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>
35 #include <linux/smp_lock.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_destroy_inode(struct inode *inode)
120 struct jfs_inode_info *ji = JFS_IP(inode);
122 BUG_ON(!list_empty(&ji->anon_inode_list));
124 spin_lock_irq(&ji->ag_lock);
125 if (ji->active_ag != -1) {
126 struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
127 atomic_dec(&bmap->db_active[ji->active_ag]);
130 spin_unlock_irq(&ji->ag_lock);
131 kmem_cache_free(jfs_inode_cachep, ji);
134 static int jfs_statfs(struct dentry *dentry, struct kstatfs *buf)
136 struct jfs_sb_info *sbi = JFS_SBI(dentry->d_sb);
138 struct inomap *imap = JFS_IP(sbi->ipimap)->i_imap;
140 jfs_info("In jfs_statfs");
141 buf->f_type = JFS_SUPER_MAGIC;
142 buf->f_bsize = sbi->bsize;
143 buf->f_blocks = sbi->bmap->db_mapsize;
144 buf->f_bfree = sbi->bmap->db_nfree;
145 buf->f_bavail = sbi->bmap->db_nfree;
147 * If we really return the number of allocated & free inodes, some
148 * applications will fail because they won't see enough free inodes.
149 * We'll try to calculate some guess as to how may inodes we can
152 * buf->f_files = atomic_read(&imap->im_numinos);
153 * buf->f_ffree = atomic_read(&imap->im_numfree);
155 maxinodes = min((s64) atomic_read(&imap->im_numinos) +
156 ((sbi->bmap->db_nfree >> imap->im_l2nbperiext)
157 << L2INOSPEREXT), (s64) 0xffffffffLL);
158 buf->f_files = maxinodes;
159 buf->f_ffree = maxinodes - (atomic_read(&imap->im_numinos) -
160 atomic_read(&imap->im_numfree));
161 buf->f_fsid.val[0] = (u32)crc32_le(0, sbi->uuid, sizeof(sbi->uuid)/2);
162 buf->f_fsid.val[1] = (u32)crc32_le(0, sbi->uuid + sizeof(sbi->uuid)/2,
163 sizeof(sbi->uuid)/2);
165 buf->f_namelen = JFS_NAME_MAX;
169 static void jfs_put_super(struct super_block *sb)
171 struct jfs_sb_info *sbi = JFS_SBI(sb);
174 jfs_info("In jfs_put_super");
180 jfs_err("jfs_umount failed with return code %d", rc);
182 unload_nls(sbi->nls_tab);
185 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
186 iput(sbi->direct_inode);
187 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
200 static const match_table_t tokens = {
201 {Opt_integrity, "integrity"},
202 {Opt_nointegrity, "nointegrity"},
203 {Opt_iocharset, "iocharset=%s"},
204 {Opt_resize, "resize=%u"},
205 {Opt_resize_nosize, "resize"},
206 {Opt_errors, "errors=%s"},
207 {Opt_ignore, "noquota"},
208 {Opt_ignore, "quota"},
209 {Opt_usrquota, "usrquota"},
210 {Opt_grpquota, "grpquota"},
213 {Opt_umask, "umask=%u"},
217 static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
220 void *nls_map = (void *)-1; /* -1: no change; NULL: none */
222 struct jfs_sb_info *sbi = JFS_SBI(sb);
229 while ((p = strsep(&options, ",")) != NULL) {
230 substring_t args[MAX_OPT_ARGS];
235 token = match_token(p, tokens, args);
238 *flag &= ~JFS_NOINTEGRITY;
240 case Opt_nointegrity:
241 *flag |= JFS_NOINTEGRITY;
244 /* Silently ignore the quota options */
245 /* Don't do anything ;-) */
248 if (nls_map && nls_map != (void *) -1)
250 if (!strcmp(args[0].from, "none"))
253 nls_map = load_nls(args[0].from);
256 "JFS: charset not found\n");
263 char *resize = args[0].from;
264 *newLVSize = simple_strtoull(resize, &resize, 0);
267 case Opt_resize_nosize:
269 *newLVSize = sb->s_bdev->bd_inode->i_size >>
270 sb->s_blocksize_bits;
273 "JFS: Cannot determine volume size\n");
278 char *errors = args[0].from;
279 if (!errors || !*errors)
281 if (!strcmp(errors, "continue")) {
