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 <asm/uaccess.h>
34 #include <linux/seq_file.h>
36 #include "jfs_incore.h"
37 #include "jfs_filsys.h"
38 #include "jfs_inode.h"
39 #include "jfs_metapage.h"
40 #include "jfs_superblock.h"
44 #include "jfs_debug.h"
46 MODULE_DESCRIPTION("The Journaled Filesystem (JFS)");
47 MODULE_AUTHOR("Steve Best/Dave Kleikamp/Barry Arndt, IBM");
48 MODULE_LICENSE("GPL");
50 static struct kmem_cache * jfs_inode_cachep;
52 static const struct super_operations jfs_super_operations;
53 static const struct export_operations jfs_export_operations;
54 static struct file_system_type jfs_fs_type;
56 #define MAX_COMMIT_THREADS 64
57 static int commit_threads = 0;
58 module_param(commit_threads, int, 0);
59 MODULE_PARM_DESC(commit_threads, "Number of commit threads");
61 static struct task_struct *jfsCommitThread[MAX_COMMIT_THREADS];
62 struct task_struct *jfsIOthread;
63 struct task_struct *jfsSyncThread;
65 #ifdef CONFIG_JFS_DEBUG
66 int jfsloglevel = JFS_LOGLEVEL_WARN;
67 module_param(jfsloglevel, int, 0644);
68 MODULE_PARM_DESC(jfsloglevel, "Specify JFS loglevel (0, 1 or 2)");
71 static void jfs_handle_error(struct super_block *sb)
73 struct jfs_sb_info *sbi = JFS_SBI(sb);
75 if (sb->s_flags & MS_RDONLY)
78 updateSuper(sb, FM_DIRTY);
80 if (sbi->flag & JFS_ERR_PANIC)
81 panic("JFS (device %s): panic forced after error\n",
83 else if (sbi->flag & JFS_ERR_REMOUNT_RO) {
84 jfs_err("ERROR: (device %s): remounting filesystem "
87 sb->s_flags |= MS_RDONLY;
90 /* nothing is done for continue beyond marking the superblock dirty */
93 void jfs_error(struct super_block *sb, const char * function, ...)
95 static char error_buf[256];
98 va_start(args, function);
99 vsnprintf(error_buf, sizeof(error_buf), function, args);
102 printk(KERN_ERR "ERROR: (device %s): %s\n", sb->s_id, error_buf);
104 jfs_handle_error(sb);
107 static struct inode *jfs_alloc_inode(struct super_block *sb)
109 struct jfs_inode_info *jfs_inode;
111 jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
114 return &jfs_inode->vfs_inode;
117 static void jfs_destroy_inode(struct inode *inode)
119 struct jfs_inode_info *ji = JFS_IP(inode);
121 BUG_ON(!list_empty(&ji->anon_inode_list));
123 spin_lock_irq(&ji->ag_lock);
124 if (ji->active_ag != -1) {
125 struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
126 atomic_dec(&bmap->db_active[ji->active_ag]);
129 spin_unlock_irq(&ji->ag_lock);
131 #ifdef CONFIG_JFS_POSIX_ACL
132 if (ji->i_acl != JFS_ACL_NOT_CACHED) {
133 posix_acl_release(ji->i_acl);
134 ji->i_acl = JFS_ACL_NOT_CACHED;
136 if (ji->i_default_acl != JFS_ACL_NOT_CACHED) {
137 posix_acl_release(ji->i_default_acl);
138 ji->i_default_acl = JFS_ACL_NOT_CACHED;
142 kmem_cache_free(jfs_inode_cachep, ji);
145 static int jfs_statfs(struct dentry *dentry, struct kstatfs *buf)
147 struct jfs_sb_info *sbi = JFS_SBI(dentry->d_sb);
149 struct inomap *imap = JFS_IP(sbi->ipimap)->i_imap;
151 jfs_info("In jfs_statfs");
152 buf->f_type = JFS_SUPER_MAGIC;
153 buf->f_bsize = sbi->bsize;
154 buf->f_blocks = sbi->bmap->db_mapsize;
155 buf->f_bfree = sbi->bmap->db_nfree;
156 buf->f_bavail = sbi->bmap->db_nfree;
158 * If we really return the number of allocated & free inodes, some
159 * applications will fail because they won't see enough free inodes.
