vm_page.h revision 80705
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 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 20 * 4. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 * 36 * from: @(#)vm_page.h 8.2 (Berkeley) 12/13/93 37 * 38 * 39 * Copyright (c) 1987, 1990 Carnegie-Mellon University. 40 * All rights reserved. 41 * 42 * Authors: Avadis Tevanian, Jr., Michael Wayne Young 43 * 44 * Permission to use, copy, modify and distribute this software and 45 * its documentation is hereby granted, provided that both the copyright 46 * notice and this permission notice appear in all copies of the 47 * software, derivative works or modified versions, and any portions 48 * thereof, and that both notices appear in supporting documentation. 49 * 50 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" 51 * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND 52 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. 53 * 54 * Carnegie Mellon requests users of this software to return to 55 * 56 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU 57 * School of Computer Science 58 * Carnegie Mellon University 59 * Pittsburgh PA 15213-3890 60 * 61 * any improvements or extensions that they make and grant Carnegie the 62 * rights to redistribute these changes. 63 * 64 * $FreeBSD: head/sys/vm/vm_page.h 80705 2001-07-31 04:09:52Z jake $ 65 */ 66 67/* 68 * Resident memory system definitions. 69 */ 70 71#ifndef _VM_PAGE_ 72#define _VM_PAGE_ 73 74#if !defined(KLD_MODULE) 75#include "opt_vmpage.h" 76#endif 77 78#include <vm/pmap.h> 79 80/* 81 * Management of resident (logical) pages. 82 * 83 * A small structure is kept for each resident 84 * page, indexed by page number. Each structure 85 * is an element of several lists: 86 * 87 * A hash table bucket used to quickly 88 * perform object/offset lookups 89 * 90 * A list of all pages for a given object, 91 * so they can be quickly deactivated at 92 * time of deallocation. 93 * 94 * An ordered list of pages due for pageout. 95 * 96 * In addition, the structure contains the object 97 * and offset to which this page belongs (for pageout), 98 * and sundry status bits. 99 * 100 * Fields in this structure are locked either by the lock on the 101 * object that the page belongs to (O) or by the lock on the page 102 * queues (P). 103 * 104 * The 'valid' and 'dirty' fields are distinct. A page may have dirty 105 * bits set without having associated valid bits set. This is used by 106 * NFS to implement piecemeal writes. 107 */ 108 109TAILQ_HEAD(pglist, vm_page); 110 111struct vm_page { 112 TAILQ_ENTRY(vm_page) pageq; /* queue info for FIFO queue or free list (P) */ 113 struct vm_page *hnext; /* hash table link (O,P) */ 114 TAILQ_ENTRY(vm_page) listq; /* pages in same object (O) */ 115 116 vm_object_t object; /* which object am I in (O,P)*/ 117 vm_pindex_t pindex; /* offset into object (O,P) */ 118 vm_offset_t phys_addr; /* physical address of page */ 119 struct md_page md; /* machine dependant stuff */ 120 u_short queue; /* page queue index */ 121 u_short flags, /* see below */ 122 pc; /* page color */ 