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 32
42 #define KVM_MEM_SLOTS_NUM KVM_USER_MEM_SLOTS
44 #ifdef CONFIG_KVM_MMIO
45 #define KVM_COALESCED_MMIO_PAGE_OFFSET 1
48 /* These values are internal and can be increased later */
49 #define KVM_NR_IRQCHIPS 1
50 #define KVM_IRQCHIP_NUM_PINS 256
52 #include <linux/mmu_notifier.h>
54 #define KVM_ARCH_WANT_MMU_NOTIFIER
56 extern int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
57 extern int kvm_unmap_hva_range(struct kvm *kvm,
58 unsigned long start, unsigned long end);
59 extern int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
60 extern int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
61 extern void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
63 static inline void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
64 unsigned long address)
68 #define HPTEG_CACHE_NUM (1 << 15)
69 #define HPTEG_HASH_BITS_PTE 13
70 #define HPTEG_HASH_BITS_PTE_LONG 12
71 #define HPTEG_HASH_BITS_VPTE 13
72 #define HPTEG_HASH_BITS_VPTE_LONG 5
73 #define HPTEG_HASH_BITS_VPTE_64K 11
74 #define HPTEG_HASH_NUM_PTE (1 << HPTEG_HASH_BITS_PTE)
75 #define HPTEG_HASH_NUM_PTE_LONG (1 << HPTEG_HASH_BITS_PTE_LONG)
76 #define HPTEG_HASH_NUM_VPTE (1 << HPTEG_HASH_BITS_VPTE)
77 #define HPTEG_HASH_NUM_VPTE_LONG (1 << HPTEG_HASH_BITS_VPTE_LONG)
78 #define HPTEG_HASH_NUM_VPTE_64K (1 << HPTEG_HASH_BITS_VPTE_64K)
80 /* Physical Address Mask - allowed range of real mode RAM access */
81 #define KVM_PAM 0x0fffffffffffffffULL
87 struct kvmppc_vcpu_book3s;
88 struct kvmppc_book3s_shadow_vcpu;
94 struct kvm_vcpu_stat {
99 /* Account for special types of light exits: */
100 u32 itlb_real_miss_exits;
101 u32 itlb_virt_miss_exits;
102 u32 dtlb_real_miss_exits;
103 u32 dtlb_virt_miss_exits;
107 u32 emulated_inst_exits;
115 #ifdef CONFIG_PPC_BOOK3S
126 enum kvm_exit_types {
129 ITLB_REAL_MISS_EXITS,
130 ITLB_VIRT_MISS_EXITS,
131 DTLB_REAL_MISS_EXITS,
132 DTLB_VIRT_MISS_EXITS,
137 EMULATED_MTMSRWE_EXITS,
138 EMULATED_WRTEE_EXITS,
139 EMULATED_MTSPR_EXITS,
140 EMULATED_MFSPR_EXITS,
141 EMULATED_MTMSR_EXITS,
142 EMULATED_MFMSR_EXITS,
143 EMULATED_TLBSX_EXITS,
144 EMULATED_TLBWE_EXITS,
157 __NUMBER_OF_KVM_EXIT_TYPES
160 /* allow access to big endian 32bit upper/lower parts and 64bit var */
161 struct kvmppc_exit_timing {
170 struct kvmppc_pginfo {
175 struct kvmppc_spapr_tce_table {
176 struct list_head list;
180 struct page *pages[0];
183 struct kvm_rma_info {
185 unsigned long base_pfn;
188 /* XICS components, defined in book3s_xics.c */
193 * The reverse mapping array has one entry for each HPTE,
194 * which stores the guest's view of the second word of the HPTE
195 * (including the guest physical address of the mapping),
196 * plus forward and backward pointers in a doubly-linked ring
197 * of HPTEs that map the same host page. The pointers in this
198 * ring are 32-bit HPTE indexes, to save space.
200 struct revmap_entry {
201 unsigned long guest_rpte;
202 unsigned int forw, back;
206 * We use the top bit of each memslot->arch.rmap entry as a lock bit,
207 * and bit 32 as a present flag. The bottom 32 bits are the
208 * index in the guest HPT of a HPTE that points to the page.
