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_subvol, Opt_nodatasum, Opt_nodatacow, Opt_max_extent,
  69         Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd, Opt_err,
  70 };
  71
  72 static match_table_t tokens = {
  73         {Opt_subvol, "subvol=%s"},
  74         {Opt_nodatasum, "nodatasum"},
  75         {Opt_nodatacow, "nodatacow"},
  76         {Opt_nobarrier, "nobarrier"},
  77         {Opt_max_extent, "max_extent=%s"},
  78         {Opt_max_inline, "max_inline=%s"},
  79         {Opt_alloc_start, "alloc_start=%s"},
  80         {Opt_ssd, "ssd"},
  81         {Opt_err, NULL}
  82 };
  83
  84 u64 btrfs_parse_size(char *str)
  85 {
  86         u64 res;
  87         int mult = 1;
  88         char *end;
  89         char last;
  90
  91         res = simple_strtoul(str, &end, 10);
  92
  93         last = end[0];
  94         if (isalpha(last)) {
  95                 last = tolower(last);
  96                 switch (last) {
  97                 case 'g':
  98                         mult *= 1024;
  99                 case 'm':
 100                         mult *= 1024;
 101                 case 'k':
 102                         mult *= 1024;
 103                 }
 104                 res = res * mult;
 105         }
 106         return res;
 107 }
 108
 109 static int parse_options (char * options,
 110                           struct btrfs_root *root,
 111                           char **subvol_name)
 112 {
 113         char * p;
 114         struct btrfs_fs_info *info = NULL;
 115         substring_t args[MAX_OPT_ARGS];
 116
 117         if (!options)
 118                 return 1;
 119
 120         /*
 121          * strsep changes the string, duplicate it because parse_options
 122          * gets called twice
 123          */
 124         options = kstrdup(options, GFP_NOFS);
 125         if (!options)
 126                 return -ENOMEM;
 127
 128         if (root)
 129                 info = root->fs_info;
 130
 131         while ((p = strsep (&options, ",")) != NULL) {
 132                 int token;
 133                 if (!*p)
 134                         continue;
 135
 136                 token = match_token(p, tokens, args);
 137                 switch (token) {
 138                 case Opt_subvol:
 139                         if (subvol_name) {
 140                                 *subvol_name = match_strdup(&args[0]);
 141                         }
 142                         break;
 143                 case Opt_nodatasum:
 144                         if (info) {
 145                                 printk("btrfs: setting nodatacsum\n");
 146                                 btrfs_set_opt(info->mount_opt, NODATASUM);
 147                         }
 148                         break;
 149                 case Opt_nodatacow:
 150                         if (info) {
 151                                 printk("btrfs: setting nodatacow\n");
 152                                 btrfs_set_opt(info->mount_opt, NODATACOW);
 153                                 btrfs_set_opt(info->mount_opt, NODATASUM);
 154                         }
 155                         break;
 156                 case Opt_ssd:
 157                         if (info) {
 158                                 printk("btrfs: use ssd allocation scheme\n");
 159                                 btrfs_set_opt(info->mount_opt, SSD);
 160                         }
 161                         break;
 162                 case Opt_nobarrier:
 163                         if (info) {
 164                                 printk("btrfs: turning off barriers\n");
 165                                 btrfs_set_opt(info->mount_opt, NOBARRIER);
 166                         }
 167                         break;
 168                 case Opt_max_extent:
 169                         if (info) {
 170                                 char *num = match_strdup(&args[0]);
 171                                 if (num) {
 172                                         info->max_extent =
 173                                                 btrfs_parse_size(num);
 174                                         kfree(num);
 175
 176                                         info->max_extent = max_t(u64,
 177                                                          info->max_extent,
 178                                                          root->sectorsize);
 179                                         printk("btrfs: max_extent at %Lu\n",
 180                                                info->max_extent);
 181                                 }
 182                         }
 183                         break;
 184                 case Opt_max_inline:
 185                         if (info) {
 186                                 char *num = match_strdup(&args[0]);
 187                                 if (num) {
