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>
37 #include "jfs_incore.h"
38 #include "jfs_filsys.h"
39 #include "jfs_inode.h"
40 #include "jfs_metapage.h"
41 #include "jfs_superblock.h"
45 #include "jfs_debug.h"
47 MODULE_DESCRIPTION("The Journaled Filesystem (JFS)");
48 MODULE_AUTHOR("Steve Best/Dave Kleikamp/Barry Arndt, IBM");
49 MODULE_LICENSE("GPL");
51 static struct kmem_cache * jfs_inode_cachep;
53 static const struct super_operations jfs_super_operations;
54 static const struct export_operations jfs_export_operations;
55 static struct file_system_type jfs_fs_type;
57 #define MAX_COMMIT_THREADS 64
58 static int commit_threads = 0;
59 module_param(commit_threads, int, 0);
60 MODULE_PARM_DESC(commit_threads, "Number of commit threads");
62 static struct task_struct *jfsCommitThread[MAX_COMMIT_THREADS];
63 struct task_struct *jfsIOthread;
64 struct task_struct *jfsSyncThread;
66 #ifdef CONFIG_JFS_DEBUG
67 int jfsloglevel = JFS_LOGLEVEL_WARN;
68 module_param(jfsloglevel, int, 0644);
69 MODULE_PARM_DESC(jfsloglevel, "Specify JFS loglevel (0, 1 or 2)");
72 static void jfs_handle_error(struct super_block *sb)
74 struct jfs_sb_info *sbi = JFS_SBI(sb);
76 if (sb->s_flags & MS_RDONLY)
79 updateSuper(sb, FM_DIRTY);
81 if (sbi->flag & JFS_ERR_PANIC)
82 panic("JFS (device %s): panic forced after error\n",
84 else if (sbi->flag & JFS_ERR_REMOUNT_RO) {
85 jfs_err("ERROR: (device %s): remounting filesystem "
88 sb->s_flags |= MS_RDONLY;
91 /* nothing is done for continue beyond marking the superblock dirty */
94 void jfs_error(struct super_block *sb, const char * function, ...)
96 static char error_buf[256];
99 va_start(args, function);
100 vsnprintf(error_buf, sizeof(error_buf), function, args);
103 printk(KERN_ERR "ERROR: (device %s): %s\n", sb->s_id, error_buf);
105 jfs_handle_error(sb);
108 static struct inode *jfs_alloc_inode(struct super_block *sb)
110 struct jfs_inode_info *jfs_inode;
112 jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
115 return &jfs_inode->vfs_inode;
118 static void jfs_i_callback(struct rcu_head *head)
120 struct inode *inode = container_of(head, struct inode, i_rcu);
121 struct jfs_inode_info *ji = JFS_IP(inode);
122 INIT_LIST_HEAD(&inode->i_dentry);
123 kmem_cache_free(jfs_inode_cachep, ji);
126 static void jfs_destroy_inode(struct inode *inode)
128 struct jfs_inode_info *ji = JFS_IP(inode);
130 BUG_ON(!list_empty(&ji->anon_inode_list));
132 spin_lock_irq(&ji->ag_lock);
133 if (ji->active_ag != -1) {
134 struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
135 atomic_dec(&bmap->db_active[ji->active_ag]);
138 spin_unlock_irq(&ji->ag_lock);
139 call_rcu(&inode->i_rcu, jfs_i_callback);
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));
169 buf->f_fsid.val[0] = (u32)crc32_le(0, sbi->uuid, sizeof(sbi->uuid)/2);
170 buf->f_fsid.val[1] = (u32)crc32_le(0, sbi->uuid + sizeof(sbi->uuid)/2,
171 sizeof(sbi->uuid)/2);
173 buf->f_namelen = JFS_NAME_MAX;
