The Design and Implementation of the FreeBSD Operating System, Second Edition
Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]

FreeBSD/Linux Kernel Cross Reference
sys/vm/vm_kern.c

Version: -  FREEBSD  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-2  -  FREEBSD-11-1  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-4  -  FREEBSD-10-3  -  FREEBSD-10-2  -  FREEBSD-10-1  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-3  -  FREEBSD-9-2  -  FREEBSD-9-1  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-4  -  FREEBSD-8-3  -  FREEBSD-8-2  -  FREEBSD-8-1  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-4  -  FREEBSD-7-3  -  FREEBSD-7-2  -  FREEBSD-7-1  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-4  -  FREEBSD-6-3  -  FREEBSD-6-2  -  FREEBSD-6-1  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-5  -  FREEBSD-5-4  -  FREEBSD-5-3  -  FREEBSD-5-2  -  FREEBSD-5-1  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  linux-2.6  -  linux-2.4.22  -  MK83  -  MK84  -  PLAN9  -  DFBSD  -  NETBSD  -  NETBSD5  -  NETBSD4  -  NETBSD3  -  NETBSD20  -  OPENBSD  -  xnu-517  -  xnu-792  -  xnu-792.6.70  -  xnu-1228  -  xnu-1456.1.26  -  xnu-1699.24.8  -  xnu-2050.18.24  -  OPENSOLARIS  -  minix-3-1-1 
SearchContext: -  none  -  3  -  10 

    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_kern.c     8.3 (Berkeley) 1/12/94
   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: releng/5.0/sys/vm/vm_kern.c 102382 2002-08-25 00:22:31Z alc $
   65  */
   66 
   67 /*
   68  *      Kernel memory management.
   69  */
   70 
   71 #include <sys/param.h>
   72 #include <sys/systm.h>
   73 #include <sys/kernel.h>         /* for ticks and hz */
   74 #include <sys/lock.h>
   75 #include <sys/mutex.h>
   76 #include <sys/proc.h>
   77 #include <sys/malloc.h>
   78 
   79 #include <vm/vm.h>
   80 #include <vm/vm_param.h>
   81 #include <vm/pmap.h>
   82 #include <vm/vm_map.h>
   83 #include <vm/vm_object.h>
   84 #include <vm/vm_page.h>
   85 #include <vm/vm_pageout.h>
   86 #include <vm/vm_extern.h>
   87 
   88 vm_map_t kernel_map=0;
   89 vm_map_t kmem_map=0;
   90 vm_map_t exec_map=0;
   91 vm_map_t clean_map=0;
   92 vm_map_t buffer_map=0;
   93 
   94 /*
   95  *      kmem_alloc_pageable:
   96  *
   97  *      Allocate pageable memory to the kernel's address map.
   98  *      "map" must be kernel_map or a submap of kernel_map.
   99  */
  100 vm_offset_t
  101 kmem_alloc_pageable(map, size)
  102         vm_map_t map;
  103         vm_size_t size;
  104 {
  105         vm_offset_t addr;
  106         int result;
  107 
  108         size = round_page(size);
  109         addr = vm_map_min(map);
  110         result = vm_map_find(map, NULL, 0,
  111             &addr, size, TRUE, VM_PROT_ALL, VM_PROT_ALL, 0);
  112         if (result != KERN_SUCCESS) {
  113                 return (0);
  114         }
  115         return (addr);
  116 }
  117 
  118 /*
  119  *      kmem_alloc_nofault:
  120  *
  121  *      Same as kmem_alloc_pageable, except that it create a nofault entry.
  122  */
  123 vm_offset_t
  124 kmem_alloc_nofault(map, size)
  125         vm_map_t map;
  126         vm_size_t size;
  127 {
  128         vm_offset_t addr;
  129         int result;
  130 
  131         size = round_page(size);
  132         addr = vm_map_min(map);
  133         result = vm_map_find(map, NULL, 0,
  134             &addr, size, TRUE, VM_PROT_ALL, VM_PROT_ALL, MAP_NOFAULT);
  135         if (result != KERN_SUCCESS) {
  136                 return (0);
  137         }
  138         return (addr);
  139 }
  140 
  141 /*
  142  *      Allocate wired-down memory in the kernel's address map
  143  *      or a submap.
