namei.c 37.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6
/*
 *   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
D
Dave Kleikamp 已提交
7
 *   the Free Software Foundation; either version 2 of the License, or
L
Linus Torvalds 已提交
8
 *   (at your option) any later version.
D
Dave Kleikamp 已提交
9
 *
L
Linus Torvalds 已提交
10 11 12 13 14 15
 *   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
D
Dave Kleikamp 已提交
16
 *   along with this program;  if not, write to the Free Software
L
Linus Torvalds 已提交
17 18 19 20 21 22
 *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
 */

#include <linux/fs.h>
#include <linux/ctype.h>
#include <linux/quotaops.h>
C
Christoph Hellwig 已提交
23
#include <linux/exportfs.h>
L
Linus Torvalds 已提交
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
#include "jfs_incore.h"
#include "jfs_superblock.h"
#include "jfs_inode.h"
#include "jfs_dinode.h"
#include "jfs_dmap.h"
#include "jfs_unicode.h"
#include "jfs_metapage.h"
#include "jfs_xattr.h"
#include "jfs_acl.h"
#include "jfs_debug.h"

/*
 * forward references
 */
struct dentry_operations jfs_ci_dentry_operations;

static s64 commitZeroLink(tid_t, struct inode *);

42 43 44
/*
 * NAME:	free_ea_wmap(inode)
 *
D
Dave Kleikamp 已提交
45
 * FUNCTION:	free uncommitted extended attributes from working map
46 47 48 49 50 51 52 53 54 55 56 57 58 59
 *
 */
static inline void free_ea_wmap(struct inode *inode)
{
	dxd_t *ea = &JFS_IP(inode)->ea;

	if (ea->flag & DXD_EXTENT) {
		/* free EA pages from cache */
		invalidate_dxd_metapages(inode, *ea);
		dbFree(inode, addressDXD(ea), lengthDXD(ea));
	}
	ea->flag = 0;
}

L
Linus Torvalds 已提交
60 61 62 63 64 65
/*
 * NAME:	jfs_create(dip, dentry, mode)
 *
 * FUNCTION:	create a regular file in the parent directory <dip>
 *		with name = <from dentry> and mode = <mode>
 *
D
Dave Kleikamp 已提交
66
 * PARAMETER:	dip	- parent directory vnode
L
Linus Torvalds 已提交
67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
 *		dentry	- dentry of new file
 *		mode	- create mode (rwxrwxrwx).
 *		nd- nd struct
 *
 * RETURN:	Errors from subroutines
 *
 */
static int jfs_create(struct inode *dip, struct dentry *dentry, int mode,
		struct nameidata *nd)
{
	int rc = 0;
	tid_t tid;		/* transaction id */
	struct inode *ip = NULL;	/* child directory inode */
	ino_t ino;
	struct component_name dname;	/* child directory name */
	struct btstack btstack;
	struct inode *iplist[2];
	struct tblock *tblk;

	jfs_info("jfs_create: dip:0x%p name:%s", dip, dentry->d_name.name);

	/*
	 * search parent directory for entry/freespace
	 * (dtSearch() returns parent directory page pinned)
	 */
	if ((rc = get_UCSname(&dname, dentry)))
		goto out1;

	/*
	 * Either iAlloc() or txBegin() may block.  Deadlock can occur if we
	 * block there while holding dtree page, so we allocate the inode &
	 * begin the transaction before we search the directory.
	 */
	ip = ialloc(dip, mode);
101 102
	if (IS_ERR(ip)) {
		rc = PTR_ERR(ip);
L
Linus Torvalds 已提交
103 104 105 106 107
		goto out2;
	}

	tid = txBegin(dip->i_sb, 0);

D
Dave Kleikamp 已提交
108 109
	mutex_lock_nested(&JFS_IP(dip)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
110

111 112 113 114
	rc = jfs_init_acl(tid, ip, dip);
	if (rc)
		goto out3;

D
Dave Kleikamp 已提交
115 116 117 118 119 120
	rc = jfs_init_security(tid, ip, dip);
	if (rc) {
		txAbort(tid, 0);
		goto out3;
	}

L
Linus Torvalds 已提交
121 122
	if ((rc = dtSearch(dip, &dname, &ino, &btstack, JFS_CREATE))) {
		jfs_err("jfs_create: dtSearch returned %d", rc);
123
		txAbort(tid, 0);
L
Linus Torvalds 已提交
124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
		goto out3;
	}

	tblk = tid_to_tblock(tid);
	tblk->xflag |= COMMIT_CREATE;
	tblk->ino = ip->i_ino;
	tblk->u.ixpxd = JFS_IP(ip)->ixpxd;

	iplist[0] = dip;
	iplist[1] = ip;

	/*
	 * initialize the child XAD tree root in-line in inode
	 */
	xtInitRoot(tid, ip);

	/*
	 * create entry in parent directory for child directory
	 * (dtInsert() releases parent directory page)
	 */
	ino = ip->i_ino;
	if ((rc = dtInsert(tid, dip, &dname, &ino, &btstack))) {
		if (rc == -EIO) {
			jfs_err("jfs_create: dtInsert returned -EIO");
			txAbort(tid, 1);	/* Marks Filesystem dirty */
		} else
			txAbort(tid, 0);	/* Filesystem full */
		goto out3;
	}

	ip->i_op = &jfs_file_inode_operations;
	ip->i_fop = &jfs_file_operations;
	ip->i_mapping->a_ops = &jfs_aops;

	insert_inode_hash(ip);
	mark_inode_dirty(ip);

	dip->i_ctime = dip->i_mtime = CURRENT_TIME;

	mark_inode_dirty(dip);

	rc = txCommit(tid, 2, &iplist[0], 0);

      out3:
	txEnd(tid);
169
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
170
	mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
171
	if (rc) {
172
		free_ea_wmap(ip);
L
Linus Torvalds 已提交
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
		ip->i_nlink = 0;
		iput(ip);
	} else
		d_instantiate(dentry, ip);

      out2:
	free_UCSname(&dname);

      out1:

	jfs_info("jfs_create: rc:%d", rc);
	return rc;
}


/*
 * NAME:	jfs_mkdir(dip, dentry, mode)
 *
 * FUNCTION:	create a child directory in the parent directory <dip>
 *		with name = <from dentry> and mode = <mode>
 *
D
Dave Kleikamp 已提交
194
 * PARAMETER:	dip	- parent directory vnode
L
Linus Torvalds 已提交
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
 *		dentry	- dentry of child directory
 *		mode	- create mode (rwxrwxrwx).
 *
 * RETURN:	Errors from subroutines
 *
 * note:
 * EACCESS: user needs search+write permission on the parent directory
 */
static int jfs_mkdir(struct inode *dip, struct dentry *dentry, int mode)
{
	int rc = 0;
	tid_t tid;		/* transaction id */
	struct inode *ip = NULL;	/* child directory inode */
	ino_t ino;
	struct component_name dname;	/* child directory name */
	struct btstack btstack;
	struct inode *iplist[2];
	struct tblock *tblk;

	jfs_info("jfs_mkdir: dip:0x%p name:%s", dip, dentry->d_name.name);

