2 * Copyright (C) International Business Machines Corp., 2000-2004
3 * Portions Copyright (C) Christoph Hellwig, 2001-2002
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
13 * the GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21 #include <linux/module.h>
22 #include <linux/parser.h>
23 #include <linux/completion.h>
24 #include <linux/vfs.h>
25 #include <linux/quotaops.h>
26 #include <linux/mount.h>
27 #include <linux/moduleparam.h>
28 #include <linux/kthread.h>
29 #include <linux/posix_acl.h>
30 #include <linux/buffer_head.h>
31 #include <linux/exportfs.h>
32 #include <linux/crc32.h>
33 #include <linux/slab.h>
34 #include <asm/uaccess.h>
35 #include <linux/seq_file.h>
36 #include <linux/smp_lock.h>
38 #include "jfs_incore.h"
39 #include "jfs_filsys.h"
40 #include "jfs_inode.h"
41 #include "jfs_metapage.h"
42 #include "jfs_superblock.h"
46 #include "jfs_debug.h"
48 MODULE_DESCRIPTION("The Journaled Filesystem (JFS)");
49 MODULE_AUTHOR("Steve Best/Dave Kleikamp/Barry Arndt, IBM");
50 MODULE_LICENSE("GPL");
52 static struct kmem_cache * jfs_inode_cachep;
54 static const struct super_operations jfs_super_operations;
55 static const struct export_operations jfs_export_operations;
56 static struct file_system_type jfs_fs_type;
58 #define MAX_COMMIT_THREADS 64
59 static int commit_threads = 0;
60 module_param(commit_threads, int, 0);
61 MODULE_PARM_DESC(commit_threads, "Number of commit threads");
63 static struct task_struct *jfsCommitThread[MAX_COMMIT_THREADS];
64 struct task_struct *jfsIOthread;
65 struct task_struct *jfsSyncThread;
67 #ifdef CONFIG_JFS_DEBUG
68 int jfsloglevel = JFS_LOGLEVEL_WARN;
69 module_param(jfsloglevel, int, 0644);
70 MODULE_PARM_DESC(jfsloglevel, "Specify JFS loglevel (0, 1 or 2)");
73 static void jfs_handle_error(struct super_block *sb)
75 struct jfs_sb_info *sbi = JFS_SBI(sb);
77 if (sb->s_flags & MS_RDONLY)
80 updateSuper(sb, FM_DIRTY);
82 if (sbi->flag & JFS_ERR_PANIC)
83 panic("JFS (device %s): panic forced after error\n",
85 else if (sbi->flag & JFS_ERR_REMOUNT_RO) {
86 jfs_err("ERROR: (device %s): remounting filesystem "
89 sb->s_flags |= MS_RDONLY;
92 /* nothing is done for continue beyond marking the superblock dirty */
95 void jfs_error(struct super_block *sb, const char * function, ...)
97 static char error_buf[256];
100 va_start(args, function);
101 vsnprintf(error_buf, sizeof(error_buf), function, args);
104 printk(KERN_ERR "ERROR: (device %s): %s\n", sb->s_id, error_buf);
106 jfs_handle_error(sb);
109 static struct inode *jfs_alloc_inode(struct super_block *sb)
111 struct jfs_inode_info *jfs_inode;
113 jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
116 return &jfs_inode->vfs_inode;
119 static void jfs_destroy_inode(struct inode *inode)
121 struct jfs_inode_info *ji = JFS_IP(inode);
123 BUG_ON(!list_empty(&ji->anon_inode_list));
125 spin_lock_irq(&ji->ag_lock);
126 if (ji->active_ag != -1) {
127 struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
128 atomic_dec(&bmap->db_active[ji->active_ag]);
131 spin_unlock_irq(&ji->ag_lock);
132 kmem_cache_free(jfs_inode_cachep, ji);
135 static void jfs_clear_inode(struct inode *inode)
140 static int jfs_statfs(struct dentry *dentry, struct kstatfs *buf)
142 struct jfs_sb_info *sbi = JFS_SBI(dentry->d_sb);
144 struct inomap *imap = JFS_IP(sbi->ipimap)->i_imap;
146 jfs_info("In jfs_statfs");
147 buf->f_type = JFS_SUPER_MAGIC;
148 buf->f_bsize = sbi->bsize;
149 buf->f_blocks = sbi->bmap->db_mapsize;
150 buf->f_bfree = sbi->bmap->db_nfree;
151 buf->f_bavail = sbi->bmap->db_nfree;
153 * If we really return the number of allocated & free inodes, some
154 * applications will fail because they won't see enough free inodes.