177 static void jfs_put_super(struct super_block *sb)
179 struct jfs_sb_info *sbi = JFS_SBI(sb);
182 jfs_info("In jfs_put_super");
184 dquot_disable(sb, -1, DQUOT_USAGE_ENABLED | DQUOT_LIMITS_ENABLED);
188 jfs_err("jfs_umount failed with return code %d", rc);
190 unload_nls(sbi->nls_tab);
192 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
193 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");
382 rc = jfs_extendfs(sb, newLVSize, 0);
387 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
389 * Invalidate any previously read metadata. fsck may have
390 * changed the on-disk data since we mounted r/o
392 truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
394 JFS_SBI(sb)->flag = flag;
395 ret = jfs_mount_rw(sb, 1);
397 /* mark the fs r/w for quota activity */
398 sb->s_flags &= ~MS_RDONLY;
400 dquot_resume(sb, -1);
403 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
404 rc = dquot_suspend(sb, -1);
408 rc = jfs_umount_rw(sb);
409 JFS_SBI(sb)->flag = flag;
412 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
413 if (!(sb->s_flags & MS_RDONLY)) {
414 rc = jfs_umount_rw(sb);
418 JFS_SBI(sb)->flag = flag;
419 ret = jfs_mount_rw(sb, 1);
422 JFS_SBI(sb)->flag = flag;
427 static int jfs_fill_super(struct super_block *sb, void *data, int silent)
429 struct jfs_sb_info *sbi;
433 int flag, ret = -EINVAL;
435 jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
437 if (!new_valid_dev(sb->s_bdev->bd_dev))
440 sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
446 sbi->uid = sbi->gid = sbi->umask = -1;
448 /* initialize the mount flag and determine the default error handler */
449 flag = JFS_ERR_REMOUNT_RO;
451 if (!parse_options((char *) data, sb, &newLVSize, &flag))
455 #ifdef CONFIG_JFS_POSIX_ACL
456 sb->s_flags |= MS_POSIXACL;
460 printk(KERN_ERR "resize option for remount only\n");
465 * Initialize blocksize to 4K.
467 sb_set_blocksize(sb, PSIZE);
470 * Set method vectors.
472 sb->s_op = &jfs_super_operations;
473 sb->s_export_op = &jfs_export_operations;
475 sb->dq_op = &dquot_operations;
476 sb->s_qcop = &dquot_quotactl_ops;
480 * Initialize direct-mapping inode/address-space
482 inode = new_inode(sb);
488 inode->i_size = sb->s_bdev->bd_inode->i_size;
489 inode->i_mapping->a_ops = &jfs_metapage_aops;
490 insert_inode_hash(inode);
491 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
493 sbi->direct_inode = inode;
498 jfs_err("jfs_mount failed w/return code = %d", rc);
500 goto out_mount_failed;
502 if (sb->s_flags & MS_RDONLY)
505 rc = jfs_mount_rw(sb, 0);
508 jfs_err("jfs_mount_rw failed, return code = %d",
515 sb->s_magic = JFS_SUPER_MAGIC;
517 if (sbi->mntflag & JFS_OS2)
518 sb->s_d_op = &jfs_ci_dentry_operations;
520 inode = jfs_iget(sb, ROOT_I);
522 ret = PTR_ERR(inode);
525 sb->s_root = d_alloc_root(inode);
529 /* logical blocks are represented by 40 bits in pxd_t, etc. */
530 sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
531 #if BITS_PER_LONG == 32
533 * Page cache is indexed by long.