  144  */
  145 vm_offset_t
  146 kmem_alloc(map, size)
  147         vm_map_t map;
  148         vm_size_t size;
  149 {
  150         vm_offset_t addr;
  151         vm_offset_t offset;
  152         vm_offset_t i;
  153 
  154         GIANT_REQUIRED;
  155 
  156         size = round_page(size);
  157 
  158         /*
  159          * Use the kernel object for wired-down kernel pages. Assume that no
  160          * region of the kernel object is referenced more than once.
  161          */
  162 
  163         /*
  164          * Locate sufficient space in the map.  This will give us the final
  165          * virtual address for the new memory, and thus will tell us the
  166          * offset within the kernel map.
  167          */
  168         vm_map_lock(map);
  169         if (vm_map_findspace(map, vm_map_min(map), size, &addr)) {
  170                 vm_map_unlock(map);
  171                 return (0);
  172         }
  173         offset = addr - VM_MIN_KERNEL_ADDRESS;
  174         vm_object_reference(kernel_object);
  175         vm_map_insert(map, kernel_object, offset, addr, addr + size,
  176                 VM_PROT_ALL, VM_PROT_ALL, 0);
  177         vm_map_unlock(map);
  178 
  179         /*
  180          * Guarantee that there are pages already in this object before
  181          * calling vm_map_pageable.  This is to prevent the following
  182          * scenario:
  183          *
  184          * 1) Threads have swapped out, so that there is a pager for the
  185          * kernel_object. 2) The kmsg zone is empty, and so we are
  186          * kmem_allocing a new page for it. 3) vm_map_pageable calls vm_fault;
  187          * there is no page, but there is a pager, so we call
  188          * pager_data_request.  But the kmsg zone is empty, so we must
  189          * kmem_alloc. 4) goto 1 5) Even if the kmsg zone is not empty: when
  190          * we get the data back from the pager, it will be (very stale)
  191          * non-zero data.  kmem_alloc is defined to return zero-filled memory.
  192          *
  193          * We're intentionally not activating the pages we allocate to prevent a
  194          * race with page-out.  vm_map_pageable will wire the pages.
  195          */
  196         for (i = 0; i < size; i += PAGE_SIZE) {
  197                 vm_page_t mem;
  198 
  199                 mem = vm_page_grab(kernel_object, OFF_TO_IDX(offset + i),
  200                                 VM_ALLOC_ZERO | VM_ALLOC_RETRY);
  201                 if ((mem->flags & PG_ZERO) == 0)
  202                         pmap_zero_page(mem);
  203                 mem->valid = VM_PAGE_BITS_ALL;
  204                 vm_page_flag_clear(mem, PG_ZERO);
  205                 vm_page_wakeup(mem);
  206         }
  207 
  208         /*
  209          * And finally, mark the data as non-pageable.
  210          */
  211         (void) vm_map_wire(map, addr, addr + size, FALSE);
  212 
  213         return (addr);
  214 }
  215 
  216 /*
  217  *      kmem_free:
  218  *
  219  *      Release a region of kernel virtual memory allocated
  220  *      with kmem_alloc, and return the physical pages
  221  *      associated with that region.
  222  *
  223  *      This routine may not block on kernel maps.
  224  */
  225 void
  226 kmem_free(map, addr, size)
  227         vm_map_t map;
  228         vm_offset_t addr;
  229         vm_size_t size;
  230 {
  231 
  232         (void) vm_map_remove(map, trunc_page(addr), round_page(addr + size));
  233 }
  234 
  235 /*
  236  *      kmem_suballoc:
  237  *
  238  *      Allocates a map to manage a subrange
  239  *      of the kernel virtual address space.