155 * We'll try to calculate some guess as to how may inodes we can
158 * buf->f_files = atomic_read(&imap->im_numinos);
159 * buf->f_ffree = atomic_read(&imap->im_numfree);
161 maxinodes = min((s64) atomic_read(&imap->im_numinos) +
162 ((sbi->bmap->db_nfree >> imap->im_l2nbperiext)
163 << L2INOSPEREXT), (s64) 0xffffffffLL);
164 buf->f_files = maxinodes;
165 buf->f_ffree = maxinodes - (atomic_read(&imap->im_numinos) -
166 atomic_read(&imap->im_numfree));
167 buf->f_fsid.val[0] = (u32)crc32_le(0, sbi->uuid, sizeof(sbi->uuid)/2);
168 buf->f_fsid.val[1] = (u32)crc32_le(0, sbi->uuid + sizeof(sbi->uuid)/2,
169 sizeof(sbi->uuid)/2);
171 buf->f_namelen = JFS_NAME_MAX;
175 static void jfs_put_super(struct super_block *sb)
177 struct jfs_sb_info *sbi = JFS_SBI(sb);
180 jfs_info("In jfs_put_super");
186 jfs_err("jfs_umount failed with return code %d", rc);
188 unload_nls(sbi->nls_tab);
190 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
191 iput(sbi->direct_inode);
199 Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
200 Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
201 Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
204 static const match_table_t tokens = {
205 {Opt_integrity, "integrity"},
206 {Opt_nointegrity, "nointegrity"},
207 {Opt_iocharset, "iocharset=%s"},
208 {Opt_resize, "resize=%u"},
209 {Opt_resize_nosize, "resize"},
210 {Opt_errors, "errors=%s"},
211 {Opt_ignore, "noquota"},
212 {Opt_ignore, "quota"},
213 {Opt_usrquota, "usrquota"},
214 {Opt_grpquota, "grpquota"},
217 {Opt_umask, "umask=%u"},
221 static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
224 void *nls_map = (void *)-1; /* -1: no change; NULL: none */
226 struct jfs_sb_info *sbi = JFS_SBI(sb);
233 while ((p = strsep(&options, ",")) != NULL) {
234 substring_t args[MAX_OPT_ARGS];
239 token = match_token(p, tokens, args);
242 *flag &= ~JFS_NOINTEGRITY;
244 case Opt_nointegrity:
245 *flag |= JFS_NOINTEGRITY;
248 /* Silently ignore the quota options */
249 /* Don't do anything ;-) */
252 if (nls_map && nls_map != (void *) -1)
254 if (!strcmp(args[0].from, "none"))
257 nls_map = load_nls(args[0].from);
260 "JFS: charset not found\n");
267 char *resize = args[0].from;
268 *newLVSize = simple_strtoull(resize, &resize, 0);
271 case Opt_resize_nosize:
273 *newLVSize = sb->s_bdev->bd_inode->i_size >>
274 sb->s_blocksize_bits;
277 "JFS: Cannot determine volume size\n");
282 char *errors = args[0].from;
283 if (!errors || !*errors)
285 if (!strcmp(errors, "continue")) {
286 *flag &= ~JFS_ERR_REMOUNT_RO;
287 *flag &= ~JFS_ERR_PANIC;
288 *flag |= JFS_ERR_CONTINUE;
289 } else if (!strcmp(errors, "remount-ro")) {
290 *flag &= ~JFS_ERR_CONTINUE;
291 *flag &= ~JFS_ERR_PANIC;
292 *flag |= JFS_ERR_REMOUNT_RO;
293 } else if (!strcmp(errors, "panic")) {
294 *flag &= ~JFS_ERR_CONTINUE;
295 *flag &= ~JFS_ERR_REMOUNT_RO;
296 *flag |= JFS_ERR_PANIC;
299 "JFS: %s is an invalid error handler\n",
309 *flag |= JFS_USRQUOTA;
312 *flag |= JFS_GRPQUOTA;
319 "JFS: quota operations not supported\n");
324 char *uid = args[0].from;
325 sbi->uid = simple_strtoul(uid, &uid, 0);
330 char *gid = args[0].from;
331 sbi->gid = simple_strtoul(gid, &gid, 0);
