Creation of Cybook 2416 (actually Gen4) repository

fs/jfs/Makefile

@@ -0,0 +1,16 @@
+#
+# Makefile for the Linux JFS filesystem routines.
+#
+
+obj-$(CONFIG_JFS_FS) += jfs.o
+
+jfs-y    := super.o file.o inode.o namei.o jfs_mount.o jfs_umount.o \
+	    jfs_xtree.o jfs_imap.o jfs_debug.o jfs_dmap.o \
+	    jfs_unicode.o jfs_dtree.o jfs_inode.o \
+	    jfs_extent.o symlink.o jfs_metapage.o \
+	    jfs_logmgr.o jfs_txnmgr.o jfs_uniupr.o \
+	    resize.o xattr.o ioctl.o
+
+jfs-$(CONFIG_JFS_POSIX_ACL) += acl.o
+
+EXTRA_CFLAGS += -D_JFS_4K

fs/jfs/acl.c

@@ -0,0 +1,246 @@
+/*
+ *   Copyright (C) International Business Machines  Corp., 2002-2004
+ *   Copyright (C) Andreas Gruenbacher, 2001
+ *   Copyright (C) Linus Torvalds, 1991, 1992
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/sched.h>
+#include <linux/fs.h>
+#include <linux/quotaops.h>
+#include <linux/posix_acl_xattr.h>
+#include "jfs_incore.h"
+#include "jfs_txnmgr.h"
+#include "jfs_xattr.h"
+#include "jfs_acl.h"
+
+static struct posix_acl *jfs_get_acl(struct inode *inode, int type)
+{
+	struct posix_acl *acl;
+	char *ea_name;
+	struct jfs_inode_info *ji = JFS_IP(inode);
+	struct posix_acl **p_acl;
+	int size;
+	char *value = NULL;
+
+	switch(type) {
+		case ACL_TYPE_ACCESS:
+			ea_name = POSIX_ACL_XATTR_ACCESS;
+			p_acl = &ji->i_acl;
+			break;
+		case ACL_TYPE_DEFAULT:
+			ea_name = POSIX_ACL_XATTR_DEFAULT;
+			p_acl = &ji->i_default_acl;
+			break;
+		default:
+			return ERR_PTR(-EINVAL);
+	}
+
+	if (*p_acl != JFS_ACL_NOT_CACHED)
+		return posix_acl_dup(*p_acl);
+
+	size = __jfs_getxattr(inode, ea_name, NULL, 0);
+
+	if (size > 0) {
+		value = kmalloc(size, GFP_KERNEL);
+		if (!value)
+			return ERR_PTR(-ENOMEM);
+		size = __jfs_getxattr(inode, ea_name, value, size);
+	}
+
+	if (size < 0) {
+		if (size == -ENODATA) {
+			*p_acl = NULL;
+			acl = NULL;
+		} else
+			acl = ERR_PTR(size);
+	} else {
+		acl = posix_acl_from_xattr(value, size);
+		if (!IS_ERR(acl))
+			*p_acl = posix_acl_dup(acl);
+	}
+	kfree(value);
+	return acl;
+}
+
+static int jfs_set_acl(tid_t tid, struct inode *inode, int type,
+		       struct posix_acl *acl)
+{
+	char *ea_name;
+	struct jfs_inode_info *ji = JFS_IP(inode);
+	struct posix_acl **p_acl;
+	int rc;
+	int size = 0;
+	char *value = NULL;
+
+	if (S_ISLNK(inode->i_mode))
+		return -EOPNOTSUPP;
+
+	switch(type) {
+		case ACL_TYPE_ACCESS:
+			ea_name = POSIX_ACL_XATTR_ACCESS;
+			p_acl = &ji->i_acl;
+			break;
+		case ACL_TYPE_DEFAULT:
+			ea_name = POSIX_ACL_XATTR_DEFAULT;
+			p_acl = &ji->i_default_acl;
+			if (!S_ISDIR(inode->i_mode))
+				return acl ? -EACCES : 0;
+			break;
+		default:
+			return -EINVAL;
+	}
+	if (acl) {
+		size = posix_acl_xattr_size(acl->a_count);
+		value = kmalloc(size, GFP_KERNEL);
+		if (!value)
+			return -ENOMEM;
+		rc = posix_acl_to_xattr(acl, value, size);
+		if (rc < 0)
+			goto out;
+	}
+	rc = __jfs_setxattr(tid, inode, ea_name, value, size, 0);
+out:
+	kfree(value);
+
+	if (!rc) {
+		if (*p_acl && (*p_acl != JFS_ACL_NOT_CACHED))
+			posix_acl_release(*p_acl);
+		*p_acl = posix_acl_dup(acl);
+	}
+	return rc;
+}
+
+static int jfs_check_acl(struct inode *inode, int mask)
+{
+	struct jfs_inode_info *ji = JFS_IP(inode);
+
+	if (ji->i_acl == JFS_ACL_NOT_CACHED) {
+		struct posix_acl *acl = jfs_get_acl(inode, ACL_TYPE_ACCESS);
+		if (IS_ERR(acl))
+			return PTR_ERR(acl);
+		posix_acl_release(acl);
+	}
+
+	if (ji->i_acl)
+		return posix_acl_permission(inode, ji->i_acl, mask);
+	return -EAGAIN;
+}
+
+int jfs_permission(struct inode *inode, int mask, struct nameidata *nd)
+{
+	return generic_permission(inode, mask, jfs_check_acl);
+}
+
+int jfs_init_acl(tid_t tid, struct inode *inode, struct inode *dir)
+{
+	struct posix_acl *acl = NULL;
+	struct posix_acl *clone;
+	mode_t mode;
+	int rc = 0;
+
+	if (S_ISLNK(inode->i_mode))
+		return 0;
+
+	acl = jfs_get_acl(dir, ACL_TYPE_DEFAULT);
+	if (IS_ERR(acl))
+		return PTR_ERR(acl);
+
+	if (acl) {
+		if (S_ISDIR(inode->i_mode)) {
+			rc = jfs_set_acl(tid, inode, ACL_TYPE_DEFAULT, acl);
+			if (rc)
+				goto cleanup;
+		}
+		clone = posix_acl_clone(acl, GFP_KERNEL);
+		if (!clone) {
+			rc = -ENOMEM;
+			goto cleanup;
+		}
+		mode = inode->i_mode;
+		rc = posix_acl_create_masq(clone, &mode);
+		if (rc >= 0) {
+			inode->i_mode = mode;
+			if (rc > 0)
+				rc = jfs_set_acl(tid, inode, ACL_TYPE_ACCESS,
+						 clone);
+		}
+		posix_acl_release(clone);
+cleanup:
+		posix_acl_release(acl);
+	} else
+		inode->i_mode &= ~current->fs->umask;
+
+	JFS_IP(inode)->mode2 = (JFS_IP(inode)->mode2 & 0xffff0000) |
+			       inode->i_mode;
+
+	return rc;
+}
+
+static int jfs_acl_chmod(struct inode *inode)
+{
+	struct posix_acl *acl, *clone;
+	int rc;
+
+	if (S_ISLNK(inode->i_mode))
+		return -EOPNOTSUPP;
+
+	acl = jfs_get_acl(inode, ACL_TYPE_ACCESS);
+	if (IS_ERR(acl) || !acl)
+		return PTR_ERR(acl);
+
+	clone = posix_acl_clone(acl, GFP_KERNEL);
+	posix_acl_release(acl);
+	if (!clone)
+		return -ENOMEM;
+
+	rc = posix_acl_chmod_masq(clone, inode->i_mode);
+	if (!rc) {
+		tid_t tid = txBegin(inode->i_sb, 0);
+		mutex_lock(&JFS_IP(inode)->commit_mutex);
+		rc = jfs_set_acl(tid, inode, ACL_TYPE_ACCESS, clone);
+		if (!rc)
+			rc = txCommit(tid, 1, &inode, 0);
+		txEnd(tid);
+		mutex_unlock(&JFS_IP(inode)->commit_mutex);
+	}
+
+	posix_acl_release(clone);
+	return rc;
+}
+
+int jfs_setattr(struct dentry *dentry, struct iattr *iattr)
+{
+	struct inode *inode = dentry->d_inode;
+	int rc;
+
+	rc = inode_change_ok(inode, iattr);
+	if (rc)
+		return rc;
+
+	if ((iattr->ia_valid & ATTR_UID && iattr->ia_uid != inode->i_uid) ||
+	    (iattr->ia_valid & ATTR_GID && iattr->ia_gid != inode->i_gid)) {
+		if (DQUOT_TRANSFER(inode, iattr))
+			return -EDQUOT;
+	}
+
+	rc = inode_setattr(inode, iattr);
+
+	if (!rc && (iattr->ia_valid & ATTR_MODE))
+		rc = jfs_acl_chmod(inode);
+
+	return rc;
+}

fs/jfs/endian24.h

@@ -0,0 +1,49 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2001
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_ENDIAN24
+#define	_H_ENDIAN24
+
+/*
+ *	endian24.h:
+ *
+ * Endian conversion for 24-byte data
+ *
+ */
+#define __swab24(x) \
+({ \
+	__u32 __x = (x); \
+	((__u32)( \
+		((__x & (__u32)0x000000ffUL) << 16) | \
+		 (__x & (__u32)0x0000ff00UL)        | \
+		((__x & (__u32)0x00ff0000UL) >> 16) )); \
+})
+
+#if (defined(__KERNEL__) && defined(__LITTLE_ENDIAN)) || (defined(__BYTE_ORDER) && (__BYTE_ORDER == __LITTLE_ENDIAN))
+	#define __cpu_to_le24(x) ((__u32)(x))
+	#define __le24_to_cpu(x) ((__u32)(x))
+#else
+	#define __cpu_to_le24(x) __swab24(x)
+	#define __le24_to_cpu(x) __swab24(x)
+#endif
+
+#ifdef __KERNEL__
+	#define cpu_to_le24 __cpu_to_le24
+	#define le24_to_cpu __le24_to_cpu
+#endif
+
+#endif				/* !_H_ENDIAN24 */

fs/jfs/file.c

@@ -0,0 +1,117 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include "jfs_incore.h"
+#include "jfs_inode.h"
+#include "jfs_dmap.h"
+#include "jfs_txnmgr.h"
+#include "jfs_xattr.h"
+#include "jfs_acl.h"
+#include "jfs_debug.h"
+
+int jfs_fsync(struct file *file, struct dentry *dentry, int datasync)
+{
+	struct inode *inode = dentry->d_inode;
+	int rc = 0;
+
+	if (!(inode->i_state & I_DIRTY) ||
+	    (datasync && !(inode->i_state & I_DIRTY_DATASYNC))) {
+		/* Make sure committed changes hit the disk */
+		jfs_flush_journal(JFS_SBI(inode->i_sb)->log, 1);
+		return rc;
+	}
+
+	rc |= jfs_commit_inode(inode, 1);
+
+	return rc ? -EIO : 0;
+}
+
+static int jfs_open(struct inode *inode, struct file *file)
+{
+	int rc;
+
+	if ((rc = generic_file_open(inode, file)))
+		return rc;
+
+	/*
+	 * We attempt to allow only one "active" file open per aggregate
+	 * group.  Otherwise, appending to files in parallel can cause
+	 * fragmentation within the files.
+	 *
+	 * If the file is empty, it was probably just created and going
+	 * to be written to.  If it has a size, we'll hold off until the
+	 * file is actually grown.
+	 */
+	if (S_ISREG(inode->i_mode) && file->f_mode & FMODE_WRITE &&
+	    (inode->i_size == 0)) {
+		struct jfs_inode_info *ji = JFS_IP(inode);
+		spin_lock_irq(&ji->ag_lock);
+		if (ji->active_ag == -1) {
+			ji->active_ag = ji->agno;
+			atomic_inc(
+			    &JFS_SBI(inode->i_sb)->bmap->db_active[ji->agno]);
+		}
+		spin_unlock_irq(&ji->ag_lock);
+	}
+
+	return 0;
+}
+static int jfs_release(struct inode *inode, struct file *file)
+{
+	struct jfs_inode_info *ji = JFS_IP(inode);
+
+	spin_lock_irq(&ji->ag_lock);
+	if (ji->active_ag != -1) {
+		struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
+		atomic_dec(&bmap->db_active[ji->active_ag]);
+		ji->active_ag = -1;
+	}
+	spin_unlock_irq(&ji->ag_lock);
+
+	return 0;
+}
+
+const struct inode_operations jfs_file_inode_operations = {
+	.truncate	= jfs_truncate,
+	.setxattr	= jfs_setxattr,
+	.getxattr	= jfs_getxattr,
+	.listxattr	= jfs_listxattr,
+	.removexattr	= jfs_removexattr,
+#ifdef CONFIG_JFS_POSIX_ACL
+	.setattr	= jfs_setattr,
+	.permission	= jfs_permission,
+#endif
+};
+
+const struct file_operations jfs_file_operations = {
+	.open		= jfs_open,
+	.llseek		= generic_file_llseek,
+	.write		= do_sync_write,
+	.read		= do_sync_read,
+	.aio_read	= generic_file_aio_read,
+	.aio_write	= generic_file_aio_write,
+	.mmap		= generic_file_mmap,
+	.sendfile	= generic_file_sendfile,
+	.splice_read	= generic_file_splice_read,
+	.splice_write	= generic_file_splice_write,
+	.fsync		= jfs_fsync,
+	.release	= jfs_release,
+	.ioctl		= jfs_ioctl,
+};

fs/jfs/inode.c

@@ -0,0 +1,365 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/mpage.h>
+#include <linux/buffer_head.h>
+#include <linux/pagemap.h>
+#include <linux/quotaops.h>
+#include "jfs_incore.h"
+#include "jfs_inode.h"
+#include "jfs_filsys.h"
+#include "jfs_imap.h"
+#include "jfs_extent.h"
+#include "jfs_unicode.h"
+#include "jfs_debug.h"
+
+
+void jfs_read_inode(struct inode *inode)
+{
+	if (diRead(inode)) {
+		make_bad_inode(inode);
+		return;
+	}
+
+	if (S_ISREG(inode->i_mode)) {
+		inode->i_op = &jfs_file_inode_operations;
+		inode->i_fop = &jfs_file_operations;
+		inode->i_mapping->a_ops = &jfs_aops;
+	} else if (S_ISDIR(inode->i_mode)) {
+		inode->i_op = &jfs_dir_inode_operations;
+		inode->i_fop = &jfs_dir_operations;
+	} else if (S_ISLNK(inode->i_mode)) {
+		if (inode->i_size >= IDATASIZE) {
+			inode->i_op = &page_symlink_inode_operations;
+			inode->i_mapping->a_ops = &jfs_aops;
+		} else
+			inode->i_op = &jfs_symlink_inode_operations;
+	} else {
+		inode->i_op = &jfs_file_inode_operations;
+		init_special_inode(inode, inode->i_mode, inode->i_rdev);
+	}
+	jfs_set_inode_flags(inode);
+}
+
+/*
+ * Workhorse of both fsync & write_inode
+ */
+int jfs_commit_inode(struct inode *inode, int wait)
+{
+	int rc = 0;
+	tid_t tid;
+	static int noisy = 5;
+
+	jfs_info("In jfs_commit_inode, inode = 0x%p", inode);
+
+	/*
+	 * Don't commit if inode has been committed since last being
+	 * marked dirty, or if it has been deleted.
+	 */
+	if (inode->i_nlink == 0 || !test_cflag(COMMIT_Dirty, inode))
+		return 0;
+
+	if (isReadOnly(inode)) {
+		/* kernel allows writes to devices on read-only
+		 * partitions and may think inode is dirty
+		 */
+		if (!special_file(inode->i_mode) && noisy) {
+			jfs_err("jfs_commit_inode(0x%p) called on "
+				   "read-only volume", inode);
+			jfs_err("Is remount racy?");
+			noisy--;
+		}
+		return 0;
+	}
+
+	tid = txBegin(inode->i_sb, COMMIT_INODE);
+	mutex_lock(&JFS_IP(inode)->commit_mutex);
+
+	/*
+	 * Retest inode state after taking commit_mutex
+	 */
+	if (inode->i_nlink && test_cflag(COMMIT_Dirty, inode))
+		rc = txCommit(tid, 1, &inode, wait ? COMMIT_SYNC : 0);
+
+	txEnd(tid);
+	mutex_unlock(&JFS_IP(inode)->commit_mutex);
+	return rc;
+}
+
+int jfs_write_inode(struct inode *inode, int wait)
+{
+	if (test_cflag(COMMIT_Nolink, inode))
+		return 0;
+	/*
+	 * If COMMIT_DIRTY is not set, the inode isn't really dirty.
+	 * It has been committed since the last change, but was still
+	 * on the dirty inode list.
+	 */
+	 if (!test_cflag(COMMIT_Dirty, inode)) {
+		/* Make sure committed changes hit the disk */
+		jfs_flush_journal(JFS_SBI(inode->i_sb)->log, wait);
+		return 0;
+	 }
+
+	if (jfs_commit_inode(inode, wait)) {
+		jfs_err("jfs_write_inode: jfs_commit_inode failed!");
+		return -EIO;
+	} else
+		return 0;
+}
+
+void jfs_delete_inode(struct inode *inode)
+{
+	jfs_info("In jfs_delete_inode, inode = 0x%p", inode);
+
+	if (!is_bad_inode(inode) &&
+	    (JFS_IP(inode)->fileset == FILESYSTEM_I)) {
+		truncate_inode_pages(&inode->i_data, 0);
+
+		if (test_cflag(COMMIT_Freewmap, inode))
+			jfs_free_zero_link(inode);
+
+		diFree(inode);
+
+		/*
+		 * Free the inode from the quota allocation.
+		 */
+		DQUOT_INIT(inode);
+		DQUOT_FREE_INODE(inode);
+		DQUOT_DROP(inode);
+	}
+
+	clear_inode(inode);
+}
+
+void jfs_dirty_inode(struct inode *inode)
+{
+	static int noisy = 5;
+
+	if (isReadOnly(inode)) {
+		if (!special_file(inode->i_mode) && noisy) {
+			/* kernel allows writes to devices on read-only
+			 * partitions and may try to mark inode dirty
+			 */
+			jfs_err("jfs_dirty_inode called on read-only volume");
+			jfs_err("Is remount racy?");
+			noisy--;
+		}
+		return;
+	}
+
+	set_cflag(COMMIT_Dirty, inode);
+}
+
+int jfs_get_block(struct inode *ip, sector_t lblock,
+		  struct buffer_head *bh_result, int create)
+{
+	s64 lblock64 = lblock;
+	int rc = 0;
+	xad_t xad;
+	s64 xaddr;
+	int xflag;
+	s32 xlen = bh_result->b_size >> ip->i_blkbits;
+
+	/*
+	 * Take appropriate lock on inode
+	 */
+	if (create)
+		IWRITE_LOCK(ip, RDWRLOCK_NORMAL);
+	else
+		IREAD_LOCK(ip, RDWRLOCK_NORMAL);
+
+	if (((lblock64 << ip->i_sb->s_blocksize_bits) < ip->i_size) &&
+	    (!xtLookup(ip, lblock64, xlen, &xflag, &xaddr, &xlen, 0)) &&
+	    xaddr) {
+		if (xflag & XAD_NOTRECORDED) {
+			if (!create)
+				/*
+				 * Allocated but not recorded, read treats
+				 * this as a hole
+				 */
+				goto unlock;
+#ifdef _JFS_4K
+			XADoffset(&xad, lblock64);
+			XADlength(&xad, xlen);
+			XADaddress(&xad, xaddr);
+#else				/* _JFS_4K */
+			/*
+			 * As long as block size = 4K, this isn't a problem.
+			 * We should mark the whole page not ABNR, but how
+			 * will we know to mark the other blocks BH_New?
+			 */
+			BUG();
+#endif				/* _JFS_4K */
+			rc = extRecord(ip, &xad);
+			if (rc)
+				goto unlock;
+			set_buffer_new(bh_result);
+		}
+
+		map_bh(bh_result, ip->i_sb, xaddr);
+		bh_result->b_size = xlen << ip->i_blkbits;
+		goto unlock;
+	}
+	if (!create)
+		goto unlock;
+
+	/*
+	 * Allocate a new block
+	 */
+#ifdef _JFS_4K
+	if ((rc = extHint(ip, lblock64 << ip->i_sb->s_blocksize_bits, &xad)))
+		goto unlock;
+	rc = extAlloc(ip, xlen, lblock64, &xad, false);
+	if (rc)
+		goto unlock;
+
+	set_buffer_new(bh_result);
+	map_bh(bh_result, ip->i_sb, addressXAD(&xad));
+	bh_result->b_size = lengthXAD(&xad) << ip->i_blkbits;
+
+#else				/* _JFS_4K */
+	/*
+	 * We need to do whatever it takes to keep all but the last buffers
+	 * in 4K pages - see jfs_write.c
+	 */
+	BUG();
+#endif				/* _JFS_4K */
+
+      unlock:
+	/*
+	 * Release lock on inode
+	 */
+	if (create)
+		IWRITE_UNLOCK(ip);
+	else
+		IREAD_UNLOCK(ip);
+	return rc;
+}
+
+static int jfs_writepage(struct page *page, struct writeback_control *wbc)
+{
+	return nobh_writepage(page, jfs_get_block, wbc);
+}
+
+static int jfs_writepages(struct address_space *mapping,
+			struct writeback_control *wbc)
+{
+	return mpage_writepages(mapping, wbc, jfs_get_block);
+}
+
+static int jfs_readpage(struct file *file, struct page *page)
+{
+	return mpage_readpage(page, jfs_get_block);
+}
+
+static int jfs_readpages(struct file *file, struct address_space *mapping,
+		struct list_head *pages, unsigned nr_pages)
+{
+	return mpage_readpages(mapping, pages, nr_pages, jfs_get_block);
+}
+
+static int jfs_prepare_write(struct file *file,
+			     struct page *page, unsigned from, unsigned to)
+{
+	return nobh_prepare_write(page, from, to, jfs_get_block);
+}
+
+static sector_t jfs_bmap(struct address_space *mapping, sector_t block)
+{
+	return generic_block_bmap(mapping, block, jfs_get_block);
+}
+
+static ssize_t jfs_direct_IO(int rw, struct kiocb *iocb,
+	const struct iovec *iov, loff_t offset, unsigned long nr_segs)
+{
+	struct file *file = iocb->ki_filp;
+	struct inode *inode = file->f_mapping->host;
+
+	return blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
+				offset, nr_segs, jfs_get_block, NULL);
+}
+
+const struct address_space_operations jfs_aops = {
+	.readpage	= jfs_readpage,
+	.readpages	= jfs_readpages,
+	.writepage	= jfs_writepage,
+	.writepages	= jfs_writepages,
+	.sync_page	= block_sync_page,
+	.prepare_write	= jfs_prepare_write,
+	.commit_write	= nobh_commit_write,
+	.bmap		= jfs_bmap,
+	.direct_IO	= jfs_direct_IO,
+};
+
+/*
+ * Guts of jfs_truncate.  Called with locks already held.  Can be called
+ * with directory for truncating directory index table.
+ */
+void jfs_truncate_nolock(struct inode *ip, loff_t length)
+{
+	loff_t newsize;
+	tid_t tid;
+
+	ASSERT(length >= 0);
+
+	if (test_cflag(COMMIT_Nolink, ip)) {
+		xtTruncate(0, ip, length, COMMIT_WMAP);
+		return;
+	}
+
+	do {
+		tid = txBegin(ip->i_sb, 0);
+
+		/*
+		 * The commit_mutex cannot be taken before txBegin.
+		 * txBegin may block and there is a chance the inode
+		 * could be marked dirty and need to be committed
+		 * before txBegin unblocks
+		 */
+		mutex_lock(&JFS_IP(ip)->commit_mutex);
+
+		newsize = xtTruncate(tid, ip, length,
+				     COMMIT_TRUNCATE | COMMIT_PWMAP);
+		if (newsize < 0) {
+			txEnd(tid);
+			mutex_unlock(&JFS_IP(ip)->commit_mutex);
+			break;
+		}
+
+		ip->i_mtime = ip->i_ctime = CURRENT_TIME;
+		mark_inode_dirty(ip);
+
+		txCommit(tid, 1, &ip, 0);
+		txEnd(tid);
+		mutex_unlock(&JFS_IP(ip)->commit_mutex);
+	} while (newsize > length);	/* Truncate isn't always atomic */
+}
+
+void jfs_truncate(struct inode *ip)
+{
+	jfs_info("jfs_truncate: size = 0x%lx", (ulong) ip->i_size);
+
+	nobh_truncate_page(ip->i_mapping, ip->i_size);
+
+	IWRITE_LOCK(ip, RDWRLOCK_NORMAL);
+	jfs_truncate_nolock(ip, ip->i_size);
+	IWRITE_UNLOCK(ip);
+}

fs/jfs/ioctl.c

@@ -0,0 +1,107 @@
+/*
+ * linux/fs/jfs/ioctl.c
+ *
+ * Copyright (C) 2006 Herbert Poetzl
+ * adapted from Remy Card's ext2/ioctl.c
+ */
+
+#include <linux/fs.h>
+#include <linux/ctype.h>
+#include <linux/capability.h>
+#include <linux/time.h>
+#include <linux/sched.h>
+#include <asm/current.h>
+#include <asm/uaccess.h>
+
+#include "jfs_incore.h"
+#include "jfs_dinode.h"
+#include "jfs_inode.h"
+
+
+static struct {
+	long jfs_flag;
+	long ext2_flag;
+} jfs_map[] = {
+	{JFS_NOATIME_FL,	FS_NOATIME_FL},
+	{JFS_DIRSYNC_FL,	FS_DIRSYNC_FL},
+	{JFS_SYNC_FL,		FS_SYNC_FL},
+	{JFS_SECRM_FL,		FS_SECRM_FL},
+	{JFS_UNRM_FL,		FS_UNRM_FL},
+	{JFS_APPEND_FL,		FS_APPEND_FL},
+	{JFS_IMMUTABLE_FL,	FS_IMMUTABLE_FL},
+	{0, 0},
+};
+
+static long jfs_map_ext2(unsigned long flags, int from)
+{
+	int index=0;
+	long mapped=0;
+
+	while (jfs_map[index].jfs_flag) {
+		if (from) {
+			if (jfs_map[index].ext2_flag & flags)
+				mapped |= jfs_map[index].jfs_flag;
+		} else {
+			if (jfs_map[index].jfs_flag & flags)
+				mapped |= jfs_map[index].ext2_flag;
+		}
+		index++;
+	}
+	return mapped;
+}
+
+
+int jfs_ioctl(struct inode * inode, struct file * filp, unsigned int cmd,
+		unsigned long arg)
+{
+	struct jfs_inode_info *jfs_inode = JFS_IP(inode);
+	unsigned int flags;
+
+	switch (cmd) {
+	case JFS_IOC_GETFLAGS:
+		flags = jfs_inode->mode2 & JFS_FL_USER_VISIBLE;
+		flags = jfs_map_ext2(flags, 0);
+		return put_user(flags, (int __user *) arg);
+	case JFS_IOC_SETFLAGS: {
+		unsigned int oldflags;
+
+		if (IS_RDONLY(inode))
+			return -EROFS;
+
+		if ((current->fsuid != inode->i_uid) && !capable(CAP_FOWNER))
+			return -EACCES;
+
+		if (get_user(flags, (int __user *) arg))
+			return -EFAULT;
+
+		flags = jfs_map_ext2(flags, 1);
+		if (!S_ISDIR(inode->i_mode))
+			flags &= ~JFS_DIRSYNC_FL;
+
+		oldflags = jfs_inode->mode2;
+
+		/*
+		 * The IMMUTABLE and APPEND_ONLY flags can only be changed by
+		 * the relevant capability.
+		 */
+		if ((oldflags & JFS_IMMUTABLE_FL) ||
+			((flags ^ oldflags) &
+			(JFS_APPEND_FL | JFS_IMMUTABLE_FL))) {
+			if (!capable(CAP_LINUX_IMMUTABLE))
+				return -EPERM;
+		}
+
+		flags = flags & JFS_FL_USER_MODIFIABLE;
+		flags |= oldflags & ~JFS_FL_USER_MODIFIABLE;
+		jfs_inode->mode2 = flags;
+
+		jfs_set_inode_flags(inode);
+		inode->i_ctime = CURRENT_TIME_SEC;
+		mark_inode_dirty(inode);
+		return 0;
+	}
+	default:
+		return -ENOTTY;
+	}
+}
+

fs/jfs/jfs_acl.h

@@ -0,0 +1,36 @@
+/*
+ *   Copyright (C) International Business Machines  Corp., 2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_ACL
+#define _H_JFS_ACL
+
+#ifdef CONFIG_JFS_POSIX_ACL
+
+int jfs_permission(struct inode *, int, struct nameidata *);
+int jfs_init_acl(tid_t, struct inode *, struct inode *);
+int jfs_setattr(struct dentry *, struct iattr *);
+
+#else
+
+static inline int jfs_init_acl(tid_t tid, struct inode *inode,
+			       struct inode *dir)
+{
+	return 0;
+}
+
+#endif
+#endif		/* _H_JFS_ACL */

fs/jfs/jfs_btree.h

@@ -0,0 +1,172 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_BTREE
+#define _H_JFS_BTREE
+
+/*
+ *	jfs_btree.h: B+-tree
+ *
+ * JFS B+-tree (dtree and xtree) common definitions
+ */
+
+/*
+ *	basic btree page - btpage
+ *
+struct btpage {
+	s64 next;		right sibling bn
+	s64 prev;		left sibling bn
+
+	u8 flag;
+	u8 rsrvd[7];		type specific
+	s64 self;		self address
+
+	u8 entry[4064];
+};						*/
+
+/* btpaget_t flag */
+#define BT_TYPE		0x07	/* B+-tree index */
+#define	BT_ROOT		0x01	/* root page */
+#define	BT_LEAF		0x02	/* leaf page */
+#define	BT_INTERNAL	0x04	/* internal page */
+#define	BT_RIGHTMOST	0x10	/* rightmost page */
+#define	BT_LEFTMOST	0x20	/* leftmost page */
+#define	BT_SWAPPED	0x80	/* used by fsck for endian swapping */
+
+/* btorder (in inode) */
+#define	BT_RANDOM		0x0000
+#define	BT_SEQUENTIAL		0x0001
+#define	BT_LOOKUP		0x0010
+#define	BT_INSERT		0x0020
+#define	BT_DELETE		0x0040
+
+/*
+ *	btree page buffer cache access
+ */
+#define BT_IS_ROOT(MP) (((MP)->xflag & COMMIT_PAGE) == 0)
+
+/* get page from buffer page */
+#define BT_PAGE(IP, MP, TYPE, ROOT)\
+	(BT_IS_ROOT(MP) ? (TYPE *)&JFS_IP(IP)->ROOT : (TYPE *)(MP)->data)
+
+/* get the page buffer and the page for specified block address */
+#define BT_GETPAGE(IP, BN, MP, TYPE, SIZE, P, RC, ROOT)\
+{\
+	if ((BN) == 0)\
+	{\
+		MP = (struct metapage *)&JFS_IP(IP)->bxflag;\
+		P = (TYPE *)&JFS_IP(IP)->ROOT;\
+		RC = 0;\
+	}\
+	else\
+	{\
+		MP = read_metapage((IP), BN, SIZE, 1);\
+		if (MP) {\
+			RC = 0;\
+			P = (MP)->data;\
+		} else {\
+			P = NULL;\
+			jfs_err("bread failed!");\
+			RC = -EIO;\
+		}\
+	}\
+}
+
+#define BT_MARK_DIRTY(MP, IP)\
+{\
+	if (BT_IS_ROOT(MP))\
+		mark_inode_dirty(IP);\
+	else\
+		mark_metapage_dirty(MP);\
+}
+
+/* put the page buffer */
+#define BT_PUTPAGE(MP)\
+{\
+	if (! BT_IS_ROOT(MP)) \
+		release_metapage(MP); \
+}
+
+
+/*
+ *	btree traversal stack
+ *
+ * record the path traversed during the search;
+ * top frame record the leaf page/entry selected.
+ */
+struct btframe {	/* stack frame */
+	s64 bn;			/* 8: */
+	s16 index;		/* 2: */
+	s16 lastindex;		/* 2: unused */
+	struct metapage *mp;	/* 4/8: */
+};				/* (16/24) */
+
+struct btstack {
+	struct btframe *top;
+	int nsplit;
+	struct btframe stack[MAXTREEHEIGHT];
+};
+
+#define BT_CLR(btstack)\
+	(btstack)->top = (btstack)->stack
+
+#define BT_STACK_FULL(btstack)\
+	( (btstack)->top == &((btstack)->stack[MAXTREEHEIGHT-1]))
+
+#define BT_PUSH(BTSTACK, BN, INDEX)\
+{\
+	assert(!BT_STACK_FULL(BTSTACK));\
+	(BTSTACK)->top->bn = BN;\
+	(BTSTACK)->top->index = INDEX;\
+	++(BTSTACK)->top;\
+}
+
+#define BT_POP(btstack)\
+	( (btstack)->top == (btstack)->stack ? NULL : --(btstack)->top )
+
+#define BT_STACK(btstack)\
+	( (btstack)->top == (btstack)->stack ? NULL : (btstack)->top )
+
+static inline void BT_STACK_DUMP(struct btstack *btstack)
+{
+	int i;
+	printk("btstack dump:\n");
+	for (i = 0; i < MAXTREEHEIGHT; i++)
+		printk(KERN_ERR "bn = %Lx, index = %d\n",
+		       (long long)btstack->stack[i].bn,
+		       btstack->stack[i].index);
+}
+
+/* retrieve search results */
+#define BT_GETSEARCH(IP, LEAF, BN, MP, TYPE, P, INDEX, ROOT)\
+{\
+	BN = (LEAF)->bn;\
+	MP = (LEAF)->mp;\
+	if (BN)\
+		P = (TYPE *)MP->data;\
+	else\
+		P = (TYPE *)&JFS_IP(IP)->ROOT;\
+	INDEX = (LEAF)->index;\
+}
+
+/* put the page buffer of search */
+#define BT_PUTSEARCH(BTSTACK)\
+{\
+	if (! BT_IS_ROOT((BTSTACK)->top->mp))\
+		release_metapage((BTSTACK)->top->mp);\
+}
+#endif				/* _H_JFS_BTREE */

