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/mount.h>
26 #include <linux/moduleparam.h>
27 #include <linux/kthread.h>
28 #include <linux/posix_acl.h>
29 #include <linux/buffer_head.h>
30 #include <asm/uaccess.h>
31 #include <linux/seq_file.h>
33 #include "jfs_incore.h"
34 #include "jfs_filsys.h"
35 #include "jfs_inode.h"
36 #include "jfs_metapage.h"
37 #include "jfs_superblock.h"
41 #include "jfs_debug.h"
43 MODULE_DESCRIPTION("The Journaled Filesystem (JFS)");
44 MODULE_AUTHOR("Steve Best/Dave Kleikamp/Barry Arndt, IBM");
45 MODULE_LICENSE("GPL");
47 static struct kmem_cache * jfs_inode_cachep;
49 static struct super_operations jfs_super_operations;
50 static struct export_operations jfs_export_operations;
51 static struct file_system_type jfs_fs_type;
53 #define MAX_COMMIT_THREADS 64
54 static int commit_threads = 0;
55 module_param(commit_threads, int, 0);
56 MODULE_PARM_DESC(commit_threads, "Number of commit threads");
58 static struct task_struct *jfsCommitThread[MAX_COMMIT_THREADS];
59 struct task_struct *jfsIOthread;
60 struct task_struct *jfsSyncThread;
62 #ifdef CONFIG_JFS_DEBUG
63 int jfsloglevel = JFS_LOGLEVEL_WARN;
64 module_param(jfsloglevel, int, 0644);
65 MODULE_PARM_DESC(jfsloglevel, "Specify JFS loglevel (0, 1 or 2)");
68 static void jfs_handle_error(struct super_block *sb)
70 struct jfs_sb_info *sbi = JFS_SBI(sb);
72 if (sb->s_flags & MS_RDONLY)
75 updateSuper(sb, FM_DIRTY);
77 if (sbi->flag & JFS_ERR_PANIC)
78 panic("JFS (device %s): panic forced after error\n",
80 else if (sbi->flag & JFS_ERR_REMOUNT_RO) {
81 jfs_err("ERROR: (device %s): remounting filesystem "
84 sb->s_flags |= MS_RDONLY;
87 /* nothing is done for continue beyond marking the superblock dirty */
90 void jfs_error(struct super_block *sb, const char * function, ...)
92 static char error_buf[256];
95 va_start(args, function);
96 vsnprintf(error_buf, sizeof(error_buf), function, args);
99 printk(KERN_ERR "ERROR: (device %s): %s\n", sb->s_id, error_buf);
101 jfs_handle_error(sb);
104 static struct inode *jfs_alloc_inode(struct super_block *sb)
106 struct jfs_inode_info *jfs_inode;
108 jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
111 return &jfs_inode->vfs_inode;
114 static void jfs_destroy_inode(struct inode *inode)
116 struct jfs_inode_info *ji = JFS_IP(inode);
118 BUG_ON(!list_empty(&ji->anon_inode_list));
120 spin_lock_irq(&ji->ag_lock);
121 if (ji->active_ag != -1) {
122 struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
123 atomic_dec(&bmap->db_active[ji->active_ag]);
126 spin_unlock_irq(&ji->ag_lock);
128 #ifdef CONFIG_JFS_POSIX_ACL
129 if (ji->i_acl != JFS_ACL_NOT_CACHED) {
130 posix_acl_release(ji->i_acl);
131 ji->i_acl = JFS_ACL_NOT_CACHED;
133 if (ji->i_default_acl != JFS_ACL_NOT_CACHED) {
134 posix_acl_release(ji->i_default_acl);
135 ji->i_default_acl = JFS_ACL_NOT_CACHED;
139 kmem_cache_free(jfs_inode_cachep, ji);
142 static int jfs_statfs(struct dentry *dentry, struct kstatfs *buf)
144 struct jfs_sb_info *sbi = JFS_SBI(dentry->d_sb);
146 struct inomap *imap = JFS_IP(sbi->ipimap)->i_imap;
148 jfs_info("In jfs_statfs");
149 buf->f_type = JFS_SUPER_MAGIC;
150 buf->f_bsize = sbi->bsize;
151 buf->f_blocks = sbi->bmap->db_mapsize;
152 buf->f_bfree = sbi->bmap->db_nfree;
153 buf->f_bavail = sbi->bmap->db_nfree;
155 * If we really return the number of allocated & free inodes, some
156 * applications will fail because they won't see enough free inodes.
