xref: /linux/fs/jfs/jfs_dtree.c (revision a1ff5a7d78a036d6c2178ee5acd6ba4946243800)
1 // SPDX-License-Identifier: GPL-2.0-or-later
2 /*
3  *   Copyright (C) International Business Machines Corp., 2000-2004
4  */
5 
6 /*
7  *	jfs_dtree.c: directory B+-tree manager
8  *
9  * B+-tree with variable length key directory:
10  *
11  * each directory page is structured as an array of 32-byte
12  * directory entry slots initialized as a freelist
13  * to avoid search/compaction of free space at insertion.
14  * when an entry is inserted, a number of slots are allocated
15  * from the freelist as required to store variable length data
16  * of the entry; when the entry is deleted, slots of the entry
17  * are returned to freelist.
18  *
19  * leaf entry stores full name as key and file serial number
20  * (aka inode number) as data.
21  * internal/router entry stores sufffix compressed name
22  * as key and simple extent descriptor as data.
23  *
24  * each directory page maintains a sorted entry index table
25  * which stores the start slot index of sorted entries
26  * to allow binary search on the table.
27  *
28  * directory starts as a root/leaf page in on-disk inode
29  * inline data area.
30  * when it becomes full, it starts a leaf of a external extent
31  * of length of 1 block. each time the first leaf becomes full,
32  * it is extended rather than split (its size is doubled),
33  * until its length becoms 4 KBytes, from then the extent is split
34  * with new 4 Kbyte extent when it becomes full
35  * to reduce external fragmentation of small directories.
36  *
37  * blah, blah, blah, for linear scan of directory in pieces by
38  * readdir().
39  *
40  *
41  *	case-insensitive directory file system
42  *
43  * names are stored in case-sensitive way in leaf entry.
44  * but stored, searched and compared in case-insensitive (uppercase) order
45  * (i.e., both search key and entry key are folded for search/compare):
46  * (note that case-sensitive order is BROKEN in storage, e.g.,
47  *  sensitive: Ad, aB, aC, aD -> insensitive: aB, aC, aD, Ad
48  *
49  *  entries which folds to the same key makes up a equivalent class
50  *  whose members are stored as contiguous cluster (may cross page boundary)
51  *  but whose order is arbitrary and acts as duplicate, e.g.,
52  *  abc, Abc, aBc, abC)
53  *
54  * once match is found at leaf, requires scan forward/backward
55  * either for, in case-insensitive search, duplicate
56  * or for, in case-sensitive search, for exact match
57  *
58  * router entry must be created/stored in case-insensitive way
59  * in internal entry:
60  * (right most key of left page and left most key of right page
61  * are folded, and its suffix compression is propagated as router
62  * key in parent)
63  * (e.g., if split occurs <abc> and <aBd>, <ABD> trather than <aB>
64  * should be made the router key for the split)
65  *
66  * case-insensitive search:
67  *
68  *	fold search key;
69  *
70  *	case-insensitive search of B-tree:
71  *	for internal entry, router key is already folded;
72  *	for leaf entry, fold the entry key before comparison.
73  *
74  *	if (leaf entry case-insensitive match found)
75  *		if (next entry satisfies case-insensitive match)
76  *			return EDUPLICATE;
77  *		if (prev entry satisfies case-insensitive match)
78  *			return EDUPLICATE;
79  *		return match;
80  *	else
81  *		return no match;
82  *
83  *	serialization:
84  * target directory inode lock is being held on entry/exit
85  * of all main directory service routines.
86  *
87  *	log based recovery:
88  */
89 
90 #include <linux/fs.h>
91 #include <linux/quotaops.h>
92 #include <linux/slab.h>
93 #include "jfs_incore.h"
94 #include "jfs_superblock.h"
95 #include "jfs_filsys.h"
96 #include "jfs_metapage.h"
97 #include "jfs_dmap.h"
98 #include "jfs_unicode.h"
99 #include "jfs_debug.h"
100 
101 /* dtree split parameter */
102 struct dtsplit {
103 	struct metapage *mp;
104 	s16 index;
105 	s16 nslot;
106 	struct component_name *key;
107 	ddata_t *data;
108 	struct pxdlist *pxdlist;
109 };
110 
111 #define DT_PAGE(IP, MP) BT_PAGE(IP, MP, dtpage_t, i_dtroot)
112 
113 /* get page buffer for specified block address */
114 #define DT_GETPAGE(IP, BN, MP, SIZE, P, RC)				\
115 do {									\
116 	BT_GETPAGE(IP, BN, MP, dtpage_t, SIZE, P, RC, i_dtroot);	\
117 	if (!(RC)) {							\
118 		if (((P)->header.nextindex >				\
119 		     (((BN) == 0) ? DTROOTMAXSLOT : (P)->header.maxslot)) || \
120 		    ((BN) && ((P)->header.maxslot > DTPAGEMAXSLOT))) {	\
121 			BT_PUTPAGE(MP);					\
122 			jfs_error((IP)->i_sb,				\
123 				  "DT_GETPAGE: dtree page corrupt\n");	\
124 			MP = NULL;					\
125 			RC = -EIO;					\
126 		}							\
127 	}								\
128 } while (0)
129 
130 /* for consistency */
131 #define DT_PUTPAGE(MP) BT_PUTPAGE(MP)
132 
133 #define DT_GETSEARCH(IP, LEAF, BN, MP, P, INDEX) \
134 	BT_GETSEARCH(IP, LEAF, BN, MP, dtpage_t, P, INDEX, i_dtroot)
135 
136 /*
137  * forward references
138  */
139 static int dtSplitUp(tid_t tid, struct inode *ip,
140 		     struct dtsplit * split, struct btstack * btstack);
141 
142 static int dtSplitPage(tid_t tid, struct inode *ip, struct dtsplit * split,
143 		       struct metapage ** rmpp, dtpage_t ** rpp, pxd_t * rxdp);
144 
145 static int dtExtendPage(tid_t tid, struct inode *ip,
146 			struct dtsplit * split, struct btstack * btstack);
147 
148 static int dtSplitRoot(tid_t tid, struct inode *ip,
149 		       struct dtsplit * split, struct metapage ** rmpp);
150 
151 static int dtDeleteUp(tid_t tid, struct inode *ip, struct metapage * fmp,
152 		      dtpage_t * fp, struct btstack * btstack);
153 
154 static int dtRelink(tid_t tid, struct inode *ip, dtpage_t * p);
155 
156 static int dtReadFirst(struct inode *ip, struct btstack * btstack);
157 
158 static int dtReadNext(struct inode *ip,
159 		      loff_t * offset, struct btstack * btstack);
160 
161 static int dtCompare(struct component_name * key, dtpage_t * p, int si);
162 
163 static int ciCompare(struct component_name * key, dtpage_t * p, int si,
164 		     int flag);
165 
166 static void dtGetKey(dtpage_t * p, int i, struct component_name * key,
167 		     int flag);
168 
169 static int ciGetLeafPrefixKey(dtpage_t * lp, int li, dtpage_t * rp,
170 			      int ri, struct component_name * key, int flag);
171 
172 static void dtInsertEntry(dtpage_t * p, int index, struct component_name * key,
173 			  ddata_t * data, struct dt_lock **);
174 
175 static void dtMoveEntry(dtpage_t * sp, int si, dtpage_t * dp,
176 			struct dt_lock ** sdtlock, struct dt_lock ** ddtlock,
177 			int do_index);
178 
179 static void dtDeleteEntry(dtpage_t * p, int fi, struct dt_lock ** dtlock);
180 
181 static void dtTruncateEntry(dtpage_t * p, int ti, struct dt_lock ** dtlock);
182 
183 static void dtLinelockFreelist(dtpage_t * p, int m, struct dt_lock ** dtlock);
184 
185 #define ciToUpper(c)	UniStrupr((c)->name)
186 
187 /*
188  *	read_index_page()
189  *
190  *	Reads a page of a directory's index table.
191  *	Having metadata mapped into the directory inode's address space
192  *	presents a multitude of problems.  We avoid this by mapping to
193  *	the absolute address space outside of the *_metapage routines
194  */
read_index_page(struct inode * inode,s64 blkno)195 static struct metapage *read_index_page(struct inode *inode, s64 blkno)
196 {
197 	int rc;
198 	s64 xaddr;
199 	int xflag;
200 	s32 xlen;
201 
202 	rc = xtLookup(inode, blkno, 1, &xflag, &xaddr, &xlen, 1);
203 	if (rc || (xaddr == 0))
204 		return NULL;
205 
206 	return read_metapage(inode, xaddr, PSIZE, 1);
207 }
208 
209 /*
210  *	get_index_page()
211  *
212  *	Same as get_index_page(), but get's a new page without reading
213  */
get_index_page(struct inode * inode,s64 blkno)214 static struct metapage *get_index_page(struct inode *inode, s64 blkno)
215 {
216 	int rc;
217 	s64 xaddr;
218 	int xflag;
219 	s32 xlen;
220 
221 	rc = xtLookup(inode, blkno, 1, &xflag, &xaddr, &xlen, 1);
222 	if (rc || (xaddr == 0))
223 		return NULL;
224 
225 	return get_metapage(inode, xaddr, PSIZE, 1);
226 }
227 
228 /*
229  *	find_index()
230  *
231  *	Returns dtree page containing directory table entry for specified
232  *	index and pointer to its entry.
233  *
234  *	mp must be released by caller.
235  */
find_index(struct inode * ip,u32 index,struct metapage ** mp,s64 * lblock)236 static struct dir_table_slot *find_index(struct inode *ip, u32 index,
237 					 struct metapage ** mp, s64 *lblock)
238 {
239 	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
240 	s64 blkno;
241 	s64 offset;
242 	int page_offset;
243 	struct dir_table_slot *slot;
244 	static int maxWarnings = 10;
245 
246 	if (index < 2) {
247 		if (maxWarnings) {
248 			jfs_warn("find_entry called with index = %d", index);
249 			maxWarnings--;
250 		}
251 		return NULL;
252 	}
253 
254 	if (index >= jfs_ip->next_index) {
255 		jfs_warn("find_entry called with index >= next_index");
256 		return NULL;
257 	}
258 
259 	if (jfs_dirtable_inline(ip)) {
260 		/*
261 		 * Inline directory table
262 		 */
263 		*mp = NULL;
264 		slot = &jfs_ip->i_dirtable[index - 2];
265 	} else {
266 		offset = (index - 2) * sizeof(struct dir_table_slot);
267 		page_offset = offset & (PSIZE - 1);
268 		blkno = ((offset + 1) >> L2PSIZE) <<
269 		    JFS_SBI(ip->i_sb)->l2nbperpage;
270 
271 		if (*mp && (*lblock != blkno)) {
272 			release_metapage(*mp);
273 			*mp = NULL;
274 		}
275 		if (!(*mp)) {
276 			*lblock = blkno;
277 			*mp = read_index_page(ip, blkno);
278 		}
279 		if (!(*mp)) {
280 			jfs_err("free_index: error reading directory table");
281 			return NULL;
282 		}
283 
284 		slot =
285 		    (struct dir_table_slot *) ((char *) (*mp)->data +
286 					       page_offset);
287 	}
288 	return slot;
289 }
290 
lock_index(tid_t tid,struct inode * ip,struct metapage * mp,u32 index)291 static inline void lock_index(tid_t tid, struct inode *ip, struct metapage * mp,
292 			      u32 index)
293 {
294 	struct tlock *tlck;
295 	struct linelock *llck;
296 	struct lv *lv;
297 
298 	tlck = txLock(tid, ip, mp, tlckDATA);
299 	llck = (struct linelock *) tlck->lock;
300 
301 	if (llck->index >= llck->maxcnt)
302 		llck = txLinelock(llck);
303 	lv = &llck->lv[llck->index];
304 
305 	/*
306 	 *	Linelock slot size is twice the size of directory table
307 	 *	slot size.  512 entries per page.
308 	 */
309 	lv->offset = ((index - 2) & 511) >> 1;
310 	lv->length = 1;
311 	llck->index++;
312 }
313 
314 /*
315  *	add_index()
316  *
317  *	Adds an entry to the directory index table.  This is used to provide
318  *	each directory entry with a persistent index in which to resume
319  *	directory traversals
320  */
add_index(tid_t tid,struct inode * ip,s64 bn,int slot)321 static u32 add_index(tid_t tid, struct inode *ip, s64 bn, int slot)
322 {
323 	struct super_block *sb = ip->i_sb;
324 	struct jfs_sb_info *sbi = JFS_SBI(sb);
325 	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
326 	u64 blkno;
327 	struct dir_table_slot *dirtab_slot;
328 	u32 index;
329 	struct linelock *llck;
330 	struct lv *lv;
331 	struct metapage *mp;
332 	s64 offset;
333 	uint page_offset;
334 	struct tlock *tlck;
335 	s64 xaddr;
336 
337 	ASSERT(DO_INDEX(ip));
338 
339 	if (jfs_ip->next_index < 2) {
340 		jfs_warn("add_index: next_index = %d.  Resetting!",
341 			   jfs_ip->next_index);
342 		jfs_ip->next_index = 2;
343 	}
344 
345 	index = jfs_ip->next_index++;
346 
347 	if (index <= MAX_INLINE_DIRTABLE_ENTRY) {
348 		/*
349 		 * i_size reflects size of index table, or 8 bytes per entry.