 188                                         info->max_inline =
 189                                                 btrfs_parse_size(num);
 190                                         kfree(num);
 191
 192                                         info->max_inline = max_t(u64,
 193                                                          info->max_inline,
 194                                                          root->sectorsize);
 195                                         printk("btrfs: max_inline at %Lu\n",
 196                                                info->max_inline);
 197                                 }
 198                         }
 199                         break;
 200                 case Opt_alloc_start:
 201                         if (info) {
 202                                 char *num = match_strdup(&args[0]);
 203                                 if (num) {
 204                                         info->alloc_start =
 205                                                 btrfs_parse_size(num);
 206                                         kfree(num);
 207                                         printk("btrfs: allocations start at "
 208                                                "%Lu\n", info->alloc_start);
 209                                 }
 210                         }
 211                         break;
 212                 default:
 213                         break;
 214                 }
 215         }
 216         kfree(options);
 217         return 1;
 218 }
 219
 220 static int btrfs_fill_super(struct super_block * sb,
 221                             struct btrfs_fs_devices *fs_devices,
 222                             void * data, int silent)
 223 {
 224         struct inode * inode;
 225         struct dentry * root_dentry;
 226         struct btrfs_super_block *disk_super;
 227         struct btrfs_root *tree_root;
 228         struct btrfs_inode *bi;
 229         int err;
 230
 231         sb->s_maxbytes = MAX_LFS_FILESIZE;
 232         sb->s_magic = BTRFS_SUPER_MAGIC;
 233         sb->s_op = &btrfs_super_ops;
 234         sb->s_xattr = btrfs_xattr_handlers;
 235         sb->s_time_gran = 1;
 236
 237         tree_root = open_ctree(sb, fs_devices);
 238
 239         if (IS_ERR(tree_root)) {
 240                 printk("btrfs: open_ctree failed\n");
 241                 return PTR_ERR(tree_root);
 242         }
 243         sb->s_fs_info = tree_root;
 244         disk_super = &tree_root->fs_info->super_copy;
 245         inode = btrfs_iget_locked(sb, btrfs_super_root_dir(disk_super),
 246                                   tree_root);
 247         bi = BTRFS_I(inode);
 248         bi->location.objectid = inode->i_ino;
 249         bi->location.offset = 0;
 250         bi->root = tree_root;
 251
 252         btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
 253
 254         if (!inode) {
 255                 err = -ENOMEM;
 256                 goto fail_close;
 257         }
 258         if (inode->i_state & I_NEW) {
 259                 btrfs_read_locked_inode(inode);
 260                 unlock_new_inode(inode);
 261         }
 262
 263         root_dentry = d_alloc_root(inode);
 264         if (!root_dentry) {
 265                 iput(inode);
 266                 err = -ENOMEM;
 267                 goto fail_close;
 268         }
 269
 270         parse_options((char *)data, tree_root, NULL);
 271
 272         /* this does the super kobj at the same time */
 273         err = btrfs_sysfs_add_super(tree_root->fs_info);
 274         if (err)
 275                 goto fail_close;
 276
 277         sb->s_root = root_dentry;
 278         btrfs_transaction_queue_work(tree_root, HZ * 30);
 279
 280 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25)
 281         save_mount_options(sb, data);
 282 #endif
 283
 284         return 0;
 285
 286 fail_close:
 287         close_ctree(tree_root);
 288         return err;
 289 }
 290
 291 static int btrfs_sync_fs(struct super_block *sb, int wait)
 292 {
 293         struct btrfs_trans_handle *trans;
 294         struct btrfs_root *root;
 295         int ret;
 296         root = btrfs_sb(sb);
 297
 298         sb->s_dirt = 0;
 299         if (!wait) {
 300                 filemap_flush(root->fs_info->btree_inode->i_mapping);
 301                 return 0;
 302         }
 303         btrfs_clean_old_snapshots(root);
 304         mutex_lock(&root->fs_info->fs_mutex);
 305         btrfs_defrag_dirty_roots(root->fs_info);
 306         trans = btrfs_start_transaction(root, 1);
 307         ret = btrfs_commit_transaction(trans, root);
 308         sb->s_dirt = 0;
 309         mutex_unlock(&root->fs_info->fs_mutex);
 310         return ret;
 311 }
 312
 313 static void btrfs_write_super(struct super_block *sb)
 314 {
 315         sb->s_dirt = 0;
 316 }
 317
 318 /*
 319  * This is almost a copy of get_sb_bdev in fs/super.c.
