vm_page.h revision 79263
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 79263 2001-07-04 23:27:09Z dillon $ 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#include <machine/atomic.h> 80 81/* 82 * Management of resident (logical) pages. 83 * 84 * A small structure is kept for each resident 85 * page, indexed by page number. Each structure 86 * is an element of several lists: 87 * 88 * A hash table bucket used to quickly 89 * perform object/offset lookups 90 * 91 * A list of all pages for a given object, 92 * so they can be quickly deactivated at 93 * time of deallocation. 94 * 95 * An ordered list of pages due for pageout. 96 * 97 * In addition, the structure contains the object 98 * and offset to which this page belongs (for pageout), 99 * and sundry status bits. 100 * 101 * Fields in this structure are locked either by the lock on the 102 * object that the page belongs to (O) or by the lock on the page 103 * queues (P). 104 * 105 * The 'valid' and 'dirty' fields are distinct. A page may have dirty 106 * bits set without having associated valid bits set. This is used by 107 * NFS to implement piecemeal writes. 108 */ 109 110TAILQ_HEAD(pglist, vm_page); 111 112struct vm_page { 113 TAILQ_ENTRY(vm_page) pageq; /* queue info for FIFO queue or free list (P) */ 114 struct vm_page *hnext; /* hash table link (O,P) */ 115 TAILQ_ENTRY(vm_page) listq; /* pages in same object (O) */ 116 117 vm_object_t object; /* which object am I in (O,P)*/ 118 vm_pindex_t pindex; /* offset into object (O,P) */ 119 vm_offset_t phys_addr; /* physical address of page */ 120 struct md_page md; /* machine dependant stuff */ 121 u_short queue; /* page queue index */ 122 u_short flags, /* see below */ 123 pc; /* page color */ 124 u_short wire_count; /* wired down maps refs (P) */ 125 short hold_count; /* page hold count */ 126 u_char act_count; /* page usage count */ 127 u_char busy; /* page busy count */ 128 /* NOTE that these must support one bit per DEV_BSIZE in a page!!! */ 129 /* so, on normal X86 kernels, they must be at least 8 bits wide */ 130#if PAGE_SIZE == 4096 131 u_char valid; /* map of valid DEV_BSIZE chunks */ 132 u_char dirty; /* map of dirty DEV_BSIZE chunks */ 133#elif PAGE_SIZE == 8192 134 u_short valid; /* map of valid DEV_BSIZE chunks */ 135 u_short dirty; /* map of dirty DEV_BSIZE chunks */ 136#endif 137}; 138 139/* 140 * note: currently use SWAPBLK_NONE as an absolute value rather then 141 * a flag bit. 142 */ 143 144#define SWAPBLK_MASK ((daddr_t)((u_daddr_t)-1 >> 1)) /* mask */ 145#define SWAPBLK_NONE ((daddr_t)((u_daddr_t)SWAPBLK_MASK + 1))/* flag */ 146 147#if !defined(KLD_MODULE) 148 149/* 150 * shared mutex array for vm_page_buckets[] 151 */ 152#ifndef BUCKET_HASH_SIZE 153#define BUCKET_HASH_SIZE 16 154#endif 155#define BUCKET_HASH_MASK (BUCKET_HASH_SIZE - 1) 156 157/* 158 * Page coloring parameters 159 */ 160/* Each of PQ_FREE, and PQ_CACHE have PQ_HASH_SIZE entries */ 161 162/* Backward compatibility for existing PQ_*CACHE config options. */ 163#if !defined(PQ_CACHESIZE) 164#if defined(PQ_HUGECACHE) 165#define PQ_CACHESIZE 1024 166#elif defined(PQ_LARGECACHE) 167#define PQ_CACHESIZE 512 168#elif defined(PQ_MEDIUMCACHE) 169#define PQ_CACHESIZE 256 170#elif defined(PQ_NORMALCACHE) 171#define PQ_CACHESIZE 64 172#elif defined(PQ_NOOPT) 173#define PQ_CACHESIZE 0 174#else 175#define PQ_CACHESIZE 128 176#endif 177#endif 178 179#if PQ_CACHESIZE >= 1024 180#define PQ_PRIME1 31 /* Prime number somewhat less than PQ_HASH_SIZE */ 181#define PQ_PRIME2 23 /* Prime number somewhat less than PQ_HASH_SIZE */ 182#define PQ_L2_SIZE 256 /* A number of colors opt for 1M cache */ 183 184#elif PQ_CACHESIZE >= 512 185#define