350 		 */
351 		ip->i_size = (loff_t) (index - 1) << 3;
352 
353 		/*
354 		 * dir table fits inline within inode
355 		 */
356 		dirtab_slot = &jfs_ip->i_dirtable[index-2];
357 		dirtab_slot->flag = DIR_INDEX_VALID;
358 		dirtab_slot->slot = slot;
359 		DTSaddress(dirtab_slot, bn);
360 
361 		set_cflag(COMMIT_Dirtable, ip);
362 
363 		return index;
364 	}
365 	if (index == (MAX_INLINE_DIRTABLE_ENTRY + 1)) {
366 		struct dir_table_slot temp_table[12];
367 
368 		/*
369 		 * It's time to move the inline table to an external
370 		 * page and begin to build the xtree
371 		 */
372 		if (dquot_alloc_block(ip, sbi->nbperpage))
373 			goto clean_up;
374 		if (dbAlloc(ip, 0, sbi->nbperpage, &xaddr)) {
375 			dquot_free_block(ip, sbi->nbperpage);
376 			goto clean_up;
377 		}
378 
379 		/*
380 		 * Save the table, we're going to overwrite it with the
381 		 * xtree root
382 		 */
383 		memcpy(temp_table, &jfs_ip->i_dirtable, sizeof(temp_table));
384 
385 		/*
386 		 * Initialize empty x-tree
387 		 */
388 		xtInitRoot(tid, ip);
389 
390 		/*
391 		 * Add the first block to the xtree
392 		 */
393 		if (xtInsert(tid, ip, 0, 0, sbi->nbperpage, &xaddr, 0)) {
394 			/* This really shouldn't fail */
395 			jfs_warn("add_index: xtInsert failed!");
396 			memcpy(&jfs_ip->i_dirtable, temp_table,
397 			       sizeof (temp_table));
398 			dbFree(ip, xaddr, sbi->nbperpage);
399 			dquot_free_block(ip, sbi->nbperpage);
400 			goto clean_up;
401 		}
402 		ip->i_size = PSIZE;
403 
404 		mp = get_index_page(ip, 0);
405 		if (!mp) {
406 			jfs_err("add_index: get_metapage failed!");
407 			xtTruncate(tid, ip, 0, COMMIT_PWMAP);
408 			memcpy(&jfs_ip->i_dirtable, temp_table,
409 			       sizeof (temp_table));
410 			goto clean_up;
411 		}
412 		tlck = txLock(tid, ip, mp, tlckDATA);
413 		llck = (struct linelock *) & tlck->lock;
414 		ASSERT(llck->index == 0);
415 		lv = &llck->lv[0];
416 
417 		lv->offset = 0;
418 		lv->length = 6;	/* tlckDATA slot size is 16 bytes */
419 		llck->index++;
420 
421 		memcpy(mp->data, temp_table, sizeof(temp_table));
422 
423 		mark_metapage_dirty(mp);
424 		release_metapage(mp);
425 
426 		/*
427 		 * Logging is now directed by xtree tlocks
428 		 */
429 		clear_cflag(COMMIT_Dirtable, ip);
430 	}
431 
432 	offset = (index - 2) * sizeof(struct dir_table_slot);
433 	page_offset = offset & (PSIZE - 1);
434 	blkno = ((offset + 1) >> L2PSIZE) << sbi->l2nbperpage;
435 	if (page_offset == 0) {
436 		/*
437 		 * This will be the beginning of a new page
438 		 */
439 		xaddr = 0;
440 		if (xtInsert(tid, ip, 0, blkno, sbi->nbperpage, &xaddr, 0)) {
441 			jfs_warn("add_index: xtInsert failed!");
442 			goto clean_up;
443 		}
444 		ip->i_size += PSIZE;
445 
446 		if ((mp = get_index_page(ip, blkno)))
447 			memset(mp->data, 0, PSIZE);	/* Just looks better */
448 		else
449 			xtTruncate(tid, ip, offset, COMMIT_PWMAP);
450 	} else
451 		mp = read_index_page(ip, blkno);
452 
453 	if (!mp) {
454 		jfs_err("add_index: get/read_metapage failed!");
455 		goto clean_up;
456 	}
457 
458 	lock_index(tid, ip, mp, index);
459 
460 	dirtab_slot =
461 	    (struct dir_table_slot *) ((char *) mp->data + page_offset);
462 	dirtab_slot->flag = DIR_INDEX_VALID;
463 	dirtab_slot->slot = slot;
464 	DTSaddress(dirtab_slot, bn);
465 
466 	mark_metapage_dirty(mp);
467 	release_metapage(mp);
468 
469 	return index;
470 
471       clean_up:
472 
473 	jfs_ip->next_index--;
474 
475 	return 0;
476 }
477 
478 /*
479  *	free_index()
480  *
481  *	Marks an entry to the directory index table as free.
482  */
free_index(tid_t tid,struct inode * ip,u32 index,u32 next)483 static void free_index(tid_t tid, struct inode *ip, u32 index, u32 next)
484 {
485 	struct dir_table_slot *dirtab_slot;
486 	s64 lblock;
487 	struct metapage *mp = NULL;
488 
489 	dirtab_slot = find_index(ip, index, &mp, &lblock);
490 
491 	if (!dirtab_slot)
492 		return;
493 
494 	dirtab_slot->flag = DIR_INDEX_FREE;
495 	dirtab_slot->slot = dirtab_slot->addr1 = 0;
496 	dirtab_slot->addr2 = cpu_to_le32(next);
497 
498 	if (mp) {
499 		lock_index(tid, ip, mp, index);
500 		mark_metapage_dirty(mp);
501 		release_metapage(mp);
502 	} else
503 		set_cflag(COMMIT_Dirtable, ip);
504 }
505 
506 /*
507  *	modify_index()
508  *
509  *	Changes an entry in the directory index table
510  */
modify_index(tid_t tid,struct inode * ip,u32 index,s64 bn,int slot,struct metapage ** mp,s64 * lblock)511 static void modify_index(tid_t tid, struct inode *ip, u32 index, s64 bn,
512 			 int slot, struct metapage ** mp, s64 *lblock)
513 {
514 	struct dir_table_slot *dirtab_slot;
515 
516 	dirtab_slot = find_index(ip, index, mp, lblock);
517 
518 	if (!dirtab_slot)
519 		return;
520 
521 	DTSaddress(dirtab_slot, bn);
522 	dirtab_slot->slot = slot;
523 
524 	if (*mp) {
525 		lock_index(tid, ip, *mp, index);
526 		mark_metapage_dirty(*mp);
527 	} else
528 		set_cflag(COMMIT_Dirtable, ip);
529 }
530 
531 /*
532  *	read_index()
533  *
534  *	reads a directory table slot
535  */
read_index(struct inode * ip,u32 index,struct dir_table_slot * dirtab_slot)536 static int read_index(struct inode *ip, u32 index,
537 		     struct dir_table_slot * dirtab_slot)
538 {
539 	s64 lblock;
540 	struct metapage *mp = NULL;
541 	struct dir_table_slot *slot;
542 
543 	slot = find_index(ip, index, &mp, &lblock);
544 	if (!slot) {
545 		return -EIO;
546 	}
547 
548 	memcpy(dirtab_slot, slot, sizeof(struct dir_table_slot));
549 
550 	if (mp)
551 		release_metapage(mp);
552 
553 	return 0;
554 }
555 
556 /*
557  *	dtSearch()
558  *
559  * function:
560  *	Search for the entry with specified key
561  *
562  * parameter:
563  *
564  * return: 0 - search result on stack, leaf page pinned;
565  *	   errno - I/O error
566  */
dtSearch(struct inode * ip,struct component_name * key,ino_t * data,struct btstack * btstack,int flag)567 int dtSearch(struct inode *ip, struct component_name * key, ino_t * data,
568 	     struct btstack * btstack, int flag)
569 {
570 	int rc = 0;
571 	int cmp = 1;		/* init for empty page */
572 	s64 bn;
573 	struct metapage *mp;
574 	dtpage_t *p;
575 	s8 *stbl;
576 	int base, index, lim;
577 	struct btframe *btsp;
578 	pxd_t *pxd;
579 	int psize = 288;	/* initial in-line directory */
580 	ino_t inumber;
581 	struct component_name ciKey;
582 	struct super_block *sb = ip->i_sb;
583 
584 	ciKey.name = kmalloc_array(JFS_NAME_MAX + 1, sizeof(wchar_t),
585 				   GFP_NOFS);
586 	if (!ciKey.name) {
587 		rc = -ENOMEM;
588 		goto dtSearch_Exit2;
589 	}
590 
591 
592 	/* uppercase search key for c-i directory */
593 	UniStrcpy(ciKey.name, key->name);
594 	ciKey.namlen = key->namlen;
595 
596 	/* only uppercase if case-insensitive support is on */
597 	if ((JFS_SBI(sb)->mntflag & JFS_OS2) == JFS_OS2) {
598 		ciToUpper(&ciKey);
599 	}
600 	BT_CLR(btstack);	/* reset stack */
601 
602 	/* init level count for max pages to split */
603 	btstack->nsplit = 1;
604 
605 	/*
606 	 *	search down tree from root:
607 	 *
608 	 * between two consecutive entries of <Ki, Pi> and <Kj, Pj> of
609 	 * internal page, child page Pi contains entry with k, Ki <= K < Kj.
610 	 *
611 	 * if entry with search key K is not found
612 	 * internal page search find the entry with largest key Ki
613 	 * less than K which point to the child page to search;
614 	 * leaf page search find the entry with smallest key Kj
615 	 * greater than K so that the returned index is the position of
616 	 * the entry to be shifted right for insertion of new entry.
617 	 * for empty tree, search key is greater than any key of the tree.
618 	 *
619 	 * by convention, root bn = 0.
620 	 */
621 	for (bn = 0;;) {
622 		/* get/pin the page to search */
623 		DT_GETPAGE(ip, bn, mp, psize, p, rc);
624 		if (rc)
625 			goto dtSearch_Exit1;
626 
627 		/* get sorted entry table of the page */
628 		stbl = DT_GETSTBL(p);
629 
630 		/*
631 		 * binary search with search key K on the current page.
632 		 */
633 		for (base = 0, lim = p->header.nextindex; lim; lim >>= 1) {
634 			index = base + (lim >> 1);
635 
636 			if (stbl[index] < 0) {
637 				rc = -EIO;
638 				goto out;
639 			}
640 
641 			if (p->header.flag & BT_LEAF) {
642 				/* uppercase leaf name to compare */
643 				cmp =
644 				    ciCompare(&ciKey, p, stbl[index],
645 					      JFS_SBI(sb)->mntflag);
646 			} else {
647 				/* router key is in uppercase */
648 
649 				cmp = dtCompare(&ciKey, p, stbl[index]);
650 
651 
652 			}
653 			if (cmp == 0) {
654 				/*
655 				 *	search hit
656 				 */
657 				/* search hit - leaf page:
658 				 * return the entry found
659 				 */
660 				if (p->header.flag & BT_LEAF) {
661 					inumber = le32_to_cpu(
662 			((struct ldtentry *) & p->slot[stbl[index]])->inumber);
663 
664 					/*
665 					 * search for JFS_LOOKUP
666 					 */
667 					if (flag == JFS_LOOKUP) {
668 						*data = inumber;
669 						rc = 0;
670 						goto out;
671 					}
672 
673 					/*
674 					 * search for JFS_CREATE
675 					 */
676 					if (flag == JFS_CREATE) {
677 						*data = inumber;
678 						rc = -EEXIST;
679 						goto out;
680 					}
681 
682 					/*
683 					 * search for JFS_REMOVE or JFS_RENAME
684 					 */
685 					if ((flag == JFS_REMOVE ||
686 					     flag == JFS_RENAME) &&
687 					    *data != inumber) {
688 						rc = -ESTALE;
689 						goto out;
690 					}
691 
692 					/*
693 					 * JFS_REMOVE|JFS_FINDDIR|JFS_RENAME
694 					 */
695 					/* save search result */
696 					*data = inumber;
697 					btsp = btstack->top;
698 					btsp->bn = bn;
699 					btsp->index = index;
700 					btsp->mp = mp;
701 
702 					rc = 0;
703 					goto dtSearch_Exit1;
704 				}
705 
706 				/* search hit - internal page:
707 				 * descend/search its child page
708 				 */
709 				goto getChild;
710 			}
711 
712 			if (cmp > 0) {
713 				base = index + 1;
714 				--lim;
715 			}
716 		}
717 
718 		/*
719 		 *	search miss
720 		 *
721 		 * base is the smallest index with key (Kj) greater than
722 		 * search key (K) and may be zero or (maxindex + 1) index.
723 		 */
724 		/*
725 		 * search miss - leaf page
726 		 *
727 		 * return location of entry (base) where new entry with
728 		 * search key K is to be inserted.
729 		 */
730 		if (p->header.flag & BT_LEAF) {
731 			/*
732 			 * search for JFS_LOOKUP, JFS_REMOVE, or JFS_RENAME
733 			 */
734 			if (flag == JFS_LOOKUP || flag == JFS_REMOVE ||
735 			    flag == JFS_RENAME) {
736 				rc = -ENOENT;
737 				goto out;
738 			}
739 
740 			/*
741 			 * search for JFS_CREATE|JFS_FINDDIR:
742 			 *
743 			 * save search result
744 			 */
745 			*data = 0;
746 			btsp = btstack->top;
747 			btsp->bn = bn;
748 			btsp->index = base;
749 			btsp->mp = mp;
750 
751 			rc = 0;
752 			goto dtSearch_Exit1;
753 		}
754 
755 		/*
756 		 * search miss - internal page
757 		 *
758 		 * if base is non-zero, decrement base by one to get the parent
759 		 * entry of the child page to search.
760 		 */
761 		index = base ? base - 1 : base;
762 
763 		/*
764 		 * go down to child page
765 		 */
766 	      getChild:
767 		/* update max. number of pages to split */
768 		if (BT_STACK_FULL(btstack)) {
769 			/* Something's corrupted, mark filesystem dirty so
770 			 * chkdsk will fix it.
771 			 */
772 			jfs_error(sb, "stack overrun!\n");
773 			BT_STACK_DUMP(btstack);
774 			rc = -EIO;
775 			goto out;
776 		}
777 		btstack->nsplit++;
778 
779 		/* push (bn, index) of the parent page/entry */
780 		BT_PUSH(btstack, bn, index);
781 
782 		/* get the child page block number */
783 		pxd = (pxd_t *) & p->slot[stbl[index]];
784 		bn = addressPXD(pxd);
785 		psize = lengthPXD(pxd) << JFS_SBI(ip->i_sb)->l2bsize;
786 
787 		/* unpin the parent page */
788 		DT_PUTPAGE(mp);
789 	}
790 
791       out:
792 	DT_PUTPAGE(mp);
793 
794       dtSearch_Exit1:
795 
796 	kfree(ciKey.name);
797 
798       dtSearch_Exit2:
799 
800 	return rc;
801 }
802 
803 
804 /*
805  *	dtInsert()
806  *
807  * function: insert an entry to directory tree
808  *
809  * parameter:
810  *
811  * return: 0 - success;
812  *	   errno - failure;
813  */
dtInsert(tid_t tid,struct inode * ip,struct component_name * name,ino_t * fsn,struct btstack * btstack)814 int dtInsert(tid_t tid, struct inode *ip,
815 	 struct component_name * name, ino_t * fsn, struct btstack * btstack)
816 {
817 	int rc = 0;
818 	struct metapage *mp;	/* meta-page buffer */
819 	dtpage_t *p;		/* base B+-tree index page */
820 	s64 bn;
821 	int index;
822 	struct dtsplit split;	/* split information */
823 	ddata_t data;
824 	struct dt_lock *dtlck;
825 	int n;
826 	struct tlock *tlck;
827 	struct lv *lv;
828 
829 	/*
830 	 *	retrieve search result
831 	 *
832 	 * dtSearch() returns (leaf page pinned, index at which to insert).