282 *flag &= ~JFS_ERR_REMOUNT_RO;
283 *flag &= ~JFS_ERR_PANIC;
284 *flag |= JFS_ERR_CONTINUE;
285 } else if (!strcmp(errors, "remount-ro")) {
286 *flag &= ~JFS_ERR_CONTINUE;
287 *flag &= ~JFS_ERR_PANIC;
288 *flag |= JFS_ERR_REMOUNT_RO;
289 } else if (!strcmp(errors, "panic")) {
290 *flag &= ~JFS_ERR_CONTINUE;
291 *flag &= ~JFS_ERR_REMOUNT_RO;
292 *flag |= JFS_ERR_PANIC;
295 "JFS: %s is an invalid error handler\n",
305 *flag |= JFS_USRQUOTA;
308 *flag |= JFS_GRPQUOTA;
315 "JFS: quota operations not supported\n");
320 char *uid = args[0].from;
321 sbi->uid = simple_strtoul(uid, &uid, 0);
326 char *gid = args[0].from;
327 sbi->gid = simple_strtoul(gid, &gid, 0);
332 char *umask = args[0].from;
333 sbi->umask = simple_strtoul(umask, &umask, 8);
334 if (sbi->umask & ~0777) {
336 "JFS: Invalid value of umask\n");
342 printk("jfs: Unrecognized mount option \"%s\" "
343 " or missing value\n", p);
348 if (nls_map != (void *) -1) {
349 /* Discard old (if remount) */
351 unload_nls(sbi->nls_tab);
352 sbi->nls_tab = nls_map;
357 if (nls_map && nls_map != (void *) -1)
362 static int jfs_remount(struct super_block *sb, int *flags, char *data)
366 int flag = JFS_SBI(sb)->flag;
369 if (!parse_options(data, sb, &newLVSize, &flag)) {
374 if (sb->s_flags & MS_RDONLY) {
376 "JFS: resize requires volume to be mounted read-write\n");
380 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);
399 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
400 rc = jfs_umount_rw(sb);
401 JFS_SBI(sb)->flag = flag;
405 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
406 if (!(sb->s_flags & MS_RDONLY)) {
407 rc = jfs_umount_rw(sb);
412 JFS_SBI(sb)->flag = flag;
413 ret = jfs_mount_rw(sb, 1);
417 JFS_SBI(sb)->flag = flag;
423 static int jfs_fill_super(struct super_block *sb, void *data, int silent)
425 struct jfs_sb_info *sbi;
429 int flag, ret = -EINVAL;
431 jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
433 if (!new_valid_dev(sb->s_bdev->bd_dev))
436 sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
441 sbi->uid = sbi->gid = sbi->umask = -1;
443 /* initialize the mount flag and determine the default error handler */
444 flag = JFS_ERR_REMOUNT_RO;
446 if (!parse_options((char *) data, sb, &newLVSize, &flag)) {
452 #ifdef CONFIG_JFS_POSIX_ACL
453 sb->s_flags |= MS_POSIXACL;
457 printk(KERN_ERR "resize option for remount only\n");
462 * Initialize blocksize to 4K.
464 sb_set_blocksize(sb, PSIZE);
467 * Set method vectors.
469 sb->s_op = &jfs_super_operations;
470 sb->s_export_op = &jfs_export_operations;
473 * Initialize direct-mapping inode/address-space
475 inode = new_inode(sb);
482 inode->i_size = sb->s_bdev->bd_inode->i_size;
483 inode->i_mapping->a_ops = &jfs_metapage_aops;
484 insert_inode_hash(inode);
485 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
487 sbi->direct_inode = inode;
492 jfs_err("jfs_mount failed w/return code = %d", rc);
494 goto out_mount_failed;
496 if (sb->s_flags & MS_RDONLY)
499 rc = jfs_mount_rw(sb, 0);
502 jfs_err("jfs_mount_rw failed, return code = %d",
509 sb->s_magic = JFS_SUPER_MAGIC;
511 inode = jfs_iget(sb, ROOT_I);
513 ret = PTR_ERR(inode);
516 sb->s_root = d_alloc_root(inode);
520 if (sbi->mntflag & JFS_OS2)
521 sb->s_root->d_op = &jfs_ci_dentry_operations;
523 /* logical blocks are represented by 40 bits in pxd_t, etc. */
524 sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
525 #if BITS_PER_LONG == 32
527 * Page cache is indexed by long.
