aboutsummaryrefslogtreecommitdiffstats
path: root/include
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2008-01-30 17:30:10 -0500
committerLinus Torvalds <torvalds@linux-foundation.org>2008-01-30 17:30:10 -0500
commit2c57ee6f924c95e4dce61ed4776fb3f62e1b9f92 (patch)
treeb9d92e52e8c0ee68a0f5012b470c6146a9f0b65a /include
parentf389e9fcecdec4c4cb890ad28ea30a87a579ec3e (diff)
parent2f52d58c92d971bf421f461ad06eb93fb4f34981 (diff)
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm: (249 commits) KVM: Move apic timer migration away from critical section KVM: Put kvm_para.h include outside __KERNEL__ KVM: Fix unbounded preemption latency KVM: Initialize the mmu caches only after verifying cpu support KVM: MMU: Fix dirty page setting for pages removed from rmap KVM: Portability: Move kvm_fpu to asm-x86/kvm.h KVM: x86 emulator: Only allow VMCALL/VMMCALL trapped by #UD KVM: MMU: Merge shadow level check in FNAME(fetch) KVM: MMU: Move kvm_free_some_pages() into critical section KVM: MMU: Switch to mmu spinlock KVM: MMU: Avoid calling gfn_to_page() in mmu_set_spte() KVM: Add kvm_read_guest_atomic() KVM: MMU: Concurrent guest walkers KVM: Disable vapic support on Intel machines with FlexPriority KVM: Accelerated apic support KVM: local APIC TPR access reporting facility KVM: Print data for unimplemented wrmsr KVM: MMU: Add cache miss statistic KVM: MMU: Coalesce remote tlb flushes KVM: Expose ioapic to ia64 save/restore APIs ...
Diffstat (limited to 'include')
-rw-r--r--include/asm-x86/Kbuild1
-rw-r--r--include/asm-x86/kvm.h191
-rw-r--r--include/asm-x86/kvm_host.h611
-rw-r--r--include/asm-x86/kvm_para.h105
-rw-r--r--include/asm-x86/kvm_x86_emulate.h186
-rw-r--r--include/linux/Kbuild2
-rw-r--r--include/linux/kvm.h203
-rw-r--r--include/linux/kvm_host.h299
-rw-r--r--include/linux/kvm_para.h82
-rw-r--r--include/linux/kvm_types.h54
10 files changed, 1518 insertions, 216 deletions
diff --git a/include/asm-x86/Kbuild b/include/asm-x86/Kbuild
index e6189b229143..3c6f0f80e827 100644
--- a/include/asm-x86/Kbuild
+++ b/include/asm-x86/Kbuild
@@ -3,6 +3,7 @@ include include/asm-generic/Kbuild.asm
3header-y += boot.h 3header-y += boot.h
4header-y += bootparam.h 4header-y += bootparam.h
5header-y += debugreg.h 5header-y += debugreg.h
6header-y += kvm.h
6header-y += ldt.h 7header-y += ldt.h
7header-y += msr-index.h 8header-y += msr-index.h
8header-y += prctl.h 9header-y += prctl.h
diff --git a/include/asm-x86/kvm.h b/include/asm-x86/kvm.h
new file mode 100644
index 000000000000..7a71120426a3
--- /dev/null
+++ b/include/asm-x86/kvm.h
@@ -0,0 +1,191 @@
1#ifndef __LINUX_KVM_X86_H
2#define __LINUX_KVM_X86_H
3
4/*
5 * KVM x86 specific structures and definitions
6 *
7 */
8
9#include <asm/types.h>
10#include <linux/ioctl.h>
11
12/* Architectural interrupt line count. */
13#define KVM_NR_INTERRUPTS 256
14
15struct kvm_memory_alias {
16 __u32 slot; /* this has a different namespace than memory slots */
17 __u32 flags;
18 __u64 guest_phys_addr;
19 __u64 memory_size;
20 __u64 target_phys_addr;
21};
22
23/* for KVM_GET_IRQCHIP and KVM_SET_IRQCHIP */
24struct kvm_pic_state {
25 __u8 last_irr; /* edge detection */
26 __u8 irr; /* interrupt request register */
27 __u8 imr; /* interrupt mask register */
28 __u8 isr; /* interrupt service register */
29 __u8 priority_add; /* highest irq priority */
30 __u8 irq_base;
31 __u8 read_reg_select;
32 __u8 poll;
33 __u8 special_mask;
34 __u8 init_state;
35 __u8 auto_eoi;
36 __u8 rotate_on_auto_eoi;
37 __u8 special_fully_nested_mode;
38 __u8 init4; /* true if 4 byte init */
39 __u8 elcr; /* PIIX edge/trigger selection */
40 __u8 elcr_mask;
41};
42
43#define KVM_IOAPIC_NUM_PINS 24
44struct kvm_ioapic_state {
45 __u64 base_address;
46 __u32 ioregsel;
47 __u32 id;
48 __u32 irr;
49 __u32 pad;
50 union {
51 __u64 bits;
52 struct {
53 __u8 vector;
54 __u8 delivery_mode:3;
55 __u8 dest_mode:1;
56 __u8 delivery_status:1;
57 __u8 polarity:1;
58 __u8 remote_irr:1;
59 __u8 trig_mode:1;
60 __u8 mask:1;
61 __u8 reserve:7;
62 __u8 reserved[4];
63 __u8 dest_id;
64 } fields;
65 } redirtbl[KVM_IOAPIC_NUM_PINS];
66};
67
68#define KVM_IRQCHIP_PIC_MASTER 0
69#define KVM_IRQCHIP_PIC_SLAVE 1
70#define KVM_IRQCHIP_IOAPIC 2
71
72/* for KVM_GET_REGS and KVM_SET_REGS */
73struct kvm_regs {
74 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
75 __u64 rax, rbx, rcx, rdx;
76 __u64 rsi, rdi, rsp, rbp;
77 __u64 r8, r9, r10, r11;
78 __u64 r12, r13, r14, r15;
79 __u64 rip, rflags;
80};
81
82/* for KVM_GET_LAPIC and KVM_SET_LAPIC */
83#define KVM_APIC_REG_SIZE 0x400
84struct kvm_lapic_state {
85 char regs[KVM_APIC_REG_SIZE];
86};
87
88struct kvm_segment {
89 __u64 base;
90 __u32 limit;
91 __u16 selector;
92 __u8 type;
93 __u8 present, dpl, db, s, l, g, avl;
94 __u8 unusable;
95 __u8 padding;
96};
97
98struct kvm_dtable {
99 __u64 base;
100 __u16 limit;
101 __u16 padding[3];
102};
103
104
105/* for KVM_GET_SREGS and KVM_SET_SREGS */
106struct kvm_sregs {
107 /* out (KVM_GET_SREGS) / in (KVM_SET_SREGS) */
108 struct kvm_segment cs, ds, es, fs, gs, ss;
109 struct kvm_segment tr, ldt;
110 struct kvm_dtable gdt, idt;
111 __u64 cr0, cr2, cr3, cr4, cr8;
112 __u64 efer;
113 __u64 apic_base;
114 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
115};
116
117/* for KVM_GET_FPU and KVM_SET_FPU */
118struct kvm_fpu {
119 __u8 fpr[8][16];
120 __u16 fcw;
121 __u16 fsw;
122 __u8 ftwx; /* in fxsave format */
123 __u8 pad1;
124 __u16 last_opcode;
125 __u64 last_ip;
126 __u64 last_dp;
127 __u8 xmm[16][16];
128 __u32 mxcsr;
129 __u32 pad2;
130};
131
132struct kvm_msr_entry {
133 __u32 index;
134 __u32 reserved;
135 __u64 data;
136};
137
138/* for KVM_GET_MSRS and KVM_SET_MSRS */
139struct kvm_msrs {
140 __u32 nmsrs; /* number of msrs in entries */
141 __u32 pad;
142
143 struct kvm_msr_entry entries[0];
144};
145
146/* for KVM_GET_MSR_INDEX_LIST */
147struct kvm_msr_list {
148 __u32 nmsrs; /* number of msrs in entries */
149 __u32 indices[0];
150};
151
152
153struct kvm_cpuid_entry {
154 __u32 function;
155 __u32 eax;
156 __u32 ebx;
157 __u32 ecx;
158 __u32 edx;
159 __u32 padding;
160};
161
162/* for KVM_SET_CPUID */
163struct kvm_cpuid {
164 __u32 nent;
165 __u32 padding;
166 struct kvm_cpuid_entry entries[0];
167};
168
169struct kvm_cpuid_entry2 {
170 __u32 function;
171 __u32 index;
172 __u32 flags;
173 __u32 eax;
174 __u32 ebx;
175 __u32 ecx;
176 __u32 edx;
177 __u32 padding[3];
178};
179
180#define KVM_CPUID_FLAG_SIGNIFCANT_INDEX 1
181#define KVM_CPUID_FLAG_STATEFUL_FUNC 2
182#define KVM_CPUID_FLAG_STATE_READ_NEXT 4
183
184/* for KVM_SET_CPUID2 */
185struct kvm_cpuid2 {
186 __u32 nent;
187 __u32 padding;
188 struct kvm_cpuid_entry2 entries[0];
189};
190
191#endif
diff --git a/include/asm-x86/kvm_host.h b/include/asm-x86/kvm_host.h
new file mode 100644
index 000000000000..4702b04b979a
--- /dev/null
+++ b/include/asm-x86/kvm_host.h
@@ -0,0 +1,611 @@
1#/*
2 * Kernel-based Virtual Machine driver for Linux
3 *
4 * This header defines architecture specific interfaces, x86 version
5 *
6 * This work is licensed under the terms of the GNU GPL, version 2. See
7 * the COPYING file in the top-level directory.
