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_PERMILLE_MMU_PAGES 20
58 #define KVM_MIN_ALLOC_MMU_PAGES 64
59 #define KVM_NUM_MMU_PAGES 1024
60 #define KVM_MIN_FREE_MMU_PAGES 5
61 #define KVM_REFILL_PAGES 25
62 #define KVM_MAX_CPUID_ENTRIES 40
64 extern spinlock_t kvm_lock
;
65 extern struct list_head vm_list
;
103 #include "x86_emulate.h"
105 #define KVM_NR_MEM_OBJS 40
108 * We don't want allocation failures within the mmu code, so we preallocate
109 * enough memory for a single page fault in a cache.
111 struct kvm_mmu_memory_cache
{
113 void *objects
[KVM_NR_MEM_OBJS
];
116 #define NR_PTE_CHAIN_ENTRIES 5
118 struct kvm_pte_chain
{
119 u64
*parent_ptes
[NR_PTE_CHAIN_ENTRIES
];
120 struct hlist_node link
;
124 * kvm_mmu_page_role, below, is defined as:
126 * bits 0:3 - total guest paging levels (2-4, or zero for real mode)
127 * bits 4:7 - page table level for this shadow (1-4)
128 * bits 8:9 - page table quadrant for 2-level guests
129 * bit 16 - "metaphysical" - gfn is not a real page (huge page/real mode)
130 * bits 17:19 - common access permissions for all ptes in this shadow page
132 union kvm_mmu_page_role
{
135 unsigned glevels
: 4;
137 unsigned quadrant
: 2;
138 unsigned pad_for_nice_hex_output
: 6;
139 unsigned metaphysical
: 1;
144 struct kvm_mmu_page
{
145 struct list_head link
;
146 struct hlist_node hash_link
;
149 * The following two entries are used to key the shadow page in the
153 union kvm_mmu_page_role role
;
156 /* hold the gfn of each spte inside spt */
158 unsigned long slot_bitmap
; /* One bit set per slot which has memory
159 * in this shadow page.
161 int multimapped
; /* More than one parent_pte? */
162 int root_count
; /* Currently serving as active root */
164 u64
*parent_pte
; /* !multimapped */
165 struct hlist_head parent_ptes
; /* multimapped, kvm_pte_chain */
170 * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
171 * 32-bit). The kvm_mmu structure abstracts the details of the current mmu
175 void (*new_cr3
)(struct kvm_vcpu
*vcpu
);
176 int (*page_fault
)(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 err
);
177 void (*free
)(struct kvm_vcpu
*vcpu
);
178 gpa_t (*gva_to_gpa
)(struct kvm_vcpu
*vcpu
, gva_t gva
);
179 void (*prefetch_page
)(struct kvm_vcpu
*vcpu
,
180 struct kvm_mmu_page
*page
);
183 int shadow_root_level
;
188 struct kvm_vcpu_arch
{
190 int interrupt_window_open
;
191 unsigned long irq_summary
; /* bit vector: 1 per word in irq_pending */
192 DECLARE_BITMAP(irq_pending
, KVM_NR_INTERRUPTS
);
193 unsigned long regs
[NR_VCPU_REGS
]; /* for rsp: vcpu_load_rsp_rip() */
194 unsigned long rip
; /* needs vcpu_load_rsp_rip() */
201 u64 pdptrs
[4]; /* pae */
204 struct kvm_lapic
*apic
; /* kernel irqchip context */
205 #define VCPU_MP_STATE_RUNNABLE 0
206 #define VCPU_MP_STATE_UNINITIALIZED 1
207 #define VCPU_MP_STATE_INIT_RECEIVED 2
208 #define VCPU_MP_STATE_SIPI_RECEIVED 3
209 #define VCPU_MP_STATE_HALTED 4
212 u64 ia32_misc_enable_msr
;
216 struct kvm_mmu_memory_cache mmu_pte_chain_cache
;
217 struct kvm_mmu_memory_cache mmu_rmap_desc_cache
;
218 struct kvm_mmu_memory_cache mmu_page_cache
;
219 struct kvm_mmu_memory_cache mmu_page_header_cache
;
221 gfn_t last_pt_write_gfn
;
222 int last_pt_write_count
;
223 u64
*last_pte_updated
;
225 struct i387_fxsave_struct host_fx_image
;
226 struct i387_fxsave_struct guest_fx_image
;