123 u_short wire_count; /* wired down maps refs (P) */ 124 short hold_count; /* page hold count */ 125 u_char act_count; /* page usage count */ 126 u_char busy; /* page busy count */ 127 /* NOTE that these must support one bit per DEV_BSIZE in a page!!! */ 128 /* so, on normal X86 kernels, they must be at least 8 bits wide */ 129#if PAGE_SIZE == 4096 130 u_char valid; /* map of valid DEV_BSIZE chunks */ 131 u_char dirty; /* map of dirty DEV_BSIZE chunks */ 132#elif PAGE_SIZE == 8192 133 u_short valid; /* map of valid DEV_BSIZE chunks */ 134 u_short dirty; /* map of dirty DEV_BSIZE chunks */ 135#endif 136}; 137 138/* 139 * note: currently use SWAPBLK_NONE as an absolute value rather then 140 * a flag bit. 141 */ 142 143#define SWAPBLK_MASK ((daddr_t)((u_daddr_t)-1 >> 1)) /* mask */ 144#define SWAPBLK_NONE ((daddr_t)((u_daddr_t)SWAPBLK_MASK + 1))/* flag */ 145 146#if !defined(KLD_MODULE) 147 148/* 149 * shared mutex array for vm_page_buckets[] 150 */ 151#ifndef BUCKET_HASH_SIZE 152#define BUCKET_HASH_SIZE 16 153#endif 154#define BUCKET_HASH_MASK (BUCKET_HASH_SIZE - 1) 155 156/* 157 * Page coloring parameters 158 */ 159/* Each of PQ_FREE, and PQ_CACHE have PQ_HASH_SIZE entries */ 160 161/* Backward compatibility for existing PQ_*CACHE config options. */ 162#if !defined(PQ_CACHESIZE) 163#if defined(PQ_HUGECACHE) 164#define PQ_CACHESIZE 1024 165#elif defined(PQ_LARGECACHE) 166#define PQ_CACHESIZE 512 167#elif defined(PQ_MEDIUMCACHE) 168#define PQ_CACHESIZE 256 169#elif defined(PQ_NORMALCACHE) 170#define PQ_CACHESIZE 64 171#elif defined(PQ_NOOPT) 172#define PQ_CACHESIZE 0 173#else 174#define PQ_CACHESIZE 128 175#endif 176#endif 177 178#if PQ_CACHESIZE >= 1024 179#define PQ_PRIME1 31 /* Prime number somewhat less than PQ_HASH_SIZE */ 180#define PQ_PRIME2 23 /* Prime number somewhat less than PQ_HASH_SIZE */ 181#define PQ_L2_SIZE 256 /* A number of colors opt for 1M cache */ 182 183#elif PQ_CACHESIZE >= 512 184#define PQ_PRIME1 31 /* Prime number somewhat less than PQ_HASH_SIZE */ 185#define PQ_PRIME2 23 /* Prime number somewhat less than PQ_HASH_SIZE */ 186#define PQ_L2_SIZE 128 /* A number of colors opt for 512K cache */ 187 188#elif PQ_CACHESIZE >= 256 189#define PQ_PRIME1 13 /* Prime number somewhat less than PQ_HASH_SIZE */ 190#define PQ_PRIME2 7 /* Prime number somewhat less than PQ_HASH_SIZE */ 191#define PQ_L2_SIZE 64 /* A number of colors opt for 256K cache */ 192 193#elif PQ_CACHESIZE >= 128 194#define PQ_PRIME1 9 /* Produces a good PQ_L2_SIZE/3 + PQ_PRIME1 */ 195#define PQ_PRIME2 5 /* Prime number somewhat less than PQ_HASH_SIZE */ 196#define PQ_L2_SIZE 32 /* A number of colors opt for 128k cache */ 197 198#elif PQ_CACHESIZE >= 64 199#define PQ_PRIME1 5 /* Prime number somewhat less than PQ_HASH_SIZE */ 200#define PQ_PRIME2 3 /* Prime number somewhat less than PQ_HASH_SIZE */ 201#define PQ_L2_SIZE 16 /* A reasonable number of colors (opt for 64K cache) */ 202 203#else 204#define PQ_PRIME1 1 /* Disable page coloring. */ 205#define PQ_PRIME2 1 206#define PQ_L2_SIZE 1 207 208#endif 209 210#define PQ_L2_MASK (PQ_L2_SIZE - 1) 211 212#if 1 213#define