[net-next-2.6.git] / fs / jfs / inode.c

/*
 *   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 <linux/writeback.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"


struct inode *jfs_iget(struct super_block *sb, unsigned long ino)
{
	struct inode *inode;
	int ret;

	inode = iget_locked(sb, ino);
	if (!inode)
		return ERR_PTR(-ENOMEM);
	if (!(inode->i_state & I_NEW))
		return inode;

	ret = diRead(inode);
	if (ret < 0) {
		iget_failed(inode);
		return ERR_PTR(ret);
	}

	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_fast_symlink_inode_operations;
			/*
			 * The inline data should be null-terminated, but
			 * don't let on-disk corruption crash the kernel
			 */
			JFS_IP(inode)->i_inline[inode->i_size] = '\0';
		}
	} else {
		inode->i_op = &jfs_file_inode_operations;
		init_special_inode(inode, inode->i_mode, inode->i_rdev);
	}
	unlock_new_inode(inode);
	return 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, struct writeback_control *wbc)
{
	int wait = wbc->sync_mode == WB_SYNC_ALL;

	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_evict_inode(struct inode *inode)
{
	jfs_info("In jfs_evict_inode, inode = 0x%p", inode);

	if (!inode->i_nlink && !is_bad_inode(inode)) {
		dquot_initialize(inode);

		if (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_initialize(inode);
			dquot_free_inode(inode);
		}
	} else {
		truncate_inode_pages(&inode->i_data, 0);
	}
	end_writeback(inode);
	dquot_drop(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 block_write_full_page(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_write_begin(struct file *file, struct address_space *mapping,
				loff_t pos, unsigned len, unsigned flags,
				struct page **pagep, void **fsdata)
{
	int ret;

	ret = nobh_write_begin(mapping, pos, len, flags, pagep, fsdata,
				jfs_get_block);
	if (unlikely(ret)) {
		loff_t isize = mapping->host->i_size;
		if (pos + len > isize)
			vmtruncate(mapping->host, isize);
	}

	return ret;
}

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;
	ssize_t ret;

	ret = blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
				offset, nr_segs, jfs_get_block, NULL);

	/*
	 * In case of error extending write may have instantiated a few
	 * blocks outside i_size. Trim these off again.
	 */
	if (unlikely((rw & WRITE) && ret < 0)) {
		loff_t isize = i_size_read(inode);
		loff_t end = offset + iov_length(iov, nr_segs);

		if (end > isize)
			vmtruncate(inode, isize);
	}

	return ret;
}

const struct address_space_operations jfs_aops = {
	.readpage	= jfs_readpage,
	.readpages	= jfs_readpages,
	.writepage	= jfs_writepage,
	.writepages	= jfs_writepages,
	.sync_page	= block_sync_page,
	.write_begin	= jfs_write_begin,
	.write_end	= nobh_write_end,
	.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, jfs_get_block);

