1 /*
2 * linux/mm/swapfile.c
3 *
4 * Copyright (C) 1991, 1992, 1993, 1994 Linus Torvalds
5 * Swap reorganised 29.12.95, Stephen Tweedie
6 */
7
8 #include <linux/malloc.h>
9 #include <linux/smp_lock.h>
10 #include <linux/kernel_stat.h>
11 #include <linux/swap.h>
12 #include <linux/swapctl.h>
13 #include <linux/blkdev.h> /* for blk_size */
14 #include <linux/vmalloc.h>
15 #include <linux/pagemap.h>
16 #include <linux/shm.h>
17
18 #include <asm/pgtable.h>
19
20 spinlock_t swaplock = SPIN_LOCK_UNLOCKED;
21 unsigned int nr_swapfiles;
22
23 struct swap_list_t swap_list = {-1, -1};
24
25 struct swap_info_struct swap_info[MAX_SWAPFILES];
26
27 #define SWAPFILE_CLUSTER 256
28
29 static inline int scan_swap_map(struct swap_info_struct *si, unsigned short count)
30 {
31 unsigned long offset;
32 /*
33 * We try to cluster swap pages by allocating them
34 * sequentially in swap. Once we've allocated
35 * SWAPFILE_CLUSTER pages this way, however, we resort to
36 * first-free allocation, starting a new cluster. This
37 * prevents us from scattering swap pages all over the entire
38 * swap partition, so that we reduce overall disk seek times
39 * between swap pages. -- sct */
40 if (si->cluster_nr) {
41 while (si->cluster_next <= si->highest_bit) {
42 offset = si->cluster_next++;
43 if (si->swap_map[offset])
44 continue;
45 si->cluster_nr--;
46 goto got_page;
47 }
48 }
49 si->cluster_nr = SWAPFILE_CLUSTER;
50
51 /* try to find an empty (even not aligned) cluster. */
52 offset = si->lowest_bit;
53 check_next_cluster:
54 if (offset+SWAPFILE_CLUSTER-1 <= si->highest_bit)
55 {
56 int nr;
57 for (nr = offset; nr < offset+SWAPFILE_CLUSTER; nr++)
58 if (si->swap_map[nr])
59 {
60 offset = nr+1;
61 goto check_next_cluster;
62 }
63 /* We found a completly empty cluster, so start
64 * using it.
65 */
66 goto got_page;
67 }
68 /* No luck, so now go finegrined as usual. -Andrea */
69 for (offset = si->lowest_bit; offset <= si->highest_bit ; offset++) {
70 if (si->swap_map[offset])
71 continue;
72 got_page:
73 if (offset == si->lowest_bit)
74 si->lowest_bit++;
75 if (offset == si->highest_bit)
76 si->highest_bit--;
77 si->swap_map[offset] = count;
78 nr_swap_pages--;
79 si->cluster_next = offset+1;
80 return offset;
81 }
82 return 0;
83 }
84
85 swp_entry_t __get_swap_page(unsigned short count)
86 {
87 struct swap_info_struct * p;
88 unsigned long offset;
89 swp_entry_t entry;
90 int type, wrapped = 0;
91
92 entry.val = 0; /* Out of memory */
93 if (count >= SWAP_MAP_MAX)
94 goto bad_count;
95 swap_list_lock();
96 type = swap_list.next;
97 if (type < 0)
98 goto out;
99 if (nr_swap_pages == 0)
100 goto out;
101
102 while (1) {
103 p = &swap_info[type];
104 if ((p->flags & SWP_WRITEOK) == SWP_WRITEOK) {
105 swap_device_lock(p);
106 offset = scan_swap_map(p, count);
107 swap_device_unlock(p);
108 if (offset) {
109 entry = SWP_ENTRY(type,offset);
110 type = swap_info[type].next;
111 if (type < 0 ||
112 p->prio != swap_info[type].prio) {
113 swap_list.next = swap_list.head;
114 } else {
115 swap_list.next = type;
116 }
117 goto out;
118 }
119 }
120 type = p->next;
121 if (!wrapped) {
122 if (type < 0 || p->prio != swap_info[type].prio) {
123 type = swap_list.head;
124 wrapped = 1;
125 }
126 } else
127 if (type < 0)
128 goto out; /* out of swap space */
129 }
130 out:
131 swap_list_unlock();
132 return entry;
133
134 bad_count:
135 printk(KERN_ERR "get_swap_page: bad count %hd from %p\n",
136 count, __builtin_return_address(0));
137 goto out;
138 }
139
140
141 /*
142 * Caller has made sure that the swapdevice corresponding to entry
143 * is still around or has not been recycled.