8 *
9 */
10
11#ifndef ASM_KVM_HOST_H
12#define ASM_KVM_HOST_H
13
14#include <linux/types.h>
15#include <linux/mm.h>
16
17#include <linux/kvm.h>
18#include <linux/kvm_para.h>
19#include <linux/kvm_types.h>
20
21#include <asm/desc.h>
22
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)
26
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 \
32 | X86_CR0_MP)
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)
37
38#define INVALID_PAGE (~(hpa_t)0)
39#define UNMAPPED_GVA (~(gpa_t)0)
40
41#define DE_VECTOR 0
42#define UD_VECTOR 6
43#define NM_VECTOR 7
44#define DF_VECTOR 8
45#define TS_VECTOR 10
46#define NP_VECTOR 11
47#define SS_VECTOR 12
48#define GP_VECTOR 13
49#define PF_VECTOR 14
50
51#define SELECTOR_TI_MASK (1 << 2)
52#define SELECTOR_RPL_MASK 0x03
53
54#define IOPL_SHIFT 12
55
56#define KVM_ALIAS_SLOTS 4
57
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
64
65extern spinlock_t kvm_lock;
66extern struct list_head vm_list;
67
68struct kvm_vcpu;
69struct kvm;
70
71enum {
72 VCPU_REGS_RAX = 0,
73 VCPU_REGS_RCX = 1,
74 VCPU_REGS_RDX = 2,
75 VCPU_REGS_RBX = 3,
76 VCPU_REGS_RSP = 4,
77 VCPU_REGS_RBP = 5,
78 VCPU_REGS_RSI = 6,
79 VCPU_REGS_RDI = 7,
80#ifdef CONFIG_X86_64
81 VCPU_REGS_R8 = 8,
82 VCPU_REGS_R9 = 9,
83 VCPU_REGS_R10 = 10,
84 VCPU_REGS_R11 = 11,
85 VCPU_REGS_R12 = 12,
86 VCPU_REGS_R13 = 13,
87 VCPU_REGS_R14 = 14,
88 VCPU_REGS_R15 = 15,
89#endif
90 NR_VCPU_REGS
91};
92
93enum {
94 VCPU_SREG_CS,
95 VCPU_SREG_DS,
96 VCPU_SREG_ES,
97 VCPU_SREG_FS,
98 VCPU_SREG_GS,
99 VCPU_SREG_SS,
100 VCPU_SREG_TR,
101 VCPU_SREG_LDTR,
102};
103
104#include <asm/kvm_x86_emulate.h>
105
106#define KVM_NR_MEM_OBJS 40
107
108/*
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.
111 */
112struct kvm_mmu_memory_cache {
113 int nobjs;
114 void *objects[KVM_NR_MEM_OBJS];
115};
116
117#define NR_PTE_CHAIN_ENTRIES 5
118
119struct kvm_pte_chain {
120 u64 *parent_ptes[NR_PTE_CHAIN_ENTRIES];
121 struct hlist_node link;
122};
123
124/*
125 * kvm_mmu_page_role, below, is defined as:
126 *
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
132 */
133union kvm_mmu_page_role {
134 unsigned word;
135 struct {
136 unsigned glevels : 4;
137 unsigned level : 4;
138 unsigned quadrant : 2;
139 unsigned pad_for_nice_hex_output : 6;
140 unsigned metaphysical : 1;
141 unsigned access : 3;
142 };
143};
144
145struct kvm_mmu_page {
146 struct list_head link;
147 struct hlist_node hash_link;
148
149 /*
150 * The following two entries are used to key the shadow page in the
151 * hash table.
152 */
153 gfn_t gfn;
154 union kvm_mmu_page_role role;
155
156 u64 *spt;
157 /* hold the gfn of each spte inside spt */
158 gfn_t *gfns;
159 unsigned long slot_bitmap; /* One bit set per slot which has memory
160 * in this shadow page.
161 */
162 int multimapped; /* More than one parent_pte? */
163 int root_count; /* Currently serving as active root */
164 union {
165 u64 *parent_pte; /* !multimapped */
166 struct hlist_head parent_ptes; /* multimapped, kvm_pte_chain */
167 };
168};
169
170/*
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
173 * mode.
174 */
175struct kvm_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);
182 hpa_t root_hpa;
183 int root_level;
184 int shadow_root_level;
185
186 u64 *pae_root;
187};
188
189struct kvm_vcpu_arch {
190 u64 host_tsc;
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() */
196
197 unsigned long cr0;
198 unsigned long cr2;
199 unsigned long cr3;
200 unsigned long cr4;
201 unsigned long cr8;
202 u64 pdptrs[4]; /* pae */
203 u64 shadow_efer;
204 u64 apic_base;
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
211 int mp_state;
212 int sipi_vector;
213 u64 ia32_misc_enable_msr;
214 bool tpr_access_reporting;
215
216 struct kvm_mmu mmu;
217
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;
222
223 gfn_t last_pt_write_gfn;
224 int last_pt_write_count;
225 u64 *last_pte_updated;
226
227 struct {
228 gfn_t gfn; /* presumed gfn during guest pte update */
229 struct page *page; /* page corresponding to that gfn */
230 } update_pte;
231
232 struct i387_fxsave_struct host_fx_image;
233 struct i387_fxsave_struct guest_fx_image;
234
235 gva_t mmio_fault_cr2;
236 struct kvm_pio_request pio;
237 void *pio_data;
238
239 struct kvm_queued_exception {
240 bool pending;
241 bool has_error_code;
242 u8 nr;
243 u32 error_code;
244 } exception;
245
246 struct {
247 int active;
248 u8 save_iopl;
249 struct kvm_save_segment {
250 u16 selector;
251 unsigned long base;
252 u32 limit;
253 u32 ar;
254 } tr, es, ds, fs, gs;
255 } rmode;
256 int halt_request; /* real mode on Intel only */
257
258 int cpuid_nent;
259 struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
260 /* emulate context */
261
262 struct x86_emulate_ctxt emulate_ctxt;
263};
264
265struct kvm_mem_alias {
266 gfn_t base_gfn;
267 unsigned long npages;
268 gfn_t target_gfn;
269};
270
271struct kvm_arch{
272 int naliases;
273 struct kvm_mem_alias aliases[KVM_ALIAS_SLOTS];
274
275 unsigned int n_free_mmu_pages;
276 unsigned int n_requested_mmu_pages;
277 unsigned int n_alloc_mmu_pages;
278 struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
279 /*
280 * Hash table of struct kvm_mmu_page.