160 * We'll try to calculate some guess as to how may inodes we can
163 * buf->f_files = atomic_read(&imap->im_numinos);
164 * buf->f_ffree = atomic_read(&imap->im_numfree);
166 maxinodes = min((s64) atomic_read(&imap->im_numinos) +
167 ((sbi->bmap->db_nfree >> imap->im_l2nbperiext)
168 << L2INOSPEREXT), (s64) 0xffffffffLL);
169 buf->f_files = maxinodes;
170 buf->f_ffree = maxinodes - (atomic_read(&imap->im_numinos) -
171 atomic_read(&imap->im_numfree));
172 buf->f_fsid.val[0] = (u32)crc32_le(0, sbi->uuid, sizeof(sbi->uuid)/2);
173 buf->f_fsid.val[1] = (u32)crc32_le(0, sbi->uuid + sizeof(sbi->uuid)/2,
174 sizeof(sbi->uuid)/2);
176 buf->f_namelen = JFS_NAME_MAX;
180 static void jfs_put_super(struct super_block *sb)
182 struct jfs_sb_info *sbi = JFS_SBI(sb);
185 jfs_info("In jfs_put_super");
191 jfs_err("jfs_umount failed with return code %d", rc);
193 unload_nls(sbi->nls_tab);
196 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
197 iput(sbi->direct_inode);
198 sbi->direct_inode = NULL;
206 Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
207 Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
208 Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
211 static const match_table_t tokens = {
212 {Opt_integrity, "integrity"},
213 {Opt_nointegrity, "nointegrity"},
214 {Opt_iocharset, "iocharset=%s"},
215 {Opt_resize, "resize=%u"},
216 {Opt_resize_nosize, "resize"},
217 {Opt_errors, "errors=%s"},
218 {Opt_ignore, "noquota"},
219 {Opt_ignore, "quota"},
220 {Opt_usrquota, "usrquota"},
221 {Opt_grpquota, "grpquota"},
224 {Opt_umask, "umask=%u"},
228 static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
231 void *nls_map = (void *)-1; /* -1: no change; NULL: none */
233 struct jfs_sb_info *sbi = JFS_SBI(sb);
240 while ((p = strsep(&options, ",")) != NULL) {
241 substring_t args[MAX_OPT_ARGS];
246 token = match_token(p, tokens, args);
249 *flag &= ~JFS_NOINTEGRITY;
251 case Opt_nointegrity:
252 *flag |= JFS_NOINTEGRITY;
255 /* Silently ignore the quota options */
256 /* Don't do anything ;-) */
259 if (nls_map && nls_map != (void *) -1)
261 if (!strcmp(args[0].from, "none"))
264 nls_map = load_nls(args[0].from);
267 "JFS: charset not found\n");
274 char *resize = args[0].from;
275 *newLVSize = simple_strtoull(resize, &resize, 0);
278 case Opt_resize_nosize:
280 *newLVSize = sb->s_bdev->bd_inode->i_size >>
281 sb->s_blocksize_bits;
284 "JFS: Cannot determine volume size\n");
289 char *errors = args[0].from;
290 if (!errors || !*errors)
292 if (!strcmp(errors, "continue")) {
293 *flag &= ~JFS_ERR_REMOUNT_RO;
294 *flag &= ~JFS_ERR_PANIC;
295 *flag |= JFS_ERR_CONTINUE;
296 } else if (!strcmp(errors, "remount-ro")) {
297 *flag &= ~JFS_ERR_CONTINUE;
298 *flag &= ~JFS_ERR_PANIC;
299 *flag |= JFS_ERR_REMOUNT_RO;
300 } else if (!strcmp(errors, "panic")) {
301 *flag &= ~JFS_ERR_CONTINUE;
302 *flag &= ~JFS_ERR_REMOUNT_RO;
303 *flag |= JFS_ERR_PANIC;
306 "JFS: %s is an invalid error handler\n",
316 *flag |= JFS_USRQUOTA;
319 *flag |= JFS_GRPQUOTA;
326 "JFS: quota operations not supported\n");
331 char *uid = args[0].from;
332 sbi->uid = simple_strtoul(uid, &uid, 0);
