2 * This program is free software; you can redistribute it and/or modify
3 * it under the terms of the GNU General Public License, version 2, as
4 * published by the Free Software Foundation.
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.
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.
15 * Copyright IBM Corp. 2007
17 * Authors: Hollis Blanchard <hollisb@us.ibm.com>
20 #ifndef __POWERPC_KVM_HOST_H__
21 #define __POWERPC_KVM_HOST_H__
23 #include <linux/mutex.h>
24 #include <linux/hrtimer.h>
25 #include <linux/interrupt.h>
26 #include <linux/types.h>
27 #include <linux/kvm_types.h>
28 #include <linux/threads.h>
29 #include <linux/spinlock.h>
30 #include <linux/kvm_para.h>
31 #include <linux/list.h>
32 #include <linux/atomic.h>
33 #include <asm/kvm_asm.h>
34 #include <asm/processor.h>
36 #include <asm/cacheflush.h>
37 #include <asm/hvcall.h>
39 #define KVM_MAX_VCPUS NR_CPUS
40 #define KVM_MAX_VCORES NR_CPUS
41 #define KVM_USER_MEM_SLOTS 512
43 #include <asm/cputhreads.h>
44 #define KVM_MAX_VCPU_ID (threads_per_subcore * KVM_MAX_VCORES)
46 #define __KVM_HAVE_ARCH_INTC_INITIALIZED
48 #ifdef CONFIG_KVM_MMIO
49 #define KVM_COALESCED_MMIO_PAGE_OFFSET 1
51 #define KVM_HALT_POLL_NS_DEFAULT 10000 /* 10 us */
53 /* These values are internal and can be increased later */
54 #define KVM_NR_IRQCHIPS 1
55 #define KVM_IRQCHIP_NUM_PINS 256
57 /* PPC-specific vcpu->requests bit members */
58 #define KVM_REQ_WATCHDOG 8
59 #define KVM_REQ_EPR_EXIT 9
61 #include <linux/mmu_notifier.h>
63 #define KVM_ARCH_WANT_MMU_NOTIFIER
65 extern int kvm_unmap_hva(struct kvm
*kvm
, unsigned long hva
);
66 extern int kvm_unmap_hva_range(struct kvm
*kvm
,
67 unsigned long start
, unsigned long end
);
68 extern int kvm_age_hva(struct kvm
*kvm
, unsigned long start
, unsigned long end
);
69 extern int kvm_test_age_hva(struct kvm
*kvm
, unsigned long hva
);
70 extern void kvm_set_spte_hva(struct kvm
*kvm
, unsigned long hva
, pte_t pte
);
72 static inline void kvm_arch_mmu_notifier_invalidate_page(struct kvm
*kvm
,
73 unsigned long address
)
77 #define HPTEG_CACHE_NUM (1 << 15)
78 #define HPTEG_HASH_BITS_PTE 13
79 #define HPTEG_HASH_BITS_PTE_LONG 12
80 #define HPTEG_HASH_BITS_VPTE 13
81 #define HPTEG_HASH_BITS_VPTE_LONG 5
82 #define HPTEG_HASH_BITS_VPTE_64K 11
83 #define HPTEG_HASH_NUM_PTE (1 << HPTEG_HASH_BITS_PTE)
84 #define HPTEG_HASH_NUM_PTE_LONG (1 << HPTEG_HASH_BITS_PTE_LONG)
85 #define HPTEG_HASH_NUM_VPTE (1 << HPTEG_HASH_BITS_VPTE)
86 #define HPTEG_HASH_NUM_VPTE_LONG (1 << HPTEG_HASH_BITS_VPTE_LONG)
87 #define HPTEG_HASH_NUM_VPTE_64K (1 << HPTEG_HASH_BITS_VPTE_64K)
89 /* Physical Address Mask - allowed range of real mode RAM access */
90 #define KVM_PAM 0x0fffffffffffffffULL
96 struct kvmppc_vcpu_book3s
;
97 struct kvmppc_book3s_shadow_vcpu
;
100 ulong remote_tlb_flush
;
103 struct kvm_vcpu_stat
{
108 /* Account for special types of light exits: */
109 u64 itlb_real_miss_exits
;
110 u64 itlb_virt_miss_exits
;
111 u64 dtlb_real_miss_exits
;
112 u64 dtlb_virt_miss_exits
;
116 u64 emulated_inst_exits
;
119 u64 halt_poll_success_ns
;
120 u64 halt_poll_fail_ns
;
122 u64 halt_successful_poll
;
123 u64 halt_attempted_poll
;
124 u64 halt_successful_wait
;
125 u64 halt_poll_invalid
;
131 #ifdef CONFIG_PPC_BOOK3S
145 enum kvm_exit_types
{
148 ITLB_REAL_MISS_EXITS
,
149 ITLB_VIRT_MISS_EXITS
,
150 DTLB_REAL_MISS_EXITS
,
151 DTLB_VIRT_MISS_EXITS
,
156 EMULATED_MTMSRWE_EXITS
,
157 EMULATED_WRTEE_EXITS
,
158 EMULATED_MTSPR_EXITS
,
159 EMULATED_MFSPR_EXITS
,
160 EMULATED_MTMSR_EXITS
,