  240  *
  241  *      Arguments are as follows:
  242  *
  243  *      parent          Map to take range from
  244  *      min, max        Returned endpoints of map
  245  *      size            Size of range to find
  246  */
  247 vm_map_t
  248 kmem_suballoc(parent, min, max, size)
  249         vm_map_t parent;
  250         vm_offset_t *min, *max;
  251         vm_size_t size;
  252 {
  253         int ret;
  254         vm_map_t result;
  255 
  256         GIANT_REQUIRED;
  257 
  258         size = round_page(size);
  259 
  260         *min = (vm_offset_t) vm_map_min(parent);
  261         ret = vm_map_find(parent, NULL, (vm_offset_t) 0,
  262             min, size, TRUE, VM_PROT_ALL, VM_PROT_ALL, 0);
  263         if (ret != KERN_SUCCESS) {
  264                 printf("kmem_suballoc: bad status return of %d.\n", ret);
  265                 panic("kmem_suballoc");
  266         }
  267         *max = *min + size;
  268         result = vm_map_create(vm_map_pmap(parent), *min, *max);
  269         if (result == NULL)
  270                 panic("kmem_suballoc: cannot create submap");
  271         if (vm_map_submap(parent, *min, *max, result) != KERN_SUCCESS)
  272                 panic("kmem_suballoc: unable to change range to submap");
  273         return (result);
  274 }
  275 
  276 /*
  277  *      kmem_malloc:
  278  *
  279  *      Allocate wired-down memory in the kernel's address map for the higher
  280  *      level kernel memory allocator (kern/kern_malloc.c).  We cannot use
  281  *      kmem_alloc() because we may need to allocate memory at interrupt
  282  *      level where we cannot block (canwait == FALSE).
  283  *
  284  *      This routine has its own private kernel submap (kmem_map) and object
  285  *      (kmem_object).  This, combined with the fact that only malloc uses
  286  *      this routine, ensures that we will never block in map or object waits.
  287  *
  288  *      Note that this still only works in a uni-processor environment and
  289  *      when called at splhigh().
  290  *
  291  *      We don't worry about expanding the map (adding entries) since entries
  292  *      for wired maps are statically allocated.
  293  *
  294  *      NOTE:  This routine is not supposed to block if M_NOWAIT is set, but
  295  *      I have not verified that it actually does not block.
  296  *
  297  *      `map' is ONLY allowed to be kmem_map or one of the mbuf submaps to
  298  *      which we never free.
  299  */
  300 vm_offset_t
  301 kmem_malloc(map, size, flags)
  302         vm_map_t map;
  303         vm_size_t size;
  304         int flags;
  305 {
  306         vm_offset_t offset, i;
  307         vm_map_entry_t entry;
  308         vm_offset_t addr;
  309         vm_page_t m;
  310         int pflags;
  311 
  312         GIANT_REQUIRED;
  313 
  314         size = round_page(size);
  315         addr = vm_map_min(map);
  316 
  317         /*
  318          * Locate sufficient space in the map.  This will give us the final
  319          * virtual address for the new memory, and thus will tell us the
  320          * offset within the kernel map.