	/* link count overflow on parent directory ? */
	if (dip->i_nlink == JFS_LINK_MAX) {
		rc = -EMLINK;
		goto out1;
	}

	/*
	 * search parent directory for entry/freespace
	 * (dtSearch() returns parent directory page pinned)
	 */
	if ((rc = get_UCSname(&dname, dentry)))
		goto out1;

	/*
	 * Either iAlloc() or txBegin() may block.  Deadlock can occur if we
	 * block there while holding dtree page, so we allocate the inode &
	 * begin the transaction before we search the directory.
	 */
	ip = ialloc(dip, S_IFDIR | mode);
235 236
	if (IS_ERR(ip)) {
		rc = PTR_ERR(ip);
L
Linus Torvalds 已提交
237 238 239 240 241
		goto out2;
	}

	tid = txBegin(dip->i_sb, 0);

D
Dave Kleikamp 已提交
242 243
	mutex_lock_nested(&JFS_IP(dip)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
244

245 246 247 248
	rc = jfs_init_acl(tid, ip, dip);
	if (rc)
		goto out3;

D
Dave Kleikamp 已提交
249 250 251 252 253 254
	rc = jfs_init_security(tid, ip, dip);
	if (rc) {
		txAbort(tid, 0);
		goto out3;
	}

L
Linus Torvalds 已提交
255 256
	if ((rc = dtSearch(dip, &dname, &ino, &btstack, JFS_CREATE))) {
		jfs_err("jfs_mkdir: dtSearch returned %d", rc);
257
		txAbort(tid, 0);
L
Linus Torvalds 已提交
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295
		goto out3;
	}

	tblk = tid_to_tblock(tid);
	tblk->xflag |= COMMIT_CREATE;
	tblk->ino = ip->i_ino;
	tblk->u.ixpxd = JFS_IP(ip)->ixpxd;

	iplist[0] = dip;
	iplist[1] = ip;

	/*
	 * initialize the child directory in-line in inode
	 */
	dtInitRoot(tid, ip, dip->i_ino);

	/*
	 * create entry in parent directory for child directory
	 * (dtInsert() releases parent directory page)
	 */
	ino = ip->i_ino;
	if ((rc = dtInsert(tid, dip, &dname, &ino, &btstack))) {
		if (rc == -EIO) {
			jfs_err("jfs_mkdir: dtInsert returned -EIO");
			txAbort(tid, 1);	/* Marks Filesystem dirty */
		} else
			txAbort(tid, 0);	/* Filesystem full */
		goto out3;
	}

	ip->i_nlink = 2;	/* for '.' */
	ip->i_op = &jfs_dir_inode_operations;
	ip->i_fop = &jfs_dir_operations;

	insert_inode_hash(ip);
	mark_inode_dirty(ip);

	/* update parent directory inode */
296
	inc_nlink(dip);		/* for '..' from child directory */
L
Linus Torvalds 已提交
297 298 299 300 301 302 303
	dip->i_ctime = dip->i_mtime = CURRENT_TIME;
	mark_inode_dirty(dip);

	rc = txCommit(tid, 2, &iplist[0], 0);

      out3:
	txEnd(tid);
304
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
305
	mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
306
	if (rc) {
307
		free_ea_wmap(ip);
L
Linus Torvalds 已提交
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
		ip->i_nlink = 0;
		iput(ip);
	} else
		d_instantiate(dentry, ip);

      out2:
	free_UCSname(&dname);


      out1:

	jfs_info("jfs_mkdir: rc:%d", rc);
	return rc;
}

/*
 * NAME:	jfs_rmdir(dip, dentry)
 *
 * FUNCTION:	remove a link to child directory
 *
D
Dave Kleikamp 已提交
328
 * PARAMETER:	dip	- parent inode
L
Linus Torvalds 已提交
329 330 331
 *		dentry	- child directory dentry
 *
 * RETURN:	-EINVAL	- if name is . or ..
332
 *		-EINVAL - if . or .. exist but are invalid.
L
Linus Torvalds 已提交
333 334 335
 *		errors from subroutines
 *
 * note:
D
Dave Kleikamp 已提交
336 337 338 339
 * if other threads have the directory open when the last link
 * is removed, the "." and ".." entries, if present, are removed before
 * rmdir() returns and no new entries may be created in the directory,
 * but the directory is not removed until the last reference to
L
Linus Torvalds 已提交
340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368
 * the directory is released (cf.unlink() of regular file).
 */
static int jfs_rmdir(struct inode *dip, struct dentry *dentry)
{
	int rc;
	tid_t tid;		/* transaction id */
	struct inode *ip = dentry->d_inode;
	ino_t ino;
	struct component_name dname;
	struct inode *iplist[2];
	struct tblock *tblk;

	jfs_info("jfs_rmdir: dip:0x%p name:%s", dip, dentry->d_name.name);

	/* Init inode for quota operations. */
	DQUOT_INIT(ip);

	/* directory must be empty to be removed */
	if (!dtEmpty(ip)) {
		rc = -ENOTEMPTY;
		goto out;
	}

	if ((rc = get_UCSname(&dname, dentry))) {
		goto out;
	}

	tid = txBegin(dip->i_sb, 0);

D
Dave Kleikamp 已提交
369 370
	mutex_lock_nested(&JFS_IP(dip)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387

	iplist[0] = dip;
	iplist[1] = ip;

	tblk = tid_to_tblock(tid);
	tblk->xflag |= COMMIT_DELETE;
	tblk->u.ip = ip;

	/*
	 * delete the entry of target directory from parent directory
	 */
	ino = ip->i_ino;
	if ((rc = dtDelete(tid, dip, &dname, &ino, JFS_REMOVE))) {
		jfs_err("jfs_rmdir: dtDelete returned %d", rc);
		if (rc == -EIO)
			txAbort(tid, 1);
		txEnd(tid);
388
		mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
389
		mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
390 391 392 393 394 395 396 397

		goto out2;
	}

	/* update parent directory's link count corresponding
	 * to ".." entry of the target directory deleted
	 */
	dip->i_ctime = dip->i_mtime = CURRENT_TIME;
398
	inode_dec_link_count(dip);
L
Linus Torvalds 已提交
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417

	/*
	 * OS/2 could have created EA and/or ACL
	 */
	/* free EA from both persistent and working map */
	if (JFS_IP(ip)->ea.flag & DXD_EXTENT) {
		/* free EA pages */
		txEA(tid, ip, &JFS_IP(ip)->ea, NULL);
	}
	JFS_IP(ip)->ea.flag = 0;

	/* free ACL from both persistent and working map */
	if (JFS_IP(ip)->acl.flag & DXD_EXTENT) {
		/* free ACL pages */
		txEA(tid, ip, &JFS_IP(ip)->acl, NULL);
	}
	JFS_IP(ip)->acl.flag = 0;

	/* mark the target directory as deleted */
418
	clear_nlink(ip);
L
Linus Torvalds 已提交
419 420 421 422 423 424
	mark_inode_dirty(ip);

	rc = txCommit(tid, 2, &iplist[0], 0);

	txEnd(tid);