281 */
282 struct list_head active_mmu_pages;
283 struct kvm_pic *vpic;
284 struct kvm_ioapic *vioapic;
285
286 int round_robin_prev_vcpu;
287 unsigned int tss_addr;
288 struct page *apic_access_page;
289};
290
291struct kvm_vm_stat {
292 u32 mmu_shadow_zapped;
293 u32 mmu_pte_write;
294 u32 mmu_pte_updated;
295 u32 mmu_pde_zapped;
296 u32 mmu_flooded;
297 u32 mmu_recycled;
298 u32 mmu_cache_miss;
299 u32 remote_tlb_flush;
300};
301
302struct kvm_vcpu_stat {
303 u32 pf_fixed;
304 u32 pf_guest;
305 u32 tlb_flush;
306 u32 invlpg;
307
308 u32 exits;
309 u32 io_exits;
310 u32 mmio_exits;
311 u32 signal_exits;
312 u32 irq_window_exits;
313 u32 halt_exits;
314 u32 halt_wakeup;
315 u32 request_irq_exits;
316 u32 irq_exits;
317 u32 host_state_reload;
318 u32 efer_reload;
319 u32 fpu_reload;
320 u32 insn_emulation;
321 u32 insn_emulation_fail;
322};
323
324struct descriptor_table {
325 u16 limit;
326 unsigned long base;
327} __attribute__((packed));
328
329struct kvm_x86_ops {
330 int (*cpu_has_kvm_support)(void); /* __init */
331 int (*disabled_by_bios)(void); /* __init */
332 void (*hardware_enable)(void *dummy); /* __init */
333 void (*hardware_disable)(void *dummy);
334 void (*check_processor_compatibility)(void *rtn);
335 int (*hardware_setup)(void); /* __init */
336 void (*hardware_unsetup)(void); /* __exit */
337 bool (*cpu_has_accelerated_tpr)(void);
338
339 /* Create, but do not attach this VCPU */
340 struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
341 void (*vcpu_free)(struct kvm_vcpu *vcpu);
342 int (*vcpu_reset)(struct kvm_vcpu *vcpu);
343
344 void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
345 void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
346 void (*vcpu_put)(struct kvm_vcpu *vcpu);
347 void (*vcpu_decache)(struct kvm_vcpu *vcpu);
348
349 int (*set_guest_debug)(struct kvm_vcpu *vcpu,
350 struct kvm_debug_guest *dbg);
351 void (*guest_debug_pre)(struct kvm_vcpu *vcpu);
352 int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
353 int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
354 u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
355 void (*get_segment)(struct kvm_vcpu *vcpu,
356 struct kvm_segment *var, int seg);
357 void (*set_segment)(struct kvm_vcpu *vcpu,
358 struct kvm_segment *var, int seg);
359 void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
360 void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
361 void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
362 void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
363 void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
364 void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
365 void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
366 void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
367 void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
368 void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
369 unsigned long (*get_dr)(struct kvm_vcpu *vcpu, int dr);
370 void (*set_dr)(struct kvm_vcpu *vcpu, int dr, unsigned long value,
371 int *exception);
372 void (*cache_regs)(struct kvm_vcpu *vcpu);
373 void (*decache_regs)(struct kvm_vcpu *vcpu);
374 unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
375 void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
376
377 void (*tlb_flush)(struct kvm_vcpu *vcpu);
378
379 void (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
380 int (*handle_exit)(struct kvm_run *run, struct kvm_vcpu *vcpu);
381 void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
382 void (*patch_hypercall)(struct kvm_vcpu *vcpu,
383 unsigned char *hypercall_addr);
384 int (*get_irq)(struct kvm_vcpu *vcpu);
385 void (*set_irq)(struct kvm_vcpu *vcpu, int vec);
386 void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
387 bool has_error_code, u32 error_code);
388 bool (*exception_injected)(struct kvm_vcpu *vcpu);
389 void (*inject_pending_irq)(struct kvm_vcpu *vcpu);
390 void (*inject_pending_vectors)(struct kvm_vcpu *vcpu,
391 struct kvm_run *run);
392
393 int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
394};
395
396extern struct kvm_x86_ops *kvm_x86_ops;
397
398int kvm_mmu_module_init(void);
399void kvm_mmu_module_exit(void);
400
401void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
402int kvm_mmu_create(struct kvm_vcpu *vcpu);
403int kvm_mmu_setup(struct kvm_vcpu *vcpu);
404void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte);
405
406int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
407void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
408void kvm_mmu_zap_all(struct kvm *kvm);
409unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
410void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
411
412enum emulation_result {
413 EMULATE_DONE, /* no further processing */
414 EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
415 EMULATE_FAIL, /* can't emulate this instruction */
416};
417
418#define EMULTYPE_NO_DECODE (1 << 0)
419#define EMULTYPE_TRAP_UD (1 << 1)
420int emulate_instruction(struct kvm_vcpu *vcpu, struct kvm_run *run,
421 unsigned long cr2, u16 error_code, int emulation_type);
422void kvm_report_emulation_failure(struct kvm_vcpu *cvpu, const char *context);
423void realmode_lgdt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
424void realmode_lidt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
425void realmode_lmsw(struct kvm_vcpu *vcpu, unsigned long msw,
426 unsigned long *rflags);
427
428unsigned long realmode_get_cr(struct kvm_vcpu *vcpu, int cr);
429void realmode_set_cr(struct kvm_vcpu *vcpu, int cr, unsigned long value,
430 unsigned long *rflags);
431int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
432int kvm_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
433
434struct x86_emulate_ctxt;
435
436int kvm_emulate_pio(struct kvm_vcpu *vcpu, struct kvm_run *run, int in,
437 int size, unsigned port);
438int kvm_emulate_pio_string(struct kvm_vcpu *vcpu, struct kvm_run *run, int in,
439 int size, unsigned long count, int down,
440 gva_t address, int rep, unsigned port);
441void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
442int kvm_emulate_halt(struct kvm_vcpu *vcpu);
443int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address);
444int emulate_clts(struct kvm_vcpu *vcpu);
445int emulator_get_dr(struct x86_emulate_ctxt *ctxt, int dr,
446 unsigned long *dest);
447int emulator_set_dr(struct x86_emulate_ctxt *ctxt, int dr,
448 unsigned long value);
449
450void set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
451void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr0);
452void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
453void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
454unsigned long get_cr8(struct kvm_vcpu *vcpu);
455void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
456void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
457
458int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
459int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data);
460
461void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
462void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
463void kvm_inject_page_fault(struct kvm_vcpu *vcpu, unsigned long cr2,
464 u32 error_code);
465
466void fx_init(struct kvm_vcpu *vcpu);
467
468int emulator_read_std(unsigned long addr,
469 void *val,
470 unsigned int bytes,
471 struct kvm_vcpu *vcpu);
472int emulator_write_emulated(unsigned long addr,
473 const void *val,
474 unsigned int bytes,
475 struct kvm_vcpu *vcpu);
476
477unsigned long segment_base(u16 selector);
478
479void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
480void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
481 const u8 *new, int bytes);
482int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
483void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
484int kvm_mmu_load(struct kvm_vcpu *vcpu);
485void kvm_mmu_unload(struct kvm_vcpu *vcpu);
486
487int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
488
489int kvm_fix_hypercall(struct kvm_vcpu *vcpu);
490
491int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code);
492
493int load_pdptrs(struct kvm_vcpu *vcpu, unsigned long cr3);
494int complete_pio(struct kvm_vcpu *vcpu);
495
496static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
497{
498 struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
499
500 return (struct kvm_mmu_page *)page_private(page);
501}
502
503static inline u16 read_fs(void)
504{
505 u16 seg;
506 asm("mov %%fs, %0" : "=g"(seg));
507 return seg;
508}
509
510static inline u16 read_gs(void)
511{
512 u16 seg;
513 asm("mov %%gs, %0" : "=g"(seg));
514 return seg;
515}
516
517static inline u16 read_ldt(void)
518{
519 u16 ldt;
520 asm("sldt %0" : "=g"(ldt));
521 return ldt;
522}
523
524static inline void load_fs(u16 sel)
525{
526 asm("mov %0, %%fs" : : "rm"(sel));
527}
528
529static inline void load_gs(u16 sel)
530{
531 asm("mov %0, %%gs" : : "rm"(sel));
532}
533
534#ifndef load_ldt
535static inline void load_ldt(u16 sel)
536{
537 asm("lldt %0" : : "rm"(sel));
538}
539#endif
540
541static inline void get_idt(struct descriptor_table *table)
542{
543 asm("sidt %0" : "=m"(*table));
544}
545
546static inline void get_gdt(struct descriptor_table *table)
547{
548 asm("sgdt %0" : "=m"(*table));
549}
550
551static inline unsigned long read_tr_base(void)
552{
553 u16 tr;
554 asm("str %0" : "=g"(tr));
555 return segment_base(tr);
556}
557
558#ifdef CONFIG_X86_64
559static inline unsigned long read_msr(unsigned long msr)
560{
561 u64 value;
562
563 rdmsrl(msr, value);
564 return value;
565}
566#endif
567
568static inline void fx_save(struct i387_fxsave_struct *image)
569{
570 asm("fxsave (%0)":: "r" (image));
571}
572
573static inline void fx_restore(struct i387_fxsave_struct *image)
574{
575 asm("fxrstor (%0)":: "r" (image));
576}
577
578static inline void fpu_init(void)
579{
580 asm("finit");
581}
582
583static inline u32 get_rdx_init_val(void)
584{
585 return 0x600; /* P6 family */
586}
587
588static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
589{
590 kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
591}
592
593#define ASM_VMX_VMCLEAR_RAX ".byte 0x66, 0x0f, 0xc7, 0x30"
594#define ASM_VMX_VMLAUNCH ".byte 0x0f, 0x01, 0xc2"
595#define ASM_VMX_VMRESUME ".byte 0x0f, 0x01, 0xc3"
596#define ASM_VMX_VMPTRLD_RAX ".byte 0x0f, 0xc7, 0x30"
597#define ASM_VMX_VMREAD_RDX_RAX ".byte 0x0f, 0x78, 0xd0"
598#define ASM_VMX_VMWRITE_RAX_RDX ".byte 0x0f, 0x79, 0xd0"
599#define ASM_VMX_VMWRITE_RSP_RDX ".byte 0x0f, 0x79, 0xd4"
600#define ASM_VMX_VMXOFF ".byte 0x0f, 0x01, 0xc4"
601#define ASM_VMX_VMXON_RAX ".byte 0xf3, 0x0f, 0xc7, 0x30"
602
603#define MSR_IA32_TIME_STAMP_COUNTER 0x010
604
605#define TSS_IOPB_BASE_OFFSET 0x66
606#define TSS_BASE_SIZE 0x68
607#define TSS_IOPB_SIZE (65536 / 8)
608#define TSS_REDIRECTION_SIZE (256 / 8)
609#define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
610
611#endif
diff --git a/include/asm-x86/kvm_para.h b/include/asm-x86/kvm_para.h
new file mode 100644
index 000000000000..c6f3fd8d8c53
--- /dev/null
+++ b/include/asm-x86/kvm_para.h
@@ -0,0 +1,105 @@
1#ifndef __X86_KVM_PARA_H
2#define __X86_KVM_PARA_H
3
4/* This CPUID returns the signature 'KVMKVMKVM' in ebx, ecx, and edx. It
5 * should be used to determine that a VM is running under KVM.