534 * I would use MAX_LFS_FILESIZE, but it's only half as big
536 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, (u64)sb->s_maxbytes);
542 jfs_err("jfs_read_super: get root dentry failed");
548 jfs_err("jfs_umount failed with return code %d", rc);
551 filemap_write_and_wait(sbi->direct_inode->i_mapping);
552 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
553 make_bad_inode(sbi->direct_inode);
554 iput(sbi->direct_inode);
555 sbi->direct_inode = NULL;
558 unload_nls(sbi->nls_tab);
564 static int jfs_freeze(struct super_block *sb)
566 struct jfs_sb_info *sbi = JFS_SBI(sb);
567 struct jfs_log *log = sbi->log;
569 if (!(sb->s_flags & MS_RDONLY)) {
572 updateSuper(sb, FM_CLEAN);
577 static int jfs_unfreeze(struct super_block *sb)
579 struct jfs_sb_info *sbi = JFS_SBI(sb);
580 struct jfs_log *log = sbi->log;
583 if (!(sb->s_flags & MS_RDONLY)) {
584 updateSuper(sb, FM_MOUNT);
585 if ((rc = lmLogInit(log)))
586 jfs_err("jfs_unlock failed with return code %d", rc);
593 static struct dentry *jfs_do_mount(struct file_system_type *fs_type,
594 int flags, const char *dev_name, void *data)
596 return mount_bdev(fs_type, flags, dev_name, data, jfs_fill_super);
599 static int jfs_sync_fs(struct super_block *sb, int wait)
601 struct jfs_log *log = JFS_SBI(sb)->log;
603 /* log == NULL indicates read-only mount */
605 jfs_flush_journal(log, wait);
612 static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
614 struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
617 seq_printf(seq, ",uid=%d", sbi->uid);
619 seq_printf(seq, ",gid=%d", sbi->gid);
620 if (sbi->umask != -1)
621 seq_printf(seq, ",umask=%03o", sbi->umask);
622 if (sbi->flag & JFS_NOINTEGRITY)
623 seq_puts(seq, ",nointegrity");
625 seq_printf(seq, ",iocharset=%s", sbi->nls_tab->charset);
626 if (sbi->flag & JFS_ERR_CONTINUE)
627 seq_printf(seq, ",errors=continue");
628 if (sbi->flag & JFS_ERR_PANIC)
629 seq_printf(seq, ",errors=panic");
632 if (sbi->flag & JFS_USRQUOTA)
633 seq_puts(seq, ",usrquota");
635 if (sbi->flag & JFS_GRPQUOTA)
636 seq_puts(seq, ",grpquota");
644 /* Read data from quotafile - avoid pagecache and such because we cannot afford
645 * acquiring the locks... As quota files are never truncated and quota code
646 * itself serializes the operations (and no one else should touch the files)
647 * we don't have to be afraid of races */
648 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
649 size_t len, loff_t off)
651 struct inode *inode = sb_dqopt(sb)->files[type];
652 sector_t blk = off >> sb->s_blocksize_bits;
654 int offset = off & (sb->s_blocksize - 1);
657 struct buffer_head tmp_bh;
658 struct buffer_head *bh;
659 loff_t i_size = i_size_read(inode);
663 if (off+len > i_size)
667 tocopy = sb->s_blocksize - offset < toread ?
668 sb->s_blocksize - offset : toread;
671 tmp_bh.b_size = 1 << inode->i_blkbits;
672 err = jfs_get_block(inode, blk, &tmp_bh, 0);
675 if (!buffer_mapped(&tmp_bh)) /* A hole? */
676 memset(data, 0, tocopy);
678 bh = sb_bread(sb, tmp_bh.b_blocknr);
681 memcpy(data, bh->b_data+offset, tocopy);
692 /* Write to quotafile */
693 static ssize_t jfs_quota_write(struct super_block *sb, int type,
694 const char *data, size_t len, loff_t off)
696 struct inode *inode = sb_dqopt(sb)->files[type];
697 sector_t blk = off >> sb->s_blocksize_bits;
699 int offset = off & (sb->s_blocksize - 1);
701 size_t towrite = len;
702 struct buffer_head tmp_bh;
703 struct buffer_head *bh;
705 mutex_lock(&inode->i_mutex);
706 while (towrite > 0) {
707 tocopy = sb->s_blocksize - offset < towrite ?