210 #define KVMPPC_RMAP_LOCK_BIT 63
211 #define KVMPPC_RMAP_RC_SHIFT 32
212 #define KVMPPC_RMAP_REFERENCED (HPTE_R_R << KVMPPC_RMAP_RC_SHIFT)
213 #define KVMPPC_RMAP_CHANGED (HPTE_R_C << KVMPPC_RMAP_RC_SHIFT)
214 #define KVMPPC_RMAP_PRESENT 0x100000000ul
215 #define KVMPPC_RMAP_INDEX 0xfffffffful
217 /* Low-order bits in memslot->arch.slot_phys[] */
218 #define KVMPPC_PAGE_ORDER_MASK 0x1f
219 #define KVMPPC_PAGE_NO_CACHE HPTE_R_I /* 0x20 */
220 #define KVMPPC_PAGE_WRITETHRU HPTE_R_W /* 0x40 */
221 #define KVMPPC_GOT_PAGE 0x80
223 struct kvm_arch_memory_slot {
224 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
226 unsigned long *slot_phys;
227 #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
232 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
233 unsigned long hpt_virt;
234 struct revmap_entry *revmap;
235 unsigned int host_lpid;
236 unsigned long host_lpcr;
238 unsigned long host_sdr1;
242 struct kvm_rma_info *rma;
243 unsigned long vrma_slb_v;
245 int using_mmu_notifiers;
247 atomic_t vcpus_running;
249 unsigned long hpt_npte;
250 unsigned long hpt_mask;
251 atomic_t hpte_mod_interest;
252 spinlock_t slot_phys_lock;
253 cpumask_t need_tlb_flush;
255 #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
256 #ifdef CONFIG_KVM_BOOK3S_PR_POSSIBLE
257 struct mutex hpt_mutex;
259 #ifdef CONFIG_PPC_BOOK3S_64
260 struct list_head spapr_tce_tables;
261 struct list_head rtas_tokens;
262 DECLARE_BITMAP(enabled_hcalls, MAX_HCALL_OPCODE/4 + 1);
264 #ifdef CONFIG_KVM_MPIC
265 struct openpic *mpic;
267 #ifdef CONFIG_KVM_XICS
268 struct kvmppc_xics *xics;
270 struct kvmppc_ops *kvm_ops;
271 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
272 /* This array can grow quite large, keep it at the end */
273 struct kvmppc_vcore *vcores[KVM_MAX_VCORES];
278 * Struct for a virtual core.
279 * Note: entry_exit_count combines an entry count in the bottom 8 bits
280 * and an exit count in the next 8 bits. This is so that we can
281 * atomically increment the entry count iff the exit count is 0
282 * without taking the lock.
284 struct kvmppc_vcore {
288 int entry_exit_count;
297 struct list_head runnable_threads;
299 wait_queue_head_t wq;
302 struct kvm_vcpu *runner;
304 u64 tb_offset; /* guest timebase - host timebase */
308 ulong dpdes; /* doorbell state (POWER8) */
309 void *mpp_buffer; /* Micro Partition Prefetch buffer */
310 bool mpp_buffer_is_valid;
313 #define VCORE_ENTRY_COUNT(vc) ((vc)->entry_exit_count & 0xff)
314 #define VCORE_EXIT_COUNT(vc) ((vc)->entry_exit_count >> 8)
316 /* Values for vcore_state */
317 #define VCORE_INACTIVE 0
318 #define VCORE_SLEEPING 1
319 #define VCORE_STARTING 2
320 #define VCORE_RUNNING 3
321 #define VCORE_EXITING 4
324 * Struct used to manage memory for a virtual processor area
325 * registered by a PAPR guest. There are three types of area
326 * that a guest can register.