fs/jfs/jfs_debug.c

@@ -0,0 +1,144 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/ctype.h>
+#include <linux/module.h>
+#include <linux/proc_fs.h>
+#include <asm/uaccess.h>
+#include "jfs_incore.h"
+#include "jfs_filsys.h"
+#include "jfs_debug.h"
+
+#ifdef CONFIG_JFS_DEBUG
+void dump_mem(char *label, void *data, int length)
+{
+	int i, j;
+	int *intptr = data;
+	char *charptr = data;
+	char buf[10], line[80];
+
+	printk("%s: dump of %d bytes of data at 0x%p\n\n", label, length,
+	       data);
+	for (i = 0; i < length; i += 16) {
+		line[0] = 0;
+		for (j = 0; (j < 4) && (i + j * 4 < length); j++) {
+			sprintf(buf, " %08x", intptr[i / 4 + j]);
+			strcat(line, buf);
+		}
+		buf[0] = ' ';
+		buf[2] = 0;
+		for (j = 0; (j < 16) && (i + j < length); j++) {
+			buf[1] =
+			    isprint(charptr[i + j]) ? charptr[i + j] : '.';
+			strcat(line, buf);
+		}
+		printk("%s\n", line);
+	}
+}
+#endif
+
+#ifdef PROC_FS_JFS /* see jfs_debug.h */
+
+static struct proc_dir_entry *base;
+#ifdef CONFIG_JFS_DEBUG
+static int loglevel_read(char *page, char **start, off_t off,
+			 int count, int *eof, void *data)
+{
+	int len;
+
+	len = sprintf(page, "%d\n", jfsloglevel);
+
+	len -= off;
+	*start = page + off;
+
+	if (len > count)
+		len = count;
+	else
+		*eof = 1;
+
+	if (len < 0)
+		len = 0;
+
+	return len;
+}
+
+static int loglevel_write(struct file *file, const char __user *buffer,
+			unsigned long count, void *data)
+{
+	char c;
+
+	if (get_user(c, buffer))
+		return -EFAULT;
+
+	/* yes, I know this is an ASCIIism.  --hch */
+	if (c < '0' || c > '9')
+		return -EINVAL;
+	jfsloglevel = c - '0';
+	return count;
+}
+#endif
+
+static struct {
+	const char	*name;
+	read_proc_t	*read_fn;
+	write_proc_t	*write_fn;
+} Entries[] = {
+#ifdef CONFIG_JFS_STATISTICS
+	{ "lmstats",	jfs_lmstats_read, },
+	{ "txstats",	jfs_txstats_read, },
+	{ "xtstat",	jfs_xtstat_read, },
+	{ "mpstat",	jfs_mpstat_read, },
+#endif
+#ifdef CONFIG_JFS_DEBUG
+	{ "TxAnchor",	jfs_txanchor_read, },
+	{ "loglevel",	loglevel_read, loglevel_write }
+#endif
+};
+#define NPROCENT	ARRAY_SIZE(Entries)
+
+void jfs_proc_init(void)
+{
+	int i;
+
+	if (!(base = proc_mkdir("jfs", proc_root_fs)))
+		return;
+	base->owner = THIS_MODULE;
+
+	for (i = 0; i < NPROCENT; i++) {
+		struct proc_dir_entry *p;
+		if ((p = create_proc_entry(Entries[i].name, 0, base))) {
+			p->read_proc = Entries[i].read_fn;
+			p->write_proc = Entries[i].write_fn;
+		}
+	}
+}
+
+void jfs_proc_clean(void)
+{
+	int i;
+
+	if (base) {
+		for (i = 0; i < NPROCENT; i++)
+			remove_proc_entry(Entries[i].name, base);
+		remove_proc_entry("jfs", proc_root_fs);
+	}
+}
+
+#endif /* PROC_FS_JFS */

fs/jfs/jfs_debug.h

@@ -0,0 +1,124 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_DEBUG
+#define _H_JFS_DEBUG
+
+/*
+ *	jfs_debug.h
+ *
+ * global debug message, data structure/macro definitions
+ * under control of CONFIG_JFS_DEBUG, CONFIG_JFS_STATISTICS;
+ */
+
+/*
+ * Create /proc/fs/jfs if procfs is enabled andeither
+ * CONFIG_JFS_DEBUG or CONFIG_JFS_STATISTICS is defined
+ */
+#if defined(CONFIG_PROC_FS) && (defined(CONFIG_JFS_DEBUG) || defined(CONFIG_JFS_STATISTICS))
+#define PROC_FS_JFS
+extern void jfs_proc_init(void);
+extern void jfs_proc_clean(void);
+#endif
+
+/*
+ *	assert with traditional printf/panic
+ */
+#define assert(p) do {	\
+	if (!(p)) {	\
+		printk(KERN_CRIT "BUG at %s:%d assert(%s)\n",	\
+		       __FILE__, __LINE__, #p);			\
+		BUG();	\
+	}		\
+} while (0)
+
+/*
+ *	debug ON
+ *	--------
+ */
+#ifdef CONFIG_JFS_DEBUG
+#define ASSERT(p) assert(p)
+
+/* printk verbosity */
+#define JFS_LOGLEVEL_ERR 1
+#define JFS_LOGLEVEL_WARN 2
+#define JFS_LOGLEVEL_DEBUG 3
+#define JFS_LOGLEVEL_INFO 4
+
+extern int jfsloglevel;
+
+extern void dump_mem(char *label, void *data, int length);
+extern int jfs_txanchor_read(char *, char **, off_t, int, int *, void *);
+
+/* information message: e.g., configuration, major event */
+#define jfs_info(fmt, arg...) do {			\
+	if (jfsloglevel >= JFS_LOGLEVEL_INFO)		\
+		printk(KERN_INFO fmt "\n", ## arg);	\
+} while (0)
+
+/* debug message: ad hoc */
+#define jfs_debug(fmt, arg...) do {			\
+	if (jfsloglevel >= JFS_LOGLEVEL_DEBUG)		\
+		printk(KERN_DEBUG fmt "\n", ## arg);	\
+} while (0)
+
+/* warn message: */
+#define jfs_warn(fmt, arg...) do {			\
+	if (jfsloglevel >= JFS_LOGLEVEL_WARN)		\
+		printk(KERN_WARNING fmt "\n", ## arg);	\
+} while (0)
+
+/* error event message: e.g., i/o error */
+#define jfs_err(fmt, arg...) do {			\
+	if (jfsloglevel >= JFS_LOGLEVEL_ERR)		\
+		printk(KERN_ERR fmt "\n", ## arg);	\
+} while (0)
+
+/*
+ *	debug OFF
+ *	---------
+ */
+#else				/* CONFIG_JFS_DEBUG */
+#define dump_mem(label,data,length) do {} while (0)
+#define ASSERT(p) do {} while (0)
+#define jfs_info(fmt, arg...) do {} while (0)
+#define jfs_debug(fmt, arg...) do {} while (0)
+#define jfs_warn(fmt, arg...) do {} while (0)
+#define jfs_err(fmt, arg...) do {} while (0)
+#endif				/* CONFIG_JFS_DEBUG */
+
+/*
+ *	statistics
+ *	----------
+ */
+#ifdef	CONFIG_JFS_STATISTICS
+extern int jfs_lmstats_read(char *, char **, off_t, int, int *, void *);
+extern int jfs_txstats_read(char *, char **, off_t, int, int *, void *);
+extern int jfs_mpstat_read(char *, char **, off_t, int, int *, void *);
+extern int jfs_xtstat_read(char *, char **, off_t, int, int *, void *);
+
+#define	INCREMENT(x)		((x)++)
+#define	DECREMENT(x)		((x)--)
+#define	HIGHWATERMARK(x,y)	((x) = max((x), (y)))
+#else
+#define	INCREMENT(x)
+#define	DECREMENT(x)
+#define	HIGHWATERMARK(x,y)
+#endif				/* CONFIG_JFS_STATISTICS */
+
+#endif				/* _H_JFS_DEBUG */

fs/jfs/jfs_dinode.h

@@ -0,0 +1,174 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2001
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_DINODE
+#define _H_JFS_DINODE
+
+/*
+ *      jfs_dinode.h: on-disk inode manager
+ */
+
+#define INODESLOTSIZE           128
+#define L2INODESLOTSIZE         7
+#define log2INODESIZE           9	/* log2(bytes per dinode) */
+
+
+/*
+ *      on-disk inode : 512 bytes
+ *
+ * note: align 64-bit fields on 8-byte boundary.
+ */
+struct dinode {
+	/*
+	 *      I. base area (128 bytes)
+	 *      ------------------------
+	 *
+	 * define generic/POSIX attributes
+	 */
+	__le32 di_inostamp;	/* 4: stamp to show inode belongs to fileset */
+	__le32 di_fileset;	/* 4: fileset number */
+	__le32 di_number;	/* 4: inode number, aka file serial number */
+	__le32 di_gen;		/* 4: inode generation number */
+
+	pxd_t di_ixpxd;		/* 8: inode extent descriptor */
+
+	__le64 di_size;		/* 8: size */
+	__le64 di_nblocks;	/* 8: number of blocks allocated */
+
+	__le32 di_nlink;	/* 4: number of links to the object */
+
+	__le32 di_uid;		/* 4: user id of owner */
+	__le32 di_gid;		/* 4: group id of owner */
+
+	__le32 di_mode;		/* 4: attribute, format and permission */
+
+	struct timestruc_t di_atime;	/* 8: time last data accessed */
+	struct timestruc_t di_ctime;	/* 8: time last status changed */
+	struct timestruc_t di_mtime;	/* 8: time last data modified */
+	struct timestruc_t di_otime;	/* 8: time created */
+
+	dxd_t di_acl;		/* 16: acl descriptor */
+
+	dxd_t di_ea;		/* 16: ea descriptor */
+
+	__le32 di_next_index;	/* 4: Next available dir_table index */
+
+	__le32 di_acltype;	/* 4: Type of ACL */
+
+	/*
+	 *      Extension Areas.
+	 *
+	 *      Historically, the inode was partitioned into 4 128-byte areas,
+	 *      the last 3 being defined as unions which could have multiple
+	 *      uses.  The first 96 bytes had been completely unused until
+	 *      an index table was added to the directory.  It is now more
+	 *      useful to describe the last 3/4 of the inode as a single
+	 *      union.  We would probably be better off redesigning the
+	 *      entire structure from scratch, but we don't want to break
+	 *      commonality with OS/2's JFS at this time.
+	 */
+	union {
+		struct {
+			/*
+			 * This table contains the information needed to
+			 * find a directory entry from a 32-bit index.
+			 * If the index is small enough, the table is inline,
+			 * otherwise, an x-tree root overlays this table
+			 */
+			struct dir_table_slot _table[12]; /* 96: inline */
+
+			dtroot_t _dtroot;		/* 288: dtree root */
+		} _dir;					/* (384) */
+#define di_dirtable	u._dir._table
+#define di_dtroot	u._dir._dtroot
+#define di_parent       di_dtroot.header.idotdot
+#define di_DASD		di_dtroot.header.DASD
+
+		struct {
+			union {
+				u8 _data[96];		/* 96: unused */
+				struct {
+					void *_imap;	/* 4: unused */
+					__le32 _gengen;	/* 4: generator */
+				} _imap;
+			} _u1;				/* 96: */
+#define di_gengen	u._file._u1._imap._gengen
+
+			union {
+				xtpage_t _xtroot;
+				struct {
+					u8 unused[16];	/* 16: */
+					dxd_t _dxd;	/* 16: */
+					union {
+						__le32 _rdev;	/* 4: */
+						u8 _fastsymlink[128];
+					} _u;
+					u8 _inlineea[128];
+				} _special;
+			} _u2;
+		} _file;
+#define di_xtroot	u._file._u2._xtroot
+#define di_dxd		u._file._u2._special._dxd
+#define di_btroot	di_xtroot
+#define di_inlinedata	u._file._u2._special._u
+#define di_rdev		u._file._u2._special._u._rdev
+#define di_fastsymlink	u._file._u2._special._u._fastsymlink
+#define di_inlineea     u._file._u2._special._inlineea
+	} u;
+};
+
+/* extended mode bits (on-disk inode di_mode) */
+#define IFJOURNAL       0x00010000	/* journalled file */
+#define ISPARSE         0x00020000	/* sparse file enabled */
+#define INLINEEA        0x00040000	/* inline EA area free */
+#define ISWAPFILE	0x00800000	/* file open for pager swap space */
+
+/* more extended mode bits: attributes for OS/2 */
+#define IREADONLY	0x02000000	/* no write access to file */
+#define IHIDDEN		0x04000000	/* hidden file */
+#define ISYSTEM		0x08000000	/* system file */
+
+#define IDIRECTORY	0x20000000	/* directory (shadow of real bit) */
+#define IARCHIVE	0x40000000	/* file archive bit */
+#define INEWNAME	0x80000000	/* non-8.3 filename format */
+
+#define IRASH		0x4E000000	/* mask for changeable attributes */
+#define ATTRSHIFT	25	/* bits to shift to move attribute
+				   specification to mode position */
+
+/* extended attributes for Linux */
+
+#define JFS_NOATIME_FL		0x00080000 /* do not update atime */
+
+#define JFS_DIRSYNC_FL		0x00100000 /* dirsync behaviour */
+#define JFS_SYNC_FL		0x00200000 /* Synchronous updates */
+#define JFS_SECRM_FL		0x00400000 /* Secure deletion */
+#define JFS_UNRM_FL		0x00800000 /* allow for undelete */
+
+#define JFS_APPEND_FL		0x01000000 /* writes to file may only append */
+#define JFS_IMMUTABLE_FL	0x02000000 /* Immutable file */
+
+#define JFS_FL_USER_VISIBLE	0x03F80000
+#define JFS_FL_USER_MODIFIABLE	0x03F80000
+#define JFS_FL_INHERIT		0x03C80000
+
+/* These are identical to EXT[23]_IOC_GETFLAGS/SETFLAGS */
+#define JFS_IOC_GETFLAGS	_IOR('f', 1, long)
+#define JFS_IOC_SETFLAGS	_IOW('f', 2, long)
+
+
+#endif /*_H_JFS_DINODE */

fs/jfs/jfs_dmap.h

@@ -0,0 +1,314 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_DMAP
+#define _H_JFS_DMAP
+
+#include "jfs_txnmgr.h"
+
+#define BMAPVERSION	1	/* version number */
+#define	TREESIZE	(256+64+16+4+1)	/* size of a dmap tree */
+#define	LEAFIND		(64+16+4+1)	/* index of 1st leaf of a dmap tree */
+#define LPERDMAP	256	/* num leaves per dmap tree */
+#define L2LPERDMAP	8	/* l2 number of leaves per dmap tree */
+#define	DBWORD		32	/* # of blks covered by a map word */
+#define	L2DBWORD	5	/* l2 # of blks covered by a mword */
+#define BUDMIN		L2DBWORD	/* max free string in a map word */
+#define BPERDMAP	(LPERDMAP * DBWORD)	/* num of blks per dmap */
+#define L2BPERDMAP	13	/* l2 num of blks per dmap */
+#define CTLTREESIZE	(1024+256+64+16+4+1)	/* size of a dmapctl tree */
+#define CTLLEAFIND	(256+64+16+4+1)	/* idx of 1st leaf of a dmapctl tree */
+#define LPERCTL		1024	/* num of leaves per dmapctl tree */
+#define L2LPERCTL	10	/* l2 num of leaves per dmapctl tree */
+#define	ROOT		0	/* index of the root of a tree */
+#define	NOFREE		((s8) -1)	/* no blocks free */
+#define	MAXAG		128	/* max number of allocation groups */
+#define L2MAXAG		7	/* l2 max num of AG */
+#define L2MINAGSZ	25	/* l2 of minimum AG size in bytes */
+#define	BMAPBLKNO	0	/* lblkno of bmap within the map */
+
+/*
+ * maximum l2 number of disk blocks at the various dmapctl levels.
+ */
+#define	L2MAXL0SIZE	(L2BPERDMAP + 1 * L2LPERCTL)
+#define	L2MAXL1SIZE	(L2BPERDMAP + 2 * L2LPERCTL)
+#define	L2MAXL2SIZE	(L2BPERDMAP + 3 * L2LPERCTL)
+
+/*
+ * maximum number of disk blocks at the various dmapctl levels.
+ */
+#define	MAXL0SIZE	((s64)1 << L2MAXL0SIZE)
+#define	MAXL1SIZE	((s64)1 << L2MAXL1SIZE)
+#define	MAXL2SIZE	((s64)1 << L2MAXL2SIZE)
+
+#define	MAXMAPSIZE	MAXL2SIZE	/* maximum aggregate map size */
+
+/*
+ * determine the maximum free string for four (lower level) nodes
+ * of the tree.
+ */
+static __inline signed char TREEMAX(signed char *cp)
+{
+	signed char tmp1, tmp2;
+
+	tmp1 = max(*(cp+2), *(cp+3));
+	tmp2 = max(*(cp), *(cp+1));
+
+	return max(tmp1, tmp2);
+}
+
+/*
+ * convert disk block number to the logical block number of the dmap
+ * describing the disk block.  s is the log2(number of logical blocks per page)
+ *
+ * The calculation figures out how many logical pages are in front of the dmap.
+ *	- the number of dmaps preceding it
+ *	- the number of L0 pages preceding its L0 page
+ *	- the number of L1 pages preceding its L1 page
+ *	- 3 is added to account for the L2, L1, and L0 page for this dmap
+ *	- 1 is added to account for the control page of the map.
+ */
+#define BLKTODMAP(b,s)    \
+        ((((b) >> 13) + ((b) >> 23) + ((b) >> 33) + 3 + 1) << (s))
+
+/*
+ * convert disk block number to the logical block number of the LEVEL 0
+ * dmapctl describing the disk block.  s is the log2(number of logical blocks
+ * per page)
+ *
+ * The calculation figures out how many logical pages are in front of the L0.
+ *	- the number of dmap pages preceding it
+ *	- the number of L0 pages preceding it
+ *	- the number of L1 pages preceding its L1 page
+ *	- 2 is added to account for the L2, and L1 page for this L0
+ *	- 1 is added to account for the control page of the map.
+ */
+#define BLKTOL0(b,s)      \
+        (((((b) >> 23) << 10) + ((b) >> 23) + ((b) >> 33) + 2 + 1) << (s))
+
+/*
+ * convert disk block number to the logical block number of the LEVEL 1
+ * dmapctl describing the disk block.  s is the log2(number of logical blocks
+ * per page)
+ *
+ * The calculation figures out how many logical pages are in front of the L1.
+ *	- the number of dmap pages preceding it
+ *	- the number of L0 pages preceding it
+ *	- the number of L1 pages preceding it
+ *	- 1 is added to account for the L2 page
+ *	- 1 is added to account for the control page of the map.
+ */
+#define BLKTOL1(b,s)      \
+     (((((b) >> 33) << 20) + (((b) >> 33) << 10) + ((b) >> 33) + 1 + 1) << (s))
+
+/*
+ * convert disk block number to the logical block number of the dmapctl
+ * at the specified level which describes the disk block.
+ */
+#define BLKTOCTL(b,s,l)   \
+        (((l) == 2) ? 1 : ((l) == 1) ? BLKTOL1((b),(s)) : BLKTOL0((b),(s)))
+
+/*
+ * convert aggregate map size to the zero origin dmapctl level of the
+ * top dmapctl.
+ */
+#define	BMAPSZTOLEV(size)	\
+	(((size) <= MAXL0SIZE) ? 0 : ((size) <= MAXL1SIZE) ? 1 : 2)
+
+/* convert disk block number to allocation group number.
+ */
+#define BLKTOAG(b,sbi)	((b) >> ((sbi)->bmap->db_agl2size))
+
+/* convert allocation group number to starting disk block
+ * number.
+ */
+#define AGTOBLK(a,ip)	\
+	((s64)(a) << (JFS_SBI((ip)->i_sb)->bmap->db_agl2size))
+
+/*
+ *	dmap summary tree
+ *
+ * dmaptree must be consistent with dmapctl.
+ */
+struct dmaptree {
+	__le32 nleafs;		/* 4: number of tree leafs      */
+	__le32 l2nleafs;	/* 4: l2 number of tree leafs   */
+	__le32 leafidx;		/* 4: index of first tree leaf  */
+	__le32 height;		/* 4: height of the tree        */
+	s8 budmin;		/* 1: min l2 tree leaf value to combine */
+	s8 stree[TREESIZE];	/* TREESIZE: tree               */
+	u8 pad[2];		/* 2: pad to word boundary      */
+};				/* - 360 -                      */
+
+/*
+ *	dmap page per 8K blocks bitmap
+ */
+struct dmap {
+	__le32 nblocks;		/* 4: num blks covered by this dmap     */
+	__le32 nfree;		/* 4: num of free blks in this dmap     */
+	__le64 start;		/* 8: starting blkno for this dmap      */
+	struct dmaptree tree;	/* 360: dmap tree                       */
+	u8 pad[1672];		/* 1672: pad to 2048 bytes              */
+	__le32 wmap[LPERDMAP];	/* 1024: bits of the working map        */
+	__le32 pmap[LPERDMAP];	/* 1024: bits of the persistent map     */
+};				/* - 4096 -                             */
+
+/*
+ *	disk map control page per level.
+ *
+ * dmapctl must be consistent with dmaptree.
+ */
+struct dmapctl {
+	__le32 nleafs;		/* 4: number of tree leafs      */
+	__le32 l2nleafs;	/* 4: l2 number of tree leafs   */
+	__le32 leafidx;		/* 4: index of the first tree leaf      */
+	__le32 height;		/* 4: height of tree            */
+	s8 budmin;		/* 1: minimum l2 tree leaf value        */
+	s8 stree[CTLTREESIZE];	/* CTLTREESIZE: dmapctl tree    */
+	u8 pad[2714];		/* 2714: pad to 4096            */
+};				/* - 4096 -                     */
+
+/*
+ *	common definition for dmaptree within dmap and dmapctl
+ */
+typedef union dmtree {
+	struct dmaptree t1;
+	struct dmapctl t2;
+} dmtree_t;
+
+/* macros for accessing fields within dmtree */
+#define	dmt_nleafs	t1.nleafs
+#define	dmt_l2nleafs	t1.l2nleafs
+#define	dmt_leafidx	t1.leafidx
+#define	dmt_height	t1.height
+#define	dmt_budmin	t1.budmin
+#define	dmt_stree	t1.stree
+
+/*
+ *	on-disk aggregate disk allocation map descriptor.
+ */
+struct dbmap_disk {
+	__le64 dn_mapsize;	/* 8: number of blocks in aggregate     */
+	__le64 dn_nfree;	/* 8: num free blks in aggregate map    */
+	__le32 dn_l2nbperpage;	/* 4: number of blks per page           */
+	__le32 dn_numag;	/* 4: total number of ags               */
+	__le32 dn_maxlevel;	/* 4: number of active ags              */
+	__le32 dn_maxag;	/* 4: max active alloc group number     */
+	__le32 dn_agpref;	/* 4: preferred alloc group (hint)      */
+	__le32 dn_aglevel;	/* 4: dmapctl level holding the AG      */
+	__le32 dn_agheigth;	/* 4: height in dmapctl of the AG       */
+	__le32 dn_agwidth;	/* 4: width in dmapctl of the AG        */
+	__le32 dn_agstart;	/* 4: start tree index at AG height     */
+	__le32 dn_agl2size;	/* 4: l2 num of blks per alloc group    */
+	__le64 dn_agfree[MAXAG];/* 8*MAXAG: per AG free count           */
+	__le64 dn_agsize;	/* 8: num of blks per alloc group       */
+	s8 dn_maxfreebud;	/* 1: max free buddy system             */
+	u8 pad[3007];		/* 3007: pad to 4096                    */
+};				/* - 4096 -                             */
+
+struct dbmap {
+	s64 dn_mapsize;		/* number of blocks in aggregate     */
+	s64 dn_nfree;		/* num free blks in aggregate map    */
+	int dn_l2nbperpage;	/* number of blks per page           */
+	int dn_numag;		/* total number of ags               */
+	int dn_maxlevel;	/* number of active ags              */
+	int dn_maxag;		/* max active alloc group number     */
+	int dn_agpref;		/* preferred alloc group (hint)      */
+	int dn_aglevel;		/* dmapctl level holding the AG      */
+	int dn_agheigth;	/* height in dmapctl of the AG       */
+	int dn_agwidth;		/* width in dmapctl of the AG        */
+	int dn_agstart;		/* start tree index at AG height     */
+	int dn_agl2size;	/* l2 num of blks per alloc group    */
+	s64 dn_agfree[MAXAG];	/* per AG free count           */
+	s64 dn_agsize;		/* num of blks per alloc group       */
+	signed char dn_maxfreebud;	/* max free buddy system             */
+};				/* - 4096 -                             */
+/*
+ *	in-memory aggregate disk allocation map descriptor.
+ */
+struct bmap {
+	struct dbmap db_bmap;		/* on-disk aggregate map descriptor */
+	struct inode *db_ipbmap;	/* ptr to aggregate map incore inode */
+	struct mutex db_bmaplock;	/* aggregate map lock */
+	atomic_t db_active[MAXAG];	/* count of active, open files in AG */
+	u32 *db_DBmap;
+};
+
+/* macros for accessing fields within in-memory aggregate map descriptor */
+#define	db_mapsize	db_bmap.dn_mapsize
+#define	db_nfree	db_bmap.dn_nfree
+#define	db_agfree	db_bmap.dn_agfree
+#define	db_agsize	db_bmap.dn_agsize
+#define	db_agl2size	db_bmap.dn_agl2size
+#define	db_agwidth	db_bmap.dn_agwidth
+#define	db_agheigth	db_bmap.dn_agheigth
+#define	db_agstart	db_bmap.dn_agstart
+#define	db_numag	db_bmap.dn_numag
+#define	db_maxlevel	db_bmap.dn_maxlevel
+#define	db_aglevel	db_bmap.dn_aglevel
+#define	db_agpref	db_bmap.dn_agpref
+#define	db_maxag	db_bmap.dn_maxag
+#define	db_maxfreebud	db_bmap.dn_maxfreebud
+#define	db_l2nbperpage	db_bmap.dn_l2nbperpage
+
+/*
+ * macros for various conversions needed by the allocators.
+ * blkstol2(), cntlz(), and cnttz() are operating system dependent functions.
+ */
+/* convert number of blocks to log2 number of blocks, rounding up to
+ * the next log2 value if blocks is not a l2 multiple.
+ */
+#define	BLKSTOL2(d)		(blkstol2(d))
+
+/* convert number of leafs to log2 leaf value */
+#define	NLSTOL2BSZ(n)		(31 - cntlz((n)) + BUDMIN)
+
+/* convert leaf index to log2 leaf value */
+#define	LITOL2BSZ(n,m,b)	((((n) == 0) ? (m) : cnttz((n))) + (b))
+
+/* convert a block number to a dmap control leaf index */
+#define BLKTOCTLLEAF(b,m)	\
+	(((b) & (((s64)1 << ((m) + L2LPERCTL)) - 1)) >> (m))
+
+/* convert log2 leaf value to buddy size */
+#define	BUDSIZE(s,m)		(1 << ((s) - (m)))
+
+/*
+ *	external references.
+ */
+extern int dbMount(struct inode *ipbmap);
+
+extern int dbUnmount(struct inode *ipbmap, int mounterror);
+
+extern int dbFree(struct inode *ipbmap, s64 blkno, s64 nblocks);
+
+extern int dbUpdatePMap(struct inode *ipbmap,
+			int free, s64 blkno, s64 nblocks, struct tblock * tblk);
+
+extern int dbNextAG(struct inode *ipbmap);
+
+extern int dbAlloc(struct inode *ipbmap, s64 hint, s64 nblocks, s64 * results);
+
+extern int dbReAlloc(struct inode *ipbmap,
+		     s64 blkno, s64 nblocks, s64 addnblocks, s64 * results);
+
+extern int dbSync(struct inode *ipbmap);
+extern int dbAllocBottomUp(struct inode *ip, s64 blkno, s64 nblocks);
+extern int dbExtendFS(struct inode *ipbmap, s64 blkno, s64 nblocks);
+extern void dbFinalizeBmap(struct inode *ipbmap);
+extern s64 dbMapFileSizeToMapSize(struct inode *ipbmap);
+#endif				/* _H_JFS_DMAP */

fs/jfs/jfs_dtree.h

@@ -0,0 +1,272 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_DTREE
+#define	_H_JFS_DTREE
+
+/*
+ *	jfs_dtree.h: directory B+-tree manager
+ */
+
+#include "jfs_btree.h"
+
+typedef union {
+	struct {
+		tid_t tid;
+		struct inode *ip;
+		u32 ino;
+	} leaf;
+	pxd_t xd;
+} ddata_t;
+
+
+/*
+ *      entry segment/slot
+ *
+ * an entry consists of type dependent head/only segment/slot and
+ * additional segments/slots linked vi next field;
+ * N.B. last/only segment of entry is terminated by next = -1;
+ */
+/*
+ *	directory page slot
+ */
+struct dtslot {
+	s8 next;		/* 1: */
+	s8 cnt;			/* 1: */
+	__le16 name[15];	/* 30: */
+};				/* (32) */
+
+
+#define DATASLOTSIZE	16
+#define L2DATASLOTSIZE	4
+#define	DTSLOTSIZE	32
+#define	L2DTSLOTSIZE	5
+#define DTSLOTHDRSIZE	2
+#define DTSLOTDATASIZE	30
+#define DTSLOTDATALEN	15
+
+/*
+ *	 internal node entry head/only segment
+ */
+struct idtentry {
+	pxd_t xd;		/* 8: child extent descriptor */
+
+	s8 next;		/* 1: */
+	u8 namlen;		/* 1: */
+	__le16 name[11];	/* 22: 2-byte aligned */
+};				/* (32) */
+
+#define DTIHDRSIZE	10
+#define DTIHDRDATALEN	11
+
+/* compute number of slots for entry */
+#define	NDTINTERNAL(klen) ( ((4 + (klen)) + (15 - 1)) / 15 )
+
+
+/*
+ *	leaf node entry head/only segment
+ *
+ *	For legacy filesystems, name contains 13 wchars -- no index field
+ */
+struct ldtentry {
+	__le32 inumber;		/* 4: 4-byte aligned */
+	s8 next;		/* 1: */
+	u8 namlen;		/* 1: */
+	__le16 name[11];	/* 22: 2-byte aligned */
+	__le32 index;		/* 4: index into dir_table */
+};				/* (32) */
+
+#define DTLHDRSIZE	6
+#define DTLHDRDATALEN_LEGACY	13	/* Old (OS/2) format */
+#define DTLHDRDATALEN	11
+
+/*
+ * dir_table used for directory traversal during readdir
+ */
+
+/*
+ * Keep persistent index for directory entries
+ */
+#define DO_INDEX(INODE) (JFS_SBI((INODE)->i_sb)->mntflag & JFS_DIR_INDEX)
+
+/*
+ * Maximum entry in inline directory table
+ */
+#define MAX_INLINE_DIRTABLE_ENTRY 13
+
+struct dir_table_slot {
+	u8 rsrvd;		/* 1: */
+	u8 flag;		/* 1: 0 if free */
+	u8 slot;		/* 1: slot within leaf page of entry */
+	u8 addr1;		/* 1: upper 8 bits of leaf page address */
+	__le32 addr2;		/* 4: lower 32 bits of leaf page address -OR-
+				   index of next entry when this entry was deleted */
+};				/* (8) */
+
+/*
+ * flag values
+ */
+#define DIR_INDEX_VALID 1
+#define DIR_INDEX_FREE 0
+
+#define DTSaddress(dir_table_slot, address64)\
+{\
+	(dir_table_slot)->addr1 = ((u64)address64) >> 32;\
+	(dir_table_slot)->addr2 = __cpu_to_le32((address64) & 0xffffffff);\
+}
+
+#define addressDTS(dts)\
+	( ((s64)((dts)->addr1)) << 32 | __le32_to_cpu((dts)->addr2) )
+
+/* compute number of slots for entry */
+#define	NDTLEAF_LEGACY(klen)	( ((2 + (klen)) + (15 - 1)) / 15 )
+#define	NDTLEAF	NDTINTERNAL
+
+
+/*
+ *	directory root page (in-line in on-disk inode):
+ *
+ * cf. dtpage_t below.
+ */
+typedef union {
+	struct {
+		struct dasd DASD; /* 16: DASD limit/usage info */
+
+		u8 flag;	/* 1: */
+		u8 nextindex;	/* 1: next free entry in stbl */
+		s8 freecnt;	/* 1: free count */
+		s8 freelist;	/* 1: freelist header */
+
+		__le32 idotdot;	/* 4: parent inode number */
+
+		s8 stbl[8];	/* 8: sorted entry index table */
+	} header;		/* (32) */
+
+	struct dtslot slot[9];
+} dtroot_t;
+
+#define PARENT(IP) \
+	(le32_to_cpu(JFS_IP(IP)->i_dtroot.header.idotdot))
+
+#define DTROOTMAXSLOT	9
+
+#define	dtEmpty(IP) (JFS_IP(IP)->i_dtroot.header.nextindex == 0)
+
+
+/*
+ *	directory regular page:
+ *
+ *	entry slot array of 32 byte slot
+ *
+ * sorted entry slot index table (stbl):
+ * contiguous slots at slot specified by stblindex,
+ * 1-byte per entry
+ *   512 byte block:  16 entry tbl (1 slot)
+ *  1024 byte block:  32 entry tbl (1 slot)
+ *  2048 byte block:  64 entry tbl (2 slot)
+ *  4096 byte block: 128 entry tbl (4 slot)
+ *
+ * data area:
+ *   512 byte block:  16 - 2 =  14 slot
+ *  1024 byte block:  32 - 2 =  30 slot
+ *  2048 byte block:  64 - 3 =  61 slot
+ *  4096 byte block: 128 - 5 = 123 slot
+ *
+ * N.B. index is 0-based; index fields refer to slot index
+ * except nextindex which refers to entry index in stbl;
+ * end of entry stot list or freelist is marked with -1.
+ */
+typedef union {
+	struct {
+		__le64 next;	/* 8: next sibling */
+		__le64 prev;	/* 8: previous sibling */
+
+		u8 flag;	/* 1: */
+		u8 nextindex;	/* 1: next entry index in stbl */
+		s8 freecnt;	/* 1: */
+		s8 freelist;	/* 1: slot index of head of freelist */
+
+		u8 maxslot;	/* 1: number of slots in page slot[] */
+		u8 stblindex;	/* 1: slot index of start of stbl */
+		u8 rsrvd[2];	/* 2: */
+
+		pxd_t self;	/* 8: self pxd */
+	} header;		/* (32) */
+
+	struct dtslot slot[128];
+} dtpage_t;
+
+#define DTPAGEMAXSLOT        128
+
+#define DT8THPGNODEBYTES     512
+#define DT8THPGNODETSLOTS      1
+#define DT8THPGNODESLOTS      16
+
+#define DTQTRPGNODEBYTES    1024
+#define DTQTRPGNODETSLOTS      1
+#define DTQTRPGNODESLOTS      32
+
+#define DTHALFPGNODEBYTES   2048
+#define DTHALFPGNODETSLOTS     2
+#define DTHALFPGNODESLOTS     64
+
+#define DTFULLPGNODEBYTES   4096
+#define DTFULLPGNODETSLOTS     4
+#define DTFULLPGNODESLOTS    128
+
+#define DTENTRYSTART	1
+
+/* get sorted entry table of the page */
+#define DT_GETSTBL(p) ( ((p)->header.flag & BT_ROOT) ?\
+	((dtroot_t *)(p))->header.stbl : \
+	(s8 *)&(p)->slot[(p)->header.stblindex] )
+
+/*
+ * Flags for dtSearch
+ */
+#define JFS_CREATE 1
+#define JFS_LOOKUP 2
+#define JFS_REMOVE 3
+#define JFS_RENAME 4
+
+#define DIRENTSIZ(namlen) \
+    ( (sizeof(struct dirent) - 2*(JFS_NAME_MAX+1) + 2*((namlen)+1) + 3) &~ 3 )
+
+/*
+ * Maximum file offset for directories.
+ */
+#define DIREND	INT_MAX
+
+/*
+ *	external declarations
+ */
+extern void dtInitRoot(tid_t tid, struct inode *ip, u32 idotdot);
+
+extern int dtSearch(struct inode *ip, struct component_name * key,
+		    ino_t * data, struct btstack * btstack, int flag);
+
+extern int dtInsert(tid_t tid, struct inode *ip, struct component_name * key,
+		    ino_t * ino, struct btstack * btstack);
+
+extern int dtDelete(tid_t tid, struct inode *ip, struct component_name * key,
+		    ino_t * data, int flag);
+
+extern int dtModify(tid_t tid, struct inode *ip, struct component_name * key,
+		    ino_t * orig_ino, ino_t new_ino, int flag);
+
+extern int jfs_readdir(struct file *filp, void *dirent, filldir_t filldir);
+#endif				/* !_H_JFS_DTREE */