228 gva_t mmio_fault_cr2
;
229 struct kvm_pio_request pio
;
232 struct kvm_queued_exception
{
242 struct kvm_save_segment
{
247 } tr
, es
, ds
, fs
, gs
;
249 int halt_request
; /* real mode on Intel only */
252 struct kvm_cpuid_entry2 cpuid_entries
[KVM_MAX_CPUID_ENTRIES
];
253 /* emulate context */
255 struct x86_emulate_ctxt emulate_ctxt
;
259 struct descriptor_table
{
262 } __attribute__((packed
));
265 int (*cpu_has_kvm_support
)(void); /* __init */
266 int (*disabled_by_bios
)(void); /* __init */
267 void (*hardware_enable
)(void *dummy
); /* __init */
268 void (*hardware_disable
)(void *dummy
);
269 void (*check_processor_compatibility
)(void *rtn
);
270 int (*hardware_setup
)(void); /* __init */
271 void (*hardware_unsetup
)(void); /* __exit */
273 /* Create, but do not attach this VCPU */
274 struct kvm_vcpu
*(*vcpu_create
)(struct kvm
*kvm
, unsigned id
);
275 void (*vcpu_free
)(struct kvm_vcpu
*vcpu
);
276 int (*vcpu_reset
)(struct kvm_vcpu
*vcpu
);
278 void (*prepare_guest_switch
)(struct kvm_vcpu
*vcpu
);
279 void (*vcpu_load
)(struct kvm_vcpu
*vcpu
, int cpu
);
280 void (*vcpu_put
)(struct kvm_vcpu
*vcpu
);
281 void (*vcpu_decache
)(struct kvm_vcpu
*vcpu
);
283 int (*set_guest_debug
)(struct kvm_vcpu
*vcpu
,
284 struct kvm_debug_guest
*dbg
);
285 void (*guest_debug_pre
)(struct kvm_vcpu
*vcpu
);
286 int (*get_msr
)(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64
*pdata
);
287 int (*set_msr
)(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64 data
);
288 u64 (*get_segment_base
)(struct kvm_vcpu
*vcpu
, int seg
);
289 void (*get_segment
)(struct kvm_vcpu
*vcpu
,
290 struct kvm_segment
*var
, int seg
);
291 void (*set_segment
)(struct kvm_vcpu
*vcpu
,
292 struct kvm_segment
*var
, int seg
);
293 void (*get_cs_db_l_bits
)(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
294 void (*decache_cr4_guest_bits
)(struct kvm_vcpu
*vcpu
);
295 void (*set_cr0
)(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
296 void (*set_cr3
)(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
297 void (*set_cr4
)(struct kvm_vcpu
*vcpu
, unsigned long cr4
);
298 void (*set_efer
)(struct kvm_vcpu
*vcpu
, u64 efer
);
299 void (*get_idt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
300 void (*set_idt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
301 void (*get_gdt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
302 void (*set_gdt
)(struct kvm_vcpu
*vcpu
, struct descriptor_table
*dt
);
303 unsigned long (*get_dr
)(struct kvm_vcpu
*vcpu
, int dr
);
304 void (*set_dr
)(struct kvm_vcpu
*vcpu
, int dr
, unsigned long value
,
306 void (*cache_regs
)(struct kvm_vcpu
*vcpu
);
307 void (*decache_regs
)(struct kvm_vcpu
*vcpu
);
308 unsigned long (*get_rflags
)(struct kvm_vcpu
*vcpu
);
309 void (*set_rflags
)(struct kvm_vcpu
*vcpu
, unsigned long rflags
);
311 void (*tlb_flush
)(struct kvm_vcpu
*vcpu
);
313 void (*run
)(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
);
314 int (*handle_exit
)(struct kvm_run
*run
, struct kvm_vcpu
*vcpu
);
315 void (*skip_emulated_instruction
)(struct kvm_vcpu
*vcpu
);
316 void (*patch_hypercall
)(struct kvm_vcpu
*vcpu
,
317 unsigned char *hypercall_addr
);
318 int (*get_irq
)(struct kvm_vcpu
*vcpu
);
319 void (*set_irq
)(struct kvm_vcpu
*vcpu
, int vec
);
320 void (*queue_exception
)(struct kvm_vcpu
*vcpu
, unsigned nr
,
321 bool has_error_code