	IWRITE_LOCK(ip, RDWRLOCK_NORMAL);
	jfs_truncate_nolock(ip, ip->i_size);
	IWRITE_UNLOCK(ip);
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Copyright (C) International Business Machines Corp., 2000-2004
	3	* Portions Copyright (C) Christoph Hellwig, 2001-2002
	4	*
	5	* This program is free software; you can redistribute it and/or modify
	6	* it under the terms of the GNU General Public License as published by
63f83c9f	7	* the Free Software Foundation; either version 2 of the License, or
1da177e4	8	* (at your option) any later version.
63f83c9f	9	*
1da177e4 LT	10	* This program is distributed in the hope that it will be useful,
	11	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
	13	* the GNU General Public License for more details.
	14	*
	15	* You should have received a copy of the GNU General Public License
63f83c9f	16	* along with this program; if not, write to the Free Software
1da177e4 LT	17	* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
	18	*/
	19
	20	#include <linux/fs.h>
	21	#include <linux/mpage.h>
	22	#include <linux/buffer_head.h>
	23	#include <linux/pagemap.h>
	24	#include <linux/quotaops.h>
a9185b41	25	#include <linux/writeback.h>
1da177e4	26	#include "jfs_incore.h"
1868f4aa	27	#include "jfs_inode.h"
1da177e4 LT	28	#include "jfs_filsys.h"
	29	#include "jfs_imap.h"
	30	#include "jfs_extent.h"
	31	#include "jfs_unicode.h"
	32	#include "jfs_debug.h"
	33
	34
eab1df71	35	struct inode jfs_iget(struct super_block sb, unsigned long ino)
1da177e4	36	{
eab1df71 DH	37	struct inode *inode;
	38	int ret;
	39
	40	inode = iget_locked(sb, ino);
	41	if (!inode)
	42	return ERR_PTR(-ENOMEM);
	43	if (!(inode->i_state & I_NEW))
	44	return inode;
	45
	46	ret = diRead(inode);
	47	if (ret < 0) {
	48	iget_failed(inode);
	49	return ERR_PTR(ret);
1da177e4 LT	50	}
	51
	52	if (S_ISREG(inode->i_mode)) {
	53	inode->i_op = &jfs_file_inode_operations;
	54	inode->i_fop = &jfs_file_operations;
	55	inode->i_mapping->a_ops = &jfs_aops;
	56	} else if (S_ISDIR(inode->i_mode)) {
	57	inode->i_op = &jfs_dir_inode_operations;
	58	inode->i_fop = &jfs_dir_operations;
	59	} else if (S_ISLNK(inode->i_mode)) {
	60	if (inode->i_size >= IDATASIZE) {
	61	inode->i_op = &page_symlink_inode_operations;
	62	inode->i_mapping->a_ops = &jfs_aops;
d69e83d9	63	} else {
c7f2e1f0	64	inode->i_op = &jfs_fast_symlink_inode_operations;
d69e83d9 DK	65	/*
	66	* The inline data should be null-terminated, but
	67	* don't let on-disk corruption crash the kernel
	68	*/
	69	JFS_IP(inode)->i_inline[inode->i_size] = '\0';
	70	}
1da177e4 LT	71	} else {
	72	inode->i_op = &jfs_file_inode_operations;
	73	init_special_inode(inode, inode->i_mode, inode->i_rdev);
	74	}
eab1df71 DH	75	unlock_new_inode(inode);
eab1df71 DH	76	return inode;
1da177e4 LT	77	}
	78
	79	/*
	80	* Workhorse of both fsync & write_inode
	81	*/
	82	int jfs_commit_inode(struct inode *inode, int wait)
	83	{
	84	int rc = 0;
	85	tid_t tid;
	86	static int noisy = 5;
	87
	88	jfs_info("In jfs_commit_inode, inode = 0x%p", inode);
	89
	90	/*
	91	* Don't commit if inode has been committed since last being
	92	* marked dirty, or if it has been deleted.
	93	*/