PQ_NONE 0 214#define PQ_FREE 1 215#define PQ_INACTIVE (1 + 1*PQ_L2_SIZE) 216#define PQ_ACTIVE (2 + 1*PQ_L2_SIZE) 217#define PQ_CACHE (3 + 1*PQ_L2_SIZE) 218#define PQ_COUNT (3 + 2*PQ_L2_SIZE) 219#else 220#define PQ_NONE PQ_COUNT 221#define PQ_FREE 0 222#define PQ_INACTIVE PQ_L2_SIZE 223#define PQ_ACTIVE (1 + PQ_L2_SIZE) 224#define PQ_CACHE (2 + PQ_L2_SIZE) 225#define PQ_COUNT (2 + 2*PQ_L2_SIZE) 226#endif 227 228struct vpgqueues { 229 struct pglist pl; 230 int *cnt; 231 int lcnt; 232}; 233 234extern struct vpgqueues vm_page_queues[PQ_COUNT]; 235 236#endif 237 238/* 239 * These are the flags defined for vm_page. 240 * 241 * Note: PG_FILLED and PG_DIRTY are added for the filesystems. 242 * 243 * Note: PG_UNMANAGED (used by OBJT_PHYS) indicates that the page is 244 * not under PV management but otherwise should be treated as a 245 * normal page. Pages not under PV management cannot be paged out 246 * via the object/vm_page_t because there is no knowledge of their 247 * pte mappings, nor can they be removed from their objects via 248 * the object, and such pages are also not on any PQ queue. 249 */ 250#define PG_BUSY 0x0001 /* page is in transit (O) */ 251#define PG_WANTED 0x0002 /* someone is waiting for page (O) */ 252#define PG_WINATCFLS 0x0004 /* flush dirty page on inactive q */ 253#define PG_FICTITIOUS 0x0008 /* physical page doesn't exist (O) */ 254#define PG_WRITEABLE 0x0010 /* page is mapped writeable */ 255#define PG_MAPPED 0x0020 /* page is mapped */ 256#define PG_ZERO 0x0040 /* page is zeroed */ 257#define PG_REFERENCED 0x0080 /* page has been referenced */ 258#define PG_CLEANCHK 0x0100 /* page will be checked for cleaning */ 259#define PG_SWAPINPROG 0x0200 /* swap I/O in progress on page */ 260#define PG_NOSYNC 0x0400 /* do not collect for syncer */ 261#define PG_UNMANAGED 0x0800 /* No PV management for page */ 262#define PG_MARKER 0x1000 /* special queue marker page */ 263 264/* 265 * Misc constants. 266 */ 267 268#define ACT_DECLINE 1 269#define ACT_ADVANCE 3 270#define ACT_INIT 5 271#define ACT_MAX 64 272#define PFCLUSTER_BEHIND 3 273#define PFCLUSTER_AHEAD 3 274 275#ifdef _KERNEL 276/* 277 * Each pageable resident page falls into one of four lists: 278 * 279 * free 280 * Available for allocation now. 281 * 282 * The following are all LRU sorted: 283 * 284 * cache 285 * Almost available for allocation. Still in an 286 * object, but clean and immediately freeable at 287 * non-interrupt times. 288 * 289 * inactive 290 * Low activity, candidates for reclamation. 291 * This is the list of pages that should be 292 * paged out next. 293 * 294 * active 295 * Pages that are "active" i.e. they have been 296 * recently referenced. 