 * bits 4:7 - page table level for this shadow (1-4)
130 * bits 8:9 - page table quadrant for 2-level guests
131 * bit 16 - "metaphysical" - gfn is not a real page (huge page/real mode)
132 * bits 17:19 - common access permissions for all ptes in this shadow page
134 union kvm_mmu_page_role
{
137 unsigned glevels
: 4;
139 unsigned quadrant
: 2;
140 unsigned pad_for_nice_hex_output
: 6;
141 unsigned metaphysical
: 1;
146 struct kvm_mmu_page
{
147 struct list_head link
;
148 struct hlist_node hash_link
;
151 * The following two entries are used to key the shadow page in the
155 union kvm_mmu_page_role role
;
158 /* hold the gfn of each spte inside spt */
160 unsigned long slot_bitmap
; /* One bit set per slot which has memory
161 * in this shadow page.
163 int multimapped
; /* More than one parent_pte? */
164 int root_count
; /* Currently serving as active root */
166 u64
*parent_pte
; /* !multimapped */
167 struct hlist_head parent_ptes
; /* multimapped, kvm_pte_chain */
172 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
173 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
177 void (*new_cr3
)(struct kvm_vcpu
*vcpu
);
178 int (*page_fault
)(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 err
);
179 void (*free
)(struct kvm_vcpu
*vcpu
);
180 gpa_t (*gva_to_gpa
)(struct kvm_vcpu
*vcpu
, gva_t gva
);
181 void (*prefetch_page
)(struct kvm_vcpu
*vcpu
,
182 struct kvm_mmu_page
*page
);
185 int shadow_root_level
;
190 struct kvm_vcpu_arch
{
192 int interrupt_window_open
;
193 unsigned long irq_summary
; /* bit vector: 1 per word in irq_pending */
194 DECLARE_BITMAP(irq_pending
, KVM_NR_INTERRUPTS
);
195 unsigned long regs
[NR_VCPU_REGS
]; /* for rsp: vcpu_load_rsp_rip() */
196 unsigned long rip
; /* needs vcpu_load_rsp_rip() */
203 u64 pdptrs
[4]; /* pae */
206 struct kvm_lapic
*apic
; /* kernel irqchip context */
207 #define VCPU_MP_STATE_RUNNABLE 0
208 #define VCPU_MP_STATE_UNINITIALIZED 1
209 #define VCPU_MP_STATE_INIT_RECEIVED 2
210 #define VCPU_MP_STATE_SIPI_RECEIVED 3
211 #define VCPU_MP_STATE_HALTED 4
214 u64 ia32_misc_enable_msr
;
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
;
280 struct kvm_vcpu_stat
{
290 u32 irq_window_exits
;
293 u32 request_irq_exits
;
295 u32 host_state_reload
;
299 u32 insn_emulation_fail
;
302 struct descriptor_table
{
305 } __attribute__((packed
));
308 int (*cpu_has_kvm_support
)(void); /* __init */
309 int (*disabled_by_bios
)(void); /* __init */
310 void (*hardware_enable
)(void *dummy
); /* __init */
311 void (*hardware_disable
)(void *dummy
);
312 void (*check_processor_compatibility
)(void *rtn
);
313 int (*hardware_setup
)(void); /* __init */
314 void (*hardware_unsetup
)(void); /* __exit */
316 /* Create, but do not attach this VCPU */
317 struct kvm_vcpu
*(*vcpu_create
)(struct kvm
*kvm
, unsigned id
);
318 void (*vcpu_free
)(struct kvm_vcpu
*vcpu
);
319 int (*vcpu_reset
)(struct kvm_vcpu
*vcpu
);
321 void (*prepare_guest_switch
)(struct kvm_vcpu
*vcpu
);
322 void (*vcpu_load
)(struct kvm_vcpu
*vcpu
, int cpu
);
323 void (*vcpu_put
)(struct kvm_vcpu
*vcpu
);
324 void (*vcpu_decache
)(struct kvm_vcpu
*vcpu
);
326 int (*set_guest_debug
)(struct kvm_vcpu
*vcpu
,
327 struct kvm_debug_guest
*dbg
);
328 void (*guest_debug_pre
)(struct kvm_vcpu
*vcpu
);
329 int (*get_msr
)(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64
*pdata
);
330 int (*set_msr
)(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64 data
);
331 u64 (*get_segment_base
)(struct kvm_vcpu
*vcpu
, int seg
);
332 void (*get_segment
)(struct kvm_vcpu
*vcpu
,
333 struct kvm_segment
*var
, int seg
);
334 void (*set_segment
)(struct kvm_vcpu
*vcpu
,
335 struct kvm_segment
*var
, int seg
);
336 void (*get_cs_db_l_bits
)(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
337 void (*decache_cr4_guest_bits
)(struct kvm_vcpu
*vcpu
);
338 void (*set_cr0
)(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
339 void (*set_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
340 void (*set_cr4
)(struct kvm_vcpu
*vcpu
, unsigned long cr4