fs/jfs/jfs_extent.c

@@ -0,0 +1,663 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/quotaops.h>
+#include "jfs_incore.h"
+#include "jfs_inode.h"
+#include "jfs_superblock.h"
+#include "jfs_dmap.h"
+#include "jfs_extent.h"
+#include "jfs_debug.h"
+
+/*
+ * forward references
+ */
+static int extBalloc(struct inode *, s64, s64 *, s64 *);
+#ifdef _NOTYET
+static int extBrealloc(struct inode *, s64, s64, s64 *, s64 *);
+#endif
+static s64 extRoundDown(s64 nb);
+
+#define DPD(a)          (printk("(a): %d\n",(a)))
+#define DPC(a)          (printk("(a): %c\n",(a)))
+#define DPL1(a)					\
+{						\
+	if ((a) >> 32)				\
+		printk("(a): %x%08x  ",(a));	\
+	else					\
+		printk("(a): %x  ",(a) << 32);	\
+}
+#define DPL(a)					\
+{						\
+	if ((a) >> 32)				\
+		printk("(a): %x%08x\n",(a));	\
+	else					\
+		printk("(a): %x\n",(a) << 32);	\
+}
+
+#define DPD1(a)         (printk("(a): %d  ",(a)))
+#define DPX(a)          (printk("(a): %08x\n",(a)))
+#define DPX1(a)         (printk("(a): %08x  ",(a)))
+#define DPS(a)          (printk("%s\n",(a)))
+#define DPE(a)          (printk("\nENTERING: %s\n",(a)))
+#define DPE1(a)          (printk("\nENTERING: %s",(a)))
+#define DPS1(a)         (printk("  %s  ",(a)))
+
+
+/*
+ * NAME:	extAlloc()
+ *
+ * FUNCTION:    allocate an extent for a specified page range within a
+ *		file.
+ *
+ * PARAMETERS:
+ *	ip	- the inode of the file.
+ *	xlen	- requested extent length.
+ *	pno	- the starting page number with the file.
+ *	xp	- pointer to an xad.  on entry, xad describes an
+ *		  extent that is used as an allocation hint if the
+ *		  xaddr of the xad is non-zero.  on successful exit,
+ *		  the xad describes the newly allocated extent.
+ *	abnr	- bool indicating whether the newly allocated extent
+ *		  should be marked as allocated but not recorded.
+ *
+ * RETURN VALUES:
+ *      0       - success
+ *      -EIO	- i/o error.
+ *      -ENOSPC	- insufficient disk resources.
+ */
+int
+extAlloc(struct inode *ip, s64 xlen, s64 pno, xad_t * xp, bool abnr)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
+	s64 nxlen, nxaddr, xoff, hint, xaddr = 0;
+	int rc;
+	int xflag;
+
+	/* This blocks if we are low on resources */
+	txBeginAnon(ip->i_sb);
+
+	/* Avoid race with jfs_commit_inode() */
+	mutex_lock(&JFS_IP(ip)->commit_mutex);
+
+	/* validate extent length */
+	if (xlen > MAXXLEN)
+		xlen = MAXXLEN;
+
+	/* get the page's starting extent offset */
+	xoff = pno << sbi->l2nbperpage;
+
+	/* check if an allocation hint was provided */
+	if ((hint = addressXAD(xp))) {
+		/* get the size of the extent described by the hint */
+		nxlen = lengthXAD(xp);
+
+		/* check if the hint is for the portion of the file
+		 * immediately previous to the current allocation
+		 * request and if hint extent has the same abnr
+		 * value as the current request.  if so, we can
+		 * extend the hint extent to include the current
+		 * extent if we can allocate the blocks immediately
+		 * following the hint extent.
+		 */
+		if (offsetXAD(xp) + nxlen == xoff &&
+		    abnr == ((xp->flag & XAD_NOTRECORDED) ? true : false))
+			xaddr = hint + nxlen;
+
+		/* adjust the hint to the last block of the extent */
+		hint += (nxlen - 1);
+	}
+
+	/* allocate the disk blocks for the extent.  initially, extBalloc()
+	 * will try to allocate disk blocks for the requested size (xlen).
+	 * if this fails (xlen contiguous free blocks not avaliable), it'll
+	 * try to allocate a smaller number of blocks (producing a smaller
+	 * extent), with this smaller number of blocks consisting of the
+	 * requested number of blocks rounded down to the next smaller
+	 * power of 2 number (i.e. 16 -> 8).  it'll continue to round down
+	 * and retry the allocation until the number of blocks to allocate
+	 * is smaller than the number of blocks per page.
+	 */
+	nxlen = xlen;
+	if ((rc = extBalloc(ip, hint ? hint : INOHINT(ip), &nxlen, &nxaddr))) {
+		mutex_unlock(&JFS_IP(ip)->commit_mutex);
+		return (rc);
+	}
+
+	/* Allocate blocks to quota. */
+	if (DQUOT_ALLOC_BLOCK(ip, nxlen)) {
+		dbFree(ip, nxaddr, (s64) nxlen);
+		mutex_unlock(&JFS_IP(ip)->commit_mutex);
+		return -EDQUOT;
+	}
+
+	/* determine the value of the extent flag */
+	xflag = abnr ? XAD_NOTRECORDED : 0;
+
+	/* if we can extend the hint extent to cover the current request,
+	 * extend it.  otherwise, insert a new extent to
+	 * cover the current request.
+	 */
+	if (xaddr && xaddr == nxaddr)
+		rc = xtExtend(0, ip, xoff, (int) nxlen, 0);
+	else
+		rc = xtInsert(0, ip, xflag, xoff, (int) nxlen, &nxaddr, 0);
+
+	/* if the extend or insert failed,
+	 * free the newly allocated blocks and return the error.
+	 */
+	if (rc) {
+		dbFree(ip, nxaddr, nxlen);
+		DQUOT_FREE_BLOCK(ip, nxlen);
+		mutex_unlock(&JFS_IP(ip)->commit_mutex);
+		return (rc);
+	}
+
+	/* set the results of the extent allocation */
+	XADaddress(xp, nxaddr);
+	XADlength(xp, nxlen);
+	XADoffset(xp, xoff);
+	xp->flag = xflag;
+
+	mark_inode_dirty(ip);
+
+	mutex_unlock(&JFS_IP(ip)->commit_mutex);
+	/*
+	 * COMMIT_SyncList flags an anonymous tlock on page that is on
+	 * sync list.
+	 * We need to commit the inode to get the page written disk.
+	 */
+	if (test_and_clear_cflag(COMMIT_Synclist,ip))
+		jfs_commit_inode(ip, 0);
+
+	return (0);
+}
+
+
+#ifdef _NOTYET
+/*
+ * NAME:        extRealloc()
+ *
+ * FUNCTION:    extend the allocation of a file extent containing a
+ *		partial back last page.
+ *
+ * PARAMETERS:
+ *	ip	- the inode of the file.
+ *	cp	- cbuf for the partial backed last page.
+ *	xlen	- request size of the resulting extent.
+ *	xp	- pointer to an xad. on successful exit, the xad
+ *		  describes the newly allocated extent.
+ *	abnr	- bool indicating whether the newly allocated extent
+ *		  should be marked as allocated but not recorded.
+ *
+ * RETURN VALUES:
+ *      0       - success
+ *      -EIO	- i/o error.
+ *      -ENOSPC	- insufficient disk resources.
+ */
+int extRealloc(struct inode *ip, s64 nxlen, xad_t * xp, bool abnr)
+{
+	struct super_block *sb = ip->i_sb;
+	s64 xaddr, xlen, nxaddr, delta, xoff;
+	s64 ntail, nextend, ninsert;
+	int rc, nbperpage = JFS_SBI(sb)->nbperpage;
+	int xflag;
+
+	/* This blocks if we are low on resources */
+	txBeginAnon(ip->i_sb);
+
+	mutex_lock(&JFS_IP(ip)->commit_mutex);
+	/* validate extent length */
+	if (nxlen > MAXXLEN)
+		nxlen = MAXXLEN;
+
+	/* get the extend (partial) page's disk block address and
+	 * number of blocks.
+	 */
+	xaddr = addressXAD(xp);
+	xlen = lengthXAD(xp);
+	xoff = offsetXAD(xp);
+
+	/* if the extend page is abnr and if the request is for
+	 * the extent to be allocated and recorded,
+	 * make the page allocated and recorded.
+	 */
+	if ((xp->flag & XAD_NOTRECORDED) && !abnr) {
+		xp->flag = 0;
+		if ((rc = xtUpdate(0, ip, xp)))
+			goto exit;
+	}
+
+	/* try to allocated the request number of blocks for the
+	 * extent.  dbRealloc() first tries to satisfy the request
+	 * by extending the allocation in place. otherwise, it will
+	 * try to allocate a new set of blocks large enough for the
+	 * request.  in satisfying a request, dbReAlloc() may allocate
+	 * less than what was request but will always allocate enough
+	 * space as to satisfy the extend page.
+	 */
+	if ((rc = extBrealloc(ip, xaddr, xlen, &nxlen, &nxaddr)))
+		goto exit;
+
+	/* Allocat blocks to quota. */
+	if (DQUOT_ALLOC_BLOCK(ip, nxlen)) {
+		dbFree(ip, nxaddr, (s64) nxlen);
+		mutex_unlock(&JFS_IP(ip)->commit_mutex);
+		return -EDQUOT;
+	}
+
+	delta = nxlen - xlen;
+
+	/* check if the extend page is not abnr but the request is abnr
+	 * and the allocated disk space is for more than one page.  if this
+	 * is the case, there is a miss match of abnr between the extend page
+	 * and the one or more pages following the extend page.  as a result,
+	 * two extents will have to be manipulated. the first will be that
+	 * of the extent of the extend page and will be manipulated thru
+	 * an xtExtend() or an xtTailgate(), depending upon whether the
+	 * disk allocation occurred as an inplace extension.  the second
+	 * extent will be manipulated (created) through an xtInsert() and
+	 * will be for the pages following the extend page.
+	 */
+	if (abnr && (!(xp->flag & XAD_NOTRECORDED)) && (nxlen > nbperpage)) {
+		ntail = nbperpage;
+		nextend = ntail - xlen;
+		ninsert = nxlen - nbperpage;
+
+		xflag = XAD_NOTRECORDED;
+	} else {
+		ntail = nxlen;
+		nextend = delta;
+		ninsert = 0;
+
+		xflag = xp->flag;
+	}
+
+	/* if we were able to extend the disk allocation in place,
+	 * extend the extent.  otherwise, move the extent to a
+	 * new disk location.
+	 */
+	if (xaddr == nxaddr) {
+		/* extend the extent */
+		if ((rc = xtExtend(0, ip, xoff + xlen, (int) nextend, 0))) {
+			dbFree(ip, xaddr + xlen, delta);
+			DQUOT_FREE_BLOCK(ip, nxlen);
+			goto exit;
+		}
+	} else {
+		/*
+		 * move the extent to a new location:
+		 *
+		 * xtTailgate() accounts for relocated tail extent;
+		 */
+		if ((rc = xtTailgate(0, ip, xoff, (int) ntail, nxaddr, 0))) {
+			dbFree(ip, nxaddr, nxlen);
+			DQUOT_FREE_BLOCK(ip, nxlen);
+			goto exit;
+		}
+	}
+
+
+	/* check if we need to also insert a new extent */
+	if (ninsert) {
+		/* perform the insert.  if it fails, free the blocks
+		 * to be inserted and make it appear that we only did
+		 * the xtExtend() or xtTailgate() above.
+		 */
+		xaddr = nxaddr + ntail;
+		if (xtInsert (0, ip, xflag, xoff + ntail, (int) ninsert,
+			      &xaddr, 0)) {
+			dbFree(ip, xaddr, (s64) ninsert);
+			delta = nextend;
+			nxlen = ntail;
+			xflag = 0;
+		}
+	}
+
+	/* set the return results */
+	XADaddress(xp, nxaddr);
+	XADlength(xp, nxlen);
+	XADoffset(xp, xoff);
+	xp->flag = xflag;
+
+	mark_inode_dirty(ip);
+exit:
+	mutex_unlock(&JFS_IP(ip)->commit_mutex);
+	return (rc);
+}
+#endif			/* _NOTYET */
+
+
+/*
+ * NAME:        extHint()
+ *
+ * FUNCTION:    produce an extent allocation hint for a file offset.
+ *
+ * PARAMETERS:
+ *	ip	- the inode of the file.
+ *	offset  - file offset for which the hint is needed.
+ *	xp	- pointer to the xad that is to be filled in with
+ *		  the hint.
+ *
+ * RETURN VALUES:
+ *      0       - success
+ *      -EIO	- i/o error.
+ */
+int extHint(struct inode *ip, s64 offset, xad_t * xp)
+{
+	struct super_block *sb = ip->i_sb;
+	struct xadlist xadl;
+	struct lxdlist lxdl;
+	lxd_t lxd;
+	s64 prev;
+	int rc, nbperpage = JFS_SBI(sb)->nbperpage;
+
+	/* init the hint as "no hint provided" */
+	XADaddress(xp, 0);
+
+	/* determine the starting extent offset of the page previous
+	 * to the page containing the offset.
+	 */
+	prev = ((offset & ~POFFSET) >> JFS_SBI(sb)->l2bsize) - nbperpage;
+
+	/* if the offsets in the first page of the file,
+	 * no hint provided.
+	 */
+	if (prev < 0)
+		return (0);
+
+	/* prepare to lookup the previous page's extent info */
+	lxdl.maxnlxd = 1;
+	lxdl.nlxd = 1;
+	lxdl.lxd = &lxd;
+	LXDoffset(&lxd, prev)
+	    LXDlength(&lxd, nbperpage);
+
+	xadl.maxnxad = 1;
+	xadl.nxad = 0;
+	xadl.xad = xp;
+
+	/* perform the lookup */
+	if ((rc = xtLookupList(ip, &lxdl, &xadl, 0)))
+		return (rc);
+
+	/* check if not extent exists for the previous page.
+	 * this is possible for sparse files.
+	 */
+	if (xadl.nxad == 0) {
+//              assert(ISSPARSE(ip));
+		return (0);
+	}
+
+	/* only preserve the abnr flag within the xad flags
+	 * of the returned hint.
+	 */
+	xp->flag &= XAD_NOTRECORDED;
+
+        if(xadl.nxad != 1 || lengthXAD(xp) != nbperpage) {
+		jfs_error(ip->i_sb, "extHint: corrupt xtree");
+		return -EIO;
+        }
+
+	return (0);
+}
+
+
+/*
+ * NAME:        extRecord()
+ *
+ * FUNCTION:    change a page with a file from not recorded to recorded.
+ *
+ * PARAMETERS:
+ *	ip	- inode of the file.
+ *	cp	- cbuf of the file page.
+ *
+ * RETURN VALUES:
+ *      0       - success
+ *      -EIO	- i/o error.
+ *      -ENOSPC	- insufficient disk resources.
+ */
+int extRecord(struct inode *ip, xad_t * xp)
+{
+	int rc;
+
+	txBeginAnon(ip->i_sb);
+
+	mutex_lock(&JFS_IP(ip)->commit_mutex);
+
+	/* update the extent */
+	rc = xtUpdate(0, ip, xp);
+
+	mutex_unlock(&JFS_IP(ip)->commit_mutex);
+	return rc;
+}
+
+
+#ifdef _NOTYET
+/*
+ * NAME:        extFill()
+ *
+ * FUNCTION:    allocate disk space for a file page that represents
+ *		a file hole.
+ *
+ * PARAMETERS:
+ *	ip	- the inode of the file.
+ *	cp	- cbuf of the file page represent the hole.
+ *
+ * RETURN VALUES:
+ *      0       - success
+ *      -EIO	- i/o error.
+ *      -ENOSPC	- insufficient disk resources.
+ */
+int extFill(struct inode *ip, xad_t * xp)
+{
+	int rc, nbperpage = JFS_SBI(ip->i_sb)->nbperpage;
+	s64 blkno = offsetXAD(xp) >> ip->i_blkbits;
+
+//      assert(ISSPARSE(ip));
+
+	/* initialize the extent allocation hint */
+	XADaddress(xp, 0);
+
+	/* allocate an extent to fill the hole */
+	if ((rc = extAlloc(ip, nbperpage, blkno, xp, false)))
+		return (rc);
+
+	assert(lengthPXD(xp) == nbperpage);
+
+	return (0);
+}
+#endif			/* _NOTYET */
+
+
+/*
+ * NAME:	extBalloc()
+ *
+ * FUNCTION:    allocate disk blocks to form an extent.
+ *
+ *		initially, we will try to allocate disk blocks for the
+ *		requested size (nblocks).  if this fails (nblocks
+ *		contiguous free blocks not avaliable), we'll try to allocate
+ *		a smaller number of blocks (producing a smaller extent), with
+ *		this smaller number of blocks consisting of the requested
+ *		number of blocks rounded down to the next smaller power of 2
+ *		number (i.e. 16 -> 8).  we'll continue to round down and
+ *		retry the allocation until the number of blocks to allocate
+ *		is smaller than the number of blocks per page.
+ *
+ * PARAMETERS:
+ *	ip	 - the inode of the file.
+ *	hint	 - disk block number to be used as an allocation hint.
+ *	*nblocks - pointer to an s64 value.  on entry, this value specifies
+ *		   the desired number of block to be allocated. on successful
+ *		   exit, this value is set to the number of blocks actually
+ *		   allocated.
+ *	blkno	 - pointer to a block address that is filled in on successful
+ *		   return with the starting block number of the newly
+ *		   allocated block range.
+ *
+ * RETURN VALUES:
+ *      0       - success
+ *      -EIO	- i/o error.
+ *      -ENOSPC	- insufficient disk resources.
+ */
+static int
+extBalloc(struct inode *ip, s64 hint, s64 * nblocks, s64 * blkno)
+{
+	struct jfs_inode_info *ji = JFS_IP(ip);
+	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
+	s64 nb, nblks, daddr, max;
+	int rc, nbperpage = sbi->nbperpage;
+	struct bmap *bmp = sbi->bmap;
+	int ag;
+
+	/* get the number of blocks to initially attempt to allocate.
+	 * we'll first try the number of blocks requested unless this
+	 * number is greater than the maximum number of contiguous free
+	 * blocks in the map. in that case, we'll start off with the
+	 * maximum free.
+	 */
+	max = (s64) 1 << bmp->db_maxfreebud;
+	if (*nblocks >= max && *nblocks > nbperpage)
+		nb = nblks = (max > nbperpage) ? max : nbperpage;
+	else
+		nb = nblks = *nblocks;
+
+	/* try to allocate blocks */
+	while ((rc = dbAlloc(ip, hint, nb, &daddr)) != 0) {
+		/* if something other than an out of space error,
+		 * stop and return this error.
+		 */
+		if (rc != -ENOSPC)
+			return (rc);
+
+		/* decrease the allocation request size */
+		nb = min(nblks, extRoundDown(nb));
+
+		/* give up if we cannot cover a page */
+		if (nb < nbperpage)
+			return (rc);
+	}
+
+	*nblocks = nb;
+	*blkno = daddr;
+
+	if (S_ISREG(ip->i_mode) && (ji->fileset == FILESYSTEM_I)) {
+		ag = BLKTOAG(daddr, sbi);
+		spin_lock_irq(&ji->ag_lock);
+		if (ji->active_ag == -1) {
+			atomic_inc(&bmp->db_active[ag]);
+			ji->active_ag = ag;
+		} else if (ji->active_ag != ag) {
+			atomic_dec(&bmp->db_active[ji->active_ag]);
+			atomic_inc(&bmp->db_active[ag]);
+			ji->active_ag = ag;
+		}
+		spin_unlock_irq(&ji->ag_lock);
+	}
+
+	return (0);
+}
+
+
+#ifdef _NOTYET
+/*
+ * NAME:	extBrealloc()
+ *
+ * FUNCTION:    attempt to extend an extent's allocation.
+ *
+ *		Initially, we will try to extend the extent's allocation
+ *		in place.  If this fails, we'll try to move the extent
+ *		to a new set of blocks.  If moving the extent, we initially
+ *		will try to allocate disk blocks for the requested size
+ *		(newnblks).  if this fails (new contiguous free blocks not
+ *		avaliable), we'll try to allocate a smaller number of
+ *		blocks (producing a smaller extent), with this smaller
+ *		number of blocks consisting of the requested number of
+ *		blocks rounded down to the next smaller power of 2
+ *		number (i.e. 16 -> 8).  We'll continue to round down and
+ *		retry the allocation until the number of blocks to allocate
+ *		is smaller than the number of blocks per page.
+ *
+ * PARAMETERS:
+ *	ip	 - the inode of the file.
+ *	blkno    - starting block number of the extents current allocation.
+ *	nblks    - number of blocks within the extents current allocation.
+ *	newnblks - pointer to a s64 value.  on entry, this value is the
+ *		   the new desired extent size (number of blocks).  on
+ *		   successful exit, this value is set to the extent's actual
+ *		   new size (new number of blocks).
+ *	newblkno - the starting block number of the extents new allocation.
+ *
+ * RETURN VALUES:
+ *      0       - success
+ *      -EIO	- i/o error.
+ *      -ENOSPC	- insufficient disk resources.
+ */
+static int
+extBrealloc(struct inode *ip,
+	    s64 blkno, s64 nblks, s64 * newnblks, s64 * newblkno)
+{
+	int rc;
+
+	/* try to extend in place */
+	if ((rc = dbExtend(ip, blkno, nblks, *newnblks - nblks)) == 0) {
+		*newblkno = blkno;
+		return (0);
+	} else {
+		if (rc != -ENOSPC)
+			return (rc);
+	}
+
+	/* in place extension not possible.
+	 * try to move the extent to a new set of blocks.
+	 */
+	return (extBalloc(ip, blkno, newnblks, newblkno));
+}
+#endif			/* _NOTYET */
+
+
+/*
+ * NAME:        extRoundDown()
+ *
+ * FUNCTION:    round down a specified number of blocks to the next
+ *		smallest power of 2 number.
+ *
+ * PARAMETERS:
+ *	nb	- the inode of the file.
+ *
+ * RETURN VALUES:
+ *      next smallest power of 2 number.
+ */
+static s64 extRoundDown(s64 nb)
+{
+	int i;
+	u64 m, k;
+
+	for (i = 0, m = (u64) 1 << 63; i < 64; i++, m >>= 1) {
+		if (m & nb)
+			break;
+	}
+
+	i = 63 - i;
+	k = (u64) 1 << i;
+	k = ((k - 1) & nb) ? k : k >> 1;
+
+	return (k);
+}

fs/jfs/jfs_extent.h

@@ -0,0 +1,31 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2001
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_EXTENT
+#define _H_JFS_EXTENT
+
+/*  get block allocation allocation hint as location of disk inode */
+#define	INOHINT(ip)	\
+	(addressPXD(&(JFS_IP(ip)->ixpxd)) + lengthPXD(&(JFS_IP(ip)->ixpxd)) - 1)
+
+extern int	extAlloc(struct inode *, s64, s64, xad_t *, bool);
+extern int	extFill(struct inode *, xad_t *);
+extern int	extHint(struct inode *, s64, xad_t *);
+extern int	extRealloc(struct inode *, s64, xad_t *, bool);
+extern int	extRecord(struct inode *, xad_t *);
+
+#endif	/* _H_JFS_EXTENT */

fs/jfs/jfs_filsys.h

@@ -0,0 +1,283 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2003
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_FILSYS
+#define _H_JFS_FILSYS
+
+/*
+ *	jfs_filsys.h
+ *
+ * file system (implementation-dependent) constants
+ *
+ * refer to <limits.h> for system wide implementation-dependent constants
+ */
+
+/*
+ *	 file system option (superblock flag)
+ */
+
+/* directory option */
+#define JFS_UNICODE	0x00000001	/* unicode name */
+
+/* mount time flags for error handling */
+#define JFS_ERR_REMOUNT_RO 0x00000002   /* remount read-only */
+#define JFS_ERR_CONTINUE   0x00000004   /* continue */
+#define JFS_ERR_PANIC      0x00000008   /* panic */
+
+/* Quota support */
+#define	JFS_USRQUOTA	0x00000010
+#define	JFS_GRPQUOTA	0x00000020
+
+/* mount time flag to disable journaling to disk */
+#define JFS_NOINTEGRITY 0x00000040
+
+/* commit option */
+#define	JFS_COMMIT	0x00000f00	/* commit option mask */
+#define	JFS_GROUPCOMMIT	0x00000100	/* group (of 1) commit */
+#define	JFS_LAZYCOMMIT	0x00000200	/* lazy commit */
+#define	JFS_TMPFS	0x00000400	/* temporary file system -
+					 * do not log/commit:
+					 * Never implemented
+					 */
+
+/* log logical volume option */
+#define	JFS_INLINELOG	0x00000800	/* inline log within file system */
+#define JFS_INLINEMOVE	0x00001000	/* inline log being moved */
+
+/* Secondary aggregate inode table */
+#define JFS_BAD_SAIT	0x00010000	/* current secondary ait is bad */
+
+/* sparse regular file support */
+#define JFS_SPARSE	0x00020000	/* sparse regular file */
+
+/* DASD Limits		F226941 */
+#define JFS_DASD_ENABLED 0x00040000	/* DASD limits enabled */
+#define	JFS_DASD_PRIME	0x00080000	/* Prime DASD usage on boot */
+
+/* big endian flag */
+#define	JFS_SWAP_BYTES	0x00100000	/* running on big endian computer */
+
+/* Directory index */
+#define JFS_DIR_INDEX	0x00200000	/* Persistent index for */
+
+/* platform options */
+#define JFS_LINUX	0x10000000	/* Linux support */
+#define JFS_DFS		0x20000000	/* DCE DFS LFS support */
+/*	Never implemented */
+
+#define JFS_OS2		0x40000000	/* OS/2 support */
+/*	case-insensitive name/directory support */
+
+#define JFS_AIX		0x80000000	/* AIX support */
+/*	POSIX name/directory  support - Never implemented*/
+
+/*
+ *	buffer cache configuration
+ */
+/* page size */
+#ifdef PSIZE
+#undef PSIZE
+#endif
+#define	PSIZE		4096	/* page size (in byte) */
+#define	L2PSIZE		12	/* log2(PSIZE) */
+#define	POFFSET		4095	/* offset within page */
+
+/* buffer page size */
+#define BPSIZE	PSIZE
+
+/*
+ *	fs fundamental size
+ *
+ * PSIZE >= file system block size >= PBSIZE >= DISIZE
+ */
+#define	PBSIZE		512	/* physical block size (in byte) */
+#define	L2PBSIZE	9	/* log2(PBSIZE) */
+
+#define DISIZE		512	/* on-disk inode size (in byte) */
+#define L2DISIZE	9	/* log2(DISIZE) */
+
+#define IDATASIZE	256	/* inode inline data size */
+#define	IXATTRSIZE	128	/* inode inline extended attribute size */
+
+#define XTPAGE_SIZE     4096
+#define log2_PAGESIZE     12
+
+#define IAG_SIZE        4096
+#define IAG_EXTENT_SIZE 4096
+#define	INOSPERIAG	4096	/* number of disk inodes per iag */
+#define	L2INOSPERIAG	12	/* l2 number of disk inodes per iag */
+#define INOSPEREXT	32	/* number of disk inode per extent */
+#define L2INOSPEREXT	5	/* l2 number of disk inode per extent */
+#define	IXSIZE		(DISIZE * INOSPEREXT)	/* inode extent size */
+#define	INOSPERPAGE	8	/* number of disk inodes per 4K page */
+#define	L2INOSPERPAGE	3	/* log2(INOSPERPAGE) */
+
+#define	IAGFREELIST_LWM	64
+
+#define INODE_EXTENT_SIZE	IXSIZE	/* inode extent size */
+#define NUM_INODE_PER_EXTENT	INOSPEREXT
+#define NUM_INODE_PER_IAG	INOSPERIAG
+
+#define MINBLOCKSIZE		512
+#define MAXBLOCKSIZE		4096
+#define	MAXFILESIZE		((s64)1 << 52)
+
+#define JFS_LINK_MAX		0xffffffff
+
+/* Minimum number of bytes supported for a JFS partition */
+#define MINJFS			(0x1000000)
+#define MINJFSTEXT		"16"
+
+/*
+ * file system block size -> physical block size
+ */
+#define LBOFFSET(x)	((x) & (PBSIZE - 1))
+#define LBNUMBER(x)	((x) >> L2PBSIZE)
+#define	LBLK2PBLK(sb,b)	((b) << (sb->s_blocksize_bits - L2PBSIZE))
+#define	PBLK2LBLK(sb,b)	((b) >> (sb->s_blocksize_bits - L2PBSIZE))
+/* size in byte -> last page number */
+#define	SIZE2PN(size)	( ((s64)((size) - 1)) >> (L2PSIZE) )
+/* size in byte -> last file system block number */
+#define	SIZE2BN(size, l2bsize) ( ((s64)((size) - 1)) >> (l2bsize) )
+
+/*
+ * fixed physical block address (physical block size = 512 byte)
+ *
+ * NOTE: since we can't guarantee a physical block size of 512 bytes the use of
+ *	 these macros should be removed and the byte offset macros used instead.
+ */
+#define SUPER1_B	64	/* primary superblock */
+#define	AIMAP_B		(SUPER1_B + 8)	/* 1st extent of aggregate inode map */
+#define	AITBL_B		(AIMAP_B + 16)	/*
+					 * 1st extent of aggregate inode table
+					 */
+#define	SUPER2_B	(AITBL_B + 32)	/* 2ndary superblock pbn */
+#define	BMAP_B		(SUPER2_B + 8)	/* block allocation map */
+
+/*
+ * SIZE_OF_SUPER defines the total amount of space reserved on disk for the
+ * superblock.  This is not the same as the superblock structure, since all of
+ * this space is not currently being used.
+ */
+#define SIZE_OF_SUPER	PSIZE
+
+/*
+ * SIZE_OF_AG_TABLE defines the amount of space reserved to hold the AG table
+ */
+#define SIZE_OF_AG_TABLE	PSIZE
+
+/*
+ * SIZE_OF_MAP_PAGE defines the amount of disk space reserved for each page of
+ * the inode allocation map (to hold iag)
+ */
+#define SIZE_OF_MAP_PAGE	PSIZE
+
+/*
+ * fixed byte offset address
+ */
+#define SUPER1_OFF	0x8000	/* primary superblock */
+#define AIMAP_OFF	(SUPER1_OFF + SIZE_OF_SUPER)
+					/*
+					 * Control page of aggregate inode map
+					 * followed by 1st extent of map
+					 */
+#define AITBL_OFF	(AIMAP_OFF + (SIZE_OF_MAP_PAGE << 1))
+					/*
+					 * 1st extent of aggregate inode table
+					 */
+#define SUPER2_OFF	(AITBL_OFF + INODE_EXTENT_SIZE)
+					/*
+					 * secondary superblock
+					 */
+#define BMAP_OFF	(SUPER2_OFF + SIZE_OF_SUPER)
+					/*
+					 * block allocation map
+					 */
+
+/*
+ * The following macro is used to indicate the number of reserved disk blocks at
+ * the front of an aggregate, in terms of physical blocks.  This value is
+ * currently defined to be 32K.  This turns out to be the same as the primary
+ * superblock's address, since it directly follows the reserved blocks.
+ */
+#define AGGR_RSVD_BLOCKS	SUPER1_B
+
+/*
+ * The following macro is used to indicate the number of reserved bytes at the
+ * front of an aggregate.  This value is currently defined to be 32K.  This
+ * turns out to be the same as the primary superblock's byte offset, since it
+ * directly follows the reserved blocks.
+ */
+#define AGGR_RSVD_BYTES	SUPER1_OFF
+
+/*
+ * The following macro defines the byte offset for the first inode extent in
+ * the aggregate inode table.  This allows us to find the self inode to find the
+ * rest of the table.  Currently this value is 44K.
+ */
+#define AGGR_INODE_TABLE_START	AITBL_OFF
+
+/*
+ *	fixed reserved inode number
+ */
+/* aggregate inode */
+#define AGGR_RESERVED_I	0	/* aggregate inode (reserved) */
+#define	AGGREGATE_I	1	/* aggregate inode map inode */
+#define	BMAP_I		2	/* aggregate block allocation map inode */
+#define	LOG_I		3	/* aggregate inline log inode */
+#define BADBLOCK_I	4	/* aggregate bad block inode */
+#define	FILESYSTEM_I	16	/* 1st/only fileset inode in ait:
+				 * fileset inode map inode
+				 */
+
+/* per fileset inode */
+#define FILESET_RSVD_I	0	/* fileset inode (reserved) */
+#define FILESET_EXT_I	1	/* fileset inode extension */
+#define	ROOT_I		2	/* fileset root inode */
+#define ACL_I		3	/* fileset ACL inode */
+
+#define FILESET_OBJECT_I 4	/* the first fileset inode available for a file
+				 * or directory or link...
+				 */
+#define FIRST_FILESET_INO 16	/* the first aggregate inode which describes
+				 * an inode.  (To fsck this is also the first
+				 * inode in part 2 of the agg inode table.)
+				 */
+
+/*
+ *	directory configuration
+ */
+#define JFS_NAME_MAX	255
+#define JFS_PATH_MAX	BPSIZE
+
+
+/*
+ *	file system state (superblock state)
+ */
+#define FM_CLEAN 0x00000000	/* file system is unmounted and clean */
+#define FM_MOUNT 0x00000001	/* file system is mounted cleanly */
+#define FM_DIRTY 0x00000002	/* file system was not unmounted and clean
+				 * when mounted or
+				 * commit failure occurred while being mounted:
+				 * fsck() must be run to repair
+				 */
+#define	FM_LOGREDO 0x00000004	/* log based recovery (logredo()) failed:
+				 * fsck() must be run to repair
+				 */
+#define	FM_EXTENDFS 0x00000008	/* file system extendfs() in progress */
+
+#endif				/* _H_JFS_FILSYS */