144 */
145 void __swap_free(swp_entry_t entry, unsigned short count)
146 {
147 struct swap_info_struct * p;
148 unsigned long offset, type;
149
150 if (!entry.val)
151 goto out;
152
153 type = SWP_TYPE(entry);
154 if (type >= nr_swapfiles)
155 goto bad_nofile;
156 p = & swap_info[type];
157 if (!(p->flags & SWP_USED))
158 goto bad_device;
159 offset = SWP_OFFSET(entry);
160 if (offset >= p->max)
161 goto bad_offset;
162 if (!p->swap_map[offset])
163 goto bad_free;
164 swap_list_lock();
165 if (p->prio > swap_info[swap_list.next].prio)
166 swap_list.next = type;
167 swap_device_lock(p);
168 if (p->swap_map[offset] < SWAP_MAP_MAX) {
169 if (p->swap_map[offset] < count)
170 goto bad_count;
171 if (!(p->swap_map[offset] -= count)) {
172 if (offset < p->lowest_bit)
173 p->lowest_bit = offset;
174 if (offset > p->highest_bit)
175 p->highest_bit = offset;
176 nr_swap_pages++;
177 }
178 }
179 swap_device_unlock(p);
180 swap_list_unlock();
181 out:
182 return;
183
184 bad_nofile:
185 printk("swap_free: Trying to free nonexistent swap-page\n");
186 goto out;
187 bad_device:
188 printk("swap_free: Trying to free swap from unused swap-device\n");
189 goto out;
190 bad_offset:
191 printk("swap_free: offset exceeds max\n");
192 goto out;
193 bad_free:
194 printk("VM: Bad swap entry %08lx\n", entry.val);
195 goto out;
196 bad_count:
197 swap_device_unlock(p);
198 swap_list_unlock();
199 printk(KERN_ERR "VM: Bad count %hd current count %hd\n", count, p->swap_map[offset]);
200 goto out;
201 }
202
203 /*
204 * The swap entry has been read in advance, and we return 1 to indicate
205 * that the page has been used or is no longer needed.
206 *
207 * Always set the resulting pte to be nowrite (the same as COW pages
208 * after one process has exited). We don't know just how many PTEs will
209 * share this swap entry, so be cautious and let do_wp_page work out
210 * what to do if a write is requested later.
211 */
212 static inline void unuse_pte(struct vm_area_struct * vma, unsigned long address,
213 pte_t *dir, swp_entry_t entry, struct page* page)
214 {
215 pte_t pte = *dir;
216
217 if (pte_none(pte))
218 return;
219 if (pte_present(pte)) {
220 /* If this entry is swap-cached, then page must already
221 hold the right address for any copies in physical
222 memory */
223 if (pte_page(pte) != page)
224 return;
225 /* We will be removing the swap cache in a moment, so... */
226 ptep_mkdirty(dir);
227 return;
228 }
229 if (pte_to_swp_entry(pte).val != entry.val)
230 return;
231 set_pte(dir, pte_mkdirty(mk_pte(page, vma->vm_page_prot)));
232 swap_free(entry);
233 get_page(page);
234 ++vma->vm_mm->rss;
235 }
236
237 static inline void unuse_pmd(struct vm_area_struct * vma, pmd_t *dir,
238 unsigned long address, unsigned long size, unsigned long offset,
239 swp_entry_t entry, struct page* page)
240 {
241 pte_t * pte;
242 unsigned long end;
243
244 if (pmd_none(*dir))
245 return;
246 if (pmd_bad(*dir)) {
247 pmd_ERROR(*dir);
248 pmd_clear(dir);
249 return;
250 }
251 pte = pte_offset(dir, address);
252 offset += address & PMD_MASK;
253 address &= ~PMD_MASK;
254 end = address + size;
255 if (end > PMD_SIZE)
256 end = PMD_SIZE;
257 do {
258 unuse_pte(vma, offset+address-vma->vm_start, pte, entry, page);
259 address += PAGE_SIZE;
260 pte++;
261 } while (address && (address < end));