6 */
7#define KVM_CPUID_SIGNATURE 0x40000000
8
9/* This CPUID returns a feature bitmap in eax. Before enabling a particular
10 * paravirtualization, the appropriate feature bit should be checked.
11 */
12#define KVM_CPUID_FEATURES 0x40000001
13
14#ifdef __KERNEL__
15#include <asm/processor.h>
16
17/* This instruction is vmcall. On non-VT architectures, it will generate a
18 * trap that we will then rewrite to the appropriate instruction.
19 */
20#define KVM_HYPERCALL ".byte 0x0f,0x01,0xc1"
21
22/* For KVM hypercalls, a three-byte sequence of either the vmrun or the vmmrun
23 * instruction. The hypervisor may replace it with something else but only the
24 * instructions are guaranteed to be supported.
25 *
26 * Up to four arguments may be passed in rbx, rcx, rdx, and rsi respectively.
27 * The hypercall number should be placed in rax and the return value will be
28 * placed in rax. No other registers will be clobbered unless explicited
29 * noted by the particular hypercall.
30 */
31
32static inline long kvm_hypercall0(unsigned int nr)
33{
34 long ret;
35 asm volatile(KVM_HYPERCALL
36 : "=a"(ret)
37 : "a"(nr));
38 return ret;
39}
40
41static inline long kvm_hypercall1(unsigned int nr, unsigned long p1)
42{
43 long ret;
44 asm volatile(KVM_HYPERCALL
45 : "=a"(ret)
46 : "a"(nr), "b"(p1));
47 return ret;
48}
49
50static inline long kvm_hypercall2(unsigned int nr, unsigned long p1,
51 unsigned long p2)
52{
53 long ret;
54 asm volatile(KVM_HYPERCALL
55 : "=a"(ret)
56 : "a"(nr), "b"(p1), "c"(p2));
57 return ret;
58}
59
60static inline long kvm_hypercall3(unsigned int nr, unsigned long p1,
61 unsigned long p2, unsigned long p3)
62{
63 long ret;
64 asm volatile(KVM_HYPERCALL
65 : "=a"(ret)
66 : "a"(nr), "b"(p1), "c"(p2), "d"(p3));
67 return ret;
68}
69
70static inline long kvm_hypercall4(unsigned int nr, unsigned long p1,
71 unsigned long p2, unsigned long p3,
72 unsigned long p4)
73{
74 long ret;
75 asm volatile(KVM_HYPERCALL
76 : "=a"(ret)
77 : "a"(nr), "b"(p1), "c"(p2), "d"(p3), "S"(p4));
78 return ret;
79}
80
81static inline int kvm_para_available(void)
82{
83 unsigned int eax, ebx, ecx, edx;
84 char signature[13];
85
86 cpuid(KVM_CPUID_SIGNATURE, &eax, &ebx, &ecx, &edx);
87 memcpy(signature + 0, &ebx, 4);
88 memcpy(signature + 4, &ecx, 4);
89 memcpy(signature + 8, &edx, 4);
90 signature[12] = 0;
91
92 if (strcmp(signature, "KVMKVMKVM") == 0)
93 return 1;
94
95 return 0;
96}
97
98static inline unsigned int kvm_arch_para_features(void)
99{
100 return cpuid_eax(KVM_CPUID_FEATURES);
101}
102
103#endif
104
105#endif
diff --git a/include/asm-x86/kvm_x86_emulate.h b/include/asm-x86/kvm_x86_emulate.h
new file mode 100644
index 000000000000..7db91b9bdcd4
--- /dev/null
+++ b/include/asm-x86/kvm_x86_emulate.h
@@ -0,0 +1,186 @@
1/******************************************************************************
2 * x86_emulate.h
3 *
4 * Generic x86 (32-bit and 64-bit) instruction decoder and emulator.
5 *
6 * Copyright (c) 2005 Keir Fraser
7 *
8 * From: xen-unstable 10676:af9809f51f81a3c43f276f00c81a52ef558afda4
9 */
10
11#ifndef __X86_EMULATE_H__
12#define __X86_EMULATE_H__
13
14struct x86_emulate_ctxt;
15
16/*
17 * x86_emulate_ops:
18 *
19 * These operations represent the instruction emulator's interface to memory.
20 * There are two categories of operation: those that act on ordinary memory
21 * regions (*_std), and those that act on memory regions known to require
22 * special treatment or emulation (*_emulated).
23 *
24 * The emulator assumes that an instruction accesses only one 'emulated memory'
25 * location, that this location is the given linear faulting address (cr2), and
26 * that this is one of the instruction's data operands. Instruction fetches and
27 * stack operations are assumed never to access emulated memory. The emulator
28 * automatically deduces which operand of a string-move operation is accessing
29 * emulated memory, and assumes that the other operand accesses normal memory.
30 *
31 * NOTES:
32 * 1. The emulator isn't very smart about emulated vs. standard memory.
33 * 'Emulated memory' access addresses should be checked for sanity.
34 * 'Normal memory' accesses may fault, and the caller must arrange to
35 * detect and handle reentrancy into the emulator via recursive faults.
36 * Accesses may be unaligned and may cross page boundaries.
37 * 2. If the access fails (cannot emulate, or a standard access faults) then
38 * it is up to the memop to propagate the fault to the guest VM via
39 * some out-of-band mechanism, unknown to the emulator. The memop signals
40 * failure by returning X86EMUL_PROPAGATE_FAULT to the emulator, which will
41 * then immediately bail.
42 * 3. Valid access sizes are 1, 2, 4 and 8 bytes. On x86/32 systems only
43 * cmpxchg8b_emulated need support 8-byte accesses.
44 * 4. The emulator cannot handle 64-bit mode emulation on an x86/32 system.
45 */
46/* Access completed successfully: continue emulation as normal. */
47#define X86EMUL_CONTINUE 0
48/* Access is unhandleable: bail from emulation and return error to caller. */
49#define X86EMUL_UNHANDLEABLE 1
50/* Terminate emulation but return success to the caller. */
51#define X86EMUL_PROPAGATE_FAULT 2 /* propagate a generated fault to guest */
52#define X86EMUL_RETRY_INSTR 2 /* retry the instruction for some reason */
53#define X86EMUL_CMPXCHG_FAILED 2 /* cmpxchg did not see expected value */
54struct x86_emulate_ops {
55 /*
56 * read_std: Read bytes of standard (non-emulated/special) memory.
57 * Used for instruction fetch, stack operations, and others.
58 * @addr: [IN ] Linear address from which to read.
59 * @val: [OUT] Value read from memory, zero-extended to 'u_long'.
60 * @bytes: [IN ] Number of bytes to read from memory.
61 */
62 int (*read_std)(unsigned long addr, void *val,
63 unsigned int bytes, struct kvm_vcpu *vcpu);
64
65 /*
66 * read_emulated: Read bytes from emulated/special memory area.
67 * @addr: [IN ] Linear address from which to read.
68 * @val: [OUT] Value read from memory, zero-extended to 'u_long'.
69 * @bytes: [IN ] Number of bytes to read from memory.
70 */
71 int (*read_emulated) (unsigned long addr,
72 void *val,
73 unsigned int bytes,
74 struct kvm_vcpu *vcpu);
75
76 /*
77 * write_emulated: Read bytes from emulated/special memory area.
78 * @addr: [IN ] Linear address to which to write.
79 * @val: [IN ] Value to write to memory (low-order bytes used as
80 * required).
81 * @bytes: [IN ] Number of bytes to write to memory.
