fs/btrfs/super.c

   1 /*
   2  * Copyright (C) 2007 Oracle.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18
  19 #include <linux/blkdev.h>
  20 #include <linux/module.h>
  21 #include <linux/buffer_head.h>
  22 #include <linux/fs.h>
  23 #include <linux/pagemap.h>
  24 #include <linux/highmem.h>
  25 #include <linux/time.h>
  26 #include <linux/init.h>
  27 #include <linux/string.h>
  28 #include <linux/smp_lock.h>
  29 #include <linux/backing-dev.h>
  30 #include <linux/mount.h>
  31 #include <linux/mpage.h>
  32 #include <linux/swap.h>
  33 #include <linux/writeback.h>
  34 #include <linux/statfs.h>
  35 #include <linux/compat.h>
  36 #include <linux/parser.h>
  37 #include <linux/ctype.h>
  38 #include <linux/namei.h>
  39 #include <linux/miscdevice.h>
  40 #include "ctree.h"
  41 #include "disk-io.h"
  42 #include "transaction.h"
  43 #include "btrfs_inode.h"
  44 #include "ioctl.h"
  45 #include "print-tree.h"
  46 #include "xattr.h"
  47 #include "volumes.h"
  48
  49 #define BTRFS_SUPER_MAGIC 0x9123683E
  50
  51 static struct super_operations btrfs_super_ops;
  52
  53 static void btrfs_put_super (struct super_block * sb)
  54 {
  55         struct btrfs_root *root = btrfs_sb(sb);
  56         struct btrfs_fs_info *fs = root->fs_info;
  57         int ret;
  58
  59         ret = close_ctree(root);
  60         if (ret) {
  61                 printk("close ctree returns %d\n", ret);
  62         }
  63         btrfs_sysfs_del_super(fs);
  64         sb->s_fs_info = NULL;
  65 }
  66
  67 enum {
  68         Opt_degraded, Opt_subvol, Opt_nodatasum, Opt_nodatacow,
  69         Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier,
  70         Opt_ssd, Opt_err,
  71 };
  72
  73 static match_table_t tokens = {
  74         {Opt_degraded, "degraded"},
  75         {Opt_subvol, "subvol=%s"},
  76         {Opt_nodatasum, "nodatasum"},
  77         {Opt_nodatacow, "nodatacow"},
  78         {Opt_nobarrier, "nobarrier"},
  79         {Opt_max_extent, "max_extent=%s"},
  80         {Opt_max_inline, "max_inline=%s"},
  81         {Opt_alloc_start, "alloc_start=%s"},
  82         {Opt_ssd, "ssd"},
  83         {Opt_err, NULL}
  84 };
  85
  86 u64 btrfs_parse_size(char *str)
  87 {
  88         u64 res;
  89         int mult = 1;
  90         char *end;
  91         char last;
  92
  93         res = simple_strtoul(str, &end, 10);
  94
  95         last = end[0];
  96         if (isalpha(last)) {
  97                 last = tolower(last);
  98                 switch (last) {
  99                 case 'g':
 100                         mult *= 1024;
 101                 case 'm':
 102                         mult *= 1024;
 103                 case 'k':
 104                         mult *= 1024;
 105                 }
 106                 res = res * mult;
 107         }
 108         return res;
 109 }
 110
 111 /*
 112  * Regular mount options parser.  Everything that is needed only when
 113  * reading in a new superblock is parsed here.
 114  */
 115 int btrfs_parse_options(struct btrfs_root *root, char *options)
 116 {
 117         struct btrfs_fs_info *info = root->fs_info;
 118         substring_t args[MAX_OPT_ARGS];
 119         char *p, *num;
 120
 121         if (!options)
 122                 return 0;
 123
 124         /*
 125          * strsep changes the string, duplicate it because parse_options
 126          * gets called twice
 127          */
 128         options = kstrdup(options, GFP_NOFS);
 129         if (!options)
 130                 return -ENOMEM;
 131
 132
 133         while ((p = strsep(&options, ",")) != NULL) {
 134                 int token;
 135                 if (!*p)
 136                         continue;
 137
 138                 token = match_token(p, tokens, args);
 139                 switch (token) {
 140                 case Opt_degraded:
 141                         printk(KERN_INFO "btrfs: allowing degraded mounts\n");
 142                         btrfs_set_opt(info->mount_opt, DEGRADED);
 143                         break;
 144                 case Opt_subvol:
 145                         /*
 146                          * This one is parsed by btrfs_parse_early_options
 147                          * and can be happily ignored here.