262 }
263
264 static inline void unuse_pgd(struct vm_area_struct * vma, pgd_t *dir,
265 unsigned long address, unsigned long size,
266 swp_entry_t entry, struct page* page)
267 {
268 pmd_t * pmd;
269 unsigned long offset, end;
270
271 if (pgd_none(*dir))
272 return;
273 if (pgd_bad(*dir)) {
274 pgd_ERROR(*dir);
275 pgd_clear(dir);
276 return;
277 }
278 pmd = pmd_offset(dir, address);
279 offset = address & PGDIR_MASK;
280 address &= ~PGDIR_MASK;
281 end = address + size;
282 if (end > PGDIR_SIZE)
283 end = PGDIR_SIZE;
284 if (address >= end)
285 BUG();
286 do {
287 unuse_pmd(vma, pmd, address, end - address, offset, entry,
288 page);
289 address = (address + PMD_SIZE) & PMD_MASK;
290 pmd++;
291 } while (address && (address < end));
292 }
293
294 static void unuse_vma(struct vm_area_struct * vma, pgd_t *pgdir,
295 swp_entry_t entry, struct page* page)
296 {
297 unsigned long start = vma->vm_start, end = vma->vm_end;
298
299 if (start >= end)
300 BUG();
301 do {
302 unuse_pgd(vma, pgdir, start, end - start, entry, page);
303 start = (start + PGDIR_SIZE) & PGDIR_MASK;
304 pgdir++;
305 } while (start && (start < end));
306 }
307
308 static void unuse_process(struct mm_struct * mm,
309 swp_entry_t entry, struct page* page)
310 {
311 struct vm_area_struct* vma;
312
313 /*
314 * Go through process' page directory.
315 */
316 if (!mm)
317 return;
318 spin_lock(&mm->page_table_lock);
319 for (vma = mm->mmap; vma; vma = vma->vm_next) {
320 pgd_t * pgd = pgd_offset(mm, vma->vm_start);
321 unuse_vma(vma, pgd, entry, page);
322 }
323 spin_unlock(&mm->page_table_lock);
324 return;
325 }
326
327 /*
328 * We completely avoid races by reading each swap page in advance,
329 * and then search for the process using it. All the necessary
330 * page table adjustments can then be made atomically.
331 */
332 static int try_to_unuse(unsigned int type)
333 {
334 struct swap_info_struct * si = &swap_info[type];
335 struct task_struct *p;
336 struct page *page;
337 swp_entry_t entry;
338 int i;
339
340 while (1) {
341 /*
342 * Find a swap page in use and read it in.
343 */
344 swap_device_lock(si);
345 for (i = 1; i < si->max ; i++) {
346 if (si->swap_map[i] > 0 && si->swap_map[i] != SWAP_MAP_BAD) {
347 /*
348 * Prevent swaphandle from being completely
349 * unused by swap_free while we are trying
350 * to read in the page - this prevents warning
351 * messages from rw_swap_page_base.
352 */
353 if (si->swap_map[i] != SWAP_MAP_MAX)
354 si->swap_map[i]++;
355 swap_device_unlock(si);
356 goto found_entry;
357 }
358 }
359 swap_device_unlock(si);
360 break;
361
362 found_entry:
363 entry = SWP_ENTRY(type, i);
364
365 /* Get a page for the entry, using the existing swap
366 cache page if there is one. Otherwise, get a clean
367 page and read the swap into it. */
368 page = read_swap_cache(entry);
369 if (!page) {
370 swap_free(entry);
371 return -ENOMEM;
372 }
373 if (PageSwapCache(page))
374 delete_from_swap_cache(page);
375 read_lock(&tasklist_lock);
376 for_each_task(p)
377 unuse_process(p->mm, entry, page);
378 read_unlock(&tasklist_lock);
379 shmem_unuse(entry, page);
380 /* Now get rid of the extra reference to the temporary
381 page we've been using. */
382 page_cache_release(page);
383 /*
384 * Check for and clear any overflowed swap map counts.