528 * I would use MAX_LFS_FILESIZE, but it's only half as big
530 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, sb->s_maxbytes);
536 jfs_err("jfs_read_super: get root dentry failed");
542 jfs_err("jfs_umount failed with return code %d", rc);
545 filemap_write_and_wait(sbi->direct_inode->i_mapping);
546 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
547 make_bad_inode(sbi->direct_inode);
548 iput(sbi->direct_inode);
549 sbi->direct_inode = NULL;
552 unload_nls(sbi->nls_tab);
557 static int jfs_freeze(struct super_block *sb)
559 struct jfs_sb_info *sbi = JFS_SBI(sb);
560 struct jfs_log *log = sbi->log;
562 if (!(sb->s_flags & MS_RDONLY)) {
565 updateSuper(sb, FM_CLEAN);
570 static int jfs_unfreeze(struct super_block *sb)
572 struct jfs_sb_info *sbi = JFS_SBI(sb);
573 struct jfs_log *log = sbi->log;
576 if (!(sb->s_flags & MS_RDONLY)) {
577 updateSuper(sb, FM_MOUNT);
578 if ((rc = lmLogInit(log)))
579 jfs_err("jfs_unlock failed with return code %d", rc);
586 static int jfs_get_sb(struct file_system_type *fs_type,
587 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
589 return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
593 static int jfs_sync_fs(struct super_block *sb, int wait)
595 struct jfs_log *log = JFS_SBI(sb)->log;
597 /* log == NULL indicates read-only mount */
599 jfs_flush_journal(log, wait);
606 static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
608 struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
611 seq_printf(seq, ",uid=%d", sbi->uid);
613 seq_printf(seq, ",gid=%d", sbi->gid);
614 if (sbi->umask != -1)
615 seq_printf(seq, ",umask=%03o", sbi->umask);
616 if (sbi->flag & JFS_NOINTEGRITY)
617 seq_puts(seq, ",nointegrity");
619 seq_printf(seq, ",iocharset=%s", sbi->nls_tab->charset);
620 if (sbi->flag & JFS_ERR_CONTINUE)
621 seq_printf(seq, ",errors=continue");
622 if (sbi->flag & JFS_ERR_PANIC)
623 seq_printf(seq, ",errors=panic");
626 if (sbi->flag & JFS_USRQUOTA)
627 seq_puts(seq, ",usrquota");
629 if (sbi->flag & JFS_GRPQUOTA)
630 seq_puts(seq, ",grpquota");
638 /* Read data from quotafile - avoid pagecache and such because we cannot afford
639 * acquiring the locks... As quota files are never truncated and quota code
640 * itself serializes the operations (and noone else should touch the files)
641 * we don't have to be afraid of races */
642 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
643 size_t len, loff_t off)
645 struct inode *inode = sb_dqopt(sb)->files[type];
646 sector_t blk = off >> sb->s_blocksize_bits;
648 int offset = off & (sb->s_blocksize - 1);
651 struct buffer_head tmp_bh;
652 struct buffer_head *bh;
653 loff_t i_size = i_size_read(inode);
657 if (off+len > i_size)
661 tocopy = sb->s_blocksize - offset < toread ?
662 sb->s_blocksize - offset : toread;
665 tmp_bh.b_size = 1 << inode->i_blkbits;
666 err = jfs_get_block(inode, blk, &tmp_bh, 0);
669 if (!buffer_mapped(&tmp_bh)) /* A hole? */
670 memset(data, 0, tocopy);
672 bh = sb_bread(sb, tmp_bh.b_blocknr);
675 memcpy(data, bh->b_data+offset, tocopy);
686 /* Write to quotafile */
687 static ssize_t jfs_quota_write(struct super_block *sb, int type,
688 const char *data, size_t len, loff_t off)
690 struct inode *inode = sb_dqopt(sb)->files[type];
691 sector_t blk = off >> sb->s_blocksize_bits;
693 int offset = off & (sb->s_blocksize - 1);
695 size_t towrite = len;
696 struct buffer_head tmp_bh;
697 struct buffer_head *bh;
699 mutex_lock(&inode->i_mutex);
700 while (towrite > 0) {
701 tocopy = sb->s_blocksize - offset < towrite ?