 148                          */
 149                         break;
 150                 case Opt_nodatasum:
 151                         printk(KERN_INFO "btrfs: setting nodatacsum\n");
 152                         btrfs_set_opt(info->mount_opt, NODATASUM);
 153                         break;
 154                 case Opt_nodatacow:
 155                         printk(KERN_INFO "btrfs: setting nodatacow\n");
 156                         btrfs_set_opt(info->mount_opt, NODATACOW);
 157                         btrfs_set_opt(info->mount_opt, NODATASUM);
 158                         break;
 159                 case Opt_ssd:
 160                         printk(KERN_INFO "btrfs: use ssd allocation scheme\n");
 161                         btrfs_set_opt(info->mount_opt, SSD);
 162                         break;
 163                 case Opt_nobarrier:
 164                         printk(KERN_INFO "btrfs: turning off barriers\n");
 165                         btrfs_set_opt(info->mount_opt, NOBARRIER);
 166                         break;
 167                 case Opt_max_extent:
 168                         num = match_strdup(&args[0]);
 169                         if (num) {
 170                                 info->max_extent = btrfs_parse_size(num);
 171                                 kfree(num);
 172
 173                                 info->max_extent = max_t(u64,
 174                                         info->max_extent, root->sectorsize);
 175                                 printk(KERN_INFO "btrfs: max_extent at %llu\n",
 176                                        info->max_extent);
 177                         }
 178                         break;
 179                 case Opt_max_inline:
 180                         num = match_strdup(&args[0]);
 181                         if (num) {
 182                                 info->max_inline = btrfs_parse_size(num);
 183                                 kfree(num);
 184
 185                                 info->max_inline = max_t(u64,
 186                                         info->max_inline, root->sectorsize);
 187                                 printk(KERN_INFO "btrfs: max_inline at %llu\n",
 188                                         info->max_inline);
 189                         }
 190                         break;
 191                 case Opt_alloc_start:
 192                         num = match_strdup(&args[0]);
 193                         if (num) {
 194                                 info->alloc_start = btrfs_parse_size(num);
 195                                 kfree(num);
 196                                 printk(KERN_INFO
 197                                         "btrfs: allocations start at %llu\n",
 198                                         info->alloc_start);
 199                         }
 200                         break;
 201                 default:
 202                         break;
 203                 }
 204         }
 205         kfree(options);
 206         return 0;
 207 }
 208
 209 /*
 210  * Parse mount options that are required early in the mount process.
 211  *
 212  * All other options will be parsed on much later in the mount process and
 213  * only when we need to allocate a new super block.
 214  */
 215 static int btrfs_parse_early_options(const char *options,
 216                         char **subvol_name)
 217 {
 218         substring_t args[MAX_OPT_ARGS];
 219         char *opts, *p;
 220         int error = 0;
 221
 222         if (!options)
 223                 goto out;
 224
 225         /*
 226          * strsep changes the string, duplicate it because parse_options
 227          * gets called twice
 228          */
 229         opts = kstrdup(options, GFP_KERNEL);
 230         if (!opts)
 231                 return -ENOMEM;
 232
 233         while ((p = strsep(&opts, ",")) != NULL) {
 234                 int token;
 235                 if (!*p)
 236                         continue;
 237
 238                 token = match_token(p, tokens, args);
 239                 switch (token) {
 240                 case Opt_subvol:
 241                         *subvol_name = match_strdup(&args[0]);
 242                         break;
 243                 default:
 244                         break;
 245                 }
 246         }
 247
 248         kfree(opts);
 249  out:
 250         /*
 251          * If no subvolume name is specified we use the default one.  Allocate
 252          * a copy of the string "default" here so that code later in the
 253          * mount path doesn't care if it's the default volume or another one.
