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 kmem_cache_t * 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 vsprintf(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,
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_tagxid, "tagxid"},
209 {Opt_ignore, "noquota"},
210 {Opt_ignore, "quota"},
211 {Opt_usrquota, "usrquota"},
212 {Opt_grpquota, "grpquota"},
215 {Opt_umask, "umask=%u"},
219 static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
222 void *nls_map = (void *)-1; /* -1: no change; NULL: none */
224 struct jfs_sb_info *sbi = JFS_SBI(sb);
231 while ((p = strsep(&options, ",")) != NULL) {
232 substring_t args[MAX_OPT_ARGS];
237 token = match_token(p, tokens, args);
240 *flag &= ~JFS_NOINTEGRITY;
242 case Opt_nointegrity:
243 *flag |= JFS_NOINTEGRITY;
246 /* Silently ignore the quota options */
247 /* Don't do anything ;-) */
250 if (nls_map && nls_map != (void *) -1)
252 if (!strcmp(args[0].from, "none"))
255 nls_map = load_nls(args[0].from);
258 "JFS: charset not found\n");
265 char *resize = args[0].from;
266 *newLVSize = simple_strtoull(resize, &resize, 0);
269 case Opt_resize_nosize:
271 *newLVSize = sb->s_bdev->bd_inode->i_size >>
272 sb->s_blocksize_bits;
275 "JFS: Cannot determine volume size\n");
280 char *errors = args[0].from;
281 if (!errors || !*errors)
283 if (!strcmp(errors, "continue")) {
284 *flag &= ~JFS_ERR_REMOUNT_RO;
285 *flag &= ~JFS_ERR_PANIC;
286 *flag |= JFS_ERR_CONTINUE;
287 } else if (!strcmp(errors, "remount-ro")) {
288 *flag &= ~JFS_ERR_CONTINUE;
289 *flag &= ~JFS_ERR_PANIC;
290 *flag |= JFS_ERR_REMOUNT_RO;
291 } else if (!strcmp(errors, "panic")) {
292 *flag &= ~JFS_ERR_CONTINUE;
293 *flag &= ~JFS_ERR_REMOUNT_RO;
294 *flag |= JFS_ERR_PANIC;
297 "JFS: %s is an invalid error handler\n",
307 *flag |= JFS_USRQUOTA;
310 *flag |= JFS_GRPQUOTA;
317 "JFS: quota operations not supported\n");
322 char *uid = args[0].from;
323 sbi->uid = simple_strtoul(uid, &uid, 0);
328 char *gid = args[0].from;
329 sbi->gid = simple_strtoul(gid, &gid, 0);
334 char *umask = args[0].from;
335 sbi->umask = simple_strtoul(umask, &umask, 8);
336 if (sbi->umask & ~0777) {
338 "JFS: Invalid value of umask\n");
343 #ifndef CONFIG_TAGGING_NONE
349 printk("jfs: Unrecognized mount option \"%s\" "
350 " or missing value\n", p);
355 if (nls_map != (void *) -1) {
356 /* Discard old (if remount) */
358 unload_nls(sbi->nls_tab);
359 sbi->nls_tab = nls_map;
364 if (nls_map && nls_map != (void *) -1)
369 static int jfs_remount(struct super_block *sb, int *flags, char *data)
373 int flag = JFS_SBI(sb)->flag;
375 if (!parse_options(data, sb, &newLVSize, &flag)) {
379 if ((flag & JFS_TAGXID) && !(sb->s_flags & MS_TAGXID)) {
380 printk(KERN_ERR "JFS: %s: tagxid not permitted on remount.\n",
386 if (sb->s_flags & MS_RDONLY) {
388 "JFS: resize requires volume to be mounted read-write\n");
391 rc = jfs_extendfs(sb, newLVSize, 0);
396 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
398 * Invalidate any previously read metadata. fsck may have
399 * changed the on-disk data since we mounted r/o
401 truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
403 JFS_SBI(sb)->flag = flag;
404 return jfs_mount_rw(sb, 1);
406 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
407 rc = jfs_umount_rw(sb);
408 JFS_SBI(sb)->flag = flag;
411 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
412 if (!(sb->s_flags & MS_RDONLY)) {
413 rc = jfs_umount_rw(sb);
416 JFS_SBI(sb)->flag = flag;
417 return jfs_mount_rw(sb, 1);
419 JFS_SBI(sb)->flag = flag;
424 static int jfs_fill_super(struct super_block *sb, void *data, int silent)
426 struct jfs_sb_info *sbi;
432 jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
434 if (!new_valid_dev(sb->s_bdev->bd_dev))
437 sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
442 sbi->uid = sbi->gid = sbi->umask = -1;
444 /* initialize the mount flag and determine the default error handler */
445 flag = JFS_ERR_REMOUNT_RO;
447 if (!parse_options((char *) data, sb, &newLVSize, &flag)) {
453 #ifdef CONFIG_JFS_POSIX_ACL
454 sb->s_flags |= MS_POSIXACL;
456 /* map mount option tagxid */
457 if (sbi->flag & JFS_TAGXID)
458 sb->s_flags |= MS_TAGXID;
461 printk(KERN_ERR "resize option for remount only\n");
466 * Initialize blocksize to 4K.