  321          */
  322         vm_map_lock(map);
  323         if (vm_map_findspace(map, vm_map_min(map), size, &addr)) {
  324                 vm_map_unlock(map);
  325                 if (map != kmem_map) {
  326                         static int last_report; /* when we did it (in ticks) */
  327                         if (ticks < last_report ||
  328                             (ticks - last_report) >= hz) {
  329                                 last_report = ticks;
  330                                 printf("Out of mbuf address space!\n");
  331                                 printf("Consider increasing NMBCLUSTERS\n");
  332                         }
  333                         goto bad;
  334                 }
  335                 if ((flags & M_NOWAIT) == 0)
  336                         panic("kmem_malloc(%ld): kmem_map too small: %ld total allocated",
  337                                 (long)size, (long)map->size);
  338                 goto bad;
  339         }
  340         offset = addr - VM_MIN_KERNEL_ADDRESS;
  341         vm_object_reference(kmem_object);
  342         vm_map_insert(map, kmem_object, offset, addr, addr + size,
  343                 VM_PROT_ALL, VM_PROT_ALL, 0);
  344 
  345         /*
  346          * Note: if M_NOWAIT specified alone, allocate from 
  347          * interrupt-safe queues only (just the free list).  If 
  348          * M_USE_RESERVE is also specified, we can also
  349          * allocate from the cache.  Neither of the latter two
  350          * flags may be specified from an interrupt since interrupts
  351          * are not allowed to mess with the cache queue.
  352          */
  353 
  354         if ((flags & (M_NOWAIT|M_USE_RESERVE)) == M_NOWAIT)
  355                 pflags = VM_ALLOC_INTERRUPT;
  356         else
  357                 pflags = VM_ALLOC_SYSTEM;
  358 
  359         if (flags & M_ZERO)
  360                 pflags |= VM_ALLOC_ZERO;
  361 
  362 
  363         for (i = 0; i < size; i += PAGE_SIZE) {
  364 retry:
  365                 m = vm_page_alloc(kmem_object, OFF_TO_IDX(offset + i), pflags);
  366 
  367                 /*
  368                  * Ran out of space, free everything up and return. Don't need
  369                  * to lock page queues here as we know that the pages we got
  370                  * aren't on any queues.
  371                  */
  372                 if (m == NULL) {
  373                         if ((flags & M_NOWAIT) == 0) {
  374                                 vm_map_unlock(map);
  375                                 VM_WAIT;
  376                                 vm_map_lock(map);
  377                                 goto retry;
  378                         }
  379                         /* 
  380                          * Free the pages before removing the map entry.
  381                          * They are already marked busy.  Calling
  382                          * vm_map_delete before the pages has been freed or
  383                          * unbusied will cause a deadlock.
  384                          */
  385                         while (i != 0) {
  386                                 i -= PAGE_SIZE;
  387                                 m = vm_page_lookup(kmem_object,
  388                                                    OFF_TO_IDX(offset + i));
  389                                 vm_page_lock_queues();
  390                                 vm_page_free(m);
  391                                 vm_page_unlock_queues();
  392                         }
  393                         vm_map_delete(map, addr, addr + size);
  394                         vm_map_unlock(map);
  395                         goto bad;
  396                 }
  397                 if (flags & M_ZERO && (m->flags & PG_ZERO) == 0)
  398                         pmap_zero_page(m);
  399                 vm_page_flag_clear(m, PG_ZERO);
  400                 m->valid = VM_PAGE_BITS_ALL;
  401         }
  402 
  403         /*
  404          * Mark map entry as non-pageable. Assert: vm_map_insert() will never
  405          * be able to extend the previous entry so there will be a new entry
  406          * exactly corresponding to this address range and it will have
  407          * wired_count == 0.
  408          */
  409         if (!vm_map_lookup_entry(map, addr, &entry) ||
  410             entry->start != addr || entry->end != addr + size ||
  411             entry->wired_count != 0)
  412                 panic("kmem_malloc: entry not found or misaligned");
  413         entry->wired_count = 1;
  414 
  415         vm_map_simplify_entry(map, entry);
  416 
  417         /*
  418          * Loop thru pages, entering them in the pmap. (We cannot add them to
  419          * the wired count without wrapping the vm_page_queue_lock in
  420          * splimp...)
  421          */
  422         for (i = 0; i < size; i += PAGE_SIZE) {
  423                 m = vm_page_lookup(kmem_object, OFF_TO_IDX(offset + i));
  424                 vm_page_lock_queues();
  425                 vm_page_wire(m);
  426                 vm_page_wakeup(m);
  427                 vm_page_unlock_queues();
  428                 /*
  429                  * Because this is kernel_pmap, this call will not block.
