Linux Audio

Check our new training course

Loading...
v4.17
 
   1/*
   2 *   Copyright (C) International Business Machines Corp., 2000-2004
   3 *
   4 *   This program is free software;  you can redistribute it and/or modify
   5 *   it under the terms of the GNU General Public License as published by
   6 *   the Free Software Foundation; either version 2 of the License, or
   7 *   (at your option) any later version.
   8 *
   9 *   This program is distributed in the hope that it will be useful,
  10 *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
  11 *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
  12 *   the GNU General Public License for more details.
  13 *
  14 *   You should have received a copy of the GNU General Public License
  15 *   along with this program;  if not, write to the Free Software
  16 *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  17 */
  18
  19/*
  20 *	jfs_imap.c: inode allocation map manager
  21 *
  22 * Serialization:
  23 *   Each AG has a simple lock which is used to control the serialization of
  24 *	the AG level lists.  This lock should be taken first whenever an AG
  25 *	level list will be modified or accessed.
  26 *
  27 *   Each IAG is locked by obtaining the buffer for the IAG page.
  28 *
  29 *   There is also a inode lock for the inode map inode.  A read lock needs to
  30 *	be taken whenever an IAG is read from the map or the global level
  31 *	information is read.  A write lock needs to be taken whenever the global
  32 *	level information is modified or an atomic operation needs to be used.
  33 *
  34 *	If more than one IAG is read at one time, the read lock may not
  35 *	be given up until all of the IAG's are read.  Otherwise, a deadlock
  36 *	may occur when trying to obtain the read lock while another thread
  37 *	holding the read lock is waiting on the IAG already being held.
  38 *
  39 *   The control page of the inode map is read into memory by diMount().
  40 *	Thereafter it should only be modified in memory and then it will be
  41 *	written out when the filesystem is unmounted by diUnmount().
  42 */
  43
  44#include <linux/fs.h>
  45#include <linux/buffer_head.h>
  46#include <linux/pagemap.h>
  47#include <linux/quotaops.h>
  48#include <linux/slab.h>
  49
  50#include "jfs_incore.h"
  51#include "jfs_inode.h"
  52#include "jfs_filsys.h"
  53#include "jfs_dinode.h"
  54#include "jfs_dmap.h"
  55#include "jfs_imap.h"
  56#include "jfs_metapage.h"
  57#include "jfs_superblock.h"
  58#include "jfs_debug.h"
  59
  60/*
  61 * imap locks
  62 */
  63/* iag free list lock */
  64#define IAGFREE_LOCK_INIT(imap)		mutex_init(&imap->im_freelock)
  65#define IAGFREE_LOCK(imap)		mutex_lock(&imap->im_freelock)
  66#define IAGFREE_UNLOCK(imap)		mutex_unlock(&imap->im_freelock)
  67
  68/* per ag iag list locks */
  69#define AG_LOCK_INIT(imap,index)	mutex_init(&(imap->im_aglock[index]))
  70#define AG_LOCK(imap,agno)		mutex_lock(&imap->im_aglock[agno])
  71#define AG_UNLOCK(imap,agno)		mutex_unlock(&imap->im_aglock[agno])
  72
  73/*
  74 * forward references
  75 */
  76static int diAllocAG(struct inomap *, int, bool, struct inode *);
  77static int diAllocAny(struct inomap *, int, bool, struct inode *);
  78static int diAllocBit(struct inomap *, struct iag *, int);
  79static int diAllocExt(struct inomap *, int, struct inode *);
  80static int diAllocIno(struct inomap *, int, struct inode *);
  81static int diFindFree(u32, int);
  82static int diNewExt(struct inomap *, struct iag *, int);
  83static int diNewIAG(struct inomap *, int *, int, struct metapage **);
  84static void duplicateIXtree(struct super_block *, s64, int, s64 *);
  85
  86static int diIAGRead(struct inomap * imap, int, struct metapage **);
  87static int copy_from_dinode(struct dinode *, struct inode *);
  88static void copy_to_dinode(struct dinode *, struct inode *);
  89
  90/*
  91 * NAME:	diMount()
  92 *
  93 * FUNCTION:	initialize the incore inode map control structures for
  94 *		a fileset or aggregate init time.
  95 *
  96 *		the inode map's control structure (dinomap) is
  97 *		brought in from disk and placed in virtual memory.
  98 *
  99 * PARAMETERS:
 100 *	ipimap	- pointer to inode map inode for the aggregate or fileset.
 101 *
 102 * RETURN VALUES:
 103 *	0	- success
 104 *	-ENOMEM	- insufficient free virtual memory.
 105 *	-EIO	- i/o error.
 106 */
 107int diMount(struct inode *ipimap)
 108{
 109	struct inomap *imap;
 110	struct metapage *mp;
 111	int index;
 112	struct dinomap_disk *dinom_le;
 113
 114	/*
 115	 * allocate/initialize the in-memory inode map control structure
 116	 */
 117	/* allocate the in-memory inode map control structure. */
 118	imap = kmalloc(sizeof(struct inomap), GFP_KERNEL);
 119	if (imap == NULL) {
 120		jfs_err("diMount: kmalloc returned NULL!");
 121		return -ENOMEM;
 122	}
 123
 124	/* read the on-disk inode map control structure. */
 125
 126	mp = read_metapage(ipimap,
 127			   IMAPBLKNO << JFS_SBI(ipimap->i_sb)->l2nbperpage,
 128			   PSIZE, 0);
 129	if (mp == NULL) {
 130		kfree(imap);
 131		return -EIO;
 132	}
 133
 134	/* copy the on-disk version to the in-memory version. */
 135	dinom_le = (struct dinomap_disk *) mp->data;
 136	imap->im_freeiag = le32_to_cpu(dinom_le->in_freeiag);
 137	imap->im_nextiag = le32_to_cpu(dinom_le->in_nextiag);
 138	atomic_set(&imap->im_numinos, le32_to_cpu(dinom_le->in_numinos));
 139	atomic_set(&imap->im_numfree, le32_to_cpu(dinom_le->in_numfree));
 140	imap->im_nbperiext = le32_to_cpu(dinom_le->in_nbperiext);
 141	imap->im_l2nbperiext = le32_to_cpu(dinom_le->in_l2nbperiext);
 142	for (index = 0; index < MAXAG; index++) {
 143		imap->im_agctl[index].inofree =
 144		    le32_to_cpu(dinom_le->in_agctl[index].inofree);
 145		imap->im_agctl[index].extfree =
 146		    le32_to_cpu(dinom_le->in_agctl[index].extfree);
 147		imap->im_agctl[index].numinos =
 148		    le32_to_cpu(dinom_le->in_agctl[index].numinos);
 149		imap->im_agctl[index].numfree =
 150		    le32_to_cpu(dinom_le->in_agctl[index].numfree);
 151	}
 152
 153	/* release the buffer. */
 154	release_metapage(mp);
 155
 156	/*
 157	 * allocate/initialize inode allocation map locks
 158	 */
 159	/* allocate and init iag free list lock */
 160	IAGFREE_LOCK_INIT(imap);
 161
 162	/* allocate and init ag list locks */
 163	for (index = 0; index < MAXAG; index++) {
 164		AG_LOCK_INIT(imap, index);
 165	}
 166
 167	/* bind the inode map inode and inode map control structure
 168	 * to each other.
 169	 */
 170	imap->im_ipimap = ipimap;
 171	JFS_IP(ipimap)->i_imap = imap;
 172
 173	return (0);
 174}
 175
 176
 177/*
 178 * NAME:	diUnmount()
 179 *
 180 * FUNCTION:	write to disk the incore inode map control structures for
 181 *		a fileset or aggregate at unmount time.
 182 *
 183 * PARAMETERS:
 184 *	ipimap	- pointer to inode map inode for the aggregate or fileset.
 185 *
 186 * RETURN VALUES:
 187 *	0	- success
 188 *	-ENOMEM	- insufficient free virtual memory.
 189 *	-EIO	- i/o error.
 190 */
 191int diUnmount(struct inode *ipimap, int mounterror)
 192{
 193	struct inomap *imap = JFS_IP(ipimap)->i_imap;
 194
 195	/*
 196	 * update the on-disk inode map control structure
 197	 */
 198
 199	if (!(mounterror || isReadOnly(ipimap)))
 200		diSync(ipimap);
 201
 202	/*
 203	 * Invalidate the page cache buffers
 204	 */
 205	truncate_inode_pages(ipimap->i_mapping, 0);
 206
 207	/*
 208	 * free in-memory control structure
 209	 */
 210	kfree(imap);
 
 211
 212	return (0);
 213}
 214
 215
 216/*
 217 *	diSync()
 218 */
 219int diSync(struct inode *ipimap)
 220{
 221	struct dinomap_disk *dinom_le;
 222	struct inomap *imp = JFS_IP(ipimap)->i_imap;
 223	struct metapage *mp;
 224	int index;
 225
 226	/*
 227	 * write imap global conrol page
 228	 */
 229	/* read the on-disk inode map control structure */
 230	mp = get_metapage(ipimap,
 231			  IMAPBLKNO << JFS_SBI(ipimap->i_sb)->l2nbperpage,
 232			  PSIZE, 0);
 233	if (mp == NULL) {
 234		jfs_err("diSync: get_metapage failed!");
 235		return -EIO;
 236	}
 237
 238	/* copy the in-memory version to the on-disk version */
 239	dinom_le = (struct dinomap_disk *) mp->data;
 240	dinom_le->in_freeiag = cpu_to_le32(imp->im_freeiag);
 241	dinom_le->in_nextiag = cpu_to_le32(imp->im_nextiag);
 242	dinom_le->in_numinos = cpu_to_le32(atomic_read(&imp->im_numinos));
 243	dinom_le->in_numfree = cpu_to_le32(atomic_read(&imp->im_numfree));
 244	dinom_le->in_nbperiext = cpu_to_le32(imp->im_nbperiext);
 245	dinom_le->in_l2nbperiext = cpu_to_le32(imp->im_l2nbperiext);
 246	for (index = 0; index < MAXAG; index++) {
 247		dinom_le->in_agctl[index].inofree =
 248		    cpu_to_le32(imp->im_agctl[index].inofree);
 249		dinom_le->in_agctl[index].extfree =
 250		    cpu_to_le32(imp->im_agctl[index].extfree);
 251		dinom_le->in_agctl[index].numinos =
 252		    cpu_to_le32(imp->im_agctl[index].numinos);
 253		dinom_le->in_agctl[index].numfree =
 254		    cpu_to_le32(imp->im_agctl[index].numfree);
 255	}
 256
 257	/* write out the control structure */
 258	write_metapage(mp);
 259
 260	/*
 261	 * write out dirty pages of imap
 262	 */
 263	filemap_write_and_wait(ipimap->i_mapping);
 264
 265	diWriteSpecial(ipimap, 0);
 266
 267	return (0);
 268}
 269
 270
 271/*
 272 * NAME:	diRead()
 273 *
 274 * FUNCTION:	initialize an incore inode from disk.
 275 *
 276 *		on entry, the specifed incore inode should itself
 277 *		specify the disk inode number corresponding to the
 278 *		incore inode (i.e. i_number should be initialized).
 279 *
 280 *		this routine handles incore inode initialization for
 281 *		both "special" and "regular" inodes.  special inodes
 282 *		are those required early in the mount process and
 283 *		require special handling since much of the file system
 284 *		is not yet initialized.  these "special" inodes are
 285 *		identified by a NULL inode map inode pointer and are
 286 *		actually initialized by a call to diReadSpecial().
 287 *
 288 *		for regular inodes, the iag describing the disk inode
 289 *		is read from disk to determine the inode extent address
 290 *		for the disk inode.  with the inode extent address in
 291 *		hand, the page of the extent that contains the disk
 292 *		inode is read and the disk inode is copied to the
 293 *		incore inode.
 294 *
 295 * PARAMETERS:
 296 *	ip	-  pointer to incore inode to be initialized from disk.
 297 *
 298 * RETURN VALUES:
 299 *	0	- success
 300 *	-EIO	- i/o error.
 301 *	-ENOMEM	- insufficient memory
 302 *
 303 */
 304int diRead(struct inode *ip)
 305{
 306	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
 307	int iagno, ino, extno, rc;
 308	struct inode *ipimap;
 309	struct dinode *dp;
 310	struct iag *iagp;
 311	struct metapage *mp;
 312	s64 blkno, agstart;
 313	struct inomap *imap;
 314	int block_offset;
 315	int inodes_left;
 316	unsigned long pageno;
 317	int rel_inode;
 318
 319	jfs_info("diRead: ino = %ld", ip->i_ino);
 320
 321	ipimap = sbi->ipimap;
 322	JFS_IP(ip)->ipimap = ipimap;
 323
 324	/* determine the iag number for this inode (number) */
 325	iagno = INOTOIAG(ip->i_ino);
 326
 327	/* read the iag */
 328	imap = JFS_IP(ipimap)->i_imap;
 329	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
 
 330	rc = diIAGRead(imap, iagno, &mp);
 331	IREAD_UNLOCK(ipimap);
 332	if (rc) {
 333		jfs_err("diRead: diIAGRead returned %d", rc);
 334		return (rc);
 335	}
 336
 337	iagp = (struct iag *) mp->data;
 338
 339	/* determine inode extent that holds the disk inode */
 340	ino = ip->i_ino & (INOSPERIAG - 1);
 341	extno = ino >> L2INOSPEREXT;
 342
 343	if ((lengthPXD(&iagp->inoext[extno]) != imap->im_nbperiext) ||
 344	    (addressPXD(&iagp->inoext[extno]) == 0)) {
 345		release_metapage(mp);
 346		return -ESTALE;
 347	}
 348
 349	/* get disk block number of the page within the inode extent
 350	 * that holds the disk inode.
 351	 */
 352	blkno = INOPBLK(&iagp->inoext[extno], ino, sbi->l2nbperpage);
 353
 354	/* get the ag for the iag */
 355	agstart = le64_to_cpu(iagp->agstart);
 
 356
 357	release_metapage(mp);
 