82 */
83 int (*write_emulated) (unsigned long addr,
84 const void *val,
85 unsigned int bytes,
86 struct kvm_vcpu *vcpu);
87
88 /*
89 * cmpxchg_emulated: Emulate an atomic (LOCKed) CMPXCHG operation on an
90 * emulated/special memory area.
91 * @addr: [IN ] Linear address to access.
92 * @old: [IN ] Value expected to be current at @addr.
93 * @new: [IN ] Value to write to @addr.
94 * @bytes: [IN ] Number of bytes to access using CMPXCHG.
95 */
96 int (*cmpxchg_emulated) (unsigned long addr,
97 const void *old,
98 const void *new,
99 unsigned int bytes,
100 struct kvm_vcpu *vcpu);
101
102};
103
104/* Type, address-of, and value of an instruction's operand. */
105struct operand {
106 enum { OP_REG, OP_MEM, OP_IMM, OP_NONE } type;
107 unsigned int bytes;
108 unsigned long val, orig_val, *ptr;
109};
110
111struct fetch_cache {
112 u8 data[15];
113 unsigned long start;
114 unsigned long end;
115};
116
117struct decode_cache {
118 u8 twobyte;
119 u8 b;
120 u8 lock_prefix;
121 u8 rep_prefix;
122 u8 op_bytes;
123 u8 ad_bytes;
124 u8 rex_prefix;
125 struct operand src;
126 struct operand dst;
127 unsigned long *override_base;
128 unsigned int d;
129 unsigned long regs[NR_VCPU_REGS];
130 unsigned long eip;
131 /* modrm */
132 u8 modrm;
133 u8 modrm_mod;
134 u8 modrm_reg;
135 u8 modrm_rm;
136 u8 use_modrm_ea;
137 unsigned long modrm_ea;
138 unsigned long modrm_val;
139 struct fetch_cache fetch;
140};
141
142struct x86_emulate_ctxt {
143 /* Register state before/after emulation. */
144 struct kvm_vcpu *vcpu;
145
146 /* Linear faulting address (if emulating a page-faulting instruction). */
147 unsigned long eflags;
148
149 /* Emulated execution mode, represented by an X86EMUL_MODE value. */
150 int mode;
151
152 unsigned long cs_base;
153 unsigned long ds_base;
154 unsigned long es_base;
155 unsigned long ss_base;
156 unsigned long gs_base;
157 unsigned long fs_base;
158
159 /* decode cache */
160
161 struct decode_cache decode;
162};
163
164/* Repeat String Operation Prefix */
165#define REPE_PREFIX 1
166#define REPNE_PREFIX 2
167
168/* Execution mode, passed to the emulator. */
169#define X86EMUL_MODE_REAL 0 /* Real mode. */
170#define X86EMUL_MODE_PROT16 2 /* 16-bit protected mode. */
171#define X86EMUL_MODE_PROT32 4 /* 32-bit protected mode. */
172#define X86EMUL_MODE_PROT64 8 /* 64-bit (long) mode. */
173
174/* Host execution mode. */
175#if defined(__i386__)
176#define X86EMUL_MODE_HOST X86EMUL_MODE_PROT32
177#elif defined(CONFIG_X86_64)
178#define X86EMUL_MODE_HOST X86EMUL_MODE_PROT64
179#endif
180
181int x86_decode_insn(struct x86_emulate_ctxt *ctxt,
182 struct x86_emulate_ops *ops);
183int x86_emulate_insn(struct x86_emulate_ctxt *ctxt,
184 struct x86_emulate_ops *ops);
185
186#endif /* __X86_EMULATE_H__ */
diff --git a/include/linux/Kbuild b/include/linux/Kbuild
index 27b9350052b4..85b2482cc736 100644
--- a/include/linux/Kbuild
+++ b/include/linux/Kbuild
@@ -100,7 +100,6 @@ header-y += iso_fs.h
100header-y += ixjuser.h 100header-y += ixjuser.h
101header-y += jffs2.h 101header-y += jffs2.h
102header-y += keyctl.h 102header-y += keyctl.h
103header-y += kvm.h
104header-y += limits.h 103header-y += limits.h
105header-y += lock_dlm_plock.h 104header-y += lock_dlm_plock.h
106header-y += magic.h 105header-y += magic.h
@@ -256,6 +255,7 @@ unifdef-y += kd.h
256unifdef-y += kernelcapi.h 255unifdef-y += kernelcapi.h
257unifdef-y += kernel.h 256unifdef-y += kernel.h
258unifdef-y += keyboard.h 257unifdef-y += keyboard.h
258unifdef-$(CONFIG_HAVE_KVM) += kvm.h
259unifdef-y += llc.h 259unifdef-y += llc.h
260unifdef-y += loop.h 260unifdef-y += loop.h
261unifdef-y += lp.h 261unifdef-y += lp.h
diff --git a/include/linux/kvm.h b/include/linux/kvm.h
index 057a7f34ee36..4de4fd2d8607 100644
--- a/include/linux/kvm.h
+++ b/include/linux/kvm.h
@@ -9,12 +9,10 @@
9 9
10#include <asm/types.h> 10#include <asm/types.h>
11#include <linux/ioctl.h> 11#include <linux/ioctl.h>
12#include <asm/kvm.h>
12 13
13#define KVM_API_VERSION 12 14#define KVM_API_VERSION 12
14 15
15/* Architectural interrupt line count. */
16#define KVM_NR_INTERRUPTS 256
17
18/* for KVM_CREATE_MEMORY_REGION */ 16/* for KVM_CREATE_MEMORY_REGION */
19struct kvm_memory_region { 17struct kvm_memory_region {
20 __u32 slot; 18 __u32 slot;
@@ -23,17 +21,19 @@ struct kvm_memory_region {
23 __u64 memory_size; /* bytes */ 21 __u64 memory_size; /* bytes */
24}; 22};
25 23
26/* for kvm_memory_region::flags */ 24/* for KVM_SET_USER_MEMORY_REGION */
27#define KVM_MEM_LOG_DIRTY_PAGES 1UL 25struct kvm_userspace_memory_region {
28 26 __u32 slot;
29struct kvm_memory_alias {
30 __u32 slot; /* this has a different namespace than memory slots */
31 __u32 flags; 27 __u32 flags;
32 __u64 guest_phys_addr; 28 __u64 guest_phys_addr;
33 __u64 memory_size; 29 __u64 memory_size; /* bytes */
34 __u64 target_phys_addr; 30 __u64 userspace_addr; /* start of the userspace allocated memory */
35}; 31};
36 32
33/* for kvm_memory_region::flags */
34#define KVM_MEM_LOG_DIRTY_PAGES 1UL
35
36
37/* for KVM_IRQ_LINE */ 37/* for KVM_IRQ_LINE */
38struct kvm_irq_level { 38struct kvm_irq_level {
39 /* 39 /*
@@ -45,62 +45,18 @@ struct kvm_irq_level {
45 __u32 level; 45 __u32 level;
46}; 46};
47 47
48/* for KVM_GET_IRQCHIP and KVM_SET_IRQCHIP */
49struct kvm_pic_state {
50 __u8 last_irr; /* edge detection */
51 __u8 irr; /* interrupt request register */
52 __u8 imr; /* interrupt mask register */
53 __u8 isr; /* interrupt service register */
54 __u8 priority_add; /* highest irq priority */
55 __u8 irq_base;
56 __u8 read_reg_select;
57 __u8 poll;
58 __u8 special_mask;
59 __u8 init_state;
60 __u8 auto_eoi;
61 __u8 rotate_on_auto_eoi;
62 __u8 special_fully_nested_mode;
63 __u8 init4; /* true if 4 byte init */
64 __u8 elcr; /* PIIX edge/trigger selection */
65 __u8 elcr_mask;
66};
67
68#define KVM_IOAPIC_NUM_PINS 24
69struct kvm_ioapic_state {
70 __u64 base_address;
71 __u32 ioregsel;
72 __u32 id;
73 __u32 irr;
74 __u32 pad;
75 union {
76 __u64 bits;
77 struct {
78 __u8 vector;
79 __u8 delivery_mode:3;
80 __u8 dest_mode:1;
81 __u8 delivery_status:1;
82 __u8 polarity:1;
83 __u8 remote_irr:1;
84 __u8 trig_mode:1;
85 __u8 mask:1;
86 __u8 reserve:7;
87 __u8 reserved[4];
88 __u8 dest_id;
89 } fields;
90 } redirtbl[KVM_IOAPIC_NUM_PINS];
91};
92
93#define KVM_IRQCHIP_PIC_MASTER 0
94#define KVM_IRQCHIP_PIC_SLAVE 1
95#define KVM_IRQCHIP_IOAPIC 2
96 48
97struct kvm_irqchip { 49struct kvm_irqchip {
98 __u32 chip_id; 50 __u32 chip_id;
99 __u32 pad; 51 __u32 pad;
100 union { 52 union {
101 char dummy[512]; /* reserving space */ 53 char dummy[512]; /* reserving space */
54#ifdef CONFIG_X86
102 struct kvm_pic_state pic; 55 struct kvm_pic_state pic;
56#endif