);
341 void (*set_efer
)(struct kvm_vcpu
*vcpu
, u64 efer
);
342 void (*get_idt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
343 void (*set_idt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
344 void (*get_gdt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
345 void (*set_gdt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
346 unsigned long (*get_dr
)(struct kvm_vcpu
*vcpu
, int dr
);
347 void (*set_dr
)(struct kvm_vcpu
*vcpu
, int dr
, unsigned long value
,
349 void (*cache_regs
)(struct kvm_vcpu
*vcpu
);
350 void (*decache_regs
)(struct kvm_vcpu
*vcpu
);
351 unsigned long (*get_rflags
)(struct kvm_vcpu
*vcpu
);
352 void (*set_rflags
)(struct kvm_vcpu
*vcpu
, unsigned long rflags
);
354 void (*tlb_flush
)(struct kvm_vcpu
*vcpu
);
356 void (*run
)(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
);
357 int (*handle_exit
)(struct kvm_run
*run
, struct kvm_vcpu
*vcpu
);
358 void (*skip_emulated_instruction
)(struct kvm_vcpu
*vcpu
);
359 void (*patch_hypercall
)(struct kvm_vcpu
*vcpu
,
360 unsigned char *hypercall_addr
);
361 int (*get_irq
)(struct kvm_vcpu
*vcpu
);
362 void (*set_irq
)(struct kvm_vcpu
*vcpu
, int vec
);
363 void (*queue_exception
)(struct kvm_vcpu
*vcpu
, unsigned nr
,
364 bool has_error_code
, u32 error_code
);
365 bool (*exception_injected
)(struct kvm_vcpu
*vcpu
);
366 void (*inject_pending_irq
)(struct kvm_vcpu
*vcpu
);
367 void (*inject_pending_vectors
)(struct kvm_vcpu
*vcpu
,
368 struct kvm_run
*run
);
370 int (*set_tss_addr
)(struct kvm
*kvm
, unsigned int addr
);
373 extern struct kvm_x86_ops
*kvm_x86_ops
;
375 int kvm_mmu_module_init(void);
376 void kvm_mmu_module_exit(void);
378 void kvm_mmu_destroy(struct kvm_vcpu
*vcpu
);
379 int kvm_mmu_create(struct kvm_vcpu
*vcpu
);
380 int kvm_mmu_setup(struct kvm_vcpu
*vcpu
);
381 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte
, u64 notrap_pte
);
383 int kvm_mmu_reset_context(struct kvm_vcpu
*vcpu
);
384 void kvm_mmu_slot_remove_write_access(struct kvm
*kvm
, int slot
);
385 void kvm_mmu_zap_all(struct kvm
*kvm
);
386 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm
*kvm
);
387 void kvm_mmu_change_mmu_pages(struct kvm
*kvm
, unsigned int kvm_nr_mmu_pages
);
389 enum emulation_result
{
390 EMULATE_DONE
, /* no further processing */
391 EMULATE_DO_MMIO
, /* kvm_run filled with mmio request */
392 EMULATE_FAIL
, /* can't emulate this instruction */
395 int emulate_instruction(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
,
396 unsigned long cr2
, u16 error_code
, int no_decode
);
397 void kvm_report_emulation_failure(struct kvm_vcpu
*cvpu
, const char *context
);
398 void realmode_lgdt(struct kvm_vcpu
*vcpu
, u16 size
, unsigned long address
);
399 void realmode_lidt(struct kvm_vcpu
*vcpu
, u16 size
, unsigned long address
);
400 void realmode_lmsw(struct kvm_vcpu
*vcpu
, unsigned long msw
,
401 unsigned long *rflags
);
403 unsigned long realmode_get_cr(struct kvm_vcpu
*vcpu
, int cr
);
404 void realmode_set_cr(struct kvm_vcpu
*vcpu
, int cr
, unsigned long value
,
405 unsigned long *rflags
);
406 int kvm_get_msr(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64
*data
);
407 int kvm_set_msr(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64 data
);
409 struct x86_emulate_ctxt
;
411 int kvm_emulate_pio(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
, int in
,
412 int size
, unsigned port
);
413 int kvm_emulate_pio_string(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
, int in
,
414 int size
, unsigned long count
, int down
,
415 gva_t address
, int rep
, unsigned port
);
416 void kvm_emulate_cpuid(struct kvm_vcpu
*vcpu
);
417 int kvm_emulate_halt(struct kvm_vcpu
*vcpu
);
418 int emulate_invlpg(struct kvm_vcpu
*vcpu
, gva_t address
);
419 int emulate_clts(struct kvm_vcpu
*vcpu
);
420 int emulator_get_dr(struct x86_emulate_ctxt
*ctxt
, int dr
,
421 unsigned long *dest
);
422 int emulator_set_dr(struct x86_emulate_ctxt
*ctxt
, int dr
,
423 unsigned long value
);
425 void set_cr0(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