	94	if (inode->i_nlink == 0 \|\| !test_cflag(COMMIT_Dirty, inode))
	95	return 0;
	96
	97	if (isReadOnly(inode)) {
	98	/* kernel allows writes to devices on read-only
	99	* partitions and may think inode is dirty
	100	*/
	101	if (!special_file(inode->i_mode) && noisy) {
	102	jfs_err("jfs_commit_inode(0x%p) called on "
	103	"read-only volume", inode);
	104	jfs_err("Is remount racy?");
	105	noisy--;
	106	}
	107	return 0;
	108	}
	109
	110	tid = txBegin(inode->i_sb, COMMIT_INODE);
1de87444	111	mutex_lock(&JFS_IP(inode)->commit_mutex);
1da177e4 LT	112
1da177e4 LT	113	/*
1de87444	114	* Retest inode state after taking commit_mutex
1da177e4 LT	115	*/
	116	if (inode->i_nlink && test_cflag(COMMIT_Dirty, inode))
	117	rc = txCommit(tid, 1, &inode, wait ? COMMIT_SYNC : 0);
	118
	119	txEnd(tid);
1de87444	120	mutex_unlock(&JFS_IP(inode)->commit_mutex);
1da177e4 LT	121	return rc;
	122	}
	123
a9185b41	124	int jfs_write_inode(struct inode inode, struct writeback_control wbc)
1da177e4	125	{
a9185b41 CH	126	int wait = wbc->sync_mode == WB_SYNC_ALL;
a9185b41 CH	127
1da177e4 LT	128	if (test_cflag(COMMIT_Nolink, inode))
	129	return 0;
	130	/*
	131	* If COMMIT_DIRTY is not set, the inode isn't really dirty.
	132	* It has been committed since the last change, but was still
	133	* on the dirty inode list.
	134	*/
	135	if (!test_cflag(COMMIT_Dirty, inode)) {
	136	/* Make sure committed changes hit the disk */
	137	jfs_flush_journal(JFS_SBI(inode->i_sb)->log, wait);
	138	return 0;
	139	}
	140
	141	if (jfs_commit_inode(inode, wait)) {
	142	jfs_err("jfs_write_inode: jfs_commit_inode failed!");
	143	return -EIO;
	144	} else
	145	return 0;
	146	}
	147
62aff86f	148	void jfs_evict_inode(struct inode *inode)
1da177e4	149	{
62aff86f	150	jfs_info("In jfs_evict_inode, inode = 0x%p", inode);
1da177e4	151
62aff86f	152	if (!inode->i_nlink && !is_bad_inode(inode)) {
871a2931	153	dquot_initialize(inode);
907f4554	154
62aff86f AV	155	if (JFS_IP(inode)->fileset == FILESYSTEM_I) {
62aff86f AV	156	truncate_inode_pages(&inode->i_data, 0);
fef26658	157
62aff86f AV	158	if (test_cflag(COMMIT_Freewmap, inode))
62aff86f AV	159	jfs_free_zero_link(inode);
1da177e4	160
62aff86f	161	diFree(inode);
1da177e4	162
62aff86f AV	163	/*
	164	* Free the inode from the quota allocation.
	165	*/
	166	dquot_initialize(inode);
	167	dquot_free_inode(inode);
	168	}
	169	} else {
	170	truncate_inode_pages(&inode->i_data, 0);
b1b5d7f9	171	}
62aff86f AV	172	end_writeback(inode);
62aff86f AV	173	dquot_drop(inode);
1da177e4 LT	174	}
	175
	176	void jfs_dirty_inode(struct inode *inode)
	177	{
	178	static int noisy = 5;
	179
	180	if (isReadOnly(inode)) {
	181	if (!special_file(inode->i_mode) && noisy) {
	182	/* kernel allows writes to devices on read-only
	183	* partitions and may try to mark inode dirty
	184	*/
	185	jfs_err("jfs_dirty_inode called on read-only volume");
	186	jfs_err("Is remount racy?");
	187	noisy--;
	188	}
	189	return;
	190	}
	191
	192	set_cflag(COMMIT_Dirty, inode);
	193	}
	194
115ff50b DK	195	int jfs_get_block(struct inode *ip, sector_t lblock,
115ff50b DK	196	struct buffer_head *bh_result, int create)
1da177e4 LT	197	{
	198	s64 lblock64 = lblock;
	199	int rc = 0;