, u32 error_code
);
322 bool (*exception_injected
)(struct kvm_vcpu
*vcpu
);
323 void (*inject_pending_irq
)(struct kvm_vcpu
*vcpu
);
324 void (*inject_pending_vectors
)(struct kvm_vcpu
*vcpu
,
325 struct kvm_run
*run
);
327 int (*set_tss_addr
)(struct kvm
*kvm
, unsigned int addr
);
330 extern struct kvm_x86_ops
*kvm_x86_ops
;
332 int kvm_mmu_module_init(void);
333 void kvm_mmu_module_exit(void);
335 void kvm_mmu_destroy(struct kvm_vcpu
*vcpu
);
336 int kvm_mmu_create(struct kvm_vcpu
*vcpu
);
337 int kvm_mmu_setup(struct kvm_vcpu
*vcpu
);
338 void kvm_mmu_set_nonpresent_ptes(u64 trap_pte
, u64 notrap_pte
);
340 int kvm_mmu_reset_context(struct kvm_vcpu
*vcpu
);
341 void kvm_mmu_slot_remove_write_access(struct kvm
*kvm
, int slot
);
342 void kvm_mmu_zap_all(struct kvm
*kvm
);
343 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm
*kvm
);
344 void kvm_mmu_change_mmu_pages(struct kvm
*kvm
, unsigned int kvm_nr_mmu_pages
);
346 enum emulation_result
{
347 EMULATE_DONE
, /* no further processing */
348 EMULATE_DO_MMIO
, /* kvm_run filled with mmio request */
349 EMULATE_FAIL
, /* can't emulate this instruction */
352 int emulate_instruction(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
,
353 unsigned long cr2
, u16 error_code
, int no_decode
);
354 void kvm_report_emulation_failure(struct kvm_vcpu
*cvpu
, const char *context
);
355 void realmode_lgdt(struct kvm_vcpu
*vcpu
, u16 size
, unsigned long address
);
356 void realmode_lidt(struct kvm_vcpu
*vcpu
, u16 size
, unsigned long address
);
357 void realmode_lmsw(struct kvm_vcpu
*vcpu
, unsigned long msw
,
358 unsigned long *rflags
);
360 unsigned long realmode_get_cr(struct kvm_vcpu
*vcpu
, int cr
);
361 void realmode_set_cr(struct kvm_vcpu
*vcpu
, int cr
, unsigned long value
,
362 unsigned long *rflags
);
363 int kvm_get_msr(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64
*data
);
364 int kvm_set_msr(struct kvm_vcpu
*vcpu
, u32 msr_index
, u64 data
);
366 struct x86_emulate_ctxt
;
368 int kvm_emulate_pio(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
, int in
,
369 int size
, unsigned port
);
370 int kvm_emulate_pio_string(struct kvm_vcpu
*vcpu
, struct kvm_run
*run
, int in
,
371 int size
, unsigned long count
, int down
,
372 gva_t address
, int rep
, unsigned port
);
373 void kvm_emulate_cpuid(struct kvm_vcpu
*vcpu
);
374 int kvm_emulate_halt(struct kvm_vcpu
*vcpu
);
375 int emulate_invlpg(struct kvm_vcpu
*vcpu
, gva_t address
);
376 int emulate_clts(struct kvm_vcpu
*vcpu
);
377 int emulator_get_dr(struct x86_emulate_ctxt
*ctxt
, int dr
,
378 unsigned long *dest
);
379 int emulator_set_dr(struct x86_emulate_ctxt
*ctxt
, int dr
,
380 unsigned long value
);
382 void set_cr0(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
383 void set_cr3(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
384 void set_cr4(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
385 void set_cr8(struct kvm_vcpu
*vcpu
, unsigned long cr0
);
386 unsigned long get_cr8(struct kvm_vcpu
*vcpu
);
387 void lmsw(struct kvm_vcpu
*vcpu
, unsigned long msw
);
388 void kvm_get_cs_db_l_bits(struct kvm_vcpu
*vcpu
, int *db
, int *l
);
390 int kvm_get_msr_common(struct kvm_vcpu
*vcpu
, u32 msr
, u64
*pdata
);
391 int kvm_set_msr_common(struct kvm_vcpu
*vcpu
, u32 msr
, u64 data
);
393 void kvm_queue_exception(struct kvm_vcpu
*vcpu
, unsigned nr
);