157 * We'll try to calculate some guess as to how may inodes we can
160 * buf->f_files = atomic_read(&imap->im_numinos);
161 * buf->f_ffree = atomic_read(&imap->im_numfree);
163 maxinodes = min((s64) atomic_read(&imap->im_numinos) +
164 ((sbi->bmap->db_nfree >> imap->im_l2nbperiext)
165 << L2INOSPEREXT), (s64) 0xffffffffLL);
166 buf->f_files = maxinodes;
167 buf->f_ffree = maxinodes - (atomic_read(&imap->im_numinos) -
168 atomic_read(&imap->im_numfree));
170 buf->f_namelen = JFS_NAME_MAX;
174 static void jfs_put_super(struct super_block *sb)
176 struct jfs_sb_info *sbi = JFS_SBI(sb);
179 jfs_info("In jfs_put_super");
182 jfs_err("jfs_umount failed with return code %d", rc);
184 unload_nls(sbi->nls_tab);
187 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
188 iput(sbi->direct_inode);
189 sbi->direct_inode = NULL;
195 Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
196 Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
197 Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask,
198 Opt_tag, Opt_notag, Opt_tagid
201 static match_table_t tokens = {
202 {Opt_integrity, "integrity"},
203 {Opt_nointegrity, "nointegrity"},
204 {Opt_iocharset, "iocharset=%s"},
205 {Opt_resize, "resize=%u"},
206 {Opt_resize_nosize, "resize"},
207 {Opt_errors, "errors=%s"},
209 {Opt_notag, "notag"},
210 {Opt_tagid, "tagid=%u"},
212 {Opt_ignore, "noquota"},
213 {Opt_ignore, "quota"},
214 {Opt_usrquota, "usrquota"},
215 {Opt_grpquota, "grpquota"},
218 {Opt_umask, "umask=%u"},
222 static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
225 void *nls_map = (void *)-1; /* -1: no change; NULL: none */
227 struct jfs_sb_info *sbi = JFS_SBI(sb);
234 while ((p = strsep(&options, ",")) != NULL) {
235 substring_t args[MAX_OPT_ARGS];
240 token = match_token(p, tokens, args);
243 *flag &= ~JFS_NOINTEGRITY;
245 case Opt_nointegrity:
246 *flag |= JFS_NOINTEGRITY;
249 /* Silently ignore the quota options */
250 /* Don't do anything ;-) */
253 if (nls_map && nls_map != (void *) -1)
255 if (!strcmp(args[0].from, "none"))
258 nls_map = load_nls(args[0].from);
261 "JFS: charset not found\n");
268 char *resize = args[0].from;
269 *newLVSize = simple_strtoull(resize, &resize, 0);
272 case Opt_resize_nosize:
274 *newLVSize = sb->s_bdev->bd_inode->i_size >>
275 sb->s_blocksize_bits;
278 "JFS: Cannot determine volume size\n");
283 char *errors = args[0].from;
284 if (!errors || !*errors)
286 if (!strcmp(errors, "continue")) {
287 *flag &= ~JFS_ERR_REMOUNT_RO;
288 *flag &= ~JFS_ERR_PANIC;
289 *flag |= JFS_ERR_CONTINUE;
290 } else if (!strcmp(errors, "remount-ro")) {
291 *flag &= ~JFS_ERR_CONTINUE;
292 *flag &= ~JFS_ERR_PANIC;
293 *flag |= JFS_ERR_REMOUNT_RO;
294 } else if (!strcmp(errors, "panic")) {
295 *flag &= ~JFS_ERR_CONTINUE;
296 *flag &= ~JFS_ERR_REMOUNT_RO;
297 *flag |= JFS_ERR_PANIC;
300 "JFS: %s is an invalid error handler\n",
310 *flag |= JFS_USRQUOTA;
313 *flag |= JFS_GRPQUOTA;
320 "JFS: quota operations not supported\n");
325 char *uid = args[0].from;
326 sbi->uid = simple_strtoul(uid, &uid, 0);