fs/jfs/jfs_imap.h

@@ -0,0 +1,175 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_IMAP
+#define _H_JFS_IMAP
+
+#include "jfs_txnmgr.h"
+
+/*
+ *	jfs_imap.h: disk inode manager
+ */
+
+#define	EXTSPERIAG	128	/* number of disk inode extent per iag  */
+#define IMAPBLKNO	0	/* lblkno of dinomap within inode map   */
+#define SMAPSZ		4	/* number of words per summary map      */
+#define	EXTSPERSUM	32	/* number of extents per summary map entry */
+#define	L2EXTSPERSUM	5	/* l2 number of extents per summary map */
+#define	PGSPERIEXT	4	/* number of 4K pages per dinode extent */
+#define	MAXIAGS		((1<<20)-1)	/* maximum number of iags       */
+#define	MAXAG		128	/* maximum number of allocation groups  */
+
+#define AMAPSIZE      512	/* bytes in the IAG allocation maps */
+#define SMAPSIZE      16	/* bytes in the IAG summary maps */
+
+/* convert inode number to iag number */
+#define	INOTOIAG(ino)	((ino) >> L2INOSPERIAG)
+
+/* convert iag number to logical block number of the iag page */
+#define IAGTOLBLK(iagno,l2nbperpg)	(((iagno) + 1) << (l2nbperpg))
+
+/* get the starting block number of the 4K page of an inode extent
+ * that contains ino.
+ */
+#define INOPBLK(pxd,ino,l2nbperpg)	(addressPXD((pxd)) +		\
+	((((ino) & (INOSPEREXT-1)) >> L2INOSPERPAGE) << (l2nbperpg)))
+
+/*
+ *	inode allocation map:
+ *
+ * inode allocation map consists of
+ * . the inode map control page and
+ * . inode allocation group pages (per 4096 inodes)
+ * which are addressed by standard JFS xtree.
+ */
+/*
+ *	inode allocation group page (per 4096 inodes of an AG)
+ */
+struct iag {
+	__le64 agstart;		/* 8: starting block of ag              */
+	__le32 iagnum;		/* 4: inode allocation group number     */
+	__le32 inofreefwd;	/* 4: ag inode free list forward        */
+	__le32 inofreeback;	/* 4: ag inode free list back           */
+	__le32 extfreefwd;	/* 4: ag inode extent free list forward */
+	__le32 extfreeback;	/* 4: ag inode extent free list back    */
+	__le32 iagfree;		/* 4: iag free list                     */
+
+	/* summary map: 1 bit per inode extent */
+	__le32 inosmap[SMAPSZ];	/* 16: sum map of mapwords w/ free inodes;
+				 *      note: this indicates free and backed
+				 *      inodes, if the extent is not backed the
+				 *      value will be 1.  if the extent is
+				 *      backed but all inodes are being used the
+				 *      value will be 1.  if the extent is
+				 *      backed but at least one of the inodes is
+				 *      free the value will be 0.
+				 */
+	__le32 extsmap[SMAPSZ];	/* 16: sum map of mapwords w/ free extents */
+	__le32 nfreeinos;		/* 4: number of free inodes             */
+	__le32 nfreeexts;		/* 4: number of free extents            */
+	/* (72) */
+	u8 pad[1976];		/* 1976: pad to 2048 bytes */
+	/* allocation bit map: 1 bit per inode (0 - free, 1 - allocated) */
+	__le32 wmap[EXTSPERIAG];	/* 512: working allocation map  */
+	__le32 pmap[EXTSPERIAG];	/* 512: persistent allocation map */
+	pxd_t inoext[EXTSPERIAG];	/* 1024: inode extent addresses */
+};				/* (4096) */
+
+/*
+ *	per AG control information (in inode map control page)
+ */
+struct iagctl_disk {
+	__le32 inofree;		/* 4: free inode list anchor            */
+	__le32 extfree;		/* 4: free extent list anchor           */
+	__le32 numinos;		/* 4: number of backed inodes           */
+	__le32 numfree;		/* 4: number of free inodes             */
+};				/* (16) */
+
+struct iagctl {
+	int inofree;		/* free inode list anchor            */
+	int extfree;		/* free extent list anchor           */
+	int numinos;		/* number of backed inodes           */
+	int numfree;		/* number of free inodes             */
+};
+
+/*
+ *	per fileset/aggregate inode map control page
+ */
+struct dinomap_disk {
+	__le32 in_freeiag;	/* 4: free iag list anchor     */
+	__le32 in_nextiag;	/* 4: next free iag number     */
+	__le32 in_numinos;	/* 4: num of backed inodes */
+	__le32 in_numfree;	/* 4: num of free backed inodes */
+	__le32 in_nbperiext;	/* 4: num of blocks per inode extent */
+	__le32 in_l2nbperiext;	/* 4: l2 of in_nbperiext */
+	__le32 in_diskblock;	/* 4: for standalone test driver  */
+	__le32 in_maxag;	/* 4: for standalone test driver  */
+	u8 pad[2016];		/* 2016: pad to 2048 */
+	struct iagctl_disk in_agctl[MAXAG]; /* 2048: AG control information */
+};				/* (4096) */
+
+struct dinomap {
+	int in_freeiag;		/* free iag list anchor     */
+	int in_nextiag;		/* next free iag number     */
+	int in_numinos;		/* num of backed inodes */
+	int in_numfree;		/* num of free backed inodes */
+	int in_nbperiext;	/* num of blocks per inode extent */
+	int in_l2nbperiext;	/* l2 of in_nbperiext */
+	int in_diskblock;	/* for standalone test driver  */
+	int in_maxag;		/* for standalone test driver  */
+	struct iagctl in_agctl[MAXAG];	/* AG control information */
+};
+
+/*
+ *	In-core inode map control page
+ */
+struct inomap {
+	struct dinomap im_imap;		/* 4096: inode allocation control */
+	struct inode *im_ipimap;	/* 4: ptr to inode for imap   */
+	struct mutex im_freelock;	/* 4: iag free list lock      */
+	struct mutex im_aglock[MAXAG];	/* 512: per AG locks          */
+	u32 *im_DBGdimap;
+	atomic_t im_numinos;	/* num of backed inodes */
+	atomic_t im_numfree;	/* num of free backed inodes */
+};
+
+#define	im_freeiag	im_imap.in_freeiag
+#define	im_nextiag	im_imap.in_nextiag
+#define	im_agctl	im_imap.in_agctl
+#define	im_nbperiext	im_imap.in_nbperiext
+#define	im_l2nbperiext	im_imap.in_l2nbperiext
+
+/* for standalone testdriver
+ */
+#define	im_diskblock	im_imap.in_diskblock
+#define	im_maxag	im_imap.in_maxag
+
+extern int diFree(struct inode *);
+extern int diAlloc(struct inode *, bool, struct inode *);
+extern int diSync(struct inode *);
+/* external references */
+extern int diUpdatePMap(struct inode *ipimap, unsigned long inum,
+			bool is_free, struct tblock * tblk);
+extern int diExtendFS(struct inode *ipimap, struct inode *ipbmap);
+extern int diMount(struct inode *);
+extern int diUnmount(struct inode *, int);
+extern int diRead(struct inode *);
+extern struct inode *diReadSpecial(struct super_block *, ino_t, int);
+extern void diWriteSpecial(struct inode *, int);
+extern void diFreeSpecial(struct inode *);
+extern int diWrite(tid_t tid, struct inode *);
+#endif				/* _H_JFS_IMAP */

fs/jfs/jfs_incore.h

@@ -0,0 +1,229 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_INCORE
+#define _H_JFS_INCORE
+
+#include <linux/mutex.h>
+#include <linux/rwsem.h>
+#include <linux/slab.h>
+#include <linux/bitops.h>
+#include "jfs_types.h"
+#include "jfs_xtree.h"
+#include "jfs_dtree.h"
+
+/*
+ * JFS magic number
+ */
+#define JFS_SUPER_MAGIC 0x3153464a /* "JFS1" */
+
+/*
+ * JFS-private inode information
+ */
+struct jfs_inode_info {
+	int	fileset;	/* fileset number (always 16)*/
+	uint	mode2;		/* jfs-specific mode		*/
+	uint	saved_uid;	/* saved for uid mount option */
+	uint	saved_gid;	/* saved for gid mount option */
+	pxd_t   ixpxd;		/* inode extent descriptor	*/
+	dxd_t	acl;		/* dxd describing acl	*/
+	dxd_t	ea;		/* dxd describing ea	*/
+	time_t	otime;		/* time created	*/
+	uint	next_index;	/* next available directory entry index */
+	int	acltype;	/* Type of ACL	*/
+	short	btorder;	/* access order	*/
+	short	btindex;	/* btpage entry index*/
+	struct inode *ipimap;	/* inode map			*/
+	long	cflag;		/* commit flags		*/
+	u16	bxflag;		/* xflag of pseudo buffer?	*/
+	unchar	agno;		/* ag number			*/
+	signed char active_ag;	/* ag currently allocating from	*/
+	lid_t	blid;		/* lid of pseudo buffer?	*/
+	lid_t	atlhead;	/* anonymous tlock list head	*/
+	lid_t	atltail;	/* anonymous tlock list tail	*/
+	spinlock_t ag_lock;	/* protects active_ag		*/
+	struct list_head anon_inode_list; /* inodes having anonymous txns */
+	/*
+	 * rdwrlock serializes xtree between reads & writes and synchronizes
+	 * changes to special inodes.  It's use would be redundant on
+	 * directories since the i_mutex taken in the VFS is sufficient.
+	 */
+	struct rw_semaphore rdwrlock;
+	/*
+	 * commit_mutex serializes transaction processing on an inode.
+	 * It must be taken after beginning a transaction (txBegin), since
+	 * dirty inodes may be committed while a new transaction on the
+	 * inode is blocked in txBegin or TxBeginAnon
+	 */
+	struct mutex commit_mutex;
+	/* xattr_sem allows us to access the xattrs without taking i_mutex */
+	struct rw_semaphore xattr_sem;
+	lid_t	xtlid;		/* lid of xtree lock on directory */
+#ifdef CONFIG_JFS_POSIX_ACL
+	struct posix_acl *i_acl;
+	struct posix_acl *i_default_acl;
+#endif
+	union {
+		struct {
+			xtpage_t _xtroot;	/* 288: xtree root */
+			struct inomap *_imap;	/* 4: inode map header	*/
+		} file;
+		struct {
+			struct dir_table_slot _table[12]; /* 96: dir index */
+			dtroot_t _dtroot;	/* 288: dtree root */
+		} dir;
+		struct {
+			unchar _unused[16];	/* 16: */
+			dxd_t _dxd;		/* 16: */
+			unchar _inline[128];	/* 128: inline symlink */
+			/* _inline_ea may overlay the last part of
+			 * file._xtroot if maxentry = XTROOTINITSLOT
+			 */
+			unchar _inline_ea[128];	/* 128: inline extended attr */
+		} link;
+	} u;
+	u32 dev;	/* will die when we get wide dev_t */
+	struct inode	vfs_inode;
+};
+#define i_xtroot u.file._xtroot
+#define i_imap u.file._imap
+#define i_dirtable u.dir._table
+#define i_dtroot u.dir._dtroot
+#define i_inline u.link._inline
+#define i_inline_ea u.link._inline_ea
+
+#define JFS_ACL_NOT_CACHED ((void *)-1)
+
+#define IREAD_LOCK(ip, subclass) \
+	down_read_nested(&JFS_IP(ip)->rdwrlock, subclass)
+#define IREAD_UNLOCK(ip)	up_read(&JFS_IP(ip)->rdwrlock)
+#define IWRITE_LOCK(ip, subclass) \
+	down_write_nested(&JFS_IP(ip)->rdwrlock, subclass)
+#define IWRITE_UNLOCK(ip)	up_write(&JFS_IP(ip)->rdwrlock)
+
+/*
+ * cflag
+ */
+enum cflags {
+	COMMIT_Nolink,		/* inode committed with zero link count */
+	COMMIT_Inlineea,	/* commit inode inline EA */
+	COMMIT_Freewmap,	/* free WMAP at iClose() */
+	COMMIT_Dirty,		/* Inode is really dirty */
+	COMMIT_Dirtable,	/* commit changes to di_dirtable */
+	COMMIT_Stale,		/* data extent is no longer valid */
+	COMMIT_Synclist,	/* metadata pages on group commit synclist */
+};
+
+/*
+ * commit_mutex nesting subclasses:
+ */
+enum commit_mutex_class
+{
+	COMMIT_MUTEX_PARENT,
+	COMMIT_MUTEX_CHILD,
+	COMMIT_MUTEX_SECOND_PARENT,	/* Renaming */
+	COMMIT_MUTEX_VICTIM		/* Inode being unlinked due to rename */
+};
+
+/*
+ * rdwrlock subclasses:
+ * The dmap inode may be locked while a normal inode or the imap inode are
+ * locked.
+ */
+enum rdwrlock_class
+{
+	RDWRLOCK_NORMAL,
+	RDWRLOCK_IMAP,
+	RDWRLOCK_DMAP
+};
+
+#define set_cflag(flag, ip)	set_bit(flag, &(JFS_IP(ip)->cflag))
+#define clear_cflag(flag, ip)	clear_bit(flag, &(JFS_IP(ip)->cflag))
+#define test_cflag(flag, ip)	test_bit(flag, &(JFS_IP(ip)->cflag))
+#define test_and_clear_cflag(flag, ip) \
+	test_and_clear_bit(flag, &(JFS_IP(ip)->cflag))
+/*
+ * JFS-private superblock information.
+ */
+struct jfs_sb_info {
+	struct super_block *sb;		/* Point back to vfs super block */
+	unsigned long	mntflag;	/* aggregate attributes	*/
+	struct inode	*ipbmap;	/* block map inode		*/
+	struct inode	*ipaimap;	/* aggregate inode map inode	*/
+	struct inode	*ipaimap2;	/* secondary aimap inode	*/
+	struct inode	*ipimap;	/* aggregate inode map inode	*/
+	struct jfs_log	*log;		/* log			*/
+	struct list_head log_list;	/* volumes associated with a journal */
+	short		bsize;		/* logical block size	*/
+	short		l2bsize;	/* log2 logical block size	*/
+	short		nbperpage;	/* blocks per page		*/
+	short		l2nbperpage;	/* log2 blocks per page	*/
+	short		l2niperblk;	/* log2 inodes per page	*/
+	dev_t		logdev;		/* external log device	*/
+	uint		aggregate;	/* volume identifier in log record */
+	pxd_t		logpxd;		/* pxd describing log	*/
+	pxd_t		fsckpxd;	/* pxd describing fsck wkspc */
+	pxd_t		ait2;		/* pxd describing AIT copy	*/
+	char		uuid[16];	/* 128-bit uuid for volume	*/
+	char		loguuid[16];	/* 128-bit uuid for log	*/
+	/*
+	 * commit_state is used for synchronization of the jfs_commit
+	 * threads.  It is protected by LAZY_LOCK().
+	 */
+	int		commit_state;	/* commit state */
+	/* Formerly in ipimap */
+	uint		gengen;		/* inode generation generator*/
+	uint		inostamp;	/* shows inode belongs to fileset*/
+
+        /* Formerly in ipbmap */
+	struct bmap	*bmap;		/* incore bmap descriptor	*/
+	struct nls_table *nls_tab;	/* current codepage		*/
+	struct inode *direct_inode;	/* metadata inode */
+	uint		state;		/* mount/recovery state	*/
+	unsigned long	flag;		/* mount time flags */
+	uint		p_state;	/* state prior to going no integrity */
+	uint		uid;		/* uid to override on-disk uid */
+	uint		gid;		/* gid to override on-disk gid */
+	uint		umask;		/* umask to override on-disk umask */
+};
+
+/* jfs_sb_info commit_state */
+#define IN_LAZYCOMMIT 1
+
+static inline struct jfs_inode_info *JFS_IP(struct inode *inode)
+{
+	return list_entry(inode, struct jfs_inode_info, vfs_inode);
+}
+
+static inline int jfs_dirtable_inline(struct inode *inode)
+{
+	return (JFS_IP(inode)->next_index <= (MAX_INLINE_DIRTABLE_ENTRY + 1));
+}
+
+static inline struct jfs_sb_info *JFS_SBI(struct super_block *sb)
+{
+	return sb->s_fs_info;
+}
+
+static inline int isReadOnly(struct inode *inode)
+{
+	if (JFS_SBI(inode->i_sb)->log)
+		return 0;
+	return 1;
+}
+#endif /* _H_JFS_INCORE */

fs/jfs/jfs_inode.c

@@ -0,0 +1,143 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/quotaops.h>
+#include "jfs_incore.h"
+#include "jfs_inode.h"
+#include "jfs_filsys.h"
+#include "jfs_imap.h"
+#include "jfs_dinode.h"
+#include "jfs_debug.h"
+
+
+void jfs_set_inode_flags(struct inode *inode)
+{
+	unsigned int flags = JFS_IP(inode)->mode2;
+
+	inode->i_flags &= ~(S_IMMUTABLE | S_APPEND |
+		S_NOATIME | S_DIRSYNC | S_SYNC);
+
+	if (flags & JFS_IMMUTABLE_FL)
+		inode->i_flags |= S_IMMUTABLE;
+	if (flags & JFS_APPEND_FL)
+		inode->i_flags |= S_APPEND;
+	if (flags & JFS_NOATIME_FL)
+		inode->i_flags |= S_NOATIME;
+	if (flags & JFS_DIRSYNC_FL)
+		inode->i_flags |= S_DIRSYNC;
+	if (flags & JFS_SYNC_FL)
+		inode->i_flags |= S_SYNC;
+}
+
+/*
+ * NAME:	ialloc()
+ *
+ * FUNCTION:	Allocate a new inode
+ *
+ */
+struct inode *ialloc(struct inode *parent, umode_t mode)
+{
+	struct super_block *sb = parent->i_sb;
+	struct inode *inode;
+	struct jfs_inode_info *jfs_inode;
+	int rc;
+
+	inode = new_inode(sb);
+	if (!inode) {
+		jfs_warn("ialloc: new_inode returned NULL!");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	jfs_inode = JFS_IP(inode);
+
+	rc = diAlloc(parent, S_ISDIR(mode), inode);
+	if (rc) {
+		jfs_warn("ialloc: diAlloc returned %d!", rc);
+		if (rc == -EIO)
+			make_bad_inode(inode);
+		iput(inode);
+		return ERR_PTR(rc);
+	}
+
+	inode->i_uid = current->fsuid;
+	if (parent->i_mode & S_ISGID) {
+		inode->i_gid = parent->i_gid;
+		if (S_ISDIR(mode))
+			mode |= S_ISGID;
+	} else
+		inode->i_gid = current->fsgid;
+
+	/*
+	 * New inodes need to save sane values on disk when
+	 * uid & gid mount options are used
+	 */
+	jfs_inode->saved_uid = inode->i_uid;
+	jfs_inode->saved_gid = inode->i_gid;
+
+	/*
+	 * Allocate inode to quota.
+	 */
+	if (DQUOT_ALLOC_INODE(inode)) {
+		DQUOT_DROP(inode);
+		inode->i_flags |= S_NOQUOTA;
+		inode->i_nlink = 0;
+		iput(inode);
+		return ERR_PTR(-EDQUOT);
+	}
+
+	inode->i_mode = mode;
+	/* inherit flags from parent */
+	jfs_inode->mode2 = JFS_IP(parent)->mode2 & JFS_FL_INHERIT;
+
+	if (S_ISDIR(mode)) {
+		jfs_inode->mode2 |= IDIRECTORY;
+		jfs_inode->mode2 &= ~JFS_DIRSYNC_FL;
+	}
+	else {
+		jfs_inode->mode2 |= INLINEEA | ISPARSE;
+		if (S_ISLNK(mode))
+			jfs_inode->mode2 &= ~(JFS_IMMUTABLE_FL|JFS_APPEND_FL);
+	}
+	jfs_inode->mode2 |= mode;
+
+	inode->i_blocks = 0;
+	inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
+	jfs_inode->otime = inode->i_ctime.tv_sec;
+	inode->i_generation = JFS_SBI(sb)->gengen++;
+
+	jfs_inode->cflag = 0;
+
+	/* Zero remaining fields */
+	memset(&jfs_inode->acl, 0, sizeof(dxd_t));
+	memset(&jfs_inode->ea, 0, sizeof(dxd_t));
+	jfs_inode->next_index = 0;
+	jfs_inode->acltype = 0;
+	jfs_inode->btorder = 0;
+	jfs_inode->btindex = 0;
+	jfs_inode->bxflag = 0;
+	jfs_inode->blid = 0;
+	jfs_inode->atlhead = 0;
+	jfs_inode->atltail = 0;
+	jfs_inode->xtlid = 0;
+	jfs_set_inode_flags(inode);
+
+	jfs_info("ialloc returns inode = 0x%p\n", inode);
+
+	return inode;
+}

fs/jfs/jfs_inode.h

@@ -0,0 +1,44 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2001
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_INODE
+#define _H_JFS_INODE
+
+extern struct inode *ialloc(struct inode *, umode_t);
+extern int jfs_fsync(struct file *, struct dentry *, int);
+extern int jfs_ioctl(struct inode *, struct file *,
+			unsigned int, unsigned long);
+extern void jfs_read_inode(struct inode *);
+extern int jfs_commit_inode(struct inode *, int);
+extern int jfs_write_inode(struct inode*, int);
+extern void jfs_delete_inode(struct inode *);
+extern void jfs_dirty_inode(struct inode *);
+extern void jfs_truncate(struct inode *);
+extern void jfs_truncate_nolock(struct inode *, loff_t);
+extern void jfs_free_zero_link(struct inode *);
+extern struct dentry *jfs_get_parent(struct dentry *dentry);
+extern void jfs_set_inode_flags(struct inode *);
+extern int jfs_get_block(struct inode *, sector_t, struct buffer_head *, int);
+
+extern const struct address_space_operations jfs_aops;
+extern const struct inode_operations jfs_dir_inode_operations;
+extern const struct file_operations jfs_dir_operations;
+extern const struct inode_operations jfs_file_inode_operations;
+extern const struct file_operations jfs_file_operations;
+extern const struct inode_operations jfs_symlink_inode_operations;
+extern struct dentry_operations jfs_ci_dentry_operations;
+#endif				/* _H_JFS_INODE */

fs/jfs/jfs_lock.h

@@ -0,0 +1,52 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2001
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_LOCK
+#define _H_JFS_LOCK
+
+#include <linux/spinlock.h>
+#include <linux/mutex.h>
+#include <linux/sched.h>
+
+/*
+ *	jfs_lock.h
+ */
+
+/*
+ * Conditional sleep where condition is protected by spinlock
+ *
+ * lock_cmd and unlock_cmd take and release the spinlock
+ */
+#define __SLEEP_COND(wq, cond, lock_cmd, unlock_cmd)	\
+do {							\
+	DECLARE_WAITQUEUE(__wait, current);		\
+							\
+	add_wait_queue(&wq, &__wait);			\
+	for (;;) {					\
+		set_current_state(TASK_UNINTERRUPTIBLE);\
+		if (cond)				\
+			break;				\
+		unlock_cmd;				\
+		io_schedule();				\
+		lock_cmd;				\
+	}						\
+	current->state = TASK_RUNNING;			\
+	remove_wait_queue(&wq, &__wait);		\
+} while (0)
+
+#endif				/* _H_JFS_LOCK */

fs/jfs/jfs_logmgr.h

@@ -0,0 +1,513 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_LOGMGR
+#define _H_JFS_LOGMGR
+
+#include "jfs_filsys.h"
+#include "jfs_lock.h"
+
+/*
+ *	log manager configuration parameters
+ */
+
+/* log page size */
+#define	LOGPSIZE	4096
+#define	L2LOGPSIZE	12
+
+#define LOGPAGES	16	/* Log pages per mounted file system */
+
+/*
+ *	log logical volume
+ *
+ * a log is used to make the commit operation on journalled
+ * files within the same logical volume group atomic.
+ * a log is implemented with a logical volume.
+ * there is one log per logical volume group.
+ *
+ * block 0 of the log logical volume is not used (ipl etc).
+ * block 1 contains a log "superblock" and is used by logFormat(),
+ * lmLogInit(), lmLogShutdown(), and logRedo() to record status
+ * of the log but is not otherwise used during normal processing.
+ * blocks 2 - (N-1) are used to contain log records.
+ *
+ * when a volume group is varied-on-line, logRedo() must have
+ * been executed before the file systems (logical volumes) in
+ * the volume group can be mounted.
+ */
+/*
+ *	log superblock (block 1 of logical volume)
+ */
+#define	LOGSUPER_B	1
+#define	LOGSTART_B	2
+
+#define	LOGMAGIC	0x87654321
+#define	LOGVERSION	1
+
+#define MAX_ACTIVE	128	/* Max active file systems sharing log */
+
+struct logsuper {
+	__le32 magic;		/* 4: log lv identifier */
+	__le32 version;		/* 4: version number */
+	__le32 serial;		/* 4: log open/mount counter */
+	__le32 size;		/* 4: size in number of LOGPSIZE blocks */
+	__le32 bsize;		/* 4: logical block size in byte */
+	__le32 l2bsize;		/* 4: log2 of bsize */
+
+	__le32 flag;		/* 4: option */
+	__le32 state;		/* 4: state - see below */
+
+	__le32 end;		/* 4: addr of last log record set by logredo */
+	char uuid[16];		/* 16: 128-bit journal uuid */
+	char label[16];		/* 16: journal label */
+	struct {
+		char uuid[16];
+	} active[MAX_ACTIVE];	/* 2048: active file systems list */
+};
+
+#define NULL_UUID "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"
+
+/* log flag: commit option (see jfs_filsys.h) */
+
+/* log state */
+#define	LOGMOUNT	0	/* log mounted by lmLogInit() */
+#define LOGREDONE	1	/* log shutdown by lmLogShutdown().
+				 * log redo completed by logredo().
+				 */
+#define LOGWRAP		2	/* log wrapped */
+#define LOGREADERR	3	/* log read error detected in logredo() */
+
+
+/*
+ *	log logical page
+ *
+ * (this comment should be rewritten !)
+ * the header and trailer structures (h,t) will normally have
+ * the same page and eor value.
+ * An exception to this occurs when a complete page write is not
+ * accomplished on a power failure. Since the hardware may "split write"
+ * sectors in the page, any out of order sequence may occur during powerfail
+ * and needs to be recognized during log replay.  The xor value is
+ * an "exclusive or" of all log words in the page up to eor.  This
+ * 32 bit eor is stored with the top 16 bits in the header and the
+ * bottom 16 bits in the trailer.  logredo can easily recognize pages
+ * that were not completed by reconstructing this eor and checking
+ * the log page.
+ *
+ * Previous versions of the operating system did not allow split
+ * writes and detected partially written records in logredo by
+ * ordering the updates to the header, trailer, and the move of data
+ * into the logdata area.  The order: (1) data is moved (2) header
+ * is updated (3) trailer is updated.  In logredo, when the header
+ * differed from the trailer, the header and trailer were reconciled
+ * as follows: if h.page != t.page they were set to the smaller of
+ * the two and h.eor and t.eor set to 8 (i.e. empty page). if (only)
+ * h.eor != t.eor they were set to the smaller of their two values.
+ */
+struct logpage {
+	struct {		/* header */
+		__le32 page;	/* 4: log sequence page number */
+		__le16 rsrvd;	/* 2: */
+		__le16 eor;	/* 2: end-of-log offset of lasrt record write */
+	} h;
+
+	__le32 data[LOGPSIZE / 4 - 4];	/* log record area */
+
+	struct {		/* trailer */
+		__le32 page;	/* 4: normally the same as h.page */
+		__le16 rsrvd;	/* 2: */
+		__le16 eor;	/* 2: normally the same as h.eor */
+	} t;
+};
+
+#define LOGPHDRSIZE	8	/* log page header size */
+#define LOGPTLRSIZE	8	/* log page trailer size */
+
+
+/*
+ *	log record
+ *
+ * (this comment should be rewritten !)
+ * jfs uses only "after" log records (only a single writer is allowed
+ * in a  page, pages are written to temporary paging space if
+ * if they must be written to disk before commit, and i/o is
+ * scheduled for modified pages to their home location after
+ * the log records containing the after values and the commit
+ * record is written to the log on disk, undo discards the copy
+ * in main-memory.)
+ *
+ * a log record consists of a data area of variable length followed by
+ * a descriptor of fixed size LOGRDSIZE bytes.
+ * the  data area is rounded up to an integral number of 4-bytes and
+ * must be no longer than LOGPSIZE.
+ * the descriptor is of size of multiple of 4-bytes and aligned on a
+ * 4-byte boundary.
+ * records are packed one after the other in the data area of log pages.
+ * (sometimes a DUMMY record is inserted so that at least one record ends
+ * on every page or the longest record is placed on at most two pages).
+ * the field eor in page header/trailer points to the byte following
+ * the last record on a page.
+ */
+
+/* log record types */
+#define LOG_COMMIT		0x8000
+#define LOG_SYNCPT		0x4000
+#define LOG_MOUNT		0x2000
+#define LOG_REDOPAGE		0x0800
+#define LOG_NOREDOPAGE		0x0080
+#define LOG_NOREDOINOEXT	0x0040
+#define LOG_UPDATEMAP		0x0008
+#define LOG_NOREDOFILE		0x0001
+
+/* REDOPAGE/NOREDOPAGE log record data type */
+#define	LOG_INODE		0x0001
+#define	LOG_XTREE		0x0002
+#define	LOG_DTREE		0x0004
+#define	LOG_BTROOT		0x0010
+#define	LOG_EA			0x0020
+#define	LOG_ACL			0x0040
+#define	LOG_DATA		0x0080
+#define	LOG_NEW			0x0100
+#define	LOG_EXTEND		0x0200
+#define LOG_RELOCATE		0x0400
+#define LOG_DIR_XTREE		0x0800	/* Xtree is in directory inode */
+
+/* UPDATEMAP log record descriptor type */
+#define	LOG_ALLOCXADLIST	0x0080
+#define	LOG_ALLOCPXDLIST	0x0040
+#define	LOG_ALLOCXAD		0x0020
+#define	LOG_ALLOCPXD		0x0010
+#define	LOG_FREEXADLIST		0x0008
+#define	LOG_FREEPXDLIST		0x0004
+#define	LOG_FREEXAD		0x0002
+#define	LOG_FREEPXD		0x0001
+
+
+struct lrd {
+	/*
+	 * type independent area
+	 */
+	__le32 logtid;		/* 4: log transaction identifier */
+	__le32 backchain;	/* 4: ptr to prev record of same transaction */
+	__le16 type;		/* 2: record type */
+	__le16 length;		/* 2: length of data in record (in byte) */
+	__le32 aggregate;	/* 4: file system lv/aggregate */
+	/* (16) */
+
+	/*
+	 * type dependent area (20)
+	 */
+	union {
+
+		/*
+		 *      COMMIT: commit
+		 *
+		 * transaction commit: no type-dependent information;
+		 */
+
+		/*
+		 *      REDOPAGE: after-image
+		 *
+		 * apply after-image;
+		 *
+		 * N.B. REDOPAGE, NOREDOPAGE, and UPDATEMAP must be same format;
+		 */
+		struct {
+			__le32 fileset;	/* 4: fileset number */
+			__le32 inode;	/* 4: inode number */
+			__le16 type;	/* 2: REDOPAGE record type */
+			__le16 l2linesize;	/* 2: log2 of line size */
+			pxd_t pxd;	/* 8: on-disk page pxd */
+		} redopage;	/* (20) */
+
+		/*
+		 *      NOREDOPAGE: the page is freed
+		 *
+		 * do not apply after-image records which precede this record
+		 * in the log with the same page block number to this page.
+		 *
+		 * N.B. REDOPAGE, NOREDOPAGE, and UPDATEMAP must be same format;
+		 */
+		struct {
+			__le32 fileset;	/* 4: fileset number */
+			__le32 inode;	/* 4: inode number */
+			__le16 type;	/* 2: NOREDOPAGE record type */
+			__le16 rsrvd;	/* 2: reserved */
+			pxd_t pxd;	/* 8: on-disk page pxd */
+		} noredopage;	/* (20) */
+
+		/*
+		 *      UPDATEMAP: update block allocation map
+		 *
+		 * either in-line PXD,
+		 * or     out-of-line  XADLIST;
+		 *
+		 * N.B. REDOPAGE, NOREDOPAGE, and UPDATEMAP must be same format;
+		 */
+		struct {
+			__le32 fileset;	/* 4: fileset number */
+			__le32 inode;	/* 4: inode number */
+			__le16 type;	/* 2: UPDATEMAP record type */
+			__le16 nxd;	/* 2: number of extents */
+			pxd_t pxd;	/* 8: pxd */
+		} updatemap;	/* (20) */
+
+		/*
+		 *      NOREDOINOEXT: the inode extent is freed
+		 *
+		 * do not apply after-image records which precede this
+		 * record in the log with the any of the 4 page block
+		 * numbers in this inode extent.
+		 *
+		 * NOTE: The fileset and pxd fields MUST remain in
+		 *       the same fields in the REDOPAGE record format.
+		 *
+		 */
+		struct {
+			__le32 fileset;	/* 4: fileset number */
+			__le32 iagnum;	/* 4: IAG number     */
+			__le32 inoext_idx;	/* 4: inode extent index */
+			pxd_t pxd;	/* 8: on-disk page pxd */
+		} noredoinoext;	/* (20) */
+
+		/*
+		 *      SYNCPT: log sync point
+		 *
+		 * replay log upto syncpt address specified;
+		 */
+		struct {
+			__le32 sync;	/* 4: syncpt address (0 = here) */
+		} syncpt;
+
+		/*
+		 *      MOUNT: file system mount
+		 *
+		 * file system mount: no type-dependent information;
+		 */
+
+		/*
+		 *      ? FREEXTENT: free specified extent(s)
+		 *
+		 * free specified extent(s) from block allocation map
+		 * N.B.: nextents should be length of data/sizeof(xad_t)
+		 */
+		struct {
+			__le32 type;	/* 4: FREEXTENT record type */
+			__le32 nextent;	/* 4: number of extents */
+
+			/* data: PXD or XAD list */
+		} freextent;
+
+		/*
+		 *      ? NOREDOFILE: this file is freed
+		 *
+		 * do not apply records which precede this record in the log
+		 * with the same inode number.
+		 *
+		 * NOREDOFILE must be the first to be written at commit
+		 * (last to be read in logredo()) - it prevents
+		 * replay of preceding updates of all preceding generations
+		 * of the inumber esp. the on-disk inode itself.
+		 */
+		struct {
+			__le32 fileset;	/* 4: fileset number */
+			__le32 inode;	/* 4: inode number */
+		} noredofile;
+
+		/*
+		 *      ? NEWPAGE:
+		 *
+		 * metadata type dependent
+		 */
+		struct {
+			__le32 fileset;	/* 4: fileset number */
+			__le32 inode;	/* 4: inode number */
+			__le32 type;	/* 4: NEWPAGE record type */
+			pxd_t pxd;	/* 8: on-disk page pxd */
+		} newpage;
+
+		/*
+		 *      ? DUMMY: filler
+		 *
+		 * no type-dependent information
+		 */
+	} log;
+};					/* (36) */
+
+#define	LOGRDSIZE	(sizeof(struct lrd))
+
+/*
+ *	line vector descriptor
+ */
+struct lvd {
+	__le16 offset;
+	__le16 length;
+};
+
+
+/*
+ *	log logical volume
+ */
+struct jfs_log {
+
+	struct list_head sb_list;/*  This is used to sync metadata
+				 *    before writing syncpt.
+				 */
+	struct list_head journal_list; /* Global list */
+	struct block_device *bdev; /* 4: log lv pointer */
+	int serial;		/* 4: log mount serial number */
+
+	s64 base;		/* @8: log extent address (inline log ) */
+	int size;		/* 4: log size in log page (in page) */
+	int l2bsize;		/* 4: log2 of bsize */
+
+	long flag;		/* 4: flag */
+
+	struct lbuf *lbuf_free;	/* 4: free lbufs */
+	wait_queue_head_t free_wait;	/* 4: */
+
+	/* log write */
+	int logtid;		/* 4: log tid */
+	int page;		/* 4: page number of eol page */
+	int eor;		/* 4: eor of last record in eol page */
+	struct lbuf *bp;	/* 4: current log page buffer */
+
+	struct mutex loglock;	/* 4: log write serialization lock */
+
+	/* syncpt */
+	int nextsync;		/* 4: bytes to write before next syncpt */
+	int active;		/* 4: */
+	wait_queue_head_t syncwait;	/* 4: */
+
+	/* commit */
+	uint cflag;		/* 4: */
+	struct list_head cqueue; /* FIFO commit queue */
+	struct tblock *flush_tblk; /* tblk we're waiting on for flush */
+	int gcrtc;		/* 4: GC_READY transaction count */
+	struct tblock *gclrt;	/* 4: latest GC_READY transaction */
+	spinlock_t gclock;	/* 4: group commit lock */
+	int logsize;		/* 4: log data area size in byte */
+	int lsn;		/* 4: end-of-log */
+	int clsn;		/* 4: clsn */
+	int syncpt;		/* 4: addr of last syncpt record */
+	int sync;		/* 4: addr from last logsync() */
+	struct list_head synclist;	/* 8: logsynclist anchor */
+	spinlock_t synclock;	/* 4: synclist lock */
+	struct lbuf *wqueue;	/* 4: log pageout queue */
+	int count;		/* 4: count */
+	char uuid[16];		/* 16: 128-bit uuid of log device */
+
+	int no_integrity;	/* 3: flag to disable journaling to disk */
+};
+
+/*
+ * Log flag
+ */
+#define log_INLINELOG	1
+#define log_SYNCBARRIER	2
+#define log_QUIESCE	3
+#define log_FLUSH	4
+
+/*
+ * group commit flag
+ */
+/* jfs_log */
+#define logGC_PAGEOUT	0x00000001
+
+/* tblock/lbuf */
+#define tblkGC_QUEUE		0x0001
+#define tblkGC_READY		0x0002
+#define tblkGC_COMMIT		0x0004
+#define tblkGC_COMMITTED	0x0008
+#define tblkGC_EOP		0x0010
+#define tblkGC_FREE		0x0020
+#define tblkGC_LEADER		0x0040
+#define tblkGC_ERROR		0x0080
+#define tblkGC_LAZY		0x0100	// D230860
+#define tblkGC_UNLOCKED		0x0200	// D230860
+
+/*
+ *		log cache buffer header
+ */
+struct lbuf {
+	struct jfs_log *l_log;	/* 4: log associated with buffer */
+
+	/*
+	 * data buffer base area
+	 */
+	uint l_flag;		/* 4: pageout control flags */
+
+	struct lbuf *l_wqnext;	/* 4: write queue link */
+	struct lbuf *l_freelist;	/* 4: freelistlink */
+
+	int l_pn;		/* 4: log page number */
+	int l_eor;		/* 4: log record eor */
+	int l_ceor;		/* 4: committed log record eor */
+
+	s64 l_blkno;		/* 8: log page block number */
+	caddr_t l_ldata;	/* 4: data page */
+	struct page *l_page;	/* The page itself */
+	uint l_offset;		/* Offset of l_ldata within the page */
+
+	wait_queue_head_t l_ioevent;	/* 4: i/o done event */
+};
+
+/* Reuse l_freelist for redrive list */
+#define l_redrive_next l_freelist
+
+/*
+ *	logsynclist block
+ *
+ * common logsyncblk prefix for jbuf_t and tblock
+ */
+struct logsyncblk {
+	u16 xflag;		/* flags */
+	u16 flag;		/* only meaninful in tblock */
+	lid_t lid;		/* lock id */
+	s32 lsn;		/* log sequence number */
+	struct list_head synclist;	/* log sync list link */
+};
+
+/*
+ *	logsynclist serialization (per log)
+ */
+
+#define LOGSYNC_LOCK_INIT(log) spin_lock_init(&(log)->synclock)
+#define LOGSYNC_LOCK(log, flags) spin_lock_irqsave(&(log)->synclock, flags)
+#define LOGSYNC_UNLOCK(log, flags) \
+	spin_unlock_irqrestore(&(log)->synclock, flags)
+
+/* compute the difference in bytes of lsn from sync point */
+#define logdiff(diff, lsn, log)\
+{\
+	diff = (lsn) - (log)->syncpt;\
+	if (diff < 0)\
+		diff += (log)->logsize;\
+}
+
+extern int lmLogOpen(struct super_block *sb);
+extern int lmLogClose(struct super_block *sb);
+extern int lmLogShutdown(struct jfs_log * log);
+extern int lmLogInit(struct jfs_log * log);
+extern int lmLogFormat(struct jfs_log *log, s64 logAddress, int logSize);
+extern int lmGroupCommit(struct jfs_log *, struct tblock *);
+extern int jfsIOWait(void *);
+extern void jfs_flush_journal(struct jfs_log * log, int wait);
+extern void jfs_syncpt(struct jfs_log *log, int hard_sync);
+
+#endif				/* _H_JFS_LOGMGR */