337 char *gid = args[0].from;
338 sbi->gid = simple_strtoul(gid, &gid, 0);
343 char *umask = args[0].from;
344 sbi->umask = simple_strtoul(umask, &umask, 8);
345 if (sbi->umask & ~0777) {
347 "JFS: Invalid value of umask\n");
353 printk("jfs: Unrecognized mount option \"%s\" "
354 " or missing value\n", p);
359 if (nls_map != (void *) -1) {
360 /* Discard old (if remount) */
362 unload_nls(sbi->nls_tab);
363 sbi->nls_tab = nls_map;
368 if (nls_map && nls_map != (void *) -1)
373 static int jfs_remount(struct super_block *sb, int *flags, char *data)
377 int flag = JFS_SBI(sb)->flag;
379 if (!parse_options(data, sb, &newLVSize, &flag)) {
383 if (sb->s_flags & MS_RDONLY) {
385 "JFS: resize requires volume to be mounted read-write\n");
388 rc = jfs_extendfs(sb, newLVSize, 0);
393 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
395 * Invalidate any previously read metadata. fsck may have
396 * changed the on-disk data since we mounted r/o
398 truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
400 JFS_SBI(sb)->flag = flag;
401 return jfs_mount_rw(sb, 1);
403 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
404 rc = jfs_umount_rw(sb);
405 JFS_SBI(sb)->flag = flag;
408 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
409 if (!(sb->s_flags & MS_RDONLY)) {
410 rc = jfs_umount_rw(sb);
413 JFS_SBI(sb)->flag = flag;
414 return jfs_mount_rw(sb, 1);
416 JFS_SBI(sb)->flag = flag;
421 static int jfs_fill_super(struct super_block *sb, void *data, int silent)
423 struct jfs_sb_info *sbi;
427 int flag, ret = -EINVAL;
429 jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
431 if (!new_valid_dev(sb->s_bdev->bd_dev))
434 sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
439 sbi->uid = sbi->gid = sbi->umask = -1;
441 /* initialize the mount flag and determine the default error handler */
442 flag = JFS_ERR_REMOUNT_RO;
444 if (!parse_options((char *) data, sb, &newLVSize, &flag)) {
450 #ifdef CONFIG_JFS_POSIX_ACL
451 sb->s_flags |= MS_POSIXACL;
455 printk(KERN_ERR "resize option for remount only\n");
460 * Initialize blocksize to 4K.
462 sb_set_blocksize(sb, PSIZE);
465 * Set method vectors.
467 sb->s_op = &jfs_super_operations;
468 sb->s_export_op = &jfs_export_operations;
471 * Initialize direct-mapping inode/address-space
473 inode = new_inode(sb);
480 inode->i_size = sb->s_bdev->bd_inode->i_size;
481 inode->i_mapping->a_ops = &jfs_metapage_aops;
482 insert_inode_hash(inode);
483 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
485 sbi->direct_inode = inode;
490 jfs_err("jfs_mount failed w/return code = %d", rc);
492 goto out_mount_failed;
494 if (sb->s_flags & MS_RDONLY)
497 rc = jfs_mount_rw(sb, 0);
500 jfs_err("jfs_mount_rw failed, return code = %d",
507 sb->s_magic = JFS_SUPER_MAGIC;
509 inode = jfs_iget(sb, ROOT_I);
511 ret = PTR_ERR(inode);
514 sb->s_root = d_alloc_root(inode);
518 if (sbi->mntflag & JFS_OS2)
519 sb->s_root->d_op = &jfs_ci_dentry_operations;
521 /* logical blocks are represented by 40 bits in pxd_t, etc. */
522 sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
523 #if BITS_PER_LONG == 32
525 * Page cache is indexed by long.