1da177e4 LT	200	xad_t xad;
	201	s64 xaddr;
	202	int xflag;
115ff50b	203	s32 xlen = bh_result->b_size >> ip->i_blkbits;
1da177e4	204
1da177e4 LT	205	/*
	206	* Take appropriate lock on inode
	207	*/
7fab479b	208	if (create)
82d5b9a7	209	IWRITE_LOCK(ip, RDWRLOCK_NORMAL);
7fab479b	210	else
82d5b9a7	211	IREAD_LOCK(ip, RDWRLOCK_NORMAL);
1da177e4 LT	212
1da177e4 LT	213	if (((lblock64 << ip->i_sb->s_blocksize_bits) < ip->i_size) &&
115ff50b	214	(!xtLookup(ip, lblock64, xlen, &xflag, &xaddr, &xlen, 0)) &&
6628465e	215	xaddr) {
1da177e4 LT	216	if (xflag & XAD_NOTRECORDED) {
	217	if (!create)
	218	/*
	219	* Allocated but not recorded, read treats
	220	* this as a hole
	221	*/
	222	goto unlock;
	223	#ifdef _JFS_4K
	224	XADoffset(&xad, lblock64);
	225	XADlength(&xad, xlen);
	226	XADaddress(&xad, xaddr);
	227	#else /* _JFS_4K */
	228	/*
	229	* As long as block size = 4K, this isn't a problem.
	230	* We should mark the whole page not ABNR, but how
	231	* will we know to mark the other blocks BH_New?
	232	*/
	233	BUG();
	234	#endif /* _JFS_4K */
	235	rc = extRecord(ip, &xad);
	236	if (rc)
	237	goto unlock;
	238	set_buffer_new(bh_result);
	239	}
	240
	241	map_bh(bh_result, ip->i_sb, xaddr);
	242	bh_result->b_size = xlen << ip->i_blkbits;
	243	goto unlock;
	244	}
	245	if (!create)
	246	goto unlock;
	247
	248	/*
	249	* Allocate a new block
	250	*/
	251	#ifdef _JFS_4K
	252	if ((rc = extHint(ip, lblock64 << ip->i_sb->s_blocksize_bits, &xad)))
	253	goto unlock;
4d81715f	254	rc = extAlloc(ip, xlen, lblock64, &xad, false);
1da177e4 LT	255	if (rc)
	256	goto unlock;
	257
	258	set_buffer_new(bh_result);
	259	map_bh(bh_result, ip->i_sb, addressXAD(&xad));
	260	bh_result->b_size = lengthXAD(&xad) << ip->i_blkbits;
	261
	262	#else /* _JFS_4K */
	263	/*
	264	* We need to do whatever it takes to keep all but the last buffers
	265	* in 4K pages - see jfs_write.c
	266	*/
	267	BUG();
	268	#endif /* _JFS_4K */
	269
	270	unlock:
	271	/*
	272	* Release lock on inode
	273	*/
7fab479b DK	274	if (create)
	275	IWRITE_UNLOCK(ip);
	276	else
	277	IREAD_UNLOCK(ip);
1da177e4 LT	278	return rc;
	279	}
	280
1da177e4 LT	281	static int jfs_writepage(struct page page, struct writeback_control wbc)
1da177e4 LT	282	{
d5c5f84b	283	return block_write_full_page(page, jfs_get_block, wbc);
1da177e4 LT	284	}
	285
	286	static int jfs_writepages(struct address_space *mapping,
	287	struct writeback_control *wbc)
	288	{
	289	return mpage_writepages(mapping, wbc, jfs_get_block);
	290	}
	291
	292	static int jfs_readpage(struct file file, struct page page)
	293	{
	294	return mpage_readpage(page, jfs_get_block);
	295	}
	296
	297	static int jfs_readpages(struct file file, struct address_space mapping,
	298	struct list_head *pages, unsigned nr_pages)
	299	{
	300	return mpage_readpages(mapping, pages, nr_pages, jfs_get_block);
	301	}
	302
d5c5f84b NP	303	static int jfs_write_begin(struct file file, struct address_space mapping,
	304	loff_t pos, unsigned len, unsigned flags,
	305	struct page pagep, void fsdata)
1da177e4	306	{
ea0f04e5 CH	307	int ret;
	308
	309	ret = nobh_write_begin(mapping, pos, len, flags, pagep, fsdata,
d5c5f84b	310	jfs_get_block);