161 EMULATED_MFMSR_EXITS
,
162 EMULATED_TLBSX_EXITS
,
163 EMULATED_TLBWE_EXITS
,
176 __NUMBER_OF_KVM_EXIT_TYPES
179 /* allow access to big endian 32bit upper/lower parts and 64bit var */
180 struct kvmppc_exit_timing
{
189 struct kvmppc_pginfo
{
194 struct kvmppc_spapr_tce_table
{
195 struct list_head list
;
200 u64 offset
; /* in pages */
201 u64 size
; /* window size in pages */
202 struct page
*pages
[0];
205 /* XICS components, defined in book3s_xics.c */
209 struct kvmppc_passthru_irqmap
;
212 * The reverse mapping array has one entry for each HPTE,
213 * which stores the guest's view of the second word of the HPTE
214 * (including the guest physical address of the mapping),
215 * plus forward and backward pointers in a doubly-linked ring
216 * of HPTEs that map the same host page. The pointers in this
217 * ring are 32-bit HPTE indexes, to save space.
219 struct revmap_entry
{
220 unsigned long guest_rpte
;
221 unsigned int forw
, back
;
225 * We use the top bit of each memslot->arch.rmap entry as a lock bit,
226 * and bit 32 as a present flag. The bottom 32 bits are the
227 * index in the guest HPT of a HPTE that points to the page.
229 #define KVMPPC_RMAP_LOCK_BIT 63
230 #define KVMPPC_RMAP_RC_SHIFT 32
231 #define KVMPPC_RMAP_CHG_SHIFT 48
232 #define KVMPPC_RMAP_REFERENCED (HPTE_R_R << KVMPPC_RMAP_RC_SHIFT)
233 #define KVMPPC_RMAP_CHANGED (HPTE_R_C << KVMPPC_RMAP_RC_SHIFT)
234 #define KVMPPC_RMAP_CHG_ORDER (0x3ful << KVMPPC_RMAP_CHG_SHIFT)
235 #define KVMPPC_RMAP_PRESENT 0x100000000ul
236 #define KVMPPC_RMAP_INDEX 0xfffffffful
238 struct kvm_arch_memory_slot
{
239 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
241 #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
244 struct kvm_hpt_info
{
245 /* Host virtual (linear mapping) address of guest HPT */
247 /* Array of reverse mapping entries for each guest HPTE */
248 struct revmap_entry
*rev
;
249 /* Guest HPT size is 2**(order) bytes */
251 /* 1 if HPT allocated with CMA, 0 otherwise */
255 struct kvm_resize_hpt
;
259 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
260 unsigned int tlb_sets
;
261 struct kvm_hpt_info hpt
;
262 atomic64_t mmio_update
;
263 unsigned int host_lpid
;
264 unsigned long host_lpcr
;
266 unsigned long host_sdr1
;
269 unsigned long vrma_slb_v
;
271 atomic_t vcpus_running
;
273 atomic_t hpte_mod_interest
;
274 cpumask_t need_tlb_flush
;
275 cpumask_t cpu_in_guest
;
279 struct dentry
*debugfs_dir
;
280 struct dentry
*htab_dentry
;
281 struct kvm_resize_hpt
*resize_hpt
; /* protected by kvm->lock */
282 #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
283 #ifdef CONFIG_KVM_BOOK3S_PR_POSSIBLE
284 struct mutex hpt_mutex
;
286 #ifdef CONFIG_PPC_BOOK3S_64
287 struct list_head spapr_tce_tables
;
288 struct list_head rtas_tokens
;
289 DECLARE_BITMAP(enabled_hcalls
, MAX_HCALL_OPCODE
/4 + 1);
291 #ifdef CONFIG_KVM_MPIC
292 struct openpic
*mpic
;
294 #ifdef CONFIG_KVM_XICS
295 struct kvmppc_xics
*xics
;
296 struct kvmppc_passthru_irqmap
*pimap
;
298 struct kvmppc_ops
*kvm_ops
;
299 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
300 /* This array can grow quite large, keep it at the end */
301 struct kvmppc_vcore
*vcores
[KVM_MAX_VCORES
];
305 #define VCORE_ENTRY_MAP(vc) ((vc)->entry_exit_map & 0xff)
306 #define VCORE_EXIT_MAP(vc) ((vc)->entry_exit_map >> 8)
307 #define VCORE_IS_EXITING(vc) (VCORE_EXIT_MAP(vc) != 0)
309 /* This bit is used when a vcore exit is triggered from outside the vcore */
310 #define VCORE_EXIT_REQ 0x10000
313 * Values for vcore_state.
