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 <linux/exportfs.h>
31 #include <asm/uaccess.h>
32 #include <linux/seq_file.h>
34 #include "jfs_incore.h"
35 #include "jfs_filsys.h"
36 #include "jfs_inode.h"
37 #include "jfs_metapage.h"
38 #include "jfs_superblock.h"
42 #include "jfs_debug.h"
44 MODULE_DESCRIPTION("The Journaled Filesystem (JFS)");
45 MODULE_AUTHOR("Steve Best/Dave Kleikamp/Barry Arndt, IBM");
46 MODULE_LICENSE("GPL");
48 static struct kmem_cache * jfs_inode_cachep;
50 static const struct super_operations jfs_super_operations;
51 static const struct export_operations jfs_export_operations;
52 static struct file_system_type jfs_fs_type;
54 #define MAX_COMMIT_THREADS 64
55 static int commit_threads = 0;
56 module_param(commit_threads, int, 0);
57 MODULE_PARM_DESC(commit_threads, "Number of commit threads");
59 static struct task_struct *jfsCommitThread[MAX_COMMIT_THREADS];
60 struct task_struct *jfsIOthread;
61 struct task_struct *jfsSyncThread;
63 #ifdef CONFIG_JFS_DEBUG
64 int jfsloglevel = JFS_LOGLEVEL_WARN;
65 module_param(jfsloglevel, int, 0644);
66 MODULE_PARM_DESC(jfsloglevel, "Specify JFS loglevel (0, 1 or 2)");
69 static void jfs_handle_error(struct super_block *sb)
71 struct jfs_sb_info *sbi = JFS_SBI(sb);
73 if (sb->s_flags & MS_RDONLY)
76 updateSuper(sb, FM_DIRTY);
78 if (sbi->flag & JFS_ERR_PANIC)
79 panic("JFS (device %s): panic forced after error\n",
81 else if (sbi->flag & JFS_ERR_REMOUNT_RO) {
82 jfs_err("ERROR: (device %s): remounting filesystem "
85 sb->s_flags |= MS_RDONLY;
88 /* nothing is done for continue beyond marking the superblock dirty */
91 void jfs_error(struct super_block *sb, const char * function, ...)
93 static char error_buf[256];
96 va_start(args, function);
97 vsnprintf(error_buf, sizeof(error_buf), function, args);
100 printk(KERN_ERR "ERROR: (device %s): %s\n", sb->s_id, error_buf);
102 jfs_handle_error(sb);
105 static struct inode *jfs_alloc_inode(struct super_block *sb)
107 struct jfs_inode_info *jfs_inode;
109 jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
112 return &jfs_inode->vfs_inode;
115 static void jfs_destroy_inode(struct inode *inode)
117 struct jfs_inode_info *ji = JFS_IP(inode);
119 BUG_ON(!list_empty(&ji->anon_inode_list));
121 spin_lock_irq(&ji->ag_lock);
122 if (ji->active_ag != -1) {
123 struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
124 atomic_dec(&bmap->db_active[ji->active_ag]);
127 spin_unlock_irq(&ji->ag_lock);
129 #ifdef CONFIG_JFS_POSIX_ACL
130 if (ji->i_acl != JFS_ACL_NOT_CACHED) {
131 posix_acl_release(ji->i_acl);
132 ji->i_acl = JFS_ACL_NOT_CACHED;
134 if (ji->i_default_acl != JFS_ACL_NOT_CACHED) {
135 posix_acl_release(ji->i_default_acl);
136 ji->i_default_acl = JFS_ACL_NOT_CACHED;
140 kmem_cache_free(jfs_inode_cachep, ji);
143 static int jfs_statfs(struct dentry *dentry, struct kstatfs *buf)
145 struct jfs_sb_info *sbi = JFS_SBI(dentry->d_sb);
147 struct inomap *imap = JFS_IP(sbi->ipimap)->i_imap;
149 jfs_info("In jfs_statfs");
150 buf->f_type = JFS_SUPER_MAGIC;
151 buf->f_bsize = sbi->bsize;
152 buf->f_blocks = sbi->bmap->db_mapsize;
153 buf->f_bfree = sbi->bmap->db_nfree;
154 buf->f_bavail = sbi->bmap->db_nfree;
156 * If we really return the number of allocated & free inodes, some
157 * applications will fail because they won't see enough free inodes.