426 void set_cr3(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
427 void set_cr4(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
428 void set_cr8(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
429 unsigned long get_cr8(struct kvm_vcpu
*vcpu
);
430 void lmsw(struct kvm_vcpu
*vcpu
, unsigned long msw
);
431 void kvm_get_cs_db_l_bits(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
433 int kvm_get_msr_common(struct kvm_vcpu
*vcpu
, u32 msr
, u64
*pdata
);
434 int kvm_set_msr_common(struct kvm_vcpu
*vcpu
, u32 msr
, u64 data
);
436 void kvm_queue_exception(struct kvm_vcpu
*vcpu
, unsigned nr
);
437 void kvm_queue_exception_e(struct kvm_vcpu
*vcpu
, unsigned nr
, u32 error_code
);
438 void kvm_inject_page_fault(struct kvm_vcpu
*vcpu
, unsigned long cr2
,
441 void fx_init(struct kvm_vcpu
*vcpu
);
443 int emulator_read_std(unsigned long addr
,
446 struct kvm_vcpu
*vcpu
);
447 int emulator_write_emulated(unsigned long addr
,
450 struct kvm_vcpu
*vcpu
);
452 unsigned long segment_base(u16 selector
);
454 void kvm_mmu_flush_tlb(struct kvm_vcpu
*vcpu
);
455 void kvm_mmu_pte_write(struct kvm_vcpu
*vcpu
, gpa_t gpa
,
456 const u8
*new, int bytes
);
457 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu
*vcpu
, gva_t gva
);
458 void __kvm_mmu_free_some_pages(struct kvm_vcpu
*vcpu
);
459 int kvm_mmu_load(struct kvm_vcpu
*vcpu
);
460 void kvm_mmu_unload(struct kvm_vcpu
*vcpu
);
462 int kvm_emulate_hypercall(struct kvm_vcpu
*vcpu
);
464 int kvm_fix_hypercall(struct kvm_vcpu
*vcpu
);
466 int kvm_mmu_page_fault(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 error_code
);
468 int load_pdptrs(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
469 int complete_pio(struct kvm_vcpu
*vcpu
);
471 static inline struct kvm_mmu_page
*page_header(hpa_t shadow_page
)
473 struct page
*page
= pfn_to_page(shadow_page
>> PAGE_SHIFT
);
475 return (struct kvm_mmu_page
*)page_private(page
);
478 static inline u16
read_fs(void)
481 asm("mov %%fs, %0" : "=g"(seg
));
485 static inline u16
read_gs(void)
488 asm("mov %%gs, %0" : "=g"(seg
));
492 static inline u16
read_ldt(void)
495 asm("sldt %0" : "=g"(ldt
));
499 static inline void load_fs(u16 sel
)
501 asm("mov %0, %%fs" : : "rm"(sel
));
504 static inline void load_gs(u16 sel
)
506 asm("mov %0, %%gs" : : "rm"(sel
));
510 static inline void load_ldt(u16 sel
)
512 asm("lldt %0" : : "rm"(sel
));
516 static inline void get_idt(struct descriptor_table
*table
)
518 asm("sidt %0" : "=m"(*table
));
521 static inline void get_gdt(struct descriptor_table
*table
)
523 asm("sgdt %0" : "=m"(*table
));
526 static inline unsigned long read_tr_base(void)
529 asm("str %0" : "=g"(tr
));
530 return segment_base(tr
);
534 static inline unsigned long read_msr(unsigned long msr
)
543 static inline void fx_save(struct i387_fxsave_struct
*image
)
545 asm("fxsave (%0)":: "r" (image
));
548 static inline void fx_restore(struct i387_fxsave_struct
*image
)
550 asm("fxrstor (%0)":: "r" (image
));
553 static inline void fpu_init(void)
558 static inline u32
get_rdx_init_val(void)
560 return 0x600; /* P6 family */
563 static inline void kvm_inject_gp(struct kvm_vcpu
*vcpu
, u32 error_code
)
565 kvm_queue_exception_e(vcpu
, GP_VECTOR
, error_code
);
568 #define ASM_VMX_VMCLEAR_RAX ".byte 0x66, 0x0f, 0xc7, 0x30"
569 #define ASM_VMX_VMLAUNCH ".byte 0x0f, 0x01, 0xc2"
570 #define ASM_VMX_VMRESUME ".byte 0x0f, 0x01, 0xc3"
571 #define ASM_VMX_VMPTRLD_RAX ".byte 0x0f, 0xc7, 0x30"
572 #define ASM_VMX_VMREAD_RDX_RAX ".byte 0x0f, 0x78, 0xd0"
573 #define ASM_VMX_VMWRITE_RAX_RDX ".byte 0x0f, 0x79, 0xd0"
574 #define ASM_VMX_VMWRITE_RSP_RDX ".byte 0x0f, 0x79, 0xd4"
575 #define ASM_VMX_VMXOFF ".byte 0x0f, 0x01, 0xc4"
576 #define ASM_VMX_VMXON_RAX ".byte 0xf3, 0x0f, 0xc7, 0x30"
578 #define MSR_IA32_TIME_STAMP_COUNTER 0x010
580 #define TSS_IOPB_BASE_OFFSET 0x66
581 #define TSS_BASE_SIZE 0x68
582 #define TSS_IOPB_SIZE (65536 / 8)
583 #define TSS_REDIRECTION_SIZE (256 / 8)
584 #define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)