PQ_PRIME1 31 /* Prime number somewhat less than PQ_HASH_SIZE */ 186#define PQ_PRIME2 23 /* Prime number somewhat less than PQ_HASH_SIZE */ 187#define PQ_L2_SIZE 128 /* A number of colors opt for 512K cache */ 188 189#elif PQ_CACHESIZE >= 256 190#define PQ_PRIME1 13 /* Prime number somewhat less than PQ_HASH_SIZE */ 191#define PQ_PRIME2 7 /* Prime number somewhat less than PQ_HASH_SIZE */ 192#define PQ_L2_SIZE 64 /* A number of colors opt for 256K cache */ 193 194#elif PQ_CACHESIZE >= 128 195#define PQ_PRIME1 9 /* Produces a good PQ_L2_SIZE/3 + PQ_PRIME1 */ 196#define PQ_PRIME2 5 /* Prime number somewhat less than PQ_HASH_SIZE */ 197#define PQ_L2_SIZE 32 /* A number of colors opt for 128k cache */ 198 199#elif PQ_CACHESIZE >= 64 200#define PQ_PRIME1 5 /* Prime number somewhat less than PQ_HASH_SIZE */ 201#define PQ_PRIME2 3 /* Prime number somewhat less than PQ_HASH_SIZE */ 202#define PQ_L2_SIZE 16 /* A reasonable number of colors (opt for 64K cache) */ 203 204#else 205#define PQ_PRIME1 1 /* Disable page coloring. */ 206#define PQ_PRIME2 1 207#define PQ_L2_SIZE 1 208 209#endif 210 211#define PQ_L2_MASK (PQ_L2_SIZE - 1) 212 213#if 1 214#define PQ_NONE 0 215#define PQ_FREE 1 216#define PQ_INACTIVE (1 + 1*PQ_L2_SIZE) 217#define PQ_ACTIVE (2 + 1*PQ_L2_SIZE) 218#define PQ_CACHE (3 + 1*PQ_L2_SIZE) 219#define PQ_COUNT (3 + 2*PQ_L2_SIZE) 220#else 221#define PQ_NONE PQ_COUNT 222#define PQ_FREE 0 223#define PQ_INACTIVE PQ_L2_SIZE 224#define PQ_ACTIVE (1 + PQ_L2_SIZE) 225#define PQ_CACHE (2 + PQ_L2_SIZE) 226#define PQ_COUNT (2 + 2*PQ_L2_SIZE) 227#endif 228 229struct vpgqueues { 230 struct pglist pl; 231 int *cnt; 232 int lcnt; 233}; 234 235extern struct vpgqueues vm_page_queues[PQ_COUNT]; 236 237#endif 238 239/* 240 * These are the flags defined for vm_page. 241 * 242 * Note: PG_FILLED and PG_DIRTY are added for the filesystems. 243 * 244 * Note: PG_UNMANAGED (used by OBJT_PHYS) indicates that the page is 245 * not under PV management but otherwise should be treated as a 246 * normal page. Pages not under PV management cannot be paged out 247 * via the object/vm_page_t because there is no knowledge of their 248 * pte mappings, nor can they be removed from their objects via 249 * the object, and such pages are also not on any PQ queue. 250 */ 251#define PG_BUSY 0x0001 /* page is in transit (O) */ 252#define PG_WANTED 0x0002 /* someone is waiting for page (O) */ 253#define PG_WINATCFLS 0x0004 /* flush dirty page on inactive q */ 254#define PG_FICTITIOUS 0x0008 /* physical page doesn't exist (O) */ 255#define PG_WRITEABLE 0x0010 /* page is mapped writeable */ 256#define PG_MAPPED 0x0020 /* page is mapped */ 257#define PG_ZERO 0x0040 /* page is zeroed */ 258#define PG_REFERENCED 0x0080 /* page has been referenced */ 259#define PG_CLEANCHK 0x0100 /* page will be checked for cleaning */ 260#define PG_SWAPINPROG 0x0200 /* swap I/O in progress on page */ 261#define PG_NOSYNC 0x0400 /* do not collect for syncer */ 262#define PG_UNMANAGED 0x0800 /* No PV management for page */ 263#define PG_MARKER 0x1000 /* special queue marker page */ 264 265/* 266 * Misc constants. 267 */ 268 269#define ACT_DECLINE 1 270#define ACT_ADVANCE 3 271#define ACT_INIT 5 272#define ACT_MAX 64 273#define PFCLUSTER_BEHIND 3 274#define PFCLUSTER_AHEAD 3 275 276#ifdef _KERNEL 277/* 278 * Each pageable resident page falls into one of four lists: 279 * 280 * free 281 * Available for allocation now. 282 * 283 * The following are all LRU sorted: 284 * 285 * cache 286 * Almost available for allocation. Still in an 287 * object, but clean and immediately freeable at 288 * non-interrupt times. 289 * 290 * inactive 291 * Low activity, candidates for reclamation. 