ea0f04e5 CH	311	if (unlikely(ret)) {
	312	loff_t isize = mapping->host->i_size;
	313	if (pos + len > isize)
	314	vmtruncate(mapping->host, isize);
	315	}
	316
	317	return ret;
1da177e4 LT	318	}
	319
	320	static sector_t jfs_bmap(struct address_space *mapping, sector_t block)
	321	{
	322	return generic_block_bmap(mapping, block, jfs_get_block);
	323	}
	324
	325	static ssize_t jfs_direct_IO(int rw, struct kiocb *iocb,
	326	const struct iovec *iov, loff_t offset, unsigned long nr_segs)
	327	{
	328	struct file *file = iocb->ki_filp;
	329	struct inode *inode = file->f_mapping->host;
eafdc7d1	330	ssize_t ret;
1da177e4	331
eafdc7d1	332	ret = blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
1d8fa7a2	333	offset, nr_segs, jfs_get_block, NULL);
eafdc7d1 CH	334
	335	/*
	336	* In case of error extending write may have instantiated a few
	337	* blocks outside i_size. Trim these off again.
	338	*/
	339	if (unlikely((rw & WRITE) && ret < 0)) {
	340	loff_t isize = i_size_read(inode);
	341	loff_t end = offset + iov_length(iov, nr_segs);
	342
	343	if (end > isize)
	344	vmtruncate(inode, isize);
	345	}
	346
	347	return ret;
1da177e4 LT	348	}
1da177e4 LT	349
f5e54d6e	350	const struct address_space_operations jfs_aops = {
1da177e4 LT	351	.readpage = jfs_readpage,
	352	.readpages = jfs_readpages,
	353	.writepage = jfs_writepage,
	354	.writepages = jfs_writepages,
	355	.sync_page = block_sync_page,
d5c5f84b	356	.write_begin = jfs_write_begin,
03158cd7	357	.write_end = nobh_write_end,
1da177e4 LT	358	.bmap = jfs_bmap,
	359	.direct_IO = jfs_direct_IO,
	360	};
	361
	362	/*
	363	* Guts of jfs_truncate. Called with locks already held. Can be called
	364	* with directory for truncating directory index table.
	365	*/
	366	void jfs_truncate_nolock(struct inode *ip, loff_t length)
	367	{
	368	loff_t newsize;
	369	tid_t tid;
	370
	371	ASSERT(length >= 0);
	372
	373	if (test_cflag(COMMIT_Nolink, ip)) {
	374	xtTruncate(0, ip, length, COMMIT_WMAP);
	375	return;
	376	}
	377
	378	do {
	379	tid = txBegin(ip->i_sb, 0);
	380
	381	/*
1de87444	382	* The commit_mutex cannot be taken before txBegin.
1da177e4 LT	383	* txBegin may block and there is a chance the inode
	384	* could be marked dirty and need to be committed
	385	* before txBegin unblocks
	386	*/
1de87444	387	mutex_lock(&JFS_IP(ip)->commit_mutex);
1da177e4 LT	388
	389	newsize = xtTruncate(tid, ip, length,
	390	COMMIT_TRUNCATE \| COMMIT_PWMAP);
	391	if (newsize < 0) {
	392	txEnd(tid);
1de87444	393	mutex_unlock(&JFS_IP(ip)->commit_mutex);
1da177e4 LT	394	break;
	395	}
	396
	397	ip->i_mtime = ip->i_ctime = CURRENT_TIME;
	398	mark_inode_dirty(ip);
	399
	400	txCommit(tid, 1, &ip, 0);
	401	txEnd(tid);
1de87444	402	mutex_unlock(&JFS_IP(ip)->commit_mutex);
1da177e4 LT	403	} while (newsize > length); /* Truncate isn't always atomic */
	404	}
	405
	406	void jfs_truncate(struct inode *ip)
	407	{
	408	jfs_info("jfs_truncate: size = 0x%lx", (ulong) ip->i_size);
	409
03158cd7	410	nobh_truncate_page(ip->i_mapping, ip->i_size, jfs_get_block);
1da177e4	411
82d5b9a7	412	IWRITE_LOCK(ip, RDWRLOCK_NORMAL);
1da177e4 LT	413	jfs_truncate_nolock(ip, ip->i_size);
	414	IWRITE_UNLOCK(ip);
	415	}