526 * I would use MAX_LFS_FILESIZE, but it's only half as big
528 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, sb->s_maxbytes);
534 jfs_err("jfs_read_super: get root dentry failed");
540 jfs_err("jfs_umount failed with return code %d", rc);
543 filemap_write_and_wait(sbi->direct_inode->i_mapping);
544 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
545 make_bad_inode(sbi->direct_inode);
546 iput(sbi->direct_inode);
547 sbi->direct_inode = NULL;
550 unload_nls(sbi->nls_tab);
555 static int jfs_freeze(struct super_block *sb)
557 struct jfs_sb_info *sbi = JFS_SBI(sb);
558 struct jfs_log *log = sbi->log;
560 if (!(sb->s_flags & MS_RDONLY)) {
563 updateSuper(sb, FM_CLEAN);
568 static int jfs_unfreeze(struct super_block *sb)
570 struct jfs_sb_info *sbi = JFS_SBI(sb);
571 struct jfs_log *log = sbi->log;
574 if (!(sb->s_flags & MS_RDONLY)) {
575 updateSuper(sb, FM_MOUNT);
576 if ((rc = lmLogInit(log)))
577 jfs_err("jfs_unlock failed with return code %d", rc);
584 static int jfs_get_sb(struct file_system_type *fs_type,
585 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
587 return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
591 static int jfs_sync_fs(struct super_block *sb, int wait)
593 struct jfs_log *log = JFS_SBI(sb)->log;
595 /* log == NULL indicates read-only mount */
597 jfs_flush_journal(log, wait);
604 static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
606 struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
609 seq_printf(seq, ",uid=%d", sbi->uid);
611 seq_printf(seq, ",gid=%d", sbi->gid);
612 if (sbi->umask != -1)
613 seq_printf(seq, ",umask=%03o", sbi->umask);
614 if (sbi->flag & JFS_NOINTEGRITY)
615 seq_puts(seq, ",nointegrity");
617 seq_printf(seq, ",iocharset=%s", sbi->nls_tab->charset);
618 if (sbi->flag & JFS_ERR_CONTINUE)
619 seq_printf(seq, ",errors=continue");
620 if (sbi->flag & JFS_ERR_PANIC)
621 seq_printf(seq, ",errors=panic");
624 if (sbi->flag & JFS_USRQUOTA)
625 seq_puts(seq, ",usrquota");
627 if (sbi->flag & JFS_GRPQUOTA)
628 seq_puts(seq, ",grpquota");
636 /* Read data from quotafile - avoid pagecache and such because we cannot afford
637 * acquiring the locks... As quota files are never truncated and quota code
638 * itself serializes the operations (and noone else should touch the files)
639 * we don't have to be afraid of races */
640 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
641 size_t len, loff_t off)
643 struct inode *inode = sb_dqopt(sb)->files[type];
644 sector_t blk = off >> sb->s_blocksize_bits;
646 int offset = off & (sb->s_blocksize - 1);
649 struct buffer_head tmp_bh;
650 struct buffer_head *bh;
651 loff_t i_size = i_size_read(inode);
655 if (off+len > i_size)
659 tocopy = sb->s_blocksize - offset < toread ?
660 sb->s_blocksize - offset : toread;
663 tmp_bh.b_size = 1 << inode->i_blkbits;
664 err = jfs_get_block(inode, blk, &tmp_bh, 0);
667 if (!buffer_mapped(&tmp_bh)) /* A hole? */
668 memset(data, 0, tocopy);
670 bh = sb_bread(sb, tmp_bh.b_blocknr);
673 memcpy(data, bh->b_data+offset, tocopy);
684 /* Write to quotafile */
685 static ssize_t jfs_quota_write(struct super_block *sb, int type,
686 const char *data, size_t len, loff_t off)
688 struct inode *inode = sb_dqopt(sb)->files[type];
689 sector_t blk = off >> sb->s_blocksize_bits;
691 int offset = off & (sb->s_blocksize - 1);
693 size_t towrite = len;
694 struct buffer_head tmp_bh;
695 struct buffer_head *bh;
697 mutex_lock(&inode->i_mutex);
698 while (towrite > 0) {
699 tocopy = sb->s_blocksize - offset < towrite ?