57#if defined(CONFIG_X86) || defined(CONFIG_IA64)
103 struct kvm_ioapic_state ioapic; 58 struct kvm_ioapic_state ioapic;
59#endif
104 } chip; 60 } chip;
105}; 61};
106 62
@@ -116,6 +72,7 @@ struct kvm_irqchip {
116#define KVM_EXIT_FAIL_ENTRY 9 72#define KVM_EXIT_FAIL_ENTRY 9
117#define KVM_EXIT_INTR 10 73#define KVM_EXIT_INTR 10
118#define KVM_EXIT_SET_TPR 11 74#define KVM_EXIT_SET_TPR 11
75#define KVM_EXIT_TPR_ACCESS 12
119 76
120/* for KVM_RUN, returned by mmap(vcpu_fd, offset=0) */ 77/* for KVM_RUN, returned by mmap(vcpu_fd, offset=0) */
121struct kvm_run { 78struct kvm_run {
@@ -174,90 +131,17 @@ struct kvm_run {
174 __u32 longmode; 131 __u32 longmode;
175 __u32 pad; 132 __u32 pad;
176 } hypercall; 133 } hypercall;
134 /* KVM_EXIT_TPR_ACCESS */
135 struct {
136 __u64 rip;
137 __u32 is_write;
138 __u32 pad;
139 } tpr_access;
177 /* Fix the size of the union. */ 140 /* Fix the size of the union. */
178 char padding[256]; 141 char padding[256];
179 }; 142 };
180}; 143};
181 144
182/* for KVM_GET_REGS and KVM_SET_REGS */
183struct kvm_regs {
184 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
185 __u64 rax, rbx, rcx, rdx;
186 __u64 rsi, rdi, rsp, rbp;
187 __u64 r8, r9, r10, r11;
188 __u64 r12, r13, r14, r15;
189 __u64 rip, rflags;
190};
191
192/* for KVM_GET_FPU and KVM_SET_FPU */
193struct kvm_fpu {
194 __u8 fpr[8][16];
195 __u16 fcw;
196 __u16 fsw;
197 __u8 ftwx; /* in fxsave format */
198 __u8 pad1;
199 __u16 last_opcode;
200 __u64 last_ip;
201 __u64 last_dp;
202 __u8 xmm[16][16];
203 __u32 mxcsr;
204 __u32 pad2;
205};
206
207/* for KVM_GET_LAPIC and KVM_SET_LAPIC */
208#define KVM_APIC_REG_SIZE 0x400
209struct kvm_lapic_state {
210 char regs[KVM_APIC_REG_SIZE];
211};
212
213struct kvm_segment {
214 __u64 base;
215 __u32 limit;
216 __u16 selector;
217 __u8 type;
218 __u8 present, dpl, db, s, l, g, avl;
219 __u8 unusable;
220 __u8 padding;
221};
222
223struct kvm_dtable {
224 __u64 base;
225 __u16 limit;
226 __u16 padding[3];
227};
228
229/* for KVM_GET_SREGS and KVM_SET_SREGS */
230struct kvm_sregs {
231 /* out (KVM_GET_SREGS) / in (KVM_SET_SREGS) */
232 struct kvm_segment cs, ds, es, fs, gs, ss;
233 struct kvm_segment tr, ldt;
234 struct kvm_dtable gdt, idt;
235 __u64 cr0, cr2, cr3, cr4, cr8;
236 __u64 efer;
237 __u64 apic_base;
238 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
239};
240
241struct kvm_msr_entry {
242 __u32 index;
243 __u32 reserved;
244 __u64 data;
245};
246
247/* for KVM_GET_MSRS and KVM_SET_MSRS */
248struct kvm_msrs {
249 __u32 nmsrs; /* number of msrs in entries */
250 __u32 pad;
251
252 struct kvm_msr_entry entries[0];
253};
254
255/* for KVM_GET_MSR_INDEX_LIST */
256struct kvm_msr_list {
257 __u32 nmsrs; /* number of msrs in entries */
258 __u32 indices[0];
259};
260
261/* for KVM_TRANSLATE */ 145/* for KVM_TRANSLATE */
262struct kvm_translation { 146struct kvm_translation {
263 /* in */ 147 /* in */
@@ -302,28 +186,24 @@ struct kvm_dirty_log {
302 }; 186 };
303}; 187};
304 188
305struct kvm_cpuid_entry {
306 __u32 function;
307 __u32 eax;
308 __u32 ebx;
309 __u32 ecx;
310 __u32 edx;
311 __u32 padding;
312};
313
314/* for KVM_SET_CPUID */
315struct kvm_cpuid {
316 __u32 nent;
317 __u32 padding;
318 struct kvm_cpuid_entry entries[0];
319};
320
321/* for KVM_SET_SIGNAL_MASK */ 189/* for KVM_SET_SIGNAL_MASK */
322struct kvm_signal_mask { 190struct kvm_signal_mask {
323 __u32 len; 191 __u32 len;
324 __u8 sigset[0]; 192 __u8 sigset[0];
325}; 193};
326 194
195/* for KVM_TPR_ACCESS_REPORTING */
196struct kvm_tpr_access_ctl {
197 __u32 enabled;
198 __u32 flags;
199 __u32 reserved[8];
200};
201
202/* for KVM_SET_VAPIC_ADDR */
203struct kvm_vapic_addr {
204 __u64 vapic_addr;
205};
206
327#define KVMIO 0xAE 207#define KVMIO 0xAE
328 208
329/* 209/*
@@ -347,11 +227,21 @@ struct kvm_signal_mask {
347 */ 227 */
348#define KVM_CAP_IRQCHIP 0 228#define KVM_CAP_IRQCHIP 0
349#define KVM_CAP_HLT 1 229#define KVM_CAP_HLT 1
230#define KVM_CAP_MMU_SHADOW_CACHE_CONTROL 2
231#define KVM_CAP_USER_MEMORY 3
232#define KVM_CAP_SET_TSS_ADDR 4
233#define KVM_CAP_EXT_CPUID 5
234#define KVM_CAP_VAPIC 6
350 235
351/* 236/*
352 * ioctls for VM fds 237 * ioctls for VM fds
353 */ 238 */
354#define KVM_SET_MEMORY_REGION _IOW(KVMIO, 0x40, struct kvm_memory_region) 239#define KVM_SET_MEMORY_REGION _IOW(KVMIO, 0x40, struct kvm_memory_region)
240#define KVM_SET_NR_MMU_PAGES _IO(KVMIO, 0x44)
241#define KVM_GET_NR_MMU_PAGES _IO(KVMIO, 0x45)
242#define KVM_SET_USER_MEMORY_REGION _IOW(KVMIO, 0x46,\
243 struct kvm_userspace_memory_region)
244#define KVM_SET_TSS_ADDR _IO(KVMIO, 0x47)
355/* 245/*
356 * KVM_CREATE_VCPU receives as a parameter the vcpu slot, and returns 246 * KVM_CREATE_VCPU receives as a parameter the vcpu slot, and returns
357 * a vcpu fd. 247 * a vcpu fd.
@@ -359,6 +249,7 @@ struct kvm_signal_mask {
359#define KVM_CREATE_VCPU _IO(KVMIO, 0x41) 249#define KVM_CREATE_VCPU _IO(KVMIO, 0x41)
360#define KVM_GET_DIRTY_LOG _IOW(KVMIO, 0x42, struct kvm_dirty_log) 250#define KVM_GET_DIRTY_LOG _IOW(KVMIO, 0x42, struct kvm_dirty_log)
361#define KVM_SET_MEMORY_ALIAS _IOW(KVMIO, 0x43, struct kvm_memory_alias) 251#define KVM_SET_MEMORY_ALIAS _IOW(KVMIO, 0x43, struct kvm_memory_alias)
252#define KVM_GET_SUPPORTED_CPUID _IOWR(KVMIO, 0x48, struct kvm_cpuid2)
362/* Device model IOC */ 253/* Device model IOC */
363#define KVM_CREATE_IRQCHIP _IO(KVMIO, 0x60) 254#define KVM_CREATE_IRQCHIP _IO(KVMIO, 0x60)
364#define KVM_IRQ_LINE _IOW(KVMIO, 0x61, struct kvm_irq_level) 255#define KVM_IRQ_LINE _IOW(KVMIO, 0x61, struct kvm_irq_level)
@@ -384,5 +275,11 @@ struct kvm_signal_mask {
384#define KVM_SET_FPU _IOW(KVMIO, 0x8d, struct kvm_fpu) 275#define KVM_SET_FPU _IOW(KVMIO, 0x8d, struct kvm_fpu)
385#define KVM_GET_LAPIC _IOR(KVMIO, 0x8e, struct kvm_lapic_state) 276#define KVM_GET_LAPIC _IOR(KVMIO, 0x8e, struct kvm_lapic_state)
386#define KVM_SET_LAPIC _IOW(KVMIO, 0x8f, struct kvm_lapic_state) 277#define KVM_SET_LAPIC _IOW(KVMIO, 0x8f, struct kvm_lapic_state)
278#define KVM_SET_CPUID2 _IOW(KVMIO, 0x90, struct kvm_cpuid2)
279#define KVM_GET_CPUID2 _IOWR(KVMIO, 0x91, struct kvm_cpuid2)
280/* Available with KVM_CAP_VAPIC */
281#define KVM_TPR_ACCESS_REPORTING _IOWR(KVMIO, 0x92, struct kvm_tpr_access_ctl)
282/* Available with KVM_CAP_VAPIC */
283#define KVM_SET_VAPIC_ADDR _IOW(KVMIO, 0x93, struct kvm_vapic_addr)
387 284
388#endif 285#endif
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
new file mode 100644
index 000000000000..ea4764b0a2f4
--- /dev/null
+++ b/include/linux/kvm_host.h
@@ -0,0 +1,299 @@
1#ifndef __KVM_HOST_H
2#define __KVM_HOST_H
3
4/*
5 * This work is licensed under the terms of the GNU GPL, version 2. See
6 * the COPYING file in the top-level directory.