331 char *gid = args[0].from;
332 sbi->gid = simple_strtoul(gid, &gid, 0);
337 char *umask = args[0].from;
338 sbi->umask = simple_strtoul(umask, &umask, 8);
339 if (sbi->umask & ~0777) {
341 "JFS: Invalid value of umask\n");
346 #ifndef CONFIG_TAGGING_NONE
354 #ifdef CONFIG_PROPAGATE
361 printk("jfs: Unrecognized mount option \"%s\" "
362 " or missing value\n", p);
367 if (nls_map != (void *) -1) {
368 /* Discard old (if remount) */
370 unload_nls(sbi->nls_tab);
371 sbi->nls_tab = nls_map;
376 if (nls_map && nls_map != (void *) -1)
381 static int jfs_remount(struct super_block *sb, int *flags, char *data)
385 int flag = JFS_SBI(sb)->flag;
387 if (!parse_options(data, sb, &newLVSize, &flag)) {
391 if ((flag & JFS_TAGGED) && !(sb->s_flags & MS_TAGGED)) {
392 printk(KERN_ERR "JFS: %s: tagging not permitted on remount.\n",
398 if (sb->s_flags & MS_RDONLY) {
400 "JFS: resize requires volume to be mounted read-write\n");
403 rc = jfs_extendfs(sb, newLVSize, 0);
408 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
410 * Invalidate any previously read metadata. fsck may have
411 * changed the on-disk data since we mounted r/o
413 truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
415 JFS_SBI(sb)->flag = flag;
416 return jfs_mount_rw(sb, 1);
418 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
419 rc = jfs_umount_rw(sb);
420 JFS_SBI(sb)->flag = flag;
423 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
424 if (!(sb->s_flags & MS_RDONLY)) {
425 rc = jfs_umount_rw(sb);
428 JFS_SBI(sb)->flag = flag;
429 return jfs_mount_rw(sb, 1);
431 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;
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)) {
465 #ifdef CONFIG_JFS_POSIX_ACL
466 sb->s_flags |= MS_POSIXACL;
468 /* map mount option tagxid */
469 if (sbi->flag & JFS_TAGGED)
470 sb->s_flags |= MS_TAGGED;
473 printk(KERN_ERR "resize option for remount only\n");
478 * Initialize blocksize to 4K.
480 sb_set_blocksize(sb, PSIZE);
483 * Set method vectors.
485 sb->s_op = &jfs_super_operations;
486 sb->s_export_op = &jfs_export_operations;
489 * Initialize direct-mapping inode/address-space
491 inode = new_inode(sb);
496 inode->i_size = sb->s_bdev->bd_inode->i_size;
497 inode->i_mapping->a_ops = &jfs_metapage_aops;
498 insert_inode_hash(inode);
499 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
501 sbi->direct_inode = inode;
506 jfs_err("jfs_mount failed w/return code = %d", rc);
508 goto out_mount_failed;
510 if (sb->s_flags & MS_RDONLY)
513 rc = jfs_mount_rw(sb, 0);
516 jfs_err("jfs_mount_rw failed, return code = %d",
523 sb->s_magic = JFS_SUPER_MAGIC;
525 inode = iget(sb, ROOT_I);
526 if (!inode || is_bad_inode(inode))
528 sb->s_root = d_alloc_root(inode);
532 if (sbi->mntflag & JFS_OS2)
533 sb->s_root->d_op = &jfs_ci_dentry_operations;
535 /* logical blocks are represented by 40 bits in pxd_t, etc. */
536 sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
537 #if BITS_PER_LONG == 32
539 * Page cache is indexed by long.