833 	 * n.b. dtSearch() may return index of (maxindex + 1) of
834 	 * the full page.
835 	 */
836 	DT_GETSEARCH(ip, btstack->top, bn, mp, p, index);
837 	if (p->header.freelist == 0)
838 		return -EINVAL;
839 
840 	/*
841 	 *	insert entry for new key
842 	 */
843 	if (DO_INDEX(ip)) {
844 		if (JFS_IP(ip)->next_index == DIREND) {
845 			DT_PUTPAGE(mp);
846 			return -EMLINK;
847 		}
848 		n = NDTLEAF(name->namlen);
849 		data.leaf.tid = tid;
850 		data.leaf.ip = ip;
851 	} else {
852 		n = NDTLEAF_LEGACY(name->namlen);
853 		data.leaf.ip = NULL;	/* signifies legacy directory format */
854 	}
855 	data.leaf.ino = *fsn;
856 
857 	/*
858 	 *	leaf page does not have enough room for new entry:
859 	 *
860 	 *	extend/split the leaf page;
861 	 *
862 	 * dtSplitUp() will insert the entry and unpin the leaf page.
863 	 */
864 	if (n > p->header.freecnt) {
865 		split.mp = mp;
866 		split.index = index;
867 		split.nslot = n;
868 		split.key = name;
869 		split.data = &data;
870 		rc = dtSplitUp(tid, ip, &split, btstack);
871 		return rc;
872 	}
873 
874 	/*
875 	 *	leaf page does have enough room for new entry:
876 	 *
877 	 *	insert the new data entry into the leaf page;
878 	 */
879 	BT_MARK_DIRTY(mp, ip);
880 	/*
881 	 * acquire a transaction lock on the leaf page
882 	 */
883 	tlck = txLock(tid, ip, mp, tlckDTREE | tlckENTRY);
884 	dtlck = (struct dt_lock *) & tlck->lock;
885 	ASSERT(dtlck->index == 0);
886 	lv = & dtlck->lv[0];
887 
888 	/* linelock header */
889 	lv->offset = 0;
890 	lv->length = 1;
891 	dtlck->index++;
892 
893 	dtInsertEntry(p, index, name, &data, &dtlck);
894 
895 	/* linelock stbl of non-root leaf page */
896 	if (!(p->header.flag & BT_ROOT)) {
897 		if (dtlck->index >= dtlck->maxcnt)
898 			dtlck = (struct dt_lock *) txLinelock(dtlck);
899 		lv = & dtlck->lv[dtlck->index];
900 		n = index >> L2DTSLOTSIZE;
901 		lv->offset = p->header.stblindex + n;
902 		lv->length =
903 		    ((p->header.nextindex - 1) >> L2DTSLOTSIZE) - n + 1;
904 		dtlck->index++;
905 	}
906 
907 	/* unpin the leaf page */
908 	DT_PUTPAGE(mp);
909 
910 	return 0;
911 }
912 
913 
914 /*
915  *	dtSplitUp()
916  *
917  * function: propagate insertion bottom up;
918  *
919  * parameter:
920  *
921  * return: 0 - success;
922  *	   errno - failure;
923  *	leaf page unpinned;
924  */
dtSplitUp(tid_t tid,struct inode * ip,struct dtsplit * split,struct btstack * btstack)925 static int dtSplitUp(tid_t tid,
926 	  struct inode *ip, struct dtsplit * split, struct btstack * btstack)
927 {
928 	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
929 	int rc = 0;
930 	struct metapage *smp;
931 	dtpage_t *sp;		/* split page */
932 	struct metapage *rmp;
933 	dtpage_t *rp;		/* new right page split from sp */
934 	pxd_t rpxd;		/* new right page extent descriptor */
935 	struct metapage *lmp;
936 	dtpage_t *lp;		/* left child page */
937 	int skip;		/* index of entry of insertion */
938 	struct btframe *parent;	/* parent page entry on traverse stack */
939 	s64 xaddr, nxaddr;
940 	int xlen, xsize;
941 	struct pxdlist pxdlist;
942 	pxd_t *pxd;
943 	struct component_name key = { 0, NULL };
944 	ddata_t *data = split->data;
945 	int n;
946 	struct dt_lock *dtlck;
947 	struct tlock *tlck;
948 	struct lv *lv;
949 	int quota_allocation = 0;
950 
951 	/* get split page */
952 	smp = split->mp;
953 	sp = DT_PAGE(ip, smp);
954 
955 	key.name = kmalloc_array(JFS_NAME_MAX + 2, sizeof(wchar_t), GFP_NOFS);
956 	if (!key.name) {
957 		DT_PUTPAGE(smp);
958 		rc = -ENOMEM;
959 		goto dtSplitUp_Exit;
960 	}
961 
962 	/*
963 	 *	split leaf page
964 	 *
965 	 * The split routines insert the new entry, and
966 	 * acquire txLock as appropriate.
967 	 */
968 	/*
969 	 *	split root leaf page:
970 	 */
971 	if (sp->header.flag & BT_ROOT) {
972 		/*
973 		 * allocate a single extent child page
974 		 */
975 		xlen = 1;
976 		n = sbi->bsize >> L2DTSLOTSIZE;
977 		n -= (n + 31) >> L2DTSLOTSIZE;	/* stbl size */
978 		n -= DTROOTMAXSLOT - sp->header.freecnt; /* header + entries */
979 		if (n <= split->nslot)
980 			xlen++;
981 		if ((rc = dbAlloc(ip, 0, (s64) xlen, &xaddr))) {
982 			DT_PUTPAGE(smp);
983 			goto freeKeyName;
984 		}
985 
986 		pxdlist.maxnpxd = 1;
987 		pxdlist.npxd = 0;
988 		pxd = &pxdlist.pxd[0];
989 		PXDaddress(pxd, xaddr);
990 		PXDlength(pxd, xlen);
991 		split->pxdlist = &pxdlist;
992 		rc = dtSplitRoot(tid, ip, split, &rmp);
993 
994 		if (rc)
995 			dbFree(ip, xaddr, xlen);
996 		else
997 			DT_PUTPAGE(rmp);
998 
999 		DT_PUTPAGE(smp);
1000 
1001 		if (!DO_INDEX(ip))
1002 			ip->i_size = xlen << sbi->l2bsize;
1003 
1004 		goto freeKeyName;
1005 	}
1006 
1007 	/*
1008 	 *	extend first leaf page
1009 	 *
1010 	 * extend the 1st extent if less than buffer page size
1011 	 * (dtExtendPage() reurns leaf page unpinned)
1012 	 */
1013 	pxd = &sp->header.self;
1014 	xlen = lengthPXD(pxd);
1015 	xsize = xlen << sbi->l2bsize;
1016 	if (xsize < PSIZE) {
1017 		xaddr = addressPXD(pxd);
1018 		n = xsize >> L2DTSLOTSIZE;
1019 		n -= (n + 31) >> L2DTSLOTSIZE;	/* stbl size */
1020 		if ((n + sp->header.freecnt) <= split->nslot)
1021 			n = xlen + (xlen << 1);
1022 		else
1023 			n = xlen;
1024 
1025 		/* Allocate blocks to quota. */
1026 		rc = dquot_alloc_block(ip, n);
1027 		if (rc)
1028 			goto extendOut;
1029 		quota_allocation += n;
1030 
1031 		if ((rc = dbReAlloc(sbi->ipbmap, xaddr, (s64) xlen,
1032 				    (s64) n, &nxaddr)))
1033 			goto extendOut;
1034 
1035 		pxdlist.maxnpxd = 1;
1036 		pxdlist.npxd = 0;
1037 		pxd = &pxdlist.pxd[0];
1038 		PXDaddress(pxd, nxaddr);
1039 		PXDlength(pxd, xlen + n);
1040 		split->pxdlist = &pxdlist;
1041 		if ((rc = dtExtendPage(tid, ip, split, btstack))) {
1042 			nxaddr = addressPXD(pxd);
1043 			if (xaddr != nxaddr) {
1044 				/* free relocated extent */
1045 				xlen = lengthPXD(pxd);
1046 				dbFree(ip, nxaddr, (s64) xlen);
1047 			} else {
1048 				/* free extended delta */
1049 				xlen = lengthPXD(pxd) - n;
1050 				xaddr = addressPXD(pxd) + xlen;
1051 				dbFree(ip, xaddr, (s64) n);
1052 			}
1053 		} else if (!DO_INDEX(ip))
1054 			ip->i_size = lengthPXD(pxd) << sbi->l2bsize;
1055 
1056 
1057 	      extendOut:
1058 		DT_PUTPAGE(smp);
1059 		goto freeKeyName;
1060 	}
1061 
1062 	/*
1063 	 *	split leaf page <sp> into <sp> and a new right page <rp>.
1064 	 *
1065 	 * return <rp> pinned and its extent descriptor <rpxd>
1066 	 */
1067 	/*
1068 	 * allocate new directory page extent and
1069 	 * new index page(s) to cover page split(s)
1070 	 *
1071 	 * allocation hint: ?
1072 	 */
1073 	n = btstack->nsplit;
1074 	pxdlist.maxnpxd = pxdlist.npxd = 0;
1075 	xlen = sbi->nbperpage;
1076 	for (pxd = pxdlist.pxd; n > 0; n--, pxd++) {
1077 		if ((rc = dbAlloc(ip, 0, (s64) xlen, &xaddr)) == 0) {
1078 			PXDaddress(pxd, xaddr);
1079 			PXDlength(pxd, xlen);
1080 			pxdlist.maxnpxd++;
1081 			continue;
1082 		}
1083 
1084 		DT_PUTPAGE(smp);
1085 
1086 		/* undo allocation */
1087 		goto splitOut;
1088 	}
1089 
1090 	split->pxdlist = &pxdlist;
1091 	if ((rc = dtSplitPage(tid, ip, split, &rmp, &rp, &rpxd))) {
1092 		DT_PUTPAGE(smp);
1093 
1094 		/* undo allocation */
1095 		goto splitOut;
1096 	}
1097 
1098 	if (!DO_INDEX(ip))
1099 		ip->i_size += PSIZE;
1100 
1101 	/*
1102 	 * propagate up the router entry for the leaf page just split
1103 	 *
1104 	 * insert a router entry for the new page into the parent page,
1105 	 * propagate the insert/split up the tree by walking back the stack
1106 	 * of (bn of parent page, index of child page entry in parent page)
1107 	 * that were traversed during the search for the page that split.
1108 	 *
1109 	 * the propagation of insert/split up the tree stops if the root
1110 	 * splits or the page inserted into doesn't have to split to hold
1111 	 * the new entry.
1112 	 *
1113 	 * the parent entry for the split page remains the same, and
1114 	 * a new entry is inserted at its right with the first key and
1115 	 * block number of the new right page.
1116 	 *
1117 	 * There are a maximum of 4 pages pinned at any time:
1118 	 * two children, left parent and right parent (when the parent splits).
1119 	 * keep the child pages pinned while working on the parent.
1120 	 * make sure that all pins are released at exit.
1121 	 */
1122 	while ((parent = BT_POP(btstack)) != NULL) {
1123 		/* parent page specified by stack frame <parent> */
1124 
1125 		/* keep current child pages (<lp>, <rp>) pinned */
1126 		lmp = smp;
1127 		lp = sp;
1128 
1129 		/*
1130 		 * insert router entry in parent for new right child page <rp>
1131 		 */
1132 		/* get the parent page <sp> */
1133 		DT_GETPAGE(ip, parent->bn, smp, PSIZE, sp, rc);
1134 		if (rc) {
1135 			DT_PUTPAGE(lmp);
1136 			DT_PUTPAGE(rmp);
1137 			goto splitOut;
1138 		}
1139 
1140 		/*
1141 		 * The new key entry goes ONE AFTER the index of parent entry,
1142 		 * because the split was to the right.
1143 		 */
1144 		skip = parent->index + 1;
1145 
1146 		/*
1147 		 * compute the key for the router entry
1148 		 *
1149 		 * key suffix compression:
1150 		 * for internal pages that have leaf pages as children,
1151 		 * retain only what's needed to distinguish between
1152 		 * the new entry and the entry on the page to its left.
1153 		 * If the keys compare equal, retain the entire key.
1154 		 *
1155 		 * note that compression is performed only at computing
1156 		 * router key at the lowest internal level.
1157 		 * further compression of the key between pairs of higher
1158 		 * level internal pages loses too much information and
1159 		 * the search may fail.
1160 		 * (e.g., two adjacent leaf pages of {a, ..., x} {xx, ...,}
1161 		 * results in two adjacent parent entries (a)(xx).
1162 		 * if split occurs between these two entries, and
1163 		 * if compression is applied, the router key of parent entry
1164 		 * of right page (x) will divert search for x into right
1165 		 * subtree and miss x in the left subtree.)
1166 		 *
1167 		 * the entire key must be retained for the next-to-leftmost
1168 		 * internal key at any level of the tree, or search may fail
1169 		 * (e.g., ?)