 
 358
 359	rel_inode = (ino & (INOSPERPAGE - 1));
 360	pageno = blkno >> sbi->l2nbperpage;
 361
 362	if ((block_offset = ((u32) blkno & (sbi->nbperpage - 1)))) {
 363		/*
 364		 * OS/2 didn't always align inode extents on page boundaries
 365		 */
 366		inodes_left =
 367		     (sbi->nbperpage - block_offset) << sbi->l2niperblk;
 368
 369		if (rel_inode < inodes_left)
 370			rel_inode += block_offset << sbi->l2niperblk;
 371		else {
 372			pageno += 1;
 373			rel_inode -= inodes_left;
 374		}
 375	}
 376
 377	/* read the page of disk inode */
 378	mp = read_metapage(ipimap, pageno << sbi->l2nbperpage, PSIZE, 1);
 379	if (!mp) {
 380		jfs_err("diRead: read_metapage failed");
 381		return -EIO;
 382	}
 383
 384	/* locate the disk inode requested */
 385	dp = (struct dinode *) mp->data;
 386	dp += rel_inode;
 387
 388	if (ip->i_ino != le32_to_cpu(dp->di_number)) {
 389		jfs_error(ip->i_sb, "i_ino != di_number\n");
 390		rc = -EIO;
 391	} else if (le32_to_cpu(dp->di_nlink) == 0)
 392		rc = -ESTALE;
 393	else
 394		/* copy the disk inode to the in-memory inode */
 395		rc = copy_from_dinode(dp, ip);
 396
 397	release_metapage(mp);
 398
 399	/* set the ag for the inode */
 400	JFS_IP(ip)->agstart = agstart;
 401	JFS_IP(ip)->active_ag = -1;
 402
 403	return (rc);
 404}
 405
 406
 407/*
 408 * NAME:	diReadSpecial()
 409 *
 410 * FUNCTION:	initialize a 'special' inode from disk.
 411 *
 412 *		this routines handles aggregate level inodes.  The
 413 *		inode cache cannot differentiate between the
 414 *		aggregate inodes and the filesystem inodes, so we
 415 *		handle these here.  We don't actually use the aggregate
 416 *		inode map, since these inodes are at a fixed location
 417 *		and in some cases the aggregate inode map isn't initialized
 418 *		yet.
 419 *
 420 * PARAMETERS:
 421 *	sb - filesystem superblock
 422 *	inum - aggregate inode number
 423 *	secondary - 1 if secondary aggregate inode table
 424 *
 425 * RETURN VALUES:
 426 *	new inode	- success
 427 *	NULL		- i/o error.
 428 */
 429struct inode *diReadSpecial(struct super_block *sb, ino_t inum, int secondary)
 430{
 431	struct jfs_sb_info *sbi = JFS_SBI(sb);
 432	uint address;
 433	struct dinode *dp;
 434	struct inode *ip;
 435	struct metapage *mp;
 436
 437	ip = new_inode(sb);
 438	if (ip == NULL) {
 439		jfs_err("diReadSpecial: new_inode returned NULL!");
 440		return ip;
 441	}
 442
 443	if (secondary) {
 444		address = addressPXD(&sbi->ait2) >> sbi->l2nbperpage;
 445		JFS_IP(ip)->ipimap = sbi->ipaimap2;
 446	} else {
 447		address = AITBL_OFF >> L2PSIZE;
 448		JFS_IP(ip)->ipimap = sbi->ipaimap;
 449	}
 450
 451	ASSERT(inum < INOSPEREXT);
 452
 453	ip->i_ino = inum;
 454
 455	address += inum >> 3;	/* 8 inodes per 4K page */
 456
 457	/* read the page of fixed disk inode (AIT) in raw mode */
 458	mp = read_metapage(ip, address << sbi->l2nbperpage, PSIZE, 1);
 459	if (mp == NULL) {
 460		set_nlink(ip, 1);	/* Don't want iput() deleting it */
 461		iput(ip);
 462		return (NULL);
 463	}
 464
 465	/* get the pointer to the disk inode of interest */
 466	dp = (struct dinode *) (mp->data);
 467	dp += inum % 8;		/* 8 inodes per 4K page */
 468
 469	/* copy on-disk inode to in-memory inode */
 470	if ((copy_from_dinode(dp, ip)) != 0) {
 471		/* handle bad return by returning NULL for ip */
 472		set_nlink(ip, 1);	/* Don't want iput() deleting it */
 473		iput(ip);
 474		/* release the page */
 475		release_metapage(mp);
 476		return (NULL);
 477
 478	}
 479
 480	ip->i_mapping->a_ops = &jfs_metapage_aops;
 481	mapping_set_gfp_mask(ip->i_mapping, GFP_NOFS);
 482
 483	/* Allocations to metadata inodes should not affect quotas */
 484	ip->i_flags |= S_NOQUOTA;
 485
 486	if ((inum == FILESYSTEM_I) && (JFS_IP(ip)->ipimap == sbi->ipaimap)) {
 487		sbi->gengen = le32_to_cpu(dp->di_gengen);
 488		sbi->inostamp = le32_to_cpu(dp->di_inostamp);
 489	}
 490
 491	/* release the page */
 492	release_metapage(mp);
 493
 494	/*
 495	 * __mark_inode_dirty expects inodes to be hashed.  Since we don't
 496	 * want special inodes in the fileset inode space, we make them
 497	 * appear hashed, but do not put on any lists.  hlist_del()
 498	 * will work fine and require no locking.
 499	 */
 500	hlist_add_fake(&ip->i_hash);
 501
 502	return (ip);
 503}
 504
 505/*
 506 * NAME:	diWriteSpecial()
 507 *
 508 * FUNCTION:	Write the special inode to disk
 509 *
 510 * PARAMETERS:
 511 *	ip - special inode
 512 *	secondary - 1 if secondary aggregate inode table
 513 *
 514 * RETURN VALUES: none
 515 */
 516
 517void diWriteSpecial(struct inode *ip, int secondary)
 518{
 519	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
 520	uint address;
 521	struct dinode *dp;
 522	ino_t inum = ip->i_ino;
 523	struct metapage *mp;
 524
 525	if (secondary)
 526		address = addressPXD(&sbi->ait2) >> sbi->l2nbperpage;
 527	else
 528		address = AITBL_OFF >> L2PSIZE;
 529
 530	ASSERT(inum < INOSPEREXT);
 531
 532	address += inum >> 3;	/* 8 inodes per 4K page */
 533
 534	/* read the page of fixed disk inode (AIT) in raw mode */
 535	mp = read_metapage(ip, address << sbi->l2nbperpage, PSIZE, 1);
 536	if (mp == NULL) {
 537		jfs_err("diWriteSpecial: failed to read aggregate inode extent!");
 538		return;
 539	}
 540
 541	/* get the pointer to the disk inode of interest */
 542	dp = (struct dinode *) (mp->data);
 543	dp += inum % 8;		/* 8 inodes per 4K page */
 544
 545	/* copy on-disk inode to in-memory inode */
 546	copy_to_dinode(dp, ip);
 547	memcpy(&dp->di_xtroot, &JFS_IP(ip)->i_xtroot, 288);
 548
 549	if (inum == FILESYSTEM_I)
 550		dp->di_gengen = cpu_to_le32(sbi->gengen);
 551
 552	/* write the page */
 553	write_metapage(mp);
 554}
 555
 556/*
 557 * NAME:	diFreeSpecial()
 558 *
 559 * FUNCTION:	Free allocated space for special inode
 560 */
 561void diFreeSpecial(struct inode *ip)
 562{
 563	if (ip == NULL) {
 564		jfs_err("diFreeSpecial called with NULL ip!");
 565		return;
 566	}
 567	filemap_write_and_wait(ip->i_mapping);
 568	truncate_inode_pages(ip->i_mapping, 0);
 569	iput(ip);
 570}
 571
 572
 573
 574/*
 575 * NAME:	diWrite()
 576 *
 577 * FUNCTION:	write the on-disk inode portion of the in-memory inode
 578 *		to its corresponding on-disk inode.
 579 *
 580 *		on entry, the specifed incore inode should itself
 581 *		specify the disk inode number corresponding to the
 582 *		incore inode (i.e. i_number should be initialized).
 583 *
 584 *		the inode contains the inode extent address for the disk
 585 *		inode.  with the inode extent address in hand, the
 586 *		page of the extent that contains the disk inode is
 587 *		read and the disk inode portion of the incore inode
 588 *		is copied to the disk inode.
 589 *
 590 * PARAMETERS:
 591 *	tid -  transacation id
 592 *	ip  -  pointer to incore inode to be written to the inode extent.
 593 *
 594 * RETURN VALUES:
 595 *	0	- success
 596 *	-EIO	- i/o error.
 597 */
 598int diWrite(tid_t tid, struct inode *ip)
 599{
 600	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
 601	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
 602	int rc = 0;
 603	s32 ino;
 604	struct dinode *dp;
 605	s64 blkno;
 606	int block_offset;
 607	int inodes_left;
 608	struct metapage *mp;
 609	unsigned long pageno;
 610	int rel_inode;
 611	int dioffset;
 612	struct inode *ipimap;
 613	uint type;
 614	lid_t lid;
 615	struct tlock *ditlck, *tlck;
 616	struct linelock *dilinelock, *ilinelock;
 617	struct lv *lv;
 618	int n;
 619
 620	ipimap = jfs_ip->ipimap;
 621
 622	ino = ip->i_ino & (INOSPERIAG - 1);
 623
 624	if (!addressPXD(&(jfs_ip->ixpxd)) ||
 625	    (lengthPXD(&(jfs_ip->ixpxd)) !=
 626	     JFS_IP(ipimap)->i_imap->im_nbperiext)) {
 627		jfs_error(ip->i_sb, "ixpxd invalid\n");
 628		return -EIO;
 629	}
 630
 631	/*
 632	 * read the page of disk inode containing the specified inode:
 633	 */
 634	/* compute the block address of the page */
 635	blkno = INOPBLK(&(jfs_ip->ixpxd), ino, sbi->l2nbperpage);
 636
 637	rel_inode = (ino & (INOSPERPAGE - 1));
 638	pageno = blkno >> sbi->l2nbperpage;
 639
 640	if ((block_offset = ((u32) blkno & (sbi->nbperpage - 1)))) {
 641		/*
 642		 * OS/2 didn't always align inode extents on page boundaries
 643		 */
 644		inodes_left =
 645		    (sbi->nbperpage - block_offset) << sbi->l2niperblk;
 646
 647		if (rel_inode < inodes_left)
 648			rel_inode += block_offset << sbi->l2niperblk;
 649		else {
 650			pageno += 1;
 651			rel_inode -= inodes_left;
 652		}
 653	}
 654	/* read the page of disk inode */
 655      retry:
 656	mp = read_metapage(ipimap, pageno << sbi->l2nbperpage, PSIZE, 1);
 657	if (!mp)
 658		return -EIO;
 659
 660	/* get the pointer to the disk inode */
 661	dp = (struct dinode *) mp->data;
 662	dp += rel_inode;
 663
 664	dioffset = (ino & (INOSPERPAGE - 1)) << L2DISIZE;
 665
 666	/*
 667	 * acquire transaction lock on the on-disk inode;
 668	 * N.B. tlock is acquired on ipimap not ip;
 669	 */
 670	if ((ditlck =
 671	     txLock(tid, ipimap, mp, tlckINODE | tlckENTRY)) == NULL)
 672		goto retry;
 673	dilinelock = (struct linelock *) & ditlck->lock;
 674
 675	/*
 676	 * copy btree root from in-memory inode to on-disk inode
 677	 *
 678	 * (tlock is taken from inline B+-tree root in in-memory
 679	 * inode when the B+-tree root is updated, which is pointed
 680	 * by jfs_ip->blid as well as being on tx tlock list)
 681	 *
 682	 * further processing of btree root is based on the copy
 683	 * in in-memory inode, where txLog() will log from, and,
 684	 * for xtree root, txUpdateMap() will update map and reset
 685	 * XAD_NEW bit;
 686	 */
 687
 688	if (S_ISDIR(ip->i_mode) && (lid = jfs_ip->xtlid)) {
 689		/*
 690		 * This is the special xtree inside the directory for storing
 691		 * the directory table
 692		 */
 693		xtpage_t *p, *xp;
 694		xad_t *xad;
 695
 696		jfs_ip->xtlid = 0;
 697		tlck = lid_to_tlock(lid);
 698		assert(tlck->type & tlckXTREE);
 699		tlck->type |= tlckBTROOT;
 700		tlck->mp = mp;
 701		ilinelock = (struct linelock *) & tlck->lock;
 702
 703		/*
 704		 * copy xtree root from inode to dinode:
 705		 */
 706		p = &jfs_ip->i_xtroot;
 707		xp = (xtpage_t *) &dp->di_dirtable;
 708		lv = ilinelock->lv;
 709		for (n = 0; n < ilinelock->index; n++, lv++) {
 710			memcpy(&xp->xad[lv->offset], &p->xad[lv->offset],
 711			       lv->length << L2XTSLOTSIZE);
 712		}
 713
 714		/* reset on-disk (metadata page) xtree XAD_NEW bit */
 715		xad = &xp->xad[XTENTRYSTART];
 716		for (n = XTENTRYSTART;
 717		     n < le16_to_cpu(xp->header.nextindex); n++, xad++)
 718			if (xad->flag & (XAD_NEW | XAD_EXTENDED))
 719				xad->flag &= ~(XAD_NEW | XAD_EXTENDED);
 720	}
 721
 722	if ((lid = jfs_ip->blid) == 0)
 723		goto inlineData;
 724	jfs_ip->blid = 0;
 725
 726	tlck = lid_to_tlock(lid);
 727	type = tlck->type;
 728	tlck->type |= tlckBTROOT;
 729	tlck->mp = mp;
 730	ilinelock = (struct linelock *) & tlck->lock;
 731
 732	/*
 733	 *	regular file: 16 byte (XAD slot) granularity
 734	 */
 735	if (type & tlckXTREE) {
 736		xtpage_t *p, *xp;
 737		xad_t *xad;
 738
 739		/*
 740		 * copy xtree root from inode to dinode:
 741		 */
 742		p = &jfs_ip->i_xtroot;
 743		xp = &dp->di_xtroot;
 744		lv = ilinelock->lv;
 745		for (n = 0; n < ilinelock->index; n++, lv++) {
 746			memcpy(&xp->xad[lv->offset], &p->xad[lv->offset],
 747			       lv->length << L2XTSLOTSIZE);
 748		}
 749
 750		/* reset on-disk (metadata page) xtree XAD_NEW bit */
 751		xad = &xp->xad[XTENTRYSTART];
 752		for (n = XTENTRYSTART;
 753		     n < le16_to_cpu(xp->header.nextindex); n++, xad++)
 754			if (xad->flag & (XAD_NEW | XAD_EXTENDED))
 755				xad->flag &= ~(XAD_NEW | XAD_EXTENDED);
 756	}
 757	/*
 758	 *	directory: 32 byte (directory entry slot) granularity
 759	 */
 760	else if (type & tlckDTREE) {
 761		dtpage_t *p, *xp;
 762
 763		/*
 764		 * copy dtree root from inode to dinode:
 765		 */
 766		p = (dtpage_t *) &jfs_ip->i_dtroot;
 767		xp = (dtpage_t *) & dp->di_dtroot;
 768		lv = ilinelock->lv;
 769		for (n = 0; n < ilinelock->index; n++, lv++) {
 770			memcpy(&xp->slot[lv->offset], &p->slot[lv->offset],
 771			       lv->length << L2DTSLOTSIZE);
 772		}
 773	} else {
 774		jfs_err("diWrite: UFO tlock");
 775	}
 776
 777      inlineData:
 778	/*
 779	 * copy inline symlink from in-memory inode to on-disk inode
 780	 */
 781	if (S_ISLNK(ip->i_mode) && ip->i_size < IDATASIZE) {
 782		lv = & dilinelock->lv[dilinelock->index];
 783		lv->offset = (dioffset + 2 * 128) >> L2INODESLOTSIZE;
 784		lv->length = 2;
 785		memcpy(&dp->di_fastsymlink, jfs_ip->i_inline, IDATASIZE);
 786		dilinelock->index++;
 787	}
 788	/*
 789	 * copy inline data from in-memory inode to on-disk inode:
 790	 * 128 byte slot granularity
 791	 */
 792	if (test_cflag(COMMIT_Inlineea, ip)) {
 793		lv = & dilinelock->lv[dilinelock->index];
 794		lv->offset = (dioffset + 3 * 128) >> L2INODESLOTSIZE;
 795		lv->length = 1;
 796		memcpy(&dp->di_inlineea, jfs_ip->i_inline_ea, INODESLOTSIZE);
 797		dilinelock->index++;
 798
 799		clear_cflag(COMMIT_Inlineea, ip);
 800	}
 801
 802	/*
 803	 *	lock/copy inode base: 128 byte slot granularity
 804	 */
 805	lv = & dilinelock->lv[dilinelock->index];
 806	lv->offset = dioffset >> L2INODESLOTSIZE;
 807	copy_to_dinode(dp, ip);
 808	if (test_and_clear_cflag(COMMIT_Dirtable, ip)) {
 809		lv->length = 2;
 810		memcpy(&dp->di_dirtable, &jfs_ip->i_dirtable, 96);
 811	} else
 812		lv->length = 1;
 813	dilinelock->index++;
 814
 815	/* release the buffer holding the updated on-disk inode.
 816	 * the buffer will be later written by commit processing.
 817	 */
 818	write_metapage(mp);
 819
 820	return (rc);
 821}
 822
 823
 824/*
 825 * NAME:	diFree(ip)
 826 *
 827 * FUNCTION:	free a specified inode from the inode working map
 828 *		for a fileset or aggregate.
 829 *
 830 *		if the inode to be freed represents the first (only)
 831 *		free inode within the iag, the iag will be placed on
 832 *		the ag free inode list.
 833 *
 834 *		freeing the inode will cause the inode extent to be
 835 *		freed if the inode is the only allocated inode within
 836 *		the extent.  in this case all the disk resource backing
 837 *		up the inode extent will be freed. in addition, the iag
 838 *		will be placed on the ag extent free list if the extent
 839 *		is the first free extent in the iag.  if freeing the
 840 *		extent also means that no free inodes will exist for
 841 *		the iag, the iag will also be removed from the ag free
 842 *		inode list.
 843 *
 844 *		the iag describing the inode will be freed if the extent
 845 *		is to be freed and it is the only backed extent within
 846 *		the iag.  in this case, the iag will be removed from the
 847 *		ag free extent list and ag free inode list and placed on
 848 *		the inode map's free iag list.
 849 *
 850 *		a careful update approach is used to provide consistency
 851 *		in the face of updates to multiple buffers.  under this
 852 *		approach, all required buffers are obtained before making
 853 *		any updates and are held until all updates are complete.
 854 *
 855 * PARAMETERS:
 856 *	ip	- inode to be freed.
 857 *
 858 * RETURN VALUES:
 859 *	0	- success
 860 *	-EIO	- i/o error.
 861 */
 862int diFree(struct inode *ip)
 863{
 864	int rc;
 865	ino_t inum = ip->i_ino;
 866	struct iag *iagp, *aiagp, *biagp, *ciagp, *diagp;
 867	struct metapage *mp, *amp, *bmp, *cmp, *dmp;
 868	int iagno, ino, extno, bitno, sword, agno;
 869	int back, fwd;
 870	u32 bitmap, mask;
 871	struct inode *ipimap = JFS_SBI(ip->i_sb)->ipimap;
 872	struct inomap *imap = JFS_IP(ipimap)->i_imap;
 873	pxd_t freepxd;
 874	tid_t tid;
 875	struct inode *iplist[3];
 876	struct tlock *tlck;
 877	struct pxd_lock *pxdlock;
 878
 879	/*
 880	 * This is just to suppress compiler warnings.  The same logic that
 881	 * references these variables is used to initialize them.
 882	 */
 883	aiagp = biagp = ciagp = diagp = NULL;
 884
 885	/* get the iag number containing the inode.
 886	 */
 887	iagno = INOTOIAG(inum);
 888
 889	/* make sure that the iag is contained within
 890	 * the map.
 891	 */
 892	if (iagno >= imap->im_nextiag) {
 893		print_hex_dump(KERN_ERR, "imap: ", DUMP_PREFIX_ADDRESS, 16, 4,
 894			       imap, 32, 0);
 895		jfs_error(ip->i_sb, "inum = %d, iagno = %d, nextiag = %d\n",
 896			  (uint) inum, iagno, imap->im_nextiag);
 897		return -EIO;
 898	}
 899
 900	/* get the allocation group for this ino.
 901	 */
 902	agno = BLKTOAG(JFS_IP(ip)->agstart, JFS_SBI(ip->i_sb));
 903
 904	/* Lock the AG specific inode map information
 905	 */
 906	AG_LOCK(imap, agno);
 907
 908	/* Obtain read lock in imap inode.  Don't release it until we have
 909	 * read all of the IAG's that we are going to.
 910	 */
 911	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
 912
 913	/* read the iag.
 914	 */
 915	if ((rc = diIAGRead(imap, iagno, &mp))) {
 916		IREAD_UNLOCK(ipimap);
 917		AG_UNLOCK(imap, agno);
 918		return (rc);
 919	}
 920	iagp = (struct iag *) mp->data;
 921
 922	/* get the inode number and extent number of the inode within
 923	 * the iag and the inode number within the extent.
 924	 */
 925	ino = inum & (INOSPERIAG - 1);
 926	extno = ino >> L2INOSPEREXT;
 927	bitno = ino & (INOSPEREXT - 1);
 928	mask = HIGHORDER >> bitno;
 929
 930	if (!(le32_to_cpu(iagp->wmap[extno]) & mask)) {
 931		jfs_error(ip->i_sb, "wmap shows inode already free\n");
 932	}
 933
 934	if (!addressPXD(&iagp->inoext[extno])) {
 935		release_metapage(mp);
 936		IREAD_UNLOCK(ipimap);
 937		AG_UNLOCK(imap, agno);
 938		jfs_error(ip->i_sb, "invalid inoext\n");
 939		return -EIO;
 940	}
 941
 942	/* compute the bitmap for the extent reflecting the freed inode.
 943	 */
 944	bitmap = le32_to_cpu(iagp->wmap[extno]) & ~mask;
 945
 946	if (imap->im_agctl[agno].numfree > imap->im_agctl[agno].numinos) {
 947		release_metapage(mp);
 948		IREAD_UNLOCK(ipimap);
 949		AG_UNLOCK(imap, agno);
 950		jfs_error(ip->i_sb, "numfree > numinos\n");
 951		return -EIO;
 952	}
 953	/*
 954	 *	inode extent still has some inodes or below low water mark:
 955	 *	keep the inode extent;
 956	 */
 957	if (bitmap ||
 958	    imap->im_agctl[agno].numfree < 96 ||
 959	    (imap->im_agctl[agno].numfree < 288 &&
 960	     (((imap->im_agctl[agno].numfree * 100) /
 961	       imap->im_agctl[agno].numinos) <= 25))) {
 962		/* if the iag currently has no free inodes (i.e.,
 963		 * the inode being freed is the first free inode of iag),
 964		 * insert the iag at head of the inode free list for the ag.
 965		 */
 966		if (iagp->nfreeinos == 0) {
 967			/* check if there are any iags on the ag inode
 968			 * free list.  if so, read the first one so that
 969			 * we can link the current iag onto the list at
 970			 * the head.
 971			 */
 972			if ((fwd = imap->im_agctl[agno].inofree) >= 0) {
 973				/* read the iag that currently is the head
 974				 * of the list.
 975				 */
 976				if ((rc = diIAGRead(imap, fwd, &amp))) {
 977					IREAD_UNLOCK(ipimap);
 978					AG_UNLOCK(imap, agno);
 979					release_metapage(mp);
 980					return (rc);
 981				}
 982				aiagp = (struct iag *) amp->data;
 983
 984				/* make current head point back to the iag.
 985				 */
 986				aiagp->inofreeback = cpu_to_le32(iagno);
 987
 988				write_metapage(amp);
 989			}
 990
 991			/* iag points forward to current head and iag
 992			 * becomes the new head of the list.
 993			 */
 994			iagp->inofreefwd =
 995			    cpu_to_le32(imap->im_agctl[agno].inofree);
 996			iagp->inofreeback = cpu_to_le32(-1);
 997			imap->im_agctl[agno].inofree = iagno;
 998		}
 999		IREAD_UNLOCK(ipimap);
1000
1001		/* update the free inode summary map for the extent if
1002		 * freeing the inode means the extent will now have free
1003		 * inodes (i.e., the inode being freed is the first free
1004		 * inode of extent),
1005		 */
1006		if (iagp->wmap[extno] == cpu_to_le32(ONES)) {
1007			sword = extno >> L2EXTSPERSUM;
1008			bitno = extno & (EXTSPERSUM - 1);
1009			iagp->inosmap[sword] &=
1010			    cpu_to_le32(~(HIGHORDER >> bitno));
1011		}
1012
1013		/* update the bitmap.
1014		 */
1015		iagp->wmap[extno] = cpu_to_le32(bitmap);
1016
1017		/* update the free inode counts at the iag, ag and
1018		 * map level.
1019		 */
1020		le32_add_cpu(&iagp->nfreeinos, 1);
1021		imap->im_agctl[agno].numfree += 1;
1022		atomic_inc(&imap->im_numfree);
1023
1024		/* release the AG inode map lock
1025		 */
1026		AG_UNLOCK(imap, agno);
1027
1028		/* write the iag */
1029		write_metapage(mp);
1030
1031		return (0);
1032	}
1033
1034
1035	/*
1036	 *	inode extent has become free and above low water mark:
1037	 *	free the inode extent;
1038	 */
1039
1040	/*
1041	 *	prepare to update iag list(s) (careful update step 1)
1042	 */
1043	amp = bmp = cmp = dmp = NULL;
1044	fwd = back = -1;
1045
1046	/* check if the iag currently has no free extents.  if so,
1047	 * it will be placed on the head of the ag extent free list.
1048	 */
1049	if (iagp->nfreeexts == 0) {
1050		/* check if the ag extent free list has any iags.
1051		 * if so, read the iag at the head of the list now.
1052		 * this (head) iag will be updated later to reflect
1053		 * the addition of the current iag at the head of
1054		 * the list.
1055		 */
1056		if ((fwd = imap->im_agctl[agno].extfree) >= 0) {
1057			if ((rc = diIAGRead(imap, fwd, &amp)))
1058				goto error_out;
1059			aiagp = (struct iag *) amp->data;
1060		}
1061	} else {
1062		/* iag has free extents. check if the addition of a free
1063		 * extent will cause all extents to be free within this
1064		 * iag.  if so, the iag will be removed from the ag extent
1065		 * free list and placed on the inode map's free iag list.
1066		 */
1067		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG - 1)) {
1068			/* in preparation for removing the iag from the
1069			 * ag extent free list, read the iags preceding
1070			 * and following the iag on the ag extent free
1071			 * list.
1072			 */
1073			if ((fwd = le32_to_cpu(iagp->extfreefwd)) >= 0) {
1074				if ((rc = diIAGRead(imap, fwd, &amp)))
1075					goto error_out;
1076				aiagp = (struct iag *) amp->data;
1077			}
1078
1079			if ((back = le32_to_cpu(iagp->extfreeback)) >= 0) {
1080				if ((rc = diIAGRead(imap, back, &bmp)))
1081					goto error_out;
1082				biagp = (struct iag *) bmp->data;
1083			}
1084		}
1085	}
1086
1087	/* remove the iag from the ag inode free list if freeing
1088	 * this extent cause the iag to have no free inodes.
1089	 */
1090	if (iagp->nfreeinos == cpu_to_le32(INOSPEREXT - 1)) {
1091		int inofreeback = le32_to_cpu(iagp->inofreeback);
1092		int inofreefwd = le32_to_cpu(iagp->inofreefwd);
1093
1094		/* in preparation for removing the iag from the
1095		 * ag inode free list, read the iags preceding
1096		 * and following the iag on the ag inode free
1097		 * list.  before reading these iags, we must make
1098		 * sure that we already don't have them in hand
1099		 * from up above, since re-reading an iag (buffer)
1100		 * we are currently holding would cause a deadlock.
1101		 */
1102		if (inofreefwd >= 0) {
1103
1104			if (inofreefwd == fwd)
1105				ciagp = (struct iag *) amp->data;
1106			else if (inofreefwd == back)
1107				ciagp = (struct iag *) bmp->data;
1108			else {
1109				if ((rc =
1110				     diIAGRead(imap, inofreefwd, &cmp)))
1111					goto error_out;
1112				ciagp = (struct iag *) cmp->data;
1113			}
1114			assert(ciagp != NULL);
1115		}
1116
1117		if (inofreeback >= 0) {
1118			if (inofreeback == fwd)
1119				diagp = (struct iag *) amp->data;
1120			else if (inofreeback == back)
1121				diagp = (struct iag *) bmp->data;
1122			else {
1123				if ((rc =
1124				     diIAGRead(imap, inofreeback, &dmp)))
1125					goto error_out;
1126				diagp = (struct iag *) dmp->data;
1127			}
1128			assert(diagp != NULL);
1129		}
1130	}
1131
1132	IREAD_UNLOCK(ipimap);
1133
1134	/*
1135	 * invalidate any page of the inode extent freed from buffer cache;
1136	 */
1137	freepxd = iagp->inoext[extno];
1138	invalidate_pxd_metapages(ip, freepxd);
1139
1140	/*
1141	 *	update iag list(s) (careful update step 2)
1142	 */
1143	/* add the iag to the ag extent free list if this is the
1144	 * first free extent for the iag.
1145	 */
1146	if (iagp->nfreeexts == 0) {
1147		if (fwd >= 0)
1148			aiagp->extfreeback = cpu_to_le32(iagno);
1149
1150		iagp->extfreefwd =
1151		    cpu_to_le32(imap->im_agctl[agno].extfree);
1152		iagp->extfreeback = cpu_to_le32(-1);
1153		imap->im_agctl[agno].extfree = iagno;
1154	} else {
1155		/* remove the iag from the ag extent list if all extents
1156		 * are now free and place it on the inode map iag free list.
1157		 */
1158		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG - 1)) {
1159			if (fwd >= 0)
1160				aiagp->extfreeback = iagp->extfreeback;
1161
1162			if (back >= 0)
1163				biagp->extfreefwd = iagp->extfreefwd;
1164			else
1165				imap->im_agctl[agno].extfree =
1166				    le32_to_cpu(iagp->extfreefwd);
1167
1168			iagp->extfreefwd = iagp->extfreeback = cpu_to_le32(-1);
1169
1170			IAGFREE_LOCK(imap);
1171			iagp->iagfree = cpu_to_le32(imap->im_freeiag);
1172			imap->im_freeiag = iagno;
1173			IAGFREE_UNLOCK(imap);
1174		}
1175	}
1176
1177	/* remove the iag from the ag inode free list if freeing
1178	 * this extent causes the iag to have no free inodes.
1179	 */
1180	if (iagp->nfreeinos == cpu_to_le32(INOSPEREXT - 1)) {
1181		if ((int) le32_to_cpu(iagp->inofreefwd) >= 0)
1182			ciagp->inofreeback = iagp->inofreeback;
1183
1184		if ((int) le32_to_cpu(iagp->inofreeback) >= 0)
1185			diagp->inofreefwd = iagp->inofreefwd;
1186		else
1187			imap->im_agctl[agno].inofree =
1188			    le32_to_cpu(iagp->inofreefwd);
1189
1190		iagp->inofreefwd = iagp->inofreeback = cpu_to_le32(-1);
1191	}
1192
1193	/* update the inode extent address and working map
1194	 * to reflect the free extent.
1195	 * the permanent map should have been updated already
1196	 * for the inode being freed.
1197	 */
1198	if (iagp->pmap[extno] != 0) {
1199		jfs_error(ip->i_sb, "the pmap does not show inode free\n");
1200	}
1201	iagp->wmap[extno] = 0;
1202	PXDlength(&iagp->inoext[extno], 0);
1203	PXDaddress(&iagp->inoext[extno], 0);
1204
1205	/* update the free extent and free inode summary maps
1206	 * to reflect the freed extent.
1207	 * the inode summary map is marked to indicate no inodes
1208	 * available for the freed extent.
1209	 */
1210	sword = extno >> L2EXTSPERSUM;
1211	bitno = extno & (EXTSPERSUM - 1);
1212	mask = HIGHORDER >> bitno;
1213	iagp->inosmap[sword] |= cpu_to_le32(mask);
1214	iagp->extsmap[sword] &= cpu_to_le32(~mask);
1215
1216	/* update the number of free inodes and number of free extents
1217	 * for the iag.
1218	 */
1219	le32_add_cpu(&iagp->nfreeinos, -(INOSPEREXT - 1));
1220	le32_add_cpu(&iagp->nfreeexts, 1);
1221
1222	/* update the number of free inodes and backed inodes
1223	 * at the ag and inode map level.
1224	 */
1225	imap->im_agctl[agno].numfree -= (INOSPEREXT - 1);
1226	imap->im_agctl[agno].numinos -= INOSPEREXT;
1227	atomic_sub(INOSPEREXT - 1, &imap->im_numfree);
1228	atomic_sub(INOSPEREXT, &imap->im_numinos);
1229
1230	if (amp)
1231		write_metapage(amp);
1232	if (bmp)
1233		write_metapage(bmp);
1234	if (cmp)
1235		write_metapage(cmp);
1236	if (dmp)
1237		write_metapage(dmp);
1238
1239	/*
1240	 * start transaction to update block allocation map
1241	 * for the inode extent freed;
1242	 *
1243	 * N.B. AG_LOCK is released and iag will be released below, and
1244	 * other thread may allocate inode from/reusing the ixad freed
1245	 * BUT with new/different backing inode extent from the extent
1246	 * to be freed by the transaction;
1247	 */
1248	tid = txBegin(ipimap->i_sb, COMMIT_FORCE);
1249	mutex_lock(&JFS_IP(ipimap)->commit_mutex);
1250
1251	/* acquire tlock of the iag page of the freed ixad
1252	 * to force the page NOHOMEOK (even though no data is
1253	 * logged from the iag page) until NOREDOPAGE|FREEXTENT log
1254	 * for the free of the extent is committed;
1255	 * write FREEXTENT|NOREDOPAGE log record
1256	 * N.B. linelock is overlaid as freed extent descriptor;
1257	 */
1258	tlck = txLock(tid, ipimap, mp, tlckINODE | tlckFREE);
1259	pxdlock = (struct pxd_lock *) & tlck->lock;
1260	pxdlock->flag = mlckFREEPXD;
1261	pxdlock->pxd = freepxd;
1262	pxdlock->index = 1;
1263
1264	write_metapage(mp);
1265
1266	iplist[0] = ipimap;
1267
1268	/*
1269	 * logredo needs the IAG number and IAG extent index in order
1270	 * to ensure that the IMap is consistent.  The least disruptive
1271	 * way to pass these values through  to the transaction manager
1272	 * is in the iplist array.
1273	 *
1274	 * It's not pretty, but it works.
1275	 */
1276	iplist[1] = (struct inode *) (size_t)iagno;
1277	iplist[2] = (struct inode *) (size_t)extno;
1278
1279	rc = txCommit(tid, 1, &iplist[0], COMMIT_FORCE);
1280
1281	txEnd(tid);
1282	mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
1283
1284	/* unlock the AG inode map information */
1285	AG_UNLOCK(imap, agno);
1286
1287	return (0);
1288
1289      error_out:
1290	IREAD_UNLOCK(ipimap);
1291
1292	if (amp)
1293		release_metapage(amp);
1294	if (bmp)
1295		release_metapage(bmp);
1296	if (cmp)
1297		release_metapage(cmp);
1298	if (dmp)
1299		release_metapage(dmp);
1300
1301	AG_UNLOCK(imap, agno);
1302
1303	release_metapage(mp);
1304
1305	return (rc);
1306}
1307
1308/*
1309 * There are several places in the diAlloc* routines where we initialize
1310 * the inode.
1311 */
1312static inline void
1313diInitInode(struct inode *ip, int iagno, int ino, int extno, struct iag * iagp)
1314{
1315	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
1316
1317	ip->i_ino = (iagno << L2INOSPERIAG) + ino;
1318	jfs_ip->ixpxd = iagp->inoext[extno];
1319	jfs_ip->agstart = le64_to_cpu(iagp->agstart);
1320	jfs_ip->active_ag = -1;
1321}
1322
1323
1324/*
1325 * NAME:	diAlloc(pip,dir,ip)
1326 *
1327 * FUNCTION:	allocate a disk inode from the inode working map
1328 *		for a fileset or aggregate.
1329 *
1330 * PARAMETERS:
1331 *	pip	- pointer to incore inode for the parent inode.
1332 *	dir	- 'true' if the new disk inode is for a directory.
1333 *	ip	- pointer to a new inode
1334 *
1335 * RETURN VALUES:
1336 *	0	- success.
1337 *	-ENOSPC	- insufficient disk resources.
1338 *	-EIO	- i/o error.
1339 */
1340int diAlloc(struct inode *pip, bool dir, struct inode *ip)
1341{
1342	int rc, ino, iagno, addext, extno, bitno, sword;
1343	int nwords, rem, i, agno;
1344	u32 mask, inosmap, extsmap;
1345	struct inode *ipimap;
1346	struct metapage *mp;
1347	ino_t inum;
1348	struct iag *iagp;
1349	struct inomap *imap;
1350
1351	/* get the pointers to the inode map inode and the
1352	 * corresponding imap control structure.
1353	 */
1354	ipimap = JFS_SBI(pip->i_sb)->ipimap;
1355	imap = JFS_IP(ipimap)->i_imap;
1356	JFS_IP(ip)->ipimap = ipimap;
1357	JFS_IP(ip)->fileset = FILESYSTEM_I;
1358
1359	/* for a directory, the allocation policy is to start
1360	 * at the ag level using the preferred ag.
1361	 */
1362	if (dir) {
1363		agno = dbNextAG(JFS_SBI(pip->i_sb)->ipbmap);
1364		AG_LOCK(imap, agno);
1365		goto tryag;
1366	}
1367
1368	/* for files, the policy starts off by trying to allocate from
1369	 * the same iag containing the parent disk inode:
1370	 * try to allocate the new disk inode close to the parent disk
1371	 * inode, using parent disk inode number + 1 as the allocation
1372	 * hint.  (we use a left-to-right policy to attempt to avoid
1373	 * moving backward on the disk.)  compute the hint within the
1374	 * file system and the iag.
1375	 */
1376
1377	/* get the ag number of this iag */
1378	agno = BLKTOAG(JFS_IP(pip)->agstart, JFS_SBI(pip->i_sb));
 
 
 