 320  * We need the local copy to allow direct mounting of
 321  * subvolumes, but this could be easily integrated back
 322  * into the generic version.  --hch
 323  */
 324
 325 /* start copy & paste */
 326 static int set_bdev_super(struct super_block *s, void *data)
 327 {
 328         s->s_bdev = data;
 329         s->s_dev = s->s_bdev->bd_dev;
 330         return 0;
 331 }
 332
 333 static int test_bdev_super(struct super_block *s, void *data)
 334 {
 335         return (void *)s->s_bdev == data;
 336 }
 337
 338 int btrfs_get_sb_bdev(struct file_system_type *fs_type,
 339         int flags, const char *dev_name, void *data,
 340         struct vfsmount *mnt, const char *subvol)
 341 {
 342         struct block_device *bdev = NULL;
 343         struct super_block *s;
 344         struct dentry *root;
 345         struct btrfs_fs_devices *fs_devices = NULL;
 346         int error = 0;
 347
 348         error = btrfs_scan_one_device(dev_name, flags, fs_type, &fs_devices);
 349         if (error)
 350                 return error;
 351
 352         error = btrfs_open_devices(fs_devices, flags, fs_type);
 353         if (error)
 354                 return error;
 355
 356         bdev = fs_devices->lowest_bdev;
 357         /*
 358          * once the super is inserted into the list by sget, s_umount
 359          * will protect the lockfs code from trying to start a snapshot
 360          * while we are mounting
 361          */
 362         down(&bdev->bd_mount_sem);
 363         s = sget(fs_type, test_bdev_super, set_bdev_super, bdev);
 364         up(&bdev->bd_mount_sem);
 365         if (IS_ERR(s))
 366                 goto error_s;
 367
 368         if (s->s_root) {
 369                 if ((flags ^ s->s_flags) & MS_RDONLY) {
 370                         up_write(&s->s_umount);
 371                         deactivate_super(s);
 372                         error = -EBUSY;
 373                         goto error_bdev;
 374                 }
 375
 376                 close_bdev_excl(bdev);
 377         } else {
 378                 char b[BDEVNAME_SIZE];
 379
 380                 s->s_flags = flags;
 381                 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
 382                 sb_set_blocksize(s, block_size(bdev));
 383                 error = btrfs_fill_super(s, fs_devices, data,
 384                                          flags & MS_SILENT ? 1 : 0);
 385                 if (error) {
 386                         up_write(&s->s_umount);
 387                         deactivate_super(s);
 388                         goto error;
 389                 }
 390
 391                 btrfs_sb(s)->fs_info->bdev_holder = fs_type;
 392                 s->s_flags |= MS_ACTIVE;
 393         }
 394
 395         if (subvol) {
 396                 root = lookup_one_len(subvol, s->s_root, strlen(subvol));
 397                 if (IS_ERR(root)) {
 398                         up_write(&s->s_umount);
 399                         deactivate_super(s);
 400                         error = PTR_ERR(root);
 401                         goto error;
 402                 }
 403                 if (!root->d_inode) {
 404                         dput(root);
 405                         up_write(&s->s_umount);
 406                         deactivate_super(s);
 407                         error = -ENXIO;
 408                         goto error;
 409                 }
 410         } else {
 411                 root = dget(s->s_root);
 412         }
 413
 414         mnt->mnt_sb = s;
 415         mnt->mnt_root = root;
 416         return 0;
 417
 418 error_s:
 419         error = PTR_ERR(s);
 420 error_bdev:
 421         btrfs_close_devices(fs_devices);
 422 error:
 423         return error;
 424 }
 425 /* end copy & paste */
 426
 427 static int btrfs_get_sb(struct file_system_type *fs_type,
 428         int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 429 {
 430         int ret;
 431         char *subvol_name = NULL;
 432
 433         parse_options((char *)data, NULL, &subvol_name);
 434         ret = btrfs_get_sb_bdev(fs_type, flags, dev_name, data, mnt,
 435                         subvol_name ? subvol_name : "default");
 436         if (subvol_name)
 437                 kfree(subvol_name);
 438         return ret;
 439 }
 440
 441 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 442 {
 443         struct btrfs_root *root = btrfs_sb(dentry->d_sb);
 444         struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
 445         int bits = dentry->d_sb->s_blocksize_bits;
 446
 447         buf->f_namelen = BTRFS_NAME_LEN;
 448         buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
 449         buf->f_bfree = buf->f_blocks -
 450                 (btrfs_super_bytes_used(disk_super) >> bits);
 451         buf->f_bavail = buf->f_bfree;