468 sb_set_blocksize(sb, PSIZE);
471 * Set method vectors.
473 sb->s_op = &jfs_super_operations;
474 sb->s_export_op = &jfs_export_operations;
477 * Initialize direct-mapping inode/address-space
479 inode = new_inode(sb);
484 inode->i_size = sb->s_bdev->bd_inode->i_size;
485 inode->i_mapping->a_ops = &jfs_metapage_aops;
486 insert_inode_hash(inode);
487 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
489 sbi->direct_inode = inode;
494 jfs_err("jfs_mount failed w/return code = %d", rc);
496 goto out_mount_failed;
498 if (sb->s_flags & MS_RDONLY)
501 rc = jfs_mount_rw(sb, 0);
504 jfs_err("jfs_mount_rw failed, return code = %d",
511 sb->s_magic = JFS_SUPER_MAGIC;
513 inode = iget(sb, ROOT_I);
514 if (!inode || is_bad_inode(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 inode failed");
543 jfs_err("jfs_umount failed with return code %d", rc);
546 filemap_write_and_wait(sbi->direct_inode->i_mapping);
547 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
548 make_bad_inode(sbi->direct_inode);
549 iput(sbi->direct_inode);
550 sbi->direct_inode = NULL;
553 unload_nls(sbi->nls_tab);
558 static void jfs_write_super_lockfs(struct super_block *sb)
560 struct jfs_sb_info *sbi = JFS_SBI(sb);
561 struct jfs_log *log = sbi->log;
563 if (!(sb->s_flags & MS_RDONLY)) {
566 updateSuper(sb, FM_CLEAN);
570 static void jfs_unlockfs(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);
585 static int jfs_get_sb(struct file_system_type *fs_type,
586 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
588 return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
592 static int jfs_sync_fs(struct super_block *sb, int wait)
594 struct jfs_log *log = JFS_SBI(sb)->log;
596 /* log == NULL indicates read-only mount */
598 jfs_flush_journal(log, wait);
605 static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
607 struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
610 seq_printf(seq, ",uid=%d", sbi->uid);
612 seq_printf(seq, ",gid=%d", sbi->gid);
613 if (sbi->umask != -1)
614 seq_printf(seq, ",umask=%03o", sbi->umask);
615 if (sbi->flag & JFS_NOINTEGRITY)
616 seq_puts(seq, ",nointegrity");
619 if (sbi->flag & JFS_USRQUOTA)
620 seq_puts(seq, ",usrquota");
622 if (sbi->flag & JFS_GRPQUOTA)
623 seq_puts(seq, ",grpquota");
631 /* Read data from quotafile - avoid pagecache and such because we cannot afford
632 * acquiring the locks... As quota files are never truncated and quota code
633 * itself serializes the operations (and noone else should touch the files)
634 * we don't have to be afraid of races */
635 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
636 size_t len, loff_t off)
638 struct inode *inode = sb_dqopt(sb)->files[type];
639 sector_t blk = off >> sb->s_blocksize_bits;
641 int offset = off & (sb->s_blocksize - 1);
644 struct buffer_head tmp_bh;
645 struct buffer_head *bh;
646 loff_t i_size = i_size_read(inode);
650 if (off+len > i_size)
654 tocopy = sb->s_blocksize - offset < toread ?
655 sb->s_blocksize - offset : toread;
658 tmp_bh.b_size = 1 << inode->i_blkbits;
659 err = jfs_get_block(inode, blk, &tmp_bh, 0);
662 if (!buffer_mapped(&tmp_bh)) /* A hole? */
663 memset(data, 0, tocopy);
665 bh = sb_bread(sb, tmp_bh.b_blocknr);
668 memcpy(data, bh->b_data+offset, tocopy);
679 /* Write to quotafile */
680 static ssize_t jfs_quota_write(struct super_block *sb, int type,
681 const char *data, size_t len, loff_t off)
683 struct inode *inode = sb_dqopt(sb)->files[type];
684 sector_t blk = off >> sb->s_blocksize_bits;
686 int offset = off & (sb->s_blocksize - 1);
688 size_t towrite = len;
689 struct buffer_head tmp_bh;
690 struct buffer_head *bh;
692 mutex_lock(&inode->i_mutex);
693 while (towrite > 0) {
694 tocopy = sb->s_blocksize - offset < towrite ?
