1 /*- 2 * SPDX-License-Identifier: (BSD-3-Clause AND MIT-CMU) 3 * 4 * Copyright (c) 1991, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * The Mach Operating System project at Carnegie-Mellon University. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 * 34 * from: @(#)vm_object.h 8.3 (Berkeley) 1/12/94 35 * 36 * 37 * Copyright (c) 1987, 1990 Carnegie-Mellon University. 38 * All rights reserved. 39 * 40 * Authors: Avadis Tevanian, Jr., Michael Wayne Young 41 * 42 * Permission to use, copy, modify and distribute this software and 43 * its documentation is hereby granted, provided that both the copyright 44 * notice and this permission notice appear in all copies of the 45 * software, derivative works or modified versions, and any portions 46 * thereof, and that both notices appear in supporting documentation. 47 * 48 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" 49 * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND 50 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. 51 * 52 * Carnegie Mellon requests users of this software to return to 53 * 54 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU 55 * School of Computer Science 56 * Carnegie Mellon University 57 * Pittsburgh PA 15213-3890 58 * 59 * any improvements or extensions that they make and grant Carnegie the 60 * rights to redistribute these changes. 61 * 62 * $FreeBSD$ 63 */ 64 65 /* 66 * Virtual memory object module definitions. 67 */ 68 69 #ifndef _VM_OBJECT_ 70 #define _VM_OBJECT_ 71 72 #include <sys/queue.h> 73 #include <sys/_lock.h> 74 #include <sys/_mutex.h> 75 #include <sys/_pctrie.h> 76 #include <sys/_rwlock.h> 77 #include <sys/_domainset.h> 78 79 #include <vm/_vm_radix.h> 80 81 /* 82 * Types defined: 83 * 84 * vm_object_t Virtual memory object. 85 * 86 * List of locks 87 * (a) atomic 88 * (c) const until freed 89 * (o) per-object lock 90 * (f) free pages queue mutex 91 * 92 */ 93 94 #ifndef VM_PAGE_HAVE_PGLIST 95 TAILQ_HEAD(pglist, vm_page); 96 #define VM_PAGE_HAVE_PGLIST 97 #endif 98 99 struct vm_object { 100 struct rwlock lock; 101 TAILQ_ENTRY(vm_object) object_list; /* list of all objects */ 102 LIST_HEAD(, vm_object) shadow_head; /* objects that this is a shadow for */ 103 LIST_ENTRY(vm_object) shadow_list; /* chain of shadow objects */ 104 struct pglist memq; /* list of resident pages */ 105 struct vm_radix rtree; /* root of the resident page radix trie*/ 106 vm_pindex_t size; /* Object size */ 107 struct domainset_ref domain; /* NUMA policy. */ 108 volatile int generation; /* generation ID */ 109 int cleangeneration; /* Generation at clean time */ 110 volatile u_int ref_count; /* How many refs?? */ 111 int shadow_count; /* how many objects that this is a shadow for */ 112 vm_memattr_t memattr; /* default memory attribute for pages */ 113 objtype_t type; /* type of pager */ 114 u_short flags; /* see below */ 115 u_short pg_color; /* (c) color of first page in obj */ 116 volatile u_int paging_in_progress; /* Paging (in or out) so don't collapse or destroy */ 117 volatile u_int busy; /* (a) object is busy, disallow page busy. */ 118 int resident_page_count; /* number of resident pages */ 119 struct vm_object *backing_object; /* object that I'm a shadow of */ 120 vm_ooffset_t backing_object_offset;/* Offset in backing object */ 121 TAILQ_ENTRY(vm_object) pager_object_list; /* list of all objects of this pager type */ 122 LIST_HEAD(, vm_reserv) rvq; /* list of reservations */ 123 void *handle; 124 union { 125 /* 126 * VNode pager 127 * 128 * vnp_size - current size of file 129 */ 130 struct { 131 off_t vnp_size; 132 vm_ooffset_t writemappings; 133 } vnp; 134 135 /* 136 * Device pager 137 * 138 * devp_pglist - list of allocated pages 139 */ 140 struct { 141 TAILQ_HEAD(, vm_page) devp_pglist; 142 struct cdev_pager_ops *ops; 143 struct cdev *dev; 144 } devp; 145 146 /* 147 * SG pager 148 * 149 * sgp_pglist - list of allocated pages 150 */ 151 struct { 152 TAILQ_HEAD(, vm_page) sgp_pglist; 153 } sgp; 154 155 /* 156 * Swap pager 157 * 158 * swp_tmpfs - back-pointer to the tmpfs vnode, 159 * if any, which uses the vm object 160 * as backing store. The handle 161 * cannot be reused for linking, 162 * because the vnode can be 163 * reclaimed and recreated, making 164 * the handle changed and hash-chain 165 * invalid. 