158 * We'll try to calculate some guess as to how may inodes we can
161 * buf->f_files = atomic_read(&imap->im_numinos);
162 * buf->f_ffree = atomic_read(&imap->im_numfree);
164 maxinodes = min((s64) atomic_read(&imap->im_numinos) +
165 ((sbi->bmap->db_nfree >> imap->im_l2nbperiext)
166 << L2INOSPEREXT), (s64) 0xffffffffLL);
167 buf->f_files = maxinodes;
168 buf->f_ffree = maxinodes - (atomic_read(&imap->im_numinos) -
169 atomic_read(&imap->im_numfree));
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");
183 jfs_err("jfs_umount failed with return code %d", rc);
185 unload_nls(sbi->nls_tab);
188 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
189 iput(sbi->direct_inode);
190 sbi->direct_inode = NULL;
196 Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
197 Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
198 Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
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"},
208 {Opt_ignore, "noquota"},
209 {Opt_ignore, "quota"},
210 {Opt_usrquota, "usrquota"},
211 {Opt_grpquota, "grpquota"},
214 {Opt_umask, "umask=%u"},
218 static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
221 void *nls_map = (void *)-1; /* -1: no change; NULL: none */
223 struct jfs_sb_info *sbi = JFS_SBI(sb);
230 while ((p = strsep(&options, ",")) != NULL) {
231 substring_t args[MAX_OPT_ARGS];
236 token = match_token(p, tokens, args);
239 *flag &= ~JFS_NOINTEGRITY;
241 case Opt_nointegrity:
242 *flag |= JFS_NOINTEGRITY;
245 /* Silently ignore the quota options */
246 /* Don't do anything ;-) */
249 if (nls_map && nls_map != (void *) -1)
251 if (!strcmp(args[0].from, "none"))
254 nls_map = load_nls(args[0].from);
257 "JFS: charset not found\n");
264 char *resize = args[0].from;
265 *newLVSize = simple_strtoull(resize, &resize, 0);
268 case Opt_resize_nosize:
270 *newLVSize = sb->s_bdev->bd_inode->i_size >>
271 sb->s_blocksize_bits;
274 "JFS: Cannot determine volume size\n");
279 char *errors = args[0].from;
280 if (!errors || !*errors)
282 if (!strcmp(errors, "continue")) {
283 *flag &= ~JFS_ERR_REMOUNT_RO;
284 *flag &= ~JFS_ERR_PANIC;
285 *flag |= JFS_ERR_CONTINUE;
286 } else if (!strcmp(errors, "remount-ro")) {
287 *flag &= ~JFS_ERR_CONTINUE;
288 *flag &= ~JFS_ERR_PANIC;
289 *flag |= JFS_ERR_REMOUNT_RO;
290 } else if (!strcmp(errors, "panic")) {
291 *flag &= ~JFS_ERR_CONTINUE;
292 *flag &= ~JFS_ERR_REMOUNT_RO;
293 *flag |= JFS_ERR_PANIC;
296 "JFS: %s is an invalid error handler\n",
306 *flag |= JFS_USRQUOTA;
309 *flag |= JFS_GRPQUOTA;
316 "JFS: quota operations not supported\n");
321 char *uid = args[0].from;
322 sbi->uid = simple_strtoul(uid, &uid, 0);
327 char *gid = args[0].from;
328 sbi->gid = simple_strtoul(gid, &gid, 0);
333 char *umask = args[0].from;
334 sbi->umask = simple_strtoul(umask, &umask, 8);
335 if (sbi->umask & ~0777) {
337 "JFS: Invalid value of umask\n");