540 * I would use MAX_LFS_FILESIZE, but it's only half as big
542 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, sb->s_maxbytes);
548 jfs_err("jfs_read_super: get root inode failed");
555 jfs_err("jfs_umount failed with return code %d", rc);
558 filemap_write_and_wait(sbi->direct_inode->i_mapping);
559 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
560 make_bad_inode(sbi->direct_inode);
561 iput(sbi->direct_inode);
562 sbi->direct_inode = NULL;
565 unload_nls(sbi->nls_tab);
570 static void jfs_write_super_lockfs(struct super_block *sb)
572 struct jfs_sb_info *sbi = JFS_SBI(sb);
573 struct jfs_log *log = sbi->log;
575 if (!(sb->s_flags & MS_RDONLY)) {
578 updateSuper(sb, FM_CLEAN);
582 static void jfs_unlockfs(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);
597 static int jfs_get_sb(struct file_system_type *fs_type,
598 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
600 return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
604 static int jfs_sync_fs(struct super_block *sb, int wait)
606 struct jfs_log *log = JFS_SBI(sb)->log;
608 /* log == NULL indicates read-only mount */
610 jfs_flush_journal(log, wait);
617 static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
619 struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
622 seq_printf(seq, ",uid=%d", sbi->uid);
624 seq_printf(seq, ",gid=%d", sbi->gid);
625 if (sbi->umask != -1)
626 seq_printf(seq, ",umask=%03o", sbi->umask);
627 if (sbi->flag & JFS_NOINTEGRITY)
628 seq_puts(seq, ",nointegrity");
631 if (sbi->flag & JFS_USRQUOTA)
632 seq_puts(seq, ",usrquota");
634 if (sbi->flag & JFS_GRPQUOTA)
635 seq_puts(seq, ",grpquota");
643 /* Read data from quotafile - avoid pagecache and such because we cannot afford
644 * acquiring the locks... As quota files are never truncated and quota code
645 * itself serializes the operations (and noone else should touch the files)
646 * we don't have to be afraid of races */
647 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
648 size_t len, loff_t off)
650 struct inode *inode = sb_dqopt(sb)->files[type];
651 sector_t blk = off >> sb->s_blocksize_bits;
653 int offset = off & (sb->s_blocksize - 1);
656 struct buffer_head tmp_bh;
657 struct buffer_head *bh;
658 loff_t i_size = i_size_read(inode);
662 if (off+len > i_size)
666 tocopy = sb->s_blocksize - offset < toread ?
667 sb->s_blocksize - offset : toread;
670 tmp_bh.b_size = 1 << inode->i_blkbits;
671 err = jfs_get_block(inode, blk, &tmp_bh, 0);
674 if (!buffer_mapped(&tmp_bh)) /* A hole? */
675 memset(data, 0, tocopy);
677 bh = sb_bread(sb, tmp_bh.b_blocknr);
680 memcpy(data, bh->b_data+offset, tocopy);
691 /* Write to quotafile */
692 static ssize_t jfs_quota_write(struct super_block *sb, int type,
693 const char *data, size_t len, loff_t off)
695 struct inode *inode = sb_dqopt(sb)->files[type];
696 sector_t blk = off >> sb->s_blocksize_bits;
698 int offset = off & (sb->s_blocksize - 1);
700 size_t towrite = len;
701 struct buffer_head tmp_bh;
702 struct buffer_head *bh;
704 mutex_lock(&inode->i_mutex);
705 while (towrite > 0) {
706 tocopy = sb->s_blocksize - offset < towrite ?