fs/jfs/jfs_metapage.c

@@ -0,0 +1,845 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2005
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/mm.h>
+#include <linux/bio.h>
+#include <linux/init.h>
+#include <linux/buffer_head.h>
+#include <linux/mempool.h>
+#include "jfs_incore.h"
+#include "jfs_superblock.h"
+#include "jfs_filsys.h"
+#include "jfs_metapage.h"
+#include "jfs_txnmgr.h"
+#include "jfs_debug.h"
+
+#ifdef CONFIG_JFS_STATISTICS
+static struct {
+	uint	pagealloc;	/* # of page allocations */
+	uint	pagefree;	/* # of page frees */
+	uint	lockwait;	/* # of sleeping lock_metapage() calls */
+} mpStat;
+#endif
+
+#define metapage_locked(mp) test_bit(META_locked, &(mp)->flag)
+#define trylock_metapage(mp) test_and_set_bit(META_locked, &(mp)->flag)
+
+static inline void unlock_metapage(struct metapage *mp)
+{
+	clear_bit(META_locked, &mp->flag);
+	wake_up(&mp->wait);
+}
+
+static inline void __lock_metapage(struct metapage *mp)
+{
+	DECLARE_WAITQUEUE(wait, current);
+	INCREMENT(mpStat.lockwait);
+	add_wait_queue_exclusive(&mp->wait, &wait);
+	do {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		if (metapage_locked(mp)) {
+			unlock_page(mp->page);
+			io_schedule();
+			lock_page(mp->page);
+		}
+	} while (trylock_metapage(mp));
+	__set_current_state(TASK_RUNNING);
+	remove_wait_queue(&mp->wait, &wait);
+}
+
+/*
+ * Must have mp->page locked
+ */
+static inline void lock_metapage(struct metapage *mp)
+{
+	if (trylock_metapage(mp))
+		__lock_metapage(mp);
+}
+
+#define METAPOOL_MIN_PAGES 32
+static struct kmem_cache *metapage_cache;
+static mempool_t *metapage_mempool;
+
+#define MPS_PER_PAGE (PAGE_CACHE_SIZE >> L2PSIZE)
+
+#if MPS_PER_PAGE > 1
+
+struct meta_anchor {
+	int mp_count;
+	atomic_t io_count;
+	struct metapage *mp[MPS_PER_PAGE];
+};
+#define mp_anchor(page) ((struct meta_anchor *)page_private(page))
+
+static inline struct metapage *page_to_mp(struct page *page, uint offset)
+{
+	if (!PagePrivate(page))
+		return NULL;
+	return mp_anchor(page)->mp[offset >> L2PSIZE];
+}
+
+static inline int insert_metapage(struct page *page, struct metapage *mp)
+{
+	struct meta_anchor *a;
+	int index;
+	int l2mp_blocks;	/* log2 blocks per metapage */
+
+	if (PagePrivate(page))
+		a = mp_anchor(page);
+	else {
+		a = kzalloc(sizeof(struct meta_anchor), GFP_NOFS);
+		if (!a)
+			return -ENOMEM;
+		set_page_private(page, (unsigned long)a);
+		SetPagePrivate(page);
+		kmap(page);
+	}
+
+	if (mp) {
+		l2mp_blocks = L2PSIZE - page->mapping->host->i_blkbits;
+		index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1);
+		a->mp_count++;
+		a->mp[index] = mp;
+	}
+
+	return 0;
+}
+
+static inline void remove_metapage(struct page *page, struct metapage *mp)
+{
+	struct meta_anchor *a = mp_anchor(page);
+	int l2mp_blocks = L2PSIZE - page->mapping->host->i_blkbits;
+	int index;
+
+	index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1);
+
+	BUG_ON(a->mp[index] != mp);
+
+	a->mp[index] = NULL;
+	if (--a->mp_count == 0) {
+		kfree(a);
+		set_page_private(page, 0);
+		ClearPagePrivate(page);
+		kunmap(page);
+	}
+}
+
+static inline void inc_io(struct page *page)
+{
+	atomic_inc(&mp_anchor(page)->io_count);
+}
+
+static inline void dec_io(struct page *page, void (*handler) (struct page *))
+{
+	if (atomic_dec_and_test(&mp_anchor(page)->io_count))
+		handler(page);
+}
+
+#else
+static inline struct metapage *page_to_mp(struct page *page, uint offset)
+{
+	return PagePrivate(page) ? (struct metapage *)page_private(page) : NULL;
+}
+
+static inline int insert_metapage(struct page *page, struct metapage *mp)
+{
+	if (mp) {
+		set_page_private(page, (unsigned long)mp);
+		SetPagePrivate(page);
+		kmap(page);
+	}
+	return 0;
+}
+
+static inline void remove_metapage(struct page *page, struct metapage *mp)
+{
+	set_page_private(page, 0);
+	ClearPagePrivate(page);
+	kunmap(page);
+}
+
+#define inc_io(page) do {} while(0)
+#define dec_io(page, handler) handler(page)
+
+#endif
+
+static void init_once(void *foo, struct kmem_cache *cachep, unsigned long flags)
+{
+	struct metapage *mp = (struct metapage *)foo;
+
+	if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) ==
+	    SLAB_CTOR_CONSTRUCTOR) {
+		mp->lid = 0;
+		mp->lsn = 0;
+		mp->flag = 0;
+		mp->data = NULL;
+		mp->clsn = 0;
+		mp->log = NULL;
+		set_bit(META_free, &mp->flag);
+		init_waitqueue_head(&mp->wait);
+	}
+}
+
+static inline struct metapage *alloc_metapage(gfp_t gfp_mask)
+{
+	return mempool_alloc(metapage_mempool, gfp_mask);
+}
+
+static inline void free_metapage(struct metapage *mp)
+{
+	mp->flag = 0;
+	set_bit(META_free, &mp->flag);
+
+	mempool_free(mp, metapage_mempool);
+}
+
+int __init metapage_init(void)
+{
+	/*
+	 * Allocate the metapage structures
+	 */
+	metapage_cache = kmem_cache_create("jfs_mp", sizeof(struct metapage),
+					   0, 0, init_once, NULL);
+	if (metapage_cache == NULL)
+		return -ENOMEM;
+
+	metapage_mempool = mempool_create_slab_pool(METAPOOL_MIN_PAGES,
+						    metapage_cache);
+
+	if (metapage_mempool == NULL) {
+		kmem_cache_destroy(metapage_cache);
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+void metapage_exit(void)
+{
+	mempool_destroy(metapage_mempool);
+	kmem_cache_destroy(metapage_cache);
+}
+
+static inline void drop_metapage(struct page *page, struct metapage *mp)
+{
+	if (mp->count || mp->nohomeok || test_bit(META_dirty, &mp->flag) ||
+	    test_bit(META_io, &mp->flag))
+		return;
+	remove_metapage(page, mp);
+	INCREMENT(mpStat.pagefree);
+	free_metapage(mp);
+}
+
+/*
+ * Metapage address space operations
+ */
+
+static sector_t metapage_get_blocks(struct inode *inode, sector_t lblock,
+				    unsigned int *len)
+{
+	int rc = 0;
+	int xflag;
+	s64 xaddr;
+	sector_t file_blocks = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
+			       inode->i_blkbits;
+
+	if (lblock >= file_blocks)
+		return 0;
+	if (lblock + *len > file_blocks)
+		*len = file_blocks - lblock;
+
+	if (inode->i_ino) {
+		rc = xtLookup(inode, (s64)lblock, *len, &xflag, &xaddr, len, 0);
+		if ((rc == 0) && *len)
+			lblock = (sector_t)xaddr;
+		else
+			lblock = 0;
+	} /* else no mapping */
+
+	return lblock;
+}
+
+static void last_read_complete(struct page *page)
+{
+	if (!PageError(page))
+		SetPageUptodate(page);
+	unlock_page(page);
+}
+
+static int metapage_read_end_io(struct bio *bio, unsigned int bytes_done,
+				int err)
+{
+	struct page *page = bio->bi_private;
+
+	if (bio->bi_size)
+		return 1;
+
+	if (!test_bit(BIO_UPTODATE, &bio->bi_flags)) {
+		printk(KERN_ERR "metapage_read_end_io: I/O error\n");
+		SetPageError(page);
+	}
+
+	dec_io(page, last_read_complete);
+	bio_put(bio);
+
+	return 0;
+}
+
+static void remove_from_logsync(struct metapage *mp)
+{
+	struct jfs_log *log = mp->log;
+	unsigned long flags;
+/*
+ * This can race.  Recheck that log hasn't been set to null, and after
+ * acquiring logsync lock, recheck lsn
+ */
+	if (!log)
+		return;
+
+	LOGSYNC_LOCK(log, flags);
+	if (mp->lsn) {
+		mp->log = NULL;
+		mp->lsn = 0;
+		mp->clsn = 0;
+		log->count--;
+		list_del(&mp->synclist);
+	}
+	LOGSYNC_UNLOCK(log, flags);
+}
+
+static void last_write_complete(struct page *page)
+{
+	struct metapage *mp;
+	unsigned int offset;
+
+	for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
+		mp = page_to_mp(page, offset);
+		if (mp && test_bit(META_io, &mp->flag)) {
+			if (mp->lsn)
+				remove_from_logsync(mp);
+			clear_bit(META_io, &mp->flag);
+		}
+		/*
+		 * I'd like to call drop_metapage here, but I don't think it's
+		 * safe unless I have the page locked
+		 */
+	}
+	end_page_writeback(page);
+}
+
+static int metapage_write_end_io(struct bio *bio, unsigned int bytes_done,
+				 int err)
+{
+	struct page *page = bio->bi_private;
+
+	BUG_ON(!PagePrivate(page));
+
+	if (bio->bi_size)
+		return 1;
+
+	if (! test_bit(BIO_UPTODATE, &bio->bi_flags)) {
+		printk(KERN_ERR "metapage_write_end_io: I/O error\n");
+		SetPageError(page);
+	}
+	dec_io(page, last_write_complete);
+	bio_put(bio);
+	return 0;
+}
+
+static int metapage_writepage(struct page *page, struct writeback_control *wbc)
+{
+	struct bio *bio = NULL;
+	unsigned int block_offset;	/* block offset of mp within page */
+	struct inode *inode = page->mapping->host;
+	unsigned int blocks_per_mp = JFS_SBI(inode->i_sb)->nbperpage;
+	unsigned int len;
+	unsigned int xlen;
+	struct metapage *mp;
+	int redirty = 0;
+	sector_t lblock;
+	sector_t pblock;
+	sector_t next_block = 0;
+	sector_t page_start;
+	unsigned long bio_bytes = 0;
+	unsigned long bio_offset = 0;
+	unsigned int offset;
+
+	page_start = (sector_t)page->index <<
+		     (PAGE_CACHE_SHIFT - inode->i_blkbits);
+	BUG_ON(!PageLocked(page));
+	BUG_ON(PageWriteback(page));
+
+	for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
+		mp = page_to_mp(page, offset);
+
+		if (!mp || !test_bit(META_dirty, &mp->flag))
+			continue;
+
+		if (mp->nohomeok && !test_bit(META_forcewrite, &mp->flag)) {
+			redirty = 1;
+			/*
+			 * Make sure this page isn't blocked indefinitely.
+			 * If the journal isn't undergoing I/O, push it
+			 */
+			if (mp->log && !(mp->log->cflag & logGC_PAGEOUT))
+				jfs_flush_journal(mp->log, 0);
+			continue;
+		}
+
+		clear_bit(META_dirty, &mp->flag);
+		block_offset = offset >> inode->i_blkbits;
+		lblock = page_start + block_offset;
+		if (bio) {
+			if (xlen && lblock == next_block) {
+				/* Contiguous, in memory & on disk */
+				len = min(xlen, blocks_per_mp);
+				xlen -= len;
+				bio_bytes += len << inode->i_blkbits;
+				set_bit(META_io, &mp->flag);
+				continue;
+			}
+			/* Not contiguous */
+			if (bio_add_page(bio, page, bio_bytes, bio_offset) <
+			    bio_bytes)
+				goto add_failed;
+			/*
+			 * Increment counter before submitting i/o to keep
+			 * count from hitting zero before we're through
+			 */
+			inc_io(page);
+			if (!bio->bi_size)
+				goto dump_bio;
+			submit_bio(WRITE, bio);
+			bio = NULL;
+		} else {
+			set_page_writeback(page);
+			inc_io(page);
+		}
+		xlen = (PAGE_CACHE_SIZE - offset) >> inode->i_blkbits;
+		pblock = metapage_get_blocks(inode, lblock, &xlen);
+		if (!pblock) {
+			/* Need better error handling */
+			printk(KERN_ERR "JFS: metapage_get_blocks failed\n");
+			dec_io(page, last_write_complete);
+			continue;
+		}
+		set_bit(META_io, &mp->flag);
+		len = min(xlen, (uint) JFS_SBI(inode->i_sb)->nbperpage);
+
+		bio = bio_alloc(GFP_NOFS, 1);
+		bio->bi_bdev = inode->i_sb->s_bdev;
+		bio->bi_sector = pblock << (inode->i_blkbits - 9);
+		bio->bi_end_io = metapage_write_end_io;
+		bio->bi_private = page;
+
+		/* Don't call bio_add_page yet, we may add to this vec */
+		bio_offset = offset;
+		bio_bytes = len << inode->i_blkbits;
+
+		xlen -= len;
+		next_block = lblock + len;
+	}
+	if (bio) {
+		if (bio_add_page(bio, page, bio_bytes, bio_offset) < bio_bytes)
+				goto add_failed;
+		if (!bio->bi_size)
+			goto dump_bio;
+
+		submit_bio(WRITE, bio);
+	}
+	if (redirty)
+		redirty_page_for_writepage(wbc, page);
+
+	unlock_page(page);
+
+	return 0;
+add_failed:
+	/* We should never reach here, since we're only adding one vec */
+	printk(KERN_ERR "JFS: bio_add_page failed unexpectedly\n");
+	goto skip;
+dump_bio:
+	dump_mem("bio", bio, sizeof(*bio));
+skip:
+	bio_put(bio);
+	unlock_page(page);
+	dec_io(page, last_write_complete);
+
+	return -EIO;
+}
+
+static int metapage_readpage(struct file *fp, struct page *page)
+{
+	struct inode *inode = page->mapping->host;
+	struct bio *bio = NULL;
+	unsigned int block_offset;
+	unsigned int blocks_per_page = PAGE_CACHE_SIZE >> inode->i_blkbits;
+	sector_t page_start;	/* address of page in fs blocks */
+	sector_t pblock;
+	unsigned int xlen;
+	unsigned int len;
+	unsigned int offset;
+
+	BUG_ON(!PageLocked(page));
+	page_start = (sector_t)page->index <<
+		     (PAGE_CACHE_SHIFT - inode->i_blkbits);
+
+	block_offset = 0;
+	while (block_offset < blocks_per_page) {
+		xlen = blocks_per_page - block_offset;
+		pblock = metapage_get_blocks(inode, page_start + block_offset,
+					     &xlen);
+		if (pblock) {
+			if (!PagePrivate(page))
+				insert_metapage(page, NULL);
+			inc_io(page);
+			if (bio)
+				submit_bio(READ, bio);
+
+			bio = bio_alloc(GFP_NOFS, 1);
+			bio->bi_bdev = inode->i_sb->s_bdev;
+			bio->bi_sector = pblock << (inode->i_blkbits - 9);
+			bio->bi_end_io = metapage_read_end_io;
+			bio->bi_private = page;
+			len = xlen << inode->i_blkbits;
+			offset = block_offset << inode->i_blkbits;
+			if (bio_add_page(bio, page, len, offset) < len)
+				goto add_failed;
+			block_offset += xlen;
+		} else
+			block_offset++;
+	}
+	if (bio)
+		submit_bio(READ, bio);
+	else
+		unlock_page(page);
+
+	return 0;
+
+add_failed:
+	printk(KERN_ERR "JFS: bio_add_page failed unexpectedly\n");
+	bio_put(bio);
+	dec_io(page, last_read_complete);
+	return -EIO;
+}
+
+static int metapage_releasepage(struct page *page, gfp_t gfp_mask)
+{
+	struct metapage *mp;
+	int ret = 1;
+	unsigned int offset;
+
+	for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
+		mp = page_to_mp(page, offset);
+
+		if (!mp)
+			continue;
+
+		jfs_info("metapage_releasepage: mp = 0x%p", mp);
+		if (mp->count || mp->nohomeok ||
+		    test_bit(META_dirty, &mp->flag)) {
+			jfs_info("count = %ld, nohomeok = %d", mp->count,
+				 mp->nohomeok);
+			ret = 0;
+			continue;
+		}
+		if (mp->lsn)
+			remove_from_logsync(mp);
+		remove_metapage(page, mp);
+		INCREMENT(mpStat.pagefree);
+		free_metapage(mp);
+	}
+	return ret;
+}
+
+static void metapage_invalidatepage(struct page *page, unsigned long offset)
+{
+	BUG_ON(offset);
+
+	BUG_ON(PageWriteback(page));
+
+	metapage_releasepage(page, 0);
+}
+
+const struct address_space_operations jfs_metapage_aops = {
+	.readpage	= metapage_readpage,
+	.writepage	= metapage_writepage,
+	.sync_page	= block_sync_page,
+	.releasepage	= metapage_releasepage,
+	.invalidatepage	= metapage_invalidatepage,
+	.set_page_dirty	= __set_page_dirty_nobuffers,
+};
+
+struct metapage *__get_metapage(struct inode *inode, unsigned long lblock,
+				unsigned int size, int absolute,
+				unsigned long new)
+{
+	int l2BlocksPerPage;
+	int l2bsize;
+	struct address_space *mapping;
+	struct metapage *mp = NULL;
+	struct page *page;
+	unsigned long page_index;
+	unsigned long page_offset;
+
+	jfs_info("__get_metapage: ino = %ld, lblock = 0x%lx, abs=%d",
+		 inode->i_ino, lblock, absolute);
+
+	l2bsize = inode->i_blkbits;
+	l2BlocksPerPage = PAGE_CACHE_SHIFT - l2bsize;
+	page_index = lblock >> l2BlocksPerPage;
+	page_offset = (lblock - (page_index << l2BlocksPerPage)) << l2bsize;
+	if ((page_offset + size) > PAGE_CACHE_SIZE) {
+		jfs_err("MetaData crosses page boundary!!");
+		jfs_err("lblock = %lx, size  = %d", lblock, size);
+		dump_stack();
+		return NULL;
+	}
+	if (absolute)
+		mapping = JFS_SBI(inode->i_sb)->direct_inode->i_mapping;
+	else {
+		/*
+		 * If an nfs client tries to read an inode that is larger
+		 * than any existing inodes, we may try to read past the
+		 * end of the inode map
+		 */
+		if ((lblock << inode->i_blkbits) >= inode->i_size)
+			return NULL;
+		mapping = inode->i_mapping;
+	}
+
+	if (new && (PSIZE == PAGE_CACHE_SIZE)) {
+		page = grab_cache_page(mapping, page_index);
+		if (!page) {
+			jfs_err("grab_cache_page failed!");
+			return NULL;
+		}
+		SetPageUptodate(page);
+	} else {
+		page = read_mapping_page(mapping, page_index, NULL);
+		if (IS_ERR(page) || !PageUptodate(page)) {
+			jfs_err("read_mapping_page failed!");
+			return NULL;
+		}
+		lock_page(page);
+	}
+
+	mp = page_to_mp(page, page_offset);
+	if (mp) {
+		if (mp->logical_size != size) {
+			jfs_error(inode->i_sb,
+				  "__get_metapage: mp->logical_size != size");
+			jfs_err("logical_size = %d, size = %d",
+				mp->logical_size, size);
+			dump_stack();
+			goto unlock;
+		}
+		mp->count++;
+		lock_metapage(mp);
+		if (test_bit(META_discard, &mp->flag)) {
+			if (!new) {
+				jfs_error(inode->i_sb,
+					  "__get_metapage: using a "
+					  "discarded metapage");
+				discard_metapage(mp);
+				goto unlock;
+			}
+			clear_bit(META_discard, &mp->flag);
+		}
+	} else {
+		INCREMENT(mpStat.pagealloc);
+		mp = alloc_metapage(GFP_NOFS);
+		mp->page = page;
+		mp->flag = 0;
+		mp->xflag = COMMIT_PAGE;
+		mp->count = 1;
+		mp->nohomeok = 0;
+		mp->logical_size = size;
+		mp->data = page_address(page) + page_offset;
+		mp->index = lblock;
+		if (unlikely(insert_metapage(page, mp))) {
+			free_metapage(mp);
+			goto unlock;
+		}
+		lock_metapage(mp);
+	}
+
+	if (new) {
+		jfs_info("zeroing mp = 0x%p", mp);
+		memset(mp->data, 0, PSIZE);
+	}
+
+	unlock_page(page);
+	jfs_info("__get_metapage: returning = 0x%p data = 0x%p", mp, mp->data);
+	return mp;
+
+unlock:
+	unlock_page(page);
+	return NULL;
+}
+
+void grab_metapage(struct metapage * mp)
+{
+	jfs_info("grab_metapage: mp = 0x%p", mp);
+	page_cache_get(mp->page);
+	lock_page(mp->page);
+	mp->count++;
+	lock_metapage(mp);
+	unlock_page(mp->page);
+}
+
+void force_metapage(struct metapage *mp)
+{
+	struct page *page = mp->page;
+	jfs_info("force_metapage: mp = 0x%p", mp);
+	set_bit(META_forcewrite, &mp->flag);
+	clear_bit(META_sync, &mp->flag);
+	page_cache_get(page);
+	lock_page(page);
+	set_page_dirty(page);
+	write_one_page(page, 1);
+	clear_bit(META_forcewrite, &mp->flag);
+	page_cache_release(page);
+}
+
+void hold_metapage(struct metapage *mp)
+{
+	lock_page(mp->page);
+}
+
+void put_metapage(struct metapage *mp)
+{
+	if (mp->count || mp->nohomeok) {
+		/* Someone else will release this */
+		unlock_page(mp->page);
+		return;
+	}
+	page_cache_get(mp->page);
+	mp->count++;
+	lock_metapage(mp);
+	unlock_page(mp->page);
+	release_metapage(mp);
+}
+
+void release_metapage(struct metapage * mp)
+{
+	struct page *page = mp->page;
+	jfs_info("release_metapage: mp = 0x%p, flag = 0x%lx", mp, mp->flag);
+
+	BUG_ON(!page);
+
+	lock_page(page);
+	unlock_metapage(mp);
+
+	assert(mp->count);
+	if (--mp->count || mp->nohomeok) {
+		unlock_page(page);
+		page_cache_release(page);
+		return;
+	}
+
+	if (test_bit(META_dirty, &mp->flag)) {
+		set_page_dirty(page);
+		if (test_bit(META_sync, &mp->flag)) {
+			clear_bit(META_sync, &mp->flag);
+			write_one_page(page, 1);
+			lock_page(page); /* write_one_page unlocks the page */
+		}
+	} else if (mp->lsn)	/* discard_metapage doesn't remove it */
+		remove_from_logsync(mp);
+
+	/* Try to keep metapages from using up too much memory */
+	drop_metapage(page, mp);
+
+	unlock_page(page);
+	page_cache_release(page);
+}
+
+void __invalidate_metapages(struct inode *ip, s64 addr, int len)
+{
+	sector_t lblock;
+	int l2BlocksPerPage = PAGE_CACHE_SHIFT - ip->i_blkbits;
+	int BlocksPerPage = 1 << l2BlocksPerPage;
+	/* All callers are interested in block device's mapping */
+	struct address_space *mapping =
+		JFS_SBI(ip->i_sb)->direct_inode->i_mapping;
+	struct metapage *mp;
+	struct page *page;
+	unsigned int offset;
+
+	/*
+	 * Mark metapages to discard.  They will eventually be
+	 * released, but should not be written.
+	 */
+	for (lblock = addr & ~(BlocksPerPage - 1); lblock < addr + len;
+	     lblock += BlocksPerPage) {
+		page = find_lock_page(mapping, lblock >> l2BlocksPerPage);
+		if (!page)
+			continue;
+		for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
+			mp = page_to_mp(page, offset);
+			if (!mp)
+				continue;
+			if (mp->index < addr)
+				continue;
+			if (mp->index >= addr + len)
+				break;
+
+			clear_bit(META_dirty, &mp->flag);
+			set_bit(META_discard, &mp->flag);
+			if (mp->lsn)
+				remove_from_logsync(mp);
+		}
+		unlock_page(page);
+		page_cache_release(page);
+	}
+}
+
+#ifdef CONFIG_JFS_STATISTICS
+int jfs_mpstat_read(char *buffer, char **start, off_t offset, int length,
+		    int *eof, void *data)
+{
+	int len = 0;
+	off_t begin;
+
+	len += sprintf(buffer,
+		       "JFS Metapage statistics\n"
+		       "=======================\n"
+		       "page allocations = %d\n"
+		       "page frees = %d\n"
+		       "lock waits = %d\n",
+		       mpStat.pagealloc,
+		       mpStat.pagefree,
+		       mpStat.lockwait);
+
+	begin = offset;
+	*start = buffer + begin;
+	len -= begin;
+
+	if (len > length)
+		len = length;
+	else
+		*eof = 1;
+
+	if (len < 0)
+		len = 0;
+
+	return len;
+}
+#endif

fs/jfs/jfs_metapage.h

@@ -0,0 +1,155 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_METAPAGE
+#define _H_JFS_METAPAGE
+
+#include <linux/pagemap.h>
+
+struct metapage {
+	/* Common logsyncblk prefix (see jfs_logmgr.h) */
+	u16 xflag;
+	u16 unused;
+	lid_t lid;
+	int lsn;
+	struct list_head synclist;
+	/* End of logsyncblk prefix */
+
+	unsigned long flag;	/* See Below */
+	unsigned long count;	/* Reference count */
+	void *data;		/* Data pointer */
+	sector_t index;		/* block address of page */
+	wait_queue_head_t wait;
+
+	/* implementation */
+	struct page *page;
+	unsigned int logical_size;
+
+	/* Journal management */
+	int clsn;
+	int nohomeok;
+	struct jfs_log *log;
+};
+
+/* metapage flag */
+#define META_locked	0
+#define META_free	1
+#define META_dirty	2
+#define META_sync	3
+#define META_discard	4
+#define META_forcewrite	5
+#define META_io		6
+
+#define mark_metapage_dirty(mp) set_bit(META_dirty, &(mp)->flag)
+
+/* function prototypes */
+extern int metapage_init(void);
+extern void metapage_exit(void);
+extern struct metapage *__get_metapage(struct inode *inode,
+				  unsigned long lblock, unsigned int size,
+				  int absolute, unsigned long new);
+
+#define read_metapage(inode, lblock, size, absolute)\
+	 __get_metapage(inode, lblock, size, absolute, false)
+
+#define get_metapage(inode, lblock, size, absolute)\
+	 __get_metapage(inode, lblock, size, absolute, true)
+
+extern void release_metapage(struct metapage *);
+extern void grab_metapage(struct metapage *);
+extern void force_metapage(struct metapage *);
+
+/*
+ * hold_metapage and put_metapage are used in conjuction.  The page lock
+ * is not dropped between the two, so no other threads can get or release
+ * the metapage
+ */
+extern void hold_metapage(struct metapage *);
+extern void put_metapage(struct metapage *);
+
+static inline void write_metapage(struct metapage *mp)
+{
+	set_bit(META_dirty, &mp->flag);
+	release_metapage(mp);
+}
+
+static inline void flush_metapage(struct metapage *mp)
+{
+	set_bit(META_sync, &mp->flag);
+	write_metapage(mp);
+}
+
+static inline void discard_metapage(struct metapage *mp)
+{
+	clear_bit(META_dirty, &mp->flag);
+	set_bit(META_discard, &mp->flag);
+	release_metapage(mp);
+}
+
+static inline void metapage_nohomeok(struct metapage *mp)
+{
+	struct page *page = mp->page;
+	lock_page(page);
+	if (!mp->nohomeok++) {
+		mark_metapage_dirty(mp);
+		page_cache_get(page);
+		wait_on_page_writeback(page);
+	}
+	unlock_page(page);
+}
+
+/*
+ * This serializes access to mp->lsn when metapages are added to logsynclist
+ * without setting nohomeok.  i.e. updating imap & dmap
+ */
+static inline void metapage_wait_for_io(struct metapage *mp)
+{
+	if (test_bit(META_io, &mp->flag))
+		wait_on_page_writeback(mp->page);
+}
+
+/*
+ * This is called when already holding the metapage
+ */
+static inline void _metapage_homeok(struct metapage *mp)
+{
+	if (!--mp->nohomeok)
+		page_cache_release(mp->page);
+}
+
+static inline void metapage_homeok(struct metapage *mp)
+{
+	hold_metapage(mp);
+	_metapage_homeok(mp);
+	put_metapage(mp);
+}
+
+extern const struct address_space_operations jfs_metapage_aops;
+
+/*
+ * This routines invalidate all pages for an extent.
+ */
+extern void __invalidate_metapages(struct inode *, s64, int);
+#define invalidate_pxd_metapages(ip, pxd) \
+	__invalidate_metapages((ip), addressPXD(&(pxd)), lengthPXD(&(pxd)))
+#define invalidate_dxd_metapages(ip, dxd) \
+	__invalidate_metapages((ip), addressDXD(&(dxd)), lengthDXD(&(dxd)))
+#define invalidate_xad_metapages(ip, xad) \
+	__invalidate_metapages((ip), addressXAD(&(xad)), lengthXAD(&(xad)))
+
+#endif				/* _H_JFS_METAPAGE */