  430                  */
  431                 pmap_enter(kernel_pmap, addr + i, m, VM_PROT_ALL, 1);
  432                 vm_page_flag_set(m, PG_WRITEABLE | PG_REFERENCED);
  433         }
  434         vm_map_unlock(map);
  435 
  436         return (addr);
  437 
  438 bad:    
  439         return (0);
  440 }
  441 
  442 /*
  443  *      kmem_alloc_wait:
  444  *
  445  *      Allocates pageable memory from a sub-map of the kernel.  If the submap
  446  *      has no room, the caller sleeps waiting for more memory in the submap.
  447  *
  448  *      This routine may block.
  449  */
  450 vm_offset_t
  451 kmem_alloc_wait(map, size)
  452         vm_map_t map;
  453         vm_size_t size;
  454 {
  455         vm_offset_t addr;
  456 
  457         size = round_page(size);
  458 
  459         for (;;) {
  460                 /*
  461                  * To make this work for more than one map, use the map's lock
  462                  * to lock out sleepers/wakers.
  463                  */
  464                 vm_map_lock(map);
  465                 if (vm_map_findspace(map, vm_map_min(map), size, &addr) == 0)
  466                         break;
  467                 /* no space now; see if we can ever get space */
  468                 if (vm_map_max(map) - vm_map_min(map) < size) {
  469                         vm_map_unlock(map);
  470                         return (0);
  471                 }
  472                 map->needs_wakeup = TRUE;
  473                 vm_map_unlock_and_wait(map, FALSE);
  474         }
  475         vm_map_insert(map, NULL, 0, addr, addr + size, VM_PROT_ALL, VM_PROT_ALL, 0);
  476         vm_map_unlock(map);
  477         return (addr);
  478 }
  479 
  480 /*
  481  *      kmem_free_wakeup:
  482  *
  483  *      Returns memory to a submap of the kernel, and wakes up any processes
  484  *      waiting for memory in that map.
  485  */
  486 void
  487 kmem_free_wakeup(map, addr, size)
  488         vm_map_t map;
  489         vm_offset_t addr;
  490         vm_size_t size;
  491 {
  492 
  493         vm_map_lock(map);
  494         (void) vm_map_delete(map, trunc_page(addr), round_page(addr + size));
  495         if (map->needs_wakeup) {
  496                 map->needs_wakeup = FALSE;
  497                 vm_map_wakeup(map);
  498         }
  499         vm_map_unlock(map);
  500 }
  501 
  502 /*
  503  *      kmem_init:
  504  *
  505  *      Create the kernel map; insert a mapping covering kernel text, 
  506  *      data, bss, and all space allocated thus far (`boostrap' data).  The 
  507  *      new map will thus map the range between VM_MIN_KERNEL_ADDRESS and 
  508  *      `start' as allocated, and the range between `start' and `end' as free.
  509  */
  510 void
  511 kmem_init(start, end)
  512         vm_offset_t start, end;
  513 {
  514         vm_map_t m;
  515 
  516         m = vm_map_create(kernel_pmap, VM_MIN_KERNEL_ADDRESS, end);
  517         vm_map_lock(m);
  518         /* N.B.: cannot use kgdb to debug, starting with this assignment ... */
  519         kernel_map = m;
  520         kernel_map->system_map = 1;
  521         (void) vm_map_insert(m, NULL, (vm_offset_t) 0,
  522             VM_MIN_KERNEL_ADDRESS, start, VM_PROT_ALL, VM_PROT_ALL, 0);
  523         /* ... and ending with the completion of the above `insert' */
  524         vm_map_unlock(m);
  525 }

Cache object: 6c74fd7eb5e3954b198fc930cfd92852


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]


This page is part of the FreeBSD/Linux Linux Kernel Cross-Reference, and was automatically generated using a modified version of the LXR engine.