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
  48 #define BTRFS_SUPER_MAGIC 0x9123683E
  49
  50 static struct super_operations btrfs_super_ops;
  51
  52 static void btrfs_put_super (struct super_block * sb)
  53 {
  54         struct btrfs_root *root = btrfs_sb(sb);
  55         struct btrfs_fs_info *fs = root->fs_info;
  56         int ret;
  57
  58         ret = close_ctree(root);
  59         if (ret) {
  60                 printk("close ctree returns %d\n", ret);
  61         }
  62         btrfs_sysfs_del_super(fs);
  63         sb->s_fs_info = NULL;
  64 }
  65
  66 enum {
  67         Opt_subvol, Opt_nodatasum, Opt_nodatacow, Opt_max_extent,
  68         Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd, Opt_err,
  69 };
  70
  71 static match_table_t tokens = {
  72         {Opt_subvol, "subvol=%s"},
  73         {Opt_nodatasum, "nodatasum"},
  74         {Opt_nodatacow, "nodatacow"},
  75         {Opt_nobarrier, "nobarrier"},
  76         {Opt_max_extent, "max_extent=%s"},
  77         {Opt_max_inline, "max_inline=%s"},
  78         {Opt_alloc_start, "alloc_start=%s"},
  79         {Opt_ssd, "ssd"},
  80         {Opt_err, NULL}
  81 };
  82
  83 u64 btrfs_parse_size(char *str)
  84 {
  85         u64 res;
  86         int mult = 1;
  87         char *end;
  88         char last;
  89
  90         res = simple_strtoul(str, &end, 10);
  91
  92         last = end[0];
  93         if (isalpha(last)) {
  94                 last = tolower(last);
  95                 switch (last) {
  96                 case 'g':
  97                         mult *= 1024;
  98                 case 'm':
  99                         mult *= 1024;
 100                 case 'k':
 101                         mult *= 1024;
 102                 }
 103                 res = res * mult;
 104         }
 105         return res;
 106 }
 107
 108 static int parse_options (char * options,
 109                           struct btrfs_root *root,
 110                           char **subvol_name)
 111 {
 112         char * p;
 113         struct btrfs_fs_info *info = NULL;
 114         substring_t args[MAX_OPT_ARGS];
 115
 116         if (!options)
 117                 return 1;
 118
 119         /*
 120          * strsep changes the string, duplicate it because parse_options
 121          * gets called twice
 122          */
 123         options = kstrdup(options, GFP_NOFS);
 124         if (!options)
 125                 return -ENOMEM;
 126
 127         if (root)
 128                 info = root->fs_info;
 129
 130         while ((p = strsep (&options, ",")) != NULL) {
 131                 int token;
 132                 if (!*p)
 133                         continue;
 134
 135                 token = match_token(p, tokens, args);
 136                 switch (token) {
 137                 case Opt_subvol:
 138                         if (subvol_name) {
 139                                 *subvol_name = match_strdup(&args[0]);
 140                         }
 141                         break;
 142                 case Opt_nodatasum:
 143                         if (info) {
 144                                 printk("btrfs: setting nodatacsum\n");
 145                                 btrfs_set_opt(info->mount_opt, NODATASUM);
 146                         }
 147                         break;
 148                 case Opt_nodatacow:
 149                         if (info) {
 150                                 printk("btrfs: setting nodatacow\n");
 151                                 btrfs_set_opt(info->mount_opt, NODATACOW);
 152                                 btrfs_set_opt(info->mount_opt, NODATASUM);
 153                         }
 154                         break;
 155                 case Opt_ssd:
 156                         if (info) {
 157                                 printk("btrfs: use ssd allocation scheme\n");
 158                                 btrfs_set_opt(info->mount_opt, SSD);
 159                         }
 160                         break;
 161                 case Opt_nobarrier:
 162                         if (info) {
 163                                 printk("btrfs: turning off barriers\n");
 164                                 btrfs_set_opt(info->mount_opt, NOBARRIER);
 165                         }
 166                         break;
 167                 case Opt_max_extent:
 168                         if (info) {
 169                                 char *num = match_strdup(&args[0]);
 170                                 if (num) {
 171                                         info->max_extent =
 172                                                 btrfs_parse_size(num);
 173                                         kfree(num);
 174
 175                                         info->max_extent = max_t(u64,
 176                                                          info->max_extent,
 177                                                          root->sectorsize);
 178                                         printk("btrfs: max_extent at %Lu\n",
 179                                                info->max_extent);