1379
1380	if (atomic_read(&JFS_SBI(pip->i_sb)->bmap->db_active[agno])) {
1381		/*
1382		 * There is an open file actively growing.  We want to
1383		 * allocate new inodes from a different ag to avoid
1384		 * fragmentation problems.
1385		 */
1386		agno = dbNextAG(JFS_SBI(pip->i_sb)->ipbmap);
1387		AG_LOCK(imap, agno);
1388		goto tryag;
1389	}
1390
1391	inum = pip->i_ino + 1;
1392	ino = inum & (INOSPERIAG - 1);
1393
1394	/* back off the hint if it is outside of the iag */
1395	if (ino == 0)
1396		inum = pip->i_ino;
1397
1398	/* lock the AG inode map information */
1399	AG_LOCK(imap, agno);
1400
1401	/* Get read lock on imap inode */
1402	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
1403
1404	/* get the iag number and read the iag */
1405	iagno = INOTOIAG(inum);
1406	if ((rc = diIAGRead(imap, iagno, &mp))) {
1407		IREAD_UNLOCK(ipimap);
1408		AG_UNLOCK(imap, agno);
1409		return (rc);
1410	}
1411	iagp = (struct iag *) mp->data;
1412
1413	/* determine if new inode extent is allowed to be added to the iag.
1414	 * new inode extent can be added to the iag if the ag
1415	 * has less than 32 free disk inodes and the iag has free extents.
1416	 */
1417	addext = (imap->im_agctl[agno].numfree < 32 && iagp->nfreeexts);
1418
1419	/*
1420	 *	try to allocate from the IAG
1421	 */
1422	/* check if the inode may be allocated from the iag
1423	 * (i.e. the inode has free inodes or new extent can be added).
1424	 */
1425	if (iagp->nfreeinos || addext) {
1426		/* determine the extent number of the hint.
1427		 */
1428		extno = ino >> L2INOSPEREXT;
1429
1430		/* check if the extent containing the hint has backed
1431		 * inodes.  if so, try to allocate within this extent.
1432		 */
1433		if (addressPXD(&iagp->inoext[extno])) {
1434			bitno = ino & (INOSPEREXT - 1);
1435			if ((bitno =
1436			     diFindFree(le32_to_cpu(iagp->wmap[extno]),
1437					bitno))
1438			    < INOSPEREXT) {
1439				ino = (extno << L2INOSPEREXT) + bitno;
1440
1441				/* a free inode (bit) was found within this
1442				 * extent, so allocate it.
1443				 */
1444				rc = diAllocBit(imap, iagp, ino);
1445				IREAD_UNLOCK(ipimap);
1446				if (rc) {
1447					assert(rc == -EIO);
1448				} else {
1449					/* set the results of the allocation
1450					 * and write the iag.
1451					 */
1452					diInitInode(ip, iagno, ino, extno,
1453						    iagp);
1454					mark_metapage_dirty(mp);
1455				}
1456				release_metapage(mp);
1457
1458				/* free the AG lock and return.
1459				 */
1460				AG_UNLOCK(imap, agno);
1461				return (rc);
1462			}
1463
1464			if (!addext)
1465				extno =
1466				    (extno ==
1467				     EXTSPERIAG - 1) ? 0 : extno + 1;
1468		}
1469
1470		/*
1471		 * no free inodes within the extent containing the hint.
1472		 *
1473		 * try to allocate from the backed extents following
1474		 * hint or, if appropriate (i.e. addext is true), allocate
1475		 * an extent of free inodes at or following the extent
1476		 * containing the hint.
1477		 *
1478		 * the free inode and free extent summary maps are used
1479		 * here, so determine the starting summary map position
1480		 * and the number of words we'll have to examine.  again,
1481		 * the approach is to allocate following the hint, so we
1482		 * might have to initially ignore prior bits of the summary
1483		 * map that represent extents prior to the extent containing
1484		 * the hint and later revisit these bits.
1485		 */
1486		bitno = extno & (EXTSPERSUM - 1);
1487		nwords = (bitno == 0) ? SMAPSZ : SMAPSZ + 1;
1488		sword = extno >> L2EXTSPERSUM;
1489
1490		/* mask any prior bits for the starting words of the
1491		 * summary map.
1492		 */
1493		mask = (bitno == 0) ? 0 : (ONES << (EXTSPERSUM - bitno));
1494		inosmap = le32_to_cpu(iagp->inosmap[sword]) | mask;
1495		extsmap = le32_to_cpu(iagp->extsmap[sword]) | mask;
1496
1497		/* scan the free inode and free extent summary maps for
1498		 * free resources.
1499		 */
1500		for (i = 0; i < nwords; i++) {
1501			/* check if this word of the free inode summary
1502			 * map describes an extent with free inodes.
1503			 */
1504			if (~inosmap) {
1505				/* an extent with free inodes has been
1506				 * found. determine the extent number
1507				 * and the inode number within the extent.
1508				 */
1509				rem = diFindFree(inosmap, 0);
1510				extno = (sword << L2EXTSPERSUM) + rem;
1511				rem = diFindFree(le32_to_cpu(iagp->wmap[extno]),
1512						 0);
1513				if (rem >= INOSPEREXT) {
1514					IREAD_UNLOCK(ipimap);
1515					release_metapage(mp);
1516					AG_UNLOCK(imap, agno);
1517					jfs_error(ip->i_sb,
1518						  "can't find free bit in wmap\n");
1519					return -EIO;
1520				}
1521
1522				/* determine the inode number within the
1523				 * iag and allocate the inode from the
1524				 * map.
1525				 */
1526				ino = (extno << L2INOSPEREXT) + rem;
1527				rc = diAllocBit(imap, iagp, ino);
1528				IREAD_UNLOCK(ipimap);
1529				if (rc)
1530					assert(rc == -EIO);
1531				else {
1532					/* set the results of the allocation
1533					 * and write the iag.
1534					 */
1535					diInitInode(ip, iagno, ino, extno,
1536						    iagp);
1537					mark_metapage_dirty(mp);
1538				}
1539				release_metapage(mp);
1540
1541				/* free the AG lock and return.
1542				 */
1543				AG_UNLOCK(imap, agno);
1544				return (rc);
1545
1546			}
1547
1548			/* check if we may allocate an extent of free
1549			 * inodes and whether this word of the free
1550			 * extents summary map describes a free extent.
1551			 */
1552			if (addext && ~extsmap) {
1553				/* a free extent has been found.  determine
1554				 * the extent number.
1555				 */
1556				rem = diFindFree(extsmap, 0);
1557				extno = (sword << L2EXTSPERSUM) + rem;
1558
1559				/* allocate an extent of free inodes.
1560				 */
1561				if ((rc = diNewExt(imap, iagp, extno))) {
1562					/* if there is no disk space for a
1563					 * new extent, try to allocate the
1564					 * disk inode from somewhere else.
1565					 */
1566					if (rc == -ENOSPC)
1567						break;
1568
1569					assert(rc == -EIO);
1570				} else {
1571					/* set the results of the allocation
1572					 * and write the iag.
1573					 */
1574					diInitInode(ip, iagno,
1575						    extno << L2INOSPEREXT,
1576						    extno, iagp);
1577					mark_metapage_dirty(mp);
1578				}
1579				release_metapage(mp);
1580				/* free the imap inode & the AG lock & return.
1581				 */
1582				IREAD_UNLOCK(ipimap);
1583				AG_UNLOCK(imap, agno);
1584				return (rc);
1585			}
1586
1587			/* move on to the next set of summary map words.
1588			 */
1589			sword = (sword == SMAPSZ - 1) ? 0 : sword + 1;
1590			inosmap = le32_to_cpu(iagp->inosmap[sword]);
1591			extsmap = le32_to_cpu(iagp->extsmap[sword]);
1592		}
1593	}
1594	/* unlock imap inode */
1595	IREAD_UNLOCK(ipimap);
1596
1597	/* nothing doing in this iag, so release it. */
1598	release_metapage(mp);
1599
1600      tryag:
1601	/*
1602	 * try to allocate anywhere within the same AG as the parent inode.
1603	 */
1604	rc = diAllocAG(imap, agno, dir, ip);
1605
1606	AG_UNLOCK(imap, agno);
1607
1608	if (rc != -ENOSPC)
1609		return (rc);
1610
1611	/*
1612	 * try to allocate in any AG.
1613	 */
1614	return (diAllocAny(imap, agno, dir, ip));
1615}
1616
1617
1618/*
1619 * NAME:	diAllocAG(imap,agno,dir,ip)
1620 *
1621 * FUNCTION:	allocate a disk inode from the allocation group.
1622 *
1623 *		this routine first determines if a new extent of free
1624 *		inodes should be added for the allocation group, with
1625 *		the current request satisfied from this extent. if this
1626 *		is the case, an attempt will be made to do just that.  if
1627 *		this attempt fails or it has been determined that a new
1628 *		extent should not be added, an attempt is made to satisfy
1629 *		the request by allocating an existing (backed) free inode
1630 *		from the allocation group.
1631 *
1632 * PRE CONDITION: Already have the AG lock for this AG.
1633 *
1634 * PARAMETERS:
1635 *	imap	- pointer to inode map control structure.
1636 *	agno	- allocation group to allocate from.
1637 *	dir	- 'true' if the new disk inode is for a directory.
1638 *	ip	- pointer to the new inode to be filled in on successful return
1639 *		  with the disk inode number allocated, its extent address
1640 *		  and the start of the ag.
1641 *
1642 * RETURN VALUES:
1643 *	0	- success.
1644 *	-ENOSPC	- insufficient disk resources.
1645 *	-EIO	- i/o error.
1646 */
1647static int
1648diAllocAG(struct inomap * imap, int agno, bool dir, struct inode *ip)
1649{
1650	int rc, addext, numfree, numinos;
1651
1652	/* get the number of free and the number of backed disk
1653	 * inodes currently within the ag.
1654	 */
1655	numfree = imap->im_agctl[agno].numfree;
1656	numinos = imap->im_agctl[agno].numinos;
1657
1658	if (numfree > numinos) {
1659		jfs_error(ip->i_sb, "numfree > numinos\n");
1660		return -EIO;
1661	}
1662
1663	/* determine if we should allocate a new extent of free inodes
1664	 * within the ag: for directory inodes, add a new extent
1665	 * if there are a small number of free inodes or number of free
1666	 * inodes is a small percentage of the number of backed inodes.
1667	 */
1668	if (dir)
1669		addext = (numfree < 64 ||
1670			  (numfree < 256
1671			   && ((numfree * 100) / numinos) <= 20));
1672	else
1673		addext = (numfree == 0);
1674
1675	/*
1676	 * try to allocate a new extent of free inodes.
1677	 */
1678	if (addext) {
1679		/* if free space is not available for this new extent, try
1680		 * below to allocate a free and existing (already backed)
1681		 * inode from the ag.
1682		 */
1683		if ((rc = diAllocExt(imap, agno, ip)) != -ENOSPC)
1684			return (rc);
1685	}
1686
1687	/*
1688	 * try to allocate an existing free inode from the ag.
1689	 */
1690	return (diAllocIno(imap, agno, ip));
1691}
1692
1693
1694/*
1695 * NAME:	diAllocAny(imap,agno,dir,iap)
1696 *
1697 * FUNCTION:	allocate a disk inode from any other allocation group.
1698 *
1699 *		this routine is called when an allocation attempt within
1700 *		the primary allocation group has failed. if attempts to
1701 *		allocate an inode from any allocation group other than the
1702 *		specified primary group.
1703 *
1704 * PARAMETERS:
1705 *	imap	- pointer to inode map control structure.
1706 *	agno	- primary allocation group (to avoid).
1707 *	dir	- 'true' if the new disk inode is for a directory.
1708 *	ip	- pointer to a new inode to be filled in on successful return
1709 *		  with the disk inode number allocated, its extent address
1710 *		  and the start of the ag.
1711 *
1712 * RETURN VALUES:
1713 *	0	- success.
1714 *	-ENOSPC	- insufficient disk resources.
1715 *	-EIO	- i/o error.
1716 */
1717static int
1718diAllocAny(struct inomap * imap, int agno, bool dir, struct inode *ip)
1719{
1720	int ag, rc;
1721	int maxag = JFS_SBI(imap->im_ipimap->i_sb)->bmap->db_maxag;
1722
1723
1724	/* try to allocate from the ags following agno up to
1725	 * the maximum ag number.
1726	 */
1727	for (ag = agno + 1; ag <= maxag; ag++) {
1728		AG_LOCK(imap, ag);
1729
1730		rc = diAllocAG(imap, ag, dir, ip);
1731
1732		AG_UNLOCK(imap, ag);
1733
1734		if (rc != -ENOSPC)
1735			return (rc);
1736	}
1737
1738	/* try to allocate from the ags in front of agno.
1739	 */
1740	for (ag = 0; ag < agno; ag++) {
1741		AG_LOCK(imap, ag);
1742
1743		rc = diAllocAG(imap, ag, dir, ip);
1744
1745		AG_UNLOCK(imap, ag);
1746
1747		if (rc != -ENOSPC)
1748			return (rc);
1749	}
1750
1751	/* no free disk inodes.
1752	 */
1753	return -ENOSPC;
1754}
1755
1756
1757/*
1758 * NAME:	diAllocIno(imap,agno,ip)
1759 *
1760 * FUNCTION:	allocate a disk inode from the allocation group's free
1761 *		inode list, returning an error if this free list is
1762 *		empty (i.e. no iags on the list).
1763 *
1764 *		allocation occurs from the first iag on the list using
1765 *		the iag's free inode summary map to find the leftmost
1766 *		free inode in the iag.
1767 *
1768 * PRE CONDITION: Already have AG lock for this AG.
1769 *
1770 * PARAMETERS:
1771 *	imap	- pointer to inode map control structure.
1772 *	agno	- allocation group.
1773 *	ip	- pointer to new inode to be filled in on successful return
1774 *		  with the disk inode number allocated, its extent address
1775 *		  and the start of the ag.
1776 *
1777 * RETURN VALUES:
1778 *	0	- success.
1779 *	-ENOSPC	- insufficient disk resources.
1780 *	-EIO	- i/o error.
1781 */
1782static int diAllocIno(struct inomap * imap, int agno, struct inode *ip)
1783{
1784	int iagno, ino, rc, rem, extno, sword;
1785	struct metapage *mp;
1786	struct iag *iagp;
1787
1788	/* check if there are iags on the ag's free inode list.
1789	 */
1790	if ((iagno = imap->im_agctl[agno].inofree) < 0)
1791		return -ENOSPC;
1792
1793	/* obtain read lock on imap inode */
1794	IREAD_LOCK(imap->im_ipimap, RDWRLOCK_IMAP);
1795
1796	/* read the iag at the head of the list.
1797	 */
1798	if ((rc = diIAGRead(imap, iagno, &mp))) {
1799		IREAD_UNLOCK(imap->im_ipimap);
1800		return (rc);
1801	}
1802	iagp = (struct iag *) mp->data;
1803
1804	/* better be free inodes in this iag if it is on the
1805	 * list.
1806	 */
1807	if (!iagp->nfreeinos) {
1808		IREAD_UNLOCK(imap->im_ipimap);
1809		release_metapage(mp);
1810		jfs_error(ip->i_sb, "nfreeinos = 0, but iag on freelist\n");
1811		return -EIO;
1812	}
1813
1814	/* scan the free inode summary map to find an extent
1815	 * with free inodes.
1816	 */
1817	for (sword = 0;; sword++) {
1818		if (sword >= SMAPSZ) {
1819			IREAD_UNLOCK(imap->im_ipimap);
1820			release_metapage(mp);
1821			jfs_error(ip->i_sb,
1822				  "free inode not found in summary map\n");
1823			return -EIO;
1824		}
1825
1826		if (~iagp->inosmap[sword])
1827			break;
1828	}
1829
1830	/* found a extent with free inodes. determine
1831	 * the extent number.
1832	 */
1833	rem = diFindFree(le32_to_cpu(iagp->inosmap[sword]), 0);
1834	if (rem >= EXTSPERSUM) {
1835		IREAD_UNLOCK(imap->im_ipimap);
1836		release_metapage(mp);
1837		jfs_error(ip->i_sb, "no free extent found\n");
1838		return -EIO;
1839	}
1840	extno = (sword << L2EXTSPERSUM) + rem;
1841
1842	/* find the first free inode in the extent.
1843	 */
1844	rem = diFindFree(le32_to_cpu(iagp->wmap[extno]), 0);
1845	if (rem >= INOSPEREXT) {
1846		IREAD_UNLOCK(imap->im_ipimap);
1847		release_metapage(mp);
1848		jfs_error(ip->i_sb, "free inode not found\n");
1849		return -EIO;
1850	}
1851
1852	/* compute the inode number within the iag.
1853	 */
1854	ino = (extno << L2INOSPEREXT) + rem;
1855
1856	/* allocate the inode.
1857	 */
1858	rc = diAllocBit(imap, iagp, ino);
1859	IREAD_UNLOCK(imap->im_ipimap);
1860	if (rc) {
1861		release_metapage(mp);
1862		return (rc);
1863	}
1864
1865	/* set the results of the allocation and write the iag.
1866	 */
1867	diInitInode(ip, iagno, ino, extno, iagp);
1868	write_metapage(mp);
1869
1870	return (0);
1871}
1872
1873
1874/*
1875 * NAME:	diAllocExt(imap,agno,ip)
1876 *
1877 * FUNCTION:	add a new extent of free inodes to an iag, allocating
1878 *		an inode from this extent to satisfy the current allocation
1879 *		request.
1880 *
1881 *		this routine first tries to find an existing iag with free
1882 *		extents through the ag free extent list.  if list is not
1883 *		empty, the head of the list will be selected as the home
1884 *		of the new extent of free inodes.  otherwise (the list is
1885 *		empty), a new iag will be allocated for the ag to contain
1886 *		the extent.
1887 *
1888 *		once an iag has been selected, the free extent summary map
1889 *		is used to locate a free extent within the iag and diNewExt()
1890 *		is called to initialize the extent, with initialization
1891 *		including the allocation of the first inode of the extent
1892 *		for the purpose of satisfying this request.
1893 *
1894 * PARAMETERS:
1895 *	imap	- pointer to inode map control structure.
1896 *	agno	- allocation group number.
1897 *	ip	- pointer to new inode to be filled in on successful return
1898 *		  with the disk inode number allocated, its extent address
1899 *		  and the start of the ag.
1900 *
1901 * RETURN VALUES:
1902 *	0	- success.
1903 *	-ENOSPC	- insufficient disk resources.
1904 *	-EIO	- i/o error.
1905 */
1906static int diAllocExt(struct inomap * imap, int agno, struct inode *ip)
1907{
1908	int rem, iagno, sword, extno, rc;
1909	struct metapage *mp;
1910	struct iag *iagp;
1911
1912	/* check if the ag has any iags with free extents.  if not,
1913	 * allocate a new iag for the ag.
1914	 */
1915	if ((iagno = imap->im_agctl[agno].extfree) < 0) {
1916		/* If successful, diNewIAG will obtain the read lock on the
1917		 * imap inode.
1918		 */
1919		if ((rc = diNewIAG(imap, &iagno, agno, &mp))) {
1920			return (rc);
1921		}
1922		iagp = (struct iag *) mp->data;
1923
1924		/* set the ag number if this a brand new iag
1925		 */
1926		iagp->agstart =
1927		    cpu_to_le64(AGTOBLK(agno, imap->im_ipimap));
1928	} else {
1929		/* read the iag.
1930		 */
1931		IREAD_LOCK(imap->im_ipimap, RDWRLOCK_IMAP);
1932		if ((rc = diIAGRead(imap, iagno, &mp))) {
1933			IREAD_UNLOCK(imap->im_ipimap);
1934			jfs_error(ip->i_sb, "error reading iag\n");
1935			return rc;
1936		}
1937		iagp = (struct iag *) mp->data;
1938	}
1939
1940	/* using the free extent summary map, find a free extent.
1941	 */
1942	for (sword = 0;; sword++) {
1943		if (sword >= SMAPSZ) {
1944			release_metapage(mp);
1945			IREAD_UNLOCK(imap->im_ipimap);
1946			jfs_error(ip->i_sb, "free ext summary map not found\n");
1947			return -EIO;
1948		}
1949		if (~iagp->extsmap[sword])
1950			break;
1951	}
1952
1953	/* determine the extent number of the free extent.
1954	 */
1955	rem = diFindFree(le32_to_cpu(iagp->extsmap[sword]), 0);
1956	if (rem >= EXTSPERSUM) {
1957		release_metapage(mp);
1958		IREAD_UNLOCK(imap->im_ipimap);
1959		jfs_error(ip->i_sb, "free extent not found\n");
1960		return -EIO;
1961	}
1962	extno = (sword << L2EXTSPERSUM) + rem;
1963
1964	/* initialize the new extent.
1965	 */
1966	rc = diNewExt(imap, iagp, extno);
1967	IREAD_UNLOCK(imap->im_ipimap);
1968	if (rc) {
1969		/* something bad happened.  if a new iag was allocated,
1970		 * place it back on the inode map's iag free list, and
1971		 * clear the ag number information.
1972		 */
1973		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
1974			IAGFREE_LOCK(imap);
1975			iagp->iagfree = cpu_to_le32(imap->im_freeiag);
1976			imap->im_freeiag = iagno;
1977			IAGFREE_UNLOCK(imap);
1978		}
1979		write_metapage(mp);
1980		return (rc);
1981	}
1982
1983	/* set the results of the allocation and write the iag.
1984	 */
1985	diInitInode(ip, iagno, extno << L2INOSPEREXT, extno, iagp);
1986
1987	write_metapage(mp);
1988
1989	return (0);
1990}
1991
1992
1993/*
1994 * NAME:	diAllocBit(imap,iagp,ino)
1995 *
1996 * FUNCTION:	allocate a backed inode from an iag.
1997 *
1998 *		this routine performs the mechanics of allocating a
1999 *		specified inode from a backed extent.
2000 *
2001 *		if the inode to be allocated represents the last free
2002 *		inode within the iag, the iag will be removed from the
2003 *		ag free inode list.
2004 *
2005 *		a careful update approach is used to provide consistency
2006 *		in the face of updates to multiple buffers.  under this
2007 *		approach, all required buffers are obtained before making
2008 *		any updates and are held all are updates are complete.
2009 *
2010 * PRE CONDITION: Already have buffer lock on iagp.  Already have AG lock on
2011 *	this AG.  Must have read lock on imap inode.
2012 *
2013 * PARAMETERS:
2014 *	imap	- pointer to inode map control structure.
2015 *	iagp	- pointer to iag.
2016 *	ino	- inode number to be allocated within the iag.
2017 *
2018 * RETURN VALUES:
2019 *	0	- success.
2020 *	-ENOSPC	- insufficient disk resources.
2021 *	-EIO	- i/o error.
2022 */
2023static int diAllocBit(struct inomap * imap, struct iag * iagp, int ino)
2024{
2025	int extno, bitno, agno, sword, rc;
2026	struct metapage *amp = NULL, *bmp = NULL;
2027	struct iag *aiagp = NULL, *biagp = NULL;
2028	u32 mask;
2029
2030	/* check if this is the last free inode within the iag.
2031	 * if so, it will have to be removed from the ag free
2032	 * inode list, so get the iags preceding and following
2033	 * it on the list.
2034	 */
2035	if (iagp->nfreeinos == cpu_to_le32(1)) {
2036		if ((int) le32_to_cpu(iagp->inofreefwd) >= 0) {
2037			if ((rc =
2038			     diIAGRead(imap, le32_to_cpu(iagp->inofreefwd),
2039				       &amp)))
2040				return (rc);
2041			aiagp = (struct iag *) amp->data;
2042		}
2043
2044		if ((int) le32_to_cpu(iagp->inofreeback) >= 0) {
2045			if ((rc =
2046			     diIAGRead(imap,
2047				       le32_to_cpu(iagp->inofreeback),
2048				       &bmp))) {
2049				if (amp)
2050					release_metapage(amp);
2051				return (rc);
2052			}
2053			biagp = (struct iag *) bmp->data;
2054		}
2055	}
2056
2057	/* get the ag number, extent number, inode number within
2058	 * the extent.
2059	 */
2060	agno = BLKTOAG(le64_to_cpu(iagp->agstart), JFS_SBI(imap->im_ipimap->i_sb));
2061	extno = ino >> L2INOSPEREXT;
2062	bitno = ino & (INOSPEREXT - 1);
2063
2064	/* compute the mask for setting the map.
2065	 */
2066	mask = HIGHORDER >> bitno;
2067
2068	/* the inode should be free and backed.
2069	 */
2070	if (((le32_to_cpu(iagp->pmap[extno]) & mask) != 0) ||
2071	    ((le32_to_cpu(iagp->wmap[extno]) & mask) != 0) ||
2072	    (addressPXD(&iagp->inoext[extno]) == 0)) {
2073		if (amp)
2074			release_metapage(amp);
2075		if (bmp)
2076			release_metapage(bmp);
2077
2078		jfs_error(imap->im_ipimap->i_sb, "iag inconsistent\n");
2079		return -EIO;
2080	}
2081
2082	/* mark the inode as allocated in the working map.
2083	 */
2084	iagp->wmap[extno] |= cpu_to_le32(mask);
2085
2086	/* check if all inodes within the extent are now
2087	 * allocated.  if so, update the free inode summary
2088	 * map to reflect this.
2089	 */
2090	if (iagp->wmap[extno] == cpu_to_le32(ONES)) {
2091		sword = extno >> L2EXTSPERSUM;
2092		bitno = extno & (EXTSPERSUM - 1);
2093		iagp->inosmap[sword] |= cpu_to_le32(HIGHORDER >> bitno);
2094	}
2095
2096	/* if this was the last free inode in the iag, remove the
2097	 * iag from the ag free inode list.
2098	 */
2099	if (iagp->nfreeinos == cpu_to_le32(1)) {
2100		if (amp) {
2101			aiagp->inofreeback = iagp->inofreeback;
2102			write_metapage(amp);
2103		}
2104
2105		if (bmp) {
2106			biagp->inofreefwd = iagp->inofreefwd;
2107			write_metapage(bmp);
2108		} else {
2109			imap->im_agctl[agno].inofree =
2110			    le32_to_cpu(iagp->inofreefwd);
2111		}
2112		iagp->inofreefwd = iagp->inofreeback = cpu_to_le32(-1);
2113	}
2114
2115	/* update the free inode count at the iag, ag, inode
2116	 * map levels.
2117	 */
2118	le32_add_cpu(&iagp->nfreeinos, -1);
2119	imap->im_agctl[agno].numfree -= 1;
2120	atomic_dec(&imap->im_numfree);
2121
2122	return (0);
2123}
2124
2125
2126/*
2127 * NAME:	diNewExt(imap,iagp,extno)
2128 *
2129 * FUNCTION:	initialize a new extent of inodes for an iag, allocating
2130 *		the first inode of the extent for use for the current
2131 *		allocation request.
2132 *
2133 *		disk resources are allocated for the new extent of inodes
2134 *		and the inodes themselves are initialized to reflect their
2135 *		existence within the extent (i.e. their inode numbers and
2136 *		inode extent addresses are set) and their initial state
2137 *		(mode and link count are set to zero).
2138 *
2139 *		if the iag is new, it is not yet on an ag extent free list
2140 *		but will now be placed on this list.
2141 *
2142 *		if the allocation of the new extent causes the iag to
2143 *		have no free extent, the iag will be removed from the
2144 *		ag extent free list.
2145 *
2146 *		if the iag has no free backed inodes, it will be placed
2147 *		on the ag free inode list, since the addition of the new
2148 *		extent will now cause it to have free inodes.
2149 *
2150 *		a careful update approach is used to provide consistency
2151 *		(i.e. list consistency) in the face of updates to multiple
2152 *		buffers.  under this approach, all required buffers are
2153 *		obtained before making any updates and are held until all
2154 *		updates are complete.
2155 *
2156 * PRE CONDITION: Already have buffer lock on iagp.  Already have AG lock on
2157 *	this AG.  Must have read lock on imap inode.
2158 *
2159 * PARAMETERS:
2160 *	imap	- pointer to inode map control structure.
2161 *	iagp	- pointer to iag.
2162 *	extno	- extent number.
2163 *
2164 * RETURN VALUES:
2165 *	0	- success.
2166 *	-ENOSPC	- insufficient disk resources.
2167 *	-EIO	- i/o error.
2168 */
2169static int diNewExt(struct inomap * imap, struct iag * iagp, int extno)
2170{
2171	int agno, iagno, fwd, back, freei = 0, sword, rc;
2172	struct iag *aiagp = NULL, *biagp = NULL, *ciagp = NULL;
2173	struct metapage *amp, *bmp, *cmp, *dmp;
2174	struct inode *ipimap;
2175	s64 blkno, hint;
2176	int i, j;
2177	u32 mask;
2178	ino_t ino;
2179	struct dinode *dp;
2180	struct jfs_sb_info *sbi;
2181
2182	/* better have free extents.
2183	 */
2184	if (!iagp->nfreeexts) {
2185		jfs_error(imap->im_ipimap->i_sb, "no free extents\n");
2186		return -EIO;
2187	}
2188
2189	/* get the inode map inode.
2190	 */
2191	ipimap = imap->im_ipimap;
2192	sbi = JFS_SBI(ipimap->i_sb);
2193
2194	amp = bmp = cmp = NULL;
2195
2196	/* get the ag and iag numbers for this iag.
2197	 */
2198	agno = BLKTOAG(le64_to_cpu(iagp->agstart), sbi);
 
 
 