292 * This is the list of pages that should be 293 * paged out next. 294 * 295 * active 296 * Pages that are "active" i.e. they have been 297 * recently referenced. 298 * 299 * zero 300 * Pages that are really free and have been pre-zeroed 301 * 302 */ 303 304extern int vm_page_zero_count; 305 306extern vm_page_t vm_page_array; /* First resident page in table */ 307extern int vm_page_array_size; /* number of vm_page_t's */ 308extern long first_page; /* first physical page number */ 309 310#define VM_PAGE_TO_PHYS(entry) ((entry)->phys_addr) 311 312#define PHYS_TO_VM_PAGE(pa) \ 313 (&vm_page_array[atop(pa) - first_page ]) 314 315 316#if PAGE_SIZE == 4096 317#define VM_PAGE_BITS_ALL 0xff 318#endif 319 320#if PAGE_SIZE == 8192 321#define VM_PAGE_BITS_ALL 0xffff 322#endif 323 324#define VM_ALLOC_NORMAL 0 325#define VM_ALLOC_INTERRUPT 1 326#define VM_ALLOC_SYSTEM 2 327#define VM_ALLOC_ZERO 3 328#define VM_ALLOC_RETRY 0x80 329 330void vm_page_flag_set(vm_page_t m, unsigned short bits); 331void vm_page_flag_clear(vm_page_t m, unsigned short bits); 332void vm_page_busy(vm_page_t m); 333void vm_page_flash(vm_page_t m); 334void vm_page_io_start(vm_page_t m); 335void vm_page_io_finish(vm_page_t m); 336void vm_page_hold(vm_page_t mem); 337void vm_page_unhold(vm_page_t mem); 338void vm_page_protect(vm_page_t mem, int prot); 339boolean_t vm_page_zero_fill(vm_page_t m); 340void vm_page_copy(vm_page_t src_m, vm_page_t dest_m); 341void vm_page_free(vm_page_t m); 342void vm_page_free_zero(vm_page_t m); 343int vm_page_sleep_busy(vm_page_t m, int also_m_busy, const char *msg); 344void vm_page_dirty(vm_page_t m); 345void vm_page_undirty(vm_page_t m); 346vm_page_t vm_page_list_find(int basequeue, int index, boolean_t prefer_zero); 347void vm_page_wakeup(vm_page_t m); 348 349void vm_pageq_init(void); 350struct vpgqueues *vm_pageq_aquire(int queue); 351void vm_pageq_release(struct vpgqueues *vpq); 352vm_page_t vm_pageq_add_new_page(vm_offset_t pa); 353void vm_pageq_remove_nowakeup(vm_page_t m); 354void vm_pageq_remove(vm_page_t m); 355vm_page_t vm_pageq_find(int basequeue, int index, boolean_t prefer_zero); 356void vm_pageq_requeue(vm_page_t m); 357 358void vm_page_activate (vm_page_t); 359vm_page_t vm_page_alloc (vm_object_t, vm_pindex_t, int); 360vm_page_t vm_page_grab (vm_object_t, vm_pindex_t, int); 361void vm_page_cache (register vm_page_t); 362int vm_page_try_to_cache (vm_page_t); 363int vm_page_try_to_free (vm_page_t); 364void vm_page_dontneed (register vm_page_t); 365void vm_page_deactivate (vm_page_t); 366void vm_page_insert (vm_page_t, vm_object_t, vm_pindex_t); 367vm_page_t vm_page_lookup (vm_object_t, vm_pindex_t); 368void vm_page_remove (vm_page_t); 369void vm_page_rename (vm_page_t, vm_object_t, vm_pindex_t); 370vm_offset_t vm_page_startup (vm_offset_t, vm_offset_t, vm_offset_t); 371vm_page_t vm_add_new_page (vm_offset_t pa); 372void vm_page_unmanage (vm_page_t); 373void vm_page_unwire (vm_page_t, int); 374void vm_page_wire (vm_page_t); 375void vm_page_unqueue (vm_page_t); 376void vm_page_unqueue_nowakeup (vm_page_t); 377void vm_page_set_validclean (vm_page_t, int, int); 378void vm_page_set_dirty (vm_page_t, int, int); 379void vm_page_clear_dirty (vm_page_t, int, int); 380void vm_page_set_invalid (vm_page_t, int, int); 381int vm_page_is_valid (vm_page_t, int, int); 382void vm_page_test_dirty (vm_page_t); 383int vm_page_bits (int, int); 384#if 0 385int vm_page_sleep(vm_page_t m, char *msg, char *busy); 386int vm_page_asleep(vm_page_t m, char *msg, char *busy); 387#endif 388void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid); 389void vm_page_free_toq(vm_page_t m); 390 391#endif /* _KERNEL */ 392#endif /* !_VM_PAGE_ */ 393