343 printk("jfs: Unrecognized mount option \"%s\" "
344 " or missing value\n", p);
349 if (nls_map != (void *) -1) {
350 /* Discard old (if remount) */
352 unload_nls(sbi->nls_tab);
353 sbi->nls_tab = nls_map;
358 if (nls_map && nls_map != (void *) -1)
363 static int jfs_remount(struct super_block *sb, int *flags, char *data)
367 int flag = JFS_SBI(sb)->flag;
369 if (!parse_options(data, sb, &newLVSize, &flag)) {
373 if (sb->s_flags & MS_RDONLY) {
375 "JFS: resize requires volume to be mounted read-write\n");
378 rc = jfs_extendfs(sb, newLVSize, 0);
383 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
385 * Invalidate any previously read metadata. fsck may have
386 * changed the on-disk data since we mounted r/o
388 truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
390 JFS_SBI(sb)->flag = flag;
391 return jfs_mount_rw(sb, 1);
393 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
394 rc = jfs_umount_rw(sb);
395 JFS_SBI(sb)->flag = flag;
398 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
399 if (!(sb->s_flags & MS_RDONLY)) {
400 rc = jfs_umount_rw(sb);
403 JFS_SBI(sb)->flag = flag;
404 return jfs_mount_rw(sb, 1);
406 JFS_SBI(sb)->flag = flag;
411 static int jfs_fill_super(struct super_block *sb, void *data, int silent)
413 struct jfs_sb_info *sbi;
417 int flag, ret = -EINVAL;
419 jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
421 if (!new_valid_dev(sb->s_bdev->bd_dev))
424 sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
429 sbi->uid = sbi->gid = sbi->umask = -1;
431 /* initialize the mount flag and determine the default error handler */
432 flag = JFS_ERR_REMOUNT_RO;
434 if (!parse_options((char *) data, sb, &newLVSize, &flag)) {
440 #ifdef CONFIG_JFS_POSIX_ACL
441 sb->s_flags |= MS_POSIXACL;
445 printk(KERN_ERR "resize option for remount only\n");
450 * Initialize blocksize to 4K.
452 sb_set_blocksize(sb, PSIZE);
455 * Set method vectors.
457 sb->s_op = &jfs_super_operations;
458 sb->s_export_op = &jfs_export_operations;
461 * Initialize direct-mapping inode/address-space
463 inode = new_inode(sb);
470 inode->i_size = sb->s_bdev->bd_inode->i_size;
471 inode->i_mapping->a_ops = &jfs_metapage_aops;
472 insert_inode_hash(inode);
473 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
475 sbi->direct_inode = inode;
480 jfs_err("jfs_mount failed w/return code = %d", rc);
482 goto out_mount_failed;
484 if (sb->s_flags & MS_RDONLY)
487 rc = jfs_mount_rw(sb, 0);
490 jfs_err("jfs_mount_rw failed, return code = %d",
497 sb->s_magic = JFS_SUPER_MAGIC;
499 inode = jfs_iget(sb, ROOT_I);
501 ret = PTR_ERR(inode);
504 sb->s_root = d_alloc_root(inode);
508 if (sbi->mntflag & JFS_OS2)
509 sb->s_root->d_op = &jfs_ci_dentry_operations;
511 /* logical blocks are represented by 40 bits in pxd_t, etc. */
512 sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
513 #if BITS_PER_LONG == 32
515 * Page cache is indexed by long.