708 sb->s_blocksize - offset : towrite;
711 tmp_bh.b_size = 1 << inode->i_blkbits;
712 err = jfs_get_block(inode, blk, &tmp_bh, 1);
715 if (offset || tocopy != sb->s_blocksize)
716 bh = sb_bread(sb, tmp_bh.b_blocknr);
718 bh = sb_getblk(sb, tmp_bh.b_blocknr);
724 memcpy(bh->b_data+offset, data, tocopy);
725 flush_dcache_page(bh->b_page);
726 set_buffer_uptodate(bh);
727 mark_buffer_dirty(bh);
736 if (len == towrite) {
737 mutex_unlock(&inode->i_mutex);
740 if (inode->i_size < off+len-towrite)
741 i_size_write(inode, off+len-towrite);
743 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
744 mark_inode_dirty(inode);
745 mutex_unlock(&inode->i_mutex);
746 return len - towrite;
751 static const struct super_operations jfs_super_operations = {
752 .alloc_inode = jfs_alloc_inode,
753 .destroy_inode = jfs_destroy_inode,
754 .dirty_inode = jfs_dirty_inode,
755 .write_inode = jfs_write_inode,
756 .evict_inode = jfs_evict_inode,
757 .put_super = jfs_put_super,
758 .sync_fs = jfs_sync_fs,
759 .freeze_fs = jfs_freeze,
760 .unfreeze_fs = jfs_unfreeze,
761 .statfs = jfs_statfs,
762 .remount_fs = jfs_remount,
763 .show_options = jfs_show_options,
765 .quota_read = jfs_quota_read,
766 .quota_write = jfs_quota_write,
770 static const struct export_operations jfs_export_operations = {
771 .fh_to_dentry = jfs_fh_to_dentry,
772 .fh_to_parent = jfs_fh_to_parent,
773 .get_parent = jfs_get_parent,
776 static struct file_system_type jfs_fs_type = {
777 .owner = THIS_MODULE,
779 .mount = jfs_do_mount,
780 .kill_sb = kill_block_super,
781 .fs_flags = FS_REQUIRES_DEV,
784 static void init_once(void *foo)
786 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
788 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
789 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
790 init_rwsem(&jfs_ip->rdwrlock);
791 mutex_init(&jfs_ip->commit_mutex);
792 init_rwsem(&jfs_ip->xattr_sem);
793 spin_lock_init(&jfs_ip->ag_lock);
794 jfs_ip->active_ag = -1;
795 inode_init_once(&jfs_ip->vfs_inode);
798 static int __init init_jfs_fs(void)
804 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
805 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
807 if (jfs_inode_cachep == NULL)
811 * Metapage initialization
813 rc = metapage_init();
815 jfs_err("metapage_init failed w/rc = %d", rc);
820 * Transaction Manager initialization
824 jfs_err("txInit failed w/rc = %d", rc);
829 * I/O completion thread (endio)
831 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
832 if (IS_ERR(jfsIOthread)) {
833 rc = PTR_ERR(jfsIOthread);
834 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
838 if (commit_threads < 1)
839 commit_threads = num_online_cpus();
840 if (commit_threads > MAX_COMMIT_THREADS)
841 commit_threads = MAX_COMMIT_THREADS;
843 for (i = 0; i < commit_threads; i++) {
844 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
845 if (IS_ERR(jfsCommitThread[i])) {
846 rc = PTR_ERR(jfsCommitThread[i]);
847 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
849 goto kill_committask;
853 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
854 if (IS_ERR(jfsSyncThread)) {
855 rc = PTR_ERR(jfsSyncThread);
856 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
857 goto kill_committask;
864 return register_filesystem(&jfs_fs_type);
867 for (i = 0; i < commit_threads; i++)
868 kthread_stop(jfsCommitThread[i]);
869 kthread_stop(jfsIOthread);
875 kmem_cache_destroy(jfs_inode_cachep);
879 static void __exit exit_jfs_fs(void)
883 jfs_info("exit_jfs_fs called");
888 kthread_stop(jfsIOthread);
889 for (i = 0; i < commit_threads; i++)
890 kthread_stop(jfsCommitThread[i]);
891 kthread_stop(jfsSyncThread);
895 unregister_filesystem(&jfs_fs_type);
896 kmem_cache_destroy(jfs_inode_cachep);
899 module_init(init_jfs_fs)
900 module_exit(exit_jfs_fs)