394 void kvm_queue_exception_e(struct kvm_vcpu
*vcpu
, unsigned nr
, u32 error_code
);
395 void kvm_inject_page_fault(struct kvm_vcpu
*vcpu
, unsigned long cr2
,
398 void fx_init(struct kvm_vcpu
*vcpu
);
400 int emulator_read_std(unsigned long addr
,
403 struct kvm_vcpu
*vcpu
);
404 int emulator_write_emulated(unsigned long addr
,
407 struct kvm_vcpu
*vcpu
);
409 unsigned long segment_base(u16 selector
);
411 void kvm_mmu_flush_tlb(struct kvm_vcpu
*vcpu
);
412 void kvm_mmu_pte_write(struct kvm_vcpu
*vcpu
, gpa_t gpa
,
413 const u8
*new, int bytes
);
414 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu
*vcpu
, gva_t gva
);
415 void __kvm_mmu_free_some_pages(struct kvm_vcpu
*vcpu
);
416 int kvm_mmu_load(struct kvm_vcpu
*vcpu
);
417 void kvm_mmu_unload(struct kvm_vcpu
*vcpu
);
419 int kvm_emulate_hypercall(struct kvm_vcpu
*vcpu
);
421 int kvm_fix_hypercall(struct kvm_vcpu
*vcpu
);
423 int kvm_mmu_page_fault(struct kvm_vcpu
*vcpu
, gva_t gva
, u32 error_code
);
425 int load_pdptrs(struct kvm_vcpu
*vcpu
, unsigned long cr3
);
426 int complete_pio(struct kvm_vcpu
*vcpu
);
428 static inline struct kvm_mmu_page
*page_header(hpa_t shadow_page
)
430 struct page
*page
= pfn_to_page(shadow_page
>> PAGE_SHIFT
);
432 return (struct kvm_mmu_page
*)page_private(page
);
435 static inline u16
read_fs(void)
438 asm("mov %%fs, %0" : "=g"(seg
));
442 static inline u16
read_gs(void)
445 asm("mov %%gs, %0" : "=g"(seg
));
449 static inline u16
read_ldt(void)
452 asm("sldt %0" : "=g"(ldt
));
456 static inline void load_fs(u16 sel
)
458 asm("mov %0, %%fs" : : "rm"(sel
));
461 static inline void load_gs(u16 sel
)
463 asm("mov %0, %%gs" : : "rm"(sel
));
467 static inline void load_ldt(u16 sel
)
469 asm("lldt %0" : : "rm"(sel
));
473 static inline void get_idt(struct descriptor_table
*table
)
475 asm("sidt %0" : "=m"(*table
));
478 static inline void get_gdt(struct descriptor_table
*table
)
480 asm("sgdt %0" : "=m"(*table
));
483 static inline unsigned long read_tr_base(void)
486 asm("str %0" : "=g"(tr
));
487 return segment_base(tr
);
491 static inline unsigned long read_msr(unsigned long msr
)
500 static inline void fx_save(struct i387_fxsave_struct
*image
)
502 asm("fxsave (%0)":: "r" (image
));
505 static inline void fx_restore(struct i387_fxsave_struct
*image
)
507 asm("fxrstor (%0)":: "r" (image
));
510 static inline void fpu_init(void)
515 static inline u32
get_rdx_init_val(void)
517 return 0x600; /* P6 family */
520 static inline void kvm_inject_gp(struct kvm_vcpu
*vcpu
, u32 error_code
)
522 kvm_queue_exception_e(vcpu
, GP_VECTOR
, error_code
);
525 #define ASM_VMX_VMCLEAR_RAX ".byte 0x66, 0x0f, 0xc7, 0x30"
526 #define ASM_VMX_VMLAUNCH ".byte 0x0f, 0x01, 0xc2"
527 #define ASM_VMX_VMRESUME ".byte 0x0f, 0x01, 0xc3"
528 #define ASM_VMX_VMPTRLD_RAX ".byte 0x0f, 0xc7, 0x30"
529 #define ASM_VMX_VMREAD_RDX_RAX ".byte 0x0f, 0x78, 0xd0"
530 #define ASM_VMX_VMWRITE_RAX_RDX ".byte 0x0f, 0x79, 0xd0"
531 #define ASM_VMX_VMWRITE_RSP_RDX ".byte 0x0f, 0x79, 0xd4"
532 #define ASM_VMX_VMXOFF ".byte 0x0f, 0x01, 0xc4"
533 #define ASM_VMX_VMXON_RAX ".byte 0xf3, 0x0f, 0xc7, 0x30"
535 #define MSR_IA32_TIME_STAMP_COUNTER 0x010
537 #define TSS_IOPB_BASE_OFFSET 0x66
538 #define TSS_BASE_SIZE 0x68
539 #define TSS_IOPB_SIZE (65536 / 8)
540 #define TSS_REDIRECTION_SIZE (256 / 8)
541 #define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)