516 * I would use MAX_LFS_FILESIZE, but it's only half as big
518 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, sb->s_maxbytes);
524 jfs_err("jfs_read_super: get root dentry failed");
530 jfs_err("jfs_umount failed with return code %d", rc);
533 filemap_write_and_wait(sbi->direct_inode->i_mapping);
534 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
535 make_bad_inode(sbi->direct_inode);
536 iput(sbi->direct_inode);
537 sbi->direct_inode = NULL;
540 unload_nls(sbi->nls_tab);
545 static void jfs_write_super_lockfs(struct super_block *sb)
547 struct jfs_sb_info *sbi = JFS_SBI(sb);
548 struct jfs_log *log = sbi->log;
550 if (!(sb->s_flags & MS_RDONLY)) {
553 updateSuper(sb, FM_CLEAN);
557 static void jfs_unlockfs(struct super_block *sb)
559 struct jfs_sb_info *sbi = JFS_SBI(sb);
560 struct jfs_log *log = sbi->log;
563 if (!(sb->s_flags & MS_RDONLY)) {
564 updateSuper(sb, FM_MOUNT);
565 if ((rc = lmLogInit(log)))
566 jfs_err("jfs_unlock failed with return code %d", rc);
572 static int jfs_get_sb(struct file_system_type *fs_type,
573 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
575 return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
579 static int jfs_sync_fs(struct super_block *sb, int wait)
581 struct jfs_log *log = JFS_SBI(sb)->log;
583 /* log == NULL indicates read-only mount */
585 jfs_flush_journal(log, wait);
592 static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
594 struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
597 seq_printf(seq, ",uid=%d", sbi->uid);
599 seq_printf(seq, ",gid=%d", sbi->gid);
600 if (sbi->umask != -1)
601 seq_printf(seq, ",umask=%03o", sbi->umask);
602 if (sbi->flag & JFS_NOINTEGRITY)
603 seq_puts(seq, ",nointegrity");
605 seq_printf(seq, ",iocharset=%s", sbi->nls_tab->charset);
606 if (sbi->flag & JFS_ERR_CONTINUE)
607 seq_printf(seq, ",errors=continue");
608 if (sbi->flag & JFS_ERR_PANIC)
609 seq_printf(seq, ",errors=panic");
612 if (sbi->flag & JFS_USRQUOTA)
613 seq_puts(seq, ",usrquota");
615 if (sbi->flag & JFS_GRPQUOTA)
616 seq_puts(seq, ",grpquota");
624 /* Read data from quotafile - avoid pagecache and such because we cannot afford
625 * acquiring the locks... As quota files are never truncated and quota code
626 * itself serializes the operations (and noone else should touch the files)
627 * we don't have to be afraid of races */
628 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
629 size_t len, loff_t off)
631 struct inode *inode = sb_dqopt(sb)->files[type];
632 sector_t blk = off >> sb->s_blocksize_bits;
634 int offset = off & (sb->s_blocksize - 1);
637 struct buffer_head tmp_bh;
638 struct buffer_head *bh;
639 loff_t i_size = i_size_read(inode);
643 if (off+len > i_size)
647 tocopy = sb->s_blocksize - offset < toread ?
648 sb->s_blocksize - offset : toread;
651 tmp_bh.b_size = 1 << inode->i_blkbits;
652 err = jfs_get_block(inode, blk, &tmp_bh, 0);
655 if (!buffer_mapped(&tmp_bh)) /* A hole? */
656 memset(data, 0, tocopy);
658 bh = sb_bread(sb, tmp_bh.b_blocknr);
661 memcpy(data, bh->b_data+offset, tocopy);
672 /* Write to quotafile */
673 static ssize_t jfs_quota_write(struct super_block *sb, int type,
674 const char *data, size_t len, loff_t off)
676 struct inode *inode = sb_dqopt(sb)->files[type];
677 sector_t blk = off >> sb->s_blocksize_bits;
679 int offset = off & (sb->s_blocksize - 1);
681 size_t towrite = len;
682 struct buffer_head tmp_bh;
683 struct buffer_head *bh;
685 mutex_lock(&inode->i_mutex);
686 while (towrite > 0) {
687 tocopy = sb->s_blocksize - offset < towrite ?