425
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
426
	mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449

	/*
	 * Truncating the directory index table is not guaranteed.  It
	 * may need to be done iteratively
	 */
	if (test_cflag(COMMIT_Stale, dip)) {
		if (dip->i_size > 1)
			jfs_truncate_nolock(dip, 0);

		clear_cflag(COMMIT_Stale, dip);
	}

      out2:
	free_UCSname(&dname);

      out:
	jfs_info("jfs_rmdir: rc:%d", rc);
	return rc;
}

/*
 * NAME:	jfs_unlink(dip, dentry)
 *
D
Dave Kleikamp 已提交
450
 * FUNCTION:	remove a link to object <vp> named by <name>
L
Linus Torvalds 已提交
451 452
 *		from parent directory <dvp>
 *
D
Dave Kleikamp 已提交
453 454
 * PARAMETER:	dip	- inode of parent directory
 *		dentry	- dentry of object to be removed
L
Linus Torvalds 已提交
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
 *
 * RETURN:	errors from subroutines
 *
 * note:
 * temporary file: if one or more processes have the file open
 * when the last link is removed, the link will be removed before
 * unlink() returns, but the removal of the file contents will be
 * postponed until all references to the files are closed.
 *
 * JFS does NOT support unlink() on directories.
 *
 */
static int jfs_unlink(struct inode *dip, struct dentry *dentry)
{
	int rc;
	tid_t tid;		/* transaction id */
	struct inode *ip = dentry->d_inode;
	ino_t ino;
	struct component_name dname;	/* object name */
	struct inode *iplist[2];
	struct tblock *tblk;
	s64 new_size = 0;
	int commit_flag;

	jfs_info("jfs_unlink: dip:0x%p name:%s", dip, dentry->d_name.name);

	/* Init inode for quota operations. */
	DQUOT_INIT(ip);

	if ((rc = get_UCSname(&dname, dentry)))
		goto out;

D
Dave Kleikamp 已提交
487
	IWRITE_LOCK(ip, RDWRLOCK_NORMAL);
L
Linus Torvalds 已提交
488 489 490

	tid = txBegin(dip->i_sb, 0);

D
Dave Kleikamp 已提交
491 492
	mutex_lock_nested(&JFS_IP(dip)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
493 494 495 496 497 498 499 500 501 502 503 504 505

	iplist[0] = dip;
	iplist[1] = ip;

	/*
	 * delete the entry of target file from parent directory
	 */
	ino = ip->i_ino;
	if ((rc = dtDelete(tid, dip, &dname, &ino, JFS_REMOVE))) {
		jfs_err("jfs_unlink: dtDelete returned %d", rc);
		if (rc == -EIO)
			txAbort(tid, 1);	/* Marks FS Dirty */
		txEnd(tid);
506
		mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
507
		mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
508 509 510 511 512 513 514 515 516 517
		IWRITE_UNLOCK(ip);
		goto out1;
	}

	ASSERT(ip->i_nlink);

	ip->i_ctime = dip->i_ctime = dip->i_mtime = CURRENT_TIME;
	mark_inode_dirty(dip);

	/* update target's inode */
518
	inode_dec_link_count(ip);
L
Linus Torvalds 已提交
519 520

	/*
521
	 *	commit zero link count object
L
Linus Torvalds 已提交
522 523 524 525 526 527 528
	 */
	if (ip->i_nlink == 0) {
		assert(!test_cflag(COMMIT_Nolink, ip));
		/* free block resources */
		if ((new_size = commitZeroLink(tid, ip)) < 0) {
			txAbort(tid, 1);	/* Marks FS Dirty */
			txEnd(tid);
529
			mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
530
			mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
			IWRITE_UNLOCK(ip);
			rc = new_size;
			goto out1;
		}
		tblk = tid_to_tblock(tid);
		tblk->xflag |= COMMIT_DELETE;
		tblk->u.ip = ip;
	}

	/*
	 * Incomplete truncate of file data can
	 * result in timing problems unless we synchronously commit the
	 * transaction.
	 */
	if (new_size)
		commit_flag = COMMIT_SYNC;
	else
		commit_flag = 0;

	/*
	 * If xtTruncate was incomplete, commit synchronously to avoid
	 * timing complications
	 */
	rc = txCommit(tid, 2, &iplist[0], commit_flag);

	txEnd(tid);

558
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
559
	mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
560 561 562

	while (new_size && (rc == 0)) {
		tid = txBegin(dip->i_sb, 0);
563
		mutex_lock(&JFS_IP(ip)->commit_mutex);
L
Linus Torvalds 已提交
564 565 566 567 568 569 570
		new_size = xtTruncate_pmap(tid, ip, new_size);
		if (new_size < 0) {
			txAbort(tid, 1);	/* Marks FS Dirty */
			rc = new_size;
		} else
			rc = txCommit(tid, 2, &iplist[0], COMMIT_SYNC);
		txEnd(tid);
571
		mutex_unlock(&JFS_IP(ip)->commit_mutex);
L
Linus Torvalds 已提交
572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599
	}

	if (ip->i_nlink == 0)
		set_cflag(COMMIT_Nolink, ip);

	IWRITE_UNLOCK(ip);

	/*
	 * Truncating the directory index table is not guaranteed.  It
	 * may need to be done iteratively
	 */
	if (test_cflag(COMMIT_Stale, dip)) {
		if (dip->i_size > 1)
			jfs_truncate_nolock(dip, 0);

		clear_cflag(COMMIT_Stale, dip);
	}

      out1:
	free_UCSname(&dname);
      out:
	jfs_info("jfs_unlink: rc:%d", rc);
	return rc;
}

/*
 * NAME:	commitZeroLink()
 *
600
 * FUNCTION:	for non-directory, called by jfs_remove(),
L
Linus Torvalds 已提交
601
 *		truncate a regular file, directory or symbolic
D
Dave Kleikamp 已提交
602
 *		link to zero length. return 0 if type is not
L
Linus Torvalds 已提交
603 604 605 606 607 608 609 610 611
 *		one of these.
 *
 *		if the file is currently associated with a VM segment
 *		only permanent disk and inode map resources are freed,
 *		and neither the inode nor indirect blocks are modified
 *		so that the resources can be later freed in the work
 *		map by ctrunc1.
 *		if there is no VM segment on entry, the resources are
 *		freed in both work and permanent map.
D
Dave Kleikamp 已提交
612
 *		(? for temporary file - memory object is cached even
L
Linus Torvalds 已提交
613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665
 *		after no reference:
 *		reference count > 0 -   )
 *
 * PARAMETERS:	cd	- pointer to commit data structure.
 *			  current inode is the one to truncate.
 *
 * RETURN:	Errors from subroutines
 */
static s64 commitZeroLink(tid_t tid, struct inode *ip)
{
	int filetype;
	struct tblock *tblk;

	jfs_info("commitZeroLink: tid = %d, ip = 0x%p", tid, ip);

	filetype = ip->i_mode & S_IFMT;
	switch (filetype) {
	case S_IFREG:
		break;
	case S_IFLNK:
		/* fast symbolic link */
		if (ip->i_size < IDATASIZE) {
			ip->i_size = 0;
			return 0;
		}
		break;
	default:
		assert(filetype != S_IFDIR);
		return 0;
	}

	set_cflag(COMMIT_Freewmap, ip);