336 char *umask = args[0].from;
337 sbi->umask = simple_strtoul(umask, &umask, 8);
338 if (sbi->umask & ~0777) {
340 "JFS: Invalid value of umask\n");
346 printk("jfs: Unrecognized mount option \"%s\" "
347 " or missing value\n", p);
352 if (nls_map != (void *) -1) {
353 /* Discard old (if remount) */
354 unload_nls(sbi->nls_tab);
355 sbi->nls_tab = nls_map;
360 if (nls_map && nls_map != (void *) -1)
365 static int jfs_remount(struct super_block *sb, int *flags, char *data)
369 int flag = JFS_SBI(sb)->flag;
372 if (!parse_options(data, sb, &newLVSize, &flag)) {
377 if (sb->s_flags & MS_RDONLY) {
379 "JFS: resize requires volume to be mounted read-write\n");
383 rc = jfs_extendfs(sb, newLVSize, 0);
390 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
392 * Invalidate any previously read metadata. fsck may have
393 * changed the on-disk data since we mounted r/o
395 truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
397 JFS_SBI(sb)->flag = flag;
398 ret = jfs_mount_rw(sb, 1);
400 /* mark the fs r/w for quota activity */
401 sb->s_flags &= ~MS_RDONLY;
404 dquot_resume(sb, -1);
407 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
408 rc = dquot_suspend(sb, -1);
413 rc = jfs_umount_rw(sb);
414 JFS_SBI(sb)->flag = flag;
418 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
419 if (!(sb->s_flags & MS_RDONLY)) {
420 rc = jfs_umount_rw(sb);
425 JFS_SBI(sb)->flag = flag;
426 ret = jfs_mount_rw(sb, 1);
430 JFS_SBI(sb)->flag = flag;
436 static int jfs_fill_super(struct super_block *sb, void *data, int silent)
438 struct jfs_sb_info *sbi;
442 int flag, ret = -EINVAL;
444 jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
446 if (!new_valid_dev(sb->s_bdev->bd_dev))
449 sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
454 sbi->uid = sbi->gid = sbi->umask = -1;
456 /* initialize the mount flag and determine the default error handler */
457 flag = JFS_ERR_REMOUNT_RO;
459 if (!parse_options((char *) data, sb, &newLVSize, &flag))
463 #ifdef CONFIG_JFS_POSIX_ACL
464 sb->s_flags |= MS_POSIXACL;
468 printk(KERN_ERR "resize option for remount only\n");
473 * Initialize blocksize to 4K.
475 sb_set_blocksize(sb, PSIZE);
478 * Set method vectors.
480 sb->s_op = &jfs_super_operations;
481 sb->s_export_op = &jfs_export_operations;
484 * Initialize direct-mapping inode/address-space
486 inode = new_inode(sb);
493 inode->i_size = sb->s_bdev->bd_inode->i_size;
494 inode->i_mapping->a_ops = &jfs_metapage_aops;
495 insert_inode_hash(inode);
496 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
498 sbi->direct_inode = inode;
503 jfs_err("jfs_mount failed w/return code = %d", rc);
505 goto out_mount_failed;
507 if (sb->s_flags & MS_RDONLY)
510 rc = jfs_mount_rw(sb, 0);
513 jfs_err("jfs_mount_rw failed, return code = %d",
520 sb->s_magic = JFS_SUPER_MAGIC;
522 inode = jfs_iget(sb, ROOT_I);
524 ret = PTR_ERR(inode);
527 sb->s_root = d_alloc_root(inode);
531 if (sbi->mntflag & JFS_OS2)
532 sb->s_root->d_op = &jfs_ci_dentry_operations;
534 /* logical blocks are represented by 40 bits in pxd_t, etc. */
535 sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
536 #if BITS_PER_LONG == 32
538 * Page cache is indexed by long.