385 */
386 swap_free(entry);
387 swap_list_lock();
388 swap_device_lock(si);
389 if (si->swap_map[i] > 0) {
390 if (si->swap_map[i] != SWAP_MAP_MAX)
391 printk("VM: Undead swap entry %08lx\n",
392 entry.val);
393 nr_swap_pages++;
394 si->swap_map[i] = 0;
395 }
396 swap_device_unlock(si);
397 swap_list_unlock();
398 }
399 return 0;
400 }
401
402 asmlinkage long sys_swapoff(const char * specialfile)
403 {
404 struct swap_info_struct * p = NULL;
405 struct nameidata nd;
406 int i, type, prev;
407 int err;
408
409 if (!capable(CAP_SYS_ADMIN))
410 return -EPERM;
411
412 err = user_path_walk(specialfile, &nd);
413 if (err)
414 goto out;
415
416 lock_kernel();
417 prev = -1;
418 swap_list_lock();
419 for (type = swap_list.head; type >= 0; type = swap_info[type].next) {
420 p = swap_info + type;
421 if ((p->flags & SWP_WRITEOK) == SWP_WRITEOK) {
422 if (p->swap_file) {
423 if (p->swap_file == nd.dentry)
424 break;
425 } else {
426 if (S_ISBLK(nd.dentry->d_inode->i_mode)
427 && (p->swap_device == nd.dentry->d_inode->i_rdev))
428 break;
429 }
430 }
431 prev = type;
432 }
433 err = -EINVAL;
434 if (type < 0) {
435 swap_list_unlock();
436 goto out_dput;
437 }
438
439 if (prev < 0) {
440 swap_list.head = p->next;
441 } else {
442 swap_info[prev].next = p->next;
443 }
444 if (type == swap_list.next) {
445 /* just pick something that's safe... */
446 swap_list.next = swap_list.head;
447 }
448 nr_swap_pages -= p->pages;
449 swap_list_unlock();
450 p->flags = SWP_USED;
451 err = try_to_unuse(type);
452 if (err) {
453 /* re-insert swap space back into swap_list */
454 swap_list_lock();
455 for (prev = -1, i = swap_list.head; i >= 0; prev = i, i = swap_info[i].next)
456 if (p->prio >= swap_info[i].prio)
457 break;
458 p->next = i;
459 if (prev < 0)
460 swap_list.head = swap_list.next = p - swap_info;
461 else
462 swap_info[prev].next = p - swap_info;
463 nr_swap_pages += p->pages;
464 swap_list_unlock();
465 p->flags = SWP_WRITEOK;
466 goto out_dput;
467 }
468 if (p->swap_device)
469 blkdev_put(nd.dentry->d_inode->i_bdev, BDEV_SWAP);
470 path_release(&nd);
471
472 nd.dentry = p->swap_file;
473 p->swap_file = NULL;
474 nd.mnt = p->swap_vfsmnt;
475 p->swap_vfsmnt = NULL;
476 p->swap_device = 0;
477 vfree(p->swap_map);
478 p->swap_map = NULL;
479 p->flags = 0;
480 err = 0;
481
482 out_dput:
483 unlock_kernel();
484 path_release(&nd);
485 out:
486 return err;
487 }
488
489 int get_swaparea_info(char *buf)
490 {
491 char * page = (char *) __get_free_page(GFP_KERNEL);
492 struct swap_info_struct *ptr = swap_info;
493 int i, j, len = 0, usedswap;
494
495 if (!page)
496 return -ENOMEM;
497
498 len += sprintf(buf, "Filename\t\t\tType\t\tSize\tUsed\tPriority\n");
499 for (i = 0 ; i < nr_swapfiles ; i++, ptr++) {
500 if (ptr->flags & SWP_USED) {
501 char * path = d_path(ptr->swap_file, ptr->swap_vfsmnt,
502 page, PAGE_SIZE);
503
504 len += sprintf(buf + len, "%-31s ", path);
505
506 if (!ptr->swap_device)
507 len += sprintf(buf + len, "file\t\t");
508 else
509 len += sprintf(buf + len, "partition\t");
510
511 usedswap = 0;
512 for (j = 0; j < ptr->max; ++j)
513 switch (ptr->swap_map[j]) {
514 case SWAP_MAP_BAD:
515 case 0:
516 continue;
517 default:
518 usedswap++;
519 }
520 len += sprintf(buf + len, "%d\t%d\t%d\n", ptr->pages << (PAGE_SHIFT - 10),
521 usedswap << (PAGE_SHIFT - 10), ptr->prio);
522 }
523 }
524 free_page((unsigned long) page);
525 return len;
526 }
527
528 int is_swap_partition(kdev_t dev) {
529 struct swap_info_struct *ptr = swap_info;
530 int i;
531
532 for (i = 0 ; i < nr_swapfiles ; i++, ptr++) {
533 if (ptr->flags & SWP_USED)
534 if (ptr->swap_device == dev)
535 return 1;
536 }
537 return 0;
538 }
539
540 /*
541 * Written 01/25/92 by Simmule Turner, heavily changed by Linus.