2199	iagno = le32_to_cpu(iagp->iagnum);
2200
2201	/* check if this is the last free extent within the
2202	 * iag.  if so, the iag must be removed from the ag
2203	 * free extent list, so get the iags preceding and
2204	 * following the iag on this list.
2205	 */
2206	if (iagp->nfreeexts == cpu_to_le32(1)) {
2207		if ((fwd = le32_to_cpu(iagp->extfreefwd)) >= 0) {
2208			if ((rc = diIAGRead(imap, fwd, &amp)))
2209				return (rc);
2210			aiagp = (struct iag *) amp->data;
2211		}
2212
2213		if ((back = le32_to_cpu(iagp->extfreeback)) >= 0) {
2214			if ((rc = diIAGRead(imap, back, &bmp)))
2215				goto error_out;
2216			biagp = (struct iag *) bmp->data;
2217		}
2218	} else {
2219		/* the iag has free extents.  if all extents are free
2220		 * (as is the case for a newly allocated iag), the iag
2221		 * must be added to the ag free extent list, so get
2222		 * the iag at the head of the list in preparation for
2223		 * adding this iag to this list.
2224		 */
2225		fwd = back = -1;
2226		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
2227			if ((fwd = imap->im_agctl[agno].extfree) >= 0) {
2228				if ((rc = diIAGRead(imap, fwd, &amp)))
2229					goto error_out;
2230				aiagp = (struct iag *) amp->data;
2231			}
2232		}
2233	}
2234
2235	/* check if the iag has no free inodes.  if so, the iag
2236	 * will have to be added to the ag free inode list, so get
2237	 * the iag at the head of the list in preparation for
2238	 * adding this iag to this list.  in doing this, we must
2239	 * check if we already have the iag at the head of
2240	 * the list in hand.
2241	 */
2242	if (iagp->nfreeinos == 0) {
2243		freei = imap->im_agctl[agno].inofree;
2244
2245		if (freei >= 0) {
2246			if (freei == fwd) {
2247				ciagp = aiagp;
2248			} else if (freei == back) {
2249				ciagp = biagp;
2250			} else {
2251				if ((rc = diIAGRead(imap, freei, &cmp)))
2252					goto error_out;
2253				ciagp = (struct iag *) cmp->data;
2254			}
2255			if (ciagp == NULL) {
2256				jfs_error(imap->im_ipimap->i_sb,
2257					  "ciagp == NULL\n");
2258				rc = -EIO;
2259				goto error_out;
2260			}
2261		}
2262	}
2263
2264	/* allocate disk space for the inode extent.
2265	 */
2266	if ((extno == 0) || (addressPXD(&iagp->inoext[extno - 1]) == 0))
2267		hint = ((s64) agno << sbi->bmap->db_agl2size) - 1;
2268	else
2269		hint = addressPXD(&iagp->inoext[extno - 1]) +
2270		    lengthPXD(&iagp->inoext[extno - 1]) - 1;
2271
2272	if ((rc = dbAlloc(ipimap, hint, (s64) imap->im_nbperiext, &blkno)))
2273		goto error_out;
2274
2275	/* compute the inode number of the first inode within the
2276	 * extent.
2277	 */
2278	ino = (iagno << L2INOSPERIAG) + (extno << L2INOSPEREXT);
2279
2280	/* initialize the inodes within the newly allocated extent a
2281	 * page at a time.
2282	 */
2283	for (i = 0; i < imap->im_nbperiext; i += sbi->nbperpage) {
2284		/* get a buffer for this page of disk inodes.
2285		 */
2286		dmp = get_metapage(ipimap, blkno + i, PSIZE, 1);
2287		if (dmp == NULL) {
2288			rc = -EIO;
2289			goto error_out;
2290		}
2291		dp = (struct dinode *) dmp->data;
2292
2293		/* initialize the inode number, mode, link count and
2294		 * inode extent address.
2295		 */
2296		for (j = 0; j < INOSPERPAGE; j++, dp++, ino++) {
2297			dp->di_inostamp = cpu_to_le32(sbi->inostamp);
2298			dp->di_number = cpu_to_le32(ino);
2299			dp->di_fileset = cpu_to_le32(FILESYSTEM_I);
2300			dp->di_mode = 0;
2301			dp->di_nlink = 0;
2302			PXDaddress(&(dp->di_ixpxd), blkno);
2303			PXDlength(&(dp->di_ixpxd), imap->im_nbperiext);
2304		}
2305		write_metapage(dmp);
2306	}
2307
2308	/* if this is the last free extent within the iag, remove the
2309	 * iag from the ag free extent list.
2310	 */
2311	if (iagp->nfreeexts == cpu_to_le32(1)) {
2312		if (fwd >= 0)
2313			aiagp->extfreeback = iagp->extfreeback;
2314
2315		if (back >= 0)
2316			biagp->extfreefwd = iagp->extfreefwd;
2317		else
2318			imap->im_agctl[agno].extfree =
2319			    le32_to_cpu(iagp->extfreefwd);
2320
2321		iagp->extfreefwd = iagp->extfreeback = cpu_to_le32(-1);
2322	} else {
2323		/* if the iag has all free extents (newly allocated iag),
2324		 * add the iag to the ag free extent list.
2325		 */
2326		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
2327			if (fwd >= 0)
2328				aiagp->extfreeback = cpu_to_le32(iagno);
2329
2330			iagp->extfreefwd = cpu_to_le32(fwd);
2331			iagp->extfreeback = cpu_to_le32(-1);
2332			imap->im_agctl[agno].extfree = iagno;
2333		}
2334	}
2335
2336	/* if the iag has no free inodes, add the iag to the
2337	 * ag free inode list.
2338	 */
2339	if (iagp->nfreeinos == 0) {
2340		if (freei >= 0)
2341			ciagp->inofreeback = cpu_to_le32(iagno);
2342
2343		iagp->inofreefwd =
2344		    cpu_to_le32(imap->im_agctl[agno].inofree);
2345		iagp->inofreeback = cpu_to_le32(-1);
2346		imap->im_agctl[agno].inofree = iagno;
2347	}
2348
2349	/* initialize the extent descriptor of the extent. */
2350	PXDlength(&iagp->inoext[extno], imap->im_nbperiext);
2351	PXDaddress(&iagp->inoext[extno], blkno);
2352
2353	/* initialize the working and persistent map of the extent.
2354	 * the working map will be initialized such that
2355	 * it indicates the first inode of the extent is allocated.
2356	 */
2357	iagp->wmap[extno] = cpu_to_le32(HIGHORDER);
2358	iagp->pmap[extno] = 0;
2359
2360	/* update the free inode and free extent summary maps
2361	 * for the extent to indicate the extent has free inodes
2362	 * and no longer represents a free extent.
2363	 */
2364	sword = extno >> L2EXTSPERSUM;
2365	mask = HIGHORDER >> (extno & (EXTSPERSUM - 1));
2366	iagp->extsmap[sword] |= cpu_to_le32(mask);
2367	iagp->inosmap[sword] &= cpu_to_le32(~mask);
2368
2369	/* update the free inode and free extent counts for the
2370	 * iag.
2371	 */
2372	le32_add_cpu(&iagp->nfreeinos, (INOSPEREXT - 1));
2373	le32_add_cpu(&iagp->nfreeexts, -1);
2374
2375	/* update the free and backed inode counts for the ag.
2376	 */
2377	imap->im_agctl[agno].numfree += (INOSPEREXT - 1);
2378	imap->im_agctl[agno].numinos += INOSPEREXT;
2379
2380	/* update the free and backed inode counts for the inode map.
2381	 */
2382	atomic_add(INOSPEREXT - 1, &imap->im_numfree);
2383	atomic_add(INOSPEREXT, &imap->im_numinos);
2384
2385	/* write the iags.
2386	 */
2387	if (amp)
2388		write_metapage(amp);
2389	if (bmp)
2390		write_metapage(bmp);
2391	if (cmp)
2392		write_metapage(cmp);
2393
2394	return (0);
2395
2396      error_out:
2397
2398	/* release the iags.
2399	 */
2400	if (amp)
2401		release_metapage(amp);
2402	if (bmp)
2403		release_metapage(bmp);
2404	if (cmp)
2405		release_metapage(cmp);
2406
2407	return (rc);
2408}
2409
2410
2411/*
2412 * NAME:	diNewIAG(imap,iagnop,agno)
2413 *
2414 * FUNCTION:	allocate a new iag for an allocation group.
2415 *
2416 *		first tries to allocate the iag from the inode map
2417 *		iagfree list:
2418 *		if the list has free iags, the head of the list is removed
2419 *		and returned to satisfy the request.
2420 *		if the inode map's iag free list is empty, the inode map
2421 *		is extended to hold a new iag. this new iag is initialized
2422 *		and returned to satisfy the request.
2423 *
2424 * PARAMETERS:
2425 *	imap	- pointer to inode map control structure.
2426 *	iagnop	- pointer to an iag number set with the number of the
2427 *		  newly allocated iag upon successful return.
2428 *	agno	- allocation group number.
2429 *	bpp	- Buffer pointer to be filled in with new IAG's buffer
2430 *
2431 * RETURN VALUES:
2432 *	0	- success.
2433 *	-ENOSPC	- insufficient disk resources.
2434 *	-EIO	- i/o error.
2435 *
2436 * serialization:
2437 *	AG lock held on entry/exit;
2438 *	write lock on the map is held inside;
2439 *	read lock on the map is held on successful completion;
2440 *
2441 * note: new iag transaction:
2442 * . synchronously write iag;
2443 * . write log of xtree and inode of imap;
2444 * . commit;
2445 * . synchronous write of xtree (right to left, bottom to top);
2446 * . at start of logredo(): init in-memory imap with one additional iag page;
2447 * . at end of logredo(): re-read imap inode to determine
2448 *   new imap size;
2449 */
2450static int
2451diNewIAG(struct inomap * imap, int *iagnop, int agno, struct metapage ** mpp)
2452{
2453	int rc;
2454	int iagno, i, xlen;
2455	struct inode *ipimap;
2456	struct super_block *sb;
2457	struct jfs_sb_info *sbi;
2458	struct metapage *mp;
2459	struct iag *iagp;
2460	s64 xaddr = 0;
2461	s64 blkno;
2462	tid_t tid;
2463	struct inode *iplist[1];
2464
2465	/* pick up pointers to the inode map and mount inodes */
2466	ipimap = imap->im_ipimap;
2467	sb = ipimap->i_sb;
2468	sbi = JFS_SBI(sb);
2469
2470	/* acquire the free iag lock */
2471	IAGFREE_LOCK(imap);
2472
2473	/* if there are any iags on the inode map free iag list,
2474	 * allocate the iag from the head of the list.
2475	 */
2476	if (imap->im_freeiag >= 0) {
2477		/* pick up the iag number at the head of the list */
2478		iagno = imap->im_freeiag;
2479
2480		/* determine the logical block number of the iag */
2481		blkno = IAGTOLBLK(iagno, sbi->l2nbperpage);
2482	} else {
2483		/* no free iags. the inode map will have to be extented
2484		 * to include a new iag.
2485		 */
2486
2487		/* acquire inode map lock */
2488		IWRITE_LOCK(ipimap, RDWRLOCK_IMAP);
2489
2490		if (ipimap->i_size >> L2PSIZE != imap->im_nextiag + 1) {
2491			IWRITE_UNLOCK(ipimap);
2492			IAGFREE_UNLOCK(imap);
2493			jfs_error(imap->im_ipimap->i_sb,
2494				  "ipimap->i_size is wrong\n");
2495			return -EIO;
2496		}
2497
2498
2499		/* get the next available iag number */
2500		iagno = imap->im_nextiag;
2501
2502		/* make sure that we have not exceeded the maximum inode
2503		 * number limit.
2504		 */
2505		if (iagno > (MAXIAGS - 1)) {
2506			/* release the inode map lock */
2507			IWRITE_UNLOCK(ipimap);
2508
2509			rc = -ENOSPC;
2510			goto out;
2511		}
2512
2513		/*
2514		 * synchronously append new iag page.
2515		 */
2516		/* determine the logical address of iag page to append */
2517		blkno = IAGTOLBLK(iagno, sbi->l2nbperpage);
2518
2519		/* Allocate extent for new iag page */
2520		xlen = sbi->nbperpage;
2521		if ((rc = dbAlloc(ipimap, 0, (s64) xlen, &xaddr))) {
2522			/* release the inode map lock */
2523			IWRITE_UNLOCK(ipimap);
2524
2525			goto out;
2526		}
2527
2528		/*
2529		 * start transaction of update of the inode map
2530		 * addressing structure pointing to the new iag page;
2531		 */
2532		tid = txBegin(sb, COMMIT_FORCE);
2533		mutex_lock(&JFS_IP(ipimap)->commit_mutex);
2534
2535		/* update the inode map addressing structure to point to it */
2536		if ((rc =
2537		     xtInsert(tid, ipimap, 0, blkno, xlen, &xaddr, 0))) {
2538			txEnd(tid);
2539			mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
2540			/* Free the blocks allocated for the iag since it was
2541			 * not successfully added to the inode map
2542			 */
2543			dbFree(ipimap, xaddr, (s64) xlen);
2544
2545			/* release the inode map lock */
2546			IWRITE_UNLOCK(ipimap);
2547
2548			goto out;
2549		}
2550
2551		/* update the inode map's inode to reflect the extension */
2552		ipimap->i_size += PSIZE;
2553		inode_add_bytes(ipimap, PSIZE);
2554
2555		/* assign a buffer for the page */
2556		mp = get_metapage(ipimap, blkno, PSIZE, 0);
2557		if (!mp) {
2558			/*
2559			 * This is very unlikely since we just created the
2560			 * extent, but let's try to handle it correctly
2561			 */
2562			xtTruncate(tid, ipimap, ipimap->i_size - PSIZE,
2563				   COMMIT_PWMAP);
2564
2565			txAbort(tid, 0);
2566			txEnd(tid);
2567			mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
2568
2569			/* release the inode map lock */
2570			IWRITE_UNLOCK(ipimap);
2571
2572			rc = -EIO;
2573			goto out;
2574		}
2575		iagp = (struct iag *) mp->data;
2576
2577		/* init the iag */
2578		memset(iagp, 0, sizeof(struct iag));
2579		iagp->iagnum = cpu_to_le32(iagno);
2580		iagp->inofreefwd = iagp->inofreeback = cpu_to_le32(-1);
2581		iagp->extfreefwd = iagp->extfreeback = cpu_to_le32(-1);
2582		iagp->iagfree = cpu_to_le32(-1);
2583		iagp->nfreeinos = 0;
2584		iagp->nfreeexts = cpu_to_le32(EXTSPERIAG);
2585
2586		/* initialize the free inode summary map (free extent
2587		 * summary map initialization handled by bzero).
2588		 */
2589		for (i = 0; i < SMAPSZ; i++)
2590			iagp->inosmap[i] = cpu_to_le32(ONES);
2591
2592		/*
2593		 * Write and sync the metapage
2594		 */
2595		flush_metapage(mp);
2596
2597		/*
2598		 * txCommit(COMMIT_FORCE) will synchronously write address
2599		 * index pages and inode after commit in careful update order
2600		 * of address index pages (right to left, bottom up);
2601		 */
2602		iplist[0] = ipimap;
2603		rc = txCommit(tid, 1, &iplist[0], COMMIT_FORCE);
2604
2605		txEnd(tid);
2606		mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
2607
2608		duplicateIXtree(sb, blkno, xlen, &xaddr);
2609
2610		/* update the next available iag number */
2611		imap->im_nextiag += 1;
2612
2613		/* Add the iag to the iag free list so we don't lose the iag
2614		 * if a failure happens now.
2615		 */
2616		imap->im_freeiag = iagno;
2617
2618		/* Until we have logredo working, we want the imap inode &
2619		 * control page to be up to date.
2620		 */
2621		diSync(ipimap);
2622
2623		/* release the inode map lock */
2624		IWRITE_UNLOCK(ipimap);
2625	}
2626
2627	/* obtain read lock on map */
2628	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
2629
2630	/* read the iag */
2631	if ((rc = diIAGRead(imap, iagno, &mp))) {
2632		IREAD_UNLOCK(ipimap);
2633		rc = -EIO;
2634		goto out;
2635	}
2636	iagp = (struct iag *) mp->data;
2637
2638	/* remove the iag from the iag free list */
2639	imap->im_freeiag = le32_to_cpu(iagp->iagfree);
2640	iagp->iagfree = cpu_to_le32(-1);
2641
2642	/* set the return iag number and buffer pointer */
2643	*iagnop = iagno;
2644	*mpp = mp;
2645
2646      out:
2647	/* release the iag free lock */
2648	IAGFREE_UNLOCK(imap);
2649
2650	return (rc);
2651}
2652
2653/*
2654 * NAME:	diIAGRead()
2655 *
2656 * FUNCTION:	get the buffer for the specified iag within a fileset
2657 *		or aggregate inode map.
2658 *
2659 * PARAMETERS:
2660 *	imap	- pointer to inode map control structure.
2661 *	iagno	- iag number.
2662 *	bpp	- point to buffer pointer to be filled in on successful
2663 *		  exit.
2664 *
2665 * SERIALIZATION:
2666 *	must have read lock on imap inode
2667 *	(When called by diExtendFS, the filesystem is quiesced, therefore
2668 *	 the read lock is unnecessary.)
2669 *
2670 * RETURN VALUES:
2671 *	0	- success.
2672 *	-EIO	- i/o error.
2673 */
2674static int diIAGRead(struct inomap * imap, int iagno, struct metapage ** mpp)
2675{
2676	struct inode *ipimap = imap->im_ipimap;
2677	s64 blkno;
2678
2679	/* compute the logical block number of the iag. */
2680	blkno = IAGTOLBLK(iagno, JFS_SBI(ipimap->i_sb)->l2nbperpage);
2681
2682	/* read the iag. */
2683	*mpp = read_metapage(ipimap, blkno, PSIZE, 0);
2684	if (*mpp == NULL) {
2685		return -EIO;
2686	}
2687
2688	return (0);
2689}
2690
2691/*
2692 * NAME:	diFindFree()
2693 *
2694 * FUNCTION:	find the first free bit in a word starting at
2695 *		the specified bit position.
2696 *
2697 * PARAMETERS:
2698 *	word	- word to be examined.
2699 *	start	- starting bit position.
2700 *
2701 * RETURN VALUES:
2702 *	bit position of first free bit in the word or 32 if
2703 *	no free bits were found.
2704 */
2705static int diFindFree(u32 word, int start)
2706{
2707	int bitno;
2708	assert(start < 32);
2709	/* scan the word for the first free bit. */
2710	for (word <<= start, bitno = start; bitno < 32;
2711	     bitno++, word <<= 1) {
2712		if ((word & HIGHORDER) == 0)
2713			break;
2714	}
2715	return (bitno);
2716}
2717
2718/*
2719 * NAME:	diUpdatePMap()
2720 *
2721 * FUNCTION: Update the persistent map in an IAG for the allocation or
2722 *	freeing of the specified inode.
2723 *
2724 * PRE CONDITIONS: Working map has already been updated for allocate.
2725 *
2726 * PARAMETERS:
2727 *	ipimap	- Incore inode map inode
2728 *	inum	- Number of inode to mark in permanent map
2729 *	is_free	- If 'true' indicates inode should be marked freed, otherwise
2730 *		  indicates inode should be marked allocated.
2731 *
2732 * RETURN VALUES:
2733 *		0 for success
2734 */
2735int
2736diUpdatePMap(struct inode *ipimap,
2737	     unsigned long inum, bool is_free, struct tblock * tblk)
2738{
2739	int rc;
2740	struct iag *iagp;
2741	struct metapage *mp;
2742	int iagno, ino, extno, bitno;
2743	struct inomap *imap;
2744	u32 mask;
2745	struct jfs_log *log;
2746	int lsn, difft, diffp;
2747	unsigned long flags;
2748
2749	imap = JFS_IP(ipimap)->i_imap;
2750	/* get the iag number containing the inode */
2751	iagno = INOTOIAG(inum);
2752	/* make sure that the iag is contained within the map */
2753	if (iagno >= imap->im_nextiag) {
2754		jfs_error(ipimap->i_sb, "the iag is outside the map\n");
2755		return -EIO;
2756	}
2757	/* read the iag */
2758	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
2759	rc = diIAGRead(imap, iagno, &mp);
2760	IREAD_UNLOCK(ipimap);
2761	if (rc)
2762		return (rc);
2763	metapage_wait_for_io(mp);
2764	iagp = (struct iag *) mp->data;
2765	/* get the inode number and extent number of the inode within
2766	 * the iag and the inode number within the extent.
2767	 */
2768	ino = inum & (INOSPERIAG - 1);
2769	extno = ino >> L2INOSPEREXT;
2770	bitno = ino & (INOSPEREXT - 1);
2771	mask = HIGHORDER >> bitno;
2772	/*
2773	 * mark the inode free in persistent map:
2774	 */
2775	if (is_free) {
2776		/* The inode should have been allocated both in working
2777		 * map and in persistent map;
2778		 * the inode will be freed from working map at the release
2779		 * of last reference release;
2780		 */
2781		if (!(le32_to_cpu(iagp->wmap[extno]) & mask)) {
2782			jfs_error(ipimap->i_sb,
2783				  "inode %ld not marked as allocated in wmap!\n",
2784				  inum);
2785		}
2786		if (!(le32_to_cpu(iagp->pmap[extno]) & mask)) {
2787			jfs_error(ipimap->i_sb,
2788				  "inode %ld not marked as allocated in pmap!\n",
2789				  inum);
2790		}
2791		/* update the bitmap for the extent of the freed inode */
2792		iagp->pmap[extno] &= cpu_to_le32(~mask);
2793	}
2794	/*
2795	 * mark the inode allocated in persistent map:
2796	 */
2797	else {
2798		/* The inode should be already allocated in the working map
2799		 * and should be free in persistent map;
2800		 */
2801		if (!(le32_to_cpu(iagp->wmap[extno]) & mask)) {
2802			release_metapage(mp);
2803			jfs_error(ipimap->i_sb,
2804				  "the inode is not allocated in the working map\n");
2805			return -EIO;
2806		}
2807		if ((le32_to_cpu(iagp->pmap[extno]) & mask) != 0) {
2808			release_metapage(mp);
2809			jfs_error(ipimap->i_sb,
2810				  "the inode is not free in the persistent map\n");
2811			return -EIO;
2812		}
2813		/* update the bitmap for the extent of the allocated inode */
2814		iagp->pmap[extno] |= cpu_to_le32(mask);
2815	}
2816	/*
2817	 * update iag lsn
2818	 */
2819	lsn = tblk->lsn;
2820	log = JFS_SBI(tblk->sb)->log;
2821	LOGSYNC_LOCK(log, flags);
2822	if (mp->lsn != 0) {
2823		/* inherit older/smaller lsn */
2824		logdiff(difft, lsn, log);
2825		logdiff(diffp, mp->lsn, log);
2826		if (difft < diffp) {
2827			mp->lsn = lsn;
2828			/* move mp after tblock in logsync list */
2829			list_move(&mp->synclist, &tblk->synclist);
2830		}
2831		/* inherit younger/larger clsn */
2832		assert(mp->clsn);
2833		logdiff(difft, tblk->clsn, log);
2834		logdiff(diffp, mp->clsn, log);
2835		if (difft > diffp)
2836			mp->clsn = tblk->clsn;
2837	} else {
2838		mp->log = log;
2839		mp->lsn = lsn;
2840		/* insert mp after tblock in logsync list */
2841		log->count++;
2842		list_add(&mp->synclist, &tblk->synclist);
2843		mp->clsn = tblk->clsn;
2844	}
2845	LOGSYNC_UNLOCK(log, flags);
2846	write_metapage(mp);
2847	return (0);
2848}
2849
2850/*
2851 *	diExtendFS()
2852 *
2853 * function: update imap for extendfs();
2854 *
2855 * note: AG size has been increased s.t. each k old contiguous AGs are
2856 * coalesced into a new AG;
2857 */
2858int diExtendFS(struct inode *ipimap, struct inode *ipbmap)
2859{
2860	int rc, rcx = 0;
2861	struct inomap *imap = JFS_IP(ipimap)->i_imap;
2862	struct iag *iagp = NULL, *hiagp = NULL;
2863	struct bmap *mp = JFS_SBI(ipbmap->i_sb)->bmap;
2864	struct metapage *bp, *hbp;
2865	int i, n, head;
2866	int numinos, xnuminos = 0, xnumfree = 0;
2867	s64 agstart;
2868
2869	jfs_info("diExtendFS: nextiag:%d numinos:%d numfree:%d",
2870		   imap->im_nextiag, atomic_read(&imap->im_numinos),
2871		   atomic_read(&imap->im_numfree));
2872
2873	/*
2874	 *	reconstruct imap
2875	 *
2876	 * coalesce contiguous k (newAGSize/oldAGSize) AGs;
2877	 * i.e., (AGi, ..., AGj) where i = k*n and j = k*(n+1) - 1 to AGn;
2878	 * note: new AG size = old AG size * (2**x).
2879	 */
2880
2881	/* init per AG control information im_agctl[] */
2882	for (i = 0; i < MAXAG; i++) {
2883		imap->im_agctl[i].inofree = -1;
2884		imap->im_agctl[i].extfree = -1;
2885		imap->im_agctl[i].numinos = 0;	/* number of backed inodes */
2886		imap->im_agctl[i].numfree = 0;	/* number of free backed inodes */
2887	}
2888
2889	/*
2890	 *	process each iag page of the map.
2891	 *
2892	 * rebuild AG Free Inode List, AG Free Inode Extent List;
2893	 */
2894	for (i = 0; i < imap->im_nextiag; i++) {
2895		if ((rc = diIAGRead(imap, i, &bp))) {
2896			rcx = rc;
2897			continue;
2898		}
2899		iagp = (struct iag *) bp->data;
2900		if (le32_to_cpu(iagp->iagnum) != i) {
2901			release_metapage(bp);
2902			jfs_error(ipimap->i_sb, "unexpected value of iagnum\n");
2903			return -EIO;
2904		}
2905
2906		/* leave free iag in the free iag list */
2907		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
2908			release_metapage(bp);
2909			continue;
2910		}
2911
2912		agstart = le64_to_cpu(iagp->agstart);
2913		n = agstart >> mp->db_agl2size;
2914		iagp->agstart = cpu_to_le64((s64)n << mp->db_agl2size);
2915
2916		/* compute backed inodes */
2917		numinos = (EXTSPERIAG - le32_to_cpu(iagp->nfreeexts))
2918		    << L2INOSPEREXT;
2919		if (numinos > 0) {
2920			/* merge AG backed inodes */
2921			imap->im_agctl[n].numinos += numinos;
2922			xnuminos += numinos;
2923		}
2924
2925		/* if any backed free inodes, insert at AG free inode list */
2926		if ((int) le32_to_cpu(iagp->nfreeinos) > 0) {
2927			if ((head = imap->im_agctl[n].inofree) == -1) {
2928				iagp->inofreefwd = cpu_to_le32(-1);
2929				iagp->inofreeback = cpu_to_le32(-1);
2930			} else {
2931				if ((rc = diIAGRead(imap, head, &hbp))) {
2932					rcx = rc;
2933					goto nextiag;
2934				}
2935				hiagp = (struct iag *) hbp->data;
2936				hiagp->inofreeback = iagp->iagnum;
2937				iagp->inofreefwd = cpu_to_le32(head);
2938				iagp->inofreeback = cpu_to_le32(-1);
2939				write_metapage(hbp);
2940			}
2941
2942			imap->im_agctl[n].inofree =
2943			    le32_to_cpu(iagp->iagnum);
2944
2945			/* merge AG backed free inodes */
2946			imap->im_agctl[n].numfree +=
2947			    le32_to_cpu(iagp->nfreeinos);
2948			xnumfree += le32_to_cpu(iagp->nfreeinos);
2949		}
2950
2951		/* if any free extents, insert at AG free extent list */
2952		if (le32_to_cpu(iagp->nfreeexts) > 0) {
2953			if ((head = imap->im_agctl[n].extfree) == -1) {
2954				iagp->extfreefwd = cpu_to_le32(-1);
2955				iagp->extfreeback = cpu_to_le32(-1);
2956			} else {
2957				if ((rc = diIAGRead(imap, head, &hbp))) {
2958					rcx = rc;
2959					goto nextiag;
2960				}
2961				hiagp = (struct iag *) hbp->data;
2962				hiagp->extfreeback = iagp->iagnum;
2963				iagp->extfreefwd = cpu_to_le32(head);
2964				iagp->extfreeback = cpu_to_le32(-1);
2965				write_metapage(hbp);
2966			}
2967
2968			imap->im_agctl[n].extfree =
2969			    le32_to_cpu(iagp->iagnum);
2970		}
2971
2972	      nextiag:
2973		write_metapage(bp);
2974	}
2975
2976	if (xnuminos != atomic_read(&imap->im_numinos) ||
2977	    xnumfree != atomic_read(&imap->im_numfree)) {
2978		jfs_error(ipimap->i_sb, "numinos or numfree incorrect\n");
2979		return -EIO;
2980	}
2981
2982	return rcx;
2983}
2984
2985
2986/*
2987 *	duplicateIXtree()
2988 *
2989 * serialization: IWRITE_LOCK held on entry/exit
2990 *
2991 * note: shadow page with regular inode (rel.2);
2992 */
2993static void duplicateIXtree(struct super_block *sb, s64 blkno,
2994			    int xlen, s64 *xaddr)
2995{
2996	struct jfs_superblock *j_sb;
2997	struct buffer_head *bh;
2998	struct inode *ip;
2999	tid_t tid;
3000
3001	/* if AIT2 ipmap2 is bad, do not try to update it */
3002	if (JFS_SBI(sb)->mntflag & JFS_BAD_SAIT)	/* s_flag */
3003		return;
3004	ip = diReadSpecial(sb, FILESYSTEM_I, 1);
3005	if (ip == NULL) {
3006		JFS_SBI(sb)->mntflag |= JFS_BAD_SAIT;
3007		if (readSuper(sb, &bh))
3008			return;
3009		j_sb = (struct jfs_superblock *)bh->b_data;
3010		j_sb->s_flag |= cpu_to_le32(JFS_BAD_SAIT);
3011
3012		mark_buffer_dirty(bh);
3013		sync_dirty_buffer(bh);
3014		brelse(bh);
3015		return;
3016	}
3017
3018	/* start transaction */
3019	tid = txBegin(sb, COMMIT_FORCE);
3020	/* update the inode map addressing structure to point to it */
3021	if (xtInsert(tid, ip, 0, blkno, xlen, xaddr, 0)) {
3022		JFS_SBI(sb)->mntflag |= JFS_BAD_SAIT;
3023		txAbort(tid, 1);
3024		goto cleanup;
3025
3026	}
3027	/* update the inode map's inode to reflect the extension */
3028	ip->i_size += PSIZE;
3029	inode_add_bytes(ip, PSIZE);
3030	txCommit(tid, 1, &ip, COMMIT_FORCE);
3031      cleanup:
3032	txEnd(tid);
3033	diFreeSpecial(ip);
3034}
3035
3036/*
3037 * NAME:	copy_from_dinode()
3038 *
3039 * FUNCTION:	Copies inode info from disk inode to in-memory inode
3040 *
3041 * RETURN VALUES:
3042 *	0	- success
3043 *	-ENOMEM	- insufficient memory
3044 */
3045static int copy_from_dinode(struct dinode * dip, struct inode *ip)
3046{
3047	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
3048	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
3049
3050	jfs_ip->fileset = le32_to_cpu(dip->di_fileset);
3051	jfs_ip->mode2 = le32_to_cpu(dip->di_mode);
3052	jfs_set_inode_flags(ip);
3053
3054	ip->i_mode = le32_to_cpu(dip->di_mode) & 0xffff;
3055	if (sbi->umask != -1) {
3056		ip->i_mode = (ip->i_mode & ~0777) | (0777 & ~sbi->umask);
3057		/* For directories, add x permission if r is allowed by umask */
3058		if (S_ISDIR(ip->i_mode)) {
3059			if (ip->i_mode & 0400)
3060				ip->i_mode |= 0100;
3061			if (ip->i_mode & 0040)
3062				ip->i_mode |= 0010;
3063			if (ip->i_mode & 0004)
3064				ip->i_mode |= 0001;
3065		}
3066	}
3067	set_nlink(ip, le32_to_cpu(dip->di_nlink));
3068
3069	jfs_ip->saved_uid = make_kuid(&init_user_ns, le32_to_cpu(dip->di_uid));
3070	if (!uid_valid(sbi->uid))
3071		ip->i_uid = jfs_ip->saved_uid;
3072	else {
3073		ip->i_uid = sbi->uid;
3074	}
3075
3076	jfs_ip->saved_gid = make_kgid(&init_user_ns, le32_to_cpu(dip->di_gid));
3077	if (!gid_valid(sbi->gid))
3078		ip->i_gid = jfs_ip->saved_gid;
3079	else {
3080		ip->i_gid = sbi->gid;
3081	}
3082
3083	ip->i_size = le64_to_cpu(dip->di_size);
3084	ip->i_atime.tv_sec = le32_to_cpu(dip->di_atime.tv_sec);
3085	ip->i_atime.tv_nsec = le32_to_cpu(dip->di_atime.tv_nsec);
3086	ip->i_mtime.tv_sec = le32_to_cpu(dip->di_mtime.tv_sec);
3087	ip->i_mtime.tv_nsec = le32_to_cpu(dip->di_mtime.tv_nsec);
3088	ip->i_ctime.tv_sec = le32_to_cpu(dip->di_ctime.tv_sec);
3089	ip->i_ctime.tv_nsec = le32_to_cpu(dip->di_ctime.tv_nsec);
3090	ip->i_blocks = LBLK2PBLK(ip->i_sb, le64_to_cpu(dip->di_nblocks));
3091	ip->i_generation = le32_to_cpu(dip->di_gen);
3092
3093	jfs_ip->ixpxd = dip->di_ixpxd;	/* in-memory pxd's are little-endian */
3094	jfs_ip->acl = dip->di_acl;	/* as are dxd's */
3095	jfs_ip->ea = dip->di_ea;
3096	jfs_ip->next_index = le32_to_cpu(dip->di_next_index);
3097	jfs_ip->otime = le32_to_cpu(dip->di_otime.tv_sec);
3098	jfs_ip->acltype = le32_to_cpu(dip->di_acltype);
3099
3100	if (S_ISCHR(ip->i_mode) || S_ISBLK(ip->i_mode)) {
3101		jfs_ip->dev = le32_to_cpu(dip->di_rdev);
3102		ip->i_rdev = new_decode_dev(jfs_ip->dev);
3103	}
3104
3105	if (S_ISDIR(ip->i_mode)) {
3106		memcpy(&jfs_ip->i_dirtable, &dip->di_dirtable, 384);
3107	} else if (S_ISREG(ip->i_mode) || S_ISLNK(ip->i_mode)) {
3108		memcpy(&jfs_ip->i_xtroot, &dip->di_xtroot, 288);
3109	} else
3110		memcpy(&jfs_ip->i_inline_ea, &dip->di_inlineea, 128);
3111
3112	/* Zero the in-memory-only stuff */
3113	jfs_ip->cflag = 0;
3114	jfs_ip->btindex = 0;
3115	jfs_ip->btorder = 0;
3116	jfs_ip->bxflag = 0;
3117	jfs_ip->blid = 0;
3118	jfs_ip->atlhead = 0;
3119	jfs_ip->atltail = 0;
3120	jfs_ip->xtlid = 0;
3121	return (0);
3122}
3123
3124/*
3125 * NAME:	copy_to_dinode()
3126 *
3127 * FUNCTION:	Copies inode info from in-memory inode to disk inode
3128 */
3129static void copy_to_dinode(struct dinode * dip, struct inode *ip)
3130{
3131	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
3132	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
3133
3134	dip->di_fileset = cpu_to_le32(jfs_ip->fileset);
3135	dip->di_inostamp = cpu_to_le32(sbi->inostamp);
3136	dip->di_number = cpu_to_le32(ip->i_ino);
3137	dip->di_gen = cpu_to_le32(ip->i_generation);
3138	dip->di_size = cpu_to_le64(ip->i_size);
3139	dip->di_nblocks = cpu_to_le64(PBLK2LBLK(ip->i_sb, ip->i_blocks));
3140	dip->di_nlink = cpu_to_le32(ip->i_nlink);
3141	if (!uid_valid(sbi->uid))
3142		dip->di_uid = cpu_to_le32(i_uid_read(ip));
3143	else
3144		dip->di_uid =cpu_to_le32(from_kuid(&init_user_ns,
3145						   jfs_ip->saved_uid));
3146	if (!gid_valid(sbi->gid))
3147		dip->di_gid = cpu_to_le32(i_gid_read(ip));
3148	else
3149		dip->di_gid = cpu_to_le32(from_kgid(&init_user_ns,
3150						    jfs_ip->saved_gid));
3151	/*
3152	 * mode2 is only needed for storing the higher order bits.
3153	 * Trust i_mode for the lower order ones
3154	 */
3155	if (sbi->umask == -1)
3156		dip->di_mode = cpu_to_le32((jfs_ip->mode2 & 0xffff0000) |
3157					   ip->i_mode);
3158	else /* Leave the original permissions alone */
3159		dip->di_mode = cpu_to_le32(jfs_ip->mode2);
3160
3161	dip->di_atime.tv_sec = cpu_to_le32(ip->i_atime.tv_sec);
3162	dip->di_atime.tv_nsec = cpu_to_le32(ip->i_atime.tv_nsec);
3163	dip->di_ctime.tv_sec = cpu_to_le32(ip->i_ctime.tv_sec);
3164	dip->di_ctime.tv_nsec = cpu_to_le32(ip->i_ctime.tv_nsec);
3165	dip->di_mtime.tv_sec = cpu_to_le32(ip->i_mtime.tv_sec);
3166	dip->di_mtime.tv_nsec = cpu_to_le32(ip->i_mtime.tv_nsec);
3167	dip->di_ixpxd = jfs_ip->ixpxd;	/* in-memory pxd's are little-endian */
3168	dip->di_acl = jfs_ip->acl;	/* as are dxd's */
3169	dip->di_ea = jfs_ip->ea;
3170	dip->di_next_index = cpu_to_le32(jfs_ip->next_index);
3171	dip->di_otime.tv_sec = cpu_to_le32(jfs_ip->otime);
3172	dip->di_otime.tv_nsec = 0;
3173	dip->di_acltype = cpu_to_le32(jfs_ip->acltype);
3174	if (S_ISCHR(ip->i_mode) || S_ISBLK(ip->i_mode))
3175		dip->di_rdev = cpu_to_le32(jfs_ip->dev);
3176}
v6.13.7
   1// SPDX-License-Identifier: GPL-2.0-or-later
   2/*
   3 *   Copyright (C) International Business Machines Corp., 2000-2004
 