314 * Note that these are arranged such that lower values
315 * (< VCORE_SLEEPING) don't require stolen time accounting
316 * on load/unload, and higher values do.
318 #define VCORE_INACTIVE 0
319 #define VCORE_PREEMPT 1
320 #define VCORE_PIGGYBACK 2
321 #define VCORE_SLEEPING 3
322 #define VCORE_RUNNING 4
323 #define VCORE_EXITING 5
324 #define VCORE_POLLING 6
327 * Struct used to manage memory for a virtual processor area
328 * registered by a PAPR guest. There are three types of area
329 * that a guest can register.
332 unsigned long gpa
; /* Current guest phys addr */
333 void *pinned_addr
; /* Address in kernel linear mapping */
334 void *pinned_end
; /* End of region */
335 unsigned long next_gpa
; /* Guest phys addr for update */
336 unsigned long len
; /* Number of bytes required */
337 u8 update_pending
; /* 1 => update pinned_addr from next_gpa */
338 bool dirty
; /* true => area has been modified by kernel */
347 bool may_execute
: 1;
348 u8 page_size
; /* MMU_PAGE_xxx */
353 void (*slbmte
)(struct kvm_vcpu
*vcpu
, u64 rb
, u64 rs
);
354 u64 (*slbmfee
)(struct kvm_vcpu
*vcpu
, u64 slb_nr
);
355 u64 (*slbmfev
)(struct kvm_vcpu
*vcpu
, u64 slb_nr
);
356 void (*slbie
)(struct kvm_vcpu
*vcpu
, u64 slb_nr
);
357 void (*slbia
)(struct kvm_vcpu
*vcpu
);
359 void (*mtsrin
)(struct kvm_vcpu
*vcpu
, u32 srnum
, ulong value
);
360 u32 (*mfsrin
)(struct kvm_vcpu
*vcpu
, u32 srnum
);
361 int (*xlate
)(struct kvm_vcpu
*vcpu
, gva_t eaddr
,
362 struct kvmppc_pte
*pte
, bool data
, bool iswrite
);
363 void (*reset_msr
)(struct kvm_vcpu
*vcpu
);
364 void (*tlbie
)(struct kvm_vcpu
*vcpu
, ulong addr
, bool large
);
365 int (*esid_to_vsid
)(struct kvm_vcpu
*vcpu
, ulong esid
, u64
*vsid
);
366 u64 (*ea_to_vp
)(struct kvm_vcpu
*vcpu
, gva_t eaddr
, bool data
);
367 bool (*is_dcbz32
)(struct kvm_vcpu
*vcpu
);
379 bool large
: 1; /* PTEs are 16MB */
380 bool tb
: 1; /* 1TB segment */
382 u8 base_page_size
; /* MMU_PAGE_xxx */
385 /* Struct used to accumulate timing information in HV real mode code */
386 struct kvmhv_tb_accumulator
{
387 u64 seqcount
; /* used to synchronize access, also count * 2 */
388 u64 tb_total
; /* total time in timebase ticks */
389 u64 tb_min
; /* min time */
390 u64 tb_max
; /* max time */
393 #ifdef CONFIG_PPC_BOOK3S_64
394 struct kvmppc_irq_map
{
397 struct irq_desc
*desc
;
400 #define KVMPPC_PIRQ_MAPPED 1024
401 struct kvmppc_passthru_irqmap
{
403 struct kvmppc_irq_map mapped
[KVMPPC_PIRQ_MAPPED
];
407 # ifdef CONFIG_PPC_FSL_BOOK3E
408 #define KVMPPC_BOOKE_IAC_NUM 2
409 #define KVMPPC_BOOKE_DAC_NUM 2
411 #define KVMPPC_BOOKE_IAC_NUM 4
412 #define KVMPPC_BOOKE_DAC_NUM 2
414 #define KVMPPC_BOOKE_MAX_IAC 4
415 #define KVMPPC_BOOKE_MAX_DAC 2
417 /* KVMPPC_EPR_USER takes precedence over KVMPPC_EPR_KERNEL */
418 #define KVMPPC_EPR_NONE 0 /* EPR not supported */