688 sb->s_blocksize - offset : towrite;
691 tmp_bh.b_size = 1 << inode->i_blkbits;
692 err = jfs_get_block(inode, blk, &tmp_bh, 1);
695 if (offset || tocopy != sb->s_blocksize)
696 bh = sb_bread(sb, tmp_bh.b_blocknr);
698 bh = sb_getblk(sb, tmp_bh.b_blocknr);
704 memcpy(bh->b_data+offset, data, tocopy);
705 flush_dcache_page(bh->b_page);
706 set_buffer_uptodate(bh);
707 mark_buffer_dirty(bh);
718 if (inode->i_size < off+len-towrite)
719 i_size_write(inode, off+len-towrite);
721 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
722 mark_inode_dirty(inode);
723 mutex_unlock(&inode->i_mutex);
724 return len - towrite;
729 static const struct super_operations jfs_super_operations = {
730 .alloc_inode = jfs_alloc_inode,
731 .destroy_inode = jfs_destroy_inode,
732 .dirty_inode = jfs_dirty_inode,
733 .write_inode = jfs_write_inode,
734 .delete_inode = jfs_delete_inode,
735 .put_super = jfs_put_super,
736 .sync_fs = jfs_sync_fs,
737 .write_super_lockfs = jfs_write_super_lockfs,
738 .unlockfs = jfs_unlockfs,
739 .statfs = jfs_statfs,
740 .remount_fs = jfs_remount,
741 .show_options = jfs_show_options,
743 .quota_read = jfs_quota_read,
744 .quota_write = jfs_quota_write,
748 static const struct export_operations jfs_export_operations = {
749 .fh_to_dentry = jfs_fh_to_dentry,
750 .fh_to_parent = jfs_fh_to_parent,
751 .get_parent = jfs_get_parent,
754 static struct file_system_type jfs_fs_type = {
755 .owner = THIS_MODULE,
757 .get_sb = jfs_get_sb,
758 .kill_sb = kill_block_super,
759 .fs_flags = FS_REQUIRES_DEV,
762 static void init_once(struct kmem_cache *cachep, void *foo)
764 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
766 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
767 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
768 init_rwsem(&jfs_ip->rdwrlock);
769 mutex_init(&jfs_ip->commit_mutex);
770 init_rwsem(&jfs_ip->xattr_sem);
771 spin_lock_init(&jfs_ip->ag_lock);
772 jfs_ip->active_ag = -1;
773 #ifdef CONFIG_JFS_POSIX_ACL
774 jfs_ip->i_acl = JFS_ACL_NOT_CACHED;
775 jfs_ip->i_default_acl = JFS_ACL_NOT_CACHED;
777 inode_init_once(&jfs_ip->vfs_inode);
780 static int __init init_jfs_fs(void)
786 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
787 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
789 if (jfs_inode_cachep == NULL)
793 * Metapage initialization
795 rc = metapage_init();
797 jfs_err("metapage_init failed w/rc = %d", rc);
802 * Transaction Manager initialization
806 jfs_err("txInit failed w/rc = %d", rc);
811 * I/O completion thread (endio)
813 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
814 if (IS_ERR(jfsIOthread)) {
815 rc = PTR_ERR(jfsIOthread);
816 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
820 if (commit_threads < 1)
821 commit_threads = num_online_cpus();
822 if (commit_threads > MAX_COMMIT_THREADS)
823 commit_threads = MAX_COMMIT_THREADS;
825 for (i = 0; i < commit_threads; i++) {
826 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
827 if (IS_ERR(jfsCommitThread[i])) {
828 rc = PTR_ERR(jfsCommitThread[i]);
829 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
831 goto kill_committask;
835 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
836 if (IS_ERR(jfsSyncThread)) {
837 rc = PTR_ERR(jfsSyncThread);
838 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
839 goto kill_committask;
846 return register_filesystem(&jfs_fs_type);
849 for (i = 0; i < commit_threads; i++)
850 kthread_stop(jfsCommitThread[i]);
851 kthread_stop(jfsIOthread);
857 kmem_cache_destroy(jfs_inode_cachep);
861 static void __exit exit_jfs_fs(void)
865 jfs_info("exit_jfs_fs called");
870 kthread_stop(jfsIOthread);
871 for (i = 0; i < commit_threads; i++)
872 kthread_stop(jfsCommitThread[i]);
873 kthread_stop(jfsSyncThread);
877 unregister_filesystem(&jfs_fs_type);
878 kmem_cache_destroy(jfs_inode_cachep);
881 module_init(init_jfs_fs)
882 module_exit(exit_jfs_fs)