542 *
543 * The swapon system call
544 */
545 asmlinkage long sys_swapon(const char * specialfile, int swap_flags)
546 {
547 struct swap_info_struct * p;
548 struct nameidata nd;
549 struct inode * swap_inode;
550 unsigned int type;
551 int i, j, prev;
552 int error;
553 static int least_priority = 0;
554 union swap_header *swap_header = 0;
555 int swap_header_version;
556 int nr_good_pages = 0;
557 unsigned long maxpages;
558 int swapfilesize;
559 struct block_device *bdev = NULL;
560
561 if (!capable(CAP_SYS_ADMIN))
562 return -EPERM;
563 lock_kernel();
564 p = swap_info;
565 for (type = 0 ; type < nr_swapfiles ; type++,p++)
566 if (!(p->flags & SWP_USED))
567 break;
568 error = -EPERM;
569 if (type >= MAX_SWAPFILES)
570 goto out;
571 if (type >= nr_swapfiles)
572 nr_swapfiles = type+1;
573 p->flags = SWP_USED;
574 p->swap_file = NULL;
575 p->swap_vfsmnt = NULL;
576 p->swap_device = 0;
577 p->swap_map = NULL;
578 p->lowest_bit = 0;
579 p->highest_bit = 0;
580 p->cluster_nr = 0;
581 p->sdev_lock = SPIN_LOCK_UNLOCKED;
582 p->max = 1;
583 p->next = -1;
584 if (swap_flags & SWAP_FLAG_PREFER) {
585 p->prio =
586 (swap_flags & SWAP_FLAG_PRIO_MASK)>>SWAP_FLAG_PRIO_SHIFT;
587 } else {
588 p->prio = --least_priority;
589 }
590 error = user_path_walk(specialfile, &nd);
591 if (error)
592 goto bad_swap_2;
593
594 p->swap_file = nd.dentry;
595 p->swap_vfsmnt = nd.mnt;
596 swap_inode = nd.dentry->d_inode;
597 error = -EINVAL;
598
599 if (S_ISBLK(swap_inode->i_mode)) {
600 kdev_t dev = swap_inode->i_rdev;
601 struct block_device_operations *bdops;
602
603 p->swap_device = dev;
604 set_blocksize(dev, PAGE_SIZE);
605
606 bdev = swap_inode->i_bdev;
607 bdops = devfs_get_ops(devfs_get_handle_from_inode(swap_inode));
608 if (bdops) bdev->bd_op = bdops;
609
610 error = blkdev_get(bdev, FMODE_READ|FMODE_WRITE, 0, BDEV_SWAP);
611 if (error)
612 goto bad_swap_2;
613 set_blocksize(dev, PAGE_SIZE);
614 error = -ENODEV;
615 if (!dev || (blk_size[MAJOR(dev)] &&
616 !blk_size[MAJOR(dev)][MINOR(dev)]))
617 goto bad_swap;
618 error = -EBUSY;
619 for (i = 0 ; i < nr_swapfiles ; i++) {
620 if (i == type)
621 continue;
622 if (dev == swap_info[i].swap_device)
623 goto bad_swap;
624 }
625 swapfilesize = 0;
626 if (blk_size[MAJOR(dev)])
627 swapfilesize = blk_size[MAJOR(dev)][MINOR(dev)]
628 >> (PAGE_SHIFT - 10);
629 } else if (S_ISREG(swap_inode->i_mode)) {
630 error = -EBUSY;
631 for (i = 0 ; i < nr_swapfiles ; i++) {
632 if (i == type || !swap_info[i].swap_file)
633 continue;
634 if (swap_inode == swap_info[i].swap_file->d_inode)
635 goto bad_swap;
636 }
637 swapfilesize = swap_inode->i_size >> PAGE_SHIFT;
638 } else
639 goto bad_swap;
640