 
 
 
 
 
 
 
 
 
 
 
 
 
   4 */
   5
   6/*
   7 *	jfs_imap.c: inode allocation map manager
   8 *
   9 * Serialization:
  10 *   Each AG has a simple lock which is used to control the serialization of
  11 *	the AG level lists.  This lock should be taken first whenever an AG
  12 *	level list will be modified or accessed.
  13 *
  14 *   Each IAG is locked by obtaining the buffer for the IAG page.
  15 *
  16 *   There is also a inode lock for the inode map inode.  A read lock needs to
  17 *	be taken whenever an IAG is read from the map or the global level
  18 *	information is read.  A write lock needs to be taken whenever the global
  19 *	level information is modified or an atomic operation needs to be used.
  20 *
  21 *	If more than one IAG is read at one time, the read lock may not
  22 *	be given up until all of the IAG's are read.  Otherwise, a deadlock
  23 *	may occur when trying to obtain the read lock while another thread
  24 *	holding the read lock is waiting on the IAG already being held.
  25 *
  26 *   The control page of the inode map is read into memory by diMount().
  27 *	Thereafter it should only be modified in memory and then it will be
  28 *	written out when the filesystem is unmounted by diUnmount().
  29 */
  30
  31#include <linux/fs.h>
  32#include <linux/buffer_head.h>
  33#include <linux/pagemap.h>
  34#include <linux/quotaops.h>
  35#include <linux/slab.h>
  36
  37#include "jfs_incore.h"
  38#include "jfs_inode.h"
  39#include "jfs_filsys.h"
  40#include "jfs_dinode.h"
  41#include "jfs_dmap.h"
  42#include "jfs_imap.h"
  43#include "jfs_metapage.h"
  44#include "jfs_superblock.h"
  45#include "jfs_debug.h"
  46
  47/*
  48 * imap locks
  49 */
  50/* iag free list lock */
  51#define IAGFREE_LOCK_INIT(imap)		mutex_init(&imap->im_freelock)
  52#define IAGFREE_LOCK(imap)		mutex_lock(&imap->im_freelock)
  53#define IAGFREE_UNLOCK(imap)		mutex_unlock(&imap->im_freelock)
  54
  55/* per ag iag list locks */
  56#define AG_LOCK_INIT(imap,index)	mutex_init(&(imap->im_aglock[index]))
  57#define AG_LOCK(imap,agno)		mutex_lock(&imap->im_aglock[agno])
  58#define AG_UNLOCK(imap,agno)		mutex_unlock(&imap->im_aglock[agno])
  59
  60/*
  61 * forward references
  62 */
  63static int diAllocAG(struct inomap *, int, bool, struct inode *);
  64static int diAllocAny(struct inomap *, int, bool, struct inode *);
  65static int diAllocBit(struct inomap *, struct iag *, int);
  66static int diAllocExt(struct inomap *, int, struct inode *);
  67static int diAllocIno(struct inomap *, int, struct inode *);
  68static int diFindFree(u32, int);
  69static int diNewExt(struct inomap *, struct iag *, int);
  70static int diNewIAG(struct inomap *, int *, int, struct metapage **);
  71static void duplicateIXtree(struct super_block *, s64, int, s64 *);
  72
  73static int diIAGRead(struct inomap * imap, int, struct metapage **);
  74static int copy_from_dinode(struct dinode *, struct inode *);
  75static void copy_to_dinode(struct dinode *, struct inode *);
  76
  77/*
  78 * NAME:	diMount()
  79 *
  80 * FUNCTION:	initialize the incore inode map control structures for
  81 *		a fileset or aggregate init time.
  82 *
  83 *		the inode map's control structure (dinomap) is
  84 *		brought in from disk and placed in virtual memory.
  85 *
  86 * PARAMETERS:
  87 *	ipimap	- pointer to inode map inode for the aggregate or fileset.
  88 *
  89 * RETURN VALUES:
  90 *	0	- success
  91 *	-ENOMEM	- insufficient free virtual memory.
  92 *	-EIO	- i/o error.
  93 */
  94int diMount(struct inode *ipimap)
  95{
  96	struct inomap *imap;
  97	struct metapage *mp;
  98	int index;
  99	struct dinomap_disk *dinom_le;
 100
 101	/*
 102	 * allocate/initialize the in-memory inode map control structure
 103	 */
 104	/* allocate the in-memory inode map control structure. */
 105	imap = kmalloc(sizeof(struct inomap), GFP_KERNEL);
 106	if (imap == NULL)
 
 107		return -ENOMEM;
 
 108
 109	/* read the on-disk inode map control structure. */
 110
 111	mp = read_metapage(ipimap,
 112			   IMAPBLKNO << JFS_SBI(ipimap->i_sb)->l2nbperpage,
 113			   PSIZE, 0);
 114	if (mp == NULL) {
 115		kfree(imap);
 116		return -EIO;
 117	}
 118
 119	/* copy the on-disk version to the in-memory version. */
 120	dinom_le = (struct dinomap_disk *) mp->data;
 121	imap->im_freeiag = le32_to_cpu(dinom_le->in_freeiag);
 122	imap->im_nextiag = le32_to_cpu(dinom_le->in_nextiag);
 123	atomic_set(&imap->im_numinos, le32_to_cpu(dinom_le->in_numinos));
 124	atomic_set(&imap->im_numfree, le32_to_cpu(dinom_le->in_numfree));
 125	imap->im_nbperiext = le32_to_cpu(dinom_le->in_nbperiext);
 126	imap->im_l2nbperiext = le32_to_cpu(dinom_le->in_l2nbperiext);
 127	for (index = 0; index < MAXAG; index++) {
 128		imap->im_agctl[index].inofree =
 129		    le32_to_cpu(dinom_le->in_agctl[index].inofree);
 130		imap->im_agctl[index].extfree =
 131		    le32_to_cpu(dinom_le->in_agctl[index].extfree);
 132		imap->im_agctl[index].numinos =
 133		    le32_to_cpu(dinom_le->in_agctl[index].numinos);
 134		imap->im_agctl[index].numfree =
 135		    le32_to_cpu(dinom_le->in_agctl[index].numfree);
 136	}
 137
 138	/* release the buffer. */
 139	release_metapage(mp);
 140
 141	/*
 142	 * allocate/initialize inode allocation map locks
 143	 */
 144	/* allocate and init iag free list lock */
 145	IAGFREE_LOCK_INIT(imap);
 146
 147	/* allocate and init ag list locks */
 148	for (index = 0; index < MAXAG; index++) {
 149		AG_LOCK_INIT(imap, index);
 150	}
 151
 152	/* bind the inode map inode and inode map control structure
 153	 * to each other.
 154	 */
 155	imap->im_ipimap = ipimap;
 156	JFS_IP(ipimap)->i_imap = imap;
 157
 158	return (0);
 159}
 160
 161
 162/*
 163 * NAME:	diUnmount()
 164 *
 165 * FUNCTION:	write to disk the incore inode map control structures for
 166 *		a fileset or aggregate at unmount time.
 167 *
 168 * PARAMETERS:
 169 *	ipimap	- pointer to inode map inode for the aggregate or fileset.
 170 *
 171 * RETURN VALUES:
 172 *	0	- success
 173 *	-ENOMEM	- insufficient free virtual memory.
 174 *	-EIO	- i/o error.
 175 */
 176int diUnmount(struct inode *ipimap, int mounterror)
 177{
 178	struct inomap *imap = JFS_IP(ipimap)->i_imap;
 179
 180	/*
 181	 * update the on-disk inode map control structure
 182	 */
 183
 184	if (!(mounterror || isReadOnly(ipimap)))
 185		diSync(ipimap);
 186
 187	/*
 188	 * Invalidate the page cache buffers
 189	 */
 190	truncate_inode_pages(ipimap->i_mapping, 0);
 191
 192	/*
 193	 * free in-memory control structure
 194	 */
 195	kfree(imap);
 196	JFS_IP(ipimap)->i_imap = NULL;
 197
 198	return (0);
 199}
 200
 201
 202/*
 203 *	diSync()
 204 */
 205int diSync(struct inode *ipimap)
 206{
 207	struct dinomap_disk *dinom_le;
 208	struct inomap *imp = JFS_IP(ipimap)->i_imap;
 209	struct metapage *mp;
 210	int index;
 211
 212	/*
 213	 * write imap global conrol page
 214	 */
 215	/* read the on-disk inode map control structure */
 216	mp = get_metapage(ipimap,
 217			  IMAPBLKNO << JFS_SBI(ipimap->i_sb)->l2nbperpage,
 218			  PSIZE, 0);
 219	if (mp == NULL) {
 220		jfs_err("diSync: get_metapage failed!");
 221		return -EIO;
 222	}
 223
 224	/* copy the in-memory version to the on-disk version */
 225	dinom_le = (struct dinomap_disk *) mp->data;
 226	dinom_le->in_freeiag = cpu_to_le32(imp->im_freeiag);
 227	dinom_le->in_nextiag = cpu_to_le32(imp->im_nextiag);
 228	dinom_le->in_numinos = cpu_to_le32(atomic_read(&imp->im_numinos));
 229	dinom_le->in_numfree = cpu_to_le32(atomic_read(&imp->im_numfree));
 230	dinom_le->in_nbperiext = cpu_to_le32(imp->im_nbperiext);
 231	dinom_le->in_l2nbperiext = cpu_to_le32(imp->im_l2nbperiext);
 232	for (index = 0; index < MAXAG; index++) {
 233		dinom_le->in_agctl[index].inofree =
 234		    cpu_to_le32(imp->im_agctl[index].inofree);
 235		dinom_le->in_agctl[index].extfree =
 236		    cpu_to_le32(imp->im_agctl[index].extfree);
 237		dinom_le->in_agctl[index].numinos =
 238		    cpu_to_le32(imp->im_agctl[index].numinos);
 239		dinom_le->in_agctl[index].numfree =
 240		    cpu_to_le32(imp->im_agctl[index].numfree);
 241	}
 242
 243	/* write out the control structure */
 244	write_metapage(mp);
 245
 246	/*
 247	 * write out dirty pages of imap
 248	 */
 249	filemap_write_and_wait(ipimap->i_mapping);
 250
 251	diWriteSpecial(ipimap, 0);
 252
 253	return (0);
 254}
 255
 256
 257/*
 258 * NAME:	diRead()
 259 *
 260 * FUNCTION:	initialize an incore inode from disk.
 261 *
 262 *		on entry, the specifed incore inode should itself
 263 *		specify the disk inode number corresponding to the
 264 *		incore inode (i.e. i_number should be initialized).
 265 *
 266 *		this routine handles incore inode initialization for
 267 *		both "special" and "regular" inodes.  special inodes
 268 *		are those required early in the mount process and
 269 *		require special handling since much of the file system
 270 *		is not yet initialized.  these "special" inodes are
 271 *		identified by a NULL inode map inode pointer and are
 272 *		actually initialized by a call to diReadSpecial().
 273 *
 274 *		for regular inodes, the iag describing the disk inode
 275 *		is read from disk to determine the inode extent address
 276 *		for the disk inode.  with the inode extent address in
 277 *		hand, the page of the extent that contains the disk
 278 *		inode is read and the disk inode is copied to the
 279 *		incore inode.
 280 *
 281 * PARAMETERS:
 282 *	ip	-  pointer to incore inode to be initialized from disk.
 283 *
 284 * RETURN VALUES:
 285 *	0	- success
 286 *	-EIO	- i/o error.
 287 *	-ENOMEM	- insufficient memory
 288 *
 289 */
 290int diRead(struct inode *ip)
 291{
 292	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
 293	int iagno, ino, extno, rc, agno;
 294	struct inode *ipimap;
 295	struct dinode *dp;
 296	struct iag *iagp;
 297	struct metapage *mp;
 298	s64 blkno, agstart;
 299	struct inomap *imap;
 300	int block_offset;
 301	int inodes_left;
 302	unsigned long pageno;
 303	int rel_inode;
 304
 305	jfs_info("diRead: ino = %ld", ip->i_ino);
 306
 307	ipimap = sbi->ipimap;
 308	JFS_IP(ip)->ipimap = ipimap;
 309
 310	/* determine the iag number for this inode (number) */
 311	iagno = INOTOIAG(ip->i_ino);
 312
 313	/* read the iag */
 
 314	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
 315	imap = JFS_IP(ipimap)->i_imap;
 316	rc = diIAGRead(imap, iagno, &mp);
 317	IREAD_UNLOCK(ipimap);
 318	if (rc) {
 319		jfs_err("diRead: diIAGRead returned %d", rc);
 320		return (rc);
 321	}
 322
 323	iagp = (struct iag *) mp->data;
 324
 325	/* determine inode extent that holds the disk inode */
 326	ino = ip->i_ino & (INOSPERIAG - 1);
 327	extno = ino >> L2INOSPEREXT;
 328
 329	if ((lengthPXD(&iagp->inoext[extno]) != imap->im_nbperiext) ||
 330	    (addressPXD(&iagp->inoext[extno]) == 0)) {
 331		release_metapage(mp);
 332		return -ESTALE;
 333	}
 334
 335	/* get disk block number of the page within the inode extent
 336	 * that holds the disk inode.
 337	 */
 338	blkno = INOPBLK(&iagp->inoext[extno], ino, sbi->l2nbperpage);
 339
 340	/* get the ag for the iag */
 341	agstart = le64_to_cpu(iagp->agstart);
 342	agno = BLKTOAG(agstart, JFS_SBI(ip->i_sb));
 343
 344	release_metapage(mp);
 345	if (agno >= MAXAG || agno < 0)
 346		return -EIO;
 347
 348	rel_inode = (ino & (INOSPERPAGE - 1));
 349	pageno = blkno >> sbi->l2nbperpage;
 350
 351	if ((block_offset = ((u32) blkno & (sbi->nbperpage - 1)))) {
 352		/*
 353		 * OS/2 didn't always align inode extents on page boundaries
 354		 */
 355		inodes_left =
 356		     (sbi->nbperpage - block_offset) << sbi->l2niperblk;
 357
 358		if (rel_inode < inodes_left)
 359			rel_inode += block_offset << sbi->l2niperblk;
 360		else {
 361			pageno += 1;
 362			rel_inode -= inodes_left;
 363		}
 364	}
 365
 366	/* read the page of disk inode */
 367	mp = read_metapage(ipimap, pageno << sbi->l2nbperpage, PSIZE, 1);
 368	if (!mp) {
 369		jfs_err("diRead: read_metapage failed");
 370		return -EIO;
 371	}
 372
 373	/* locate the disk inode requested */
 374	dp = (struct dinode *) mp->data;
 375	dp += rel_inode;
 376
 377	if (ip->i_ino != le32_to_cpu(dp->di_number)) {
 378		jfs_error(ip->i_sb, "i_ino != di_number\n");
 379		rc = -EIO;
 380	} else if (le32_to_cpu(dp->di_nlink) == 0)
 381		rc = -ESTALE;
 382	else
 383		/* copy the disk inode to the in-memory inode */
 384		rc = copy_from_dinode(dp, ip);
 385
 386	release_metapage(mp);
 387
 388	/* set the ag for the inode */
 389	JFS_IP(ip)->agstart = agstart;
 390	JFS_IP(ip)->active_ag = -1;
 391
 392	return (rc);
 393}
 394
 395
 396/*
 397 * NAME:	diReadSpecial()
 398 *
 399 * FUNCTION:	initialize a 'special' inode from disk.
 400 *
 401 *		this routines handles aggregate level inodes.  The
 402 *		inode cache cannot differentiate between the
 403 *		aggregate inodes and the filesystem inodes, so we
 404 *		handle these here.  We don't actually use the aggregate
 405 *		inode map, since these inodes are at a fixed location
 406 *		and in some cases the aggregate inode map isn't initialized
 407 *		yet.
 408 *
 409 * PARAMETERS:
 410 *	sb - filesystem superblock
 411 *	inum - aggregate inode number
 412 *	secondary - 1 if secondary aggregate inode table
 413 *
 414 * RETURN VALUES:
 415 *	new inode	- success
 416 *	NULL		- i/o error.
 417 */
 418struct inode *diReadSpecial(struct super_block *sb, ino_t inum, int secondary)
 419{
 420	struct jfs_sb_info *sbi = JFS_SBI(sb);
 421	uint address;
 422	struct dinode *dp;
 423	struct inode *ip;
 424	struct metapage *mp;
 425
 426	ip = new_inode(sb);
 427	if (ip == NULL) {
 428		jfs_err("diReadSpecial: new_inode returned NULL!");
 429		return ip;
 430	}
 431
 432	if (secondary) {
 433		address = addressPXD(&sbi->ait2) >> sbi->l2nbperpage;
 434		JFS_IP(ip)->ipimap = sbi->ipaimap2;
 435	} else {
 436		address = AITBL_OFF >> L2PSIZE;
 437		JFS_IP(ip)->ipimap = sbi->ipaimap;
 438	}
 439
 440	ASSERT(inum < INOSPEREXT);
 441
 442	ip->i_ino = inum;
 443
 444	address += inum >> 3;	/* 8 inodes per 4K page */
 445
 446	/* read the page of fixed disk inode (AIT) in raw mode */
 447	mp = read_metapage(ip, address << sbi->l2nbperpage, PSIZE, 1);
 448	if (mp == NULL) {
 449		set_nlink(ip, 1);	/* Don't want iput() deleting it */
 450		iput(ip);
 451		return (NULL);
 452	}
 453
 454	/* get the pointer to the disk inode of interest */
 455	dp = (struct dinode *) (mp->data);
 456	dp += inum % 8;		/* 8 inodes per 4K page */
 457
 458	/* copy on-disk inode to in-memory inode */
 459	if ((copy_from_dinode(dp, ip)) != 0) {
 460		/* handle bad return by returning NULL for ip */
 461		set_nlink(ip, 1);	/* Don't want iput() deleting it */
 462		iput(ip);
 463		/* release the page */
 464		release_metapage(mp);
 465		return (NULL);
 466
 467	}
 468
 469	ip->i_mapping->a_ops = &jfs_metapage_aops;
 470	mapping_set_gfp_mask(ip->i_mapping, GFP_NOFS);
 471
 472	/* Allocations to metadata inodes should not affect quotas */
 473	ip->i_flags |= S_NOQUOTA;
 474
 475	if ((inum == FILESYSTEM_I) && (JFS_IP(ip)->ipimap == sbi->ipaimap)) {
 476		sbi->gengen = le32_to_cpu(dp->di_gengen);
 477		sbi->inostamp = le32_to_cpu(dp->di_inostamp);
 478	}
 479
 480	/* release the page */
 481	release_metapage(mp);
 482
 483	inode_fake_hash(ip);
 
 
 
 
 