1170 		 */
1171 		switch (rp->header.flag & BT_TYPE) {
1172 		case BT_LEAF:
1173 			/*
1174 			 * compute the length of prefix for suffix compression
1175 			 * between last entry of left page and first entry
1176 			 * of right page
1177 			 */
1178 			if ((sp->header.flag & BT_ROOT && skip > 1) ||
1179 			    sp->header.prev != 0 || skip > 1) {
1180 				/* compute uppercase router prefix key */
1181 				rc = ciGetLeafPrefixKey(lp,
1182 							lp->header.nextindex-1,
1183 							rp, 0, &key,
1184 							sbi->mntflag);
1185 				if (rc) {
1186 					DT_PUTPAGE(lmp);
1187 					DT_PUTPAGE(rmp);
1188 					DT_PUTPAGE(smp);
1189 					goto splitOut;
1190 				}
1191 			} else {
1192 				/* next to leftmost entry of
1193 				   lowest internal level */
1194 
1195 				/* compute uppercase router key */
1196 				dtGetKey(rp, 0, &key, sbi->mntflag);
1197 				key.name[key.namlen] = 0;
1198 
1199 				if ((sbi->mntflag & JFS_OS2) == JFS_OS2)
1200 					ciToUpper(&key);
1201 			}
1202 
1203 			n = NDTINTERNAL(key.namlen);
1204 			break;
1205 
1206 		case BT_INTERNAL:
1207 			dtGetKey(rp, 0, &key, sbi->mntflag);
1208 			n = NDTINTERNAL(key.namlen);
1209 			break;
1210 
1211 		default:
1212 			jfs_err("dtSplitUp(): UFO!");
1213 			break;
1214 		}
1215 
1216 		/* unpin left child page */
1217 		DT_PUTPAGE(lmp);
1218 
1219 		/*
1220 		 * compute the data for the router entry
1221 		 */
1222 		data->xd = rpxd;	/* child page xd */
1223 
1224 		/*
1225 		 * parent page is full - split the parent page
1226 		 */
1227 		if (n > sp->header.freecnt) {
1228 			/* init for parent page split */
1229 			split->mp = smp;
1230 			split->index = skip;	/* index at insert */
1231 			split->nslot = n;
1232 			split->key = &key;
1233 			/* split->data = data; */
1234 
1235 			/* unpin right child page */
1236 			DT_PUTPAGE(rmp);
1237 
1238 			/* The split routines insert the new entry,
1239 			 * acquire txLock as appropriate.
1240 			 * return <rp> pinned and its block number <rbn>.
1241 			 */
1242 			rc = (sp->header.flag & BT_ROOT) ?
1243 			    dtSplitRoot(tid, ip, split, &rmp) :
1244 			    dtSplitPage(tid, ip, split, &rmp, &rp, &rpxd);
1245 			if (rc) {
1246 				DT_PUTPAGE(smp);
1247 				goto splitOut;
1248 			}
1249 
1250 			/* smp and rmp are pinned */
1251 		}
1252 		/*
1253 		 * parent page is not full - insert router entry in parent page
1254 		 */
1255 		else {
1256 			BT_MARK_DIRTY(smp, ip);
1257 			/*
1258 			 * acquire a transaction lock on the parent page
1259 			 */
1260 			tlck = txLock(tid, ip, smp, tlckDTREE | tlckENTRY);
1261 			dtlck = (struct dt_lock *) & tlck->lock;
1262 			ASSERT(dtlck->index == 0);
1263 			lv = & dtlck->lv[0];
1264 
1265 			/* linelock header */
1266 			lv->offset = 0;
1267 			lv->length = 1;
1268 			dtlck->index++;
1269 
1270 			/* linelock stbl of non-root parent page */
1271 			if (!(sp->header.flag & BT_ROOT)) {
1272 				lv++;
1273 				n = skip >> L2DTSLOTSIZE;
1274 				lv->offset = sp->header.stblindex + n;
1275 				lv->length =
1276 				    ((sp->header.nextindex -
1277 				      1) >> L2DTSLOTSIZE) - n + 1;
1278 				dtlck->index++;
1279 			}
1280 
1281 			dtInsertEntry(sp, skip, &key, data, &dtlck);
1282 
1283 			/* exit propagate up */
1284 			break;
1285 		}
1286 	}
1287 
1288 	/* unpin current split and its right page */
1289 	DT_PUTPAGE(smp);
1290 	DT_PUTPAGE(rmp);
1291 
1292 	/*
1293 	 * free remaining extents allocated for split
1294 	 */
1295       splitOut:
1296 	n = pxdlist.npxd;
1297 	pxd = &pxdlist.pxd[n];
1298 	for (; n < pxdlist.maxnpxd; n++, pxd++)
1299 		dbFree(ip, addressPXD(pxd), (s64) lengthPXD(pxd));
1300 
1301       freeKeyName:
1302 	kfree(key.name);
1303 
1304 	/* Rollback quota allocation */
1305 	if (rc && quota_allocation)
1306 		dquot_free_block(ip, quota_allocation);
1307 
1308       dtSplitUp_Exit:
1309 
1310 	return rc;
1311 }
1312 
1313 
1314 /*
1315  *	dtSplitPage()
1316  *
1317  * function: Split a non-root page of a btree.
1318  *
1319  * parameter:
1320  *
1321  * return: 0 - success;
1322  *	   errno - failure;
1323  *	return split and new page pinned;
1324  */
dtSplitPage(tid_t tid,struct inode * ip,struct dtsplit * split,struct metapage ** rmpp,dtpage_t ** rpp,pxd_t * rpxdp)1325 static int dtSplitPage(tid_t tid, struct inode *ip, struct dtsplit * split,
1326 	    struct metapage ** rmpp, dtpage_t ** rpp, pxd_t * rpxdp)
1327 {
1328 	int rc = 0;
1329 	struct metapage *smp;
1330 	dtpage_t *sp;
1331 	struct metapage *rmp;
1332 	dtpage_t *rp;		/* new right page allocated */
1333 	s64 rbn;		/* new right page block number */
1334 	struct metapage *mp;
1335 	dtpage_t *p;
1336 	s64 nextbn;
1337 	struct pxdlist *pxdlist;
1338 	pxd_t *pxd;
1339 	int skip, nextindex, half, left, nxt, off, si;
1340 	struct ldtentry *ldtentry;
1341 	struct idtentry *idtentry;
1342 	u8 *stbl;
1343 	struct dtslot *f;
1344 	int fsi, stblsize;
1345 	int n;
1346 	struct dt_lock *sdtlck, *rdtlck;
1347 	struct tlock *tlck;
1348 	struct dt_lock *dtlck;
1349 	struct lv *slv, *rlv, *lv;
1350 
1351 	/* get split page */
1352 	smp = split->mp;
1353 	sp = DT_PAGE(ip, smp);
1354 
1355 	/*
1356 	 * allocate the new right page for the split
1357 	 */
1358 	pxdlist = split->pxdlist;
1359 	pxd = &pxdlist->pxd[pxdlist->npxd];
1360 	pxdlist->npxd++;
1361 	rbn = addressPXD(pxd);
1362 	rmp = get_metapage(ip, rbn, PSIZE, 1);
1363 	if (rmp == NULL)
1364 		return -EIO;
1365 
1366 	/* Allocate blocks to quota. */
1367 	rc = dquot_alloc_block(ip, lengthPXD(pxd));
1368 	if (rc) {
1369 		release_metapage(rmp);
1370 		return rc;
1371 	}
1372 
1373 	jfs_info("dtSplitPage: ip:0x%p smp:0x%p rmp:0x%p", ip, smp, rmp);
1374 
1375 	BT_MARK_DIRTY(rmp, ip);
1376 	/*
1377 	 * acquire a transaction lock on the new right page
1378 	 */
1379 	tlck = txLock(tid, ip, rmp, tlckDTREE | tlckNEW);
1380 	rdtlck = (struct dt_lock *) & tlck->lock;
1381 
1382 	rp = (dtpage_t *) rmp->data;
1383 	*rpp = rp;
1384 	rp->header.self = *pxd;
1385 
1386 	BT_MARK_DIRTY(smp, ip);
1387 	/*
1388 	 * acquire a transaction lock on the split page
1389 	 *
1390 	 * action:
1391 	 */
1392 	tlck = txLock(tid, ip, smp, tlckDTREE | tlckENTRY);
1393 	sdtlck = (struct dt_lock *) & tlck->lock;
1394 
1395 	/* linelock header of split page */
1396 	ASSERT(sdtlck->index == 0);
1397 	slv = & sdtlck->lv[0];
1398 	slv->offset = 0;
1399 	slv->length = 1;
1400 	sdtlck->index++;
1401 
1402 	/*
1403 	 * initialize/update sibling pointers between sp and rp
1404 	 */
1405 	nextbn = le64_to_cpu(sp->header.next);
1406 	rp->header.next = cpu_to_le64(nextbn);
1407 	rp->header.prev = cpu_to_le64(addressPXD(&sp->header.self));
1408 	sp->header.next = cpu_to_le64(rbn);
1409 
1410 	/*
1411 	 * initialize new right page
1412 	 */
1413 	rp->header.flag = sp->header.flag;
1414 
1415 	/* compute sorted entry table at start of extent data area */
1416 	rp->header.nextindex = 0;
1417 	rp->header.stblindex = 1;
1418 
1419 	n = PSIZE >> L2DTSLOTSIZE;
1420 	rp->header.maxslot = n;
1421 	stblsize = (n + 31) >> L2DTSLOTSIZE;	/* in unit of slot */
1422 
1423 	/* init freelist */
1424 	fsi = rp->header.stblindex + stblsize;
1425 	rp->header.freelist = fsi;
1426 	rp->header.freecnt = rp->header.maxslot - fsi;
1427 
1428 	/*
1429 	 *	sequential append at tail: append without split
1430 	 *
1431 	 * If splitting the last page on a level because of appending
1432 	 * a entry to it (skip is maxentry), it's likely that the access is
1433 	 * sequential. Adding an empty page on the side of the level is less
1434 	 * work and can push the fill factor much higher than normal.
1435 	 * If we're wrong it's no big deal, we'll just do the split the right
1436 	 * way next time.
1437 	 * (It may look like it's equally easy to do a similar hack for
1438 	 * reverse sorted data, that is, split the tree left,
1439 	 * but it's not. Be my guest.)
1440 	 */
1441 	if (nextbn == 0 && split->index == sp->header.nextindex) {
1442 		/* linelock header + stbl (first slot) of new page */
1443 		rlv = & rdtlck->lv[rdtlck->index];
1444 		rlv->offset = 0;
1445 		rlv->length = 2;
1446 		rdtlck->index++;
1447 
1448 		/*
1449 		 * initialize freelist of new right page
1450 		 */
1451 		f = &rp->slot[fsi];
1452 		for (fsi++; fsi < rp->header.maxslot; f++, fsi++)
1453 			f->next = fsi;
1454 		f->next = -1;
1455 
1456 		/* insert entry at the first entry of the new right page */
1457 		dtInsertEntry(rp, 0, split->key, split->data, &rdtlck);
1458 
1459 		goto out;
1460 	}
1461 
1462 	/*
1463 	 *	non-sequential insert (at possibly middle page)
1464 	 */
1465 
1466 	/*
1467 	 * update prev pointer of previous right sibling page;
1468 	 */
1469 	if (nextbn != 0) {
1470 		DT_GETPAGE(ip, nextbn, mp, PSIZE, p, rc);
1471 		if (rc) {
1472 			discard_metapage(rmp);
1473 			return rc;
1474 		}
1475 
1476 		BT_MARK_DIRTY(mp, ip);
1477 		/*
1478 		 * acquire a transaction lock on the next page
1479 		 */
1480 		tlck = txLock(tid, ip, mp, tlckDTREE | tlckRELINK);
1481 		jfs_info("dtSplitPage: tlck = 0x%p, ip = 0x%p, mp=0x%p",
1482 			tlck, ip, mp);
1483 		dtlck = (struct dt_lock *) & tlck->lock;
1484 
1485 		/* linelock header of previous right sibling page */
1486 		lv = & dtlck->lv[dtlck->index];
1487 		lv->offset = 0;
1488 		lv->length = 1;
1489 		dtlck->index++;
1490 
1491 		p->header.prev = cpu_to_le64(rbn);
1492 
1493 		DT_PUTPAGE(mp);
1494 	}
1495 
1496 	/*
1497 	 * split the data between the split and right pages.