 254          */
 255         if (!*subvol_name) {
 256                 *subvol_name = kstrdup("default", GFP_KERNEL);
 257                 if (!*subvol_name)
 258                         return -ENOMEM;
 259         }
 260         return error;
 261 }
 262
 263 static int btrfs_fill_super(struct super_block * sb,
 264                             struct btrfs_fs_devices *fs_devices,
 265                             void * data, int silent)
 266 {
 267         struct inode * inode;
 268         struct dentry * root_dentry;
 269         struct btrfs_super_block *disk_super;
 270         struct btrfs_root *tree_root;
 271         struct btrfs_inode *bi;
 272         int err;
 273
 274         sb->s_maxbytes = MAX_LFS_FILESIZE;
 275         sb->s_magic = BTRFS_SUPER_MAGIC;
 276         sb->s_op = &btrfs_super_ops;
 277         sb->s_xattr = btrfs_xattr_handlers;
 278         sb->s_time_gran = 1;
 279
 280         tree_root = open_ctree(sb, fs_devices, (char *)data);
 281
 282         if (IS_ERR(tree_root)) {
 283                 printk("btrfs: open_ctree failed\n");
 284                 return PTR_ERR(tree_root);
 285         }
 286         sb->s_fs_info = tree_root;
 287         disk_super = &tree_root->fs_info->super_copy;
 288         inode = btrfs_iget_locked(sb, btrfs_super_root_dir(disk_super),
 289                                   tree_root);
 290         bi = BTRFS_I(inode);
 291         bi->location.objectid = inode->i_ino;
 292         bi->location.offset = 0;
 293         bi->root = tree_root;
 294
 295         btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
 296
 297         if (!inode) {
 298                 err = -ENOMEM;
 299                 goto fail_close;
 300         }
 301         if (inode->i_state & I_NEW) {
 302                 btrfs_read_locked_inode(inode);
 303                 unlock_new_inode(inode);
 304         }
 305
 306         root_dentry = d_alloc_root(inode);
 307         if (!root_dentry) {
 308                 iput(inode);
 309                 err = -ENOMEM;
 310                 goto fail_close;
 311         }
 312
 313         /* this does the super kobj at the same time */
 314         err = btrfs_sysfs_add_super(tree_root->fs_info);
 315         if (err)
 316                 goto fail_close;
 317
 318         sb->s_root = root_dentry;
 319         btrfs_transaction_queue_work(tree_root, HZ * 30);
 320
 321 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25)
 322         save_mount_options(sb, data);
 323 #endif
 324
 325         return 0;
 326
 327 fail_close:
 328         close_ctree(tree_root);
 329         return err;
 330 }
 331
 332 int btrfs_sync_fs(struct super_block *sb, int wait)
 333 {
 334         struct btrfs_trans_handle *trans;
 335         struct btrfs_root *root;
 336         int ret;
 337         root = btrfs_sb(sb);
 338
 339         sb->s_dirt = 0;
 340         if (!wait) {
 341                 filemap_flush(root->fs_info->btree_inode->i_mapping);
 342                 return 0;
 343         }
 344         btrfs_clean_old_snapshots(root);
 345         mutex_lock(&root->fs_info->fs_mutex);
 346         btrfs_defrag_dirty_roots(root->fs_info);
 347         trans = btrfs_start_transaction(root, 1);
 348         ret = btrfs_commit_transaction(trans, root);
 349         sb->s_dirt = 0;
 350         mutex_unlock(&root->fs_info->fs_mutex);
 351         return ret;
 352 }
 353
 354 static void btrfs_write_super(struct super_block *sb)
 355 {
 356         sb->s_dirt = 0;
 357 }
 358
 359 static int btrfs_test_super(struct super_block *s, void *data)
 360 {
 361         struct btrfs_fs_devices *test_fs_devices = data;
 362         struct btrfs_root *root = btrfs_sb(s);
 363
 364         return root->fs_info->fs_devices == test_fs_devices;
 365 }
 366
 367 /*
 368  * Find a superblock for the given device / mount point.
 369  *
 370  * Note:  This is based on get_sb_bdev from fs/super.c with a few additions
 371  *        for multiple device setup.  Make sure to keep it in sync.