	/* mark transaction of block map update type */
	tblk = tid_to_tblock(tid);
	tblk->xflag |= COMMIT_PMAP;

	/*
	 * free EA
	 */
	if (JFS_IP(ip)->ea.flag & DXD_EXTENT)
		/* acquire maplock on EA to be freed from block map */
		txEA(tid, ip, &JFS_IP(ip)->ea, NULL);

	/*
	 * free ACL
	 */
	if (JFS_IP(ip)->acl.flag & DXD_EXTENT)
		/* acquire maplock on EA to be freed from block map */
		txEA(tid, ip, &JFS_IP(ip)->acl, NULL);

	/*
	 * free xtree/data (truncate to zero length):
D
Dave Kleikamp 已提交
666
	 * free xtree/data pages from cache if COMMIT_PWMAP,
L
Linus Torvalds 已提交
667 668 669 670 671 672 673 674 675 676 677
	 * free xtree/data blocks from persistent block map, and
	 * free xtree/data blocks from working block map if COMMIT_PWMAP;
	 */
	if (ip->i_size)
		return xtTruncate_pmap(tid, ip, 0);

	return 0;
}


/*
678
 * NAME:	jfs_free_zero_link()
L
Linus Torvalds 已提交
679
 *
680
 * FUNCTION:	for non-directory, called by iClose(),
D
Dave Kleikamp 已提交
681
 *		free resources of a file from cache and WORKING map
L
Linus Torvalds 已提交
682 683 684 685 686
 *		for a file previously committed with zero link count
 *		while associated with a pager object,
 *
 * PARAMETER:	ip	- pointer to inode of file.
 */
687
void jfs_free_zero_link(struct inode *ip)
L
Linus Torvalds 已提交
688 689 690
{
	int type;

691
	jfs_info("jfs_free_zero_link: ip = 0x%p", ip);
L
Linus Torvalds 已提交
692 693 694 695 696 697 698 699 700 701 702 703

	/* return if not reg or symbolic link or if size is
	 * already ok.
	 */
	type = ip->i_mode & S_IFMT;

	switch (type) {
	case S_IFREG:
		break;
	case S_IFLNK:
		/* if its contained in inode nothing to do */
		if (ip->i_size < IDATASIZE)
704
			return;
L
Linus Torvalds 已提交
705 706
		break;
	default:
707
		return;
L
Linus Torvalds 已提交
708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756
	}

	/*
	 * free EA
	 */
	if (JFS_IP(ip)->ea.flag & DXD_EXTENT) {
		s64 xaddr = addressDXD(&JFS_IP(ip)->ea);
		int xlen = lengthDXD(&JFS_IP(ip)->ea);
		struct maplock maplock;	/* maplock for COMMIT_WMAP */
		struct pxd_lock *pxdlock;	/* maplock for COMMIT_WMAP */

		/* free EA pages from cache */
		invalidate_dxd_metapages(ip, JFS_IP(ip)->ea);

		/* free EA extent from working block map */
		maplock.index = 1;
		pxdlock = (struct pxd_lock *) & maplock;
		pxdlock->flag = mlckFREEPXD;
		PXDaddress(&pxdlock->pxd, xaddr);
		PXDlength(&pxdlock->pxd, xlen);
		txFreeMap(ip, pxdlock, NULL, COMMIT_WMAP);
	}

	/*
	 * free ACL
	 */
	if (JFS_IP(ip)->acl.flag & DXD_EXTENT) {
		s64 xaddr = addressDXD(&JFS_IP(ip)->acl);
		int xlen = lengthDXD(&JFS_IP(ip)->acl);
		struct maplock maplock;	/* maplock for COMMIT_WMAP */
		struct pxd_lock *pxdlock;	/* maplock for COMMIT_WMAP */

		invalidate_dxd_metapages(ip, JFS_IP(ip)->acl);

		/* free ACL extent from working block map */
		maplock.index = 1;
		pxdlock = (struct pxd_lock *) & maplock;
		pxdlock->flag = mlckFREEPXD;
		PXDaddress(&pxdlock->pxd, xaddr);
		PXDlength(&pxdlock->pxd, xlen);
		txFreeMap(ip, pxdlock, NULL, COMMIT_WMAP);
	}

	/*
	 * free xtree/data (truncate to zero length):
	 * free xtree/data pages from cache, and
	 * free xtree/data blocks from working block map;
	 */
	if (ip->i_size)
757
		xtTruncate(0, ip, 0, COMMIT_WMAP);
L
Linus Torvalds 已提交
758 759 760 761 762 763 764 765
}

/*
 * NAME:	jfs_link(vp, dvp, name, crp)
 *
 * FUNCTION:	create a link to <vp> by the name = <name>
 *		in the parent directory <dvp>
 *
D
Dave Kleikamp 已提交
766
 * PARAMETER:	vp	- target object
L
Linus Torvalds 已提交
767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
 *		dvp	- parent directory of new link
 *		name	- name of new link to target object
 *		crp	- credential
 *
 * RETURN:	Errors from subroutines
 *
 * note:
 * JFS does NOT support link() on directories (to prevent circular
 * path in the directory hierarchy);
 * EPERM: the target object is a directory, and either the caller
 * does not have appropriate privileges or the implementation prohibits
 * using link() on directories [XPG4.2].
 *
 * JFS does NOT support links between file systems:
 * EXDEV: target object and new link are on different file systems and
 * implementation does not support links between file systems [XPG4.2].
 */
static int jfs_link(struct dentry *old_dentry,
	     struct inode *dir, struct dentry *dentry)
{
	int rc;
	tid_t tid;
	struct inode *ip = old_dentry->d_inode;
	ino_t ino;
	struct component_name dname;
	struct btstack btstack;
	struct inode *iplist[2];

	jfs_info("jfs_link: %s %s", old_dentry->d_name.name,
		 dentry->d_name.name);

	if (ip->i_nlink == JFS_LINK_MAX)
		return -EMLINK;

	if (ip->i_nlink == 0)
		return -ENOENT;

	tid = txBegin(ip->i_sb, 0);

D
Dave Kleikamp 已提交
806 807
	mutex_lock_nested(&JFS_IP(dir)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825

	/*
	 * scan parent directory for entry/freespace
	 */
	if ((rc = get_UCSname(&dname, dentry)))
		goto out;

	if ((rc = dtSearch(dir, &dname, &ino, &btstack, JFS_CREATE)))
		goto free_dname;

	/*
	 * create entry for new link in parent directory
	 */
	ino = ip->i_ino;
	if ((rc = dtInsert(tid, dir, &dname, &ino, &btstack)))
		goto free_dname;