7 */
8
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>
16#include <linux/mm.h>
17#include <linux/preempt.h>
18#include <asm/signal.h>
19
20#include <linux/kvm.h>
21#include <linux/kvm_para.h>
22
23#include <linux/kvm_types.h>
24
25#include <asm/kvm_host.h>
26
27#define KVM_MAX_VCPUS 4
28#define KVM_MEMORY_SLOTS 8
29/* memory slots that does not exposed to userspace */
30#define KVM_PRIVATE_MEM_SLOTS 4
31
32#define KVM_PIO_PAGE_OFFSET 1
33
34/*
35 * vcpu->requests bit members
36 */
37#define KVM_REQ_TLB_FLUSH 0
38#define KVM_REQ_MIGRATE_TIMER 1
39#define KVM_REQ_REPORT_TPR_ACCESS 2
40
41struct kvm_vcpu;
42extern struct kmem_cache *kvm_vcpu_cache;
43
44struct kvm_guest_debug {
45 int enabled;
46 unsigned long bp[4];
47 int singlestep;
48};
49
50/*
51 * It would be nice to use something smarter than a linear search, TBD...
52 * Thankfully we dont expect many devices to register (famous last words :),
53 * so until then it will suffice. At least its abstracted so we can change
54 * in one place.
55 */
56struct kvm_io_bus {
57 int dev_count;
58#define NR_IOBUS_DEVS 6
59 struct kvm_io_device *devs[NR_IOBUS_DEVS];
60};
61
62void kvm_io_bus_init(struct kvm_io_bus *bus);
63void kvm_io_bus_destroy(struct kvm_io_bus *bus);
64struct kvm_io_device *kvm_io_bus_find_dev(struct kvm_io_bus *bus, gpa_t addr);
65void kvm_io_bus_register_dev(struct kvm_io_bus *bus,
66 struct kvm_io_device *dev);
67
68struct kvm_vcpu {
69 struct kvm *kvm;
70 struct preempt_notifier preempt_notifier;
71 int vcpu_id;
72 struct mutex mutex;
73 int cpu;
74 struct kvm_run *run;
75 int guest_mode;
76 unsigned long requests;
77 struct kvm_guest_debug guest_debug;
78 int fpu_active;
79 int guest_fpu_loaded;
80 wait_queue_head_t wq;
81 int sigset_active;
82 sigset_t sigset;
83 struct kvm_vcpu_stat stat;
84
85#ifdef CONFIG_HAS_IOMEM
86 int mmio_needed;
87 int mmio_read_completed;
88 int mmio_is_write;
89 int mmio_size;
90 unsigned char mmio_data[8];
91 gpa_t mmio_phys_addr;
92#endif
93
94 struct kvm_vcpu_arch arch;
95};
96
97struct kvm_memory_slot {
98 gfn_t base_gfn;
99 unsigned long npages;
100 unsigned long flags;
101 unsigned long *rmap;
102 unsigned long *dirty_bitmap;
103 unsigned long userspace_addr;
104 int user_alloc;
105};
106
107struct kvm {
108 struct mutex lock; /* protects the vcpus array and APIC accesses */
109 spinlock_t mmu_lock;
110 struct mm_struct *mm; /* userspace tied to this vm */
111 int nmemslots;
112 struct kvm_memory_slot memslots[KVM_MEMORY_SLOTS +
113 KVM_PRIVATE_MEM_SLOTS];
114 struct kvm_vcpu *vcpus[KVM_MAX_VCPUS];
115 struct list_head vm_list;
116 struct file *filp;
117 struct kvm_io_bus mmio_bus;
118 struct kvm_io_bus pio_bus;
119 struct kvm_vm_stat stat;
120 struct kvm_arch arch;
121};
122
123/* The guest did something we don't support. */
124#define pr_unimpl(vcpu, fmt, ...) \
125 do { \
126 if (printk_ratelimit()) \
127 printk(KERN_ERR "kvm: %i: cpu%i " fmt, \
128 current->tgid, (vcpu)->vcpu_id , ## __VA_ARGS__); \
129 } while (0)
130
131#define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
132#define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
133
134int kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id);
135void kvm_vcpu_uninit(struct kvm_vcpu *vcpu);
136
137void vcpu_load(struct kvm_vcpu *vcpu);
138void vcpu_put(struct kvm_vcpu *vcpu);
139
140void decache_vcpus_on_cpu(int cpu);
141
142
143int kvm_init(void *opaque, unsigned int vcpu_size,
144 struct module *module);
145void kvm_exit(void);
146
147#define HPA_MSB ((sizeof(hpa_t) * 8) - 1)
148#define HPA_ERR_MASK ((hpa_t)1 << HPA_MSB)
149static inline int is_error_hpa(hpa_t hpa) { return hpa >> HPA_MSB; }
150struct page *gva_to_page(struct kvm_vcpu *vcpu, gva_t gva);
151
152extern struct page *bad_page;
153
154int is_error_page(struct page *page);
155int kvm_is_error_hva(unsigned long addr);
156int kvm_set_memory_region(struct kvm *kvm,
157 struct kvm_userspace_memory_region *mem,
158 int user_alloc);
159int __kvm_set_memory_region(struct kvm *kvm,
160 struct kvm_userspace_memory_region *mem,
161 int user_alloc);
162int kvm_arch_set_memory_region(struct kvm *kvm,
163 struct kvm_userspace_memory_region *mem,
164 struct kvm_memory_slot old,
165 int user_alloc);
166gfn_t unalias_gfn(struct kvm *kvm, gfn_t gfn);
167struct page *gfn_to_page(struct kvm *kvm, gfn_t gfn);
168void kvm_release_page_clean(struct page *page);
169void kvm_release_page_dirty(struct page *page);
170int kvm_read_guest_page(struct kvm *kvm, gfn_t gfn, void *data, int offset,
171 int len);
172int kvm_read_guest_atomic(struct kvm *kvm, gpa_t gpa, void *data,
173 unsigned long len);
174int kvm_read_guest(struct kvm *kvm, gpa_t gpa, void *data, unsigned long len);
175int kvm_write_guest_page(struct kvm *kvm, gfn_t gfn, const void *data,
176 int offset, int len);
177int kvm_write_guest(struct kvm *kvm, gpa_t gpa, const void *data,
178 unsigned long len);
179int kvm_clear_guest_page(struct kvm *kvm, gfn_t gfn, int offset, int len);
180int kvm_clear_guest(struct kvm *kvm, gpa_t gpa, unsigned long len);
181struct kvm_memory_slot *gfn_to_memslot(struct kvm *kvm, gfn_t gfn);
182int kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn);
183void mark_page_dirty(struct kvm *kvm, gfn_t gfn);
184
185void kvm_vcpu_block(struct kvm_vcpu *vcpu);
186void kvm_resched(struct kvm_vcpu *vcpu);
187void kvm_load_guest_fpu(struct kvm_vcpu *vcpu);
188void kvm_put_guest_fpu(struct kvm_vcpu *vcpu);
189void kvm_flush_remote_tlbs(struct kvm *kvm);
190
191long kvm_arch_dev_ioctl(struct file *filp,
192 unsigned int ioctl, unsigned long arg);
193long kvm_arch_vcpu_ioctl(struct file *filp,
194 unsigned int ioctl, unsigned long arg);
195void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
196void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
197
198int kvm_dev_ioctl_check_extension(long ext);
199
200int kvm_get_dirty_log(struct kvm *kvm,
201 struct kvm_dirty_log *log, int *is_dirty);
202int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
203 struct kvm_dirty_log *log);
204
205int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
206 struct
207 kvm_userspace_memory_region *mem,
208 int user_alloc);
209long kvm_arch_vm_ioctl(struct file *filp,
210 unsigned int ioctl, unsigned long arg);
211void kvm_arch_destroy_vm(struct kvm *kvm);
212
213int kvm_arch_vcpu_ioctl_get_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu);
214int kvm_arch_vcpu_ioctl_set_fpu(struct kvm_vcpu *vcpu, struct kvm_fpu *fpu);
215
216int kvm_arch_vcpu_ioctl_translate(struct kvm_vcpu *vcpu,
217 struct kvm_translation *tr);
218
219int kvm_arch_vcpu_ioctl_get_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs);
220int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs);
221int kvm_arch_vcpu_ioctl_get_sregs(struct kvm_vcpu *vcpu,
222 struct kvm_sregs *sregs);
223int kvm_arch_vcpu_ioctl_set_sregs(struct kvm_vcpu *vcpu,
224 struct kvm_sregs *sregs);