 372  */
 373 static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
 374                 const char *dev_name, void *data, struct vfsmount *mnt)
 375 {
 376         char *subvol_name = NULL;
 377         struct block_device *bdev = NULL;
 378         struct super_block *s;
 379         struct dentry *root;
 380         struct btrfs_fs_devices *fs_devices = NULL;
 381         int error = 0;
 382
 383         error = btrfs_parse_early_options(data, &subvol_name);
 384         if (error)
 385                 goto error;
 386
 387         error = btrfs_scan_one_device(dev_name, flags, fs_type, &fs_devices);
 388         if (error)
 389                 goto error_free_subvol_name;
 390
 391         error = btrfs_open_devices(fs_devices, flags, fs_type);
 392         if (error)
 393                 goto error_free_subvol_name;
 394
 395         bdev = fs_devices->latest_bdev;
 396         btrfs_lock_volumes();
 397         s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
 398         btrfs_unlock_volumes();
 399         if (IS_ERR(s))
 400                 goto error_s;
 401
 402         if (s->s_root) {
 403                 if ((flags ^ s->s_flags) & MS_RDONLY) {
 404                         up_write(&s->s_umount);
 405                         deactivate_super(s);
 406                         error = -EBUSY;
 407                         goto error_bdev;
 408                 }
 409
 410         } else {
 411                 char b[BDEVNAME_SIZE];
 412
 413                 s->s_flags = flags;
 414                 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
 415                 error = btrfs_fill_super(s, fs_devices, data,
 416                                          flags & MS_SILENT ? 1 : 0);
 417                 if (error) {
 418                         up_write(&s->s_umount);
 419                         deactivate_super(s);
 420                         goto error;
 421                 }
 422
 423                 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
 424                 s->s_flags |= MS_ACTIVE;
 425         }
 426
 427         root = lookup_one_len(subvol_name, s->s_root, strlen(subvol_name));
 428         if (IS_ERR(root)) {
 429                 up_write(&s->s_umount);
 430                 deactivate_super(s);
 431                 error = PTR_ERR(root);
 432                 goto error;
 433         }
 434         if (!root->d_inode) {
 435                 dput(root);
 436                 up_write(&s->s_umount);
 437                 deactivate_super(s);
 438                 error = -ENXIO;
 439                 goto error;
 440         }
 441
 442         mnt->mnt_sb = s;
 443         mnt->mnt_root = root;
 444
 445         kfree(subvol_name);
 446         return 0;
 447
 448 error_s:
 449         error = PTR_ERR(s);
 450 error_bdev:
 451         btrfs_close_devices(fs_devices);
 452 error_free_subvol_name:
 453         kfree(subvol_name);
 454 error:
 455         return error;
 456 }
 457
 458 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 459 {
 460         struct btrfs_root *root = btrfs_sb(dentry->d_sb);
 461         struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
 462         int bits = dentry->d_sb->s_blocksize_bits;
 463
 464         buf->f_namelen = BTRFS_NAME_LEN;
 465         buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
 466         buf->f_bfree = buf->f_blocks -
 467                 (btrfs_super_bytes_used(disk_super) >> bits);
 468         buf->f_bavail = buf->f_bfree;
 469         buf->f_bsize = dentry->d_sb->s_blocksize;
 470         buf->f_type = BTRFS_SUPER_MAGIC;
 471         return 0;
 472 }
 473
 474 static struct file_system_type btrfs_fs_type = {
 475         .owner          = THIS_MODULE,
 476         .name           = "btrfs",
 477         .get_sb         = btrfs_get_sb,
 478         .kill_sb        = kill_anon_super,
 479         .fs_flags       = FS_REQUIRES_DEV,
 480 };
 481
 482 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
 483                                 unsigned long arg)
 484 {
 485         struct btrfs_ioctl_vol_args *vol;
 486         struct btrfs_fs_devices *fs_devices;