 180                                 }
 181                         }
 182                         break;
 183                 case Opt_max_inline:
 184                         if (info) {
 185                                 char *num = match_strdup(&args[0]);
 186                                 if (num) {
 187                                         info->max_inline =
 188                                                 btrfs_parse_size(num);
 189                                         kfree(num);
 190
 191                                         info->max_inline = max_t(u64,
 192                                                          info->max_inline,
 193                                                          root->sectorsize);
 194                                         printk("btrfs: max_inline at %Lu\n",
 195                                                info->max_inline);
 196                                 }
 197                         }
 198                         break;
 199                 case Opt_alloc_start:
 200                         if (info) {
 201                                 char *num = match_strdup(&args[0]);
 202                                 if (num) {
 203                                         info->alloc_start =
 204                                                 btrfs_parse_size(num);
 205                                         kfree(num);
 206                                         printk("btrfs: allocations start at "
 207                                                "%Lu\n", info->alloc_start);
 208                                 }
 209                         }
 210                         break;
 211                 default:
 212                         break;
 213                 }
 214         }
 215         kfree(options);
 216         return 1;
 217 }
 218
 219 static int btrfs_fill_super(struct super_block * sb, void * data, int silent)
 220 {
 221         struct inode * inode;
 222         struct dentry * root_dentry;
 223         struct btrfs_super_block *disk_super;
 224         struct btrfs_root *tree_root;
 225         struct btrfs_inode *bi;
 226         int err;
 227
 228         sb->s_maxbytes = MAX_LFS_FILESIZE;
 229         sb->s_magic = BTRFS_SUPER_MAGIC;
 230         sb->s_op = &btrfs_super_ops;
 231         sb->s_xattr = btrfs_xattr_handlers;
 232         sb->s_time_gran = 1;
 233
 234         tree_root = open_ctree(sb);
 235
 236         if (!tree_root || IS_ERR(tree_root)) {
 237                 printk("btrfs: open_ctree failed\n");
 238                 return -EIO;
 239         }
 240         sb->s_fs_info = tree_root;
 241         disk_super = &tree_root->fs_info->super_copy;
 242         inode = btrfs_iget_locked(sb, btrfs_super_root_dir(disk_super),
 243                                   tree_root);
 244         bi = BTRFS_I(inode);
 245         bi->location.objectid = inode->i_ino;
 246         bi->location.offset = 0;
 247         bi->root = tree_root;
 248
 249         btrfs_set_key_type(&bi->location, BTRFS_INODE_ITEM_KEY);
 250
 251         if (!inode) {
 252                 err = -ENOMEM;
 253                 goto fail_close;
 254         }
 255         if (inode->i_state & I_NEW) {
 256                 btrfs_read_locked_inode(inode);
 257                 unlock_new_inode(inode);
 258         }
 259
 260         root_dentry = d_alloc_root(inode);
 261         if (!root_dentry) {
 262                 iput(inode);
 263                 err = -ENOMEM;
 264                 goto fail_close;
 265         }
 266
 267         parse_options((char *)data, tree_root, NULL);
 268
 269         /* this does the super kobj at the same time */
 270         err = btrfs_sysfs_add_super(tree_root->fs_info);
 271         if (err)
 272                 goto fail_close;
 273
 274         sb->s_root = root_dentry;
 275         btrfs_transaction_queue_work(tree_root, HZ * 30);
 276
 277 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25)
 278         save_mount_options(sb, data);
 279 #endif
 280
 281         return 0;
 282
 283 fail_close:
 284         close_ctree(tree_root);
 285         return err;
 286 }
 287
 288 static int btrfs_sync_fs(struct super_block *sb, int wait)
 289 {
 290         struct btrfs_trans_handle *trans;
 291         struct btrfs_root *root;
 292         int ret;
 293         root = btrfs_sb(sb);
 294
 295         sb->s_dirt = 0;
 296         if (!wait) {
 297                 filemap_flush(root->fs_info->btree_inode->i_mapping);
 298                 return 0;
 299         }
 300         btrfs_clean_old_snapshots(root);
 301         mutex_lock(&root->fs_info->fs_mutex);
 302         btrfs_defrag_dirty_roots(root->fs_info);
 303         trans = btrfs_start_transaction(root, 1);
 304         ret = btrfs_commit_transaction(trans, root);
 305         sb->s_dirt = 0;
 306         mutex_unlock(&root->fs_info->fs_mutex);
 307         return ret;
 308 }
 309
 310 static void btrfs_write_super(struct super_block *sb)
 311 {
 312         sb->s_dirt = 0;
 313 }
 314
 315 /*
 316  * This is almost a copy of get_sb_bdev in fs/super.c.