707 sb->s_blocksize - offset : towrite;
710 tmp_bh.b_size = 1 << inode->i_blkbits;
711 err = jfs_get_block(inode, blk, &tmp_bh, 1);
714 if (offset || tocopy != sb->s_blocksize)
715 bh = sb_bread(sb, tmp_bh.b_blocknr);
717 bh = sb_getblk(sb, tmp_bh.b_blocknr);
723 memcpy(bh->b_data+offset, data, tocopy);
724 flush_dcache_page(bh->b_page);
725 set_buffer_uptodate(bh);
726 mark_buffer_dirty(bh);
737 if (inode->i_size < off+len-towrite)
738 i_size_write(inode, off+len-towrite);
740 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
741 mark_inode_dirty(inode);
742 mutex_unlock(&inode->i_mutex);
743 return len - towrite;
748 static struct super_operations jfs_super_operations = {
749 .alloc_inode = jfs_alloc_inode,
750 .destroy_inode = jfs_destroy_inode,
751 .read_inode = jfs_read_inode,
752 .dirty_inode = jfs_dirty_inode,
753 .write_inode = jfs_write_inode,
754 .delete_inode = jfs_delete_inode,
755 .put_super = jfs_put_super,
756 .sync_fs = jfs_sync_fs,
757 .write_super_lockfs = jfs_write_super_lockfs,
758 .unlockfs = jfs_unlockfs,
759 .statfs = jfs_statfs,
760 .remount_fs = jfs_remount,
761 .show_options = jfs_show_options,
763 .quota_read = jfs_quota_read,
764 .quota_write = jfs_quota_write,
768 static struct export_operations jfs_export_operations = {
769 .get_parent = jfs_get_parent,
772 static struct file_system_type jfs_fs_type = {
773 .owner = THIS_MODULE,
775 .get_sb = jfs_get_sb,
776 .kill_sb = kill_block_super,
777 .fs_flags = FS_REQUIRES_DEV,
780 static void init_once(void *foo, struct kmem_cache * cachep, unsigned long flags)
782 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
784 if ((flags & (SLAB_CTOR_VERIFY | SLAB_CTOR_CONSTRUCTOR)) ==
785 SLAB_CTOR_CONSTRUCTOR) {
786 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
787 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
788 init_rwsem(&jfs_ip->rdwrlock);
789 mutex_init(&jfs_ip->commit_mutex);
790 init_rwsem(&jfs_ip->xattr_sem);
791 spin_lock_init(&jfs_ip->ag_lock);
792 jfs_ip->active_ag = -1;
793 #ifdef CONFIG_JFS_POSIX_ACL
794 jfs_ip->i_acl = JFS_ACL_NOT_CACHED;
795 jfs_ip->i_default_acl = JFS_ACL_NOT_CACHED;
797 inode_init_once(&jfs_ip->vfs_inode);
801 static int __init init_jfs_fs(void)
807 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
808 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
810 if (jfs_inode_cachep == NULL)
814 * Metapage initialization
816 rc = metapage_init();
818 jfs_err("metapage_init failed w/rc = %d", rc);
823 * Transaction Manager initialization
827 jfs_err("txInit failed w/rc = %d", rc);
832 * I/O completion thread (endio)
834 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
835 if (IS_ERR(jfsIOthread)) {
836 rc = PTR_ERR(jfsIOthread);
837 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
841 if (commit_threads < 1)
842 commit_threads = num_online_cpus();
843 if (commit_threads > MAX_COMMIT_THREADS)
844 commit_threads = MAX_COMMIT_THREADS;
846 for (i = 0; i < commit_threads; i++) {
847 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
848 if (IS_ERR(jfsCommitThread[i])) {
849 rc = PTR_ERR(jfsCommitThread[i]);
850 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
852 goto kill_committask;
856 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
857 if (IS_ERR(jfsSyncThread)) {
858 rc = PTR_ERR(jfsSyncThread);
859 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
860 goto kill_committask;
867 return register_filesystem(&jfs_fs_type);
870 for (i = 0; i < commit_threads; i++)
871 kthread_stop(jfsCommitThread[i]);
872 kthread_stop(jfsIOthread);
878 kmem_cache_destroy(jfs_inode_cachep);
882 static void __exit exit_jfs_fs(void)
886 jfs_info("exit_jfs_fs called");
891 kthread_stop(jfsIOthread);
892 for (i = 0; i < commit_threads; i++)
893 kthread_stop(jfsCommitThread[i]);
894 kthread_stop(jfsSyncThread);
898 unregister_filesystem(&jfs_fs_type);
899 kmem_cache_destroy(jfs_inode_cachep);
902 module_init(init_jfs_fs)
903 module_exit(exit_jfs_fs)