539 * I would use MAX_LFS_FILESIZE, but it's only half as big
541 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, (u64)sb->s_maxbytes);
547 jfs_err("jfs_read_super: get root dentry failed");
553 jfs_err("jfs_umount failed with return code %d", rc);
556 filemap_write_and_wait(sbi->direct_inode->i_mapping);
557 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
558 make_bad_inode(sbi->direct_inode);
559 iput(sbi->direct_inode);
560 sbi->direct_inode = NULL;
563 unload_nls(sbi->nls_tab);
569 static int jfs_freeze(struct super_block *sb)
571 struct jfs_sb_info *sbi = JFS_SBI(sb);
572 struct jfs_log *log = sbi->log;
574 if (!(sb->s_flags & MS_RDONLY)) {
577 updateSuper(sb, FM_CLEAN);
582 static int jfs_unfreeze(struct super_block *sb)
584 struct jfs_sb_info *sbi = JFS_SBI(sb);
585 struct jfs_log *log = sbi->log;
588 if (!(sb->s_flags & MS_RDONLY)) {
589 updateSuper(sb, FM_MOUNT);
590 if ((rc = lmLogInit(log)))
591 jfs_err("jfs_unlock failed with return code %d", rc);
598 static int jfs_get_sb(struct file_system_type *fs_type,
599 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
601 return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
605 static int jfs_sync_fs(struct super_block *sb, int wait)
607 struct jfs_log *log = JFS_SBI(sb)->log;
609 /* log == NULL indicates read-only mount */
611 jfs_flush_journal(log, wait);
618 static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
620 struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
623 seq_printf(seq, ",uid=%d", sbi->uid);
625 seq_printf(seq, ",gid=%d", sbi->gid);
626 if (sbi->umask != -1)
627 seq_printf(seq, ",umask=%03o", sbi->umask);
628 if (sbi->flag & JFS_NOINTEGRITY)
629 seq_puts(seq, ",nointegrity");
631 seq_printf(seq, ",iocharset=%s", sbi->nls_tab->charset);
632 if (sbi->flag & JFS_ERR_CONTINUE)
633 seq_printf(seq, ",errors=continue");
634 if (sbi->flag & JFS_ERR_PANIC)
635 seq_printf(seq, ",errors=panic");
638 if (sbi->flag & JFS_USRQUOTA)
639 seq_puts(seq, ",usrquota");
641 if (sbi->flag & JFS_GRPQUOTA)
642 seq_puts(seq, ",grpquota");
650 /* Read data from quotafile - avoid pagecache and such because we cannot afford
651 * acquiring the locks... As quota files are never truncated and quota code
652 * itself serializes the operations (and noone else should touch the files)
653 * we don't have to be afraid of races */
654 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
655 size_t len, loff_t off)
657 struct inode *inode = sb_dqopt(sb)->files[type];
658 sector_t blk = off >> sb->s_blocksize_bits;
660 int offset = off & (sb->s_blocksize - 1);
663 struct buffer_head tmp_bh;
664 struct buffer_head *bh;
665 loff_t i_size = i_size_read(inode);
669 if (off+len > i_size)
673 tocopy = sb->s_blocksize - offset < toread ?
674 sb->s_blocksize - offset : toread;
677 tmp_bh.b_size = 1 << inode->i_blkbits;
678 err = jfs_get_block(inode, blk, &tmp_bh, 0);
681 if (!buffer_mapped(&tmp_bh)) /* A hole? */
682 memset(data, 0, tocopy);
684 bh = sb_bread(sb, tmp_bh.b_blocknr);
687 memcpy(data, bh->b_data+offset, tocopy);
698 /* Write to quotafile */
699 static ssize_t jfs_quota_write(struct super_block *sb, int type,
700 const char *data, size_t len, loff_t off)
702 struct inode *inode = sb_dqopt(sb)->files[type];
703 sector_t blk = off >> sb->s_blocksize_bits;
705 int offset = off & (sb->s_blocksize - 1);
707 size_t towrite = len;
708 struct buffer_head tmp_bh;
709 struct buffer_head *bh;
711 mutex_lock(&inode->i_mutex);
712 while (towrite > 0) {
713 tocopy = sb->s_blocksize - offset < towrite ?