fs/jfs/jfs_mount.c

@@ -0,0 +1,507 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+/*
+ * Module: jfs_mount.c
+ *
+ * note: file system in transition to aggregate/fileset:
+ *
+ * file system mount is interpreted as the mount of aggregate,
+ * if not already mounted, and mount of the single/only fileset in
+ * the aggregate;
+ *
+ * a file system/aggregate is represented by an internal inode
+ * (aka mount inode) initialized with aggregate superblock;
+ * each vfs represents a fileset, and points to its "fileset inode
+ * allocation map inode" (aka fileset inode):
+ * (an aggregate itself is structured recursively as a filset:
+ * an internal vfs is constructed and points to its "fileset inode
+ * allocation map inode" (aka aggregate inode) where each inode
+ * represents a fileset inode) so that inode number is mapped to
+ * on-disk inode in uniform way at both aggregate and fileset level;
+ *
+ * each vnode/inode of a fileset is linked to its vfs (to facilitate
+ * per fileset inode operations, e.g., unmount of a fileset, etc.);
+ * each inode points to the mount inode (to facilitate access to
+ * per aggregate information, e.g., block size, etc.) as well as
+ * its file set inode.
+ *
+ *   aggregate
+ *   ipmnt
+ *   mntvfs -> fileset ipimap+ -> aggregate ipbmap -> aggregate ipaimap;
+ *             fileset vfs     -> vp(1) <-> ... <-> vp(n) <->vproot;
+ */
+
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+
+#include "jfs_incore.h"
+#include "jfs_filsys.h"
+#include "jfs_superblock.h"
+#include "jfs_dmap.h"
+#include "jfs_imap.h"
+#include "jfs_metapage.h"
+#include "jfs_debug.h"
+
+
+/*
+ * forward references
+ */
+static int chkSuper(struct super_block *);
+static int logMOUNT(struct super_block *sb);
+
+/*
+ * NAME:	jfs_mount(sb)
+ *
+ * FUNCTION:	vfs_mount()
+ *
+ * PARAMETER:	sb	- super block
+ *
+ * RETURN:	-EBUSY	- device already mounted or open for write
+ *		-EBUSY	- cvrdvp already mounted;
+ *		-EBUSY	- mount table full
+ *		-ENOTDIR- cvrdvp not directory on a device mount
+ *		-ENXIO	- device open failure
+ */
+int jfs_mount(struct super_block *sb)
+{
+	int rc = 0;		/* Return code          */
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct inode *ipaimap = NULL;
+	struct inode *ipaimap2 = NULL;
+	struct inode *ipimap = NULL;
+	struct inode *ipbmap = NULL;
+
+	/*
+	 * read/validate superblock
+	 * (initialize mount inode from the superblock)
+	 */
+	if ((rc = chkSuper(sb))) {
+		goto errout20;
+	}
+
+	ipaimap = diReadSpecial(sb, AGGREGATE_I, 0);
+	if (ipaimap == NULL) {
+		jfs_err("jfs_mount: Faild to read AGGREGATE_I");
+		rc = -EIO;
+		goto errout20;
+	}
+	sbi->ipaimap = ipaimap;
+
+	jfs_info("jfs_mount: ipaimap:0x%p", ipaimap);
+
+	/*
+	 * initialize aggregate inode allocation map
+	 */
+	if ((rc = diMount(ipaimap))) {
+		jfs_err("jfs_mount: diMount(ipaimap) failed w/rc = %d", rc);
+		goto errout21;
+	}
+
+	/*
+	 * open aggregate block allocation map
+	 */
+	ipbmap = diReadSpecial(sb, BMAP_I, 0);
+	if (ipbmap == NULL) {
+		rc = -EIO;
+		goto errout22;
+	}
+
+	jfs_info("jfs_mount: ipbmap:0x%p", ipbmap);
+
+	sbi->ipbmap = ipbmap;
+
+	/*
+	 * initialize aggregate block allocation map
+	 */
+	if ((rc = dbMount(ipbmap))) {
+		jfs_err("jfs_mount: dbMount failed w/rc = %d", rc);
+		goto errout22;
+	}
+
+	/*
+	 * open the secondary aggregate inode allocation map
+	 *
+	 * This is a duplicate of the aggregate inode allocation map.
+	 *
+	 * hand craft a vfs in the same fashion as we did to read ipaimap.
+	 * By adding INOSPEREXT (32) to the inode number, we are telling
+	 * diReadSpecial that we are reading from the secondary aggregate
+	 * inode table.  This also creates a unique entry in the inode hash
+	 * table.
+	 */
+	if ((sbi->mntflag & JFS_BAD_SAIT) == 0) {
+		ipaimap2 = diReadSpecial(sb, AGGREGATE_I, 1);
+		if (ipaimap2 == 0) {
+			jfs_err("jfs_mount: Faild to read AGGREGATE_I");
+			rc = -EIO;
+			goto errout35;
+		}
+		sbi->ipaimap2 = ipaimap2;
+
+		jfs_info("jfs_mount: ipaimap2:0x%p", ipaimap2);
+
+		/*
+		 * initialize secondary aggregate inode allocation map
+		 */
+		if ((rc = diMount(ipaimap2))) {
+			jfs_err("jfs_mount: diMount(ipaimap2) failed, rc = %d",
+				rc);
+			goto errout35;
+		}
+	} else
+		/* Secondary aggregate inode table is not valid */
+		sbi->ipaimap2 = NULL;
+
+	/*
+	 *      mount (the only/single) fileset
+	 */
+	/*
+	 * open fileset inode allocation map (aka fileset inode)
+	 */
+	ipimap = diReadSpecial(sb, FILESYSTEM_I, 0);
+	if (ipimap == NULL) {
+		jfs_err("jfs_mount: Failed to read FILESYSTEM_I");
+		/* open fileset secondary inode allocation map */
+		rc = -EIO;
+		goto errout40;
+	}
+	jfs_info("jfs_mount: ipimap:0x%p", ipimap);
+
+	/* map further access of per fileset inodes by the fileset inode */
+	sbi->ipimap = ipimap;
+
+	/* initialize fileset inode allocation map */
+	if ((rc = diMount(ipimap))) {
+		jfs_err("jfs_mount: diMount failed w/rc = %d", rc);
+		goto errout41;
+	}
+
+	goto out;
+
+	/*
+	 *      unwind on error
+	 */
+      errout41:		/* close fileset inode allocation map inode */
+	diFreeSpecial(ipimap);
+
+      errout40:		/* fileset closed */
+
+	/* close secondary aggregate inode allocation map */
+	if (ipaimap2) {
+		diUnmount(ipaimap2, 1);
+		diFreeSpecial(ipaimap2);
+	}
+
+      errout35:
+
+	/* close aggregate block allocation map */
+	dbUnmount(ipbmap, 1);
+	diFreeSpecial(ipbmap);
+
+      errout22:		/* close aggregate inode allocation map */
+
+	diUnmount(ipaimap, 1);
+
+      errout21:		/* close aggregate inodes */
+	diFreeSpecial(ipaimap);
+      errout20:		/* aggregate closed */
+
+      out:
+
+	if (rc)
+		jfs_err("Mount JFS Failure: %d", rc);
+
+	return rc;
+}
+
+/*
+ * NAME:	jfs_mount_rw(sb, remount)
+ *
+ * FUNCTION:	Completes read-write mount, or remounts read-only volume
+ *		as read-write
+ */
+int jfs_mount_rw(struct super_block *sb, int remount)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	int rc;
+
+	/*
+	 * If we are re-mounting a previously read-only volume, we want to
+	 * re-read the inode and block maps, since fsck.jfs may have updated
+	 * them.
+	 */
+	if (remount) {
+		if (chkSuper(sb) || (sbi->state != FM_CLEAN))
+			return -EINVAL;
+
+		truncate_inode_pages(sbi->ipimap->i_mapping, 0);
+		truncate_inode_pages(sbi->ipbmap->i_mapping, 0);
+		diUnmount(sbi->ipimap, 1);
+		if ((rc = diMount(sbi->ipimap))) {
+			jfs_err("jfs_mount_rw: diMount failed!");
+			return rc;
+		}
+
+		dbUnmount(sbi->ipbmap, 1);
+		if ((rc = dbMount(sbi->ipbmap))) {
+			jfs_err("jfs_mount_rw: dbMount failed!");
+			return rc;
+		}
+	}
+
+	/*
+	 * open/initialize log
+	 */
+	if ((rc = lmLogOpen(sb)))
+		return rc;
+
+	/*
+	 * update file system superblock;
+	 */
+	if ((rc = updateSuper(sb, FM_MOUNT))) {
+		jfs_err("jfs_mount: updateSuper failed w/rc = %d", rc);
+		lmLogClose(sb);
+		return rc;
+	}
+
+	/*
+	 * write MOUNT log record of the file system
+	 */
+	logMOUNT(sb);
+
+	return rc;
+}
+
+/*
+ *	chkSuper()
+ *
+ * validate the superblock of the file system to be mounted and
+ * get the file system parameters.
+ *
+ * returns
+ *	0 with fragsize set if check successful
+ *	error code if not successful
+ */
+static int chkSuper(struct super_block *sb)
+{
+	int rc = 0;
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct jfs_superblock *j_sb;
+	struct buffer_head *bh;
+	int AIM_bytesize, AIT_bytesize;
+	int expected_AIM_bytesize, expected_AIT_bytesize;
+	s64 AIM_byte_addr, AIT_byte_addr, fsckwsp_addr;
+	s64 byte_addr_diff0, byte_addr_diff1;
+	s32 bsize;
+
+	if ((rc = readSuper(sb, &bh)))
+		return rc;
+	j_sb = (struct jfs_superblock *)bh->b_data;
+
+	/*
+	 * validate superblock
+	 */
+	/* validate fs signature */
+	if (strncmp(j_sb->s_magic, JFS_MAGIC, 4) ||
+	    le32_to_cpu(j_sb->s_version) > JFS_VERSION) {
+		rc = -EINVAL;
+		goto out;
+	}
+
+	bsize = le32_to_cpu(j_sb->s_bsize);
+#ifdef _JFS_4K
+	if (bsize != PSIZE) {
+		jfs_err("Currently only 4K block size supported!");
+		rc = -EINVAL;
+		goto out;
+	}
+#endif				/* _JFS_4K */
+
+	jfs_info("superblock: flag:0x%08x state:0x%08x size:0x%Lx",
+		 le32_to_cpu(j_sb->s_flag), le32_to_cpu(j_sb->s_state),
+		 (unsigned long long) le64_to_cpu(j_sb->s_size));
+
+	/* validate the descriptors for Secondary AIM and AIT */
+	if ((j_sb->s_flag & cpu_to_le32(JFS_BAD_SAIT)) !=
+	    cpu_to_le32(JFS_BAD_SAIT)) {
+		expected_AIM_bytesize = 2 * PSIZE;
+		AIM_bytesize = lengthPXD(&(j_sb->s_aim2)) * bsize;
+		expected_AIT_bytesize = 4 * PSIZE;
+		AIT_bytesize = lengthPXD(&(j_sb->s_ait2)) * bsize;
+		AIM_byte_addr = addressPXD(&(j_sb->s_aim2)) * bsize;
+		AIT_byte_addr = addressPXD(&(j_sb->s_ait2)) * bsize;
+		byte_addr_diff0 = AIT_byte_addr - AIM_byte_addr;
+		fsckwsp_addr = addressPXD(&(j_sb->s_fsckpxd)) * bsize;
+		byte_addr_diff1 = fsckwsp_addr - AIT_byte_addr;
+		if ((AIM_bytesize != expected_AIM_bytesize) ||
+		    (AIT_bytesize != expected_AIT_bytesize) ||
+		    (byte_addr_diff0 != AIM_bytesize) ||
+		    (byte_addr_diff1 <= AIT_bytesize))
+			j_sb->s_flag |= cpu_to_le32(JFS_BAD_SAIT);
+	}
+
+	if ((j_sb->s_flag & cpu_to_le32(JFS_GROUPCOMMIT)) !=
+	    cpu_to_le32(JFS_GROUPCOMMIT))
+		j_sb->s_flag |= cpu_to_le32(JFS_GROUPCOMMIT);
+
+	/* validate fs state */
+	if (j_sb->s_state != cpu_to_le32(FM_CLEAN) &&
+	    !(sb->s_flags & MS_RDONLY)) {
+		jfs_err("jfs_mount: Mount Failure: File System Dirty.");
+		rc = -EINVAL;
+		goto out;
+	}
+
+	sbi->state = le32_to_cpu(j_sb->s_state);
+	sbi->mntflag = le32_to_cpu(j_sb->s_flag);
+
+	/*
+	 * JFS always does I/O by 4K pages.  Don't tell the buffer cache
+	 * that we use anything else (leave s_blocksize alone).
+	 */
+	sbi->bsize = bsize;
+	sbi->l2bsize = le16_to_cpu(j_sb->s_l2bsize);
+
+	/*
+	 * For now, ignore s_pbsize, l2bfactor.  All I/O going through buffer
+	 * cache.
+	 */
+	sbi->nbperpage = PSIZE >> sbi->l2bsize;
+	sbi->l2nbperpage = L2PSIZE - sbi->l2bsize;
+	sbi->l2niperblk = sbi->l2bsize - L2DISIZE;
+	if (sbi->mntflag & JFS_INLINELOG)
+		sbi->logpxd = j_sb->s_logpxd;
+	else {
+		sbi->logdev = new_decode_dev(le32_to_cpu(j_sb->s_logdev));
+		memcpy(sbi->uuid, j_sb->s_uuid, sizeof(sbi->uuid));
+		memcpy(sbi->loguuid, j_sb->s_loguuid, sizeof(sbi->uuid));
+	}
+	sbi->fsckpxd = j_sb->s_fsckpxd;
+	sbi->ait2 = j_sb->s_ait2;
+
+      out:
+	brelse(bh);
+	return rc;
+}
+
+
+/*
+ *	updateSuper()
+ *
+ * update synchronously superblock if it is mounted read-write.
+ */
+int updateSuper(struct super_block *sb, uint state)
+{
+	struct jfs_superblock *j_sb;
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct buffer_head *bh;
+	int rc;
+
+	if (sbi->flag & JFS_NOINTEGRITY) {
+		if (state == FM_DIRTY) {
+			sbi->p_state = state;
+			return 0;
+		} else if (state == FM_MOUNT) {
+			sbi->p_state = sbi->state;
+			state = FM_DIRTY;
+		} else if (state == FM_CLEAN) {
+			state = sbi->p_state;
+		} else
+			jfs_err("updateSuper: bad state");
+	} else if (sbi->state == FM_DIRTY)
+		return 0;
+
+	if ((rc = readSuper(sb, &bh)))
+		return rc;
+
+	j_sb = (struct jfs_superblock *)bh->b_data;
+
+	j_sb->s_state = cpu_to_le32(state);
+	sbi->state = state;
+
+	if (state == FM_MOUNT) {
+		/* record log's dev_t and mount serial number */
+		j_sb->s_logdev = cpu_to_le32(new_encode_dev(sbi->log->bdev->bd_dev));
+		j_sb->s_logserial = cpu_to_le32(sbi->log->serial);
+	} else if (state == FM_CLEAN) {
+		/*
+		 * If this volume is shared with OS/2, OS/2 will need to
+		 * recalculate DASD usage, since we don't deal with it.
+		 */
+		if (j_sb->s_flag & cpu_to_le32(JFS_DASD_ENABLED))
+			j_sb->s_flag |= cpu_to_le32(JFS_DASD_PRIME);
+	}
+
+	mark_buffer_dirty(bh);
+	sync_dirty_buffer(bh);
+	brelse(bh);
+
+	return 0;
+}
+
+
+/*
+ *	readSuper()
+ *
+ * read superblock by raw sector address
+ */
+int readSuper(struct super_block *sb, struct buffer_head **bpp)
+{
+	/* read in primary superblock */
+	*bpp = sb_bread(sb, SUPER1_OFF >> sb->s_blocksize_bits);
+	if (*bpp)
+		return 0;
+
+	/* read in secondary/replicated superblock */
+	*bpp = sb_bread(sb, SUPER2_OFF >> sb->s_blocksize_bits);
+	if (*bpp)
+		return 0;
+
+	return -EIO;
+}
+
+
+/*
+ *	logMOUNT()
+ *
+ * function: write a MOUNT log record for file system.
+ *
+ * MOUNT record keeps logredo() from processing log records
+ * for this file system past this point in log.
+ * it is harmless if mount fails.
+ *
+ * note: MOUNT record is at aggregate level, not at fileset level,
+ * since log records of previous mounts of a fileset
+ * (e.g., AFTER record of extent allocation) have to be processed
+ * to update block allocation map at aggregate level.
+ */
+static int logMOUNT(struct super_block *sb)
+{
+	struct jfs_log *log = JFS_SBI(sb)->log;
+	struct lrd lrd;
+
+	lrd.logtid = 0;
+	lrd.backchain = 0;
+	lrd.type = cpu_to_le16(LOG_MOUNT);
+	lrd.length = 0;
+	lrd.aggregate = cpu_to_le32(new_encode_dev(sb->s_bdev->bd_dev));
+	lmLog(log, NULL, &lrd, NULL);
+
+	return 0;
+}

fs/jfs/jfs_superblock.h

@@ -0,0 +1,121 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2003
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef	_H_JFS_SUPERBLOCK
+#define _H_JFS_SUPERBLOCK
+
+/*
+ * make the magic number something a human could read
+ */
+#define JFS_MAGIC	"JFS1"	/* Magic word */
+
+#define JFS_VERSION	2	/* Version number: Version 2 */
+
+#define LV_NAME_SIZE	11	/* MUST BE 11 for OS/2 boot sector */
+
+/*
+ *	aggregate superblock
+ *
+ * The name superblock is too close to super_block, so the name has been
+ * changed to jfs_superblock.  The utilities are still using the old name.
+ */
+struct jfs_superblock {
+	char s_magic[4];	/* 4: magic number */
+	__le32 s_version;	/* 4: version number */
+
+	__le64 s_size;		/* 8: aggregate size in hardware/LVM blocks;
+				 * VFS: number of blocks
+				 */
+	__le32 s_bsize;		/* 4: aggregate block size in bytes;
+				 * VFS: fragment size
+				 */
+	__le16 s_l2bsize;	/* 2: log2 of s_bsize */
+	__le16 s_l2bfactor;	/* 2: log2(s_bsize/hardware block size) */
+	__le32 s_pbsize;	/* 4: hardware/LVM block size in bytes */
+	__le16 s_l2pbsize;	/* 2: log2 of s_pbsize */
+	__le16 pad;		/* 2: padding necessary for alignment */
+
+	__le32 s_agsize;	/* 4: allocation group size in aggr. blocks */
+
+	__le32 s_flag;		/* 4: aggregate attributes:
+				 *    see jfs_filsys.h
+				 */
+	__le32 s_state;		/* 4: mount/unmount/recovery state:
+				 *    see jfs_filsys.h
+				 */
+	__le32 s_compress;		/* 4: > 0 if data compression */
+
+	pxd_t s_ait2;		/* 8: first extent of secondary
+				 *    aggregate inode table
+				 */
+
+	pxd_t s_aim2;		/* 8: first extent of secondary
+				 *    aggregate inode map
+				 */
+	__le32 s_logdev;		/* 4: device address of log */
+	__le32 s_logserial;	/* 4: log serial number at aggregate mount */
+	pxd_t s_logpxd;		/* 8: inline log extent */
+
+	pxd_t s_fsckpxd;	/* 8: inline fsck work space extent */
+
+	struct timestruc_t s_time;	/* 8: time last updated */
+
+	__le32 s_fsckloglen;	/* 4: Number of filesystem blocks reserved for
+				 *    the fsck service log.
+				 *    N.B. These blocks are divided among the
+				 *         versions kept.  This is not a per
+				 *         version size.
+				 *    N.B. These blocks are included in the
+				 *         length field of s_fsckpxd.
+				 */
+	s8 s_fscklog;		/* 1: which fsck service log is most recent
+				 *    0 => no service log data yet
+				 *    1 => the first one
+				 *    2 => the 2nd one
+				 */
+	char s_fpack[11];	/* 11: file system volume name
+				 *     N.B. This must be 11 bytes to
+				 *          conform with the OS/2 BootSector
+				 *          requirements
+				 *          Only used when s_version is 1
+				 */
+
+	/* extendfs() parameter under s_state & FM_EXTENDFS */
+	__le64 s_xsize;		/* 8: extendfs s_size */
+	pxd_t s_xfsckpxd;	/* 8: extendfs fsckpxd */
+	pxd_t s_xlogpxd;	/* 8: extendfs logpxd */
+	/* - 128 byte boundary - */
+
+	char s_uuid[16];	/* 16: 128-bit uuid for volume */
+	char s_label[16];	/* 16: volume label */
+	char s_loguuid[16];	/* 16: 128-bit uuid for log device */
+
+};
+
+extern int readSuper(struct super_block *, struct buffer_head **);
+extern int updateSuper(struct super_block *, uint);
+extern void jfs_error(struct super_block *, const char *, ...);
+extern int jfs_mount(struct super_block *);
+extern int jfs_mount_rw(struct super_block *, int);
+extern int jfs_umount(struct super_block *);
+extern int jfs_umount_rw(struct super_block *);
+extern int jfs_extendfs(struct super_block *, s64, int);
+
+extern struct task_struct *jfsIOthread;
+extern struct task_struct *jfsSyncThread;
+
+#endif /*_H_JFS_SUPERBLOCK */

fs/jfs/jfs_txnmgr.h

@@ -0,0 +1,311 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_TXNMGR
+#define _H_JFS_TXNMGR
+
+#include "jfs_logmgr.h"
+
+/*
+ * Hide implementation of TxBlock and TxLock
+ */
+#define tid_to_tblock(tid) (&TxBlock[tid])
+
+#define lid_to_tlock(lid) (&TxLock[lid])
+
+/*
+ *	transaction block
+ */
+struct tblock {
+	/*
+	 * tblock and jbuf_t common area: struct logsyncblk
+	 *
+	 * the following 5 fields are the same as struct logsyncblk
+	 * which is common to tblock and jbuf to form logsynclist
+	 */
+	u16 xflag;		/* tx commit type */
+	u16 flag;		/* tx commit state */
+	lid_t dummy;		/* Must keep structures common */
+	s32 lsn;		/* recovery lsn */
+	struct list_head synclist;	/* logsynclist link */
+
+	/* lock management */
+	struct super_block *sb;	/* super block */
+	lid_t next;		/* index of first tlock of tid */
+	lid_t last;		/* index of last tlock of tid */
+	wait_queue_head_t waitor;	/* tids waiting on this tid */
+
+	/* log management */
+	u32 logtid;		/* log transaction id */
+
+	/* commit management */
+	struct list_head cqueue;	/* commit queue list */
+	s32 clsn;		/* commit lsn */
+	struct lbuf *bp;
+	s32 pn;			/* commit record log page number */
+	s32 eor;		/* commit record eor */
+	wait_queue_head_t gcwait;	/* group commit event list:
+					 * ready transactions wait on this
+					 * event for group commit completion.
+					 */
+	union {
+		struct inode *ip; /* inode being deleted */
+		pxd_t ixpxd;	/* pxd of inode extent for created inode */
+	} u;
+	u32 ino;		/* inode number being created */
+};
+
+extern struct tblock *TxBlock;	/* transaction block table */
+
+/* commit flags: tblk->xflag */
+#define	COMMIT_SYNC	0x0001	/* synchronous commit */
+#define	COMMIT_FORCE	0x0002	/* force pageout at end of commit */
+#define	COMMIT_FLUSH	0x0004	/* init flush at end of commit */
+#define COMMIT_MAP	0x00f0
+#define	COMMIT_PMAP	0x0010	/* update pmap */
+#define	COMMIT_WMAP	0x0020	/* update wmap */
+#define	COMMIT_PWMAP	0x0040	/* update pwmap */
+#define	COMMIT_FREE	0x0f00
+#define	COMMIT_DELETE	0x0100	/* inode delete */
+#define	COMMIT_TRUNCATE	0x0200	/* file truncation */
+#define	COMMIT_CREATE	0x0400	/* inode create */
+#define	COMMIT_LAZY	0x0800	/* lazy commit */
+#define COMMIT_PAGE	0x1000	/* Identifies element as metapage */
+#define COMMIT_INODE	0x2000	/* Identifies element as inode */
+
+/* group commit flags tblk->flag: see jfs_logmgr.h */
+
+/*
+ *	transaction lock
+ */
+struct tlock {
+	lid_t next;		/* 2: index next lockword on tid locklist
+				 *          next lockword on freelist
+				 */
+	tid_t tid;		/* 2: transaction id holding lock */
+
+	u16 flag;		/* 2: lock control */
+	u16 type;		/* 2: log type */
+
+	struct metapage *mp;	/* 4/8: object page buffer locked */
+	struct inode *ip;	/* 4/8: object */
+	/* (16) */
+
+	s16 lock[24];		/* 48: overlay area */
+};				/* (64) */
+
+extern struct tlock *TxLock;	/* transaction lock table */
+
+/*
+ * tlock flag
+ */
+/* txLock state */
+#define tlckPAGELOCK		0x8000
+#define tlckINODELOCK		0x4000
+#define tlckLINELOCK		0x2000
+#define tlckINLINELOCK		0x1000
+/* lmLog state */
+#define tlckLOG			0x0800
+/* updateMap state */
+#define	tlckUPDATEMAP		0x0080
+#define	tlckDIRECTORY		0x0040
+/* freeLock state */
+#define tlckFREELOCK		0x0008
+#define tlckWRITEPAGE		0x0004
+#define tlckFREEPAGE		0x0002
+
+/*
+ * tlock type
+ */
+#define	tlckTYPE		0xfe00
+#define	tlckINODE		0x8000
+#define	tlckXTREE		0x4000
+#define	tlckDTREE		0x2000
+#define	tlckMAP			0x1000
+#define	tlckEA			0x0800
+#define	tlckACL			0x0400
+#define	tlckDATA		0x0200
+#define	tlckBTROOT		0x0100
+
+#define	tlckOPERATION		0x00ff
+#define tlckGROW		0x0001	/* file grow */
+#define tlckREMOVE		0x0002	/* file delete */
+#define tlckTRUNCATE		0x0004	/* file truncate */
+#define tlckRELOCATE		0x0008	/* file/directory relocate */
+#define tlckENTRY		0x0001	/* directory insert/delete */
+#define tlckEXTEND		0x0002	/* directory extend in-line */
+#define tlckSPLIT		0x0010	/* splited page */
+#define tlckNEW			0x0020	/* new page from split */
+#define tlckFREE		0x0040	/* free page */
+#define tlckRELINK		0x0080	/* update sibling pointer */
+
+/*
+ *	linelock for lmLog()
+ *
+ * note: linelock and its variations are overlaid
+ * at tlock.lock: watch for alignment;
+ */
+struct lv {
+	u8 offset;		/* 1: */
+	u8 length;		/* 1: */
+};				/* (2) */
+
+#define	TLOCKSHORT	20
+#define	TLOCKLONG	28
+
+struct linelock {
+	lid_t next;		/* 2: next linelock */
+
+	s8 maxcnt;		/* 1: */
+	s8 index;		/* 1: */
+
+	u16 flag;		/* 2: */
+	u8 type;		/* 1: */
+	u8 l2linesize;		/* 1: log2 of linesize */
+	/* (8) */
+
+	struct lv lv[20];	/* 40: */
+};				/* (48) */
+
+#define dt_lock	linelock
+
+struct xtlock {
+	lid_t next;		/* 2: */
+
+	s8 maxcnt;		/* 1: */
+	s8 index;		/* 1: */
+
+	u16 flag;		/* 2: */
+	u8 type;		/* 1: */
+	u8 l2linesize;		/* 1: log2 of linesize */
+				/* (8) */
+
+	struct lv header;	/* 2: */
+	struct lv lwm;		/* 2: low water mark */
+	struct lv hwm;		/* 2: high water mark */
+	struct lv twm;		/* 2: */
+				/* (16) */
+
+	s32 pxdlock[8];		/* 32: */
+};				/* (48) */
+
+
+/*
+ *	maplock for txUpdateMap()
+ *
+ * note: maplock and its variations are overlaid
+ * at tlock.lock/linelock: watch for alignment;
+ * N.B. next field may be set by linelock, and should not
+ * be modified by maplock;
+ * N.B. index of the first pxdlock specifies index of next
+ * free maplock (i.e., number of maplock) in the tlock;
+ */
+struct maplock {
+	lid_t next;		/* 2: */
+
+	u8 maxcnt;		/* 2: */
+	u8 index;		/* 2: next free maplock index */
+
+	u16 flag;		/* 2: */
+	u8 type;		/* 1: */
+	u8 count;		/* 1: number of pxd/xad */
+				/* (8) */
+
+	pxd_t pxd;		/* 8: */
+};				/* (16): */
+
+/* maplock flag */
+#define	mlckALLOC		0x00f0
+#define	mlckALLOCXADLIST	0x0080
+#define	mlckALLOCPXDLIST	0x0040
+#define	mlckALLOCXAD		0x0020
+#define	mlckALLOCPXD		0x0010
+#define	mlckFREE		0x000f
+#define	mlckFREEXADLIST		0x0008
+#define	mlckFREEPXDLIST		0x0004
+#define	mlckFREEXAD		0x0002
+#define	mlckFREEPXD		0x0001
+
+#define	pxd_lock	maplock
+
+struct xdlistlock {
+	lid_t next;		/* 2: */
+
+	u8 maxcnt;		/* 2: */
+	u8 index;		/* 2: */
+
+	u16 flag;		/* 2: */
+	u8 type;		/* 1: */
+	u8 count;		/* 1: number of pxd/xad */
+				/* (8) */
+
+	/*
+	 * We need xdlist to be 64 bits (8 bytes), regardless of
+	 * whether void * is 32 or 64 bits
+	 */
+	union {
+		void *_xdlist;	/* pxd/xad list */
+		s64 pad;	/* 8: Force 64-bit xdlist size */
+	} union64;
+};				/* (16): */
+
+#define xdlist union64._xdlist
+
+/*
+ *	commit
+ *
+ * parameter to the commit manager routines
+ */
+struct commit {
+	tid_t tid;		/* tid = index of tblock */
+	int flag;		/* flags */
+	struct jfs_log *log;	/* log */
+	struct super_block *sb;	/* superblock */
+
+	int nip;		/* number of entries in iplist */
+	struct inode **iplist;	/* list of pointers to inodes */
+
+	/* log record descriptor on 64-bit boundary */
+	struct lrd lrd;		/* : log record descriptor */
+};
+
+/*
+ * external declarations
+ */
+extern int jfs_tlocks_low;
+
+extern int txInit(void);
+extern void txExit(void);
+extern struct tlock *txLock(tid_t, struct inode *, struct metapage *, int);
+extern struct tlock *txMaplock(tid_t, struct inode *, int);
+extern int txCommit(tid_t, int, struct inode **, int);
+extern tid_t txBegin(struct super_block *, int);
+extern void txBeginAnon(struct super_block *);
+extern void txEnd(tid_t);
+extern void txAbort(tid_t, int);
+extern struct linelock *txLinelock(struct linelock *);
+extern void txFreeMap(struct inode *, struct maplock *, struct tblock *, int);
+extern void txEA(tid_t, struct inode *, dxd_t *, dxd_t *);
+extern void txFreelock(struct inode *);
+extern int lmLog(struct jfs_log *, struct tblock *, struct lrd *,
+		 struct tlock *);
+extern void txQuiesce(struct super_block *);
+extern void txResume(struct super_block *);
+extern void txLazyUnlock(struct tblock *);
+extern int jfs_lazycommit(void *);
+extern int jfs_sync(void *);
+#endif				/* _H_JFS_TXNMGR */

fs/jfs/jfs_types.h

@@ -0,0 +1,188 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_TYPES
+#define	_H_JFS_TYPES
+
+/*
+ *	jfs_types.h:
+ *
+ * basic type/utility  definitions
+ *
+ * note: this header file must be the 1st include file
+ * of JFS include list in all JFS .c file.
+ */
+
+#include <linux/types.h>
+#include <linux/nls.h>
+
+#include "endian24.h"
+
+/*
+ * transaction and lock id's
+ *
+ * Don't change these without carefully considering the impact on the
+ * size and alignment of all of the linelock variants
+ */
+typedef u16 tid_t;
+typedef u16 lid_t;
+
+/*
+ * Almost identical to Linux's timespec, but not quite
+ */
+struct timestruc_t {
+	__le32 tv_sec;
+	__le32 tv_nsec;
+};
+
+/*
+ *	handy
+ */
+
+#define LEFTMOSTONE	0x80000000
+#define	HIGHORDER	0x80000000u	/* high order bit on            */
+#define	ONES		0xffffffffu	/* all bit on                   */
+
+/*
+ *	logical xd (lxd)
+ */
+typedef struct {
+	unsigned len:24;
+	unsigned off1:8;
+	u32 off2;
+} lxd_t;
+
+/* lxd_t field construction */
+#define	LXDlength(lxd, length32)	( (lxd)->len = length32 )
+#define	LXDoffset(lxd, offset64)\
+{\
+	(lxd)->off1 = ((s64)offset64) >> 32;\
+	(lxd)->off2 = (offset64) & 0xffffffff;\
+}
+
+/* lxd_t field extraction */
+#define	lengthLXD(lxd)	( (lxd)->len )
+#define	offsetLXD(lxd)\
+	( ((s64)((lxd)->off1)) << 32 | (lxd)->off2 )
+
+/* lxd list */
+struct lxdlist {
+	s16 maxnlxd;
+	s16 nlxd;
+	lxd_t *lxd;
+};
+
+/*
+ *	physical xd (pxd)
+ */
+typedef struct {
+	unsigned len:24;
+	unsigned addr1:8;
+	__le32 addr2;
+} pxd_t;
+
+/* xd_t field construction */
+
+#define	PXDlength(pxd, length32)	((pxd)->len = __cpu_to_le24(length32))
+#define	PXDaddress(pxd, address64)\
+{\
+	(pxd)->addr1 = ((s64)address64) >> 32;\
+	(pxd)->addr2 = __cpu_to_le32((address64) & 0xffffffff);\
+}
+
+/* xd_t field extraction */
+#define	lengthPXD(pxd)	__le24_to_cpu((pxd)->len)
+#define	addressPXD(pxd)\
+	( ((s64)((pxd)->addr1)) << 32 | __le32_to_cpu((pxd)->addr2))
+
+#define MAXTREEHEIGHT 8
+/* pxd list */
+struct pxdlist {
+	s16 maxnpxd;
+	s16 npxd;
+	pxd_t pxd[MAXTREEHEIGHT];
+};
+
+
+/*
+ *	data extent descriptor (dxd)
+ */
+typedef struct {
+	unsigned flag:8;	/* 1: flags */
+	unsigned rsrvd:24;
+	__le32 size;		/* 4: size in byte */
+	unsigned len:24;	/* 3: length in unit of fsblksize */
+	unsigned addr1:8;	/* 1: address in unit of fsblksize */
+	__le32 addr2;		/* 4: address in unit of fsblksize */
+} dxd_t;			/* - 16 - */
+
+/* dxd_t flags */
+#define	DXD_INDEX	0x80	/* B+-tree index */
+#define	DXD_INLINE	0x40	/* in-line data extent */
+#define	DXD_EXTENT	0x20	/* out-of-line single extent */
+#define	DXD_FILE	0x10	/* out-of-line file (inode) */
+#define DXD_CORRUPT	0x08	/* Inconsistency detected */
+
+/* dxd_t field construction
+ *	Conveniently, the PXD macros work for DXD
+ */
+#define	DXDlength	PXDlength
+#define	DXDaddress	PXDaddress
+#define	lengthDXD	lengthPXD
+#define	addressDXD	addressPXD
+#define DXDsize(dxd, size32) ((dxd)->size = cpu_to_le32(size32))
+#define sizeDXD(dxd)	le32_to_cpu((dxd)->size)
+
+/*
+ *      directory entry argument
+ */
+struct component_name {
+	int namlen;
+	wchar_t *name;
+};
+
+
+/*
+ *	DASD limit information - stored in directory inode
+ */
+struct dasd {
+	u8 thresh;		/* Alert Threshold (in percent) */
+	u8 delta;		/* Alert Threshold delta (in percent)   */
+	u8 rsrvd1;
+	u8 limit_hi;		/* DASD limit (in logical blocks)       */
+	__le32 limit_lo;	/* DASD limit (in logical blocks)       */
+	u8 rsrvd2[3];
+	u8 used_hi;		/* DASD usage (in logical blocks)       */
+	__le32 used_lo;		/* DASD usage (in logical blocks)       */
+};
+
+#define DASDLIMIT(dasdp) \
+	(((u64)((dasdp)->limit_hi) << 32) + __le32_to_cpu((dasdp)->limit_lo))
+#define setDASDLIMIT(dasdp, limit)\
+{\
+	(dasdp)->limit_hi = ((u64)limit) >> 32;\
+	(dasdp)->limit_lo = __cpu_to_le32(limit);\
+}
+#define DASDUSED(dasdp) \
+	(((u64)((dasdp)->used_hi) << 32) + __le32_to_cpu((dasdp)->used_lo))
+#define setDASDUSED(dasdp, used)\
+{\
+	(dasdp)->used_hi = ((u64)used) >> 32;\
+	(dasdp)->used_lo = __cpu_to_le32(used);\
+}
+
+#endif				/* !_H_JFS_TYPES */

