1 /*- 2 * Copyright (c) 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * The Mach Operating System project at Carnegie-Mellon University. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 4. Neither the name of the University nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 * 32 * from: @(#)vm_page.h 8.2 (Berkeley) 12/13/93 33 * 34 * 35 * Copyright (c) 1987, 1990 Carnegie-Mellon University. 36 * All rights reserved. 37 * 38 * Authors: Avadis Tevanian, Jr., Michael Wayne Young 39 * 40 * Permission to use, copy, modify and distribute this software and 41 * its documentation is hereby granted, provided that both the copyright 42 * notice and this permission notice appear in all copies of the 43 * software, derivative works or modified versions, and any portions 44 * thereof, and that both notices appear in supporting documentation. 45 * 46 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" 47 * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND 48 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. 49 * 50 * Carnegie Mellon requests users of this software to return to 51 * 52 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU 53 * School of Computer Science 54 * Carnegie Mellon University 55 * Pittsburgh PA 15213-3890 56 * 57 * any improvements or extensions that they make and grant Carnegie the 58 * rights to redistribute these changes. 59 * 60 * $FreeBSD$ 61 */ 62 63 /* 64 * Resident memory system definitions. 65 */ 66 67 #ifndef _VM_PAGE_ 68 #define _VM_PAGE_ 69 70 #include <vm/pmap.h> 71 72 /* 73 * Management of resident (logical) pages. 74 * 75 * A small structure is kept for each resident 76 * page, indexed by page number. Each structure 77 * is an element of several lists: 78 * 79 * A hash table bucket used to quickly 80 * perform object/offset lookups 81 * 82 * A list of all pages for a given object, 83 * so they can be quickly deactivated at 84 * time of deallocation. 85 * 86 * An ordered list of pages due for pageout. 87 * 88 * In addition, the structure contains the object 89 * and offset to which this page belongs (for pageout), 90 * and sundry status bits. 91 * 92 * Fields in this structure are locked either by the lock on the 93 * object that the page belongs to (O) or by the lock on the page 94 * queues (P). 95 * 96 * The 'valid' and 'dirty' fields are distinct. A page may have dirty 97 * bits set without having associated valid bits set. This is used by 98 * NFS to implement piecemeal writes. 99 */ 100 101 TAILQ_HEAD(pglist, vm_page); 102 103 struct vm_page { 104 TAILQ_ENTRY(vm_page) pageq; /* queue info for FIFO queue or free list (P) */ 105 TAILQ_ENTRY(vm_page) listq; /* pages in same object (O) */ 106 struct vm_page *left; /* splay tree link (O) */ 107 struct vm_page *right; /* splay tree link (O) */ 108 109 vm_object_t object; /* which object am I in (O,P)*/ 110 vm_pindex_t pindex; /* offset into object (O,P) */ 111 vm_paddr_t phys_addr; /* physical address of page */ 112 struct md_page md; /* machine dependant stuff */ 113 uint8_t queue; /* page queue index */ 114 int8_t segind; 115 u_short flags; /* see below */ 116 uint8_t order; /* index of the buddy queue */ 117 uint8_t pool; 118 u_short wire_count; /* wired down maps refs (P) */ 119 u_int cow; /* page cow mapping count */ 120 short hold_count; /* page hold count */ 121 u_short oflags; /* page flags (O) */ 122 u_char act_count; /* page usage count */ 123 u_char busy; /* page busy count (O) */ 124 /* NOTE that these must support one bit per DEV_BSIZE in a page!!! */ 125 /* so, on normal X86 kernels, they must be at least 8 bits wide */ 126 #if PAGE_SIZE == 4096 127 u_char valid; /* map of valid DEV_BSIZE chunks (O) */ 128 u_char dirty; /* map of dirty DEV_BSIZE chunks */ 129 #elif PAGE_SIZE == 8192 130 u_short valid; /* map of valid DEV_BSIZE chunks (O) */ 131 u_short dirty; /* map of dirty DEV_BSIZE chunks */ 132 #elif PAGE_SIZE == 16384 133 u_int valid; /* map of valid DEV_BSIZE chunks (O) */ 134 u_int dirty; /* map of dirty DEV_BSIZE chunks */ 135 #elif PAGE_SIZE == 32768 136 u_long valid; /* map of valid DEV_BSIZE chunks (O) */ 137 u_long dirty; /* map of dirty DEV_BSIZE chunks */ 138 #endif 139 }; 140 141 /* 142 * Page flags stored in oflags: 143 * 144 * Access to these page flags is synchronized by the lock on the object 145 * containing the page (O). 