641 swap_header = (void *) __get_free_page(GFP_USER);
642 if (!swap_header) {
643 printk("Unable to start swapping: out of memory :-)\n");
644 error = -ENOMEM;
645 goto bad_swap;
646 }
647
648 lock_page(virt_to_page(swap_header));
649 rw_swap_page_nolock(READ, SWP_ENTRY(type,0), (char *) swap_header, 1);
650
651 if (!memcmp("SWAP-SPACE",swap_header->magic.magic,10))
652 swap_header_version = 1;
653 else if (!memcmp("SWAPSPACE2",swap_header->magic.magic,10))
654 swap_header_version = 2;
655 else {
656 printk("Unable to find swap-space signature\n");
657 error = -EINVAL;
658 goto bad_swap;
659 }
660
661 switch (swap_header_version) {
662 case 1:
663 memset(((char *) swap_header)+PAGE_SIZE-10,0,10);
664 j = 0;
665 p->lowest_bit = 0;
666 p->highest_bit = 0;
667 for (i = 1 ; i < 8*PAGE_SIZE ; i++) {
668 if (test_bit(i,(char *) swap_header)) {
669 if (!p->lowest_bit)
670 p->lowest_bit = i;
671 p->highest_bit = i;
672 p->max = i+1;
673 j++;
674 }
675 }
676 nr_good_pages = j;
677 p->swap_map = vmalloc(p->max * sizeof(short));
678 if (!p->swap_map) {
679 error = -ENOMEM;
680 goto bad_swap;
681 }
682 for (i = 1 ; i < p->max ; i++) {
683 if (test_bit(i,(char *) swap_header))
684 p->swap_map[i] = 0;
685 else
686 p->swap_map[i] = SWAP_MAP_BAD;
687 }
688 break;
689
690 case 2:
691 /* Check the swap header's sub-version and the size of
692 the swap file and bad block lists */
693 if (swap_header->info.version != 1) {
694 printk(KERN_WARNING
695 "Unable to handle swap header version %d\n",
696 swap_header->info.version);
697 error = -EINVAL;
698 goto bad_swap;
699 }
700
701 p->lowest_bit = 1;
702 p->highest_bit = swap_header->info.last_page - 1;
703 p->max = swap_header->info.last_page;
704
705 maxpages = SWP_OFFSET(SWP_ENTRY(0,~0UL));
706 if (p->max >= maxpages)
707 p->max = maxpages-1;
708
709 error = -EINVAL;
710 if (swap_header->info.nr_badpages > MAX_SWAP_BADPAGES)
711 goto bad_swap;
712
713 /* OK, set up the swap map and apply the bad block list */
714 if (!(p->swap_map = vmalloc (p->max * sizeof(short)))) {
715 error = -ENOMEM;
716 goto bad_swap;
717 }
718
719 error = 0;
720 memset(p->swap_map, 0, p->max * sizeof(short));
721 for (i=0; i<swap_header->info.nr_badpages; i++) {
722 int page = swap_header->info.badpages[i];
723 if (page <= 0 || page >= swap_header->info.last_page)
724 error = -EINVAL;
725 else
726 p->swap_map[page] = SWAP_MAP_BAD;
727 }
728 nr_good_pages = swap_header->info.last_page -
729 swap_header->info.nr_badpages -
730 1 /* header page */;
731 if (error)
732 goto bad_swap;
733 }
734
735 if (swapfilesize && p->max > swapfilesize) {
736 printk(KERN_WARNING
737 "Swap area shorter than signature indicates\n");
738 error = -EINVAL;