 
 484
 485	return (ip);
 486}
 487
 488/*
 489 * NAME:	diWriteSpecial()
 490 *
 491 * FUNCTION:	Write the special inode to disk
 492 *
 493 * PARAMETERS:
 494 *	ip - special inode
 495 *	secondary - 1 if secondary aggregate inode table
 496 *
 497 * RETURN VALUES: none
 498 */
 499
 500void diWriteSpecial(struct inode *ip, int secondary)
 501{
 502	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
 503	uint address;
 504	struct dinode *dp;
 505	ino_t inum = ip->i_ino;
 506	struct metapage *mp;
 507
 508	if (secondary)
 509		address = addressPXD(&sbi->ait2) >> sbi->l2nbperpage;
 510	else
 511		address = AITBL_OFF >> L2PSIZE;
 512
 513	ASSERT(inum < INOSPEREXT);
 514
 515	address += inum >> 3;	/* 8 inodes per 4K page */
 516
 517	/* read the page of fixed disk inode (AIT) in raw mode */
 518	mp = read_metapage(ip, address << sbi->l2nbperpage, PSIZE, 1);
 519	if (mp == NULL) {
 520		jfs_err("diWriteSpecial: failed to read aggregate inode extent!");
 521		return;
 522	}
 523
 524	/* get the pointer to the disk inode of interest */
 525	dp = (struct dinode *) (mp->data);
 526	dp += inum % 8;		/* 8 inodes per 4K page */
 527
 528	/* copy on-disk inode to in-memory inode */
 529	copy_to_dinode(dp, ip);
 530	memcpy(&dp->di_xtroot, &JFS_IP(ip)->i_xtroot, 288);
 531
 532	if (inum == FILESYSTEM_I)
 533		dp->di_gengen = cpu_to_le32(sbi->gengen);
 534
 535	/* write the page */
 536	write_metapage(mp);
 537}
 538
 539/*
 540 * NAME:	diFreeSpecial()
 541 *
 542 * FUNCTION:	Free allocated space for special inode
 543 */
 544void diFreeSpecial(struct inode *ip)
 545{
 546	if (ip == NULL) {
 547		jfs_err("diFreeSpecial called with NULL ip!");
 548		return;
 549	}
 550	filemap_write_and_wait(ip->i_mapping);
 551	truncate_inode_pages(ip->i_mapping, 0);
 552	iput(ip);
 553}
 554
 555
 556
 557/*
 558 * NAME:	diWrite()
 559 *
 560 * FUNCTION:	write the on-disk inode portion of the in-memory inode
 561 *		to its corresponding on-disk inode.
 562 *
 563 *		on entry, the specifed incore inode should itself
 564 *		specify the disk inode number corresponding to the
 565 *		incore inode (i.e. i_number should be initialized).
 566 *
 567 *		the inode contains the inode extent address for the disk
 568 *		inode.  with the inode extent address in hand, the
 569 *		page of the extent that contains the disk inode is
 570 *		read and the disk inode portion of the incore inode
 571 *		is copied to the disk inode.
 572 *
 573 * PARAMETERS:
 574 *	tid -  transacation id
 575 *	ip  -  pointer to incore inode to be written to the inode extent.
 576 *
 577 * RETURN VALUES:
 578 *	0	- success
 579 *	-EIO	- i/o error.
 580 */
 581int diWrite(tid_t tid, struct inode *ip)
 582{
 583	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
 584	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
 585	int rc = 0;
 586	s32 ino;
 587	struct dinode *dp;
 588	s64 blkno;
 589	int block_offset;
 590	int inodes_left;
 591	struct metapage *mp;
 592	unsigned long pageno;
 593	int rel_inode;
 594	int dioffset;
 595	struct inode *ipimap;
 596	uint type;
 597	lid_t lid;
 598	struct tlock *ditlck, *tlck;
 599	struct linelock *dilinelock, *ilinelock;
 600	struct lv *lv;
 601	int n;
 602
 603	ipimap = jfs_ip->ipimap;
 604
 605	ino = ip->i_ino & (INOSPERIAG - 1);
 606
 607	if (!addressPXD(&(jfs_ip->ixpxd)) ||
 608	    (lengthPXD(&(jfs_ip->ixpxd)) !=
 609	     JFS_IP(ipimap)->i_imap->im_nbperiext)) {
 610		jfs_error(ip->i_sb, "ixpxd invalid\n");
 611		return -EIO;
 612	}
 613
 614	/*
 615	 * read the page of disk inode containing the specified inode:
 616	 */
 617	/* compute the block address of the page */
 618	blkno = INOPBLK(&(jfs_ip->ixpxd), ino, sbi->l2nbperpage);
 619
 620	rel_inode = (ino & (INOSPERPAGE - 1));
 621	pageno = blkno >> sbi->l2nbperpage;
 622
 623	if ((block_offset = ((u32) blkno & (sbi->nbperpage - 1)))) {
 624		/*
 625		 * OS/2 didn't always align inode extents on page boundaries
 626		 */
 627		inodes_left =
 628		    (sbi->nbperpage - block_offset) << sbi->l2niperblk;
 629
 630		if (rel_inode < inodes_left)
 631			rel_inode += block_offset << sbi->l2niperblk;
 632		else {
 633			pageno += 1;
 634			rel_inode -= inodes_left;
 635		}
 636	}
 637	/* read the page of disk inode */
 638      retry:
 639	mp = read_metapage(ipimap, pageno << sbi->l2nbperpage, PSIZE, 1);
 640	if (!mp)
 641		return -EIO;
 642
 643	/* get the pointer to the disk inode */
 644	dp = (struct dinode *) mp->data;
 645	dp += rel_inode;
 646
 647	dioffset = (ino & (INOSPERPAGE - 1)) << L2DISIZE;
 648
 649	/*
 650	 * acquire transaction lock on the on-disk inode;
 651	 * N.B. tlock is acquired on ipimap not ip;
 652	 */
 653	if ((ditlck =
 654	     txLock(tid, ipimap, mp, tlckINODE | tlckENTRY)) == NULL)
 655		goto retry;
 656	dilinelock = (struct linelock *) & ditlck->lock;
 657
 658	/*
 659	 * copy btree root from in-memory inode to on-disk inode
 660	 *
 661	 * (tlock is taken from inline B+-tree root in in-memory
 662	 * inode when the B+-tree root is updated, which is pointed
 663	 * by jfs_ip->blid as well as being on tx tlock list)
 664	 *
 665	 * further processing of btree root is based on the copy
 666	 * in in-memory inode, where txLog() will log from, and,
 667	 * for xtree root, txUpdateMap() will update map and reset
 668	 * XAD_NEW bit;
 669	 */
 670
 671	if (S_ISDIR(ip->i_mode) && (lid = jfs_ip->xtlid)) {
 672		/*
 673		 * This is the special xtree inside the directory for storing
 674		 * the directory table
 675		 */
 676		xtroot_t *p, *xp;
 677		xad_t *xad;
 678
 679		jfs_ip->xtlid = 0;
 680		tlck = lid_to_tlock(lid);
 681		assert(tlck->type & tlckXTREE);
 682		tlck->type |= tlckBTROOT;
 683		tlck->mp = mp;
 684		ilinelock = (struct linelock *) & tlck->lock;
 685
 686		/*
 687		 * copy xtree root from inode to dinode:
 688		 */
 689		p = &jfs_ip->i_xtroot;
 690		xp = (xtroot_t *) &dp->di_dirtable;
 691		lv = ilinelock->lv;
 692		for (n = 0; n < ilinelock->index; n++, lv++) {
 693			memcpy(&xp->xad[lv->offset], &p->xad[lv->offset],
 694			       lv->length << L2XTSLOTSIZE);
 695		}
 696
 697		/* reset on-disk (metadata page) xtree XAD_NEW bit */
 698		xad = &xp->xad[XTENTRYSTART];
 699		for (n = XTENTRYSTART;
 700		     n < le16_to_cpu(xp->header.nextindex); n++, xad++)
 701			if (xad->flag & (XAD_NEW | XAD_EXTENDED))
 702				xad->flag &= ~(XAD_NEW | XAD_EXTENDED);
 703	}
 704
 705	if ((lid = jfs_ip->blid) == 0)
 706		goto inlineData;
 707	jfs_ip->blid = 0;
 708
 709	tlck = lid_to_tlock(lid);
 710	type = tlck->type;
 711	tlck->type |= tlckBTROOT;
 712	tlck->mp = mp;
 713	ilinelock = (struct linelock *) & tlck->lock;
 714
 715	/*
 716	 *	regular file: 16 byte (XAD slot) granularity
 717	 */
 718	if (type & tlckXTREE) {
 719		xtroot_t *p, *xp;
 720		xad_t *xad;
 721
 722		/*
 723		 * copy xtree root from inode to dinode:
 724		 */
 725		p = &jfs_ip->i_xtroot;
 726		xp = &dp->di_xtroot;
 727		lv = ilinelock->lv;
 728		for (n = 0; n < ilinelock->index; n++, lv++) {
 729			memcpy(&xp->xad[lv->offset], &p->xad[lv->offset],
 730			       lv->length << L2XTSLOTSIZE);
 731		}
 732
 733		/* reset on-disk (metadata page) xtree XAD_NEW bit */
 734		xad = &xp->xad[XTENTRYSTART];
 735		for (n = XTENTRYSTART;
 736		     n < le16_to_cpu(xp->header.nextindex); n++, xad++)
 737			if (xad->flag & (XAD_NEW | XAD_EXTENDED))
 738				xad->flag &= ~(XAD_NEW | XAD_EXTENDED);
 739	}
 740	/*
 741	 *	directory: 32 byte (directory entry slot) granularity
 742	 */
 743	else if (type & tlckDTREE) {
 744		dtpage_t *p, *xp;
 745
 746		/*
 747		 * copy dtree root from inode to dinode:
 748		 */
 749		p = (dtpage_t *) &jfs_ip->i_dtroot;
 750		xp = (dtpage_t *) & dp->di_dtroot;
 751		lv = ilinelock->lv;
 752		for (n = 0; n < ilinelock->index; n++, lv++) {
 753			memcpy(&xp->slot[lv->offset], &p->slot[lv->offset],
 754			       lv->length << L2DTSLOTSIZE);
 755		}
 756	} else {
 757		jfs_err("diWrite: UFO tlock");
 758	}
 759
 760      inlineData:
 761	/*
 762	 * copy inline symlink from in-memory inode to on-disk inode
 763	 */
 764	if (S_ISLNK(ip->i_mode) && ip->i_size < IDATASIZE) {
 765		lv = & dilinelock->lv[dilinelock->index];
 766		lv->offset = (dioffset + 2 * 128) >> L2INODESLOTSIZE;
 767		lv->length = 2;
 768		memcpy(&dp->di_inline_all, jfs_ip->i_inline_all, IDATASIZE);
 769		dilinelock->index++;
 770	}
 771	/*
 772	 * copy inline data from in-memory inode to on-disk inode:
 773	 * 128 byte slot granularity
 774	 */
 775	if (test_cflag(COMMIT_Inlineea, ip)) {
 776		lv = & dilinelock->lv[dilinelock->index];
 777		lv->offset = (dioffset + 3 * 128) >> L2INODESLOTSIZE;
 778		lv->length = 1;
 779		memcpy(&dp->di_inlineea, jfs_ip->i_inline_ea, INODESLOTSIZE);
 780		dilinelock->index++;
 781
 782		clear_cflag(COMMIT_Inlineea, ip);
 783	}
 784
 785	/*
 786	 *	lock/copy inode base: 128 byte slot granularity
 787	 */
 788	lv = & dilinelock->lv[dilinelock->index];
 789	lv->offset = dioffset >> L2INODESLOTSIZE;
 790	copy_to_dinode(dp, ip);
 791	if (test_and_clear_cflag(COMMIT_Dirtable, ip)) {
 792		lv->length = 2;
 793		memcpy(&dp->di_dirtable, &jfs_ip->i_dirtable, 96);
 794	} else
 795		lv->length = 1;
 796	dilinelock->index++;
 797
 798	/* release the buffer holding the updated on-disk inode.
 799	 * the buffer will be later written by commit processing.
 800	 */
 801	write_metapage(mp);
 802
 803	return (rc);
 804}
 805
 806
 807/*
 808 * NAME:	diFree(ip)
 809 *
 810 * FUNCTION:	free a specified inode from the inode working map
 811 *		for a fileset or aggregate.
 812 *
 813 *		if the inode to be freed represents the first (only)
 814 *		free inode within the iag, the iag will be placed on
 815 *		the ag free inode list.
 816 *
 817 *		freeing the inode will cause the inode extent to be
 818 *		freed if the inode is the only allocated inode within
 819 *		the extent.  in this case all the disk resource backing
 820 *		up the inode extent will be freed. in addition, the iag
 821 *		will be placed on the ag extent free list if the extent
 822 *		is the first free extent in the iag.  if freeing the
 823 *		extent also means that no free inodes will exist for
 824 *		the iag, the iag will also be removed from the ag free
 825 *		inode list.
 826 *
 827 *		the iag describing the inode will be freed if the extent
 828 *		is to be freed and it is the only backed extent within
 829 *		the iag.  in this case, the iag will be removed from the
 830 *		ag free extent list and ag free inode list and placed on
 831 *		the inode map's free iag list.
 832 *
 833 *		a careful update approach is used to provide consistency
 834 *		in the face of updates to multiple buffers.  under this
 835 *		approach, all required buffers are obtained before making
 836 *		any updates and are held until all updates are complete.
 837 *
 838 * PARAMETERS:
 839 *	ip	- inode to be freed.
 840 *
 841 * RETURN VALUES:
 842 *	0	- success
 843 *	-EIO	- i/o error.
 844 */
 845int diFree(struct inode *ip)
 846{
 847	int rc;
 848	ino_t inum = ip->i_ino;
 849	struct iag *iagp, *aiagp, *biagp, *ciagp, *diagp;
 850	struct metapage *mp, *amp, *bmp, *cmp, *dmp;
 851	int iagno, ino, extno, bitno, sword, agno;
 852	int back, fwd;
 853	u32 bitmap, mask;
 854	struct inode *ipimap = JFS_SBI(ip->i_sb)->ipimap;
 855	struct inomap *imap = JFS_IP(ipimap)->i_imap;
 856	pxd_t freepxd;
 857	tid_t tid;
 858	struct inode *iplist[3];
 859	struct tlock *tlck;
 860	struct pxd_lock *pxdlock;
 861
 862	/*
 863	 * This is just to suppress compiler warnings.  The same logic that
 864	 * references these variables is used to initialize them.
 865	 */
 866	aiagp = biagp = ciagp = diagp = NULL;
 867
 868	/* get the iag number containing the inode.
 869	 */
 870	iagno = INOTOIAG(inum);
 871
 872	/* make sure that the iag is contained within
 873	 * the map.
 874	 */
 875	if (iagno >= imap->im_nextiag) {
 876		print_hex_dump(KERN_ERR, "imap: ", DUMP_PREFIX_ADDRESS, 16, 4,
 877			       imap, 32, 0);
 878		jfs_error(ip->i_sb, "inum = %d, iagno = %d, nextiag = %d\n",
 879			  (uint) inum, iagno, imap->im_nextiag);
 880		return -EIO;
 881	}
 882
 883	/* get the allocation group for this ino.
 884	 */
 885	agno = BLKTOAG(JFS_IP(ip)->agstart, JFS_SBI(ip->i_sb));
 886
 887	/* Lock the AG specific inode map information
 888	 */
 889	AG_LOCK(imap, agno);
 890
 891	/* Obtain read lock in imap inode.  Don't release it until we have
 892	 * read all of the IAG's that we are going to.
 893	 */
 894	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
 895
 896	/* read the iag.
 897	 */
 898	if ((rc = diIAGRead(imap, iagno, &mp))) {
 899		IREAD_UNLOCK(ipimap);
 900		AG_UNLOCK(imap, agno);
 901		return (rc);
 902	}
 903	iagp = (struct iag *) mp->data;
 904
 905	/* get the inode number and extent number of the inode within
 906	 * the iag and the inode number within the extent.
 907	 */
 908	ino = inum & (INOSPERIAG - 1);
 909	extno = ino >> L2INOSPEREXT;
 910	bitno = ino & (INOSPEREXT - 1);
 911	mask = HIGHORDER >> bitno;
 912
 913	if (!(le32_to_cpu(iagp->wmap[extno]) & mask)) {
 914		jfs_error(ip->i_sb, "wmap shows inode already free\n");
 915	}
 916
 917	if (!addressPXD(&iagp->inoext[extno])) {
 918		release_metapage(mp);
 919		IREAD_UNLOCK(ipimap);
 920		AG_UNLOCK(imap, agno);
 921		jfs_error(ip->i_sb, "invalid inoext\n");
 922		return -EIO;
 923	}
 924
 925	/* compute the bitmap for the extent reflecting the freed inode.
 926	 */
 927	bitmap = le32_to_cpu(iagp->wmap[extno]) & ~mask;
 928
 929	if (imap->im_agctl[agno].numfree > imap->im_agctl[agno].numinos) {
 930		release_metapage(mp);
 931		IREAD_UNLOCK(ipimap);
 932		AG_UNLOCK(imap, agno);
 933		jfs_error(ip->i_sb, "numfree > numinos\n");
 934		return -EIO;
 935	}
 936	/*
 937	 *	inode extent still has some inodes or below low water mark:
 938	 *	keep the inode extent;
 939	 */
 940	if (bitmap ||
 941	    imap->im_agctl[agno].numfree < 96 ||
 942	    (imap->im_agctl[agno].numfree < 288 &&
 943	     (((imap->im_agctl[agno].numfree * 100) /
 944	       imap->im_agctl[agno].numinos) <= 25))) {
 945		/* if the iag currently has no free inodes (i.e.,
 946		 * the inode being freed is the first free inode of iag),
 947		 * insert the iag at head of the inode free list for the ag.
 948		 */
 949		if (iagp->nfreeinos == 0) {
 950			/* check if there are any iags on the ag inode
 951			 * free list.  if so, read the first one so that
 952			 * we can link the current iag onto the list at
 953			 * the head.
 954			 */
 955			if ((fwd = imap->im_agctl[agno].inofree) >= 0) {
 956				/* read the iag that currently is the head
 957				 * of the list.
 958				 */
 959				if ((rc = diIAGRead(imap, fwd, &amp))) {
 960					IREAD_UNLOCK(ipimap);
 961					AG_UNLOCK(imap, agno);
 962					release_metapage(mp);
 963					return (rc);
 964				}
 965				aiagp = (struct iag *) amp->data;
 966
 967				/* make current head point back to the iag.
 968				 */
 969				aiagp->inofreeback = cpu_to_le32(iagno);
 970
 971				write_metapage(amp);
 972			}
 973
 974			/* iag points forward to current head and iag
 975			 * becomes the new head of the list.
 976			 */
 977			iagp->inofreefwd =
 978			    cpu_to_le32(imap->im_agctl[agno].inofree);
 979			iagp->inofreeback = cpu_to_le32(-1);
 980			imap->im_agctl[agno].inofree = iagno;
 981		}
 982		IREAD_UNLOCK(ipimap);
 983
 984		/* update the free inode summary map for the extent if
 985		 * freeing the inode means the extent will now have free
 986		 * inodes (i.e., the inode being freed is the first free
 987		 * inode of extent),
 988		 */
 989		if (iagp->wmap[extno] == cpu_to_le32(ONES)) {
 990			sword = extno >> L2EXTSPERSUM;
 991			bitno = extno & (EXTSPERSUM - 1);
 992			iagp->inosmap[sword] &=
 993			    cpu_to_le32(~(HIGHORDER >> bitno));
 994		}
 995
 996		/* update the bitmap.
 997		 */
 998		iagp->wmap[extno] = cpu_to_le32(bitmap);
 999
1000		/* update the free inode counts at the iag, ag and
1001		 * map level.
1002		 */
1003		le32_add_cpu(&iagp->nfreeinos, 1);
1004		imap->im_agctl[agno].numfree += 1;
1005		atomic_inc(&imap->im_numfree);
1006
1007		/* release the AG inode map lock
1008		 */
1009		AG_UNLOCK(imap, agno);
1010
1011		/* write the iag */
1012		write_metapage(mp);
1013
1014		return (0);
1015	}
1016
1017
1018	/*
1019	 *	inode extent has become free and above low water mark:
1020	 *	free the inode extent;
1021	 */
1022
1023	/*
1024	 *	prepare to update iag list(s) (careful update step 1)
1025	 */
1026	amp = bmp = cmp = dmp = NULL;
1027	fwd = back = -1;
1028
1029	/* check if the iag currently has no free extents.  if so,
1030	 * it will be placed on the head of the ag extent free list.
1031	 */
1032	if (iagp->nfreeexts == 0) {
1033		/* check if the ag extent free list has any iags.
1034		 * if so, read the iag at the head of the list now.
1035		 * this (head) iag will be updated later to reflect
1036		 * the addition of the current iag at the head of
1037		 * the list.
1038		 */
1039		if ((fwd = imap->im_agctl[agno].extfree) >= 0) {
1040			if ((rc = diIAGRead(imap, fwd, &amp)))
1041				goto error_out;
1042			aiagp = (struct iag *) amp->data;
1043		}
1044	} else {
1045		/* iag has free extents. check if the addition of a free
1046		 * extent will cause all extents to be free within this
1047		 * iag.  if so, the iag will be removed from the ag extent
1048		 * free list and placed on the inode map's free iag list.
1049		 */
1050		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG - 1)) {
1051			/* in preparation for removing the iag from the
1052			 * ag extent free list, read the iags preceding
1053			 * and following the iag on the ag extent free
1054			 * list.
1055			 */
1056			if ((fwd = le32_to_cpu(iagp->extfreefwd)) >= 0) {
1057				if ((rc = diIAGRead(imap, fwd, &amp)))
1058					goto error_out;
1059				aiagp = (struct iag *) amp->data;
1060			}
1061
1062			if ((back = le32_to_cpu(iagp->extfreeback)) >= 0) {
1063				if ((rc = diIAGRead(imap, back, &bmp)))
1064					goto error_out;
1065				biagp = (struct iag *) bmp->data;
1066			}
1067		}
1068	}
1069
1070	/* remove the iag from the ag inode free list if freeing
1071	 * this extent cause the iag to have no free inodes.
1072	 */
1073	if (iagp->nfreeinos == cpu_to_le32(INOSPEREXT - 1)) {
1074		int inofreeback = le32_to_cpu(iagp->inofreeback);
1075		int inofreefwd = le32_to_cpu(iagp->inofreefwd);
1076
1077		/* in preparation for removing the iag from the
1078		 * ag inode free list, read the iags preceding
1079		 * and following the iag on the ag inode free
1080		 * list.  before reading these iags, we must make
1081		 * sure that we already don't have them in hand
1082		 * from up above, since re-reading an iag (buffer)
1083		 * we are currently holding would cause a deadlock.
1084		 */
1085		if (inofreefwd >= 0) {
1086
1087			if (inofreefwd == fwd)
1088				ciagp = (struct iag *) amp->data;
1089			else if (inofreefwd == back)
1090				ciagp = (struct iag *) bmp->data;
1091			else {
1092				if ((rc =
1093				     diIAGRead(imap, inofreefwd, &cmp)))
1094					goto error_out;
1095				ciagp = (struct iag *) cmp->data;
1096			}
1097			assert(ciagp != NULL);
1098		}
1099
1100		if (inofreeback >= 0) {
1101			if (inofreeback == fwd)
1102				diagp = (struct iag *) amp->data;
1103			else if (inofreeback == back)
1104				diagp = (struct iag *) bmp->data;
1105			else {
1106				if ((rc =
1107				     diIAGRead(imap, inofreeback, &dmp)))
1108					goto error_out;
1109				diagp = (struct iag *) dmp->data;
1110			}
1111			assert(diagp != NULL);
1112		}
1113	}
1114
1115	IREAD_UNLOCK(ipimap);
1116
1117	/*
1118	 * invalidate any page of the inode extent freed from buffer cache;
1119	 */
1120	freepxd = iagp->inoext[extno];
1121	invalidate_pxd_metapages(ip, freepxd);
1122
1123	/*
1124	 *	update iag list(s) (careful update step 2)
1125	 */
1126	/* add the iag to the ag extent free list if this is the
1127	 * first free extent for the iag.
1128	 */
1129	if (iagp->nfreeexts == 0) {
1130		if (fwd >= 0)
1131			aiagp->extfreeback = cpu_to_le32(iagno);
1132
1133		iagp->extfreefwd =
1134		    cpu_to_le32(imap->im_agctl[agno].extfree);
1135		iagp->extfreeback = cpu_to_le32(-1);
1136		imap->im_agctl[agno].extfree = iagno;
1137	} else {
1138		/* remove the iag from the ag extent list if all extents
1139		 * are now free and place it on the inode map iag free list.
1140		 */
1141		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG - 1)) {
1142			if (fwd >= 0)
1143				aiagp->extfreeback = iagp->extfreeback;
1144
1145			if (back >= 0)
1146				biagp->extfreefwd = iagp->extfreefwd;
1147			else
1148				imap->im_agctl[agno].extfree =
1149				    le32_to_cpu(iagp->extfreefwd);
1150
1151			iagp->extfreefwd = iagp->extfreeback = cpu_to_le32(-1);
1152
1153			IAGFREE_LOCK(imap);
1154			iagp->iagfree = cpu_to_le32(imap->im_freeiag);
1155			imap->im_freeiag = iagno;
1156			IAGFREE_UNLOCK(imap);
1157		}
1158	}
1159
1160	/* remove the iag from the ag inode free list if freeing
1161	 * this extent causes the iag to have no free inodes.
1162	 */
1163	if (iagp->nfreeinos == cpu_to_le32(INOSPEREXT - 1)) {
1164		if ((int) le32_to_cpu(iagp->inofreefwd) >= 0)
1165			ciagp->inofreeback = iagp->inofreeback;
1166
1167		if ((int) le32_to_cpu(iagp->inofreeback) >= 0)
1168			diagp->inofreefwd = iagp->inofreefwd;
1169		else
1170			imap->im_agctl[agno].inofree =
1171			    le32_to_cpu(iagp->inofreefwd);
1172
1173		iagp->inofreefwd = iagp->inofreeback = cpu_to_le32(-1);
1174	}
1175
1176	/* update the inode extent address and working map
1177	 * to reflect the free extent.
1178	 * the permanent map should have been updated already
1179	 * for the inode being freed.
1180	 */
1181	if (iagp->pmap[extno] != 0) {
1182		jfs_error(ip->i_sb, "the pmap does not show inode free\n");
1183	}
1184	iagp->wmap[extno] = 0;
1185	PXDlength(&iagp->inoext[extno], 0);
1186	PXDaddress(&iagp->inoext[extno], 0);
1187
1188	/* update the free extent and free inode summary maps
1189	 * to reflect the freed extent.
1190	 * the inode summary map is marked to indicate no inodes
1191	 * available for the freed extent.
1192	 */
1193	sword = extno >> L2EXTSPERSUM;
1194	bitno = extno & (EXTSPERSUM - 1);
1195	mask = HIGHORDER >> bitno;
1196	iagp->inosmap[sword] |= cpu_to_le32(mask);
1197	iagp->extsmap[sword] &= cpu_to_le32(~mask);
1198
1199	/* update the number of free inodes and number of free extents
1200	 * for the iag.
1201	 */
1202	le32_add_cpu(&iagp->nfreeinos, -(INOSPEREXT - 1));
1203	le32_add_cpu(&iagp->nfreeexts, 1);
1204
1205	/* update the number of free inodes and backed inodes
1206	 * at the ag and inode map level.
1207	 */
1208	imap->im_agctl[agno].numfree -= (INOSPEREXT - 1);
1209	imap->im_agctl[agno].numinos -= INOSPEREXT;
1210	atomic_sub(INOSPEREXT - 1, &imap->im_numfree);
1211	atomic_sub(INOSPEREXT, &imap->im_numinos);
1212
1213	if (amp)
1214		write_metapage(amp);
1215	if (bmp)
1216		write_metapage(bmp);
1217	if (cmp)
1218		write_metapage(cmp);
1219	if (dmp)
1220		write_metapage(dmp);
1221
1222	/*
1223	 * start transaction to update block allocation map
1224	 * for the inode extent freed;
1225	 *
1226	 * N.B. AG_LOCK is released and iag will be released below, and
1227	 * other thread may allocate inode from/reusing the ixad freed
1228	 * BUT with new/different backing inode extent from the extent
1229	 * to be freed by the transaction;
1230	 */
1231	tid = txBegin(ipimap->i_sb, COMMIT_FORCE);
1232	mutex_lock(&JFS_IP(ipimap)->commit_mutex);
1233
1234	/* acquire tlock of the iag page of the freed ixad
1235	 * to force the page NOHOMEOK (even though no data is
1236	 * logged from the iag page) until NOREDOPAGE|FREEXTENT log
1237	 * for the free of the extent is committed;
1238	 * write FREEXTENT|NOREDOPAGE log record
1239	 * N.B. linelock is overlaid as freed extent descriptor;
1240	 */
1241	tlck = txLock(tid, ipimap, mp, tlckINODE | tlckFREE);
1242	pxdlock = (struct pxd_lock *) & tlck->lock;
1243	pxdlock->flag = mlckFREEPXD;
1244	pxdlock->pxd = freepxd;
1245	pxdlock->index = 1;
1246
1247	write_metapage(mp);
1248
1249	iplist[0] = ipimap;
1250
1251	/*
1252	 * logredo needs the IAG number and IAG extent index in order
1253	 * to ensure that the IMap is consistent.  The least disruptive
1254	 * way to pass these values through  to the transaction manager
1255	 * is in the iplist array.
1256	 *
1257	 * It's not pretty, but it works.
1258	 */
1259	iplist[1] = (struct inode *) (size_t)iagno;
1260	iplist[2] = (struct inode *) (size_t)extno;
1261
1262	rc = txCommit(tid, 1, &iplist[0], COMMIT_FORCE);
1263
1264	txEnd(tid);
1265	mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
1266
1267	/* unlock the AG inode map information */
1268	AG_UNLOCK(imap, agno);
1269
1270	return (0);
1271
1272      error_out:
1273	IREAD_UNLOCK(ipimap);
1274
1275	if (amp)
1276		release_metapage(amp);
1277	if (bmp)
1278		release_metapage(bmp);
1279	if (cmp)
1280		release_metapage(cmp);
1281	if (dmp)
1282		release_metapage(dmp);
1283
1284	AG_UNLOCK(imap, agno);
1285
1286	release_metapage(mp);
1287
1288	return (rc);
1289}
1290
1291/*
1292 * There are several places in the diAlloc* routines where we initialize
1293 * the inode.
1294 */
1295static inline void
1296diInitInode(struct inode *ip, int iagno, int ino, int extno, struct iag * iagp)
1297{
1298	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
1299
1300	ip->i_ino = (iagno << L2INOSPERIAG) + ino;
1301	jfs_ip->ixpxd = iagp->inoext[extno];
1302	jfs_ip->agstart = le64_to_cpu(iagp->agstart);
1303	jfs_ip->active_ag = -1;
1304}
1305
1306
1307/*
1308 * NAME:	diAlloc(pip,dir,ip)
1309 *
1310 * FUNCTION:	allocate a disk inode from the inode working map
1311 *		for a fileset or aggregate.
1312 *
1313 * PARAMETERS:
1314 *	pip	- pointer to incore inode for the parent inode.
1315 *	dir	- 'true' if the new disk inode is for a directory.
1316 *	ip	- pointer to a new inode
1317 *
1318 * RETURN VALUES:
1319 *	0	- success.
1320 *	-ENOSPC	- insufficient disk resources.
1321 *	-EIO	- i/o error.
1322 */
1323int diAlloc(struct inode *pip, bool dir, struct inode *ip)
1324{
1325	int rc, ino, iagno, addext, extno, bitno, sword;
1326	int nwords, rem, i, agno, dn_numag;
1327	u32 mask, inosmap, extsmap;
1328	struct inode *ipimap;
1329	struct metapage *mp;
1330	ino_t inum;
1331	struct iag *iagp;
1332	struct inomap *imap;
1333
1334	/* get the pointers to the inode map inode and the
1335	 * corresponding imap control structure.
1336	 */
1337	ipimap = JFS_SBI(pip->i_sb)->ipimap;
1338	imap = JFS_IP(ipimap)->i_imap;
1339	JFS_IP(ip)->ipimap = ipimap;
1340	JFS_IP(ip)->fileset = FILESYSTEM_I;
1341
1342	/* for a directory, the allocation policy is to start
1343	 * at the ag level using the preferred ag.
1344	 */
1345	if (dir) {
1346		agno = dbNextAG(JFS_SBI(pip->i_sb)->ipbmap);
1347		AG_LOCK(imap, agno);
1348		goto tryag;
1349	}
1350
1351	/* for files, the policy starts off by trying to allocate from
1352	 * the same iag containing the parent disk inode:
1353	 * try to allocate the new disk inode close to the parent disk
1354	 * inode, using parent disk inode number + 1 as the allocation
1355	 * hint.  (we use a left-to-right policy to attempt to avoid
1356	 * moving backward on the disk.)  compute the hint within the
1357	 * file system and the iag.
1358	 */
1359
1360	/* get the ag number of this iag */
1361	agno = BLKTOAG(JFS_IP(pip)->agstart, JFS_SBI(pip->i_sb));
1362	dn_numag = JFS_SBI(pip->i_sb)->bmap->db_numag;
1363	if (agno < 0 || agno > dn_numag || agno >= MAXAG)
1364		return -EIO;
1365
1366	if (atomic_read(&JFS_SBI(pip->i_sb)->bmap->db_active[agno])) {
1367		/*
1368		 * There is an open file actively growing.  We want to
1369		 * allocate new inodes from a different ag to avoid
1370		 * fragmentation problems.
1371		 */
1372		agno = dbNextAG(JFS_SBI(pip->i_sb)->ipbmap);
1373		AG_LOCK(imap, agno);
1374		goto tryag;
1375	}
1376