297 * 298 * zero 299 * Pages that are really free and have been pre-zeroed 300 * 301 */ 302 303extern int vm_page_zero_count; 304 305extern vm_page_t vm_page_array; /* First resident page in table */ 306extern int vm_page_array_size; /* number of vm_page_t's */ 307extern long first_page; /* first physical page number */ 308 309#define VM_PAGE_TO_PHYS(entry) ((entry)->phys_addr) 310 311#define PHYS_TO_VM_PAGE(pa) \ 312 (&vm_page_array[atop(pa) - first_page ]) 313 314 315#if PAGE_SIZE == 4096 316#define VM_PAGE_BITS_ALL 0xff 317#endif 318 319#if PAGE_SIZE == 8192 320#define VM_PAGE_BITS_ALL 0xffff 321#endif 322 323#define VM_ALLOC_NORMAL 0 324#define VM_ALLOC_INTERRUPT 1 325#define VM_ALLOC_SYSTEM 2 326#define VM_ALLOC_ZERO 3 327#define VM_ALLOC_RETRY 0x80 328 329void vm_page_flag_set(vm_page_t m, unsigned short bits); 330void vm_page_flag_clear(vm_page_t m, unsigned short bits); 331void vm_page_busy(vm_page_t m); 332void vm_page_flash(vm_page_t m); 333void vm_page_io_start(vm_page_t m); 334void vm_page_io_finish(vm_page_t m); 335void vm_page_hold(vm_page_t mem); 336void vm_page_unhold(vm_page_t mem); 337void vm_page_protect(vm_page_t mem, int prot); 338boolean_t vm_page_zero_fill(vm_page_t m); 339void vm_page_copy(vm_page_t src_m, vm_page_t dest_m); 340void vm_page_free(vm_page_t m); 341void vm_page_free_zero(vm_page_t m); 342int vm_page_sleep_busy(vm_page_t m, int also_m_busy, const char *msg); 343void vm_page_dirty(vm_page_t m); 344void vm_page_undirty(vm_page_t m); 345vm_page_t vm_page_list_find(int basequeue, int index, boolean_t prefer_zero); 346void vm_page_wakeup(vm_page_t m); 347 348void vm_pageq_init(void); 349struct vpgqueues *vm_pageq_aquire(int queue); 350void vm_pageq_release(struct vpgqueues *vpq); 351vm_page_t vm_pageq_add_new_page(vm_offset_t pa); 352void vm_pageq_remove_nowakeup(vm_page_t m); 353void vm_pageq_remove(vm_page_t m); 354vm_page_t vm_pageq_find(int basequeue, int index, boolean_t prefer_zero); 355void vm_pageq_requeue(vm_page_t m); 356 357void vm_page_activate (vm_page_t); 358vm_page_t vm_page_alloc (vm_object_t, vm_pindex_t, int); 359vm_page_t vm_page_grab (vm_object_t, vm_pindex_t, int); 360void vm_page_cache (register vm_page_t); 361int vm_page_try_to_cache (vm_page_t); 362int vm_page_try_to_free (vm_page_t); 363void vm_page_dontneed (register vm_page_t); 364void vm_page_deactivate (vm_page_t); 365void vm_page_insert (vm_page_t, vm_object_t, vm_pindex_t); 366vm_page_t vm_page_lookup (vm_object_t, vm_pindex_t); 367void vm_page_remove (vm_page_t); 368void vm_page_rename (vm_page_t, vm_object_t, vm_pindex_t); 369vm_offset_t vm_page_startup (vm_offset_t, vm_offset_t, vm_offset_t); 370vm_page_t vm_add_new_page (vm_offset_t pa); 371void vm_page_unmanage (vm_page_t); 372void vm_page_unwire (vm_page_t, int); 373void vm_page_wire (vm_page_t); 374void vm_page_unqueue (vm_page_t); 375void vm_page_unqueue_nowakeup (vm_page_t); 376void vm_page_set_validclean (vm_page_t, int, int); 377void vm_page_set_dirty (vm_page_t, int, int); 378void vm_page_clear_dirty (vm_page_t, int, int); 379void vm_page_set_invalid (vm_page_t, int, int); 380int vm_page_is_valid (vm_page_t, int, int); 381void vm_page_test_dirty (vm_page_t); 382int vm_page_bits (int, int); 383#if 0 384int vm_page_sleep(vm_page_t m, char *msg, char *busy); 385int vm_page_asleep(vm_page_t m, char *msg, char *busy); 386#endif 387void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid); 388void vm_page_free_toq(vm_page_t m); 389 390#endif /* _KERNEL */ 391#endif /* !_VM_PAGE_ */ 392