739 goto bad_swap;
740 }
741 if (!nr_good_pages) {
742 printk(KERN_WARNING "Empty swap-file\n");
743 error = -EINVAL;
744 goto bad_swap;
745 }
746 p->swap_map[0] = SWAP_MAP_BAD;
747 p->flags = SWP_WRITEOK;
748 p->pages = nr_good_pages;
749 swap_list_lock();
750 nr_swap_pages += nr_good_pages;
751 printk(KERN_INFO "Adding Swap: %dk swap-space (priority %d)\n",
752 nr_good_pages<<(PAGE_SHIFT-10), p->prio);
753
754 /* insert swap space into swap_list: */
755 prev = -1;
756 for (i = swap_list.head; i >= 0; i = swap_info[i].next) {
757 if (p->prio >= swap_info[i].prio) {
758 break;
759 }
760 prev = i;
761 }
762 p->next = i;
763 if (prev < 0) {
764 swap_list.head = swap_list.next = p - swap_info;
765 } else {
766 swap_info[prev].next = p - swap_info;
767 }
768 swap_list_unlock();
769 error = 0;
770 goto out;
771 bad_swap:
772 if (bdev)
773 blkdev_put(bdev, BDEV_SWAP);
774 bad_swap_2:
775 if (p->swap_map)
776 vfree(p->swap_map);
777 nd.mnt = p->swap_vfsmnt;
778 nd.dentry = p->swap_file;
779 p->swap_device = 0;
780 p->swap_file = NULL;
781 p->swap_vfsmnt = NULL;
782 p->swap_map = NULL;
783 p->flags = 0;
784 if (!(swap_flags & SWAP_FLAG_PREFER))
785 ++least_priority;
786 path_release(&nd);
787 out:
788 if (swap_header)
789 free_page((long) swap_header);
790 unlock_kernel();
791 return error;
792 }
793
794 void si_swapinfo(struct sysinfo *val)
795 {
796 unsigned int i;
797 unsigned long freeswap = 0;
798 unsigned long totalswap = 0;
799
800 for (i = 0; i < nr_swapfiles; i++) {
801 unsigned int j;
802 if ((swap_info[i].flags & SWP_WRITEOK) != SWP_WRITEOK)
803 continue;
804 for (j = 0; j < swap_info[i].max; ++j) {
805 switch (swap_info[i].swap_map[j]) {
806 case SWAP_MAP_BAD:
807 continue;
808 case 0:
809 freeswap++;
810 default:
811 totalswap++;
812 }
813 }
814 }
815 val->freeswap = freeswap;
816 val->totalswap = totalswap;
817 return;
818 }
819
820 /*
821 * Verify that a swap entry is valid and increment its swap map count.
822 * Kernel_lock is held, which guarantees existance of swap device.
823 *
824 * Note: if swap_map[] reaches SWAP_MAP_MAX the entries are treated as
825 * "permanent", but will be reclaimed by the next swapoff.
826 */
827 int swap_duplicate(swp_entry_t entry)
828 {
829 struct swap_info_struct * p;
830 unsigned long offset, type;
831 int result = 0;
832
833 /* Swap entry 0 is illegal */
834 if (!entry.val)
835 goto out;
836 type = SWP_TYPE(entry);
837 if (type >= nr_swapfiles)
838 goto bad_file;
839 p = type + swap_info;
840 offset = SWP_OFFSET(entry);
841 if (offset >= p->max)
842 goto bad_offset;
843 if (!p->swap_map[offset])
844 goto bad_unused;
845 /*
846 * Entry is valid, so increment the map count.