	/* update object inode */
826
	inc_nlink(ip);		/* for new link */
L
Linus Torvalds 已提交
827
	ip->i_ctime = CURRENT_TIME;
828
	dir->i_ctime = dir->i_mtime = CURRENT_TIME;
L
Linus Torvalds 已提交
829 830 831 832 833 834 835 836
	mark_inode_dirty(dir);
	atomic_inc(&ip->i_count);

	iplist[0] = ip;
	iplist[1] = dir;
	rc = txCommit(tid, 2, &iplist[0], 0);

	if (rc) {
837
		ip->i_nlink--; /* never instantiated */
L
Linus Torvalds 已提交
838 839 840 841 842 843 844 845 846 847
		iput(ip);
	} else
		d_instantiate(dentry, ip);

      free_dname:
	free_UCSname(&dname);

      out:
	txEnd(tid);

848
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
849
	mutex_unlock(&JFS_IP(dir)->commit_mutex);
L
Linus Torvalds 已提交
850 851 852 853 854 855 856 857 858

	jfs_info("jfs_link: rc:%d", rc);
	return rc;
}

/*
 * NAME:	jfs_symlink(dip, dentry, name)
 *
 * FUNCTION:	creates a symbolic link to <symlink> by name <name>
859
 *			in directory <dip>
L
Linus Torvalds 已提交
860
 *
861 862 863 864
 * PARAMETER:	dip	- parent directory vnode
 *		dentry	- dentry of symbolic link
 *		name	- the path name of the existing object
 *			  that will be the source of the link
L
Linus Torvalds 已提交
865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909
 *
 * RETURN:	errors from subroutines
 *
 * note:
 * ENAMETOOLONG: pathname resolution of a symbolic link produced
 * an intermediate result whose length exceeds PATH_MAX [XPG4.2]
*/

static int jfs_symlink(struct inode *dip, struct dentry *dentry,
		const char *name)
{
	int rc;
	tid_t tid;
	ino_t ino = 0;
	struct component_name dname;
	int ssize;		/* source pathname size */
	struct btstack btstack;
	struct inode *ip = dentry->d_inode;
	unchar *i_fastsymlink;
	s64 xlen = 0;
	int bmask = 0, xsize;
	s64 extent = 0, xaddr;
	struct metapage *mp;
	struct super_block *sb;
	struct tblock *tblk;

	struct inode *iplist[2];

	jfs_info("jfs_symlink: dip:0x%p name:%s", dip, name);

	ssize = strlen(name) + 1;

	/*
	 * search parent directory for entry/freespace
	 * (dtSearch() returns parent directory page pinned)
	 */

	if ((rc = get_UCSname(&dname, dentry)))
		goto out1;

	/*
	 * allocate on-disk/in-memory inode for symbolic link:
	 * (iAlloc() returns new, locked inode)
	 */
	ip = ialloc(dip, S_IFLNK | 0777);
910 911
	if (IS_ERR(ip)) {
		rc = PTR_ERR(ip);
L
Linus Torvalds 已提交
912 913 914 915 916
		goto out2;
	}

	tid = txBegin(dip->i_sb, 0);

D
Dave Kleikamp 已提交
917 918
	mutex_lock_nested(&JFS_IP(dip)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
919

D
Dave Kleikamp 已提交
920 921 922 923
	rc = jfs_init_security(tid, ip, dip);
	if (rc)
		goto out3;

L
Linus Torvalds 已提交
924 925 926 927 928 929
	tblk = tid_to_tblock(tid);
	tblk->xflag |= COMMIT_CREATE;
	tblk->ino = ip->i_ino;
	tblk->u.ixpxd = JFS_IP(ip)->ixpxd;

	/* fix symlink access permission
D
Dave Kleikamp 已提交
930
	 * (dir_create() ANDs in the u.u_cmask,
L
Linus Torvalds 已提交
931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970
	 * but symlinks really need to be 777 access)
	 */
	ip->i_mode |= 0777;

	/*
	 * write symbolic link target path name
	 */
	xtInitRoot(tid, ip);

	/*
	 * write source path name inline in on-disk inode (fast symbolic link)
	 */

	if (ssize <= IDATASIZE) {
		ip->i_op = &jfs_symlink_inode_operations;

		i_fastsymlink = JFS_IP(ip)->i_inline;
		memcpy(i_fastsymlink, name, ssize);
		ip->i_size = ssize - 1;

		/*
		 * if symlink is > 128 bytes, we don't have the space to
		 * store inline extended attributes
		 */
		if (ssize > sizeof (JFS_IP(ip)->i_inline))
			JFS_IP(ip)->mode2 &= ~INLINEEA;

		jfs_info("jfs_symlink: fast symlink added  ssize:%d name:%s ",
			 ssize, name);
	}
	/*
	 * write source path name in a single extent
	 */
	else {
		jfs_info("jfs_symlink: allocate extent ip:0x%p", ip);

		ip->i_op = &page_symlink_inode_operations;
		ip->i_mapping->a_ops = &jfs_aops;

		/*
D
Dave Kleikamp 已提交
971
		 * even though the data of symlink object (source
L
Linus Torvalds 已提交
972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
		 * path name) is treated as non-journaled user data,
		 * it is read/written thru buffer cache for performance.
		 */
		sb = ip->i_sb;
		bmask = JFS_SBI(sb)->bsize - 1;
		xsize = (ssize + bmask) & ~bmask;
		xaddr = 0;
		xlen = xsize >> JFS_SBI(sb)->l2bsize;
		if ((rc = xtInsert(tid, ip, 0, 0, xlen, &xaddr, 0))) {
			txAbort(tid, 0);
			goto out3;
		}
		extent = xaddr;
		ip->i_size = ssize - 1;
		while (ssize) {
			/* This is kind of silly since PATH_MAX == 4K */
			int copy_size = min(ssize, PSIZE);

			mp = get_metapage(ip, xaddr, PSIZE, 1);

			if (mp == NULL) {
				xtTruncate(tid, ip, 0, COMMIT_PWMAP);
				rc = -EIO;
				txAbort(tid, 0);
				goto out3;
			}
			memcpy(mp->data, name, copy_size);
			flush_metapage(mp);
			ssize -= copy_size;
			name += copy_size;
			xaddr += JFS_SBI(sb)->nbperpage;
		}
	}

	/*
	 * create entry for symbolic link in parent directory
	 */
	rc = dtSearch(dip, &dname, &ino, &btstack, JFS_CREATE);
	if (rc == 0) {
		ino = ip->i_ino;
		rc = dtInsert(tid, dip, &dname, &ino, &btstack);
	}
	if (rc) {
		if (xlen)
			xtTruncate(tid, ip, 0, COMMIT_PWMAP);
		txAbort(tid, 0);
		/* discard new inode */
		goto out3;
	}

	insert_inode_hash(ip);
	mark_inode_dirty(ip);