146 */ 147 #define VPO_BUSY 0x0001 /* page is in transit */ 148 #define VPO_WANTED 0x0002 /* someone is waiting for page */ 149 #define VPO_CLEANCHK 0x0100 /* page will be checked for cleaning */ 150 #define VPO_SWAPINPROG 0x0200 /* swap I/O in progress on page */ 151 #define VPO_NOSYNC 0x0400 /* do not collect for syncer */ 152 153 /* Make sure that u_long is at least 64 bits when PAGE_SIZE is 32K. */ 154 #if PAGE_SIZE == 32768 155 #ifdef CTASSERT 156 CTASSERT(sizeof(u_long) >= 8); 157 #endif 158 #endif 159 160 #define PQ_NONE 0 161 #define PQ_INACTIVE 1 162 #define PQ_ACTIVE 2 163 #define PQ_HOLD 3 164 #define PQ_COUNT 4 165 166 /* Returns the real queue a page is on. */ 167 #define VM_PAGE_GETQUEUE(m) ((m)->queue) 168 169 /* Returns the well known queue a page is on. */ 170 #define VM_PAGE_GETKNOWNQUEUE2(m) VM_PAGE_GETQUEUE(m) 171 172 /* Returns true if the page is in the named well known queue. */ 173 #define VM_PAGE_INQUEUE2(m, q) (VM_PAGE_GETKNOWNQUEUE2(m) == (q)) 174 175 /* Sets the queue a page is on. */ 176 #define VM_PAGE_SETQUEUE2(m, q) (VM_PAGE_GETQUEUE(m) = (q)) 177 178 struct vpgqueues { 179 struct pglist pl; 180 int *cnt; 181 }; 182 183 extern struct vpgqueues vm_page_queues[PQ_COUNT]; 184 extern struct mtx vm_page_queue_free_mtx; 185 186 /* 187 * These are the flags defined for vm_page. 188 * 189 * Note: PG_UNMANAGED (used by OBJT_PHYS) indicates that the page is 190 * not under PV management but otherwise should be treated as a 191 * normal page. Pages not under PV management cannot be paged out 192 * via the object/vm_page_t because there is no knowledge of their 193 * pte mappings, nor can they be removed from their objects via 194 * the object, and such pages are also not on any PQ queue. 195 */ 196 #define PG_CACHED 0x0001 /* page is cached */ 197 #define PG_FREE 0x0002 /* page is free */ 198 #define PG_WINATCFLS 0x0004 /* flush dirty page on inactive q */ 199 #define PG_FICTITIOUS 0x0008 /* physical page doesn't exist (O) */ 200 #define PG_WRITEABLE 0x0010 /* page is mapped writeable */ 201 #define PG_ZERO 0x0040 /* page is zeroed */ 202 #define PG_REFERENCED 0x0080 /* page has been referenced */ 203 #define PG_UNMANAGED 0x0800 /* No PV management for page */ 204 #define PG_MARKER 0x1000 /* special queue marker page */ 205 #define PG_SLAB 0x2000 /* object pointer is actually a slab */ 206 207 /* 208 * Misc constants. 209 */ 210 #define ACT_DECLINE 1 211 #define ACT_ADVANCE 3 212 #define ACT_INIT 5 213 #define ACT_MAX 64 214 215 #ifdef _KERNEL 216 217 #include <vm/vm_param.h> 218 219 /* 220 * Each pageable resident page falls into one of five lists: 221 * 222 * free 223 * Available for allocation now. 224 * 225 * cache 226 * Almost available for allocation. Still associated with 227 * an object, but clean and immediately freeable. 228 * 229 * hold 230 * Will become free after a pending I/O operation 231 * completes. 232 * 233 * The following lists are LRU sorted: 234 * 235 * inactive 236 * Low activity, candidates for reclamation. 237 * This is the list of pages that should be 238 * paged out next. 239 * 240 * active 241 * Pages that are "active" i.e. they have been 242 * recently referenced. 243 * 244 */ 245 246 extern int vm_page_zero_count; 247 248 extern vm_page_t vm_page_array; /* First resident page in table */ 249 extern int vm_page_array_size; /* number of vm_page_t's */ 250 extern long first_page; /* first physical page number */ 251 252 #define VM_PAGE_IS_FREE(m) (((m)->flags & PG_FREE) != 0) 253 254 #define VM_PAGE_TO_PHYS(entry) ((entry)->phys_addr) 255 256 vm_page_t vm_phys_paddr_to_vm_page(vm_paddr_t pa); 257 258 static __inline vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa); 259 260 static __inline vm_page_t 261 PHYS_TO_VM_PAGE(vm_paddr_t pa) 262 { 263 #ifdef VM_PHYSSEG_SPARSE 264 return (vm_phys_paddr_to_vm_page(pa)); 265 #elif defined(VM_PHYSSEG_DENSE) 266 return (&vm_page_array[atop(pa) - first_page]); 267 #else 268 #error "Either VM_PHYSSEG_DENSE or VM_PHYSSEG_SPARSE must be defined." 