700 sb->s_blocksize - offset : towrite;
703 tmp_bh.b_size = 1 << inode->i_blkbits;
704 err = jfs_get_block(inode, blk, &tmp_bh, 1);
707 if (offset || tocopy != sb->s_blocksize)
708 bh = sb_bread(sb, tmp_bh.b_blocknr);
710 bh = sb_getblk(sb, tmp_bh.b_blocknr);
716 memcpy(bh->b_data+offset, data, tocopy);
717 flush_dcache_page(bh->b_page);
718 set_buffer_uptodate(bh);
719 mark_buffer_dirty(bh);
728 if (len == towrite) {
729 mutex_unlock(&inode->i_mutex);
732 if (inode->i_size < off+len-towrite)
733 i_size_write(inode, off+len-towrite);
735 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
736 mark_inode_dirty(inode);
737 mutex_unlock(&inode->i_mutex);
738 return len - towrite;
743 static const struct super_operations jfs_super_operations = {
744 .alloc_inode = jfs_alloc_inode,
745 .destroy_inode = jfs_destroy_inode,
746 .dirty_inode = jfs_dirty_inode,
747 .write_inode = jfs_write_inode,
748 .delete_inode = jfs_delete_inode,
749 .put_super = jfs_put_super,
750 .sync_fs = jfs_sync_fs,
751 .freeze_fs = jfs_freeze,
752 .unfreeze_fs = jfs_unfreeze,
753 .statfs = jfs_statfs,
754 .remount_fs = jfs_remount,
755 .show_options = jfs_show_options,
757 .quota_read = jfs_quota_read,
758 .quota_write = jfs_quota_write,
762 static const struct export_operations jfs_export_operations = {
763 .fh_to_dentry = jfs_fh_to_dentry,
764 .fh_to_parent = jfs_fh_to_parent,
765 .get_parent = jfs_get_parent,
768 static struct file_system_type jfs_fs_type = {
769 .owner = THIS_MODULE,
771 .get_sb = jfs_get_sb,
772 .kill_sb = kill_block_super,
773 .fs_flags = FS_REQUIRES_DEV,
776 static void init_once(void *foo)
778 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
780 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
781 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
782 init_rwsem(&jfs_ip->rdwrlock);
783 mutex_init(&jfs_ip->commit_mutex);
784 init_rwsem(&jfs_ip->xattr_sem);
785 spin_lock_init(&jfs_ip->ag_lock);
786 jfs_ip->active_ag = -1;
787 #ifdef CONFIG_JFS_POSIX_ACL
788 jfs_ip->i_acl = JFS_ACL_NOT_CACHED;
789 jfs_ip->i_default_acl = JFS_ACL_NOT_CACHED;
791 inode_init_once(&jfs_ip->vfs_inode);
794 static int __init init_jfs_fs(void)
800 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
801 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
803 if (jfs_inode_cachep == NULL)
807 * Metapage initialization
809 rc = metapage_init();
811 jfs_err("metapage_init failed w/rc = %d", rc);
816 * Transaction Manager initialization
820 jfs_err("txInit failed w/rc = %d", rc);
825 * I/O completion thread (endio)
827 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
828 if (IS_ERR(jfsIOthread)) {
829 rc = PTR_ERR(jfsIOthread);
830 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
834 if (commit_threads < 1)
835 commit_threads = num_online_cpus();
836 if (commit_threads > MAX_COMMIT_THREADS)
837 commit_threads = MAX_COMMIT_THREADS;
839 for (i = 0; i < commit_threads; i++) {
840 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
841 if (IS_ERR(jfsCommitThread[i])) {
842 rc = PTR_ERR(jfsCommitThread[i]);
843 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
845 goto kill_committask;
849 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
850 if (IS_ERR(jfsSyncThread)) {
851 rc = PTR_ERR(jfsSyncThread);
852 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
853 goto kill_committask;
860 return register_filesystem(&jfs_fs_type);
863 for (i = 0; i < commit_threads; i++)
864 kthread_stop(jfsCommitThread[i]);
865 kthread_stop(jfsIOthread);
871 kmem_cache_destroy(jfs_inode_cachep);
875 static void __exit exit_jfs_fs(void)
879 jfs_info("exit_jfs_fs called");
884 kthread_stop(jfsIOthread);
885 for (i = 0; i < commit_threads; i++)
886 kthread_stop(jfsCommitThread[i]);
887 kthread_stop(jfsSyncThread);
891 unregister_filesystem(&jfs_fs_type);
892 kmem_cache_destroy(jfs_inode_cachep);
895 module_init(init_jfs_fs)
896 module_exit(exit_jfs_fs)