1025 1026
	dip->i_ctime = dip->i_mtime = CURRENT_TIME;
	mark_inode_dirty(dip);
L
Linus Torvalds 已提交
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
	/*
	 * commit update of parent directory and link object
	 */

	iplist[0] = dip;
	iplist[1] = ip;
	rc = txCommit(tid, 2, &iplist[0], 0);

      out3:
	txEnd(tid);
1037
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
1038
	mutex_unlock(&JFS_IP(dip)->commit_mutex);
L
Linus Torvalds 已提交
1039
	if (rc) {
1040
		free_ea_wmap(ip);
L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
		ip->i_nlink = 0;
		iput(ip);
	} else
		d_instantiate(dentry, ip);

      out2:
	free_UCSname(&dname);

      out1:
	jfs_info("jfs_symlink: rc:%d", rc);
	return rc;
}


/*
1056
 * NAME:	jfs_rename
L
Linus Torvalds 已提交
1057
 *
1058
 * FUNCTION:	rename a file or directory
L
Linus Torvalds 已提交
1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130
 */
static int jfs_rename(struct inode *old_dir, struct dentry *old_dentry,
	       struct inode *new_dir, struct dentry *new_dentry)
{
	struct btstack btstack;
	ino_t ino;
	struct component_name new_dname;
	struct inode *new_ip;
	struct component_name old_dname;
	struct inode *old_ip;
	int rc;
	tid_t tid;
	struct tlock *tlck;
	struct dt_lock *dtlck;
	struct lv *lv;
	int ipcount;
	struct inode *iplist[4];
	struct tblock *tblk;
	s64 new_size = 0;
	int commit_flag;


	jfs_info("jfs_rename: %s %s", old_dentry->d_name.name,
		 new_dentry->d_name.name);

	old_ip = old_dentry->d_inode;
	new_ip = new_dentry->d_inode;

	if ((rc = get_UCSname(&old_dname, old_dentry)))
		goto out1;

	if ((rc = get_UCSname(&new_dname, new_dentry)))
		goto out2;

	/*
	 * Make sure source inode number is what we think it is
	 */
	rc = dtSearch(old_dir, &old_dname, &ino, &btstack, JFS_LOOKUP);
	if (rc || (ino != old_ip->i_ino)) {
		rc = -ENOENT;
		goto out3;
	}

	/*
	 * Make sure dest inode number (if any) is what we think it is
	 */
	rc = dtSearch(new_dir, &new_dname, &ino, &btstack, JFS_LOOKUP);
	if (rc == 0) {
		if ((new_ip == 0) || (ino != new_ip->i_ino)) {
			rc = -ESTALE;
			goto out3;
		}
	} else if (rc != -ENOENT)
		goto out3;
	else if (new_ip) {
		/* no entry exists, but one was expected */
		rc = -ESTALE;
		goto out3;
	}

	if (S_ISDIR(old_ip->i_mode)) {
		if (new_ip) {
			if (!dtEmpty(new_ip)) {
				rc = -ENOTEMPTY;
				goto out3;
			}
		} else if ((new_dir != old_dir) &&
			   (new_dir->i_nlink == JFS_LINK_MAX)) {
			rc = -EMLINK;
			goto out3;
		}
	} else if (new_ip) {
D
Dave Kleikamp 已提交
1131
		IWRITE_LOCK(new_ip, RDWRLOCK_NORMAL);
L
Linus Torvalds 已提交
1132 1133 1134 1135 1136 1137 1138 1139 1140
		/* Init inode for quota operations. */
		DQUOT_INIT(new_ip);
	}

	/*
	 * The real work starts here
	 */
	tid = txBegin(new_dir->i_sb, 0);

D
Dave Kleikamp 已提交
1141 1142 1143 1144 1145 1146 1147 1148
	/*
	 * How do we know the locking is safe from deadlocks?
	 * The vfs does the hard part for us.  Any time we are taking nested
	 * commit_mutexes, the vfs already has i_mutex held on the parent.
	 * Here, the vfs has already taken i_mutex on both old_dir and new_dir.
	 */
	mutex_lock_nested(&JFS_IP(new_dir)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(old_ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
1149
	if (old_dir != new_dir)
D
Dave Kleikamp 已提交
1150 1151
		mutex_lock_nested(&JFS_IP(old_dir)->commit_mutex,
				  COMMIT_MUTEX_SECOND_PARENT);
L
Linus Torvalds 已提交
1152 1153

	if (new_ip) {
D
Dave Kleikamp 已提交
1154 1155
		mutex_lock_nested(&JFS_IP(new_ip)->commit_mutex,
				  COMMIT_MUTEX_VICTIM);
L
Linus Torvalds 已提交
1156 1157 1158 1159 1160 1161 1162 1163
		/*
		 * Change existing directory entry to new inode number
		 */
		ino = new_ip->i_ino;
		rc = dtModify(tid, new_dir, &new_dname, &ino,
			      old_ip->i_ino, JFS_RENAME);
		if (rc)
			goto out4;
1164
		drop_nlink(new_ip);
L
Linus Torvalds 已提交
1165
		if (S_ISDIR(new_ip->i_mode)) {
1166
			drop_nlink(new_ip);
L
Linus Torvalds 已提交
1167
			if (new_ip->i_nlink) {
E
Evgeniy Dushistov 已提交
1168
				mutex_unlock(&JFS_IP(new_ip)->commit_mutex);
L
Linus Torvalds 已提交
1169
				if (old_dir != new_dir)
1170
					mutex_unlock(&JFS_IP(old_dir)->commit_mutex);
E
Evgeniy Dushistov 已提交
1171 1172
				mutex_unlock(&JFS_IP(old_ip)->commit_mutex);
				mutex_unlock(&JFS_IP(new_dir)->commit_mutex);
L
Linus Torvalds 已提交
1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186
				if (!S_ISDIR(old_ip->i_mode) && new_ip)
					IWRITE_UNLOCK(new_ip);
				jfs_error(new_ip->i_sb,
					  "jfs_rename: new_ip->i_nlink != 0");
				return -EIO;
			}
			tblk = tid_to_tblock(tid);
			tblk->xflag |= COMMIT_DELETE;
			tblk->u.ip = new_ip;
		} else if (new_ip->i_nlink == 0) {
			assert(!test_cflag(COMMIT_Nolink, new_ip));
			/* free block resources */
			if ((new_size = commitZeroLink(tid, new_ip)) < 0) {
				txAbort(tid, 1);	/* Marks FS Dirty */
D
Dave Kleikamp 已提交
1187
				rc = new_size;
L
Linus Torvalds 已提交
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216
				goto out4;
			}
			tblk = tid_to_tblock(tid);
			tblk->xflag |= COMMIT_DELETE;
			tblk->u.ip = new_ip;
		} else {
			new_ip->i_ctime = CURRENT_TIME;
			mark_inode_dirty(new_ip);
		}
	} else {
		/*
		 * Add new directory entry
		 */
		rc = dtSearch(new_dir, &new_dname, &ino, &btstack,
			      JFS_CREATE);
		if (rc) {
			jfs_err("jfs_rename didn't expect dtSearch to fail "
				"w/rc = %d", rc);
			goto out4;
		}

		ino = old_ip->i_ino;
		rc = dtInsert(tid, new_dir, &new_dname, &ino, &btstack);
		if (rc) {
			if (rc == -EIO)
				jfs_err("jfs_rename: dtInsert returned -EIO");
			goto out4;
		}
		if (S_ISDIR(old_ip->i_mode))
1217
			inc_nlink(new_dir);
L
Linus Torvalds 已提交
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
	}
	/*
	 * Remove old directory entry
	 */

	ino = old_ip->i_ino;
	rc = dtDelete(tid, old_dir, &old_dname, &ino, JFS_REMOVE);
	if (rc) {
		jfs_err("jfs_rename did not expect dtDelete to return rc = %d",
			rc);
		txAbort(tid, 1);	/* Marks Filesystem dirty */
		goto out4;
	}
	if (S_ISDIR(old_ip->i_mode)) {
1232
		drop_nlink(old_dir);
L
Linus Torvalds 已提交
1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290
		if (old_dir != new_dir) {
			/*
			 * Change inode number of parent for moved directory
			 */