fs/jfs/jfs_umount.c

@@ -0,0 +1,168 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+/*
+ *	jfs_umount.c
+ *
+ * note: file system in transition to aggregate/fileset:
+ * (ref. jfs_mount.c)
+ *
+ * file system unmount is interpreted as mount of the single/only
+ * fileset in the aggregate and, if unmount of the last fileset,
+ * as unmount of the aggerate;
+ */
+
+#include <linux/fs.h>
+#include "jfs_incore.h"
+#include "jfs_filsys.h"
+#include "jfs_superblock.h"
+#include "jfs_dmap.h"
+#include "jfs_imap.h"
+#include "jfs_metapage.h"
+#include "jfs_debug.h"
+
+/*
+ * NAME:	jfs_umount(vfsp, flags, crp)
+ *
+ * FUNCTION:	vfs_umount()
+ *
+ * PARAMETERS:	vfsp	- virtual file system pointer
+ *		flags	- unmount for shutdown
+ *		crp	- credential
+ *
+ * RETURN :	EBUSY	- device has open files
+ */
+int jfs_umount(struct super_block *sb)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct inode *ipbmap = sbi->ipbmap;
+	struct inode *ipimap = sbi->ipimap;
+	struct inode *ipaimap = sbi->ipaimap;
+	struct inode *ipaimap2 = sbi->ipaimap2;
+	struct jfs_log *log;
+	int rc = 0;
+
+	jfs_info("UnMount JFS: sb:0x%p", sb);
+
+	/*
+	 *      update superblock and close log
+	 *
+	 * if mounted read-write and log based recovery was enabled
+	 */
+	if ((log = sbi->log))
+		/*
+		 * Wait for outstanding transactions to be written to log:
+		 */
+		jfs_flush_journal(log, 2);
+
+	/*
+	 * close fileset inode allocation map (aka fileset inode)
+	 */
+	diUnmount(ipimap, 0);
+
+	diFreeSpecial(ipimap);
+	sbi->ipimap = NULL;
+
+	/*
+	 * close secondary aggregate inode allocation map
+	 */
+	ipaimap2 = sbi->ipaimap2;
+	if (ipaimap2) {
+		diUnmount(ipaimap2, 0);
+		diFreeSpecial(ipaimap2);
+		sbi->ipaimap2 = NULL;
+	}
+
+	/*
+	 * close aggregate inode allocation map
+	 */
+	ipaimap = sbi->ipaimap;
+	diUnmount(ipaimap, 0);
+	diFreeSpecial(ipaimap);
+	sbi->ipaimap = NULL;
+
+	/*
+	 * close aggregate block allocation map
+	 */
+	dbUnmount(ipbmap, 0);
+
+	diFreeSpecial(ipbmap);
+	sbi->ipimap = NULL;
+
+	/*
+	 * Make sure all metadata makes it to disk before we mark
+	 * the superblock as clean
+	 */
+	filemap_write_and_wait(sbi->direct_inode->i_mapping);
+
+	/*
+	 * ensure all file system file pages are propagated to their
+	 * home blocks on disk (and their in-memory buffer pages are
+	 * invalidated) BEFORE updating file system superblock state
+	 * (to signify file system is unmounted cleanly, and thus in
+	 * consistent state) and log superblock active file system
+	 * list (to signify skip logredo()).
+	 */
+	if (log) {		/* log = NULL if read-only mount */
+		updateSuper(sb, FM_CLEAN);
+
+		/*
+		 * close log:
+		 *
+		 * remove file system from log active file system list.
+		 */
+		rc = lmLogClose(sb);
+	}
+	jfs_info("UnMount JFS Complete: rc = %d", rc);
+	return rc;
+}
+
+
+int jfs_umount_rw(struct super_block *sb)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct jfs_log *log = sbi->log;
+
+	if (!log)
+		return 0;
+
+	/*
+	 * close log:
+	 *
+	 * remove file system from log active file system list.
+	 */
+	jfs_flush_journal(log, 2);
+
+	/*
+	 * Make sure all metadata makes it to disk
+	 */
+	dbSync(sbi->ipbmap);
+	diSync(sbi->ipimap);
+
+	/*
+	 * Note that we have to do this even if sync_blockdev() will
+	 * do exactly the same a few instructions later:  We can't
+	 * mark the superblock clean before everything is flushed to
+	 * disk.
+	 */
+	filemap_write_and_wait(sbi->direct_inode->i_mapping);
+
+	updateSuper(sb, FM_CLEAN);
+
+	return lmLogClose(sb);
+}

fs/jfs/jfs_unicode.c

@@ -0,0 +1,138 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/slab.h>
+#include "jfs_incore.h"
+#include "jfs_filsys.h"
+#include "jfs_unicode.h"
+#include "jfs_debug.h"
+
+/*
+ * NAME:	jfs_strfromUCS()
+ *
+ * FUNCTION:	Convert little-endian unicode string to character string
+ *
+ */
+int jfs_strfromUCS_le(char *to, const __le16 * from,
+		      int len, struct nls_table *codepage)
+{
+	int i;
+	int outlen = 0;
+	static int warn_again = 5;	/* Only warn up to 5 times total */
+	int warn = !!warn_again;	/* once per string */
+
+	if (codepage) {
+		for (i = 0; (i < len) && from[i]; i++) {
+			int charlen;
+			charlen =
+			    codepage->uni2char(le16_to_cpu(from[i]),
+					       &to[outlen],
+					       NLS_MAX_CHARSET_SIZE);
+			if (charlen > 0)
+				outlen += charlen;
+			else
+				to[outlen++] = '?';
+		}
+	} else {
+		for (i = 0; (i < len) && from[i]; i++) {
+			if (unlikely(le16_to_cpu(from[i]) & 0xff00)) {
+				to[i] = '?';
+				if (unlikely(warn)) {
+					warn--;
+					warn_again--;
+					printk(KERN_ERR
+			"non-latin1 character 0x%x found in JFS file name\n",
+					       le16_to_cpu(from[i]));
+					printk(KERN_ERR
+				"mount with iocharset=utf8 to access\n");
+				}
+
+			}
+			else
+				to[i] = (char) (le16_to_cpu(from[i]));
+		}
+		outlen = i;
+	}
+	to[outlen] = 0;
+	return outlen;
+}
+
+/*
+ * NAME:	jfs_strtoUCS()
+ *
+ * FUNCTION:	Convert character string to unicode string
+ *
+ */
+static int jfs_strtoUCS(wchar_t * to, const unsigned char *from, int len,
+		struct nls_table *codepage)
+{
+	int charlen;
+	int i;
+
+	if (codepage) {
+		for (i = 0; len && *from; i++, from += charlen, len -= charlen)
+		{
+			charlen = codepage->char2uni(from, len, &to[i]);
+			if (charlen < 1) {
+				jfs_err("jfs_strtoUCS: char2uni returned %d.",
+					charlen);
+				jfs_err("charset = %s, char = 0x%x",
+					codepage->charset, *from);
+				return charlen;
+			}
+		}
+	} else {
+		for (i = 0; (i < len) && from[i]; i++)
+			to[i] = (wchar_t) from[i];
+	}
+
+	to[i] = 0;
+	return i;
+}
+
+/*
+ * NAME:	get_UCSname()
+ *
+ * FUNCTION:	Allocate and translate to unicode string
+ *
+ */
+int get_UCSname(struct component_name * uniName, struct dentry *dentry)
+{
+	struct nls_table *nls_tab = JFS_SBI(dentry->d_sb)->nls_tab;
+	int length = dentry->d_name.len;
+
+	if (length > JFS_NAME_MAX)
+		return -ENAMETOOLONG;
+
+	uniName->name =
+	    kmalloc((length + 1) * sizeof(wchar_t), GFP_NOFS);
+
+	if (uniName->name == NULL)
+		return -ENOMEM;
+
+	uniName->namlen = jfs_strtoUCS(uniName->name, dentry->d_name.name,
+				       length, nls_tab);
+
+	if (uniName->namlen < 0) {
+		kfree(uniName->name);
+		return uniName->namlen;
+	}
+
+	return 0;
+}

fs/jfs/jfs_unicode.h

@@ -0,0 +1,155 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_UNICODE
+#define _H_JFS_UNICODE
+
+#include <asm/byteorder.h>
+#include "jfs_types.h"
+
+typedef struct {
+	wchar_t start;
+	wchar_t end;
+	signed char *table;
+} UNICASERANGE;
+
+extern signed char UniUpperTable[512];
+extern UNICASERANGE UniUpperRange[];
+extern int get_UCSname(struct component_name *, struct dentry *);
+extern int jfs_strfromUCS_le(char *, const __le16 *, int, struct nls_table *);
+
+#define free_UCSname(COMP) kfree((COMP)->name)
+
+/*
+ * UniStrcpy:  Copy a string
+ */
+static inline wchar_t *UniStrcpy(wchar_t * ucs1, const wchar_t * ucs2)
+{
+	wchar_t *anchor = ucs1;	/* save the start of result string */
+
+	while ((*ucs1++ = *ucs2++));
+	return anchor;
+}
+
+
+
+/*
+ * UniStrncpy:  Copy length limited string with pad
+ */
+static inline __le16 *UniStrncpy_le(__le16 * ucs1, const __le16 * ucs2,
+				  size_t n)
+{
+	__le16 *anchor = ucs1;
+
+	while (n-- && *ucs2)	/* Copy the strings */
+		*ucs1++ = *ucs2++;
+
+	n++;
+	while (n--)		/* Pad with nulls */
+		*ucs1++ = 0;
+	return anchor;
+}
+
+/*
+ * UniStrncmp_le:  Compare length limited string - native to little-endian
+ */
+static inline int UniStrncmp_le(const wchar_t * ucs1, const __le16 * ucs2,
+				size_t n)
+{
+	if (!n)
+		return 0;	/* Null strings are equal */
+	while ((*ucs1 == __le16_to_cpu(*ucs2)) && *ucs1 && --n) {
+		ucs1++;
+		ucs2++;
+	}
+	return (int) *ucs1 - (int) __le16_to_cpu(*ucs2);
+}
+
+/*
+ * UniStrncpy_to_le:  Copy length limited string with pad to little-endian
+ */
+static inline __le16 *UniStrncpy_to_le(__le16 * ucs1, const wchar_t * ucs2,
+				       size_t n)
+{
+	__le16 *anchor = ucs1;
+
+	while (n-- && *ucs2)	/* Copy the strings */
+		*ucs1++ = cpu_to_le16(*ucs2++);
+
+	n++;
+	while (n--)		/* Pad with nulls */
+		*ucs1++ = 0;
+	return anchor;
+}
+
+/*
+ * UniStrncpy_from_le:  Copy length limited string with pad from little-endian
+ */
+static inline wchar_t *UniStrncpy_from_le(wchar_t * ucs1, const __le16 * ucs2,
+					  size_t n)
+{
+	wchar_t *anchor = ucs1;
+
+	while (n-- && *ucs2)	/* Copy the strings */
+		*ucs1++ = __le16_to_cpu(*ucs2++);
+
+	n++;
+	while (n--)		/* Pad with nulls */
+		*ucs1++ = 0;
+	return anchor;
+}
+
+/*
+ * UniToupper:  Convert a unicode character to upper case
+ */
+static inline wchar_t UniToupper(wchar_t uc)
+{
+	UNICASERANGE *rp;
+
+	if (uc < sizeof(UniUpperTable)) {	/* Latin characters */
+		return uc + UniUpperTable[uc];	/* Use base tables */
+	} else {
+		rp = UniUpperRange;	/* Use range tables */
+		while (rp->start) {
+			if (uc < rp->start)	/* Before start of range */
+				return uc;	/* Uppercase = input */
+			if (uc <= rp->end)	/* In range */
+				return uc + rp->table[uc - rp->start];
+			rp++;	/* Try next range */
+		}
+	}
+	return uc;		/* Past last range */
+}
+
+
+/*
+ * UniStrupr:  Upper case a unicode string
+ */
+static inline wchar_t *UniStrupr(wchar_t * upin)
+{
+	wchar_t *up;
+
+	up = upin;
+	while (*up) {		/* For all characters */
+		*up = UniToupper(*up);
+		up++;
+	}
+	return upin;		/* Return input pointer */
+}
+
+#endif				/* !_H_JFS_UNICODE */

fs/jfs/jfs_uniupr.c

@@ -0,0 +1,134 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include "jfs_unicode.h"
+
+/*
+ * Latin upper case
+ */
+signed char UniUpperTable[512] = {
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 000-00f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 010-01f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 020-02f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 030-03f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 040-04f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 050-05f */
+   0,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32, /* 060-06f */
+ -32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,  0,  0,  0,  0,  0, /* 070-07f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 080-08f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 090-09f */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 0a0-0af */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 0b0-0bf */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 0c0-0cf */
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 0d0-0df */
+ -32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32, /* 0e0-0ef */
+ -32,-32,-32,-32,-32,-32,-32,  0,-32,-32,-32,-32,-32,-32,-32,121, /* 0f0-0ff */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 100-10f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 110-11f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 120-12f */
+   0,  0,  0, -1,  0, -1,  0, -1,  0,  0, -1,  0, -1,  0, -1,  0, /* 130-13f */
+  -1,  0, -1,  0, -1,  0, -1,  0, -1,  0,  0, -1,  0, -1,  0, -1, /* 140-14f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 150-15f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 160-16f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0,  0, -1,  0, -1,  0, -1,  0, /* 170-17f */
+   0,  0,  0, -1,  0, -1,  0,  0, -1,  0,  0,  0, -1,  0,  0,  0, /* 180-18f */
+   0,  0, -1,  0,  0,  0,  0,  0,  0, -1,  0,  0,  0,  0,  0,  0, /* 190-19f */
+   0, -1,  0, -1,  0, -1,  0,  0, -1,  0,  0,  0,  0, -1,  0,  0, /* 1a0-1af */
+  -1,  0,  0,  0, -1,  0, -1,  0,  0, -1,  0,  0,  0, -1,  0,  0, /* 1b0-1bf */
+   0,  0,  0,  0,  0, -1, -2,  0, -1, -2,  0, -1, -2,  0, -1,  0, /* 1c0-1cf */
+  -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,-79,  0, -1, /* 1d0-1df */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e0-1ef */
+   0,  0, -1, -2,  0, -1,  0,  0,  0, -1,  0, -1,  0, -1,  0, -1, /* 1f0-1ff */
+};
+
+/* Upper case range - Greek */
+static signed char UniCaseRangeU03a0[47] = {
+   0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,-38,-37,-37,-37, /* 3a0-3af */
+   0,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32, /* 3b0-3bf */
+ -32,-32,-31,-32,-32,-32,-32,-32,-32,-32,-32,-32,-64,-63,-63,
+};
+
+/* Upper case range - Cyrillic */
+static signed char UniCaseRangeU0430[48] = {
+ -32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32, /* 430-43f */
+ -32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32, /* 440-44f */
+   0,-80,-80,-80,-80,-80,-80,-80,-80,-80,-80,-80,-80,  0,-80,-80, /* 450-45f */
+};
+
+/* Upper case range - Extended cyrillic */
+static signed char UniCaseRangeU0490[61] = {
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 490-49f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 4a0-4af */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 4b0-4bf */
+   0,  0, -1,  0, -1,  0,  0,  0, -1,  0,  0,  0, -1,
+};
+
+/* Upper case range - Extended latin and greek */
+static signed char UniCaseRangeU1e00[509] = {
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e00-1e0f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e10-1e1f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e20-1e2f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e30-1e3f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e40-1e4f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e50-1e5f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e60-1e6f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e70-1e7f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1e80-1e8f */
+   0, -1,  0, -1,  0, -1,  0,  0,  0,  0,  0,-59,  0, -1,  0, -1, /* 1e90-1e9f */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1ea0-1eaf */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1eb0-1ebf */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1ec0-1ecf */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1ed0-1edf */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0, -1, /* 1ee0-1eef */
+   0, -1,  0, -1,  0, -1,  0, -1,  0, -1,  0,  0,  0,  0,  0,  0, /* 1ef0-1eff */
+   8,  8,  8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f00-1f0f */
+   8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f10-1f1f */
+   8,  8,  8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f20-1f2f */
+   8,  8,  8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f30-1f3f */
+   8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f40-1f4f */
+   0,  8,  0,  8,  0,  8,  0,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f50-1f5f */
+   8,  8,  8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f60-1f6f */
+  74, 74, 86, 86, 86, 86,100,100,  0,  0,112,112,126,126,  0,  0, /* 1f70-1f7f */
+   8,  8,  8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f80-1f8f */
+   8,  8,  8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1f90-1f9f */
+   8,  8,  8,  8,  8,  8,  8,  8,  0,  0,  0,  0,  0,  0,  0,  0, /* 1fa0-1faf */
+   8,  8,  0,  9,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 1fb0-1fbf */
+   0,  0,  0,  9,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 1fc0-1fcf */
+   8,  8,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 1fd0-1fdf */
+   8,  8,  0,  0,  0,  7,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, /* 1fe0-1fef */
+   0,  0,  0,  9,  0,  0,  0,  0,  0,  0,  0,  0,  0,
+};
+
+/* Upper case range - Wide latin */
+static signed char UniCaseRangeUff40[27] = {
+   0,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32, /* ff40-ff4f */
+ -32,-32,-32,-32,-32,-32,-32,-32,-32,-32,-32,
+};
+
+/*
+ * Upper Case Range
+ */
+UNICASERANGE UniUpperRange[] = {
+    { 0x03a0,  0x03ce,  UniCaseRangeU03a0 },
+    { 0x0430,  0x045f,  UniCaseRangeU0430 },
+    { 0x0490,  0x04cc,  UniCaseRangeU0490 },
+    { 0x1e00,  0x1ffc,  UniCaseRangeU1e00 },
+    { 0xff40,  0xff5a,  UniCaseRangeUff40 },
+    { 0 }
+};

fs/jfs/jfs_xattr.h

@@ -0,0 +1,74 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef H_JFS_XATTR
+#define H_JFS_XATTR
+
+/*
+ * jfs_ea_list describe the on-disk format of the extended attributes.
+ * I know the null-terminator is redundant since namelen is stored, but
+ * I am maintaining compatibility with OS/2 where possible.
+ */
+struct jfs_ea {
+	u8 flag;	/* Unused? */
+	u8 namelen;	/* Length of name */
+	__le16 valuelen;	/* Length of value */
+	char name[0];	/* Attribute name (includes null-terminator) */
+};			/* Value immediately follows name */
+
+struct jfs_ea_list {
+	__le32 size;		/* overall size */
+	struct jfs_ea ea[0];	/* Variable length list */
+};
+
+/* Macros for defining maxiumum number of bytes supported for EAs */
+#define MAXEASIZE	65535
+#define MAXEALISTSIZE	MAXEASIZE
+
+/*
+ * some macros for dealing with variable length EA lists.
+ */
+#define EA_SIZE(ea) \
+	(sizeof (struct jfs_ea) + (ea)->namelen + 1 + \
+	 le16_to_cpu((ea)->valuelen))
+#define	NEXT_EA(ea) ((struct jfs_ea *) (((char *) (ea)) + (EA_SIZE (ea))))
+#define	FIRST_EA(ealist) ((ealist)->ea)
+#define	EALIST_SIZE(ealist) le32_to_cpu((ealist)->size)
+#define	END_EALIST(ealist) \
+	((struct jfs_ea *) (((char *) (ealist)) + EALIST_SIZE(ealist)))
+
+extern int __jfs_setxattr(tid_t, struct inode *, const char *, const void *,
+			  size_t, int);
+extern int jfs_setxattr(struct dentry *, const char *, const void *, size_t,
+			int);
+extern ssize_t __jfs_getxattr(struct inode *, const char *, void *, size_t);
+extern ssize_t jfs_getxattr(struct dentry *, const char *, void *, size_t);
+extern ssize_t jfs_listxattr(struct dentry *, char *, size_t);
+extern int jfs_removexattr(struct dentry *, const char *);
+
+#ifdef CONFIG_JFS_SECURITY
+extern int jfs_init_security(tid_t, struct inode *, struct inode *);
+#else
+static inline int jfs_init_security(tid_t tid, struct inode *inode,
+				    struct inode *dir)
+{
+	return 0;
+}
+#endif
+
+#endif	/* H_JFS_XATTR */

fs/jfs/jfs_xtree.h

@@ -0,0 +1,134 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+#ifndef _H_JFS_XTREE
+#define _H_JFS_XTREE
+
+/*
+ *      jfs_xtree.h: extent allocation descriptor B+-tree manager
+ */
+
+#include "jfs_btree.h"
+
+
+/*
+ *      extent allocation descriptor (xad)
+ */
+typedef struct xad {
+	unsigned flag:8;	/* 1: flag */
+	unsigned rsvrd:16;	/* 2: reserved */
+	unsigned off1:8;	/* 1: offset in unit of fsblksize */
+	__le32 off2;		/* 4: offset in unit of fsblksize */
+	unsigned len:24;	/* 3: length in unit of fsblksize */
+	unsigned addr1:8;	/* 1: address in unit of fsblksize */
+	__le32 addr2;		/* 4: address in unit of fsblksize */
+} xad_t;			/* (16) */
+
+#define MAXXLEN         ((1 << 24) - 1)
+
+#define XTSLOTSIZE      16
+#define L2XTSLOTSIZE    4
+
+/* xad_t field construction */
+#define XADoffset(xad, offset64)\
+{\
+        (xad)->off1 = ((u64)offset64) >> 32;\
+        (xad)->off2 = __cpu_to_le32((offset64) & 0xffffffff);\
+}
+#define XADaddress(xad, address64)\
+{\
+        (xad)->addr1 = ((u64)address64) >> 32;\
+        (xad)->addr2 = __cpu_to_le32((address64) & 0xffffffff);\
+}
+#define XADlength(xad, length32)        (xad)->len = __cpu_to_le24(length32)
+
+/* xad_t field extraction */
+#define offsetXAD(xad)\
+        ( ((s64)((xad)->off1)) << 32 | __le32_to_cpu((xad)->off2))
+#define addressXAD(xad)\
+        ( ((s64)((xad)->addr1)) << 32 | __le32_to_cpu((xad)->addr2))
+#define lengthXAD(xad)  __le24_to_cpu((xad)->len)
+
+/* xad list */
+struct xadlist {
+	s16 maxnxad;
+	s16 nxad;
+	xad_t *xad;
+};
+
+/* xad_t flags */
+#define XAD_NEW         0x01	/* new */
+#define XAD_EXTENDED    0x02	/* extended */
+#define XAD_COMPRESSED  0x04	/* compressed with recorded length */
+#define XAD_NOTRECORDED 0x08	/* allocated but not recorded */
+#define XAD_COW         0x10	/* copy-on-write */
+
+
+/* possible values for maxentry */
+#define XTROOTINITSLOT_DIR  6
+#define XTROOTINITSLOT  10
+#define XTROOTMAXSLOT   18
+#define XTPAGEMAXSLOT   256
+#define XTENTRYSTART    2
+
+/*
+ *      xtree page:
+ */
+typedef union {
+	struct xtheader {
+		__le64 next;	/* 8: */
+		__le64 prev;	/* 8: */
+
+		u8 flag;	/* 1: */
+		u8 rsrvd1;	/* 1: */
+		__le16 nextindex;	/* 2: next index = number of entries */
+		__le16 maxentry;	/* 2: max number of entries */
+		__le16 rsrvd2;	/* 2: */
+
+		pxd_t self;	/* 8: self */
+	} header;		/* (32) */
+
+	xad_t xad[XTROOTMAXSLOT];	/* 16 * maxentry: xad array */
+} xtpage_t;
+
+/*
+ *      external declaration
+ */
+extern int xtLookup(struct inode *ip, s64 lstart, s64 llen,
+		    int *pflag, s64 * paddr, int *plen, int flag);
+extern int xtLookupList(struct inode *ip, struct lxdlist * lxdlist,
+			struct xadlist * xadlist, int flag);
+extern void xtInitRoot(tid_t tid, struct inode *ip);
+extern int xtInsert(tid_t tid, struct inode *ip,
+		    int xflag, s64 xoff, int xlen, s64 * xaddrp, int flag);
+extern int xtExtend(tid_t tid, struct inode *ip, s64 xoff, int xlen,
+		    int flag);
+#ifdef _NOTYET
+extern int xtTailgate(tid_t tid, struct inode *ip,
+		      s64 xoff, int xlen, s64 xaddr, int flag);
+#endif
+extern int xtUpdate(tid_t tid, struct inode *ip, struct xad *nxad);
+extern int xtDelete(tid_t tid, struct inode *ip, s64 xoff, int xlen,
+		    int flag);
+extern s64 xtTruncate(tid_t tid, struct inode *ip, s64 newsize, int type);
+extern s64 xtTruncate_pmap(tid_t tid, struct inode *ip, s64 committed_size);
+extern int xtRelocate(tid_t tid, struct inode *ip,
+		      xad_t * oxad, s64 nxaddr, int xtype);
+extern int xtAppend(tid_t tid,
+		    struct inode *ip, int xflag, s64 xoff, int maxblocks,
+		    int *xlenp, s64 * xaddrp, int flag);
+#endif				/* !_H_JFS_XTREE */

fs/jfs/resize.c

@@ -0,0 +1,539 @@
+/*
+ *   Copyright (C) International Business Machines  Corp., 2000-2004
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+*/
+
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include <linux/quotaops.h>
+#include "jfs_incore.h"
+#include "jfs_filsys.h"
+#include "jfs_metapage.h"
+#include "jfs_dinode.h"
+#include "jfs_imap.h"
+#include "jfs_dmap.h"
+#include "jfs_superblock.h"
+#include "jfs_txnmgr.h"
+#include "jfs_debug.h"
+
+#define BITSPERPAGE     (PSIZE << 3)
+#define L2MEGABYTE      20
+#define MEGABYTE        (1 << L2MEGABYTE)
+#define MEGABYTE32     (MEGABYTE << 5)
+
+/* convert block number to bmap file page number */
+#define BLKTODMAPN(b)\
+        (((b) >> 13) + ((b) >> 23) + ((b) >> 33) + 3 + 1)
+
+/*
+ *      jfs_extendfs()
+ *
+ * function: extend file system;
+ *
+ *   |-------------------------------|----------|----------|
+ *   file system space               fsck       inline log
+ *                                   workspace  space
+ *
+ * input:
+ *      new LVSize: in LV blocks (required)
+ *      new LogSize: in LV blocks (optional)
+ *      new FSSize: in LV blocks (optional)
+ *
+ * new configuration:
+ * 1. set new LogSize as specified or default from new LVSize;
+ * 2. compute new FSCKSize from new LVSize;
+ * 3. set new FSSize as MIN(FSSize, LVSize-(LogSize+FSCKSize)) where
+ *    assert(new FSSize >= old FSSize),
+ *    i.e., file system must not be shrinked;
+ */
+int jfs_extendfs(struct super_block *sb, s64 newLVSize, int newLogSize)
+{
+	int rc = 0;
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct inode *ipbmap = sbi->ipbmap;
+	struct inode *ipbmap2;
+	struct inode *ipimap = sbi->ipimap;
+	struct jfs_log *log = sbi->log;
+	struct bmap *bmp = sbi->bmap;
+	s64 newLogAddress, newFSCKAddress;
+	int newFSCKSize;
+	s64 newMapSize = 0, mapSize;
+	s64 XAddress, XSize, nblocks, xoff, xaddr, t64;
+	s64 oldLVSize;
+	s64 newFSSize;
+	s64 VolumeSize;
+	int newNpages = 0, nPages, newPage, xlen, t32;
+	int tid;
+	int log_formatted = 0;
+	struct inode *iplist[1];
+	struct jfs_superblock *j_sb, *j_sb2;
+	uint old_agsize;
+	struct buffer_head *bh, *bh2;
+
+	/* If the volume hasn't grown, get out now */
+
+	if (sbi->mntflag & JFS_INLINELOG)
+		oldLVSize = addressPXD(&sbi->logpxd) + lengthPXD(&sbi->logpxd);
+	else
+		oldLVSize = addressPXD(&sbi->fsckpxd) +
+		    lengthPXD(&sbi->fsckpxd);
+
+	if (oldLVSize >= newLVSize) {
+		printk(KERN_WARNING
+		       "jfs_extendfs: volume hasn't grown, returning\n");
+		goto out;
+	}
+
+	VolumeSize = sb->s_bdev->bd_inode->i_size >> sb->s_blocksize_bits;
+
+	if (VolumeSize) {
+		if (newLVSize > VolumeSize) {
+			printk(KERN_WARNING "jfs_extendfs: invalid size\n");
+			rc = -EINVAL;
+			goto out;
+		}
+	} else {
+		/* check the device */
+		bh = sb_bread(sb, newLVSize - 1);
+		if (!bh) {
+			printk(KERN_WARNING "jfs_extendfs: invalid size\n");
+			rc = -EINVAL;
+			goto out;
+		}
+		bforget(bh);
+	}
+
+	/* Can't extend write-protected drive */
+
+	if (isReadOnly(ipbmap)) {
+		printk(KERN_WARNING "jfs_extendfs: read-only file system\n");
+		rc = -EROFS;
+		goto out;
+	}
+
+	/*
+	 *      reconfigure LV spaces
+	 *      ---------------------
+	 *
+	 * validate new size, or, if not specified, determine new size
+	 */
+
+	/*
+	 * reconfigure inline log space:
+	 */
+	if ((sbi->mntflag & JFS_INLINELOG)) {
+		if (newLogSize == 0) {
+			/*
+			 * no size specified: default to 1/256 of aggregate
+			 * size; rounded up to a megabyte boundary;
+			 */
+			newLogSize = newLVSize >> 8;
+			t32 = (1 << (20 - sbi->l2bsize)) - 1;
+			newLogSize = (newLogSize + t32) & ~t32;
+			newLogSize =
+			    min(newLogSize, MEGABYTE32 >> sbi->l2bsize);
+		} else {
+			/*
+			 * convert the newLogSize to fs blocks.
+			 *
+			 * Since this is given in megabytes, it will always be
+			 * an even number of pages.
+			 */
+			newLogSize = (newLogSize * MEGABYTE) >> sbi->l2bsize;
+		}
+
+	} else
+		newLogSize = 0;
+
+	newLogAddress = newLVSize - newLogSize;
+
+	/*
+	 * reconfigure fsck work space:
+	 *
+	 * configure it to the end of the logical volume regardless of
+	 * whether file system extends to the end of the aggregate;
+	 * Need enough 4k pages to cover:
+	 *  - 1 bit per block in aggregate rounded up to BPERDMAP boundary
+	 *  - 1 extra page to handle control page and intermediate level pages
+	 *  - 50 extra pages for the chkdsk service log
+	 */
+	t64 = ((newLVSize - newLogSize + BPERDMAP - 1) >> L2BPERDMAP)
+	    << L2BPERDMAP;
+	t32 = ((t64 + (BITSPERPAGE - 1)) / BITSPERPAGE) + 1 + 50;
+	newFSCKSize = t32 << sbi->l2nbperpage;
+	newFSCKAddress = newLogAddress - newFSCKSize;
+
+	/*
+	 * compute new file system space;
+	 */
+	newFSSize = newLVSize - newLogSize - newFSCKSize;
+
+	/* file system cannot be shrinked */
+	if (newFSSize < bmp->db_mapsize) {
+		rc = -EINVAL;
+		goto out;
+	}
+
+	/*
+	 * If we're expanding enough that the inline log does not overlap
+	 * the old one, we can format the new log before we quiesce the
+	 * filesystem.
+	 */
+	if ((sbi->mntflag & JFS_INLINELOG) && (newLogAddress > oldLVSize)) {
+		if ((rc = lmLogFormat(log, newLogAddress, newLogSize)))
+			goto out;
+		log_formatted = 1;
+	}
+	/*
+	 *      quiesce file system
+	 *
+	 * (prepare to move the inline log and to prevent map update)
+	 *
+	 * block any new transactions and wait for completion of
+	 * all wip transactions and flush modified pages s.t.
+	 * on-disk file system is in consistent state and
+	 * log is not required for recovery.
+	 */
+	txQuiesce(sb);
+
+	/* Reset size of direct inode */
+	sbi->direct_inode->i_size =  sb->s_bdev->bd_inode->i_size;
+
+	if (sbi->mntflag & JFS_INLINELOG) {
+		/*
+		 * deactivate old inline log
+		 */
+		lmLogShutdown(log);
+
+		/*
+		 * mark on-disk super block for fs in transition;
+		 *
+		 * update on-disk superblock for the new space configuration
+		 * of inline log space and fsck work space descriptors:
+		 * N.B. FS descriptor is NOT updated;
+		 *
+		 * crash recovery:
+		 * logredo(): if FM_EXTENDFS, return to fsck() for cleanup;
+		 * fsck(): if FM_EXTENDFS, reformat inline log and fsck
+		 * workspace from superblock inline log descriptor and fsck
+		 * workspace descriptor;
+		 */
+
+		/* read in superblock */
+		if ((rc = readSuper(sb, &bh)))
+			goto error_out;
+		j_sb = (struct jfs_superblock *)bh->b_data;
+
+		/* mark extendfs() in progress */
+		j_sb->s_state |= cpu_to_le32(FM_EXTENDFS);
+		j_sb->s_xsize = cpu_to_le64(newFSSize);
+		PXDaddress(&j_sb->s_xfsckpxd, newFSCKAddress);
+		PXDlength(&j_sb->s_xfsckpxd, newFSCKSize);
+		PXDaddress(&j_sb->s_xlogpxd, newLogAddress);
+		PXDlength(&j_sb->s_xlogpxd, newLogSize);
+
+		/* synchronously update superblock */
+		mark_buffer_dirty(bh);
+		sync_dirty_buffer(bh);
+		brelse(bh);
+
+		/*
+		 * format new inline log synchronously;
+		 *
+		 * crash recovery: if log move in progress,
+		 * reformat log and exit success;
+		 */
+		if (!log_formatted)
+			if ((rc = lmLogFormat(log, newLogAddress, newLogSize)))
+				goto error_out;
+
+		/*
+		 * activate new log
+		 */
+		log->base = newLogAddress;
+		log->size = newLogSize >> (L2LOGPSIZE - sb->s_blocksize_bits);
+		if ((rc = lmLogInit(log)))
+			goto error_out;
+	}
+
+	/*
+	 *      extend block allocation map
+	 *      ---------------------------
+	 *
+	 * extendfs() for new extension, retry after crash recovery;
+	 *
+	 * note: both logredo() and fsck() rebuild map from
+	 * the bitmap and configuration parameter from superblock
+	 * (disregarding all other control information in the map);
+	 *
+	 * superblock:
+	 *  s_size: aggregate size in physical blocks;
+	 */
+	/*
+	 *      compute the new block allocation map configuration
+	 *
+	 * map dinode:
+	 *  di_size: map file size in byte;
+	 *  di_nblocks: number of blocks allocated for map file;
+	 *  di_mapsize: number of blocks in aggregate (covered by map);
+	 * map control page:
+	 *  db_mapsize: number of blocks in aggregate (covered by map);
+	 */
+	newMapSize = newFSSize;
+	/* number of data pages of new bmap file:
+	 * roundup new size to full dmap page boundary and
+	 * add 1 extra dmap page for next extendfs()
+	 */
+	t64 = (newMapSize - 1) + BPERDMAP;
+	newNpages = BLKTODMAPN(t64) + 1;
+
+	/*
+	 *      extend map from current map (WITHOUT growing mapfile)
+	 *
+	 * map new extension with unmapped part of the last partial
+	 * dmap page, if applicable, and extra page(s) allocated
+	 * at end of bmap by mkfs() or previous extendfs();
+	 */
+      extendBmap:
+	/* compute number of blocks requested to extend */
+	mapSize = bmp->db_mapsize;
+	XAddress = mapSize;	/* eXtension Address */
+	XSize = newMapSize - mapSize;	/* eXtension Size */
+	old_agsize = bmp->db_agsize;	/* We need to know if this changes */
+
+	/* compute number of blocks that can be extended by current mapfile */
+	t64 = dbMapFileSizeToMapSize(ipbmap);
+	if (mapSize > t64) {
+		printk(KERN_ERR "jfs_extendfs: mapSize (0x%Lx) > t64 (0x%Lx)\n",
+		       (long long) mapSize, (long long) t64);
+		rc = -EIO;
+		goto error_out;
+	}
+	nblocks = min(t64 - mapSize, XSize);
+
+	/*
+	 * update map pages for new extension:
+	 *
+	 * update/init dmap and bubble up the control hierarchy
+	 * incrementally fold up dmaps into upper levels;
+	 * update bmap control page;
+	 */
+	if ((rc = dbExtendFS(ipbmap, XAddress, nblocks)))
+		goto error_out;
+	/*
+	 * the map now has extended to cover additional nblocks:
+	 * dn_mapsize = oldMapsize + nblocks;
+	 */
+	/* ipbmap->i_mapsize += nblocks; */
+	XSize -= nblocks;
+
+	/*
+	 *      grow map file to cover remaining extension
+	 *      and/or one extra dmap page for next extendfs();
+	 *
+	 * allocate new map pages and its backing blocks, and
+	 * update map file xtree
+	 */
+	/* compute number of data pages of current bmap file */
+	nPages = ipbmap->i_size >> L2PSIZE;
+
+	/* need to grow map file ? */
+	if (nPages == newNpages)
+		goto finalizeBmap;
+
+	/*
+	 * grow bmap file for the new map pages required:
+	 *
+	 * allocate growth at the start of newly extended region;
+	 * bmap file only grows sequentially, i.e., both data pages
+	 * and possibly xtree index pages may grow in append mode,
+	 * s.t. logredo() can reconstruct pre-extension state
+	 * by washing away bmap file of pages outside s_size boundary;
+	 */
+	/*
+	 * journal map file growth as if a regular file growth:
+	 * (note: bmap is created with di_mode = IFJOURNAL|IFREG);
+	 *
+	 * journaling of bmap file growth is not required since
+	 * logredo() do/can not use log records of bmap file growth
+	 * but it provides careful write semantics, pmap update, etc.;
+	 */
+	/* synchronous write of data pages: bmap data pages are
+	 * cached in meta-data cache, and not written out
+	 * by txCommit();
+	 */
+	filemap_fdatawait(ipbmap->i_mapping);
+	filemap_write_and_wait(ipbmap->i_mapping);
+	diWriteSpecial(ipbmap, 0);
+
+	newPage = nPages;	/* first new page number */
+	xoff = newPage << sbi->l2nbperpage;
+	xlen = (newNpages - nPages) << sbi->l2nbperpage;
+	xlen = min(xlen, (int) nblocks) & ~(sbi->nbperpage - 1);
+	xaddr = XAddress;
+
+	tid = txBegin(sb, COMMIT_FORCE);
+
+	if ((rc = xtAppend(tid, ipbmap, 0, xoff, nblocks, &xlen, &xaddr, 0))) {
+		txEnd(tid);
+		goto error_out;
+	}
+	/* update bmap file size */
+	ipbmap->i_size += xlen << sbi->l2bsize;
+	inode_add_bytes(ipbmap, xlen << sbi->l2bsize);
+
+	iplist[0] = ipbmap;
+	rc = txCommit(tid, 1, &iplist[0], COMMIT_FORCE);
+
+	txEnd(tid);
+
+	if (rc)
+		goto error_out;
+
+	/*
+	 * map file has been grown now to cover extension to further out;
+	 * di_size = new map file size;
+	 *
+	 * if huge extension, the previous extension based on previous
+	 * map file size may not have been sufficient to cover whole extension
+	 * (it could have been used up for new map pages),
+	 * but the newly grown map file now covers lot bigger new free space
+	 * available for further extension of map;
+	 */
+	/* any more blocks to extend ? */
+	if (XSize)
+		goto extendBmap;
+
+      finalizeBmap:
+	/* finalize bmap */
+	dbFinalizeBmap(ipbmap);
+
+	/*
+	 *      update inode allocation map
+	 *      ---------------------------
+	 *
+	 * move iag lists from old to new iag;
+	 * agstart field is not updated for logredo() to reconstruct
+	 * iag lists if system crash occurs.
+	 * (computation of ag number from agstart based on agsize
+	 * will correctly identify the new ag);
+	 */
+	/* if new AG size the same as old AG size, done! */
+	if (bmp->db_agsize != old_agsize) {
+		if ((rc = diExtendFS(ipimap, ipbmap)))
+			goto error_out;
+
+		/* finalize imap */
+		if ((rc = diSync(ipimap)))
+			goto error_out;
+	}
+
+	/*
+	 *      finalize
+	 *      --------
+	 *
+	 * extension is committed when on-disk super block is
+	 * updated with new descriptors: logredo will recover
+	 * crash before it to pre-extension state;
+	 */
+
+	/* sync log to skip log replay of bmap file growth transaction; */
+	/* lmLogSync(log, 1); */
+
+	/*
+	 * synchronous write bmap global control page;
+	 * for crash before completion of write
+	 * logredo() will recover to pre-extendfs state;
+	 * for crash after completion of write,
+	 * logredo() will recover post-extendfs state;
+	 */
+	if ((rc = dbSync(ipbmap)))
+		goto error_out;
+
+	/*
+	 * copy primary bmap inode to secondary bmap inode
+	 */
+
+	ipbmap2 = diReadSpecial(sb, BMAP_I, 1);
+	if (ipbmap2 == NULL) {
+		printk(KERN_ERR "jfs_extendfs: diReadSpecial(bmap) failed\n");
+		goto error_out;
+	}
+	memcpy(&JFS_IP(ipbmap2)->i_xtroot, &JFS_IP(ipbmap)->i_xtroot, 288);
+	ipbmap2->i_size = ipbmap->i_size;
+	ipbmap2->i_blocks = ipbmap->i_blocks;
+
+	diWriteSpecial(ipbmap2, 1);
+	diFreeSpecial(ipbmap2);
+
+	/*
+	 *      update superblock
+	 */
+	if ((rc = readSuper(sb, &bh)))
+		goto error_out;
+	j_sb = (struct jfs_superblock *)bh->b_data;
+
+	/* mark extendfs() completion */
+	j_sb->s_state &= cpu_to_le32(~FM_EXTENDFS);
+	j_sb->s_size = cpu_to_le64(bmp->db_mapsize <<
+				   le16_to_cpu(j_sb->s_l2bfactor));
+	j_sb->s_agsize = cpu_to_le32(bmp->db_agsize);
+
+	/* update inline log space descriptor */
+	if (sbi->mntflag & JFS_INLINELOG) {
+		PXDaddress(&(j_sb->s_logpxd), newLogAddress);
+		PXDlength(&(j_sb->s_logpxd), newLogSize);
+	}
+
+	/* record log's mount serial number */
+	j_sb->s_logserial = cpu_to_le32(log->serial);
+
+	/* update fsck work space descriptor */
+	PXDaddress(&(j_sb->s_fsckpxd), newFSCKAddress);
+	PXDlength(&(j_sb->s_fsckpxd), newFSCKSize);
+	j_sb->s_fscklog = 1;
+	/* sb->s_fsckloglen remains the same */
+
+	/* Update secondary superblock */
+	bh2 = sb_bread(sb, SUPER2_OFF >> sb->s_blocksize_bits);
+	if (bh2) {
+		j_sb2 = (struct jfs_superblock *)bh2->b_data;
+		memcpy(j_sb2, j_sb, sizeof (struct jfs_superblock));
+
+		mark_buffer_dirty(bh);
+		sync_dirty_buffer(bh2);
+		brelse(bh2);
+	}
+
+	/* write primary superblock */
+	mark_buffer_dirty(bh);
+	sync_dirty_buffer(bh);
+	brelse(bh);
+
+	goto resume;
+
+      error_out:
+	jfs_error(sb, "jfs_extendfs");
+
+      resume:
+	/*
+	 *      resume file system transactions
+	 */
+	txResume(sb);
+
+      out:
+	return rc;
+}