419 #define KVMPPC_EPR_USER 1 /* exit to userspace to fill EPR */
420 #define KVMPPC_EPR_KERNEL 2 /* in-kernel irqchip */
422 #define KVMPPC_IRQ_DEFAULT 0
423 #define KVMPPC_IRQ_MPIC 1
424 #define KVMPPC_IRQ_XICS 2
426 #define MMIO_HPTE_CACHE_SIZE 4
428 struct mmio_hpte_cache_entry
{
429 unsigned long hpte_v
;
430 unsigned long hpte_r
;
432 unsigned long pte_index
;
436 unsigned int slb_base_pshift
;
439 struct mmio_hpte_cache
{
440 struct mmio_hpte_cache_entry entry
[MMIO_HPTE_CACHE_SIZE
];
446 struct kvm_vcpu_arch
{
449 #ifdef CONFIG_PPC_BOOK3S
450 struct kvmppc_slb slb
[64];
451 int slb_max
; /* 1 + index of last valid entry in slb[] */
452 int slb_nr
; /* total number of entries in SLB */
453 struct kvmppc_mmu mmu
;
454 struct kvmppc_vcpu_book3s
*book3s
;
456 #ifdef CONFIG_PPC_BOOK3S_32
457 struct kvmppc_book3s_shadow_vcpu
*shadow_vcpu
;
462 struct thread_fp_state fp
;
470 #ifdef CONFIG_ALTIVEC
471 struct thread_vr_state vr
;
474 #ifdef CONFIG_KVM_BOOKE_HV
484 #if defined(CONFIG_BOOKE)
485 #if defined(CONFIG_KVM_BOOKE_HV) || defined(CONFIG_64BIT)
490 #ifdef CONFIG_PPC_BOOK3S
491 /* For Gekko paired singles */
498 #ifdef CONFIG_PPC_BOOK3S
505 #ifdef CONFIG_PPC_BOOK3S
507 ulong guest_owned_ext
;
538 u32 vrsave
; /* also USPRG0 */
540 /* shadow_msr is unused for BookE HV */
553 /* Time base value when we entered the guest */
558 ulong tsr
; /* we need to perform set/clr_bits() which requires ulong */
578 #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
594 struct thread_fp_state fp_tm
;
596 struct thread_vr_state vr_tm
;
597 u32 vrsave_tm
; /* also USPRG0 */
601 #ifdef CONFIG_KVM_EXIT_TIMING
602 struct mutex exit_timing_lock
;
603 struct kvmppc_exit_timing timing_exit
;
604 struct kvmppc_exit_timing timing_last_enter
;
606 u32 timing_count_type
[__NUMBER_OF_KVM_EXIT_TYPES
];
607 u64 timing_sum_duration
[__NUMBER_OF_KVM_EXIT_TYPES
];
608 u64 timing_sum_quad_duration
[__NUMBER_OF_KVM_EXIT_TYPES
];
609 u64 timing_min_duration
[__NUMBER_OF_KVM_EXIT_TYPES
];
610 u64 timing_max_duration
[__NUMBER_OF_KVM_EXIT_TYPES
];
611 u64 timing_last_exit
;
612 struct dentry
*debugfs_exit_timing
;
615 #ifdef CONFIG_PPC_BOOK3S
618 unsigned long intr_msr
;
619 ulong fault_gpa
; /* guest real address of page fault (POWER9) */
628 struct timer_list wdt_timer
;
637 /* guest debug registers*/
638 struct debug_reg dbg_reg
;
640 gpa_t paddr_accessed
;
641 gva_t vaddr_accessed
;
644 u8 io_gpr
; /* GPR used as IO source/target */
645 u8 mmio_host_swabbed
;
654 u8 epr_flags
; /* KVMPPC_EPR_xxx */
657 u32 cpr0_cfgaddr
; /* holds the last set cpr0_cfgaddr */
659 struct hrtimer dec_timer
;
662 unsigned long pending_exceptions
;
667 struct swait_queue_head
*wqp
;
668 struct kvmppc_vcore
*vcore
;
676 wait_queue_head_t cpu_run
;
678 struct kvm_vcpu_arch_shared
*shared
;
679 #if defined(CONFIG_PPC_BOOK3S_64) && defined(CONFIG_KVM_BOOK3S_PR_POSSIBLE)