 487         int ret = 0;
 488         int len;
 489
 490         vol = kmalloc(sizeof(*vol), GFP_KERNEL);
 491         if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
 492                 ret = -EFAULT;
 493                 goto out;
 494         }
 495         len = strnlen(vol->name, BTRFS_PATH_NAME_MAX);
 496         switch (cmd) {
 497         case BTRFS_IOC_SCAN_DEV:
 498                 ret = btrfs_scan_one_device(vol->name, MS_RDONLY,
 499                                             &btrfs_fs_type, &fs_devices);
 500                 break;
 501         }
 502 out:
 503         kfree(vol);
 504         return ret;
 505 }
 506
 507 static void btrfs_write_super_lockfs(struct super_block *sb)
 508 {
 509         struct btrfs_root *root = btrfs_sb(sb);
 510         btrfs_transaction_flush_work(root);
 511 }
 512
 513 static void btrfs_unlockfs(struct super_block *sb)
 514 {
 515         struct btrfs_root *root = btrfs_sb(sb);
 516         btrfs_transaction_queue_work(root, HZ * 30);
 517 }
 518
 519 static struct super_operations btrfs_super_ops = {
 520         .delete_inode   = btrfs_delete_inode,
 521         .put_super      = btrfs_put_super,
 522         .write_super    = btrfs_write_super,
 523         .sync_fs        = btrfs_sync_fs,
 524 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25)
 525         .read_inode     = btrfs_read_locked_inode,
 526 #else
 527         .show_options   = generic_show_options,
 528 #endif
 529         .write_inode    = btrfs_write_inode,
 530         .dirty_inode    = btrfs_dirty_inode,
 531         .alloc_inode    = btrfs_alloc_inode,
 532         .destroy_inode  = btrfs_destroy_inode,
 533         .statfs         = btrfs_statfs,
 534         .write_super_lockfs = btrfs_write_super_lockfs,
 535         .unlockfs       = btrfs_unlockfs,
 536 };
 537
 538 static const struct file_operations btrfs_ctl_fops = {
 539         .unlocked_ioctl  = btrfs_control_ioctl,
 540         .compat_ioctl = btrfs_control_ioctl,
 541         .owner   = THIS_MODULE,
 542 };
 543
 544 static struct miscdevice btrfs_misc = {
 545         .minor          = MISC_DYNAMIC_MINOR,
 546         .name           = "btrfs-control",
 547         .fops           = &btrfs_ctl_fops
 548 };
 549
 550 static int btrfs_interface_init(void)
 551 {
 552         return misc_register(&btrfs_misc);
 553 }
 554
 555 void btrfs_interface_exit(void)
 556 {
 557         if (misc_deregister(&btrfs_misc) < 0)
 558                 printk("misc_deregister failed for control device");
 559 }
 560
 561 static int __init init_btrfs_fs(void)
 562 {
 563         int err;
 564
 565         err = btrfs_init_sysfs();
 566         if (err)
 567                 return err;
 568
 569         btrfs_init_transaction_sys();
 570         err = btrfs_init_cachep();
 571         if (err)
 572                 goto free_transaction_sys;
 573
 574         err = extent_io_init();
 575         if (err)
 576                 goto free_cachep;
 577
 578         err = extent_map_init();
 579         if (err)
 580                 goto free_extent_io;
 581
 582         err = btrfs_interface_init();
 583         if (err)
 584                 goto free_extent_map;
 585         err = register_filesystem(&btrfs_fs_type);
 586         if (err)
 587                 goto unregister_ioctl;
 588         return 0;
 589
 590 unregister_ioctl:
 591         btrfs_interface_exit();
 592 free_extent_map:
 593         extent_map_exit();
 594 free_extent_io:
 595         extent_io_exit();
 596 free_cachep:
 597         btrfs_destroy_cachep();
 598 free_transaction_sys:
 599         btrfs_exit_transaction_sys();
 600         btrfs_exit_sysfs();
 601         return err;
 602 }
 603
 604 static void __exit exit_btrfs_fs(void)
 605 {
 606         btrfs_exit_transaction_sys();
 607         btrfs_destroy_cachep();
 608         extent_map_exit();
 609         extent_io_exit();
 610         btrfs_interface_exit();
 611         unregister_filesystem(&btrfs_fs_type);
 612         btrfs_exit_sysfs();
 613         btrfs_cleanup_fs_uuids();
 614 }
 615
 616 module_init(init_btrfs_fs)
 617 module_exit(exit_btrfs_fs)
 618
 619 MODULE_LICENSE("GPL");