fs/jfs/super.c

@@ -0,0 +1,874 @@
+/*
+ *   Copyright (C) International Business Machines Corp., 2000-2004
+ *   Portions Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/module.h>
+#include <linux/parser.h>
+#include <linux/completion.h>
+#include <linux/vfs.h>
+#include <linux/mount.h>
+#include <linux/moduleparam.h>
+#include <linux/kthread.h>
+#include <linux/posix_acl.h>
+#include <linux/buffer_head.h>
+#include <asm/uaccess.h>
+#include <linux/seq_file.h>
+
+#include "jfs_incore.h"
+#include "jfs_filsys.h"
+#include "jfs_inode.h"
+#include "jfs_metapage.h"
+#include "jfs_superblock.h"
+#include "jfs_dmap.h"
+#include "jfs_imap.h"
+#include "jfs_acl.h"
+#include "jfs_debug.h"
+
+MODULE_DESCRIPTION("The Journaled Filesystem (JFS)");
+MODULE_AUTHOR("Steve Best/Dave Kleikamp/Barry Arndt, IBM");
+MODULE_LICENSE("GPL");
+
+static struct kmem_cache * jfs_inode_cachep;
+
+static const struct super_operations jfs_super_operations;
+static struct export_operations jfs_export_operations;
+static struct file_system_type jfs_fs_type;
+
+#define MAX_COMMIT_THREADS 64
+static int commit_threads = 0;
+module_param(commit_threads, int, 0);
+MODULE_PARM_DESC(commit_threads, "Number of commit threads");
+
+static struct task_struct *jfsCommitThread[MAX_COMMIT_THREADS];
+struct task_struct *jfsIOthread;
+struct task_struct *jfsSyncThread;
+
+#ifdef CONFIG_JFS_DEBUG
+int jfsloglevel = JFS_LOGLEVEL_WARN;
+module_param(jfsloglevel, int, 0644);
+MODULE_PARM_DESC(jfsloglevel, "Specify JFS loglevel (0, 1 or 2)");
+#endif
+
+static void jfs_handle_error(struct super_block *sb)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+
+	if (sb->s_flags & MS_RDONLY)
+		return;
+
+	updateSuper(sb, FM_DIRTY);
+
+	if (sbi->flag & JFS_ERR_PANIC)
+		panic("JFS (device %s): panic forced after error\n",
+			sb->s_id);
+	else if (sbi->flag & JFS_ERR_REMOUNT_RO) {
+		jfs_err("ERROR: (device %s): remounting filesystem "
+			"as read-only\n",
+			sb->s_id);
+		sb->s_flags |= MS_RDONLY;
+	}
+
+	/* nothing is done for continue beyond marking the superblock dirty */
+}
+
+void jfs_error(struct super_block *sb, const char * function, ...)
+{
+	static char error_buf[256];
+	va_list args;
+
+	va_start(args, function);
+	vsnprintf(error_buf, sizeof(error_buf), function, args);
+	va_end(args);
+
+	printk(KERN_ERR "ERROR: (device %s): %s\n", sb->s_id, error_buf);
+
+	jfs_handle_error(sb);
+}
+
+static struct inode *jfs_alloc_inode(struct super_block *sb)
+{
+	struct jfs_inode_info *jfs_inode;
+
+	jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
+	if (!jfs_inode)
+		return NULL;
+	return &jfs_inode->vfs_inode;
+}
+
+static void jfs_destroy_inode(struct inode *inode)
+{
+	struct jfs_inode_info *ji = JFS_IP(inode);
+
+	BUG_ON(!list_empty(&ji->anon_inode_list));
+
+	spin_lock_irq(&ji->ag_lock);
+	if (ji->active_ag != -1) {
+		struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
+		atomic_dec(&bmap->db_active[ji->active_ag]);
+		ji->active_ag = -1;
+	}
+	spin_unlock_irq(&ji->ag_lock);
+
+#ifdef CONFIG_JFS_POSIX_ACL
+	if (ji->i_acl != JFS_ACL_NOT_CACHED) {
+		posix_acl_release(ji->i_acl);
+		ji->i_acl = JFS_ACL_NOT_CACHED;
+	}
+	if (ji->i_default_acl != JFS_ACL_NOT_CACHED) {
+		posix_acl_release(ji->i_default_acl);
+		ji->i_default_acl = JFS_ACL_NOT_CACHED;
+	}
+#endif
+
+	kmem_cache_free(jfs_inode_cachep, ji);
+}
+
+static int jfs_statfs(struct dentry *dentry, struct kstatfs *buf)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(dentry->d_sb);
+	s64 maxinodes;
+	struct inomap *imap = JFS_IP(sbi->ipimap)->i_imap;
+
+	jfs_info("In jfs_statfs");
+	buf->f_type = JFS_SUPER_MAGIC;
+	buf->f_bsize = sbi->bsize;
+	buf->f_blocks = sbi->bmap->db_mapsize;
+	buf->f_bfree = sbi->bmap->db_nfree;
+	buf->f_bavail = sbi->bmap->db_nfree;
+	/*
+	 * If we really return the number of allocated & free inodes, some
+	 * applications will fail because they won't see enough free inodes.
+	 * We'll try to calculate some guess as to how may inodes we can
+	 * really allocate
+	 *
+	 * buf->f_files = atomic_read(&imap->im_numinos);
+	 * buf->f_ffree = atomic_read(&imap->im_numfree);
+	 */
+	maxinodes = min((s64) atomic_read(&imap->im_numinos) +
+			((sbi->bmap->db_nfree >> imap->im_l2nbperiext)
+			 << L2INOSPEREXT), (s64) 0xffffffffLL);
+	buf->f_files = maxinodes;
+	buf->f_ffree = maxinodes - (atomic_read(&imap->im_numinos) -
+				    atomic_read(&imap->im_numfree));
+
+	buf->f_namelen = JFS_NAME_MAX;
+	return 0;
+}
+
+static void jfs_put_super(struct super_block *sb)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	int rc;
+
+	jfs_info("In jfs_put_super");
+	rc = jfs_umount(sb);
+	if (rc)
+		jfs_err("jfs_umount failed with return code %d", rc);
+	if (sbi->nls_tab)
+		unload_nls(sbi->nls_tab);
+	sbi->nls_tab = NULL;
+
+	truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
+	iput(sbi->direct_inode);
+	sbi->direct_inode = NULL;
+
+	kfree(sbi);
+}
+
+enum {
+	Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
+	Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
+	Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
+};
+
+static match_table_t tokens = {
+	{Opt_integrity, "integrity"},
+	{Opt_nointegrity, "nointegrity"},
+	{Opt_iocharset, "iocharset=%s"},
+	{Opt_resize, "resize=%u"},
+	{Opt_resize_nosize, "resize"},
+	{Opt_errors, "errors=%s"},
+	{Opt_ignore, "noquota"},
+	{Opt_ignore, "quota"},
+	{Opt_usrquota, "usrquota"},
+	{Opt_grpquota, "grpquota"},
+	{Opt_uid, "uid=%u"},
+	{Opt_gid, "gid=%u"},
+	{Opt_umask, "umask=%u"},
+	{Opt_err, NULL}
+};
+
+static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
+			 int *flag)
+{
+	void *nls_map = (void *)-1;	/* -1: no change;  NULL: none */
+	char *p;
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+
+	*newLVSize = 0;
+
+	if (!options)
+		return 1;
+
+	while ((p = strsep(&options, ",")) != NULL) {
+		substring_t args[MAX_OPT_ARGS];
+		int token;
+		if (!*p)
+			continue;
+
+		token = match_token(p, tokens, args);
+		switch (token) {
+		case Opt_integrity:
+			*flag &= ~JFS_NOINTEGRITY;
+			break;
+		case Opt_nointegrity:
+			*flag |= JFS_NOINTEGRITY;
+			break;
+		case Opt_ignore:
+			/* Silently ignore the quota options */
+			/* Don't do anything ;-) */
+			break;
+		case Opt_iocharset:
+			if (nls_map && nls_map != (void *) -1)
+				unload_nls(nls_map);
+			if (!strcmp(args[0].from, "none"))
+				nls_map = NULL;
+			else {
+				nls_map = load_nls(args[0].from);
+				if (!nls_map) {
+					printk(KERN_ERR
+					       "JFS: charset not found\n");
+					goto cleanup;
+				}
+			}
+			break;
+		case Opt_resize:
+		{
+			char *resize = args[0].from;
+			*newLVSize = simple_strtoull(resize, &resize, 0);
+			break;
+		}
+		case Opt_resize_nosize:
+		{
+			*newLVSize = sb->s_bdev->bd_inode->i_size >>
+				sb->s_blocksize_bits;
+			if (*newLVSize == 0)
+				printk(KERN_ERR
+				       "JFS: Cannot determine volume size\n");
+			break;
+		}
+		case Opt_errors:
+		{
+			char *errors = args[0].from;
+			if (!errors || !*errors)
+				goto cleanup;
+			if (!strcmp(errors, "continue")) {
+				*flag &= ~JFS_ERR_REMOUNT_RO;
+				*flag &= ~JFS_ERR_PANIC;
+				*flag |= JFS_ERR_CONTINUE;
+			} else if (!strcmp(errors, "remount-ro")) {
+				*flag &= ~JFS_ERR_CONTINUE;
+				*flag &= ~JFS_ERR_PANIC;
+				*flag |= JFS_ERR_REMOUNT_RO;
+			} else if (!strcmp(errors, "panic")) {
+				*flag &= ~JFS_ERR_CONTINUE;
+				*flag &= ~JFS_ERR_REMOUNT_RO;
+				*flag |= JFS_ERR_PANIC;
+			} else {
+				printk(KERN_ERR
+				       "JFS: %s is an invalid error handler\n",
+				       errors);
+				goto cleanup;
+			}
+			break;
+		}
+
+#ifdef CONFIG_QUOTA
+		case Opt_quota:
+		case Opt_usrquota:
+			*flag |= JFS_USRQUOTA;
+			break;
+		case Opt_grpquota:
+			*flag |= JFS_GRPQUOTA;
+			break;
+#else
+		case Opt_usrquota:
+		case Opt_grpquota:
+		case Opt_quota:
+			printk(KERN_ERR
+			       "JFS: quota operations not supported\n");
+			break;
+#endif
+		case Opt_uid:
+		{
+			char *uid = args[0].from;
+			sbi->uid = simple_strtoul(uid, &uid, 0);
+			break;
+		}
+		case Opt_gid:
+		{
+			char *gid = args[0].from;
+			sbi->gid = simple_strtoul(gid, &gid, 0);
+			break;
+		}
+		case Opt_umask:
+		{
+			char *umask = args[0].from;
+			sbi->umask = simple_strtoul(umask, &umask, 8);
+			if (sbi->umask & ~0777) {
+				printk(KERN_ERR
+				       "JFS: Invalid value of umask\n");
+				goto cleanup;
+			}
+			break;
+		}
+		default:
+			printk("jfs: Unrecognized mount option \"%s\" "
+					" or missing value\n", p);
+			goto cleanup;
+		}
+	}
+
+	if (nls_map != (void *) -1) {
+		/* Discard old (if remount) */
+		if (sbi->nls_tab)
+			unload_nls(sbi->nls_tab);
+		sbi->nls_tab = nls_map;
+	}
+	return 1;
+
+cleanup:
+	if (nls_map && nls_map != (void *) -1)
+		unload_nls(nls_map);
+	return 0;
+}
+
+static int jfs_remount(struct super_block *sb, int *flags, char *data)
+{
+	s64 newLVSize = 0;
+	int rc = 0;
+	int flag = JFS_SBI(sb)->flag;
+
+	if (!parse_options(data, sb, &newLVSize, &flag)) {
+		return -EINVAL;
+	}
+	if (newLVSize) {
+		if (sb->s_flags & MS_RDONLY) {
+			printk(KERN_ERR
+		  "JFS: resize requires volume to be mounted read-write\n");
+			return -EROFS;
+		}
+		rc = jfs_extendfs(sb, newLVSize, 0);
+		if (rc)
+			return rc;
+	}
+
+	if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
+		/*
+		 * Invalidate any previously read metadata.  fsck may have
+		 * changed the on-disk data since we mounted r/o
+		 */
+		truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
+
+		JFS_SBI(sb)->flag = flag;
+		return jfs_mount_rw(sb, 1);
+	}
+	if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
+		rc = jfs_umount_rw(sb);
+		JFS_SBI(sb)->flag = flag;
+		return rc;
+	}
+	if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
+		if (!(sb->s_flags & MS_RDONLY)) {
+			rc = jfs_umount_rw(sb);
+			if (rc)
+				return rc;
+			JFS_SBI(sb)->flag = flag;
+			return jfs_mount_rw(sb, 1);
+		}
+	JFS_SBI(sb)->flag = flag;
+
+	return 0;
+}
+
+static int jfs_fill_super(struct super_block *sb, void *data, int silent)
+{
+	struct jfs_sb_info *sbi;
+	struct inode *inode;
+	int rc;
+	s64 newLVSize = 0;
+	int flag;
+
+	jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
+
+	if (!new_valid_dev(sb->s_bdev->bd_dev))
+		return -EOVERFLOW;
+
+	sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
+	if (!sbi)
+		return -ENOMEM;
+	sb->s_fs_info = sbi;
+	sbi->sb = sb;
+	sbi->uid = sbi->gid = sbi->umask = -1;
+
+	/* initialize the mount flag and determine the default error handler */
+	flag = JFS_ERR_REMOUNT_RO;
+
+	if (!parse_options((char *) data, sb, &newLVSize, &flag)) {
+		kfree(sbi);
+		return -EINVAL;
+	}
+	sbi->flag = flag;
+
+#ifdef CONFIG_JFS_POSIX_ACL
+	sb->s_flags |= MS_POSIXACL;
+#endif
+
+	if (newLVSize) {
+		printk(KERN_ERR "resize option for remount only\n");
+		return -EINVAL;
+	}
+
+	/*
+	 * Initialize blocksize to 4K.
+	 */
+	sb_set_blocksize(sb, PSIZE);
+
+	/*
+	 * Set method vectors.
+	 */
+	sb->s_op = &jfs_super_operations;
+	sb->s_export_op = &jfs_export_operations;
+
+	/*
+	 * Initialize direct-mapping inode/address-space
+	 */
+	inode = new_inode(sb);
+	if (inode == NULL)
+		goto out_kfree;
+	inode->i_ino = 0;
+	inode->i_nlink = 1;
+	inode->i_size = sb->s_bdev->bd_inode->i_size;
+	inode->i_mapping->a_ops = &jfs_metapage_aops;
+	insert_inode_hash(inode);
+	mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
+
+	sbi->direct_inode = inode;
+
+	rc = jfs_mount(sb);
+	if (rc) {
+		if (!silent) {
+			jfs_err("jfs_mount failed w/return code = %d", rc);
+		}
+		goto out_mount_failed;
+	}
+	if (sb->s_flags & MS_RDONLY)
+		sbi->log = NULL;
+	else {
+		rc = jfs_mount_rw(sb, 0);
+		if (rc) {
+			if (!silent) {
+				jfs_err("jfs_mount_rw failed, return code = %d",
+					rc);
+			}
+			goto out_no_rw;
+		}
+	}
+
+	sb->s_magic = JFS_SUPER_MAGIC;
+
+	inode = iget(sb, ROOT_I);
+	if (!inode || is_bad_inode(inode))
+		goto out_no_root;
+	sb->s_root = d_alloc_root(inode);
+	if (!sb->s_root)
+		goto out_no_root;
+
+	if (sbi->mntflag & JFS_OS2)
+		sb->s_root->d_op = &jfs_ci_dentry_operations;
+
+	/* logical blocks are represented by 40 bits in pxd_t, etc. */
+	sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
+#if BITS_PER_LONG == 32
+	/*
+	 * Page cache is indexed by long.
+	 * I would use MAX_LFS_FILESIZE, but it's only half as big
+	 */
+	sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, sb->s_maxbytes);
+#endif
+	sb->s_time_gran = 1;
+	return 0;
+
+out_no_root:
+	jfs_err("jfs_read_super: get root inode failed");
+	if (inode)
+		iput(inode);
+
+out_no_rw:
+	rc = jfs_umount(sb);
+	if (rc) {
+		jfs_err("jfs_umount failed with return code %d", rc);
+	}
+out_mount_failed:
+	filemap_write_and_wait(sbi->direct_inode->i_mapping);
+	truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
+	make_bad_inode(sbi->direct_inode);
+	iput(sbi->direct_inode);
+	sbi->direct_inode = NULL;
+out_kfree:
+	if (sbi->nls_tab)
+		unload_nls(sbi->nls_tab);
+	kfree(sbi);
+	return -EINVAL;
+}
+
+static void jfs_write_super_lockfs(struct super_block *sb)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct jfs_log *log = sbi->log;
+
+	if (!(sb->s_flags & MS_RDONLY)) {
+		txQuiesce(sb);
+		lmLogShutdown(log);
+		updateSuper(sb, FM_CLEAN);
+	}
+}
+
+static void jfs_unlockfs(struct super_block *sb)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(sb);
+	struct jfs_log *log = sbi->log;
+	int rc = 0;
+
+	if (!(sb->s_flags & MS_RDONLY)) {
+		updateSuper(sb, FM_MOUNT);
+		if ((rc = lmLogInit(log)))
+			jfs_err("jfs_unlock failed with return code %d", rc);
+		else
+			txResume(sb);
+	}
+}
+
+static int jfs_get_sb(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
+{
+	return get_sb_bdev(fs_type, flags, dev_name, data, jfs_fill_super,
+			   mnt);
+}
+
+static int jfs_sync_fs(struct super_block *sb, int wait)
+{
+	struct jfs_log *log = JFS_SBI(sb)->log;
+
+	/* log == NULL indicates read-only mount */
+	if (log) {
+		jfs_flush_journal(log, wait);
+		jfs_syncpt(log, 0);
+	}
+
+	return 0;
+}
+
+static int jfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
+{
+	struct jfs_sb_info *sbi = JFS_SBI(vfs->mnt_sb);
+
+	if (sbi->uid != -1)
+		seq_printf(seq, ",uid=%d", sbi->uid);
+	if (sbi->gid != -1)
+		seq_printf(seq, ",gid=%d", sbi->gid);
+	if (sbi->umask != -1)
+		seq_printf(seq, ",umask=%03o", sbi->umask);
+	if (sbi->flag & JFS_NOINTEGRITY)
+		seq_puts(seq, ",nointegrity");
+
+#ifdef CONFIG_QUOTA
+	if (sbi->flag & JFS_USRQUOTA)
+		seq_puts(seq, ",usrquota");
+
+	if (sbi->flag & JFS_GRPQUOTA)
+		seq_puts(seq, ",grpquota");
+#endif
+
+	return 0;
+}
+
+#ifdef CONFIG_QUOTA
+
+/* Read data from quotafile - avoid pagecache and such because we cannot afford
+ * acquiring the locks... As quota files are never truncated and quota code
+ * itself serializes the operations (and noone else should touch the files)
+ * we don't have to be afraid of races */
+static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
+			      size_t len, loff_t off)
+{
+	struct inode *inode = sb_dqopt(sb)->files[type];
+	sector_t blk = off >> sb->s_blocksize_bits;
+	int err = 0;
+	int offset = off & (sb->s_blocksize - 1);
+	int tocopy;
+	size_t toread;
+	struct buffer_head tmp_bh;
+	struct buffer_head *bh;
+	loff_t i_size = i_size_read(inode);
+
+	if (off > i_size)
+		return 0;
+	if (off+len > i_size)
+		len = i_size-off;
+	toread = len;
+	while (toread > 0) {
+		tocopy = sb->s_blocksize - offset < toread ?
+				sb->s_blocksize - offset : toread;
+
+		tmp_bh.b_state = 0;
+		tmp_bh.b_size = 1 << inode->i_blkbits;
+		err = jfs_get_block(inode, blk, &tmp_bh, 0);
+		if (err)
+			return err;
+		if (!buffer_mapped(&tmp_bh))	/* A hole? */
+			memset(data, 0, tocopy);
+		else {
+			bh = sb_bread(sb, tmp_bh.b_blocknr);
+			if (!bh)
+				return -EIO;
+			memcpy(data, bh->b_data+offset, tocopy);
+			brelse(bh);
+		}
+		offset = 0;
+		toread -= tocopy;
+		data += tocopy;
+		blk++;
+	}
+	return len;
+}
+
+/* Write to quotafile */
+static ssize_t jfs_quota_write(struct super_block *sb, int type,
+			       const char *data, size_t len, loff_t off)
+{
+	struct inode *inode = sb_dqopt(sb)->files[type];
+	sector_t blk = off >> sb->s_blocksize_bits;
+	int err = 0;
+	int offset = off & (sb->s_blocksize - 1);
+	int tocopy;
+	size_t towrite = len;
+	struct buffer_head tmp_bh;
+	struct buffer_head *bh;
+
+	mutex_lock(&inode->i_mutex);
+	while (towrite > 0) {
+		tocopy = sb->s_blocksize - offset < towrite ?
+				sb->s_blocksize - offset : towrite;
+
+		tmp_bh.b_state = 0;
+		tmp_bh.b_size = 1 << inode->i_blkbits;
+		err = jfs_get_block(inode, blk, &tmp_bh, 1);
+		if (err)
+			goto out;
+		if (offset || tocopy != sb->s_blocksize)
+			bh = sb_bread(sb, tmp_bh.b_blocknr);
+		else
+			bh = sb_getblk(sb, tmp_bh.b_blocknr);
+		if (!bh) {
+			err = -EIO;
+			goto out;
+		}
+		lock_buffer(bh);
+		memcpy(bh->b_data+offset, data, tocopy);
+		flush_dcache_page(bh->b_page);
+		set_buffer_uptodate(bh);
+		mark_buffer_dirty(bh);
+		unlock_buffer(bh);
+		brelse(bh);
+		offset = 0;
+		towrite -= tocopy;
+		data += tocopy;
+		blk++;
+	}
+out:
+	if (len == towrite)
+		return err;
+	if (inode->i_size < off+len-towrite)
+		i_size_write(inode, off+len-towrite);
+	inode->i_version++;
+	inode->i_mtime = inode->i_ctime = CURRENT_TIME;
+	mark_inode_dirty(inode);
+	mutex_unlock(&inode->i_mutex);
+	return len - towrite;
+}
+
+#endif
+
+static const struct super_operations jfs_super_operations = {
+	.alloc_inode	= jfs_alloc_inode,
+	.destroy_inode	= jfs_destroy_inode,
+	.read_inode	= jfs_read_inode,
+	.dirty_inode	= jfs_dirty_inode,
+	.write_inode	= jfs_write_inode,
+	.delete_inode	= jfs_delete_inode,
+	.put_super	= jfs_put_super,
+	.sync_fs	= jfs_sync_fs,
+	.write_super_lockfs = jfs_write_super_lockfs,
+	.unlockfs       = jfs_unlockfs,
+	.statfs		= jfs_statfs,
+	.remount_fs	= jfs_remount,
+	.show_options	= jfs_show_options,
+#ifdef CONFIG_QUOTA
+	.quota_read	= jfs_quota_read,
+	.quota_write	= jfs_quota_write,
+#endif
+};
+
+static struct export_operations jfs_export_operations = {
+	.get_parent	= jfs_get_parent,
+};
+
+static struct file_system_type jfs_fs_type = {
+	.owner		= THIS_MODULE,
+	.name		= "jfs",
+	.get_sb		= jfs_get_sb,
+	.kill_sb	= kill_block_super,
+	.fs_flags	= FS_REQUIRES_DEV,
+};
+
+static void init_once(void *foo, struct kmem_cache * cachep, unsigned long flags)
+{
+	struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
+
+	if ((flags & (SLAB_CTOR_VERIFY | SLAB_CTOR_CONSTRUCTOR)) ==
+	    SLAB_CTOR_CONSTRUCTOR) {
+		memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
+		INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
+		init_rwsem(&jfs_ip->rdwrlock);
+		mutex_init(&jfs_ip->commit_mutex);
+		init_rwsem(&jfs_ip->xattr_sem);
+		spin_lock_init(&jfs_ip->ag_lock);
+		jfs_ip->active_ag = -1;
+#ifdef CONFIG_JFS_POSIX_ACL
+		jfs_ip->i_acl = JFS_ACL_NOT_CACHED;
+		jfs_ip->i_default_acl = JFS_ACL_NOT_CACHED;
+#endif
+		inode_init_once(&jfs_ip->vfs_inode);
+	}
+}
+
+static int __init init_jfs_fs(void)
+{
+	int i;
+	int rc;
+
+	jfs_inode_cachep =
+	    kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
+			    SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
+			    init_once, NULL);
+	if (jfs_inode_cachep == NULL)
+		return -ENOMEM;
+
+	/*
+	 * Metapage initialization
+	 */
+	rc = metapage_init();
+	if (rc) {
+		jfs_err("metapage_init failed w/rc = %d", rc);
+		goto free_slab;
+	}
+
+	/*
+	 * Transaction Manager initialization
+	 */
+	rc = txInit();
+	if (rc) {
+		jfs_err("txInit failed w/rc = %d", rc);
+		goto free_metapage;
+	}
+
+	/*
+	 * I/O completion thread (endio)
+	 */
+	jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
+	if (IS_ERR(jfsIOthread)) {
+		rc = PTR_ERR(jfsIOthread);
+		jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
+		goto end_txmngr;
+	}
+
+	if (commit_threads < 1)
+		commit_threads = num_online_cpus();
+	if (commit_threads > MAX_COMMIT_THREADS)
+		commit_threads = MAX_COMMIT_THREADS;
+
+	for (i = 0; i < commit_threads; i++) {
+		jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
+		if (IS_ERR(jfsCommitThread[i])) {
+			rc = PTR_ERR(jfsCommitThread[i]);
+			jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
+			commit_threads = i;
+			goto kill_committask;
+		}
+	}
+
+	jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
+	if (IS_ERR(jfsSyncThread)) {
+		rc = PTR_ERR(jfsSyncThread);
+		jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
+		goto kill_committask;
+	}
+
+#ifdef PROC_FS_JFS
+	jfs_proc_init();
+#endif
+
+	return register_filesystem(&jfs_fs_type);
+
+kill_committask:
+	for (i = 0; i < commit_threads; i++)
+		kthread_stop(jfsCommitThread[i]);
+	kthread_stop(jfsIOthread);
+end_txmngr:
+	txExit();
+free_metapage:
+	metapage_exit();
+free_slab:
+	kmem_cache_destroy(jfs_inode_cachep);
+	return rc;
+}
+
+static void __exit exit_jfs_fs(void)
+{
+	int i;
+
+	jfs_info("exit_jfs_fs called");
+
+	txExit();
+	metapage_exit();
+
+	kthread_stop(jfsIOthread);
+	for (i = 0; i < commit_threads; i++)
+		kthread_stop(jfsCommitThread[i]);
+	kthread_stop(jfsSyncThread);
+#ifdef PROC_FS_JFS
+	jfs_proc_clean();
+#endif
+	unregister_filesystem(&jfs_fs_type);
+	kmem_cache_destroy(jfs_inode_cachep);
+}
+
+module_init(init_jfs_fs)
+module_exit(exit_jfs_fs)

fs/jfs/symlink.c

@@ -0,0 +1,40 @@
+/*
+ *   Copyright (C) Christoph Hellwig, 2001-2002
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/fs.h>
+#include <linux/namei.h>
+#include "jfs_incore.h"
+#include "jfs_inode.h"
+#include "jfs_xattr.h"
+
+static void *jfs_follow_link(struct dentry *dentry, struct nameidata *nd)
+{
+	char *s = JFS_IP(dentry->d_inode)->i_inline;
+	nd_set_link(nd, s);
+	return NULL;
+}
+
+const struct inode_operations jfs_symlink_inode_operations = {
+	.readlink	= generic_readlink,
+	.follow_link	= jfs_follow_link,
+	.setxattr	= jfs_setxattr,
+	.getxattr	= jfs_getxattr,
+	.listxattr	= jfs_listxattr,
+	.removexattr	= jfs_removexattr,
+};
+