702 sb->s_blocksize - offset : towrite;
705 tmp_bh.b_size = 1 << inode->i_blkbits;
706 err = jfs_get_block(inode, blk, &tmp_bh, 1);
709 if (offset || tocopy != sb->s_blocksize)
710 bh = sb_bread(sb, tmp_bh.b_blocknr);
712 bh = sb_getblk(sb, tmp_bh.b_blocknr);
718 memcpy(bh->b_data+offset, data, tocopy);
719 flush_dcache_page(bh->b_page);
720 set_buffer_uptodate(bh);
721 mark_buffer_dirty(bh);
730 if (len == towrite) {
731 mutex_unlock(&inode->i_mutex);
734 if (inode->i_size < off+len-towrite)
735 i_size_write(inode, off+len-towrite);
737 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
738 mark_inode_dirty(inode);
739 mutex_unlock(&inode->i_mutex);
740 return len - towrite;
745 static const struct super_operations jfs_super_operations = {
746 .alloc_inode = jfs_alloc_inode,
747 .destroy_inode = jfs_destroy_inode,
748 .dirty_inode = jfs_dirty_inode,
749 .write_inode = jfs_write_inode,
750 .delete_inode = jfs_delete_inode,
751 .put_super = jfs_put_super,
752 .sync_fs = jfs_sync_fs,
753 .freeze_fs = jfs_freeze,
754 .unfreeze_fs = jfs_unfreeze,
755 .statfs = jfs_statfs,
756 .remount_fs = jfs_remount,
757 .show_options = jfs_show_options,
759 .quota_read = jfs_quota_read,
760 .quota_write = jfs_quota_write,
764 static const struct export_operations jfs_export_operations = {
765 .fh_to_dentry = jfs_fh_to_dentry,
766 .fh_to_parent = jfs_fh_to_parent,
767 .get_parent = jfs_get_parent,
770 static struct file_system_type jfs_fs_type = {
771 .owner = THIS_MODULE,
773 .get_sb = jfs_get_sb,
774 .kill_sb = kill_block_super,
775 .fs_flags = FS_REQUIRES_DEV,
778 static void init_once(void *foo)
780 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
782 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
783 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
784 init_rwsem(&jfs_ip->rdwrlock);
785 mutex_init(&jfs_ip->commit_mutex);
786 init_rwsem(&jfs_ip->xattr_sem);
787 spin_lock_init(&jfs_ip->ag_lock);
788 jfs_ip->active_ag = -1;
789 inode_init_once(&jfs_ip->vfs_inode);
792 static int __init init_jfs_fs(void)
798 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
799 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
801 if (jfs_inode_cachep == NULL)
805 * Metapage initialization
807 rc = metapage_init();
809 jfs_err("metapage_init failed w/rc = %d", rc);
814 * Transaction Manager initialization
818 jfs_err("txInit failed w/rc = %d", rc);
823 * I/O completion thread (endio)
825 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
826 if (IS_ERR(jfsIOthread)) {
827 rc = PTR_ERR(jfsIOthread);
828 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
832 if (commit_threads < 1)
833 commit_threads = num_online_cpus();
834 if (commit_threads > MAX_COMMIT_THREADS)
835 commit_threads = MAX_COMMIT_THREADS;
837 for (i = 0; i < commit_threads; i++) {
838 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
839 if (IS_ERR(jfsCommitThread[i])) {
840 rc = PTR_ERR(jfsCommitThread[i]);
841 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
843 goto kill_committask;
847 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
848 if (IS_ERR(jfsSyncThread)) {
849 rc = PTR_ERR(jfsSyncThread);
850 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
851 goto kill_committask;
858 return register_filesystem(&jfs_fs_type);
861 for (i = 0; i < commit_threads; i++)
862 kthread_stop(jfsCommitThread[i]);
863 kthread_stop(jfsIOthread);
869 kmem_cache_destroy(jfs_inode_cachep);
873 static void __exit exit_jfs_fs(void)
877 jfs_info("exit_jfs_fs called");
882 kthread_stop(jfsIOthread);
883 for (i = 0; i < commit_threads; i++)
884 kthread_stop(jfsCommitThread[i]);
885 kthread_stop(jfsSyncThread);
889 unregister_filesystem(&jfs_fs_type);
890 kmem_cache_destroy(jfs_inode_cachep);
893 module_init(init_jfs_fs)
894 module_exit(exit_jfs_fs)