329 unsigned long gpa; /* Current guest phys addr */
330 void *pinned_addr; /* Address in kernel linear mapping */
331 void *pinned_end; /* End of region */
332 unsigned long next_gpa; /* Guest phys addr for update */
333 unsigned long len; /* Number of bytes required */
334 u8 update_pending; /* 1 => update pinned_addr from next_gpa */
335 bool dirty; /* true => area has been modified by kernel */
344 bool may_execute : 1;
345 u8 page_size; /* MMU_PAGE_xxx */
350 void (*slbmte)(struct kvm_vcpu *vcpu, u64 rb, u64 rs);
351 u64 (*slbmfee)(struct kvm_vcpu *vcpu, u64 slb_nr);
352 u64 (*slbmfev)(struct kvm_vcpu *vcpu, u64 slb_nr);
353 void (*slbie)(struct kvm_vcpu *vcpu, u64 slb_nr);
354 void (*slbia)(struct kvm_vcpu *vcpu);
356 void (*mtsrin)(struct kvm_vcpu *vcpu, u32 srnum, ulong value);
357 u32 (*mfsrin)(struct kvm_vcpu *vcpu, u32 srnum);
358 int (*xlate)(struct kvm_vcpu *vcpu, gva_t eaddr,
359 struct kvmppc_pte *pte, bool data, bool iswrite);
360 void (*reset_msr)(struct kvm_vcpu *vcpu);
361 void (*tlbie)(struct kvm_vcpu *vcpu, ulong addr, bool large);
362 int (*esid_to_vsid)(struct kvm_vcpu *vcpu, ulong esid, u64 *vsid);
363 u64 (*ea_to_vp)(struct kvm_vcpu *vcpu, gva_t eaddr, bool data);
364 bool (*is_dcbz32)(struct kvm_vcpu *vcpu);
376 bool large : 1; /* PTEs are 16MB */
377 bool tb : 1; /* 1TB segment */
379 u8 base_page_size; /* MMU_PAGE_xxx */
382 # ifdef CONFIG_PPC_FSL_BOOK3E
383 #define KVMPPC_BOOKE_IAC_NUM 2
384 #define KVMPPC_BOOKE_DAC_NUM 2
386 #define KVMPPC_BOOKE_IAC_NUM 4
387 #define KVMPPC_BOOKE_DAC_NUM 2
389 #define KVMPPC_BOOKE_MAX_IAC 4
390 #define KVMPPC_BOOKE_MAX_DAC 2
392 /* KVMPPC_EPR_USER takes precedence over KVMPPC_EPR_KERNEL */
393 #define KVMPPC_EPR_NONE 0 /* EPR not supported */
394 #define KVMPPC_EPR_USER 1 /* exit to userspace to fill EPR */
395 #define KVMPPC_EPR_KERNEL 2 /* in-kernel irqchip */
397 #define KVMPPC_IRQ_DEFAULT 0
398 #define KVMPPC_IRQ_MPIC 1
399 #define KVMPPC_IRQ_XICS 2
403 struct kvm_vcpu_arch {
406 #ifdef CONFIG_PPC_BOOK3S
407 struct kvmppc_slb slb[64];
408 int slb_max; /* 1 + index of last valid entry in slb[] */
409 int slb_nr; /* total number of entries in SLB */
410 struct kvmppc_mmu mmu;
411 struct kvmppc_vcpu_book3s *book3s;
413 #ifdef CONFIG_PPC_BOOK3S_32
414 struct kvmppc_book3s_shadow_vcpu *shadow_vcpu;
419 struct thread_fp_state fp;
427 #ifdef CONFIG_ALTIVEC
428 struct thread_vr_state vr;
431 #ifdef CONFIG_KVM_BOOKE_HV
441 #if defined(CONFIG_BOOKE)
442 #if defined(CONFIG_KVM_BOOKE_HV) || defined(CONFIG_64BIT)
447 #ifdef CONFIG_PPC_BOOK3S
448 /* For Gekko paired singles */
455 #ifdef CONFIG_PPC_BOOK3S
462 #ifdef CONFIG_PPC_BOOK3S
464 ulong guest_owned_ext;
494 u32 vrsave; /* also USPRG0 */
496 /* shadow_msr is unused for BookE HV */
509 /* Time base value when we entered the guest */
514 ulong tsr; /* we need to perform set/clr_bits() which requires ulong */
534 #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
549 struct thread_fp_state fp_tm;
551 struct thread_vr_state vr_tm;
552 u32 vrsave_tm; /* also USPRG0 */