847 */
848 swap_device_lock(p);
849 if (p->swap_map[offset] < SWAP_MAP_MAX)
850 p->swap_map[offset]++;
851 else {
852 static int overflow = 0;
853 if (overflow++ < 5)
854 printk("VM: swap entry overflow\n");
855 p->swap_map[offset] = SWAP_MAP_MAX;
856 }
857 swap_device_unlock(p);
858 result = 1;
859 out:
860 return result;
861
862 bad_file:
863 printk("Bad swap file entry %08lx\n", entry.val);
864 goto out;
865 bad_offset:
866 printk("Bad swap offset entry %08lx\n", entry.val);
867 goto out;
868 bad_unused:
869 printk("Unused swap offset entry in swap_dup %08lx\n", entry.val);
870 goto out;
871 }
872
873 /*
874 * Page lock needs to be held in all cases to prevent races with
875 * swap file deletion.
876 */
877 int swap_count(struct page *page)
878 {
879 struct swap_info_struct * p;
880 unsigned long offset, type;
881 swp_entry_t entry;
882 int retval = 0;
883
884 entry.val = page->index;
885 if (!entry.val)
886 goto bad_entry;
887 type = SWP_TYPE(entry);
888 if (type >= nr_swapfiles)
889 goto bad_file;
890 p = type + swap_info;
891 offset = SWP_OFFSET(entry);
892 if (offset >= p->max)
893 goto bad_offset;
894 if (!p->swap_map[offset])
895 goto bad_unused;
896 retval = p->swap_map[offset];
897 out:
898 return retval;
899
900 bad_entry:
901 printk(KERN_ERR "swap_count: null entry!\n");
902 goto out;
903 bad_file:
904 printk("Bad swap file entry %08lx\n", entry.val);
905 goto out;
906 bad_offset:
907 printk("Bad swap offset entry %08lx\n", entry.val);
908 goto out;
909 bad_unused:
910 printk("Unused swap offset entry in swap_count %08lx\n", entry.val);
911 goto out;
912 }
913
914 /*
915 * Kernel_lock protects against swap device deletion.
916 */
917 void get_swaphandle_info(swp_entry_t entry, unsigned long *offset,
918 kdev_t *dev, struct inode **swapf)
919 {
920 unsigned long type;
921 struct swap_info_struct *p;
922
923 type = SWP_TYPE(entry);
924 if (type >= nr_swapfiles) {
925 printk("Internal error: bad swap-device\n");
926 return;
927 }
928
929 p = &swap_info[type];
930 *offset = SWP_OFFSET(entry);
931 if (*offset >= p->max) {
932 printk("rw_swap_page: weirdness\n");
933 return;
934 }
935 if (p->swap_map && !p->swap_map[*offset]) {
936 printk("VM: Bad swap entry %08lx\n", entry.val);
937 return;
938 }
939 if (!(p->flags & SWP_USED)) {
940 printk(KERN_ERR "rw_swap_page: "
941 "Trying to swap to unused swap-device\n");
942 return;
943 }
944
945 if (p->swap_device) {
946 *dev = p->swap_device;
947 } else if (p->swap_file) {
948 *swapf = p->swap_file->d_inode;
949 } else {
950 printk(KERN_ERR "rw_swap_page: no swap file or device\n");
951 }
952 return;
953 }
954
955 /*
956 * Kernel_lock protects against swap device deletion. Grab an extra
957 * reference on the swaphandle so that it dos not become unused.
958 */
959 int valid_swaphandles(swp_entry_t entry, unsigned long *offset)
960 {
961 int ret = 0, i = 1 << page_cluster;
962 unsigned long toff;
963 struct swap_info_struct *swapdev = SWP_TYPE(entry) + swap_info;
964
965 *offset = SWP_OFFSET(entry);
966 toff = *offset = (*offset >> page_cluster) << page_cluster;
967
968 swap_device_lock(swapdev);
969 do {
970 /* Don't read-ahead past the end of the swap area */
971 if (toff >= swapdev->max)
972 break;
973 /* Don't read in bad or busy pages */
974 if (!swapdev->swap_map[toff])
975 break;
976 if (swapdev->swap_map[toff] == SWAP_MAP_BAD)
977 break;
978 swapdev->swap_map[toff]++;
979 toff++;
980 ret++;
981 } while (--i);
982 swap_device_unlock(swapdev);
983 return ret;
984 }
985
This page was automatically generated by the
LXR engine.
Visit the LXR main site for more
information.