166 * 167 * swp_blks - pc-trie of the allocated swap blocks. 168 * 169 */ 170 struct { 171 void *swp_tmpfs; 172 struct pctrie swp_blks; 173 vm_ooffset_t writemappings; 174 } swp; 175 } un_pager; 176 struct ucred *cred; 177 vm_ooffset_t charge; 178 void *umtx_data; 179 }; 180 181 /* 182 * Flags 183 */ 184 #define OBJ_FICTITIOUS 0x0001 /* (c) contains fictitious pages */ 185 #define OBJ_UNMANAGED 0x0002 /* (c) contains unmanaged pages */ 186 #define OBJ_POPULATE 0x0004 /* pager implements populate() */ 187 #define OBJ_DEAD 0x0008 /* dead objects (during rundown) */ 188 #define OBJ_ANON 0x0010 /* (c) contains anonymous memory */ 189 #define OBJ_UMTXDEAD 0x0020 /* umtx pshared was terminated */ 190 #define OBJ_SIZEVNLOCK 0x0040 /* lock vnode to check obj size */ 191 #define OBJ_PG_DTOR 0x0080 /* dont reset object, leave that for dtor */ 192 #define OBJ_TMPFS_NODE 0x0200 /* object belongs to tmpfs VREG node */ 193 #define OBJ_COLORED 0x1000 /* pg_color is defined */ 194 #define OBJ_ONEMAPPING 0x2000 /* One USE (a single, non-forked) mapping flag */ 195 #define OBJ_SHADOWLIST 0x4000 /* Object is on the shadow list. */ 196 #define OBJ_TMPFS 0x8000 /* has tmpfs vnode allocated */ 197 198 /* 199 * Helpers to perform conversion between vm_object page indexes and offsets. 200 * IDX_TO_OFF() converts an index into an offset. 201 * OFF_TO_IDX() converts an offset into an index. 202 * OBJ_MAX_SIZE specifies the maximum page index corresponding to the 203 * maximum unsigned offset. 204 */ 205 #define IDX_TO_OFF(idx) (((vm_ooffset_t)(idx)) << PAGE_SHIFT) 206 #define OFF_TO_IDX(off) ((vm_pindex_t)(((vm_ooffset_t)(off)) >> PAGE_SHIFT)) 207 #define OBJ_MAX_SIZE (OFF_TO_IDX(UINT64_MAX) + 1) 208 209 #ifdef _KERNEL 210 211 #define OBJPC_SYNC 0x1 /* sync I/O */ 212 #define OBJPC_INVAL 0x2 /* invalidate */ 213 #define OBJPC_NOSYNC 0x4 /* skip if PGA_NOSYNC */ 214 215 /* 216 * The following options are supported by vm_object_page_remove(). 217 */ 218 #define OBJPR_CLEANONLY 0x1 /* Don't remove dirty pages. */ 219 #define OBJPR_NOTMAPPED 0x2 /* Don't unmap pages. */ 220 221 TAILQ_HEAD(object_q, vm_object); 222 223 extern struct object_q vm_object_list; /* list of allocated objects */ 224 extern struct mtx vm_object_list_mtx; /* lock for object list and count */ 225 226 extern struct vm_object kernel_object_store; 227 228 /* kernel and kmem are aliased for backwards KPI compat. */ 229 #define kernel_object (&kernel_object_store) 230 #define kmem_object (&kernel_object_store) 231 232 #define VM_OBJECT_ASSERT_LOCKED(object) \ 233 rw_assert(&(object)->lock, RA_LOCKED) 234 #define VM_OBJECT_ASSERT_RLOCKED(object) \ 235 rw_assert(&(object)->lock, RA_RLOCKED) 236 #define VM_OBJECT_ASSERT_WLOCKED(object) \ 237 rw_assert(&(object)->lock, RA_WLOCKED) 238 #define VM_OBJECT_ASSERT_UNLOCKED(object) \ 239 rw_assert(&(object)->lock, RA_UNLOCKED) 240 #define VM_OBJECT_LOCK_DOWNGRADE(object) \ 241 rw_downgrade(&(object)->lock) 242 #define VM_OBJECT_RLOCK(object) \ 243 rw_rlock(&(object)->lock) 244 #define VM_OBJECT_RUNLOCK(object) \ 245 rw_runlock(&(object)->lock) 246 #define VM_OBJECT_SLEEP(object, wchan, pri, wmesg, timo) \ 247 rw_sleep((wchan), &(object)->lock, (pri), (wmesg), (timo)) 248 #define VM_OBJECT_TRYRLOCK(object) \ 249 rw_try_rlock(&(object)->lock) 250 #define VM_OBJECT_TRYWLOCK(object) \ 251 rw_try_wlock(&(object)->lock) 252 #define VM_OBJECT_TRYUPGRADE(object) \ 253 rw_try_upgrade(&(object)->lock) 254 #define VM_OBJECT_WLOCK(object) \ 255 rw_wlock(&(object)->lock) 256 #define VM_OBJECT_WOWNED(object) \ 257 rw_wowned(&(object)->lock) 258 #define VM_OBJECT_WUNLOCK(object) \ 259 rw_wunlock(&(object)->lock) 260 #define VM_OBJECT_DROP(object) \ 261 lock_class_rw.lc_unlock(&(object)->lock.lock_object) 262 #define VM_OBJECT_PICKUP(object, state) \ 263 lock_class_rw.lc_lock(&(object)->lock.lock_object, (state)) 264 265 struct vnode; 266 267 /* 268 * The object must be locked or thread private. 