 317  * We need the local copy to allow direct mounting of
 318  * subvolumes, but this could be easily integrated back
 319  * into the generic version.  --hch
 320  */
 321
 322 /* start copy & paste */
 323 static int set_bdev_super(struct super_block *s, void *data)
 324 {
 325         s->s_bdev = data;
 326         s->s_dev = s->s_bdev->bd_dev;
 327         return 0;
 328 }
 329
 330 static int test_bdev_super(struct super_block *s, void *data)
 331 {
 332         return (void *)s->s_bdev == data;
 333 }
 334
 335 int btrfs_get_sb_bdev(struct file_system_type *fs_type,
 336         int flags, const char *dev_name, void *data,
 337         int (*fill_super)(struct super_block *, void *, int),
 338         struct vfsmount *mnt, const char *subvol)
 339 {
 340         struct block_device *bdev = NULL;
 341         struct super_block *s;
 342         struct dentry *root;
 343         int error = 0;
 344
 345         bdev = open_bdev_excl(dev_name, flags, fs_type);
 346         if (IS_ERR(bdev))
 347                 return PTR_ERR(bdev);
 348
 349         /*
 350          * once the super is inserted into the list by sget, s_umount
 351          * will protect the lockfs code from trying to start a snapshot
 352          * while we are mounting
 353          */
 354         down(&bdev->bd_mount_sem);
 355         s = sget(fs_type, test_bdev_super, set_bdev_super, bdev);
 356         up(&bdev->bd_mount_sem);
 357         if (IS_ERR(s))
 358                 goto error_s;
 359
 360         if (s->s_root) {
 361                 if ((flags ^ s->s_flags) & MS_RDONLY) {
 362                         up_write(&s->s_umount);
 363                         deactivate_super(s);
 364                         error = -EBUSY;
 365                         goto error_bdev;
 366                 }
 367
 368                 close_bdev_excl(bdev);
 369         } else {
 370                 char b[BDEVNAME_SIZE];
 371
 372                 s->s_flags = flags;
 373                 strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
 374                 sb_set_blocksize(s, block_size(bdev));
 375                 error = fill_super(s, data, flags & MS_SILENT ? 1 : 0);
 376                 if (error) {
 377                         up_write(&s->s_umount);
 378                         deactivate_super(s);
 379                         goto error;
 380                 }
 381
 382                 s->s_flags |= MS_ACTIVE;
 383         }
 384
 385         if (subvol) {
 386                 root = lookup_one_len(subvol, s->s_root, strlen(subvol));
 387                 if (IS_ERR(root)) {
 388                         up_write(&s->s_umount);
 389                         deactivate_super(s);
 390                         error = PTR_ERR(root);
 391                         goto error;
 392                 }
 393                 if (!root->d_inode) {
 394                         dput(root);
 395                         up_write(&s->s_umount);
 396                         deactivate_super(s);
 397                         error = -ENXIO;
 398                         goto error;
 399                 }
 400         } else {
 401                 root = dget(s->s_root);
 402         }
 403
 404         mnt->mnt_sb = s;
 405         mnt->mnt_root = root;
 406         return 0;
 407
 408 error_s:
 409         error = PTR_ERR(s);
 410 error_bdev:
 411         close_bdev_excl(bdev);
 412 error:
 413         return error;
 414 }
 415 /* end copy & paste */
 416
 417 static int btrfs_get_sb(struct file_system_type *fs_type,
 418         int flags, const char *dev_name, void *data, struct vfsmount *mnt)
 419 {
 420         int ret;
 421         char *subvol_name = NULL;
 422
 423         parse_options((char *)data, NULL, &subvol_name);
 424         ret = btrfs_get_sb_bdev(fs_type, flags, dev_name, data,
 425                         btrfs_fill_super, mnt,
 426                         subvol_name ? subvol_name : "default");
 427         if (subvol_name)
 428                 kfree(subvol_name);
 429         return ret;
 430 }
 431
 432 static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 433 {
 434         struct btrfs_root *root = btrfs_sb(dentry->d_sb);
 435         struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
 436         int bits = dentry->d_sb->s_blocksize_bits;
 437
 438         buf->f_namelen = BTRFS_NAME_LEN;