1498 	 */
1499 	skip = split->index;
1500 	half = (PSIZE >> L2DTSLOTSIZE) >> 1;	/* swag */
1501 	left = 0;
1502 
1503 	/*
1504 	 *	compute fill factor for split pages
1505 	 *
1506 	 * <nxt> traces the next entry to move to rp
1507 	 * <off> traces the next entry to stay in sp
1508 	 */
1509 	stbl = (u8 *) & sp->slot[sp->header.stblindex];
1510 	nextindex = sp->header.nextindex;
1511 	for (nxt = off = 0; nxt < nextindex; ++off) {
1512 		if (off == skip)
1513 			/* check for fill factor with new entry size */
1514 			n = split->nslot;
1515 		else {
1516 			si = stbl[nxt];
1517 			switch (sp->header.flag & BT_TYPE) {
1518 			case BT_LEAF:
1519 				ldtentry = (struct ldtentry *) & sp->slot[si];
1520 				if (DO_INDEX(ip))
1521 					n = NDTLEAF(ldtentry->namlen);
1522 				else
1523 					n = NDTLEAF_LEGACY(ldtentry->
1524 							   namlen);
1525 				break;
1526 
1527 			case BT_INTERNAL:
1528 				idtentry = (struct idtentry *) & sp->slot[si];
1529 				n = NDTINTERNAL(idtentry->namlen);
1530 				break;
1531 
1532 			default:
1533 				break;
1534 			}
1535 
1536 			++nxt;	/* advance to next entry to move in sp */
1537 		}
1538 
1539 		left += n;
1540 		if (left >= half)
1541 			break;
1542 	}
1543 
1544 	/* <nxt> poins to the 1st entry to move */
1545 
1546 	/*
1547 	 *	move entries to right page
1548 	 *
1549 	 * dtMoveEntry() initializes rp and reserves entry for insertion
1550 	 *
1551 	 * split page moved out entries are linelocked;
1552 	 * new/right page moved in entries are linelocked;
1553 	 */
1554 	/* linelock header + stbl of new right page */
1555 	rlv = & rdtlck->lv[rdtlck->index];
1556 	rlv->offset = 0;
1557 	rlv->length = 5;
1558 	rdtlck->index++;
1559 
1560 	dtMoveEntry(sp, nxt, rp, &sdtlck, &rdtlck, DO_INDEX(ip));
1561 
1562 	sp->header.nextindex = nxt;
1563 
1564 	/*
1565 	 * finalize freelist of new right page
1566 	 */
1567 	fsi = rp->header.freelist;
1568 	f = &rp->slot[fsi];
1569 	for (fsi++; fsi < rp->header.maxslot; f++, fsi++)
1570 		f->next = fsi;
1571 	f->next = -1;
1572 
1573 	/*
1574 	 * Update directory index table for entries now in right page
1575 	 */
1576 	if ((rp->header.flag & BT_LEAF) && DO_INDEX(ip)) {
1577 		s64 lblock;
1578 
1579 		mp = NULL;
1580 		stbl = DT_GETSTBL(rp);
1581 		for (n = 0; n < rp->header.nextindex; n++) {
1582 			ldtentry = (struct ldtentry *) & rp->slot[stbl[n]];
1583 			modify_index(tid, ip, le32_to_cpu(ldtentry->index),
1584 				     rbn, n, &mp, &lblock);
1585 		}
1586 		if (mp)
1587 			release_metapage(mp);
1588 	}
1589 
1590 	/*
1591 	 * the skipped index was on the left page,
1592 	 */
1593 	if (skip <= off) {
1594 		/* insert the new entry in the split page */
1595 		dtInsertEntry(sp, skip, split->key, split->data, &sdtlck);
1596 
1597 		/* linelock stbl of split page */
1598 		if (sdtlck->index >= sdtlck->maxcnt)
1599 			sdtlck = (struct dt_lock *) txLinelock(sdtlck);
1600 		slv = & sdtlck->lv[sdtlck->index];
1601 		n = skip >> L2DTSLOTSIZE;
1602 		slv->offset = sp->header.stblindex + n;
1603 		slv->length =
1604 		    ((sp->header.nextindex - 1) >> L2DTSLOTSIZE) - n + 1;
1605 		sdtlck->index++;
1606 	}
1607 	/*
1608 	 * the skipped index was on the right page,
1609 	 */
1610 	else {
1611 		/* adjust the skip index to reflect the new position */
1612 		skip -= nxt;
1613 
1614 		/* insert the new entry in the right page */
1615 		dtInsertEntry(rp, skip, split->key, split->data, &rdtlck);
1616 	}
1617 
1618       out:
1619 	*rmpp = rmp;
1620 	*rpxdp = *pxd;
1621 
1622 	return rc;
1623 }
1624 
1625 
1626 /*
1627  *	dtExtendPage()
1628  *
1629  * function: extend 1st/only directory leaf page
1630  *
1631  * parameter:
1632  *
1633  * return: 0 - success;
1634  *	   errno - failure;
1635  *	return extended page pinned;
1636  */
dtExtendPage(tid_t tid,struct inode * ip,struct dtsplit * split,struct btstack * btstack)1637 static int dtExtendPage(tid_t tid,
1638 	     struct inode *ip, struct dtsplit * split, struct btstack * btstack)
1639 {
1640 	struct super_block *sb = ip->i_sb;
1641 	int rc;
1642 	struct metapage *smp, *pmp, *mp;
1643 	dtpage_t *sp, *pp;
1644 	struct pxdlist *pxdlist;
1645 	pxd_t *pxd, *tpxd;
1646 	int xlen, xsize;
1647 	int newstblindex, newstblsize;
1648 	int oldstblindex, oldstblsize;
1649 	int fsi, last;
1650 	struct dtslot *f;
1651 	struct btframe *parent;
1652 	int n;
1653 	struct dt_lock *dtlck;
1654 	s64 xaddr, txaddr;
1655 	struct tlock *tlck;
1656 	struct pxd_lock *pxdlock;
1657 	struct lv *lv;
1658 	uint type;
1659 	struct ldtentry *ldtentry;
1660 	u8 *stbl;
1661 
1662 	/* get page to extend */
1663 	smp = split->mp;
1664 	sp = DT_PAGE(ip, smp);
1665 
1666 	/* get parent/root page */
1667 	parent = BT_POP(btstack);
1668 	DT_GETPAGE(ip, parent->bn, pmp, PSIZE, pp, rc);
1669 	if (rc)
1670 		return (rc);
1671 
1672 	/*
1673 	 *	extend the extent
1674 	 */
1675 	pxdlist = split->pxdlist;
1676 	pxd = &pxdlist->pxd[pxdlist->npxd];
1677 	pxdlist->npxd++;
1678 
1679 	xaddr = addressPXD(pxd);
1680 	tpxd = &sp->header.self;
1681 	txaddr = addressPXD(tpxd);
1682 	/* in-place extension */
1683 	if (xaddr == txaddr) {
1684 		type = tlckEXTEND;
1685 	}
1686 	/* relocation */
1687 	else {
1688 		type = tlckNEW;
1689 
1690 		/* save moved extent descriptor for later free */
1691 		tlck = txMaplock(tid, ip, tlckDTREE | tlckRELOCATE);
1692 		pxdlock = (struct pxd_lock *) & tlck->lock;
1693 		pxdlock->flag = mlckFREEPXD;
1694 		pxdlock->pxd = sp->header.self;
1695 		pxdlock->index = 1;
1696 
1697 		/*
1698 		 * Update directory index table to reflect new page address
1699 		 */
1700 		if (DO_INDEX(ip)) {
1701 			s64 lblock;
1702 
1703 			mp = NULL;
1704 			stbl = DT_GETSTBL(sp);
1705 			for (n = 0; n < sp->header.nextindex; n++) {
1706 				ldtentry =
1707 				    (struct ldtentry *) & sp->slot[stbl[n]];
1708 				modify_index(tid, ip,
1709 					     le32_to_cpu(ldtentry->index),
1710 					     xaddr, n, &mp, &lblock);
1711 			}
1712 			if (mp)
1713 				release_metapage(mp);
1714 		}
1715 	}
1716 
1717 	/*
1718 	 *	extend the page
1719 	 */
1720 	sp->header.self = *pxd;
1721 
1722 	jfs_info("dtExtendPage: ip:0x%p smp:0x%p sp:0x%p", ip, smp, sp);
1723 
1724 	BT_MARK_DIRTY(smp, ip);
1725 	/*
1726 	 * acquire a transaction lock on the extended/leaf page
1727 	 */
1728 	tlck = txLock(tid, ip, smp, tlckDTREE | type);
1729 	dtlck = (struct dt_lock *) & tlck->lock;
1730 	lv = & dtlck->lv[0];
1731 
1732 	/* update buffer extent descriptor of extended page */
1733 	xlen = lengthPXD(pxd);
1734 	xsize = xlen << JFS_SBI(sb)->l2bsize;
1735 
1736 	/*
1737 	 * copy old stbl to new stbl at start of extended area
1738 	 */
1739 	oldstblindex = sp->header.stblindex;
1740 	oldstblsize = (sp->header.maxslot + 31) >> L2DTSLOTSIZE;
1741 	newstblindex = sp->header.maxslot;
1742 	n = xsize >> L2DTSLOTSIZE;
1743 	newstblsize = (n + 31) >> L2DTSLOTSIZE;
1744 	memcpy(&sp->slot[newstblindex], &sp->slot[oldstblindex],
1745 	       sp->header.nextindex);
1746 
1747 	/*
1748 	 * in-line extension: linelock old area of extended page
1749 	 */
1750 	if (type == tlckEXTEND) {
1751 		/* linelock header */
1752 		lv->offset = 0;
1753 		lv->length = 1;
1754 		dtlck->index++;
1755 		lv++;
1756 
1757 		/* linelock new stbl of extended page */
1758 		lv->offset = newstblindex;
1759 		lv->length = newstblsize;
1760 	}
1761 	/*
1762 	 * relocation: linelock whole relocated area
1763 	 */
1764 	else {
1765 		lv->offset = 0;
1766 		lv->length = sp->header.maxslot + newstblsize;
1767 	}
1768 
1769 	dtlck->index++;
1770 
1771 	sp->header.maxslot = n;
1772 	sp->header.stblindex = newstblindex;
1773 	/* sp->header.nextindex remains the same */
1774 
1775 	/*
1776 	 * add old stbl region at head of freelist
1777 	 */
1778 	fsi = oldstblindex;
1779 	f = &sp->slot[fsi];
1780 	last = sp->header.freelist;
1781 	for (n = 0; n < oldstblsize; n++, fsi++, f++) {
1782 		f->next = last;
1783 		last = fsi;
1784 	}
1785 	sp->header.freelist = last;
1786 	sp->header.freecnt += oldstblsize;
1787 
1788 	/*
1789 	 * append free region of newly extended area at tail of freelist
1790 	 */
1791 	/* init free region of newly extended area */
1792 	fsi = n = newstblindex + newstblsize;
1793 	f = &sp->slot[fsi];
1794 	for (fsi++; fsi < sp->header.maxslot; f++, fsi++)
1795 		f->next = fsi;
1796 	f->next = -1;
1797 
1798 	/* append new free region at tail of old freelist */
1799 	fsi = sp->header.freelist;
1800 	if (fsi == -1)
1801 		sp->header.freelist = n;
1802 	else {
1803 		do {
1804 			f = &sp->slot[fsi];
1805 			fsi = f->next;
1806 		} while (fsi != -1);
1807 
1808 		f->next = n;
1809 	}
1810 
1811 	sp->header.freecnt += sp->header.maxslot - n;
1812 
1813 	/*
1814 	 * insert the new entry
1815 	 */
1816 	dtInsertEntry(sp, split->index, split->key, split->data, &dtlck);
1817 
1818 	BT_MARK_DIRTY(pmp, ip);
1819 	/*
1820 	 * linelock any freeslots residing in old extent
1821 	 */
1822 	if (type == tlckEXTEND) {
1823 		n = sp->header.maxslot >> 2;
1824 		if (sp->header.freelist < n)
1825 			dtLinelockFreelist(sp, n, &dtlck);
1826 	}
1827 
1828 	/*
1829 	 *	update parent entry on the parent/root page
1830 	 */
1831 	/*
1832 	 * acquire a transaction lock on the parent/root page
1833 	 */
1834 	tlck = txLock(tid, ip, pmp, tlckDTREE | tlckENTRY);
1835 	dtlck = (struct dt_lock *) & tlck->lock;
1836 	lv = & dtlck->lv[dtlck->index];
1837 
1838 	/* linelock parent entry - 1st slot */
1839 	lv->offset = 1;
1840 	lv->length = 1;
1841 	dtlck->index++;
1842 
1843 	/* update the parent pxd for page extension */
1844 	tpxd = (pxd_t *) & pp->slot[1];
1845 	*tpxd = *pxd;
1846 
1847 	DT_PUTPAGE(pmp);
1848 	return 0;
1849 }
1850 
1851 
1852 /*
1853  *	dtSplitRoot()
1854  *
1855  * function:
1856  *	split the full root page into
1857  *	original/root/split page and new right page
1858  *	i.e., root remains fixed in tree anchor (inode) and
1859  *	the root is copied to a single new right child page
1860  *	since root page << non-root page, and
1861  *	the split root page contains a single entry for the
1862  *	new right child page.
1863  *
1864  * parameter:
1865  *
1866  * return: 0 - success;
1867  *	   errno - failure;
1868  *	return new page pinned;
1869  */
dtSplitRoot(tid_t tid,struct inode * ip,struct dtsplit * split,struct metapage ** rmpp)1870 static int dtSplitRoot(tid_t tid,
1871 	    struct inode *ip, struct dtsplit * split, struct metapage ** rmpp)
1872 {
1873 	struct super_block *sb = ip->i_sb;
1874 	struct metapage *smp;
1875 	dtroot_t *sp;
1876 	struct metapage *rmp;
1877 	dtpage_t *rp;
1878 	s64 rbn;
1879 	int xlen;
1880 	int xsize;
1881 	struct dtslot *f;
1882 	s8 *stbl;
1883 	int fsi, stblsize, n;
1884 	struct idtentry *s;
1885 	pxd_t *ppxd;
1886 	struct pxdlist *pxdlist;
1887 	pxd_t *pxd;
1888 	struct dt_lock *dtlck;
1889 	struct tlock *tlck;
1890 	struct lv *lv;
1891 	int rc;
1892 
1893 	/* get split root page */
1894 	smp = split->mp;
1895 	sp = &JFS_IP(ip)->i_dtroot;
1896 
1897 	/*
1898 	 *	allocate/initialize a single (right) child page
1899 	 *
1900 	 * N.B. at first split, a one (or two) block to fit new entry
1901 	 * is allocated; at subsequent split, a full page is allocated;
1902 	 */
1903 	pxdlist = split->pxdlist;
1904 	pxd = &pxdlist->pxd[pxdlist->npxd];
1905 	pxdlist->npxd++;
1906 	rbn = addressPXD(pxd);
1907 	xlen = lengthPXD(pxd);
1908 	xsize = xlen << JFS_SBI(sb)->l2bsize;
1909 	rmp = get_metapage(ip, rbn, xsize, 1);
1910 	if (!rmp)
1911 		return -EIO;
1912 
1913 	rp = rmp->data;
1914 
1915 	/* Allocate blocks to quota. */
1916 	rc = dquot_alloc_block(ip, lengthPXD(pxd));
1917 	if (rc) {
1918 		release_metapage(rmp);
1919 		return rc;
1920 	}
1921 
1922 	BT_MARK_DIRTY(rmp, ip);
1923 	/*
1924 	 * acquire a transaction lock on the new right page
1925 	 */
1926 	tlck = txLock(tid, ip, rmp, tlckDTREE | tlckNEW);
1927 	dtlck = (struct dt_lock *) & tlck->lock;
1928 
1929 	rp->header.flag =
1930 	    (sp->header.flag & BT_LEAF) ? BT_LEAF : BT_INTERNAL;
1931 	rp->header.self = *pxd;
1932 
1933 	/* initialize sibling pointers */
1934 	rp->header.next = 0;
1935 	rp->header.prev = 0;
1936 
1937 	/*
1938 	 *	move in-line root page into new right page extent
1939 	 */
1940 	/* linelock header + copied entries + new stbl (1st slot) in new page */
1941 	ASSERT(dtlck->index == 0);
1942 	lv = & dtlck->lv[0];
1943 	lv->offset = 0;
1944 	lv->length = 10;	/* 1 + 8 + 1 */
1945 	dtlck->index++;
1946 
1947 	n = xsize >> L2DTSLOTSIZE;
1948 	rp->header.maxslot = n;
1949 	stblsize = (n + 31) >> L2DTSLOTSIZE;
1950 
1951 	/* copy old stbl to new stbl at start of extended area */
1952 	rp->header.stblindex = DTROOTMAXSLOT;
1953 	stbl = (s8 *) & rp->slot[DTROOTMAXSLOT];
1954 	memcpy(stbl, sp->header.stbl, sp->header.nextindex);
1955 	rp->header.nextindex = sp->header.nextindex;
1956 
1957 	/* copy old data area to start of new data area */
1958 	memcpy(&rp->slot[1], &sp->slot[1], IDATASIZE);
1959 
1960 	/*
1961 	 * append free region of newly extended area at tail of freelist
1962 	 */
1963 	/* init free region of newly extended area */
1964 	fsi = n = DTROOTMAXSLOT + stblsize;
1965 	f = &rp->slot[fsi];
1966 	for (fsi++; fsi < rp->header.maxslot; f++, fsi++)
1967 		f->next = fsi;
1968 	f->next = -1;
1969 
1970 	/* append new free region at tail of old freelist */
1971 	fsi = sp->header.freelist;
1972 	if (fsi == -1)
1973 		rp->header.freelist = n;
1974 	else {
1975 		rp->header.freelist = fsi;
1976 
1977 		do {
1978 			f = &rp->slot[fsi];
1979 			fsi = f->next;
1980 		} while (fsi >= 0);
1981 
1982 		f->next = n;
1983 	}
1984 
1985 	rp->header.freecnt = sp->header.freecnt + rp->header.maxslot - n;
1986 
1987 	/*
1988 	 * Update directory index table for entries now in right page
1989 	 */
1990 	if ((rp->header.flag & BT_LEAF) && DO_INDEX(ip)) {
1991 		s64 lblock;
1992 		struct metapage *mp = NULL;
1993 		struct ldtentry *ldtentry;
1994 
1995 		stbl = DT_GETSTBL(rp);
1996 		for (n = 0; n < rp->header.nextindex; n++) {
1997 			ldtentry = (struct ldtentry *) & rp->slot[stbl[n]];
1998 			modify_index(tid, ip, le32_to_cpu(ldtentry->index),
1999 				     rbn, n, &mp, &lblock);
2000 		}
2001 		if (mp)
2002 			release_metapage(mp);
2003 	}
2004 	/*
2005 	 * insert the new entry into the new right/child page
2006 	 * (skip index in the new right page will not change)
2007 	 */
2008 	dtInsertEntry(rp, split->index, split->key, split->data, &dtlck);
2009 
2010 	/*
2011 	 *	reset parent/root page
2012 	 *
2013 	 * set the 1st entry offset to 0, which force the left-most key
2014 	 * at any level of the tree to be less than any search key.