269 */ 270 static __inline void 271 vm_object_set_flag(vm_object_t object, u_short bits) 272 { 273 274 object->flags |= bits; 275 } 276 277 /* 278 * Conditionally set the object's color, which (1) enables the allocation 279 * of physical memory reservations for anonymous objects and larger-than- 280 * superpage-sized named objects and (2) determines the first page offset 281 * within the object at which a reservation may be allocated. In other 282 * words, the color determines the alignment of the object with respect 283 * to the largest superpage boundary. When mapping named objects, like 284 * files or POSIX shared memory objects, the color should be set to zero 285 * before a virtual address is selected for the mapping. In contrast, 286 * for anonymous objects, the color may be set after the virtual address 287 * is selected. 288 * 289 * The object must be locked. 290 */ 291 static __inline void 292 vm_object_color(vm_object_t object, u_short color) 293 { 294 295 if ((object->flags & OBJ_COLORED) == 0) { 296 object->pg_color = color; 297 object->flags |= OBJ_COLORED; 298 } 299 } 300 301 static __inline bool 302 vm_object_reserv(vm_object_t object) 303 { 304 305 if (object != NULL && 306 (object->flags & (OBJ_COLORED | OBJ_FICTITIOUS)) == OBJ_COLORED) { 307 return (true); 308 } 309 return (false); 310 } 311 312 static __inline bool 313 vm_object_mightbedirty(vm_object_t object) 314 { 315 316 return (object->type == OBJT_VNODE && 317 object->generation != object->cleangeneration); 318 } 319 320 void vm_object_clear_flag(vm_object_t object, u_short bits); 321 void vm_object_pip_add(vm_object_t object, short i); 322 void vm_object_pip_wakeup(vm_object_t object); 323 void vm_object_pip_wakeupn(vm_object_t object, short i); 324 void vm_object_pip_wait(vm_object_t object, char *waitid); 325 void vm_object_pip_wait_unlocked(vm_object_t object, char *waitid); 326 327 void vm_object_busy(vm_object_t object); 328 void vm_object_unbusy(vm_object_t object); 329 void vm_object_busy_wait(vm_object_t object, const char *wmesg); 330 331 static inline bool 332 vm_object_busied(vm_object_t object) 333 { 334 335 return (object->busy != 0); 336 } 337 #define VM_OBJECT_ASSERT_BUSY(object) MPASS(vm_object_busied((object))) 338 339 void umtx_shm_object_init(vm_object_t object); 340 void umtx_shm_object_terminated(vm_object_t object); 341 extern int umtx_shm_vnobj_persistent; 342 343 vm_object_t vm_object_allocate (objtype_t, vm_pindex_t); 344 vm_object_t vm_object_allocate_anon(vm_pindex_t, vm_object_t, struct ucred *, 345 vm_size_t); 346 boolean_t vm_object_coalesce(vm_object_t, vm_ooffset_t, vm_size_t, vm_size_t, 347 boolean_t); 348 void vm_object_collapse (vm_object_t); 349 void vm_object_deallocate (vm_object_t); 350 void vm_object_destroy (vm_object_t); 351 void vm_object_terminate (vm_object_t); 352 void vm_object_set_writeable_dirty (vm_object_t); 353 void vm_object_init (void); 354 int vm_object_kvme_type(vm_object_t object, struct vnode **vpp); 355 void vm_object_madvise(vm_object_t, vm_pindex_t, vm_pindex_t, int); 356 boolean_t vm_object_page_clean(vm_object_t object, vm_ooffset_t start, 357 vm_ooffset_t end, int flags); 358 void vm_object_page_noreuse(vm_object_t object, vm_pindex_t start, 359 vm_pindex_t end); 360 void vm_object_page_remove(vm_object_t object, vm_pindex_t start, 361 vm_pindex_t end, int options); 362 boolean_t vm_object_populate(vm_object_t, vm_pindex_t, vm_pindex_t); 363 void vm_object_print(long addr, boolean_t have_addr, long count, char *modif); 364 void vm_object_reference (vm_object_t); 365 void vm_object_reference_locked(vm_object_t); 366 int vm_object_set_memattr(vm_object_t object, vm_memattr_t memattr); 367 void vm_object_shadow(vm_object_t *, vm_ooffset_t *, vm_size_t, struct ucred *, 368 bool); 369 void vm_object_split(vm_map_entry_t); 370 boolean_t vm_object_sync(vm_object_t, vm_ooffset_t, vm_size_t, boolean_t, 371 boolean_t); 372 void vm_object_unwire(vm_object_t object, vm_ooffset_t offset, 373 vm_size_t length, uint8_t queue); 374 struct vnode *vm_object_vnode(vm_object_t object); 375 #endif /* _KERNEL */ 376 377 #endif /* _VM_OBJECT_ */ 378