			JFS_IP(old_ip)->i_dtroot.header.idotdot =
				cpu_to_le32(new_dir->i_ino);

			/* Linelock header of dtree */
			tlck = txLock(tid, old_ip,
				    (struct metapage *) &JFS_IP(old_ip)->bxflag,
				      tlckDTREE | tlckBTROOT | tlckRELINK);
			dtlck = (struct dt_lock *) & tlck->lock;
			ASSERT(dtlck->index == 0);
			lv = & dtlck->lv[0];
			lv->offset = 0;
			lv->length = 1;
			dtlck->index++;
		}
	}

	/*
	 * Update ctime on changed/moved inodes & mark dirty
	 */
	old_ip->i_ctime = CURRENT_TIME;
	mark_inode_dirty(old_ip);

	new_dir->i_ctime = new_dir->i_mtime = current_fs_time(new_dir->i_sb);
	mark_inode_dirty(new_dir);

	/* Build list of inodes modified by this transaction */
	ipcount = 0;
	iplist[ipcount++] = old_ip;
	if (new_ip)
		iplist[ipcount++] = new_ip;
	iplist[ipcount++] = old_dir;

	if (old_dir != new_dir) {
		iplist[ipcount++] = new_dir;
		old_dir->i_ctime = old_dir->i_mtime = CURRENT_TIME;
		mark_inode_dirty(old_dir);
	}

	/*
	 * Incomplete truncate of file data can
	 * result in timing problems unless we synchronously commit the
	 * transaction.
	 */
	if (new_size)
		commit_flag = COMMIT_SYNC;
	else
		commit_flag = 0;

	rc = txCommit(tid, ipcount, iplist, commit_flag);

      out4:
	txEnd(tid);
	if (new_ip)
1291
		mutex_unlock(&JFS_IP(new_ip)->commit_mutex);
E
Evgeniy Dushistov 已提交
1292 1293 1294 1295
	if (old_dir != new_dir)
		mutex_unlock(&JFS_IP(old_dir)->commit_mutex);
	mutex_unlock(&JFS_IP(old_ip)->commit_mutex);
	mutex_unlock(&JFS_IP(new_dir)->commit_mutex);
L
Linus Torvalds 已提交
1296 1297 1298

	while (new_size && (rc == 0)) {
		tid = txBegin(new_ip->i_sb, 0);
1299
		mutex_lock(&JFS_IP(new_ip)->commit_mutex);
L
Linus Torvalds 已提交
1300 1301 1302
		new_size = xtTruncate_pmap(tid, new_ip, new_size);
		if (new_size < 0) {
			txAbort(tid, 1);
D
Dave Kleikamp 已提交
1303
			rc = new_size;
L
Linus Torvalds 已提交
1304 1305 1306
		} else
			rc = txCommit(tid, 1, &new_ip, COMMIT_SYNC);
		txEnd(tid);
1307
		mutex_unlock(&JFS_IP(new_ip)->commit_mutex);
L
Linus Torvalds 已提交
1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
	}
	if (new_ip && (new_ip->i_nlink == 0))
		set_cflag(COMMIT_Nolink, new_ip);
      out3:
	free_UCSname(&new_dname);
      out2:
	free_UCSname(&old_dname);
      out1:
	if (new_ip && !S_ISDIR(new_ip->i_mode))
		IWRITE_UNLOCK(new_ip);
	/*
	 * Truncating the directory index table is not guaranteed.  It
	 * may need to be done iteratively
	 */
	if (test_cflag(COMMIT_Stale, old_dir)) {
		if (old_dir->i_size > 1)
			jfs_truncate_nolock(old_dir, 0);

		clear_cflag(COMMIT_Stale, old_dir);
	}

	jfs_info("jfs_rename: returning %d", rc);
	return rc;
}


/*
1335
 * NAME:	jfs_mknod
L
Linus Torvalds 已提交
1336
 *
1337
 * FUNCTION:	Create a special file (device)
L
Linus Torvalds 已提交
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
 */
static int jfs_mknod(struct inode *dir, struct dentry *dentry,
		int mode, dev_t rdev)
{
	struct jfs_inode_info *jfs_ip;
	struct btstack btstack;
	struct component_name dname;
	ino_t ino;
	struct inode *ip;
	struct inode *iplist[2];
	int rc;
	tid_t tid;
	struct tblock *tblk;

	if (!new_valid_dev(rdev))
		return -EINVAL;

	jfs_info("jfs_mknod: %s", dentry->d_name.name);

	if ((rc = get_UCSname(&dname, dentry)))
		goto out;

	ip = ialloc(dir, mode);
1361 1362
	if (IS_ERR(ip)) {
		rc = PTR_ERR(ip);
L
Linus Torvalds 已提交
1363 1364 1365 1366 1367 1368
		goto out1;
	}
	jfs_ip = JFS_IP(ip);

	tid = txBegin(dir->i_sb, 0);

D
Dave Kleikamp 已提交
1369 1370
	mutex_lock_nested(&JFS_IP(dir)->commit_mutex, COMMIT_MUTEX_PARENT);
	mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
L
Linus Torvalds 已提交
1371

1372 1373
	rc = jfs_init_acl(tid, ip, dir);
	if (rc)
L
Linus Torvalds 已提交
1374 1375
		goto out3;

D
Dave Kleikamp 已提交
1376 1377 1378 1379 1380 1381
	rc = jfs_init_security(tid, ip, dir);
	if (rc) {
		txAbort(tid, 0);
		goto out3;
	}

1382 1383 1384 1385 1386
	if ((rc = dtSearch(dir, &dname, &ino, &btstack, JFS_CREATE))) {
		txAbort(tid, 0);
		goto out3;
	}