556 #ifdef CONFIG_KVM_EXIT_TIMING
557 struct mutex exit_timing_lock;
558 struct kvmppc_exit_timing timing_exit;
559 struct kvmppc_exit_timing timing_last_enter;
561 u32 timing_count_type[__NUMBER_OF_KVM_EXIT_TYPES];
562 u64 timing_sum_duration[__NUMBER_OF_KVM_EXIT_TYPES];
563 u64 timing_sum_quad_duration[__NUMBER_OF_KVM_EXIT_TYPES];
564 u64 timing_min_duration[__NUMBER_OF_KVM_EXIT_TYPES];
565 u64 timing_max_duration[__NUMBER_OF_KVM_EXIT_TYPES];
566 u64 timing_last_exit;
567 struct dentry *debugfs_exit_timing;
570 #ifdef CONFIG_PPC_BOOK3S
573 unsigned long intr_msr;
582 struct timer_list wdt_timer;
591 /* guest debug registers*/
592 struct debug_reg dbg_reg;
594 gpa_t paddr_accessed;
595 gva_t vaddr_accessed;
598 u8 io_gpr; /* GPR used as IO source/target */
599 u8 mmio_is_bigendian;
608 u8 epr_flags; /* KVMPPC_EPR_xxx */
611 u32 cpr0_cfgaddr; /* holds the last set cpr0_cfgaddr */
613 struct hrtimer dec_timer;
616 unsigned long pending_exceptions;
621 wait_queue_head_t *wqp;
622 struct kvmppc_vcore *vcore;
628 wait_queue_head_t cpu_run;
630 struct kvm_vcpu_arch_shared *shared;
631 #if defined(CONFIG_PPC_BOOK3S_64) && defined(CONFIG_KVM_BOOK3S_PR_POSSIBLE)
632 bool shared_big_endian;
634 unsigned long magic_page_pa; /* phys addr to map the magic page to */
635 unsigned long magic_page_ea; /* effect. addr to map the magic page to */
636 bool disable_kernel_nx;
638 int irq_type; /* one of KVM_IRQ_* */
640 struct openpic *mpic; /* KVM_IRQ_MPIC */
641 #ifdef CONFIG_KVM_XICS
642 struct kvmppc_icp *icp; /* XICS presentation controller */
645 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
646 struct kvm_vcpu_arch_shared shregs;
648 unsigned long pgfault_addr;
650 unsigned long pgfault_hpte[2];
652 struct list_head run_list;
653 struct task_struct *run_task;
654 struct kvm_run *kvm_run;
656 spinlock_t vpa_update_lock;
657 struct kvmppc_vpa vpa;
658 struct kvmppc_vpa dtl;
659 struct dtl_entry *dtl_ptr;
660 unsigned long dtl_index;
662 struct kvmppc_vpa slb_shadow;
664 spinlock_t tbacct_lock;
670 #define VCPU_FPR(vcpu, i) (vcpu)->arch.fp.fpr[i][TS_FPROFFSET]
672 /* Values for vcpu->arch.state */
673 #define KVMPPC_VCPU_NOTREADY 0
674 #define KVMPPC_VCPU_RUNNABLE 1
675 #define KVMPPC_VCPU_BUSY_IN_HOST 2
677 /* Values for vcpu->arch.io_gpr */
678 #define KVM_MMIO_REG_MASK 0x001f
679 #define KVM_MMIO_REG_EXT_MASK 0xffe0
680 #define KVM_MMIO_REG_GPR 0x0000
681 #define KVM_MMIO_REG_FPR 0x0020
682 #define KVM_MMIO_REG_QPR 0x0040
683 #define KVM_MMIO_REG_FQPR 0x0060
685 #define __KVM_HAVE_ARCH_WQP
686 #define __KVM_HAVE_CREATE_DEVICE
688 static inline void kvm_arch_hardware_disable(void) {}
689 static inline void kvm_arch_hardware_unsetup(void) {}
690 static inline void kvm_arch_sync_events(struct kvm *kvm) {}
691 static inline void kvm_arch_memslots_updated(struct kvm *kvm) {}
692 static inline void kvm_arch_flush_shadow_all(struct kvm *kvm) {}
693 static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
694 static inline void kvm_arch_exit(void) {}
696 #endif /* __POWERPC_KVM_HOST_H__ */