 452         buf->f_bsize = dentry->d_sb->s_blocksize;
 453         buf->f_type = BTRFS_SUPER_MAGIC;
 454         return 0;
 455 }
 456
 457 static struct file_system_type btrfs_fs_type = {
 458         .owner          = THIS_MODULE,
 459         .name           = "btrfs",
 460         .get_sb         = btrfs_get_sb,
 461         .kill_sb        = kill_block_super,
 462         .fs_flags       = FS_REQUIRES_DEV,
 463 };
 464
 465 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
 466                                 unsigned long arg)
 467 {
 468         struct btrfs_ioctl_vol_args *vol;
 469         struct btrfs_fs_devices *fs_devices;
 470         int ret;
 471         int len;
 472
 473         vol = kmalloc(sizeof(*vol), GFP_KERNEL);
 474         if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) {
 475                 ret = -EFAULT;
 476                 goto out;
 477         }
 478         len = strnlen(vol->name, BTRFS_PATH_NAME_MAX);
 479         switch (cmd) {
 480         case BTRFS_IOC_SCAN_DEV:
 481                 ret = btrfs_scan_one_device(vol->name, MS_RDONLY,
 482                                             &btrfs_fs_type, &fs_devices);
 483                 break;
 484         }
 485 out:
 486         kfree(vol);
 487         return 0;
 488 }
 489
 490 static void btrfs_write_super_lockfs(struct super_block *sb)
 491 {
 492         struct btrfs_root *root = btrfs_sb(sb);
 493         btrfs_transaction_flush_work(root);
 494 }
 495
 496 static void btrfs_unlockfs(struct super_block *sb)
 497 {
 498         struct btrfs_root *root = btrfs_sb(sb);
 499         btrfs_transaction_queue_work(root, HZ * 30);
 500 }
 501
 502 static struct super_operations btrfs_super_ops = {
 503         .delete_inode   = btrfs_delete_inode,
 504         .put_inode      = btrfs_put_inode,
 505         .put_super      = btrfs_put_super,
 506         .write_super    = btrfs_write_super,
 507         .sync_fs        = btrfs_sync_fs,
 508 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25)
 509         .read_inode     = btrfs_read_locked_inode,
 510 #else
 511         .show_options   = generic_show_options,
 512 #endif
 513         .write_inode    = btrfs_write_inode,
 514         .dirty_inode    = btrfs_dirty_inode,
 515         .alloc_inode    = btrfs_alloc_inode,
 516         .destroy_inode  = btrfs_destroy_inode,
 517         .statfs         = btrfs_statfs,
 518         .write_super_lockfs = btrfs_write_super_lockfs,
 519         .unlockfs       = btrfs_unlockfs,
 520 };
 521
 522 static const struct file_operations btrfs_ctl_fops = {
 523         .unlocked_ioctl  = btrfs_control_ioctl,
 524         .compat_ioctl = btrfs_control_ioctl,
 525         .owner   = THIS_MODULE,
 526 };
 527
 528 static struct miscdevice btrfs_misc = {
 529         .minor          = MISC_DYNAMIC_MINOR,
 530         .name           = "btrfs-control",
 531         .fops           = &btrfs_ctl_fops
 532 };
 533
 534 static int btrfs_interface_init(void)
 535 {
 536         return misc_register(&btrfs_misc);
 537 }
 538
 539 void btrfs_interface_exit(void)
 540 {
 541         if (misc_deregister(&btrfs_misc) < 0)
 542                 printk("misc_deregister failed for control device");
 543 }
 544
 545 static int __init init_btrfs_fs(void)
 546 {
 547         int err;
 548
 549         err = btrfs_init_sysfs();
 550         if (err)
 551                 return err;
 552
 553         btrfs_init_transaction_sys();
 554         err = btrfs_init_cachep();
 555         if (err)
 556                 goto free_transaction_sys;
 557
 558         err = extent_io_init();
 559         if (err)
 560                 goto free_cachep;
 561
 562         err = extent_map_init();
 563         if (err)
 564                 goto free_extent_io;
 565
 566         err = btrfs_interface_init();
 567         if (err)
 568                 goto free_extent_map;
 569         err = register_filesystem(&btrfs_fs_type);
 570         if (err)
 571                 goto unregister_ioctl;
 572         return 0;
 573
 574 unregister_ioctl:
 575         btrfs_interface_exit();
 576 free_extent_map:
 577         extent_map_exit();
 578 free_extent_io:
 579         extent_io_exit();
 580 free_cachep:
 581         btrfs_destroy_cachep();
 582 free_transaction_sys:
 583         btrfs_exit_transaction_sys();
 584         btrfs_exit_sysfs();
 585         return err;
 586 }
 587
 588 static void __exit exit_btrfs_fs(void)
 589 {
 590         btrfs_exit_transaction_sys();
 591         btrfs_destroy_cachep();
 592         extent_map_exit();
 593         extent_io_exit();
 594         btrfs_interface_exit();
 595         unregister_filesystem(&btrfs_fs_type);
 596         btrfs_exit_sysfs();
 597         btrfs_cleanup_fs_uuids();
 598 }
 599
 600 module_init(init_btrfs_fs)
 601 module_exit(exit_btrfs_fs)
 602
 603 MODULE_LICENSE("GPL");