2015 	 *
2016 	 * The btree comparison code guarantees that the left-most key on any
2017 	 * level of the tree is never used, so it doesn't need to be filled in.
2018 	 */
2019 	BT_MARK_DIRTY(smp, ip);
2020 	/*
2021 	 * acquire a transaction lock on the root page (in-memory inode)
2022 	 */
2023 	tlck = txLock(tid, ip, smp, tlckDTREE | tlckNEW | tlckBTROOT);
2024 	dtlck = (struct dt_lock *) & tlck->lock;
2025 
2026 	/* linelock root */
2027 	ASSERT(dtlck->index == 0);
2028 	lv = & dtlck->lv[0];
2029 	lv->offset = 0;
2030 	lv->length = DTROOTMAXSLOT;
2031 	dtlck->index++;
2032 
2033 	/* update page header of root */
2034 	if (sp->header.flag & BT_LEAF) {
2035 		sp->header.flag &= ~BT_LEAF;
2036 		sp->header.flag |= BT_INTERNAL;
2037 	}
2038 
2039 	/* init the first entry */
2040 	s = (struct idtentry *) & sp->slot[DTENTRYSTART];
2041 	ppxd = (pxd_t *) s;
2042 	*ppxd = *pxd;
2043 	s->next = -1;
2044 	s->namlen = 0;
2045 
2046 	stbl = sp->header.stbl;
2047 	stbl[0] = DTENTRYSTART;
2048 	sp->header.nextindex = 1;
2049 
2050 	/* init freelist */
2051 	fsi = DTENTRYSTART + 1;
2052 	f = &sp->slot[fsi];
2053 
2054 	/* init free region of remaining area */
2055 	for (fsi++; fsi < DTROOTMAXSLOT; f++, fsi++)
2056 		f->next = fsi;
2057 	f->next = -1;
2058 
2059 	sp->header.freelist = DTENTRYSTART + 1;
2060 	sp->header.freecnt = DTROOTMAXSLOT - (DTENTRYSTART + 1);
2061 
2062 	*rmpp = rmp;
2063 
2064 	return 0;
2065 }
2066 
2067 
2068 /*
2069  *	dtDelete()
2070  *
2071  * function: delete the entry(s) referenced by a key.
2072  *
2073  * parameter:
2074  *
2075  * return:
2076  */
dtDelete(tid_t tid,struct inode * ip,struct component_name * key,ino_t * ino,int flag)2077 int dtDelete(tid_t tid,
2078 	 struct inode *ip, struct component_name * key, ino_t * ino, int flag)
2079 {
2080 	int rc = 0;
2081 	s64 bn;
2082 	struct metapage *mp, *imp;
2083 	dtpage_t *p;
2084 	int index;
2085 	struct btstack btstack;
2086 	struct dt_lock *dtlck;
2087 	struct tlock *tlck;
2088 	struct lv *lv;
2089 	int i;
2090 	struct ldtentry *ldtentry;
2091 	u8 *stbl;
2092 	u32 table_index, next_index;
2093 	struct metapage *nmp;
2094 	dtpage_t *np;
2095 
2096 	/*
2097 	 *	search for the entry to delete:
2098 	 *
2099 	 * dtSearch() returns (leaf page pinned, index at which to delete).
2100 	 */
2101 	if ((rc = dtSearch(ip, key, ino, &btstack, flag)))
2102 		return rc;
2103 
2104 	/* retrieve search result */
2105 	DT_GETSEARCH(ip, btstack.top, bn, mp, p, index);
2106 
2107 	/*
2108 	 * We need to find put the index of the next entry into the
2109 	 * directory index table in order to resume a readdir from this
2110 	 * entry.
2111 	 */
2112 	if (DO_INDEX(ip)) {
2113 		stbl = DT_GETSTBL(p);
2114 		ldtentry = (struct ldtentry *) & p->slot[stbl[index]];
2115 		table_index = le32_to_cpu(ldtentry->index);
2116 		if (index == (p->header.nextindex - 1)) {
2117 			/*
2118 			 * Last entry in this leaf page
2119 			 */
2120 			if ((p->header.flag & BT_ROOT)
2121 			    || (p->header.next == 0))
2122 				next_index = -1;
2123 			else {
2124 				/* Read next leaf page */
2125 				DT_GETPAGE(ip, le64_to_cpu(p->header.next),
2126 					   nmp, PSIZE, np, rc);
2127 				if (rc)
2128 					next_index = -1;
2129 				else {
2130 					stbl = DT_GETSTBL(np);
2131 					ldtentry =
2132 					    (struct ldtentry *) & np->
2133 					    slot[stbl[0]];
2134 					next_index =
2135 					    le32_to_cpu(ldtentry->index);
2136 					DT_PUTPAGE(nmp);
2137 				}
2138 			}
2139 		} else {
2140 			ldtentry =
2141 			    (struct ldtentry *) & p->slot[stbl[index + 1]];
2142 			next_index = le32_to_cpu(ldtentry->index);
2143 		}
2144 		free_index(tid, ip, table_index, next_index);
2145 	}
2146 	/*
2147 	 * the leaf page becomes empty, delete the page
2148 	 */
2149 	if (p->header.nextindex == 1) {
2150 		/* delete empty page */
2151 		rc = dtDeleteUp(tid, ip, mp, p, &btstack);
2152 	}
2153 	/*
2154 	 * the leaf page has other entries remaining:
2155 	 *
2156 	 * delete the entry from the leaf page.
2157 	 */
2158 	else {
2159 		BT_MARK_DIRTY(mp, ip);
2160 		/*
2161 		 * acquire a transaction lock on the leaf page
2162 		 */
2163 		tlck = txLock(tid, ip, mp, tlckDTREE | tlckENTRY);
2164 		dtlck = (struct dt_lock *) & tlck->lock;
2165 
2166 		/*
2167 		 * Do not assume that dtlck->index will be zero.  During a
2168 		 * rename within a directory, this transaction may have
2169 		 * modified this page already when adding the new entry.
2170 		 */
2171 
2172 		/* linelock header */
2173 		if (dtlck->index >= dtlck->maxcnt)
2174 			dtlck = (struct dt_lock *) txLinelock(dtlck);
2175 		lv = & dtlck->lv[dtlck->index];
2176 		lv->offset = 0;
2177 		lv->length = 1;
2178 		dtlck->index++;
2179 
2180 		/* linelock stbl of non-root leaf page */
2181 		if (!(p->header.flag & BT_ROOT)) {
2182 			if (dtlck->index >= dtlck->maxcnt)
2183 				dtlck = (struct dt_lock *) txLinelock(dtlck);
2184 			lv = & dtlck->lv[dtlck->index];
2185 			i = index >> L2DTSLOTSIZE;
2186 			lv->offset = p->header.stblindex + i;
2187 			lv->length =
2188 			    ((p->header.nextindex - 1) >> L2DTSLOTSIZE) -
2189 			    i + 1;
2190 			dtlck->index++;
2191 		}
2192 
2193 		/* free the leaf entry */
2194 		dtDeleteEntry(p, index, &dtlck);
2195 
2196 		/*
2197 		 * Update directory index table for entries moved in stbl
2198 		 */
2199 		if (DO_INDEX(ip) && index < p->header.nextindex) {
2200 			s64 lblock;
2201 
2202 			imp = NULL;
2203 			stbl = DT_GETSTBL(p);
2204 			for (i = index; i < p->header.nextindex; i++) {
2205 				ldtentry =
2206 				    (struct ldtentry *) & p->slot[stbl[i]];
2207 				modify_index(tid, ip,
2208 					     le32_to_cpu(ldtentry->index),
2209 					     bn, i, &imp, &lblock);
2210 			}
2211 			if (imp)
2212 				release_metapage(imp);
2213 		}
2214 
2215 		DT_PUTPAGE(mp);
2216 	}
2217 
2218 	return rc;
2219 }
2220 
2221 
2222 /*
2223  *	dtDeleteUp()
2224  *
2225  * function:
2226  *	free empty pages as propagating deletion up the tree
2227  *
2228  * parameter:
2229  *
2230  * return:
2231  */
dtDeleteUp(tid_t tid,struct inode * ip,struct metapage * fmp,dtpage_t * fp,struct btstack * btstack)2232 static int dtDeleteUp(tid_t tid, struct inode *ip,
2233 	   struct metapage * fmp, dtpage_t * fp, struct btstack * btstack)
2234 {
2235 	int rc = 0;
2236 	struct metapage *mp;
2237 	dtpage_t *p;
2238 	int index, nextindex;
2239 	int xlen;
2240 	struct btframe *parent;
2241 	struct dt_lock *dtlck;
2242 	struct tlock *tlck;
2243 	struct lv *lv;
2244 	struct pxd_lock *pxdlock;
2245 	int i;
2246 
2247 	/*
2248 	 *	keep the root leaf page which has become empty
2249 	 */
2250 	if (BT_IS_ROOT(fmp)) {
2251 		/*
2252 		 * reset the root
2253 		 *
2254 		 * dtInitRoot() acquires txlock on the root
2255 		 */
2256 		dtInitRoot(tid, ip, PARENT(ip));
2257 
2258 		DT_PUTPAGE(fmp);
2259 
2260 		return 0;
2261 	}
2262 
2263 	/*
2264 	 *	free the non-root leaf page
2265 	 */
2266 	/*
2267 	 * acquire a transaction lock on the page
2268 	 *
2269 	 * write FREEXTENT|NOREDOPAGE log record
2270 	 * N.B. linelock is overlaid as freed extent descriptor, and
2271 	 * the buffer page is freed;
2272 	 */
2273 	tlck = txMaplock(tid, ip, tlckDTREE | tlckFREE);
2274 	pxdlock = (struct pxd_lock *) & tlck->lock;
2275 	pxdlock->flag = mlckFREEPXD;
2276 	pxdlock->pxd = fp->header.self;
2277 	pxdlock->index = 1;
2278 
2279 	/* update sibling pointers */
2280 	if ((rc = dtRelink(tid, ip, fp))) {
2281 		BT_PUTPAGE(fmp);
2282 		return rc;
2283 	}
2284 
2285 	xlen = lengthPXD(&fp->header.self);
2286 
2287 	/* Free quota allocation. */
2288 	dquot_free_block(ip, xlen);
2289 
2290 	/* free/invalidate its buffer page */
2291 	discard_metapage(fmp);
2292 
2293 	/*
2294 	 *	propagate page deletion up the directory tree
2295 	 *
2296 	 * If the delete from the parent page makes it empty,
2297 	 * continue all the way up the tree.
2298 	 * stop if the root page is reached (which is never deleted) or
2299 	 * if the entry deletion does not empty the page.
2300 	 */
2301 	while ((parent = BT_POP(btstack)) != NULL) {
2302 		/* pin the parent page <sp> */
2303 		DT_GETPAGE(ip, parent->bn, mp, PSIZE, p, rc);
2304 		if (rc)
2305 			return rc;
2306 
2307 		/*
2308 		 * free the extent of the child page deleted
2309 		 */
2310 		index = parent->index;
2311 
2312 		/*
2313 		 * delete the entry for the child page from parent
2314 		 */
2315 		nextindex = p->header.nextindex;
2316 
2317 		/*
2318 		 * the parent has the single entry being deleted:
2319 		 *
2320 		 * free the parent page which has become empty.