695 sb->s_blocksize - offset : towrite;
698 tmp_bh.b_size = 1 << inode->i_blkbits;
699 err = jfs_get_block(inode, blk, &tmp_bh, 1);
702 if (offset || tocopy != sb->s_blocksize)
703 bh = sb_bread(sb, tmp_bh.b_blocknr);
705 bh = sb_getblk(sb, tmp_bh.b_blocknr);
711 memcpy(bh->b_data+offset, data, tocopy);
712 flush_dcache_page(bh->b_page);
713 set_buffer_uptodate(bh);
714 mark_buffer_dirty(bh);
725 if (inode->i_size < off+len-towrite)
726 i_size_write(inode, off+len-towrite);
728 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
729 mark_inode_dirty(inode);
730 mutex_unlock(&inode->i_mutex);
731 return len - towrite;
736 static struct super_operations jfs_super_operations = {
737 .alloc_inode = jfs_alloc_inode,
738 .destroy_inode = jfs_destroy_inode,
739 .read_inode = jfs_read_inode,
740 .dirty_inode = jfs_dirty_inode,
741 .write_inode = jfs_write_inode,
742 .delete_inode = jfs_delete_inode,
743 .put_super = jfs_put_super,
744 .sync_fs = jfs_sync_fs,
745 .write_super_lockfs = jfs_write_super_lockfs,
746 .unlockfs = jfs_unlockfs,
747 .statfs = jfs_statfs,
748 .remount_fs = jfs_remount,
749 .show_options = jfs_show_options,
751 .quota_read = jfs_quota_read,
752 .quota_write = jfs_quota_write,
756 static struct export_operations jfs_export_operations = {
757 .get_parent = jfs_get_parent,
760 static struct file_system_type jfs_fs_type = {
761 .owner = THIS_MODULE,
763 .get_sb = jfs_get_sb,
764 .kill_sb = kill_block_super,
765 .fs_flags = FS_REQUIRES_DEV,
768 static void init_once(void *foo, kmem_cache_t * cachep, unsigned long flags)
770 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
772 if ((flags & (SLAB_CTOR_VERIFY | SLAB_CTOR_CONSTRUCTOR)) ==
773 SLAB_CTOR_CONSTRUCTOR) {
774 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
775 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
776 init_rwsem(&jfs_ip->rdwrlock);
777 mutex_init(&jfs_ip->commit_mutex);
778 init_rwsem(&jfs_ip->xattr_sem);
779 spin_lock_init(&jfs_ip->ag_lock);
780 jfs_ip->active_ag = -1;
781 #ifdef CONFIG_JFS_POSIX_ACL
782 jfs_ip->i_acl = JFS_ACL_NOT_CACHED;
783 jfs_ip->i_default_acl = JFS_ACL_NOT_CACHED;
785 inode_init_once(&jfs_ip->vfs_inode);
789 static int __init init_jfs_fs(void)
795 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
796 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
798 if (jfs_inode_cachep == NULL)
802 * Metapage initialization
804 rc = metapage_init();
806 jfs_err("metapage_init failed w/rc = %d", rc);
811 * Transaction Manager initialization
815 jfs_err("txInit failed w/rc = %d", rc);
820 * I/O completion thread (endio)
822 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
823 if (IS_ERR(jfsIOthread)) {
824 rc = PTR_ERR(jfsIOthread);
825 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
829 if (commit_threads < 1)
830 commit_threads = num_online_cpus();
831 if (commit_threads > MAX_COMMIT_THREADS)
832 commit_threads = MAX_COMMIT_THREADS;
834 for (i = 0; i < commit_threads; i++) {
835 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
836 if (IS_ERR(jfsCommitThread[i])) {
837 rc = PTR_ERR(jfsCommitThread[i]);
838 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
840 goto kill_committask;
844 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
845 if (IS_ERR(jfsSyncThread)) {
846 rc = PTR_ERR(jfsSyncThread);
847 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
848 goto kill_committask;
855 return register_filesystem(&jfs_fs_type);
858 for (i = 0; i < commit_threads; i++)
859 kthread_stop(jfsCommitThread[i]);
860 kthread_stop(jfsIOthread);
866 kmem_cache_destroy(jfs_inode_cachep);
870 static void __exit exit_jfs_fs(void)
874 jfs_info("exit_jfs_fs called");
879 kthread_stop(jfsIOthread);
880 for (i = 0; i < commit_threads; i++)
881 kthread_stop(jfsCommitThread[i]);
882 kthread_stop(jfsSyncThread);
886 unregister_filesystem(&jfs_fs_type);
887 kmem_cache_destroy(jfs_inode_cachep);
890 module_init(init_jfs_fs)
891 module_exit(exit_jfs_fs)