vm_page.h revision 33936
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 * $Id: vm_page.h,v 1.36 1998/02/05 03:32:47 dyson Exp $ 65 */ 66 67/* 68 * Resident memory system definitions. 69 */ 70 71#ifndef _VM_PAGE_ 72#define _VM_PAGE_ 73 74#include <vm/pmap.h> 75/* 76 * Management of resident (logical) pages. 77 * 78 * A small structure is kept for each resident 79 * page, indexed by page number. Each structure 80 * is an element of several lists: 81 * 82 * A hash table bucket used to quickly 83 * perform object/offset lookups 84 * 85 * A list of all pages for a given object, 86 * so they can be quickly deactivated at 87 * time of deallocation. 88 * 89 * An ordered list of pages due for pageout. 90 * 91 * In addition, the structure contains the object 92 * and offset to which this page belongs (for pageout), 93 * and sundry status bits. 94 * 95 * Fields in this structure are locked either by the lock on the 96 * object that the page belongs to (O) or by the lock on the page 97 * queues (P). 98 */ 99 100TAILQ_HEAD(pglist, vm_page); 101 102struct vm_page { 103 TAILQ_ENTRY(vm_page) pageq; /* queue info for FIFO queue or free list (P) */ 104 TAILQ_ENTRY(vm_page) hashq; /* hash table links (O) */ 105 TAILQ_ENTRY(vm_page) listq; /* pages in same object (O) */ 106 107 vm_object_t object; /* which object am I in (O,P) */ 108 vm_pindex_t pindex; /* offset into object (O,P) */ 109 vm_offset_t phys_addr; /* physical address of page */ 110 u_short queue; /* page queue index */ 111 u_short flags, /* see below */ 112 pc; /* page color */ 113 u_short wire_count; /* wired down maps refs (P) */ 114 short hold_count; /* page hold count */ 115 u_char act_count; /* page usage count */ 116 u_char busy; /* page busy count */ 117 /* NOTE that these must support one bit per DEV_BSIZE in a page!!! */ 118 /* so, on normal X86 kernels, they must be at least 8 bits wide */ 119 u_char valid; /* map of valid DEV_BSIZE chunks */ 120 u_char dirty; /* map of dirty DEV_BSIZE chunks */ 121}; 122 123/* 124 * Page coloring parameters 125 */ 126/* Each of PQ_FREE, PQ_ZERO and PQ_CACHE have PQ_HASH_SIZE entries */ 127 128/* Define one of the following */ 129#if defined(PQ_LARGECACHE) 130#define PQ_PRIME1 31 /* Prime number somewhat less than PQ_HASH_SIZE */ 131#define PQ_PRIME2 23 /* Prime number somewhat less than PQ_HASH_SIZE */ 132#define PQ_PRIME3 17 /* Prime number somewhat less than PQ_HASH_SIZE */ 133#define PQ_L2_SIZE 128 /* A number of colors opt for 512K cache */ 134#define PQ_L1_SIZE 2 /* Two page L1 cache */ 135#endif 136 137 138/* 139 * Use 'options PQ_NOOPT' to disable page coloring 140 */ 141#if defined(PQ_NOOPT) 142#define PQ_PRIME1 1 143#define PQ_PRIME2 1 144#define PQ_PRIME3 1 145#define PQ_L2_SIZE 1 146#define PQ_L1_SIZE 1 147#endif 148 149#if defined(PQ_NORMALCACHE) 150#define PQ_PRIME1 5 /* Prime number somewhat less than PQ_HASH_SIZE */ 151#define PQ_PRIME2 3 /* Prime number somewhat less than PQ_HASH_SIZE */ 152#define PQ_PRIME3 11 /* Prime number somewhat less than PQ_HASH_SIZE */ 153#define PQ_L2_SIZE 16 /* A reasonable number of colors (opt for 64K cache) */ 154#define PQ_L1_SIZE 2 /* Two page L1 cache */ 155#endif 156 157#if defined(PQ_MEDIUMCACHE) || !defined(PQ_L2_SIZE) 158#define PQ_PRIME1 13 /* Prime number somewhat less than PQ_HASH_SIZE */ 159#define PQ_PRIME2 7 /* Prime number somewhat less than PQ_HASH_SIZE */ 160#define PQ_PRIME3 5 /* Prime number somewhat less than PQ_HASH_SIZE */ 161#define PQ_L2_SIZE 64 /* A number of colors opt for 256K cache */ 162#define PQ_L1_SIZE 2 /* Two page L1 cache */ 163#endif 164 165#define PQ_L2_MASK (PQ_L2_SIZE - 1) 166 167#define PQ_NONE 0 168#define PQ_FREE 1 169#define PQ_ZERO (1 + PQ_L2_SIZE) 170#define PQ_INACTIVE (1 + 2*PQ_L2_SIZE) 171#define PQ_ACTIVE (2 + 2*PQ_L2_SIZE) 172#define PQ_CACHE (3 + 2*PQ_L2_SIZE) 173#define PQ_COUNT (3 + 3*PQ_L2_SIZE) 174 175extern struct vpgqueues { 176 struct pglist *pl; 177 int *cnt; 178 int *lcnt; 179} vm_page_queues[PQ_COUNT]; 180 181/* 182 * These are the flags defined for vm_page. 183 * 184 * Note: PG_FILLED and PG_DIRTY are added for the filesystems. 185 */ 186#define PG_BUSY 0x01 /* page is in transit (O) */ 187#define PG_WANTED 0x02 /* someone is waiting for page (O) */ 188#define PG_TABLED 0x04 /* page is in VP table (O) */ 189#define PG_FICTITIOUS 0x08 /* physical page doesn't exist (O) */ 190#define PG_WRITEABLE 0x10 /* page is mapped writeable */ 191#define PG_MAPPED 0x20 /* page is mapped */ 192#define PG_ZERO 0x40 /* page is zeroed */ 193#define PG_REFERENCED 0x80 /* page has been referenced */ 194#define PG_CLEANCHK 0x100 /* page has been checked for cleaning */ 195 196/* 197 * Misc constants. 198 */ 199 200#define ACT_DECLINE 1 201#define ACT_ADVANCE 3 202#define ACT_INIT 5 203#define ACT_MAX 64 204#define PFCLUSTER_BEHIND 3 205#define PFCLUSTER_AHEAD 3 206 207#ifdef KERNEL 208/* 209 * Each pageable resident page falls into one of four lists: 210 * 211 * free 212 * Available for allocation now. 213 * 214 * The following are all LRU sorted: 215 * 216 * cache 217 * Almost available for allocation. Still in an 218 * object, but clean and immediately freeable at 219 * non-interrupt times. 220 * 221 * inactive 222 * Low activity, candidates for reclamation. 223 * This is the list of pages that should be 224 * paged out next. 225 * 226 * active 227 * Pages that are "active" i.e. they have been 228 * recently referenced. 229 * 230 * zero 231 * Pages that are really free and have been pre-zeroed 232 * 233 */ 234 235extern struct pglist vm_page_queue_free[PQ_L2_SIZE];/* memory free queue */ 236extern struct pglist vm_page_queue_zero[PQ_L2_SIZE];/* zeroed memory free queue */ 237extern struct pglist vm_page_queue_active; /* active memory queue */ 238extern struct pglist vm_page_queue_inactive; /* inactive memory queue */ 239extern struct pglist vm_page_queue_cache[PQ_L2_SIZE];/* cache memory queue */ 240 241extern int vm_page_zero_count; 242 243extern vm_page_t vm_page_array; /* First resident page in table */ 244extern long first_page; /* first physical page number */ 245 246 /* ... represented in vm_page_array */ 247extern long last_page; /* last physical page number */ 248 249 /* ... represented in vm_page_array */ 250 /* [INCLUSIVE] */ 251extern vm_offset_t first_phys_addr; /* physical address for first_page */ 252extern vm_offset_t last_phys_addr; /* physical address for last_page */ 253 254#define VM_PAGE_TO_PHYS(entry) ((entry)->phys_addr) 255 256#define IS_VM_PHYSADDR(pa) \ 257 ((pa) >= first_phys_addr && (pa) <= last_phys_addr) 258 259#define PHYS_TO_VM_PAGE(pa) \ 260 (&vm_page_array[atop(pa) - first_page ]) 261 262/* 263 * Functions implemented as macros 264 */ 265 266#define PAGE_ASSERT_WAIT(m, interruptible) { \ 267 (m)->flags |= PG_WANTED; \ 268 assert_wait((int) (m), (interruptible)); \ 269} 270 271#define PAGE_WAKEUP(m) { \ 272 (m)->flags &= ~PG_BUSY; \ 273 if (((m)->flags & PG_WANTED) && ((m)->busy == 0)) { \ 274 (m)->flags &= ~PG_WANTED; \ 275 wakeup((m)); \ 276 } \ 277} 278 279#if PAGE_SIZE == 4096 280#define VM_PAGE_BITS_ALL 0xff 281#endif 282 283#if PAGE_SIZE == 8192 284#define VM_PAGE_BITS_ALL 0xffff 285#endif 286 287#define VM_ALLOC_NORMAL 0 288#define VM_ALLOC_INTERRUPT 1 289#define VM_ALLOC_SYSTEM 2 290#define VM_ALLOC_ZERO 3 291#define VM_ALLOC_RETRY 0x80 292 293void vm_page_activate __P((vm_page_t)); 294vm_page_t vm_page_alloc __P((vm_object_t, vm_pindex_t, int)); 295vm_page_t vm_page_grab __P((vm_object_t, vm_pindex_t, int)); 296void vm_page_cache __P((register vm_page_t)); 297static __inline void vm_page_copy __P((vm_page_t, vm_page_t)); 298void vm_page_deactivate __P((vm_page_t)); 299void vm_page_free __P((vm_page_t)); 300void vm_page_free_zero __P((vm_page_t)); 301void vm_page_insert __P((vm_page_t, vm_object_t, vm_pindex_t)); 302vm_page_t vm_page_lookup __P((vm_object_t, vm_pindex_t)); 303void vm_page_remove __P((vm_page_t)); 304void vm_page_rename __P((vm_page_t, vm_object_t, vm_pindex_t)); 305vm_offset_t vm_page_startup __P((vm_offset_t, vm_offset_t, vm_offset_t)); 306void vm_page_unwire __P((vm_page_t)); 307void vm_page_wire __P((vm_page_t)); 308void vm_page_unqueue __P((vm_page_t)); 309void vm_page_unqueue_nowakeup __P((vm_page_t)); 310void vm_page_set_validclean __P((vm_page_t, int, int)); 311void vm_page_set_invalid __P((vm_page_t, int, int)); 312static __inline boolean_t vm_page_zero_fill __P((vm_page_t)); 313int vm_page_is_valid __P((vm_page_t, int, int)); 314void vm_page_test_dirty __P((vm_page_t)); 315int vm_page_bits __P((int, int)); 316vm_page_t vm_page_list_find __P((int, int)); 317int vm_page_queue_index __P((vm_offset_t, int)); 318vm_page_t vm_page_select __P((vm_object_t, vm_pindex_t, int)); 319int vm_page_sleep(vm_page_t m, char *msg, char *busy); 320 321/* 322 * Keep page from being freed by the page daemon 323 * much of the same effect as wiring, except much lower 324 * overhead and should be used only for *very* temporary 325 * holding ("wiring"). 326 */ 327static __inline void 328vm_page_hold(vm_page_t mem) 329{ 330 mem->hold_count++; 331} 332 333#ifdef DIAGNOSTIC 334#include <sys/systm.h> /* make GCC shut up */ 335#endif 336 337static __inline void 338vm_page_unhold(vm_page_t mem) 339{ 340#ifdef DIAGNOSTIC 341 if (--mem->hold_count < 0) 342 panic("vm_page_unhold: hold count < 0!!!"); 343#else 344 --mem->hold_count; 345#endif 346} 347 348static __inline void 349vm_page_protect(vm_page_t mem, int prot) 350{ 351 if (prot == VM_PROT_NONE) { 352 if (mem->flags & (PG_WRITEABLE|PG_MAPPED)) { 353 pmap_page_protect(VM_PAGE_TO_PHYS(mem), prot); 354 mem->flags &= ~(PG_WRITEABLE|PG_MAPPED); 355 } 356 } else if ((prot == VM_PROT_READ) && (mem->flags & PG_WRITEABLE)) { 357 pmap_page_protect(VM_PAGE_TO_PHYS(mem), prot); 358 mem->flags &= ~PG_WRITEABLE; 359 } 360} 361 362/* 363 * vm_page_zero_fill: 364 * 365 * Zero-fill the specified page. 366 * Written as a standard pagein routine, to 367 * be used by the zero-fill object. 368 */ 369static __inline boolean_t 370vm_page_zero_fill(m) 371 vm_page_t m; 372{ 373 pmap_zero_page(VM_PAGE_TO_PHYS(m)); 374 return (TRUE); 375} 376 377/* 378 * vm_page_copy: 379 * 380 * Copy one page to another 381 */ 382static __inline void 383vm_page_copy(src_m, dest_m) 384 vm_page_t src_m; 385 vm_page_t dest_m; 386{ 387 pmap_copy_page(VM_PAGE_TO_PHYS(src_m), VM_PAGE_TO_PHYS(dest_m)); 388 dest_m->valid = VM_PAGE_BITS_ALL; 389} 390 391#endif /* KERNEL */ 392#endif /* !_VM_PAGE_ */ 393