 439         buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
 440         buf->f_bfree = buf->f_blocks -
 441                 (btrfs_super_bytes_used(disk_super) >> bits);
 442         buf->f_bavail = buf->f_bfree;
 443         buf->f_bsize = dentry->d_sb->s_blocksize;
 444         buf->f_type = BTRFS_SUPER_MAGIC;
 445         return 0;
 446 }
 447
 448 static long btrfs_control_ioctl(struct file *file, unsigned int cmd,
 449                                 unsigned long arg)
 450 {
 451         printk("btrfs control ioctl %d\n", cmd);
 452         return 0;
 453 }
 454
 455 static struct file_system_type btrfs_fs_type = {
 456         .owner          = THIS_MODULE,
 457         .name           = "btrfs",
 458         .get_sb         = btrfs_get_sb,
 459         .kill_sb        = kill_block_super,
 460         .fs_flags       = FS_REQUIRES_DEV,
 461 };
 462
 463 static void btrfs_write_super_lockfs(struct super_block *sb)
 464 {
 465         struct btrfs_root *root = btrfs_sb(sb);
 466         btrfs_transaction_flush_work(root);
 467 }
 468
 469 static void btrfs_unlockfs(struct super_block *sb)
 470 {
 471         struct btrfs_root *root = btrfs_sb(sb);
 472         btrfs_transaction_queue_work(root, HZ * 30);
 473 }
 474
 475 static struct super_operations btrfs_super_ops = {
 476         .delete_inode   = btrfs_delete_inode,
 477         .put_inode      = btrfs_put_inode,
 478         .put_super      = btrfs_put_super,
 479         .write_super    = btrfs_write_super,
 480         .sync_fs        = btrfs_sync_fs,
 481 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25)
 482         .read_inode     = btrfs_read_locked_inode,
 483 #else
 484         .show_options   = generic_show_options,
 485 #endif
 486         .write_inode    = btrfs_write_inode,
 487         .dirty_inode    = btrfs_dirty_inode,
 488         .alloc_inode    = btrfs_alloc_inode,
 489         .destroy_inode  = btrfs_destroy_inode,
 490         .statfs         = btrfs_statfs,
 491         .write_super_lockfs = btrfs_write_super_lockfs,
 492         .unlockfs       = btrfs_unlockfs,
 493 };
 494
 495 static const struct file_operations btrfs_ctl_fops = {
 496         .unlocked_ioctl  = btrfs_control_ioctl,
 497         .compat_ioctl = btrfs_control_ioctl,
 498         .owner   = THIS_MODULE,
 499 };
 500
 501 static struct miscdevice btrfs_misc = {
 502         .minor          = MISC_DYNAMIC_MINOR,
 503         .name           = "btrfs-control",
 504         .fops           = &btrfs_ctl_fops
 505 };
 506
 507 static int btrfs_interface_init(void)
 508 {
 509         return misc_register(&btrfs_misc);
 510 }
 511
 512 void btrfs_interface_exit(void)
 513 {
 514         if (misc_deregister(&btrfs_misc) < 0)
 515                 printk("misc_deregister failed for control device");
 516 }
 517
 518 static int __init init_btrfs_fs(void)
 519 {
 520         int err;
 521
 522         err = btrfs_init_sysfs();
 523         if (err)
 524                 return err;
 525
 526         btrfs_init_transaction_sys();
 527         err = btrfs_init_cachep();
 528         if (err)
 529                 goto free_transaction_sys;
 530
 531         err = extent_io_init();
 532         if (err)
 533                 goto free_cachep;
 534
 535         err = extent_map_init();
 536         if (err)
 537                 goto free_extent_io;
 538
 539         err = btrfs_interface_init();
 540         if (err)
 541                 goto free_extent_map;
 542         err = register_filesystem(&btrfs_fs_type);
 543         if (err)
 544                 goto unregister_ioctl;
 545         return 0;
 546
 547 unregister_ioctl:
 548         btrfs_interface_exit();
 549 free_extent_map:
 550         extent_map_exit();
 551 free_extent_io:
 552         extent_io_exit();
 553 free_cachep:
 554         btrfs_destroy_cachep();
 555 free_transaction_sys:
 556         btrfs_exit_transaction_sys();
 557         btrfs_exit_sysfs();
 558         return err;
 559 }
 560
 561 static void __exit exit_btrfs_fs(void)
 562 {
 563         btrfs_exit_transaction_sys();
 564         btrfs_destroy_cachep();
 565         extent_map_exit();
 566         extent_io_exit();
 567         btrfs_interface_exit();
 568         unregister_filesystem(&btrfs_fs_type);
 569         btrfs_exit_sysfs();
 570 }
 571
 572 module_init(init_btrfs_fs)
 573 module_exit(exit_btrfs_fs)
 574
 575 MODULE_LICENSE("GPL");