269 #endif 270 } 271 272 extern struct mtx vm_page_queue_mtx; 273 #define vm_page_lock_queues() mtx_lock(&vm_page_queue_mtx) 274 #define vm_page_unlock_queues() mtx_unlock(&vm_page_queue_mtx) 275 276 #if PAGE_SIZE == 4096 277 #define VM_PAGE_BITS_ALL 0xffu 278 #elif PAGE_SIZE == 8192 279 #define VM_PAGE_BITS_ALL 0xffffu 280 #elif PAGE_SIZE == 16384 281 #define VM_PAGE_BITS_ALL 0xffffffffu 282 #elif PAGE_SIZE == 32768 283 #define VM_PAGE_BITS_ALL 0xfffffffffffffffflu 284 #endif 285 286 /* page allocation classes: */ 287 #define VM_ALLOC_NORMAL 0 288 #define VM_ALLOC_INTERRUPT 1 289 #define VM_ALLOC_SYSTEM 2 290 #define VM_ALLOC_CLASS_MASK 3 291 /* page allocation flags: */ 292 #define VM_ALLOC_WIRED 0x0020 /* non pageable */ 293 #define VM_ALLOC_ZERO 0x0040 /* Try to obtain a zeroed page */ 294 #define VM_ALLOC_RETRY 0x0080 /* vm_page_grab() only */ 295 #define VM_ALLOC_NOOBJ 0x0100 /* No associated object */ 296 #define VM_ALLOC_NOBUSY 0x0200 /* Do not busy the page */ 297 #define VM_ALLOC_IFCACHED 0x0400 /* Fail if the page is not cached */ 298 #define VM_ALLOC_IFNOTCACHED 0x0800 /* Fail if the page is cached */ 299 300 void vm_page_flag_set(vm_page_t m, unsigned short bits); 301 void vm_page_flag_clear(vm_page_t m, unsigned short bits); 302 void vm_page_busy(vm_page_t m); 303 void vm_page_flash(vm_page_t m); 304 void vm_page_io_start(vm_page_t m); 305 void vm_page_io_finish(vm_page_t m); 306 void vm_page_hold(vm_page_t mem); 307 void vm_page_unhold(vm_page_t mem); 308 void vm_page_free(vm_page_t m); 309 void vm_page_free_zero(vm_page_t m); 310 void vm_page_dirty(vm_page_t m); 311 void vm_page_wakeup(vm_page_t m); 312 313 void vm_pageq_remove(vm_page_t m); 314 315 void vm_page_activate (vm_page_t); 316 vm_page_t vm_page_alloc (vm_object_t, vm_pindex_t, int); 317 vm_page_t vm_page_grab (vm_object_t, vm_pindex_t, int); 318 void vm_page_cache (register vm_page_t); 319 void vm_page_cache_free(vm_object_t, vm_pindex_t, vm_pindex_t); 320 void vm_page_cache_remove(vm_page_t); 321 void vm_page_cache_transfer(vm_object_t, vm_pindex_t, vm_object_t); 322 int vm_page_try_to_cache (vm_page_t); 323 int vm_page_try_to_free (vm_page_t); 324 void vm_page_dontneed (register vm_page_t); 325 void vm_page_deactivate (vm_page_t); 326 void vm_page_insert (vm_page_t, vm_object_t, vm_pindex_t); 327 vm_page_t vm_page_lookup (vm_object_t, vm_pindex_t); 328 void vm_page_remove (vm_page_t); 329 void vm_page_rename (vm_page_t, vm_object_t, vm_pindex_t); 330 void vm_page_requeue(vm_page_t m); 331 void vm_page_sleep(vm_page_t m, const char *msg); 332 vm_page_t vm_page_splay(vm_pindex_t, vm_page_t); 333 vm_offset_t vm_page_startup(vm_offset_t vaddr); 334 void vm_page_unwire (vm_page_t, int); 335 void vm_page_wire (vm_page_t); 336 void vm_page_set_validclean (vm_page_t, int, int); 337 void vm_page_clear_dirty (vm_page_t, int, int); 338 void vm_page_set_invalid (vm_page_t, int, int); 339 int vm_page_is_valid (vm_page_t, int, int); 340 void vm_page_test_dirty (vm_page_t); 341 int vm_page_bits (int, int); 342 void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid); 343 void vm_page_free_toq(vm_page_t m); 344 void vm_page_zero_idle_wakeup(void); 345 void vm_page_cowfault (vm_page_t); 346 void vm_page_cowsetup (vm_page_t); 347 void vm_page_cowclear (vm_page_t); 348 349 /* 350 * vm_page_sleep_if_busy: 351 * 352 * Sleep and release the page queues lock if VPO_BUSY is set or, 353 * if also_m_busy is TRUE, busy is non-zero. Returns TRUE if the 354 * thread slept and the page queues lock was released. 355 * Otherwise, retains the page queues lock and returns FALSE. 356 * 357 * The object containing the given page must be locked. 358 */ 359 static __inline int 360 vm_page_sleep_if_busy(vm_page_t m, int also_m_busy, const char *msg) 361 { 362 363 if ((m->oflags & VPO_BUSY) || (also_m_busy && m->busy)) { 364 vm_page_sleep(m, msg); 365 return (TRUE); 366 } 367 return (FALSE); 368 } 369 370 /* 371 * vm_page_undirty: 372 * 373 * Set page to not be dirty. Note: does not clear pmap modify bits 374 */ 375 static __inline void 376 vm_page_undirty(vm_page_t m) 377 { 378 m->dirty = 0; 379 } 380 381 #endif /* _KERNEL */ 382 #endif /* !_VM_PAGE_ */ 383