1377	inum = pip->i_ino + 1;
1378	ino = inum & (INOSPERIAG - 1);
1379
1380	/* back off the hint if it is outside of the iag */
1381	if (ino == 0)
1382		inum = pip->i_ino;
1383
1384	/* lock the AG inode map information */
1385	AG_LOCK(imap, agno);
1386
1387	/* Get read lock on imap inode */
1388	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
1389
1390	/* get the iag number and read the iag */
1391	iagno = INOTOIAG(inum);
1392	if ((rc = diIAGRead(imap, iagno, &mp))) {
1393		IREAD_UNLOCK(ipimap);
1394		AG_UNLOCK(imap, agno);
1395		return (rc);
1396	}
1397	iagp = (struct iag *) mp->data;
1398
1399	/* determine if new inode extent is allowed to be added to the iag.
1400	 * new inode extent can be added to the iag if the ag
1401	 * has less than 32 free disk inodes and the iag has free extents.
1402	 */
1403	addext = (imap->im_agctl[agno].numfree < 32 && iagp->nfreeexts);
1404
1405	/*
1406	 *	try to allocate from the IAG
1407	 */
1408	/* check if the inode may be allocated from the iag
1409	 * (i.e. the inode has free inodes or new extent can be added).
1410	 */
1411	if (iagp->nfreeinos || addext) {
1412		/* determine the extent number of the hint.
1413		 */
1414		extno = ino >> L2INOSPEREXT;
1415
1416		/* check if the extent containing the hint has backed
1417		 * inodes.  if so, try to allocate within this extent.
1418		 */
1419		if (addressPXD(&iagp->inoext[extno])) {
1420			bitno = ino & (INOSPEREXT - 1);
1421			if ((bitno =
1422			     diFindFree(le32_to_cpu(iagp->wmap[extno]),
1423					bitno))
1424			    < INOSPEREXT) {
1425				ino = (extno << L2INOSPEREXT) + bitno;
1426
1427				/* a free inode (bit) was found within this
1428				 * extent, so allocate it.
1429				 */
1430				rc = diAllocBit(imap, iagp, ino);
1431				IREAD_UNLOCK(ipimap);
1432				if (rc) {
1433					assert(rc == -EIO);
1434				} else {
1435					/* set the results of the allocation
1436					 * and write the iag.
1437					 */
1438					diInitInode(ip, iagno, ino, extno,
1439						    iagp);
1440					mark_metapage_dirty(mp);
1441				}
1442				release_metapage(mp);
1443
1444				/* free the AG lock and return.
1445				 */
1446				AG_UNLOCK(imap, agno);
1447				return (rc);
1448			}
1449
1450			if (!addext)
1451				extno =
1452				    (extno ==
1453				     EXTSPERIAG - 1) ? 0 : extno + 1;
1454		}
1455
1456		/*
1457		 * no free inodes within the extent containing the hint.
1458		 *
1459		 * try to allocate from the backed extents following
1460		 * hint or, if appropriate (i.e. addext is true), allocate
1461		 * an extent of free inodes at or following the extent
1462		 * containing the hint.
1463		 *
1464		 * the free inode and free extent summary maps are used
1465		 * here, so determine the starting summary map position
1466		 * and the number of words we'll have to examine.  again,
1467		 * the approach is to allocate following the hint, so we
1468		 * might have to initially ignore prior bits of the summary
1469		 * map that represent extents prior to the extent containing
1470		 * the hint and later revisit these bits.
1471		 */
1472		bitno = extno & (EXTSPERSUM - 1);
1473		nwords = (bitno == 0) ? SMAPSZ : SMAPSZ + 1;
1474		sword = extno >> L2EXTSPERSUM;
1475
1476		/* mask any prior bits for the starting words of the
1477		 * summary map.
1478		 */
1479		mask = (bitno == 0) ? 0 : (ONES << (EXTSPERSUM - bitno));
1480		inosmap = le32_to_cpu(iagp->inosmap[sword]) | mask;
1481		extsmap = le32_to_cpu(iagp->extsmap[sword]) | mask;
1482
1483		/* scan the free inode and free extent summary maps for
1484		 * free resources.
1485		 */
1486		for (i = 0; i < nwords; i++) {
1487			/* check if this word of the free inode summary
1488			 * map describes an extent with free inodes.
1489			 */
1490			if (~inosmap) {
1491				/* an extent with free inodes has been
1492				 * found. determine the extent number
1493				 * and the inode number within the extent.
1494				 */
1495				rem = diFindFree(inosmap, 0);
1496				extno = (sword << L2EXTSPERSUM) + rem;
1497				rem = diFindFree(le32_to_cpu(iagp->wmap[extno]),
1498						 0);
1499				if (rem >= INOSPEREXT) {
1500					IREAD_UNLOCK(ipimap);
1501					release_metapage(mp);
1502					AG_UNLOCK(imap, agno);
1503					jfs_error(ip->i_sb,
1504						  "can't find free bit in wmap\n");
1505					return -EIO;
1506				}
1507
1508				/* determine the inode number within the
1509				 * iag and allocate the inode from the
1510				 * map.
1511				 */
1512				ino = (extno << L2INOSPEREXT) + rem;
1513				rc = diAllocBit(imap, iagp, ino);
1514				IREAD_UNLOCK(ipimap);
1515				if (rc)
1516					assert(rc == -EIO);
1517				else {
1518					/* set the results of the allocation
1519					 * and write the iag.
1520					 */
1521					diInitInode(ip, iagno, ino, extno,
1522						    iagp);
1523					mark_metapage_dirty(mp);
1524				}
1525				release_metapage(mp);
1526
1527				/* free the AG lock and return.
1528				 */
1529				AG_UNLOCK(imap, agno);
1530				return (rc);
1531
1532			}
1533
1534			/* check if we may allocate an extent of free
1535			 * inodes and whether this word of the free
1536			 * extents summary map describes a free extent.
1537			 */
1538			if (addext && ~extsmap) {
1539				/* a free extent has been found.  determine
1540				 * the extent number.
1541				 */
1542				rem = diFindFree(extsmap, 0);
1543				extno = (sword << L2EXTSPERSUM) + rem;
1544
1545				/* allocate an extent of free inodes.
1546				 */
1547				if ((rc = diNewExt(imap, iagp, extno))) {
1548					/* if there is no disk space for a
1549					 * new extent, try to allocate the
1550					 * disk inode from somewhere else.
1551					 */
1552					if (rc == -ENOSPC)
1553						break;
1554
1555					assert(rc == -EIO);
1556				} else {
1557					/* set the results of the allocation
1558					 * and write the iag.
1559					 */
1560					diInitInode(ip, iagno,
1561						    extno << L2INOSPEREXT,
1562						    extno, iagp);
1563					mark_metapage_dirty(mp);
1564				}
1565				release_metapage(mp);
1566				/* free the imap inode & the AG lock & return.
1567				 */
1568				IREAD_UNLOCK(ipimap);
1569				AG_UNLOCK(imap, agno);
1570				return (rc);
1571			}
1572
1573			/* move on to the next set of summary map words.
1574			 */
1575			sword = (sword == SMAPSZ - 1) ? 0 : sword + 1;
1576			inosmap = le32_to_cpu(iagp->inosmap[sword]);
1577			extsmap = le32_to_cpu(iagp->extsmap[sword]);
1578		}
1579	}
1580	/* unlock imap inode */
1581	IREAD_UNLOCK(ipimap);
1582
1583	/* nothing doing in this iag, so release it. */
1584	release_metapage(mp);
1585
1586      tryag:
1587	/*
1588	 * try to allocate anywhere within the same AG as the parent inode.
1589	 */
1590	rc = diAllocAG(imap, agno, dir, ip);
1591
1592	AG_UNLOCK(imap, agno);
1593
1594	if (rc != -ENOSPC)
1595		return (rc);
1596
1597	/*
1598	 * try to allocate in any AG.
1599	 */
1600	return (diAllocAny(imap, agno, dir, ip));
1601}
1602
1603
1604/*
1605 * NAME:	diAllocAG(imap,agno,dir,ip)
1606 *
1607 * FUNCTION:	allocate a disk inode from the allocation group.
1608 *
1609 *		this routine first determines if a new extent of free
1610 *		inodes should be added for the allocation group, with
1611 *		the current request satisfied from this extent. if this
1612 *		is the case, an attempt will be made to do just that.  if
1613 *		this attempt fails or it has been determined that a new
1614 *		extent should not be added, an attempt is made to satisfy
1615 *		the request by allocating an existing (backed) free inode
1616 *		from the allocation group.
1617 *
1618 * PRE CONDITION: Already have the AG lock for this AG.
1619 *
1620 * PARAMETERS:
1621 *	imap	- pointer to inode map control structure.
1622 *	agno	- allocation group to allocate from.
1623 *	dir	- 'true' if the new disk inode is for a directory.
1624 *	ip	- pointer to the new inode to be filled in on successful return
1625 *		  with the disk inode number allocated, its extent address
1626 *		  and the start of the ag.
1627 *
1628 * RETURN VALUES:
1629 *	0	- success.
1630 *	-ENOSPC	- insufficient disk resources.
1631 *	-EIO	- i/o error.
1632 */
1633static int
1634diAllocAG(struct inomap * imap, int agno, bool dir, struct inode *ip)
1635{
1636	int rc, addext, numfree, numinos;
1637
1638	/* get the number of free and the number of backed disk
1639	 * inodes currently within the ag.
1640	 */
1641	numfree = imap->im_agctl[agno].numfree;
1642	numinos = imap->im_agctl[agno].numinos;
1643
1644	if (numfree > numinos) {
1645		jfs_error(ip->i_sb, "numfree > numinos\n");
1646		return -EIO;
1647	}
1648
1649	/* determine if we should allocate a new extent of free inodes
1650	 * within the ag: for directory inodes, add a new extent
1651	 * if there are a small number of free inodes or number of free
1652	 * inodes is a small percentage of the number of backed inodes.
1653	 */
1654	if (dir)
1655		addext = (numfree < 64 ||
1656			  (numfree < 256
1657			   && ((numfree * 100) / numinos) <= 20));
1658	else
1659		addext = (numfree == 0);
1660
1661	/*
1662	 * try to allocate a new extent of free inodes.
1663	 */
1664	if (addext) {
1665		/* if free space is not available for this new extent, try
1666		 * below to allocate a free and existing (already backed)
1667		 * inode from the ag.
1668		 */
1669		if ((rc = diAllocExt(imap, agno, ip)) != -ENOSPC)
1670			return (rc);
1671	}
1672
1673	/*
1674	 * try to allocate an existing free inode from the ag.
1675	 */
1676	return (diAllocIno(imap, agno, ip));
1677}
1678
1679
1680/*
1681 * NAME:	diAllocAny(imap,agno,dir,iap)
1682 *
1683 * FUNCTION:	allocate a disk inode from any other allocation group.
1684 *
1685 *		this routine is called when an allocation attempt within
1686 *		the primary allocation group has failed. if attempts to
1687 *		allocate an inode from any allocation group other than the
1688 *		specified primary group.
1689 *
1690 * PARAMETERS:
1691 *	imap	- pointer to inode map control structure.
1692 *	agno	- primary allocation group (to avoid).
1693 *	dir	- 'true' if the new disk inode is for a directory.
1694 *	ip	- pointer to a new inode to be filled in on successful return
1695 *		  with the disk inode number allocated, its extent address
1696 *		  and the start of the ag.
1697 *
1698 * RETURN VALUES:
1699 *	0	- success.
1700 *	-ENOSPC	- insufficient disk resources.
1701 *	-EIO	- i/o error.
1702 */
1703static int
1704diAllocAny(struct inomap * imap, int agno, bool dir, struct inode *ip)
1705{
1706	int ag, rc;
1707	int maxag = JFS_SBI(imap->im_ipimap->i_sb)->bmap->db_maxag;
1708
1709
1710	/* try to allocate from the ags following agno up to
1711	 * the maximum ag number.
1712	 */
1713	for (ag = agno + 1; ag <= maxag; ag++) {
1714		AG_LOCK(imap, ag);
1715
1716		rc = diAllocAG(imap, ag, dir, ip);
1717
1718		AG_UNLOCK(imap, ag);
1719
1720		if (rc != -ENOSPC)
1721			return (rc);
1722	}
1723
1724	/* try to allocate from the ags in front of agno.
1725	 */
1726	for (ag = 0; ag < agno; ag++) {
1727		AG_LOCK(imap, ag);
1728
1729		rc = diAllocAG(imap, ag, dir, ip);
1730
1731		AG_UNLOCK(imap, ag);
1732
1733		if (rc != -ENOSPC)
1734			return (rc);
1735	}
1736
1737	/* no free disk inodes.
1738	 */
1739	return -ENOSPC;
1740}
1741
1742
1743/*
1744 * NAME:	diAllocIno(imap,agno,ip)
1745 *
1746 * FUNCTION:	allocate a disk inode from the allocation group's free
1747 *		inode list, returning an error if this free list is
1748 *		empty (i.e. no iags on the list).
1749 *
1750 *		allocation occurs from the first iag on the list using
1751 *		the iag's free inode summary map to find the leftmost
1752 *		free inode in the iag.
1753 *
1754 * PRE CONDITION: Already have AG lock for this AG.
1755 *
1756 * PARAMETERS:
1757 *	imap	- pointer to inode map control structure.
1758 *	agno	- allocation group.
1759 *	ip	- pointer to new inode to be filled in on successful return
1760 *		  with the disk inode number allocated, its extent address
1761 *		  and the start of the ag.
1762 *
1763 * RETURN VALUES:
1764 *	0	- success.
1765 *	-ENOSPC	- insufficient disk resources.
1766 *	-EIO	- i/o error.
1767 */
1768static int diAllocIno(struct inomap * imap, int agno, struct inode *ip)
1769{
1770	int iagno, ino, rc, rem, extno, sword;
1771	struct metapage *mp;
1772	struct iag *iagp;
1773
1774	/* check if there are iags on the ag's free inode list.
1775	 */
1776	if ((iagno = imap->im_agctl[agno].inofree) < 0)
1777		return -ENOSPC;
1778
1779	/* obtain read lock on imap inode */
1780	IREAD_LOCK(imap->im_ipimap, RDWRLOCK_IMAP);
1781
1782	/* read the iag at the head of the list.
1783	 */
1784	if ((rc = diIAGRead(imap, iagno, &mp))) {
1785		IREAD_UNLOCK(imap->im_ipimap);
1786		return (rc);
1787	}
1788	iagp = (struct iag *) mp->data;
1789
1790	/* better be free inodes in this iag if it is on the
1791	 * list.
1792	 */
1793	if (!iagp->nfreeinos) {
1794		IREAD_UNLOCK(imap->im_ipimap);
1795		release_metapage(mp);
1796		jfs_error(ip->i_sb, "nfreeinos = 0, but iag on freelist\n");
1797		return -EIO;
1798	}
1799
1800	/* scan the free inode summary map to find an extent
1801	 * with free inodes.
1802	 */
1803	for (sword = 0;; sword++) {
1804		if (sword >= SMAPSZ) {
1805			IREAD_UNLOCK(imap->im_ipimap);
1806			release_metapage(mp);
1807			jfs_error(ip->i_sb,
1808				  "free inode not found in summary map\n");
1809			return -EIO;
1810		}
1811
1812		if (~iagp->inosmap[sword])
1813			break;
1814	}
1815
1816	/* found a extent with free inodes. determine
1817	 * the extent number.
1818	 */
1819	rem = diFindFree(le32_to_cpu(iagp->inosmap[sword]), 0);
1820	if (rem >= EXTSPERSUM) {
1821		IREAD_UNLOCK(imap->im_ipimap);
1822		release_metapage(mp);
1823		jfs_error(ip->i_sb, "no free extent found\n");
1824		return -EIO;
1825	}
1826	extno = (sword << L2EXTSPERSUM) + rem;
1827
1828	/* find the first free inode in the extent.
1829	 */
1830	rem = diFindFree(le32_to_cpu(iagp->wmap[extno]), 0);
1831	if (rem >= INOSPEREXT) {
1832		IREAD_UNLOCK(imap->im_ipimap);
1833		release_metapage(mp);
1834		jfs_error(ip->i_sb, "free inode not found\n");
1835		return -EIO;
1836	}
1837
1838	/* compute the inode number within the iag.
1839	 */
1840	ino = (extno << L2INOSPEREXT) + rem;
1841
1842	/* allocate the inode.
1843	 */
1844	rc = diAllocBit(imap, iagp, ino);
1845	IREAD_UNLOCK(imap->im_ipimap);
1846	if (rc) {
1847		release_metapage(mp);
1848		return (rc);
1849	}
1850
1851	/* set the results of the allocation and write the iag.
1852	 */
1853	diInitInode(ip, iagno, ino, extno, iagp);
1854	write_metapage(mp);
1855
1856	return (0);
1857}
1858
1859
1860/*
1861 * NAME:	diAllocExt(imap,agno,ip)
1862 *
1863 * FUNCTION:	add a new extent of free inodes to an iag, allocating
1864 *		an inode from this extent to satisfy the current allocation
1865 *		request.
1866 *
1867 *		this routine first tries to find an existing iag with free
1868 *		extents through the ag free extent list.  if list is not
1869 *		empty, the head of the list will be selected as the home
1870 *		of the new extent of free inodes.  otherwise (the list is
1871 *		empty), a new iag will be allocated for the ag to contain
1872 *		the extent.
1873 *
1874 *		once an iag has been selected, the free extent summary map
1875 *		is used to locate a free extent within the iag and diNewExt()
1876 *		is called to initialize the extent, with initialization
1877 *		including the allocation of the first inode of the extent
1878 *		for the purpose of satisfying this request.
1879 *
1880 * PARAMETERS:
1881 *	imap	- pointer to inode map control structure.
1882 *	agno	- allocation group number.
1883 *	ip	- pointer to new inode to be filled in on successful return
1884 *		  with the disk inode number allocated, its extent address
1885 *		  and the start of the ag.
1886 *
1887 * RETURN VALUES:
1888 *	0	- success.
1889 *	-ENOSPC	- insufficient disk resources.
1890 *	-EIO	- i/o error.
1891 */
1892static int diAllocExt(struct inomap * imap, int agno, struct inode *ip)
1893{
1894	int rem, iagno, sword, extno, rc;
1895	struct metapage *mp;
1896	struct iag *iagp;
1897
1898	/* check if the ag has any iags with free extents.  if not,
1899	 * allocate a new iag for the ag.
1900	 */
1901	if ((iagno = imap->im_agctl[agno].extfree) < 0) {
1902		/* If successful, diNewIAG will obtain the read lock on the
1903		 * imap inode.
1904		 */
1905		if ((rc = diNewIAG(imap, &iagno, agno, &mp))) {
1906			return (rc);
1907		}
1908		iagp = (struct iag *) mp->data;
1909
1910		/* set the ag number if this a brand new iag
1911		 */
1912		iagp->agstart =
1913		    cpu_to_le64(AGTOBLK(agno, imap->im_ipimap));
1914	} else {
1915		/* read the iag.
1916		 */
1917		IREAD_LOCK(imap->im_ipimap, RDWRLOCK_IMAP);
1918		if ((rc = diIAGRead(imap, iagno, &mp))) {
1919			IREAD_UNLOCK(imap->im_ipimap);
1920			jfs_error(ip->i_sb, "error reading iag\n");
1921			return rc;
1922		}
1923		iagp = (struct iag *) mp->data;
1924	}
1925
1926	/* using the free extent summary map, find a free extent.
1927	 */
1928	for (sword = 0;; sword++) {
1929		if (sword >= SMAPSZ) {
1930			release_metapage(mp);
1931			IREAD_UNLOCK(imap->im_ipimap);
1932			jfs_error(ip->i_sb, "free ext summary map not found\n");
1933			return -EIO;
1934		}
1935		if (~iagp->extsmap[sword])
1936			break;
1937	}
1938
1939	/* determine the extent number of the free extent.
1940	 */
1941	rem = diFindFree(le32_to_cpu(iagp->extsmap[sword]), 0);
1942	if (rem >= EXTSPERSUM) {
1943		release_metapage(mp);
1944		IREAD_UNLOCK(imap->im_ipimap);
1945		jfs_error(ip->i_sb, "free extent not found\n");
1946		return -EIO;
1947	}
1948	extno = (sword << L2EXTSPERSUM) + rem;
1949
1950	/* initialize the new extent.
1951	 */
1952	rc = diNewExt(imap, iagp, extno);
1953	IREAD_UNLOCK(imap->im_ipimap);
1954	if (rc) {
1955		/* something bad happened.  if a new iag was allocated,
1956		 * place it back on the inode map's iag free list, and
1957		 * clear the ag number information.
1958		 */
1959		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
1960			IAGFREE_LOCK(imap);
1961			iagp->iagfree = cpu_to_le32(imap->im_freeiag);
1962			imap->im_freeiag = iagno;
1963			IAGFREE_UNLOCK(imap);
1964		}
1965		write_metapage(mp);
1966		return (rc);
1967	}
1968
1969	/* set the results of the allocation and write the iag.
1970	 */
1971	diInitInode(ip, iagno, extno << L2INOSPEREXT, extno, iagp);
1972
1973	write_metapage(mp);
1974
1975	return (0);
1976}
1977
1978
1979/*
1980 * NAME:	diAllocBit(imap,iagp,ino)
1981 *
1982 * FUNCTION:	allocate a backed inode from an iag.
1983 *
1984 *		this routine performs the mechanics of allocating a
1985 *		specified inode from a backed extent.
1986 *
1987 *		if the inode to be allocated represents the last free
1988 *		inode within the iag, the iag will be removed from the
1989 *		ag free inode list.
1990 *
1991 *		a careful update approach is used to provide consistency
1992 *		in the face of updates to multiple buffers.  under this
1993 *		approach, all required buffers are obtained before making
1994 *		any updates and are held all are updates are complete.
1995 *
1996 * PRE CONDITION: Already have buffer lock on iagp.  Already have AG lock on
1997 *	this AG.  Must have read lock on imap inode.
1998 *
1999 * PARAMETERS:
2000 *	imap	- pointer to inode map control structure.
2001 *	iagp	- pointer to iag.
2002 *	ino	- inode number to be allocated within the iag.
2003 *
2004 * RETURN VALUES:
2005 *	0	- success.
2006 *	-ENOSPC	- insufficient disk resources.
2007 *	-EIO	- i/o error.
2008 */
2009static int diAllocBit(struct inomap * imap, struct iag * iagp, int ino)
2010{
2011	int extno, bitno, agno, sword, rc;
2012	struct metapage *amp = NULL, *bmp = NULL;
2013	struct iag *aiagp = NULL, *biagp = NULL;
2014	u32 mask;
2015
2016	/* check if this is the last free inode within the iag.
2017	 * if so, it will have to be removed from the ag free
2018	 * inode list, so get the iags preceding and following
2019	 * it on the list.
2020	 */
2021	if (iagp->nfreeinos == cpu_to_le32(1)) {
2022		if ((int) le32_to_cpu(iagp->inofreefwd) >= 0) {
2023			if ((rc =
2024			     diIAGRead(imap, le32_to_cpu(iagp->inofreefwd),
2025				       &amp)))
2026				return (rc);
2027			aiagp = (struct iag *) amp->data;
2028		}
2029
2030		if ((int) le32_to_cpu(iagp->inofreeback) >= 0) {
2031			if ((rc =
2032			     diIAGRead(imap,
2033				       le32_to_cpu(iagp->inofreeback),
2034				       &bmp))) {
2035				if (amp)
2036					release_metapage(amp);
2037				return (rc);
2038			}
2039			biagp = (struct iag *) bmp->data;
2040		}
2041	}
2042
2043	/* get the ag number, extent number, inode number within
2044	 * the extent.
2045	 */
2046	agno = BLKTOAG(le64_to_cpu(iagp->agstart), JFS_SBI(imap->im_ipimap->i_sb));
2047	extno = ino >> L2INOSPEREXT;
2048	bitno = ino & (INOSPEREXT - 1);
2049
2050	/* compute the mask for setting the map.
2051	 */
2052	mask = HIGHORDER >> bitno;
2053
2054	/* the inode should be free and backed.
2055	 */
2056	if (((le32_to_cpu(iagp->pmap[extno]) & mask) != 0) ||
2057	    ((le32_to_cpu(iagp->wmap[extno]) & mask) != 0) ||
2058	    (addressPXD(&iagp->inoext[extno]) == 0)) {
2059		if (amp)
2060			release_metapage(amp);
2061		if (bmp)
2062			release_metapage(bmp);
2063
2064		jfs_error(imap->im_ipimap->i_sb, "iag inconsistent\n");
2065		return -EIO;
2066	}
2067
2068	/* mark the inode as allocated in the working map.
2069	 */
2070	iagp->wmap[extno] |= cpu_to_le32(mask);
2071
2072	/* check if all inodes within the extent are now
2073	 * allocated.  if so, update the free inode summary
2074	 * map to reflect this.
2075	 */
2076	if (iagp->wmap[extno] == cpu_to_le32(ONES)) {
2077		sword = extno >> L2EXTSPERSUM;
2078		bitno = extno & (EXTSPERSUM - 1);
2079		iagp->inosmap[sword] |= cpu_to_le32(HIGHORDER >> bitno);
2080	}
2081
2082	/* if this was the last free inode in the iag, remove the
2083	 * iag from the ag free inode list.
2084	 */
2085	if (iagp->nfreeinos == cpu_to_le32(1)) {
2086		if (amp) {
2087			aiagp->inofreeback = iagp->inofreeback;
2088			write_metapage(amp);
2089		}
2090
2091		if (bmp) {
2092			biagp->inofreefwd = iagp->inofreefwd;
2093			write_metapage(bmp);
2094		} else {
2095			imap->im_agctl[agno].inofree =
2096			    le32_to_cpu(iagp->inofreefwd);
2097		}
2098		iagp->inofreefwd = iagp->inofreeback = cpu_to_le32(-1);
2099	}
2100
2101	/* update the free inode count at the iag, ag, inode
2102	 * map levels.
2103	 */
2104	le32_add_cpu(&iagp->nfreeinos, -1);
2105	imap->im_agctl[agno].numfree -= 1;
2106	atomic_dec(&imap->im_numfree);
2107
2108	return (0);
2109}
2110
2111
2112/*
2113 * NAME:	diNewExt(imap,iagp,extno)
2114 *
2115 * FUNCTION:	initialize a new extent of inodes for an iag, allocating
2116 *		the first inode of the extent for use for the current
2117 *		allocation request.
2118 *
2119 *		disk resources are allocated for the new extent of inodes
2120 *		and the inodes themselves are initialized to reflect their
2121 *		existence within the extent (i.e. their inode numbers and
2122 *		inode extent addresses are set) and their initial state
2123 *		(mode and link count are set to zero).
2124 *
2125 *		if the iag is new, it is not yet on an ag extent free list
2126 *		but will now be placed on this list.
2127 *
2128 *		if the allocation of the new extent causes the iag to
2129 *		have no free extent, the iag will be removed from the
2130 *		ag extent free list.
2131 *
2132 *		if the iag has no free backed inodes, it will be placed
2133 *		on the ag free inode list, since the addition of the new
2134 *		extent will now cause it to have free inodes.
2135 *
2136 *		a careful update approach is used to provide consistency
2137 *		(i.e. list consistency) in the face of updates to multiple
2138 *		buffers.  under this approach, all required buffers are
2139 *		obtained before making any updates and are held until all
2140 *		updates are complete.
2141 *
2142 * PRE CONDITION: Already have buffer lock on iagp.  Already have AG lock on
2143 *	this AG.  Must have read lock on imap inode.
2144 *
2145 * PARAMETERS:
2146 *	imap	- pointer to inode map control structure.
2147 *	iagp	- pointer to iag.
2148 *	extno	- extent number.
2149 *
2150 * RETURN VALUES:
2151 *	0	- success.
2152 *	-ENOSPC	- insufficient disk resources.
2153 *	-EIO	- i/o error.
2154 */
2155static int diNewExt(struct inomap * imap, struct iag * iagp, int extno)
2156{
2157	int agno, iagno, fwd, back, freei = 0, sword, rc;
2158	struct iag *aiagp = NULL, *biagp = NULL, *ciagp = NULL;
2159	struct metapage *amp, *bmp, *cmp, *dmp;
2160	struct inode *ipimap;
2161	s64 blkno, hint;
2162	int i, j;
2163	u32 mask;
2164	ino_t ino;
2165	struct dinode *dp;
2166	struct jfs_sb_info *sbi;
2167
2168	/* better have free extents.
2169	 */
2170	if (!iagp->nfreeexts) {
2171		jfs_error(imap->im_ipimap->i_sb, "no free extents\n");
2172		return -EIO;
2173	}
2174
2175	/* get the inode map inode.
2176	 */
2177	ipimap = imap->im_ipimap;
2178	sbi = JFS_SBI(ipimap->i_sb);
2179
2180	amp = bmp = cmp = NULL;
2181
2182	/* get the ag and iag numbers for this iag.
2183	 */
2184	agno = BLKTOAG(le64_to_cpu(iagp->agstart), sbi);
2185	if (agno >= MAXAG || agno < 0)
2186		return -EIO;
2187
2188	iagno = le32_to_cpu(iagp->iagnum);
2189
2190	/* check if this is the last free extent within the
2191	 * iag.  if so, the iag must be removed from the ag
2192	 * free extent list, so get the iags preceding and
2193	 * following the iag on this list.
2194	 */
2195	if (iagp->nfreeexts == cpu_to_le32(1)) {
2196		if ((fwd = le32_to_cpu(iagp->extfreefwd)) >= 0) {
2197			if ((rc = diIAGRead(imap, fwd, &amp)))
2198				return (rc);
2199			aiagp = (struct iag *) amp->data;
2200		}
2201
2202		if ((back = le32_to_cpu(iagp->extfreeback)) >= 0) {
2203			if ((rc = diIAGRead(imap, back, &bmp)))
2204				goto error_out;
2205			biagp = (struct iag *) bmp->data;
2206		}
2207	} else {
2208		/* the iag has free extents.  if all extents are free
2209		 * (as is the case for a newly allocated iag), the iag
2210		 * must be added to the ag free extent list, so get
2211		 * the iag at the head of the list in preparation for
2212		 * adding this iag to this list.
2213		 */
2214		fwd = back = -1;
2215		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
2216			if ((fwd = imap->im_agctl[agno].extfree) >= 0) {
2217				if ((rc = diIAGRead(imap, fwd, &amp)))
2218					goto error_out;
2219				aiagp = (struct iag *) amp->data;
2220			}
2221		}
2222	}
2223
2224	/* check if the iag has no free inodes.  if so, the iag
2225	 * will have to be added to the ag free inode list, so get
2226	 * the iag at the head of the list in preparation for
2227	 * adding this iag to this list.  in doing this, we must
2228	 * check if we already have the iag at the head of
2229	 * the list in hand.
2230	 */
2231	if (iagp->nfreeinos == 0) {
2232		freei = imap->im_agctl[agno].inofree;
2233
2234		if (freei >= 0) {
2235			if (freei == fwd) {
2236				ciagp = aiagp;
2237			} else if (freei == back) {
2238				ciagp = biagp;
2239			} else {
2240				if ((rc = diIAGRead(imap, freei, &cmp)))
2241					goto error_out;
2242				ciagp = (struct iag *) cmp->data;
2243			}
2244			if (ciagp == NULL) {
2245				jfs_error(imap->im_ipimap->i_sb,
2246					  "ciagp == NULL\n");
2247				rc = -EIO;
2248				goto error_out;
2249			}
2250		}
2251	}
2252
2253	/* allocate disk space for the inode extent.
2254	 */
2255	if ((extno == 0) || (addressPXD(&iagp->inoext[extno - 1]) == 0))
2256		hint = ((s64) agno << sbi->bmap->db_agl2size) - 1;
2257	else
2258		hint = addressPXD(&iagp->inoext[extno - 1]) +
2259		    lengthPXD(&iagp->inoext[extno - 1]) - 1;
2260
2261	if ((rc = dbAlloc(ipimap, hint, (s64) imap->im_nbperiext, &blkno)))
2262		goto error_out;
2263
2264	/* compute the inode number of the first inode within the
2265	 * extent.
2266	 */
2267	ino = (iagno << L2INOSPERIAG) + (extno << L2INOSPEREXT);
2268
2269	/* initialize the inodes within the newly allocated extent a
2270	 * page at a time.