714 sb->s_blocksize - offset : towrite;
717 tmp_bh.b_size = 1 << inode->i_blkbits;
718 err = jfs_get_block(inode, blk, &tmp_bh, 1);
721 if (offset || tocopy != sb->s_blocksize)
722 bh = sb_bread(sb, tmp_bh.b_blocknr);
724 bh = sb_getblk(sb, tmp_bh.b_blocknr);
730 memcpy(bh->b_data+offset, data, tocopy);
731 flush_dcache_page(bh->b_page);
732 set_buffer_uptodate(bh);
733 mark_buffer_dirty(bh);
742 if (len == towrite) {
743 mutex_unlock(&inode->i_mutex);
746 if (inode->i_size < off+len-towrite)
747 i_size_write(inode, off+len-towrite);
749 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
750 mark_inode_dirty(inode);
751 mutex_unlock(&inode->i_mutex);
752 return len - towrite;
757 static const struct super_operations jfs_super_operations = {
758 .alloc_inode = jfs_alloc_inode,
759 .destroy_inode = jfs_destroy_inode,
760 .dirty_inode = jfs_dirty_inode,
761 .write_inode = jfs_write_inode,
762 .delete_inode = jfs_delete_inode,
763 .clear_inode = jfs_clear_inode,
764 .put_super = jfs_put_super,
765 .sync_fs = jfs_sync_fs,
766 .freeze_fs = jfs_freeze,
767 .unfreeze_fs = jfs_unfreeze,
768 .statfs = jfs_statfs,
769 .remount_fs = jfs_remount,
770 .show_options = jfs_show_options,
772 .quota_read = jfs_quota_read,
773 .quota_write = jfs_quota_write,
777 static const struct export_operations jfs_export_operations = {
778 .fh_to_dentry = jfs_fh_to_dentry,
779 .fh_to_parent = jfs_fh_to_parent,
780 .get_parent = jfs_get_parent,
783 static struct file_system_type jfs_fs_type = {
784 .owner = THIS_MODULE,
786 .get_sb = jfs_get_sb,
787 .kill_sb = kill_block_super,
788 .fs_flags = FS_REQUIRES_DEV,
791 static void init_once(void *foo)
793 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
795 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
796 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
797 init_rwsem(&jfs_ip->rdwrlock);
798 mutex_init(&jfs_ip->commit_mutex);
799 init_rwsem(&jfs_ip->xattr_sem);
800 spin_lock_init(&jfs_ip->ag_lock);
801 jfs_ip->active_ag = -1;
802 inode_init_once(&jfs_ip->vfs_inode);
805 static int __init init_jfs_fs(void)
811 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
812 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
814 if (jfs_inode_cachep == NULL)
818 * Metapage initialization
820 rc = metapage_init();
822 jfs_err("metapage_init failed w/rc = %d", rc);
827 * Transaction Manager initialization
831 jfs_err("txInit failed w/rc = %d", rc);
836 * I/O completion thread (endio)
838 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
839 if (IS_ERR(jfsIOthread)) {
840 rc = PTR_ERR(jfsIOthread);
841 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
845 if (commit_threads < 1)
846 commit_threads = num_online_cpus();
847 if (commit_threads > MAX_COMMIT_THREADS)
848 commit_threads = MAX_COMMIT_THREADS;
850 for (i = 0; i < commit_threads; i++) {
851 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
852 if (IS_ERR(jfsCommitThread[i])) {
853 rc = PTR_ERR(jfsCommitThread[i]);
854 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
856 goto kill_committask;
860 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
861 if (IS_ERR(jfsSyncThread)) {
862 rc = PTR_ERR(jfsSyncThread);
863 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
864 goto kill_committask;
871 return register_filesystem(&jfs_fs_type);
874 for (i = 0; i < commit_threads; i++)
875 kthread_stop(jfsCommitThread[i]);
876 kthread_stop(jfsIOthread);
882 kmem_cache_destroy(jfs_inode_cachep);
886 static void __exit exit_jfs_fs(void)
890 jfs_info("exit_jfs_fs called");
895 kthread_stop(jfsIOthread);
896 for (i = 0; i < commit_threads; i++)
897 kthread_stop(jfsCommitThread[i]);
898 kthread_stop(jfsSyncThread);
902 unregister_filesystem(&jfs_fs_type);
903 kmem_cache_destroy(jfs_inode_cachep);
906 module_init(init_jfs_fs)
907 module_exit(exit_jfs_fs)