225int kvm_arch_vcpu_ioctl_debug_guest(struct kvm_vcpu *vcpu,
226 struct kvm_debug_guest *dbg);
227int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run);
228
229int kvm_arch_init(void *opaque);
230void kvm_arch_exit(void);
231
232int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu);
233void kvm_arch_vcpu_uninit(struct kvm_vcpu *vcpu);
234
235void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu);
236void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
237void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
238struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id);
239int kvm_arch_vcpu_setup(struct kvm_vcpu *vcpu);
240void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu);
241
242int kvm_arch_vcpu_reset(struct kvm_vcpu *vcpu);
243void kvm_arch_hardware_enable(void *garbage);
244void kvm_arch_hardware_disable(void *garbage);
245int kvm_arch_hardware_setup(void);
246void kvm_arch_hardware_unsetup(void);
247void kvm_arch_check_processor_compat(void *rtn);
248int kvm_arch_vcpu_runnable(struct kvm_vcpu *vcpu);
249
250void kvm_free_physmem(struct kvm *kvm);
251
252struct kvm *kvm_arch_create_vm(void);
253void kvm_arch_destroy_vm(struct kvm *kvm);
254
255int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
256int kvm_cpu_has_interrupt(struct kvm_vcpu *v);
257void kvm_vcpu_kick(struct kvm_vcpu *vcpu);
258
259static inline void kvm_guest_enter(void)
260{
261 account_system_vtime(current);
262 current->flags |= PF_VCPU;
263}
264
265static inline void kvm_guest_exit(void)
266{
267 account_system_vtime(current);
268 current->flags &= ~PF_VCPU;
269}
270
271static inline int memslot_id(struct kvm *kvm, struct kvm_memory_slot *slot)
272{
273 return slot - kvm->memslots;
274}
275
276static inline gpa_t gfn_to_gpa(gfn_t gfn)
277{
278 return (gpa_t)gfn << PAGE_SHIFT;
279}
280
281static inline void kvm_migrate_apic_timer(struct kvm_vcpu *vcpu)
282{
283 set_bit(KVM_REQ_MIGRATE_TIMER, &vcpu->requests);
284}
285
286enum kvm_stat_kind {
287 KVM_STAT_VM,
288 KVM_STAT_VCPU,
289};
290
291struct kvm_stats_debugfs_item {
292 const char *name;
293 int offset;
294 enum kvm_stat_kind kind;
295 struct dentry *dentry;
296};
297extern struct kvm_stats_debugfs_item debugfs_entries[];
298
299#endif
diff --git a/include/linux/kvm_para.h b/include/linux/kvm_para.h
index 3b292565a693..5497aac0d2f8 100644
--- a/include/linux/kvm_para.h
+++ b/include/linux/kvm_para.h
@@ -2,72 +2,30 @@
2#define __LINUX_KVM_PARA_H 2#define __LINUX_KVM_PARA_H
3 3
4/* 4/*
5 * Guest OS interface for KVM paravirtualization 5 * This header file provides a method for making a hypercall to the host
6 * 6 * Architectures should define:
7 * Note: this interface is totally experimental, and is certain to change 7 * - kvm_hypercall0, kvm_hypercall1...
8 * as we make progress. 8 * - kvm_arch_para_features
9 * - kvm_para_available
9 */ 10 */
10 11
11/* 12/* Return values for hypercalls */
12 * Per-VCPU descriptor area shared between guest and host. Writable to 13#define KVM_ENOSYS 1000
13 * both guest and host. Registered with the host by the guest when
14 * a guest acknowledges paravirtual mode.
15 *
16 * NOTE: all addresses are guest-physical addresses (gpa), to make it
17 * easier for the hypervisor to map between the various addresses.
18 */
19struct kvm_vcpu_para_state {
20 /*
21 * API version information for compatibility. If there's any support
22 * mismatch (too old host trying to execute too new guest) then
23 * the host will deny entry into paravirtual mode. Any other
24 * combination (new host + old guest and new host + new guest)
25 * is supposed to work - new host versions will support all old
26 * guest API versions.
27 */
28 u32 guest_version;
29 u32 host_version;
30 u32 size;
31 u32 ret;
32
33 /*
34 * The address of the vm exit instruction (VMCALL or VMMCALL),
35 * which the host will patch according to the CPU model the
36 * VM runs on:
37 */
38 u64 hypercall_gpa;
39
40} __attribute__ ((aligned(PAGE_SIZE)));
41
42#define KVM_PARA_API_VERSION 1
43
44/*
45 * This is used for an RDMSR's ECX parameter to probe for a KVM host.
46 * Hopefully no CPU vendor will use up this number. This is placed well
47 * out of way of the typical space occupied by CPU vendors' MSR indices,
48 * and we think (or at least hope) it wont be occupied in the future
49 * either.
50 */
51#define MSR_KVM_API_MAGIC 0x87655678
52 14
53#define KVM_EINVAL 1 15#define KVM_HC_VAPIC_POLL_IRQ 1
54 16
55/* 17/*
56 * Hypercall calling convention: 18 * hypercalls use architecture specific
57 *
58 * Each hypercall may have 0-6 parameters.
59 *
60 * 64-bit hypercall index is in RAX, goes from 0 to __NR_hypercalls-1
61 *
62 * 64-bit parameters 1-6 are in the standard gcc x86_64 calling convention
63 * order: RDI, RSI, RDX, RCX, R8, R9.
64 *
65 * 32-bit index is EBX, parameters are: EAX, ECX, EDX, ESI, EDI, EBP.
66 * (the first 3 are according to the gcc regparm calling convention)
67 *
68 * No registers are clobbered by the hypercall, except that the
69 * return value is in RAX.
70 */ 19 */
71#define __NR_hypercalls 0 20#include <asm/kvm_para.h>
21
22#ifdef __KERNEL__
23static inline int kvm_para_has_feature(unsigned int feature)
24{
25 if (kvm_arch_para_features() & (1UL << feature))
26 return 1;
27 return 0;
28}
29#endif /* __KERNEL__ */
30#endif /* __LINUX_KVM_PARA_H */
72 31
73#endif
diff --git a/include/linux/kvm_types.h b/include/linux/kvm_types.h
new file mode 100644
index 000000000000..1c4e46decb22
--- /dev/null
+++ b/include/linux/kvm_types.h
@@ -0,0 +1,54 @@
1/*
2 * This program is free software; you can redistribute it and/or modify
3 * it under the terms of the GNU General Public License as published by
4 * the Free Software Foundation; either version 2 of the License.
5 *
6 * This program is distributed in the hope that it will be useful,
7 * but WITHOUT ANY WARRANTY; without even the implied warranty of
8 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
9 * GNU General Public License for more details.
10 *
11 * You should have received a copy of the GNU General Public License
12 * along with this program; if not, write to the Free Software
13 * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
14 *
15 */
16
17#ifndef __KVM_TYPES_H__
18#define __KVM_TYPES_H__
19
20#include <asm/types.h>
21
22/*
23 * Address types:
24 *
25 * gva - guest virtual address
26 * gpa - guest physical address
27 * gfn - guest frame number
28 * hva - host virtual address
29 * hpa - host physical address
30 * hfn - host frame number
31 */
32
33typedef unsigned long gva_t;
34typedef u64 gpa_t;
35typedef unsigned long gfn_t;
36
37typedef unsigned long hva_t;
38typedef u64 hpa_t;
39typedef unsigned long hfn_t;
40
41struct kvm_pio_request {
42 unsigned long count;
43 int cur_count;
44 struct page *guest_pages[2];
45 unsigned guest_page_offset;
46 int in;
47 int port;
48 int size;
49 int string;
50 int down;
51 int rep;
52};
53
54#endif /* __KVM_TYPES_H__ */