680 bool shared_big_endian
;
682 unsigned long magic_page_pa
; /* phys addr to map the magic page to */
683 unsigned long magic_page_ea
; /* effect. addr to map the magic page to */
684 bool disable_kernel_nx
;
686 int irq_type
; /* one of KVM_IRQ_* */
688 struct openpic
*mpic
; /* KVM_IRQ_MPIC */
689 #ifdef CONFIG_KVM_XICS
690 struct kvmppc_icp
*icp
; /* XICS presentation controller */
693 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
694 struct kvm_vcpu_arch_shared shregs
;
696 struct mmio_hpte_cache mmio_cache
;
697 unsigned long pgfault_addr
;
699 unsigned long pgfault_hpte
[2];
700 struct mmio_hpte_cache_entry
*pgfault_cache
;
702 struct task_struct
*run_task
;
703 struct kvm_run
*kvm_run
;
705 spinlock_t vpa_update_lock
;
706 struct kvmppc_vpa vpa
;
707 struct kvmppc_vpa dtl
;
708 struct dtl_entry
*dtl_ptr
;
709 unsigned long dtl_index
;
711 struct kvmppc_vpa slb_shadow
;
713 spinlock_t tbacct_lock
;
720 #ifdef CONFIG_KVM_BOOK3S_HV_EXIT_TIMING
721 struct kvmhv_tb_accumulator
*cur_activity
; /* What we're timing */
722 u64 cur_tb_start
; /* when it started */
723 struct kvmhv_tb_accumulator rm_entry
; /* real-mode entry code */
724 struct kvmhv_tb_accumulator rm_intr
; /* real-mode intr handling */
725 struct kvmhv_tb_accumulator rm_exit
; /* real-mode exit code */
726 struct kvmhv_tb_accumulator guest_time
; /* guest execution */
727 struct kvmhv_tb_accumulator cede_time
; /* time napping inside guest */
729 struct dentry
*debugfs_dir
;
730 struct dentry
*debugfs_timings
;
731 #endif /* CONFIG_KVM_BOOK3S_HV_EXIT_TIMING */
734 #define VCPU_FPR(vcpu, i) (vcpu)->arch.fp.fpr[i][TS_FPROFFSET]
736 /* Values for vcpu->arch.state */
737 #define KVMPPC_VCPU_NOTREADY 0
738 #define KVMPPC_VCPU_RUNNABLE 1
739 #define KVMPPC_VCPU_BUSY_IN_HOST 2
741 /* Values for vcpu->arch.io_gpr */
742 #define KVM_MMIO_REG_MASK 0x001f
743 #define KVM_MMIO_REG_EXT_MASK 0xffe0
744 #define KVM_MMIO_REG_GPR 0x0000
745 #define KVM_MMIO_REG_FPR 0x0020
746 #define KVM_MMIO_REG_QPR 0x0040
747 #define KVM_MMIO_REG_FQPR 0x0060
749 #define __KVM_HAVE_ARCH_WQP
750 #define __KVM_HAVE_CREATE_DEVICE
752 static inline void kvm_arch_hardware_disable(void) {}
753 static inline void kvm_arch_hardware_unsetup(void) {}
754 static inline void kvm_arch_sync_events(struct kvm
*kvm
) {}
755 static inline void kvm_arch_memslots_updated(struct kvm
*kvm
, struct kvm_memslots
*slots
) {}
756 static inline void kvm_arch_flush_shadow_all(struct kvm
*kvm
) {}
757 static inline void kvm_arch_sched_in(struct kvm_vcpu
*vcpu
, int cpu
) {}
758 static inline void kvm_arch_exit(void) {}
759 static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu
*vcpu
) {}
760 static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu
*vcpu
) {}
761 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu
*vcpu
) {}
763 #endif /* __POWERPC_KVM_HOST_H__ */