2271	 */
2272	for (i = 0; i < imap->im_nbperiext; i += sbi->nbperpage) {
2273		/* get a buffer for this page of disk inodes.
2274		 */
2275		dmp = get_metapage(ipimap, blkno + i, PSIZE, 1);
2276		if (dmp == NULL) {
2277			rc = -EIO;
2278			goto error_out;
2279		}
2280		dp = (struct dinode *) dmp->data;
2281
2282		/* initialize the inode number, mode, link count and
2283		 * inode extent address.
2284		 */
2285		for (j = 0; j < INOSPERPAGE; j++, dp++, ino++) {
2286			dp->di_inostamp = cpu_to_le32(sbi->inostamp);
2287			dp->di_number = cpu_to_le32(ino);
2288			dp->di_fileset = cpu_to_le32(FILESYSTEM_I);
2289			dp->di_mode = 0;
2290			dp->di_nlink = 0;
2291			PXDaddress(&(dp->di_ixpxd), blkno);
2292			PXDlength(&(dp->di_ixpxd), imap->im_nbperiext);
2293		}
2294		write_metapage(dmp);
2295	}
2296
2297	/* if this is the last free extent within the iag, remove the
2298	 * iag from the ag free extent list.
2299	 */
2300	if (iagp->nfreeexts == cpu_to_le32(1)) {
2301		if (fwd >= 0)
2302			aiagp->extfreeback = iagp->extfreeback;
2303
2304		if (back >= 0)
2305			biagp->extfreefwd = iagp->extfreefwd;
2306		else
2307			imap->im_agctl[agno].extfree =
2308			    le32_to_cpu(iagp->extfreefwd);
2309
2310		iagp->extfreefwd = iagp->extfreeback = cpu_to_le32(-1);
2311	} else {
2312		/* if the iag has all free extents (newly allocated iag),
2313		 * add the iag to the ag free extent list.
2314		 */
2315		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
2316			if (fwd >= 0)
2317				aiagp->extfreeback = cpu_to_le32(iagno);
2318
2319			iagp->extfreefwd = cpu_to_le32(fwd);
2320			iagp->extfreeback = cpu_to_le32(-1);
2321			imap->im_agctl[agno].extfree = iagno;
2322		}
2323	}
2324
2325	/* if the iag has no free inodes, add the iag to the
2326	 * ag free inode list.
2327	 */
2328	if (iagp->nfreeinos == 0) {
2329		if (freei >= 0)
2330			ciagp->inofreeback = cpu_to_le32(iagno);
2331
2332		iagp->inofreefwd =
2333		    cpu_to_le32(imap->im_agctl[agno].inofree);
2334		iagp->inofreeback = cpu_to_le32(-1);
2335		imap->im_agctl[agno].inofree = iagno;
2336	}
2337
2338	/* initialize the extent descriptor of the extent. */
2339	PXDlength(&iagp->inoext[extno], imap->im_nbperiext);
2340	PXDaddress(&iagp->inoext[extno], blkno);
2341
2342	/* initialize the working and persistent map of the extent.
2343	 * the working map will be initialized such that
2344	 * it indicates the first inode of the extent is allocated.
2345	 */
2346	iagp->wmap[extno] = cpu_to_le32(HIGHORDER);
2347	iagp->pmap[extno] = 0;
2348
2349	/* update the free inode and free extent summary maps
2350	 * for the extent to indicate the extent has free inodes
2351	 * and no longer represents a free extent.
2352	 */
2353	sword = extno >> L2EXTSPERSUM;
2354	mask = HIGHORDER >> (extno & (EXTSPERSUM - 1));
2355	iagp->extsmap[sword] |= cpu_to_le32(mask);
2356	iagp->inosmap[sword] &= cpu_to_le32(~mask);
2357
2358	/* update the free inode and free extent counts for the
2359	 * iag.
2360	 */
2361	le32_add_cpu(&iagp->nfreeinos, (INOSPEREXT - 1));
2362	le32_add_cpu(&iagp->nfreeexts, -1);
2363
2364	/* update the free and backed inode counts for the ag.
2365	 */
2366	imap->im_agctl[agno].numfree += (INOSPEREXT - 1);
2367	imap->im_agctl[agno].numinos += INOSPEREXT;
2368
2369	/* update the free and backed inode counts for the inode map.
2370	 */
2371	atomic_add(INOSPEREXT - 1, &imap->im_numfree);
2372	atomic_add(INOSPEREXT, &imap->im_numinos);
2373
2374	/* write the iags.
2375	 */
2376	if (amp)
2377		write_metapage(amp);
2378	if (bmp)
2379		write_metapage(bmp);
2380	if (cmp)
2381		write_metapage(cmp);
2382
2383	return (0);
2384
2385      error_out:
2386
2387	/* release the iags.
2388	 */
2389	if (amp)
2390		release_metapage(amp);
2391	if (bmp)
2392		release_metapage(bmp);
2393	if (cmp)
2394		release_metapage(cmp);
2395
2396	return (rc);
2397}
2398
2399
2400/*
2401 * NAME:	diNewIAG(imap,iagnop,agno)
2402 *
2403 * FUNCTION:	allocate a new iag for an allocation group.
2404 *
2405 *		first tries to allocate the iag from the inode map
2406 *		iagfree list:
2407 *		if the list has free iags, the head of the list is removed
2408 *		and returned to satisfy the request.
2409 *		if the inode map's iag free list is empty, the inode map
2410 *		is extended to hold a new iag. this new iag is initialized
2411 *		and returned to satisfy the request.
2412 *
2413 * PARAMETERS:
2414 *	imap	- pointer to inode map control structure.
2415 *	iagnop	- pointer to an iag number set with the number of the
2416 *		  newly allocated iag upon successful return.
2417 *	agno	- allocation group number.
2418 *	bpp	- Buffer pointer to be filled in with new IAG's buffer
2419 *
2420 * RETURN VALUES:
2421 *	0	- success.
2422 *	-ENOSPC	- insufficient disk resources.
2423 *	-EIO	- i/o error.
2424 *
2425 * serialization:
2426 *	AG lock held on entry/exit;
2427 *	write lock on the map is held inside;
2428 *	read lock on the map is held on successful completion;
2429 *
2430 * note: new iag transaction:
2431 * . synchronously write iag;
2432 * . write log of xtree and inode of imap;
2433 * . commit;
2434 * . synchronous write of xtree (right to left, bottom to top);
2435 * . at start of logredo(): init in-memory imap with one additional iag page;
2436 * . at end of logredo(): re-read imap inode to determine
2437 *   new imap size;
2438 */
2439static int
2440diNewIAG(struct inomap * imap, int *iagnop, int agno, struct metapage ** mpp)
2441{
2442	int rc;
2443	int iagno, i, xlen;
2444	struct inode *ipimap;
2445	struct super_block *sb;
2446	struct jfs_sb_info *sbi;
2447	struct metapage *mp;
2448	struct iag *iagp;
2449	s64 xaddr = 0;
2450	s64 blkno;
2451	tid_t tid;
2452	struct inode *iplist[1];
2453
2454	/* pick up pointers to the inode map and mount inodes */
2455	ipimap = imap->im_ipimap;
2456	sb = ipimap->i_sb;
2457	sbi = JFS_SBI(sb);
2458
2459	/* acquire the free iag lock */
2460	IAGFREE_LOCK(imap);
2461
2462	/* if there are any iags on the inode map free iag list,
2463	 * allocate the iag from the head of the list.
2464	 */
2465	if (imap->im_freeiag >= 0) {
2466		/* pick up the iag number at the head of the list */
2467		iagno = imap->im_freeiag;
2468
2469		/* determine the logical block number of the iag */
2470		blkno = IAGTOLBLK(iagno, sbi->l2nbperpage);
2471	} else {
2472		/* no free iags. the inode map will have to be extented
2473		 * to include a new iag.
2474		 */
2475
2476		/* acquire inode map lock */
2477		IWRITE_LOCK(ipimap, RDWRLOCK_IMAP);
2478
2479		if (ipimap->i_size >> L2PSIZE != imap->im_nextiag + 1) {
2480			IWRITE_UNLOCK(ipimap);
2481			IAGFREE_UNLOCK(imap);
2482			jfs_error(imap->im_ipimap->i_sb,
2483				  "ipimap->i_size is wrong\n");
2484			return -EIO;
2485		}
2486
2487
2488		/* get the next available iag number */
2489		iagno = imap->im_nextiag;
2490
2491		/* make sure that we have not exceeded the maximum inode
2492		 * number limit.
2493		 */
2494		if (iagno > (MAXIAGS - 1)) {
2495			/* release the inode map lock */
2496			IWRITE_UNLOCK(ipimap);
2497
2498			rc = -ENOSPC;
2499			goto out;
2500		}
2501
2502		/*
2503		 * synchronously append new iag page.
2504		 */
2505		/* determine the logical address of iag page to append */
2506		blkno = IAGTOLBLK(iagno, sbi->l2nbperpage);
2507
2508		/* Allocate extent for new iag page */
2509		xlen = sbi->nbperpage;
2510		if ((rc = dbAlloc(ipimap, 0, (s64) xlen, &xaddr))) {
2511			/* release the inode map lock */
2512			IWRITE_UNLOCK(ipimap);
2513
2514			goto out;
2515		}
2516
2517		/*
2518		 * start transaction of update of the inode map
2519		 * addressing structure pointing to the new iag page;
2520		 */
2521		tid = txBegin(sb, COMMIT_FORCE);
2522		mutex_lock(&JFS_IP(ipimap)->commit_mutex);
2523
2524		/* update the inode map addressing structure to point to it */
2525		if ((rc =
2526		     xtInsert(tid, ipimap, 0, blkno, xlen, &xaddr, 0))) {
2527			txEnd(tid);
2528			mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
2529			/* Free the blocks allocated for the iag since it was
2530			 * not successfully added to the inode map
2531			 */
2532			dbFree(ipimap, xaddr, (s64) xlen);
2533
2534			/* release the inode map lock */
2535			IWRITE_UNLOCK(ipimap);
2536
2537			goto out;
2538		}
2539
2540		/* update the inode map's inode to reflect the extension */
2541		ipimap->i_size += PSIZE;
2542		inode_add_bytes(ipimap, PSIZE);
2543
2544		/* assign a buffer for the page */
2545		mp = get_metapage(ipimap, blkno, PSIZE, 0);
2546		if (!mp) {
2547			/*
2548			 * This is very unlikely since we just created the
2549			 * extent, but let's try to handle it correctly
2550			 */
2551			xtTruncate(tid, ipimap, ipimap->i_size - PSIZE,
2552				   COMMIT_PWMAP);
2553
2554			txAbort(tid, 0);
2555			txEnd(tid);
2556			mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
2557
2558			/* release the inode map lock */
2559			IWRITE_UNLOCK(ipimap);
2560
2561			rc = -EIO;
2562			goto out;
2563		}
2564		iagp = (struct iag *) mp->data;
2565
2566		/* init the iag */
2567		memset(iagp, 0, sizeof(struct iag));
2568		iagp->iagnum = cpu_to_le32(iagno);
2569		iagp->inofreefwd = iagp->inofreeback = cpu_to_le32(-1);
2570		iagp->extfreefwd = iagp->extfreeback = cpu_to_le32(-1);
2571		iagp->iagfree = cpu_to_le32(-1);
2572		iagp->nfreeinos = 0;
2573		iagp->nfreeexts = cpu_to_le32(EXTSPERIAG);
2574
2575		/* initialize the free inode summary map (free extent
2576		 * summary map initialization handled by bzero).
2577		 */
2578		for (i = 0; i < SMAPSZ; i++)
2579			iagp->inosmap[i] = cpu_to_le32(ONES);
2580
2581		/*
2582		 * Write and sync the metapage
2583		 */
2584		flush_metapage(mp);
2585
2586		/*
2587		 * txCommit(COMMIT_FORCE) will synchronously write address
2588		 * index pages and inode after commit in careful update order
2589		 * of address index pages (right to left, bottom up);
2590		 */
2591		iplist[0] = ipimap;
2592		rc = txCommit(tid, 1, &iplist[0], COMMIT_FORCE);
2593
2594		txEnd(tid);
2595		mutex_unlock(&JFS_IP(ipimap)->commit_mutex);
2596
2597		duplicateIXtree(sb, blkno, xlen, &xaddr);
2598
2599		/* update the next available iag number */
2600		imap->im_nextiag += 1;
2601
2602		/* Add the iag to the iag free list so we don't lose the iag
2603		 * if a failure happens now.
2604		 */
2605		imap->im_freeiag = iagno;
2606
2607		/* Until we have logredo working, we want the imap inode &
2608		 * control page to be up to date.
2609		 */
2610		diSync(ipimap);
2611
2612		/* release the inode map lock */
2613		IWRITE_UNLOCK(ipimap);
2614	}
2615
2616	/* obtain read lock on map */
2617	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
2618
2619	/* read the iag */
2620	if ((rc = diIAGRead(imap, iagno, &mp))) {
2621		IREAD_UNLOCK(ipimap);
2622		rc = -EIO;
2623		goto out;
2624	}
2625	iagp = (struct iag *) mp->data;
2626
2627	/* remove the iag from the iag free list */
2628	imap->im_freeiag = le32_to_cpu(iagp->iagfree);
2629	iagp->iagfree = cpu_to_le32(-1);
2630
2631	/* set the return iag number and buffer pointer */
2632	*iagnop = iagno;
2633	*mpp = mp;
2634
2635      out:
2636	/* release the iag free lock */
2637	IAGFREE_UNLOCK(imap);
2638
2639	return (rc);
2640}
2641
2642/*
2643 * NAME:	diIAGRead()
2644 *
2645 * FUNCTION:	get the buffer for the specified iag within a fileset
2646 *		or aggregate inode map.
2647 *
2648 * PARAMETERS:
2649 *	imap	- pointer to inode map control structure.
2650 *	iagno	- iag number.
2651 *	bpp	- point to buffer pointer to be filled in on successful
2652 *		  exit.
2653 *
2654 * SERIALIZATION:
2655 *	must have read lock on imap inode
2656 *	(When called by diExtendFS, the filesystem is quiesced, therefore
2657 *	 the read lock is unnecessary.)
2658 *
2659 * RETURN VALUES:
2660 *	0	- success.
2661 *	-EIO	- i/o error.
2662 */
2663static int diIAGRead(struct inomap * imap, int iagno, struct metapage ** mpp)
2664{
2665	struct inode *ipimap = imap->im_ipimap;
2666	s64 blkno;
2667
2668	/* compute the logical block number of the iag. */
2669	blkno = IAGTOLBLK(iagno, JFS_SBI(ipimap->i_sb)->l2nbperpage);
2670
2671	/* read the iag. */
2672	*mpp = read_metapage(ipimap, blkno, PSIZE, 0);
2673	if (*mpp == NULL) {
2674		return -EIO;
2675	}
2676
2677	return (0);
2678}
2679
2680/*
2681 * NAME:	diFindFree()
2682 *
2683 * FUNCTION:	find the first free bit in a word starting at
2684 *		the specified bit position.
2685 *
2686 * PARAMETERS:
2687 *	word	- word to be examined.
2688 *	start	- starting bit position.
2689 *
2690 * RETURN VALUES:
2691 *	bit position of first free bit in the word or 32 if
2692 *	no free bits were found.
2693 */
2694static int diFindFree(u32 word, int start)
2695{
2696	int bitno;
2697	assert(start < 32);
2698	/* scan the word for the first free bit. */
2699	for (word <<= start, bitno = start; bitno < 32;
2700	     bitno++, word <<= 1) {
2701		if ((word & HIGHORDER) == 0)
2702			break;
2703	}
2704	return (bitno);
2705}
2706
2707/*
2708 * NAME:	diUpdatePMap()
2709 *
2710 * FUNCTION: Update the persistent map in an IAG for the allocation or
2711 *	freeing of the specified inode.
2712 *
2713 * PRE CONDITIONS: Working map has already been updated for allocate.
2714 *
2715 * PARAMETERS:
2716 *	ipimap	- Incore inode map inode
2717 *	inum	- Number of inode to mark in permanent map
2718 *	is_free	- If 'true' indicates inode should be marked freed, otherwise
2719 *		  indicates inode should be marked allocated.
2720 *
2721 * RETURN VALUES:
2722 *		0 for success
2723 */
2724int
2725diUpdatePMap(struct inode *ipimap,
2726	     unsigned long inum, bool is_free, struct tblock * tblk)
2727{
2728	int rc;
2729	struct iag *iagp;
2730	struct metapage *mp;
2731	int iagno, ino, extno, bitno;
2732	struct inomap *imap;
2733	u32 mask;
2734	struct jfs_log *log;
2735	int lsn, difft, diffp;
2736	unsigned long flags;
2737
2738	imap = JFS_IP(ipimap)->i_imap;
2739	/* get the iag number containing the inode */
2740	iagno = INOTOIAG(inum);
2741	/* make sure that the iag is contained within the map */
2742	if (iagno >= imap->im_nextiag) {
2743		jfs_error(ipimap->i_sb, "the iag is outside the map\n");
2744		return -EIO;
2745	}
2746	/* read the iag */
2747	IREAD_LOCK(ipimap, RDWRLOCK_IMAP);
2748	rc = diIAGRead(imap, iagno, &mp);
2749	IREAD_UNLOCK(ipimap);
2750	if (rc)
2751		return (rc);
2752	metapage_wait_for_io(mp);
2753	iagp = (struct iag *) mp->data;
2754	/* get the inode number and extent number of the inode within
2755	 * the iag and the inode number within the extent.
2756	 */
2757	ino = inum & (INOSPERIAG - 1);
2758	extno = ino >> L2INOSPEREXT;
2759	bitno = ino & (INOSPEREXT - 1);
2760	mask = HIGHORDER >> bitno;
2761	/*
2762	 * mark the inode free in persistent map:
2763	 */
2764	if (is_free) {
2765		/* The inode should have been allocated both in working
2766		 * map and in persistent map;
2767		 * the inode will be freed from working map at the release
2768		 * of last reference release;
2769		 */
2770		if (!(le32_to_cpu(iagp->wmap[extno]) & mask)) {
2771			jfs_error(ipimap->i_sb,
2772				  "inode %ld not marked as allocated in wmap!\n",
2773				  inum);
2774		}
2775		if (!(le32_to_cpu(iagp->pmap[extno]) & mask)) {
2776			jfs_error(ipimap->i_sb,
2777				  "inode %ld not marked as allocated in pmap!\n",
2778				  inum);
2779		}
2780		/* update the bitmap for the extent of the freed inode */
2781		iagp->pmap[extno] &= cpu_to_le32(~mask);
2782	}
2783	/*
2784	 * mark the inode allocated in persistent map:
2785	 */
2786	else {
2787		/* The inode should be already allocated in the working map
2788		 * and should be free in persistent map;
2789		 */
2790		if (!(le32_to_cpu(iagp->wmap[extno]) & mask)) {
2791			release_metapage(mp);
2792			jfs_error(ipimap->i_sb,
2793				  "the inode is not allocated in the working map\n");
2794			return -EIO;
2795		}
2796		if ((le32_to_cpu(iagp->pmap[extno]) & mask) != 0) {
2797			release_metapage(mp);
2798			jfs_error(ipimap->i_sb,
2799				  "the inode is not free in the persistent map\n");
2800			return -EIO;
2801		}
2802		/* update the bitmap for the extent of the allocated inode */
2803		iagp->pmap[extno] |= cpu_to_le32(mask);
2804	}
2805	/*
2806	 * update iag lsn
2807	 */
2808	lsn = tblk->lsn;
2809	log = JFS_SBI(tblk->sb)->log;
2810	LOGSYNC_LOCK(log, flags);
2811	if (mp->lsn != 0) {
2812		/* inherit older/smaller lsn */
2813		logdiff(difft, lsn, log);
2814		logdiff(diffp, mp->lsn, log);
2815		if (difft < diffp) {
2816			mp->lsn = lsn;
2817			/* move mp after tblock in logsync list */
2818			list_move(&mp->synclist, &tblk->synclist);
2819		}
2820		/* inherit younger/larger clsn */
2821		assert(mp->clsn);
2822		logdiff(difft, tblk->clsn, log);
2823		logdiff(diffp, mp->clsn, log);
2824		if (difft > diffp)
2825			mp->clsn = tblk->clsn;
2826	} else {
2827		mp->log = log;
2828		mp->lsn = lsn;
2829		/* insert mp after tblock in logsync list */
2830		log->count++;
2831		list_add(&mp->synclist, &tblk->synclist);
2832		mp->clsn = tblk->clsn;
2833	}
2834	LOGSYNC_UNLOCK(log, flags);
2835	write_metapage(mp);
2836	return (0);
2837}
2838
2839/*
2840 *	diExtendFS()
2841 *
2842 * function: update imap for extendfs();
2843 *
2844 * note: AG size has been increased s.t. each k old contiguous AGs are
2845 * coalesced into a new AG;
2846 */
2847int diExtendFS(struct inode *ipimap, struct inode *ipbmap)
2848{
2849	int rc, rcx = 0;
2850	struct inomap *imap = JFS_IP(ipimap)->i_imap;
2851	struct iag *iagp = NULL, *hiagp = NULL;
2852	struct bmap *mp = JFS_SBI(ipbmap->i_sb)->bmap;
2853	struct metapage *bp, *hbp;
2854	int i, n, head;
2855	int numinos, xnuminos = 0, xnumfree = 0;
2856	s64 agstart;
2857
2858	jfs_info("diExtendFS: nextiag:%d numinos:%d numfree:%d",
2859		   imap->im_nextiag, atomic_read(&imap->im_numinos),
2860		   atomic_read(&imap->im_numfree));
2861
2862	/*
2863	 *	reconstruct imap
2864	 *
2865	 * coalesce contiguous k (newAGSize/oldAGSize) AGs;
2866	 * i.e., (AGi, ..., AGj) where i = k*n and j = k*(n+1) - 1 to AGn;
2867	 * note: new AG size = old AG size * (2**x).
2868	 */
2869
2870	/* init per AG control information im_agctl[] */
2871	for (i = 0; i < MAXAG; i++) {
2872		imap->im_agctl[i].inofree = -1;
2873		imap->im_agctl[i].extfree = -1;
2874		imap->im_agctl[i].numinos = 0;	/* number of backed inodes */
2875		imap->im_agctl[i].numfree = 0;	/* number of free backed inodes */
2876	}
2877
2878	/*
2879	 *	process each iag page of the map.
2880	 *
2881	 * rebuild AG Free Inode List, AG Free Inode Extent List;
2882	 */
2883	for (i = 0; i < imap->im_nextiag; i++) {
2884		if ((rc = diIAGRead(imap, i, &bp))) {
2885			rcx = rc;
2886			continue;
2887		}
2888		iagp = (struct iag *) bp->data;
2889		if (le32_to_cpu(iagp->iagnum) != i) {
2890			release_metapage(bp);
2891			jfs_error(ipimap->i_sb, "unexpected value of iagnum\n");
2892			return -EIO;
2893		}
2894
2895		/* leave free iag in the free iag list */
2896		if (iagp->nfreeexts == cpu_to_le32(EXTSPERIAG)) {
2897			release_metapage(bp);
2898			continue;
2899		}
2900
2901		agstart = le64_to_cpu(iagp->agstart);
2902		n = agstart >> mp->db_agl2size;
2903		iagp->agstart = cpu_to_le64((s64)n << mp->db_agl2size);
2904
2905		/* compute backed inodes */
2906		numinos = (EXTSPERIAG - le32_to_cpu(iagp->nfreeexts))
2907		    << L2INOSPEREXT;
2908		if (numinos > 0) {
2909			/* merge AG backed inodes */
2910			imap->im_agctl[n].numinos += numinos;
2911			xnuminos += numinos;
2912		}
2913
2914		/* if any backed free inodes, insert at AG free inode list */
2915		if ((int) le32_to_cpu(iagp->nfreeinos) > 0) {
2916			if ((head = imap->im_agctl[n].inofree) == -1) {
2917				iagp->inofreefwd = cpu_to_le32(-1);
2918				iagp->inofreeback = cpu_to_le32(-1);
2919			} else {
2920				if ((rc = diIAGRead(imap, head, &hbp))) {
2921					rcx = rc;
2922					goto nextiag;
2923				}
2924				hiagp = (struct iag *) hbp->data;
2925				hiagp->inofreeback = iagp->iagnum;
2926				iagp->inofreefwd = cpu_to_le32(head);
2927				iagp->inofreeback = cpu_to_le32(-1);
2928				write_metapage(hbp);
2929			}
2930
2931			imap->im_agctl[n].inofree =
2932			    le32_to_cpu(iagp->iagnum);
2933
2934			/* merge AG backed free inodes */
2935			imap->im_agctl[n].numfree +=
2936			    le32_to_cpu(iagp->nfreeinos);
2937			xnumfree += le32_to_cpu(iagp->nfreeinos);
2938		}
2939
2940		/* if any free extents, insert at AG free extent list */
2941		if (le32_to_cpu(iagp->nfreeexts) > 0) {
2942			if ((head = imap->im_agctl[n].extfree) == -1) {
2943				iagp->extfreefwd = cpu_to_le32(-1);
2944				iagp->extfreeback = cpu_to_le32(-1);
2945			} else {
2946				if ((rc = diIAGRead(imap, head, &hbp))) {
2947					rcx = rc;
2948					goto nextiag;
2949				}
2950				hiagp = (struct iag *) hbp->data;
2951				hiagp->extfreeback = iagp->iagnum;
2952				iagp->extfreefwd = cpu_to_le32(head);
2953				iagp->extfreeback = cpu_to_le32(-1);
2954				write_metapage(hbp);
2955			}
2956
2957			imap->im_agctl[n].extfree =
2958			    le32_to_cpu(iagp->iagnum);
2959		}
2960
2961	      nextiag:
2962		write_metapage(bp);
2963	}
2964
2965	if (xnuminos != atomic_read(&imap->im_numinos) ||
2966	    xnumfree != atomic_read(&imap->im_numfree)) {
2967		jfs_error(ipimap->i_sb, "numinos or numfree incorrect\n");
2968		return -EIO;
2969	}
2970
2971	return rcx;
2972}
2973
2974
2975/*
2976 *	duplicateIXtree()
2977 *
2978 * serialization: IWRITE_LOCK held on entry/exit
2979 *
2980 * note: shadow page with regular inode (rel.2);
2981 */
2982static void duplicateIXtree(struct super_block *sb, s64 blkno,
2983			    int xlen, s64 *xaddr)
2984{
2985	struct jfs_superblock *j_sb;
2986	struct buffer_head *bh;
2987	struct inode *ip;
2988	tid_t tid;
2989
2990	/* if AIT2 ipmap2 is bad, do not try to update it */
2991	if (JFS_SBI(sb)->mntflag & JFS_BAD_SAIT)	/* s_flag */
2992		return;
2993	ip = diReadSpecial(sb, FILESYSTEM_I, 1);
2994	if (ip == NULL) {
2995		JFS_SBI(sb)->mntflag |= JFS_BAD_SAIT;
2996		if (readSuper(sb, &bh))
2997			return;
2998		j_sb = (struct jfs_superblock *)bh->b_data;
2999		j_sb->s_flag |= cpu_to_le32(JFS_BAD_SAIT);
3000
3001		mark_buffer_dirty(bh);
3002		sync_dirty_buffer(bh);
3003		brelse(bh);
3004		return;
3005	}
3006
3007	/* start transaction */
3008	tid = txBegin(sb, COMMIT_FORCE);
3009	/* update the inode map addressing structure to point to it */
3010	if (xtInsert(tid, ip, 0, blkno, xlen, xaddr, 0)) {
3011		JFS_SBI(sb)->mntflag |= JFS_BAD_SAIT;
3012		txAbort(tid, 1);
3013		goto cleanup;
3014
3015	}
3016	/* update the inode map's inode to reflect the extension */
3017	ip->i_size += PSIZE;
3018	inode_add_bytes(ip, PSIZE);
3019	txCommit(tid, 1, &ip, COMMIT_FORCE);
3020      cleanup:
3021	txEnd(tid);
3022	diFreeSpecial(ip);
3023}
3024
3025/*
3026 * NAME:	copy_from_dinode()
3027 *
3028 * FUNCTION:	Copies inode info from disk inode to in-memory inode
3029 *
3030 * RETURN VALUES:
3031 *	0	- success
3032 *	-ENOMEM	- insufficient memory
3033 */
3034static int copy_from_dinode(struct dinode * dip, struct inode *ip)
3035{
3036	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
3037	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
3038
3039	jfs_ip->fileset = le32_to_cpu(dip->di_fileset);
3040	jfs_ip->mode2 = le32_to_cpu(dip->di_mode);
3041	jfs_set_inode_flags(ip);
3042
3043	ip->i_mode = le32_to_cpu(dip->di_mode) & 0xffff;
3044	if (sbi->umask != -1) {
3045		ip->i_mode = (ip->i_mode & ~0777) | (0777 & ~sbi->umask);
3046		/* For directories, add x permission if r is allowed by umask */
3047		if (S_ISDIR(ip->i_mode)) {
3048			if (ip->i_mode & 0400)
3049				ip->i_mode |= 0100;
3050			if (ip->i_mode & 0040)
3051				ip->i_mode |= 0010;
3052			if (ip->i_mode & 0004)
3053				ip->i_mode |= 0001;
3054		}
3055	}
3056	set_nlink(ip, le32_to_cpu(dip->di_nlink));
3057
3058	jfs_ip->saved_uid = make_kuid(&init_user_ns, le32_to_cpu(dip->di_uid));
3059	if (!uid_valid(sbi->uid))
3060		ip->i_uid = jfs_ip->saved_uid;
3061	else {
3062		ip->i_uid = sbi->uid;
3063	}
3064
3065	jfs_ip->saved_gid = make_kgid(&init_user_ns, le32_to_cpu(dip->di_gid));
3066	if (!gid_valid(sbi->gid))
3067		ip->i_gid = jfs_ip->saved_gid;
3068	else {
3069		ip->i_gid = sbi->gid;
3070	}
3071
3072	ip->i_size = le64_to_cpu(dip->di_size);
3073	inode_set_atime(ip, le32_to_cpu(dip->di_atime.tv_sec),
3074			le32_to_cpu(dip->di_atime.tv_nsec));
3075	inode_set_mtime(ip, le32_to_cpu(dip->di_mtime.tv_sec),
3076			le32_to_cpu(dip->di_mtime.tv_nsec));
3077	inode_set_ctime(ip, le32_to_cpu(dip->di_ctime.tv_sec),
3078			le32_to_cpu(dip->di_ctime.tv_nsec));
3079	ip->i_blocks = LBLK2PBLK(ip->i_sb, le64_to_cpu(dip->di_nblocks));
3080	ip->i_generation = le32_to_cpu(dip->di_gen);
3081
3082	jfs_ip->ixpxd = dip->di_ixpxd;	/* in-memory pxd's are little-endian */
3083	jfs_ip->acl = dip->di_acl;	/* as are dxd's */
3084	jfs_ip->ea = dip->di_ea;
3085	jfs_ip->next_index = le32_to_cpu(dip->di_next_index);
3086	jfs_ip->otime = le32_to_cpu(dip->di_otime.tv_sec);
3087	jfs_ip->acltype = le32_to_cpu(dip->di_acltype);
3088
3089	if (S_ISCHR(ip->i_mode) || S_ISBLK(ip->i_mode)) {
3090		jfs_ip->dev = le32_to_cpu(dip->di_rdev);
3091		ip->i_rdev = new_decode_dev(jfs_ip->dev);
3092	}
3093
3094	if (S_ISDIR(ip->i_mode)) {
3095		memcpy(&jfs_ip->u.dir, &dip->u._dir, 384);
3096	} else if (S_ISREG(ip->i_mode) || S_ISLNK(ip->i_mode)) {
3097		memcpy(&jfs_ip->i_xtroot, &dip->di_xtroot, 288);
3098	} else
3099		memcpy(&jfs_ip->i_inline_ea, &dip->di_inlineea, 128);
3100
3101	/* Zero the in-memory-only stuff */
3102	jfs_ip->cflag = 0;
3103	jfs_ip->btindex = 0;
3104	jfs_ip->btorder = 0;
3105	jfs_ip->bxflag = 0;
3106	jfs_ip->blid = 0;
3107	jfs_ip->atlhead = 0;
3108	jfs_ip->atltail = 0;
3109	jfs_ip->xtlid = 0;
3110	return (0);
3111}
3112
3113/*
3114 * NAME:	copy_to_dinode()
3115 *
3116 * FUNCTION:	Copies inode info from in-memory inode to disk inode
3117 */
3118static void copy_to_dinode(struct dinode * dip, struct inode *ip)
3119{
3120	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
3121	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
3122
3123	dip->di_fileset = cpu_to_le32(jfs_ip->fileset);
3124	dip->di_inostamp = cpu_to_le32(sbi->inostamp);
3125	dip->di_number = cpu_to_le32(ip->i_ino);
3126	dip->di_gen = cpu_to_le32(ip->i_generation);
3127	dip->di_size = cpu_to_le64(ip->i_size);
3128	dip->di_nblocks = cpu_to_le64(PBLK2LBLK(ip->i_sb, ip->i_blocks));
3129	dip->di_nlink = cpu_to_le32(ip->i_nlink);
3130	if (!uid_valid(sbi->uid))
3131		dip->di_uid = cpu_to_le32(i_uid_read(ip));
3132	else
3133		dip->di_uid =cpu_to_le32(from_kuid(&init_user_ns,
3134						   jfs_ip->saved_uid));
3135	if (!gid_valid(sbi->gid))
3136		dip->di_gid = cpu_to_le32(i_gid_read(ip));
3137	else
3138		dip->di_gid = cpu_to_le32(from_kgid(&init_user_ns,
3139						    jfs_ip->saved_gid));
3140	/*
3141	 * mode2 is only needed for storing the higher order bits.
3142	 * Trust i_mode for the lower order ones
3143	 */
3144	if (sbi->umask == -1)
3145		dip->di_mode = cpu_to_le32((jfs_ip->mode2 & 0xffff0000) |
3146					   ip->i_mode);
3147	else /* Leave the original permissions alone */
3148		dip->di_mode = cpu_to_le32(jfs_ip->mode2);
3149
3150	dip->di_atime.tv_sec = cpu_to_le32(inode_get_atime_sec(ip));
3151	dip->di_atime.tv_nsec = cpu_to_le32(inode_get_atime_nsec(ip));
3152	dip->di_ctime.tv_sec = cpu_to_le32(inode_get_ctime_sec(ip));
3153	dip->di_ctime.tv_nsec = cpu_to_le32(inode_get_ctime_nsec(ip));
3154	dip->di_mtime.tv_sec = cpu_to_le32(inode_get_mtime_sec(ip));
3155	dip->di_mtime.tv_nsec = cpu_to_le32(inode_get_mtime_nsec(ip));
3156	dip->di_ixpxd = jfs_ip->ixpxd;	/* in-memory pxd's are little-endian */
3157	dip->di_acl = jfs_ip->acl;	/* as are dxd's */
3158	dip->di_ea = jfs_ip->ea;
3159	dip->di_next_index = cpu_to_le32(jfs_ip->next_index);
3160	dip->di_otime.tv_sec = cpu_to_le32(jfs_ip->otime);
3161	dip->di_otime.tv_nsec = 0;
3162	dip->di_acltype = cpu_to_le32(jfs_ip->acltype);
3163	if (S_ISCHR(ip->i_mode) || S_ISBLK(ip->i_mode))
3164		dip->di_rdev = cpu_to_le32(jfs_ip->dev);
3165}