L
Linus Torvalds 已提交
1387 1388 1389 1390 1391 1392
	tblk = tid_to_tblock(tid);
	tblk->xflag |= COMMIT_CREATE;
	tblk->ino = ip->i_ino;
	tblk->u.ixpxd = JFS_IP(ip)->ixpxd;

	ino = ip->i_ino;
1393 1394
	if ((rc = dtInsert(tid, dir, &dname, &ino, &btstack))) {
		txAbort(tid, 0);
L
Linus Torvalds 已提交
1395
		goto out3;
1396
	}
L
Linus Torvalds 已提交
1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414

	ip->i_op = &jfs_file_inode_operations;
	jfs_ip->dev = new_encode_dev(rdev);
	init_special_inode(ip, ip->i_mode, rdev);

	insert_inode_hash(ip);
	mark_inode_dirty(ip);

	dir->i_ctime = dir->i_mtime = CURRENT_TIME;

	mark_inode_dirty(dir);

	iplist[0] = dir;
	iplist[1] = ip;
	rc = txCommit(tid, 2, iplist, 0);

      out3:
	txEnd(tid);
1415 1416
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
	mutex_unlock(&JFS_IP(dir)->commit_mutex);
L
Linus Torvalds 已提交
1417
	if (rc) {
1418
		free_ea_wmap(ip);
L
Linus Torvalds 已提交
1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443
		ip->i_nlink = 0;
		iput(ip);
	} else
		d_instantiate(dentry, ip);

      out1:
	free_UCSname(&dname);

      out:
	jfs_info("jfs_mknod: returning %d", rc);
	return rc;
}

static struct dentry *jfs_lookup(struct inode *dip, struct dentry *dentry, struct nameidata *nd)
{
	struct btstack btstack;
	ino_t inum;
	struct inode *ip;
	struct component_name key;
	const char *name = dentry->d_name.name;
	int len = dentry->d_name.len;
	int rc;

	jfs_info("jfs_lookup: name = %s", name);

1444 1445
	if (JFS_SBI(dip->i_sb)->mntflag & JFS_OS2)
		dentry->d_op = &jfs_ci_dentry_operations;
L
Linus Torvalds 已提交
1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480

	if ((name[0] == '.') && (len == 1))
		inum = dip->i_ino;
	else if (strcmp(name, "..") == 0)
		inum = PARENT(dip);
	else {
		if ((rc = get_UCSname(&key, dentry)))
			return ERR_PTR(rc);
		rc = dtSearch(dip, &key, &inum, &btstack, JFS_LOOKUP);
		free_UCSname(&key);
		if (rc == -ENOENT) {
			d_add(dentry, NULL);
			return ERR_PTR(0);
		} else if (rc) {
			jfs_err("jfs_lookup: dtSearch returned %d", rc);
			return ERR_PTR(rc);
		}
	}

	ip = iget(dip->i_sb, inum);
	if (ip == NULL || is_bad_inode(ip)) {
		jfs_err("jfs_lookup: iget failed on inum %d", (uint) inum);
		if (ip)
			iput(ip);
		return ERR_PTR(-EACCES);
	}

	dentry = d_splice_alias(ip, dentry);

	if (dentry && (JFS_SBI(dip->i_sb)->mntflag & JFS_OS2))
		dentry->d_op = &jfs_ci_dentry_operations;

	return dentry;
}

C
Christoph Hellwig 已提交
1481 1482
static struct inode *jfs_nfs_get_inode(struct super_block *sb,
		u64 ino, u32 generation)
1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
{
	struct inode *inode;

	if (ino == 0)
		return ERR_PTR(-ESTALE);
	inode = iget(sb, ino);
	if (inode == NULL)
		return ERR_PTR(-ENOMEM);

	if (is_bad_inode(inode) ||
	    (generation && inode->i_generation != generation)) {
C
Christoph Hellwig 已提交
1494 1495
		iput(inode);
		return ERR_PTR(-ESTALE);
1496 1497
	}

C
Christoph Hellwig 已提交
1498 1499
	return inode;
}
1500

C
Christoph Hellwig 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512
struct dentry *jfs_fh_to_dentry(struct super_block *sb, struct fid *fid,
		int fh_len, int fh_type)
{
	return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
				    jfs_nfs_get_inode);
}

struct dentry *jfs_fh_to_parent(struct super_block *sb, struct fid *fid,
		int fh_len, int fh_type)
{
	return generic_fh_to_parent(sb, fid, fh_len, fh_type,
				    jfs_nfs_get_inode);
1513 1514
}

L
Linus Torvalds 已提交
1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540
struct dentry *jfs_get_parent(struct dentry *dentry)
{
	struct super_block *sb = dentry->d_inode->i_sb;
	struct dentry *parent = ERR_PTR(-ENOENT);
	struct inode *inode;
	unsigned long parent_ino;

	parent_ino =
		le32_to_cpu(JFS_IP(dentry->d_inode)->i_dtroot.header.idotdot);
	inode = iget(sb, parent_ino);
	if (inode) {
		if (is_bad_inode(inode)) {
			iput(inode);
			parent = ERR_PTR(-EACCES);
		} else {
			parent = d_alloc_anon(inode);
			if (!parent) {
				parent = ERR_PTR(-ENOMEM);
				iput(inode);
			}
		}
	}

	return parent;
}

1541
const struct inode_operations jfs_dir_inode_operations = {
L
Linus Torvalds 已提交
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
	.create		= jfs_create,
	.lookup		= jfs_lookup,
	.link		= jfs_link,
	.unlink		= jfs_unlink,
	.symlink	= jfs_symlink,
	.mkdir		= jfs_mkdir,
	.rmdir		= jfs_rmdir,
	.mknod		= jfs_mknod,
	.rename		= jfs_rename,
	.setxattr	= jfs_setxattr,
	.getxattr	= jfs_getxattr,
	.listxattr	= jfs_listxattr,
	.removexattr	= jfs_removexattr,
#ifdef CONFIG_JFS_POSIX_ACL
	.setattr	= jfs_setattr,
	.permission	= jfs_permission,
#endif
};

1561
const struct file_operations jfs_dir_operations = {
L
Linus Torvalds 已提交
1562 1563 1564
	.read		= generic_read_dir,
	.readdir	= jfs_readdir,
	.fsync		= jfs_fsync,
1565
	.ioctl		= jfs_ioctl,
L
Linus Torvalds 已提交
1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609
};

static int jfs_ci_hash(struct dentry *dir, struct qstr *this)
{
	unsigned long hash;
	int i;

	hash = init_name_hash();
	for (i=0; i < this->len; i++)
		hash = partial_name_hash(tolower(this->name[i]), hash);
	this->hash = end_name_hash(hash);

	return 0;
}

static int jfs_ci_compare(struct dentry *dir, struct qstr *a, struct qstr *b)
{
	int i, result = 1;

	if (a->len != b->len)
		goto out;
	for (i=0; i < a->len; i++) {
		if (tolower(a->name[i]) != tolower(b->name[i]))
			goto out;
	}
	result = 0;

	/*
	 * We want creates to preserve case.  A negative dentry, a, that
	 * has a different case than b may cause a new entry to be created
	 * with the wrong case.  Since we can't tell if a comes from a negative
	 * dentry, we blindly replace it with b.  This should be harmless if
	 * a is not a negative dentry.
	 */
	memcpy((unsigned char *)a->name, b->name, a->len);
out:
	return result;
}

struct dentry_operations jfs_ci_dentry_operations =
{
	.d_hash = jfs_ci_hash,
	.d_compare = jfs_ci_compare,
};