2321 		 */
2322 		if (nextindex == 1) {
2323 			/*
2324 			 * keep the root internal page which has become empty
2325 			 */
2326 			if (p->header.flag & BT_ROOT) {
2327 				/*
2328 				 * reset the root
2329 				 *
2330 				 * dtInitRoot() acquires txlock on the root
2331 				 */
2332 				dtInitRoot(tid, ip, PARENT(ip));
2333 
2334 				DT_PUTPAGE(mp);
2335 
2336 				return 0;
2337 			}
2338 			/*
2339 			 * free the parent page
2340 			 */
2341 			else {
2342 				/*
2343 				 * acquire a transaction lock on the page
2344 				 *
2345 				 * write FREEXTENT|NOREDOPAGE log record
2346 				 */
2347 				tlck =
2348 				    txMaplock(tid, ip,
2349 					      tlckDTREE | tlckFREE);
2350 				pxdlock = (struct pxd_lock *) & tlck->lock;
2351 				pxdlock->flag = mlckFREEPXD;
2352 				pxdlock->pxd = p->header.self;
2353 				pxdlock->index = 1;
2354 
2355 				/* update sibling pointers */
2356 				if ((rc = dtRelink(tid, ip, p))) {
2357 					DT_PUTPAGE(mp);
2358 					return rc;
2359 				}
2360 
2361 				xlen = lengthPXD(&p->header.self);
2362 
2363 				/* Free quota allocation */
2364 				dquot_free_block(ip, xlen);
2365 
2366 				/* free/invalidate its buffer page */
2367 				discard_metapage(mp);
2368 
2369 				/* propagate up */
2370 				continue;
2371 			}
2372 		}
2373 
2374 		/*
2375 		 * the parent has other entries remaining:
2376 		 *
2377 		 * delete the router entry from the parent page.
2378 		 */
2379 		BT_MARK_DIRTY(mp, ip);
2380 		/*
2381 		 * acquire a transaction lock on the page
2382 		 *
2383 		 * action: router entry deletion
2384 		 */
2385 		tlck = txLock(tid, ip, mp, tlckDTREE | tlckENTRY);
2386 		dtlck = (struct dt_lock *) & tlck->lock;
2387 
2388 		/* linelock header */
2389 		if (dtlck->index >= dtlck->maxcnt)
2390 			dtlck = (struct dt_lock *) txLinelock(dtlck);
2391 		lv = & dtlck->lv[dtlck->index];
2392 		lv->offset = 0;
2393 		lv->length = 1;
2394 		dtlck->index++;
2395 
2396 		/* linelock stbl of non-root leaf page */
2397 		if (!(p->header.flag & BT_ROOT)) {
2398 			if (dtlck->index < dtlck->maxcnt)
2399 				lv++;
2400 			else {
2401 				dtlck = (struct dt_lock *) txLinelock(dtlck);
2402 				lv = & dtlck->lv[0];
2403 			}
2404 			i = index >> L2DTSLOTSIZE;
2405 			lv->offset = p->header.stblindex + i;
2406 			lv->length =
2407 			    ((p->header.nextindex - 1) >> L2DTSLOTSIZE) -
2408 			    i + 1;
2409 			dtlck->index++;
2410 		}
2411 
2412 		/* free the router entry */
2413 		dtDeleteEntry(p, index, &dtlck);
2414 
2415 		/* reset key of new leftmost entry of level (for consistency) */
2416 		if (index == 0 &&
2417 		    ((p->header.flag & BT_ROOT) || p->header.prev == 0))
2418 			dtTruncateEntry(p, 0, &dtlck);
2419 
2420 		/* unpin the parent page */
2421 		DT_PUTPAGE(mp);
2422 
2423 		/* exit propagation up */
2424 		break;
2425 	}
2426 
2427 	if (!DO_INDEX(ip))
2428 		ip->i_size -= PSIZE;
2429 
2430 	return 0;
2431 }
2432 
2433 /*
2434  *	dtRelink()
2435  *
2436  * function:
2437  *	link around a freed page.
2438  *
2439  * parameter:
2440  *	fp:	page to be freed
2441  *
2442  * return:
2443  */
dtRelink(tid_t tid,struct inode * ip,dtpage_t * p)2444 static int dtRelink(tid_t tid, struct inode *ip, dtpage_t * p)
2445 {
2446 	int rc;
2447 	struct metapage *mp;
2448 	s64 nextbn, prevbn;
2449 	struct tlock *tlck;
2450 	struct dt_lock *dtlck;
2451 	struct lv *lv;
2452 
2453 	nextbn = le64_to_cpu(p->header.next);
2454 	prevbn = le64_to_cpu(p->header.prev);
2455 
2456 	/* update prev pointer of the next page */
2457 	if (nextbn != 0) {
2458 		DT_GETPAGE(ip, nextbn, mp, PSIZE, p, rc);
2459 		if (rc)
2460 			return rc;
2461 
2462 		BT_MARK_DIRTY(mp, ip);
2463 		/*
2464 		 * acquire a transaction lock on the next page
2465 		 *
2466 		 * action: update prev pointer;
2467 		 */
2468 		tlck = txLock(tid, ip, mp, tlckDTREE | tlckRELINK);
2469 		jfs_info("dtRelink nextbn: tlck = 0x%p, ip = 0x%p, mp=0x%p",
2470 			tlck, ip, mp);
2471 		dtlck = (struct dt_lock *) & tlck->lock;
2472 
2473 		/* linelock header */
2474 		if (dtlck->index >= dtlck->maxcnt)
2475 			dtlck = (struct dt_lock *) txLinelock(dtlck);
2476 		lv = & dtlck->lv[dtlck->index];
2477 		lv->offset = 0;
2478 		lv->length = 1;
2479 		dtlck->index++;
2480 
2481 		p->header.prev = cpu_to_le64(prevbn);
2482 		DT_PUTPAGE(mp);
2483 	}
2484 
2485 	/* update next pointer of the previous page */
2486 	if (prevbn != 0) {
2487 		DT_GETPAGE(ip, prevbn, mp, PSIZE, p, rc);
2488 		if (rc)
2489 			return rc;
2490 
2491 		BT_MARK_DIRTY(mp, ip);
2492 		/*
2493 		 * acquire a transaction lock on the prev page
2494 		 *
2495 		 * action: update next pointer;
2496 		 */
2497 		tlck = txLock(tid, ip, mp, tlckDTREE | tlckRELINK);
2498 		jfs_info("dtRelink prevbn: tlck = 0x%p, ip = 0x%p, mp=0x%p",
2499 			tlck, ip, mp);
2500 		dtlck = (struct dt_lock *) & tlck->lock;
2501 
2502 		/* linelock header */
2503 		if (dtlck->index >= dtlck->maxcnt)
2504 			dtlck = (struct dt_lock *) txLinelock(dtlck);
2505 		lv = & dtlck->lv[dtlck->index];
2506 		lv->offset = 0;
2507 		lv->length = 1;
2508 		dtlck->index++;
2509 
2510 		p->header.next = cpu_to_le64(nextbn);
2511 		DT_PUTPAGE(mp);
2512 	}
2513 
2514 	return 0;
2515 }
2516 
2517 
2518 /*
2519  *	dtInitRoot()
2520  *
2521  * initialize directory root (inline in inode)
2522  */
dtInitRoot(tid_t tid,struct inode * ip,u32 idotdot)2523 void dtInitRoot(tid_t tid, struct inode *ip, u32 idotdot)
2524 {
2525 	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
2526 	dtroot_t *p;
2527 	int fsi;
2528 	struct dtslot *f;
2529 	struct tlock *tlck;
2530 	struct dt_lock *dtlck;
2531 	struct lv *lv;
2532 	u16 xflag_save;
2533 
2534 	/*
2535 	 * If this was previously an non-empty directory, we need to remove
2536 	 * the old directory table.
2537 	 */
2538 	if (DO_INDEX(ip)) {
2539 		if (!jfs_dirtable_inline(ip)) {
2540 			struct tblock *tblk = tid_to_tblock(tid);
2541 			/*
2542 			 * We're playing games with the tid's xflag.  If
2543 			 * we're removing a regular file, the file's xtree
2544 			 * is committed with COMMIT_PMAP, but we always
2545 			 * commit the directories xtree with COMMIT_PWMAP.
2546 			 */
2547 			xflag_save = tblk->xflag;
2548 			tblk->xflag = 0;
2549 			/*
2550 			 * xtTruncate isn't guaranteed to fully truncate
2551 			 * the xtree.  The caller needs to check i_size
2552 			 * after committing the transaction to see if
2553 			 * additional truncation is needed.  The
2554 			 * COMMIT_Stale flag tells caller that we
2555 			 * initiated the truncation.
2556 			 */
2557 			xtTruncate(tid, ip, 0, COMMIT_PWMAP);
2558 			set_cflag(COMMIT_Stale, ip);
2559 
2560 			tblk->xflag = xflag_save;
2561 		} else
2562 			ip->i_size = 1;
2563 
2564 		jfs_ip->next_index = 2;
2565 	} else
2566 		ip->i_size = IDATASIZE;
2567 
2568 	/*
2569 	 * acquire a transaction lock on the root
2570 	 *
2571 	 * action: directory initialization;
2572 	 */
2573 	tlck = txLock(tid, ip, (struct metapage *) & jfs_ip->bxflag,
2574 		      tlckDTREE | tlckENTRY | tlckBTROOT);
2575 	dtlck = (struct dt_lock *) & tlck->lock;
2576 
2577 	/* linelock root */
2578 	ASSERT(dtlck->index == 0);
2579 	lv = & dtlck->lv[0];
2580 	lv->offset = 0;
2581 	lv->length = DTROOTMAXSLOT;
2582 	dtlck->index++;
2583 
2584 	p = &jfs_ip->i_dtroot;
2585 
2586 	p->header.flag = DXD_INDEX | BT_ROOT | BT_LEAF;
2587 
2588 	p->header.nextindex = 0;
2589 
2590 	/* init freelist */
2591 	fsi = 1;
2592 	f = &p->slot[fsi];
2593 
2594 	/* init data area of root */
2595 	for (fsi++; fsi < DTROOTMAXSLOT; f++, fsi++)
2596 		f->next = fsi;
2597 	f->next = -1;
2598 
2599 	p->header.freelist = 1;
2600 	p->header.freecnt = 8;
2601 
2602 	/* init '..' entry */
2603 	p->header.idotdot = cpu_to_le32(idotdot);
2604 
2605 	return;
2606 }
2607 
2608 /*
2609  *	add_missing_indices()
2610  *
2611  * function: Fix dtree page in which one or more entries has an invalid index.
2612  *	     fsck.jfs should really fix this, but it currently does not.
2613  *	     Called from jfs_readdir when bad index is detected.
2614  */
add_missing_indices(struct inode * inode,s64 bn)2615 static void add_missing_indices(struct inode *inode, s64 bn)
2616 {
2617 	struct ldtentry *d;
2618 	struct dt_lock *dtlck;
2619 	int i;
2620 	uint index;
2621 	struct lv *lv;
2622 	struct metapage *mp;
2623 	dtpage_t *p;
2624 	int rc;
2625 	s8 *stbl;
2626 	tid_t tid;
2627 	struct tlock *tlck;
2628 
2629 	tid = txBegin(inode->i_sb, 0);
2630 
2631 	DT_GETPAGE(inode, bn, mp, PSIZE, p, rc);
2632 
2633 	if (rc) {
2634 		printk(KERN_ERR "DT_GETPAGE failed!\n");
2635 		goto end;
2636 	}
2637 	BT_MARK_DIRTY(mp, inode);
2638 
2639 	ASSERT(p->header.flag & BT_LEAF);
2640 
2641 	tlck = txLock(tid, inode, mp, tlckDTREE | tlckENTRY);
2642 	if (BT_IS_ROOT(mp))
2643 		tlck->type |= tlckBTROOT;
2644 
2645 	dtlck = (struct dt_lock *) &tlck->lock;
2646 
2647 	stbl = DT_GETSTBL(p);
2648 	for (i = 0; i < p->header.nextindex; i++) {
2649 		d = (struct ldtentry *) &p->slot[stbl[i]];
2650 		index = le32_to_cpu(d->index);
2651 		if ((index < 2) || (index >= JFS_IP(inode)->next_index)) {
2652 			d->index = cpu_to_le32(add_index(tid, inode, bn, i));
2653 			if (dtlck->index >= dtlck->maxcnt)
2654 				dtlck = (struct dt_lock *) txLinelock(dtlck);
2655 			lv = &dtlck->lv[dtlck->index];
2656 			lv->offset = stbl[i];
2657 			lv->length = 1;
2658 			dtlck->index++;
2659 		}
2660 	}
2661 
2662 	DT_PUTPAGE(mp);
2663 	(void) txCommit(tid, 1, &inode, 0);
2664 end:
2665 	txEnd(tid);
2666 }
2667 
2668 /*
2669  * Buffer to hold directory entry info while traversing a dtree page
2670  * before being fed to the filldir function
2671  */
2672 struct jfs_dirent {
2673 	loff_t position;
2674 	int ino;
2675 	u16 name_len;
2676 	char name[];
2677 };
2678 
2679 /*
2680  * function to determine next variable-sized jfs_dirent in buffer
2681  */
next_jfs_dirent(struct jfs_dirent * dirent)2682 static inline struct jfs_dirent *next_jfs_dirent(struct jfs_dirent *dirent)
2683 {
2684 	return (struct jfs_dirent *)
2685 		((char *)dirent +
2686 		 ((sizeof (struct jfs_dirent) + dirent->name_len + 1 +
2687 		   sizeof (loff_t) - 1) &
2688 		  ~(sizeof (loff_t) - 1)));
2689 }
2690 
2691 /*
2692  *	jfs_readdir()
2693  *
2694  * function: read directory entries sequentially
2695  *	from the specified entry offset
2696  *
2697  * parameter:
2698  *
2699  * return: offset = (pn, index) of start entry
2700  *	of next jfs_readdir()/dtRead()
2701  */
jfs_readdir(struct file * file,struct dir_context * ctx)2702 int jfs_readdir(struct file *file, struct dir_context *ctx)
2703 {
2704 	struct inode *ip = file_inode(file);
2705 	struct nls_table *codepage = JFS_SBI(ip->i_sb)->nls_tab;
2706 	int rc = 0;
2707 	loff_t dtpos;	/* legacy OS/2 style position */
2708 	struct dtoffset {
2709 		s16 pn;
2710 		s16 index;
2711 		s32 unused;
2712 	} *dtoffset = (struct dtoffset *) &dtpos;
2713 	s64 bn;
2714 	struct metapage *mp;
2715 	dtpage_t *p;
2716 	int index;
2717 	s8 *stbl;
2718 	struct btstack btstack;
2719 	int i, next;
2720 	struct ldtentry *d;
2721 	struct dtslot *t;
2722 	int d_namleft, len, outlen;
2723 	unsigned long dirent_buf;
2724 	char *name_ptr;
2725 	u32 dir_index;
2726 	int do_index = 0;
2727 	uint loop_count = 0;
2728 	struct jfs_dirent *jfs_dirent;
2729 	int jfs_dirents;
2730 	int overflow, fix_page, page_fixed = 0;
2731 	static int unique_pos = 2;	/* If we can't fix broken index */
2732 
2733 	if (ctx->pos == DIREND)
2734 		return 0;
2735 
2736 	if (DO_INDEX(ip)) {
2737 		/*
2738 		 * persistent index is stored in directory entries.
2739 		 * Special cases:	 0 = .
2740 		 *			 1 = ..
2741 		 *			-1 = End of directory
2742 		 */
2743 		do_index = 1;
2744 
2745 		dir_index = (u32) ctx->pos;
2746 
2747 		/*
2748 		 * NFSv4 reserves cookies 1 and 2 for . and .. so the value
2749 		 * we return to the vfs is one greater than the one we use
2750 		 * internally.
2751 		 */
2752 		if (dir_index)
2753 			dir_index--;
2754 
2755 		if (dir_index > 1) {
2756 			struct dir_table_slot dirtab_slot;
2757 
2758 			if (dtEmpty(ip) ||
2759 			    (dir_index >= JFS_IP(ip)->next_index)) {
2760 				/* Stale position.  Directory has shrunk */
2761 				ctx->pos = DIREND;
2762 				return 0;
2763 			}
2764 		      repeat:
2765 			rc = read_index(ip, dir_index, &dirtab_slot);
2766 			if (rc) {
2767 				ctx->pos = DIREND;
2768 				return rc;
2769 			}
2770 			if (dirtab_slot.flag == DIR_INDEX_FREE) {
2771 				if (loop_count++ > JFS_IP(ip)->next_index) {
2772 					jfs_err("jfs_readdir detected infinite loop!");
2773 					ctx->pos = DIREND;
2774 					return 0;
2775 				}
2776 				dir_index = le32_to_cpu(dirtab_slot.addr2);
2777 				if (dir_index == -1) {
2778 					ctx->pos = DIREND;
2779 					return 0;
2780 				}
2781 				goto repeat;
2782 			}
2783 			bn = addressDTS(&dirtab_slot);
2784 			index = dirtab_slot.slot;
2785 			DT_GETPAGE(ip, bn, mp, PSIZE, p, rc);
2786 			if (rc) {
2787 				ctx->pos = DIREND;
2788 				return 0;
2789 			}
2790 			if (p->header.flag & BT_INTERNAL) {
2791 				jfs_err("jfs_readdir: bad index table");
2792 				DT_PUTPAGE(mp);
2793 				ctx->pos = DIREND;
2794 				return 0;
2795 			}
2796 		} else {
2797 			if (dir_index == 0) {
2798 				/*
2799 				 * self "."
2800 				 */
2801 				ctx->pos = 1;
2802 				if (!dir_emit(ctx, ".", 1, ip->i_ino, DT_DIR))
2803 					return 0;
2804 			}
2805 			/*
2806 			 * parent ".."
2807 			 */
2808 			ctx->pos = 2;
2809 			if (!dir_emit(ctx, "..", 2, PARENT(ip), DT_DIR))
2810 				return 0;
2811 
2812 			/*
2813 			 * Find first entry of left-most leaf
2814 			 */
2815 			if (dtEmpty(ip)) {
2816 				ctx->pos = DIREND;
2817 				return 0;
2818 			}
2819 
2820 			if ((rc = dtReadFirst(ip, &btstack)))
2821 				return rc;
2822 
2823 			DT_GETSEARCH(ip, btstack.top, bn, mp, p, index);
2824 		}
2825 	} else {
2826 		/*
2827 		 * Legacy filesystem - OS/2 & Linux JFS < 0.3.6
2828 		 *
2829 		 * pn = 0; index = 1:	First entry "."
2830 		 * pn = 0; index = 2:	Second entry ".."
2831 		 * pn > 0:		Real entries, pn=1 -> leftmost page
2832 		 * pn = index = -1:	No more entries
2833 		 */
2834 		dtpos = ctx->pos;
2835 		if (dtpos < 2) {
2836 			/* build "." entry */
2837 			ctx->pos = 1;
2838 			if (!dir_emit(ctx, ".", 1, ip->i_ino, DT_DIR))
2839 				return 0;
2840 			dtoffset->index = 2;
2841 			ctx->pos = dtpos;
2842 		}
2843 
2844 		if (dtoffset->pn == 0) {
2845 			if (dtoffset->index == 2) {
2846 				/* build ".." entry */
2847 				if (!dir_emit(ctx, "..", 2, PARENT(ip), DT_DIR))
2848 					return 0;
2849 			} else {
2850 				jfs_err("jfs_readdir called with invalid offset!");
2851 			}
2852 			dtoffset->pn = 1;
2853 			dtoffset->index = 0;
2854 			ctx->pos = dtpos;
2855 		}
2856 
2857 		if (dtEmpty(ip)) {
2858 			ctx->pos = DIREND;
2859 			return 0;
2860 		}
2861 
2862 		if ((rc = dtReadNext(ip, &ctx->pos, &btstack))) {
2863 			jfs_err("jfs_readdir: unexpected rc = %d from dtReadNext",
2864 				rc);
2865 			ctx->pos = DIREND;
2866 			return 0;
2867 		}
2868 		/* get start leaf page and index */
2869 		DT_GETSEARCH(ip, btstack.top, bn, mp, p, index);
2870 
2871 		/* offset beyond directory eof ? */
2872 		if (bn < 0) {
2873 			ctx->pos = DIREND;
2874 			return 0;
2875 		}
2876 	}
2877 
2878 	dirent_buf = __get_free_page(GFP_KERNEL);
2879 	if (dirent_buf == 0) {
2880 		DT_PUTPAGE(mp);
2881 		jfs_warn("jfs_readdir: __get_free_page failed!");
2882 		ctx->pos = DIREND;
2883 		return -ENOMEM;
2884 	}
2885 
2886 	while (1) {
2887 		jfs_dirent = (struct jfs_dirent *) dirent_buf;
2888 		jfs_dirents = 0;
2889 		overflow = fix_page = 0;
2890 
2891 		stbl = DT_GETSTBL(p);
2892 
2893 		for (i = index; i < p->header.nextindex; i++) {
2894 			d = (struct ldtentry *) & p->slot[stbl[i]];
2895 
2896 			if (((long) jfs_dirent + d->namlen + 1) >
2897 			    (dirent_buf + PAGE_SIZE)) {
2898 				/* DBCS codepages could overrun dirent_buf */
2899 				index = i;
2900 				overflow = 1;
2901 				break;
2902 			}
2903 
2904 			d_namleft = d->namlen;
2905 			name_ptr = jfs_dirent->name;
2906 			jfs_dirent->ino = le32_to_cpu(d->inumber);
2907 
2908 			if (do_index) {
2909 				len = min(d_namleft, DTLHDRDATALEN);
2910 				jfs_dirent->position = le32_to_cpu(d->index);
2911 				/*
2912 				 * d->index should always be valid, but it
2913 				 * isn't.  fsck.jfs doesn't create the
2914 				 * directory index for the lost+found
2915 				 * directory.  Rather than let it go,
2916 				 * we can try to fix it.
2917 				 */
2918 				if ((jfs_dirent->position < 2) ||
2919 				    (jfs_dirent->position >=
2920 				     JFS_IP(ip)->next_index)) {
2921 					if (!page_fixed && !isReadOnly(ip)) {
2922 						fix_page = 1;
2923 						/*
2924 						 * setting overflow and setting
2925 						 * index to i will cause the
2926 						 * same page to be processed
2927 						 * again starting here
2928 						 */
2929 						overflow = 1;
2930 						index = i;
2931 						break;
2932 					}
2933 					jfs_dirent->position = unique_pos++;
2934 				}
2935 				/*
2936 				 * We add 1 to the index because we may
2937 				 * use a value of 2 internally, and NFSv4
2938 				 * doesn't like that.
2939 				 */
2940 				jfs_dirent->position++;
2941 			} else {
2942 				jfs_dirent->position = dtpos;
2943 				len = min(d_namleft, DTLHDRDATALEN_LEGACY);
2944 			}
2945 
2946 			/* copy the name of head/only segment */
2947 			outlen = jfs_strfromUCS_le(name_ptr, d->name, len,
2948 						   codepage);
2949 			jfs_dirent->name_len = outlen;
2950 
2951 			/* copy name in the additional segment(s) */
2952 			next = d->next;
2953 			while (next >= 0) {
2954 				t = (struct dtslot *) & p->slot[next];
2955 				name_ptr += outlen;
2956 				d_namleft -= len;
2957 				/* Sanity Check */
2958 				if (d_namleft == 0) {
2959 					jfs_error(ip->i_sb,
2960 						  "JFS:Dtree error: ino = %ld, bn=%lld, index = %d\n",
2961 						  (long)ip->i_ino,
2962 						  (long long)bn,
2963 						  i);
2964 					goto skip_one;
2965 				}
2966 				len = min(d_namleft, DTSLOTDATALEN);
2967 				outlen = jfs_strfromUCS_le(name_ptr, t->name,
2968 							   len, codepage);
2969 				jfs_dirent->name_len += outlen;
2970 
2971 				next = t->next;
2972 			}
2973 
2974 			jfs_dirents++;
2975 			jfs_dirent = next_jfs_dirent(jfs_dirent);
2976 skip_one:
2977 			if (!do_index)
2978 				dtoffset->index++;
2979 		}
2980 
2981 		if (!overflow) {
2982 			/* Point to next leaf page */
2983 			if (p->header.flag & BT_ROOT)
2984 				bn = 0;
2985 			else {
2986 				bn = le64_to_cpu(p->header.next);
2987 				index = 0;
2988 				/* update offset (pn:index) for new page */
2989 				if (!do_index) {
2990 					dtoffset->pn++;
2991 					dtoffset->index = 0;
2992 				}
2993 			}
2994 			page_fixed = 0;
2995 		}
2996 
2997 		/* unpin previous leaf page */
2998 		DT_PUTPAGE(mp);
2999 
3000 		jfs_dirent = (struct jfs_dirent *) dirent_buf;
3001 		while (jfs_dirents--) {
3002 			ctx->pos = jfs_dirent->position;
3003 			if (!dir_emit(ctx, jfs_dirent->name,
3004 				    jfs_dirent->name_len,
3005 				    jfs_dirent->ino, DT_UNKNOWN))
3006 				goto out;
3007 			jfs_dirent = next_jfs_dirent(jfs_dirent);
3008 		}
3009 
3010 		if (fix_page) {
3011 			add_missing_indices(ip, bn);
3012 			page_fixed = 1;
3013 		}
3014 
3015 		if (!overflow && (bn == 0)) {
3016 			ctx->pos = DIREND;
3017 			break;
3018 		}
3019 
3020 		DT_GETPAGE(ip, bn, mp, PSIZE, p, rc);
3021 		if (rc) {
3022 			free_page(dirent_buf);
3023 			return rc;
3024 		}
3025 	}
3026 
3027       out:
3028 	free_page(dirent_buf);
3029 
3030 	return rc;
3031 }
3032 
3033 
3034 /*
3035  *	dtReadFirst()
3036  *
3037  * function: get the leftmost page of the directory
3038  */
dtReadFirst(struct inode * ip,struct btstack * btstack)3039 static int dtReadFirst(struct inode *ip, struct btstack * btstack)
3040 {
3041 	int rc = 0;
3042 	s64 bn;
3043 	int psize = 288;	/* initial in-line directory */
3044 	struct metapage *mp;
3045 	dtpage_t *p;
3046 	s8 *stbl;
3047 	struct btframe *btsp;
3048 	pxd_t *xd;
3049 
3050 	BT_CLR(btstack);	/* reset stack */
3051 
3052 	/*
3053 	 *	descend leftmost path of the tree
3054 	 *
3055 	 * by convention, root bn = 0.
3056 	 */
3057 	for (bn = 0;;) {
3058 		DT_GETPAGE(ip, bn, mp, psize, p, rc);
3059 		if (rc)
3060 			return rc;
3061 
3062 		/*
3063 		 * leftmost leaf page
3064 		 */
3065 		if (p->header.flag & BT_LEAF) {
3066 			/* return leftmost entry */
3067 			btsp = btstack->top;
3068 			btsp->bn = bn;
3069 			btsp->index = 0;
3070 			btsp->mp = mp;
3071 
3072 			return 0;
3073 		}
3074 
3075 		/*
3076 		 * descend down to leftmost child page
3077 		 */
3078 		if (BT_STACK_FULL(btstack)) {
3079 			DT_PUTPAGE(mp);
3080 			jfs_error(ip->i_sb, "btstack overrun\n");
3081 			BT_STACK_DUMP(btstack);
3082 			return -EIO;
3083 		}
3084 		/* push (bn, index) of the parent page/entry */
3085 		BT_PUSH(btstack, bn, 0);
3086 
3087 		/* get the leftmost entry */
3088 		stbl = DT_GETSTBL(p);
3089 		xd = (pxd_t *) & p->slot[stbl[0]];
3090 
3091 		/* get the child page block address */
3092 		bn = addressPXD(xd);
3093 		psize = lengthPXD(xd) << JFS_SBI(ip->i_sb)->l2bsize;
3094 
3095 		/* unpin the parent page */
3096 		DT_PUTPAGE(mp);
3097 	}
3098 }
3099 
3100 
3101 /*
3102  *	dtReadNext()
3103  *
3104  * function: get the page of the specified offset (pn:index)
3105  *
3106  * return: if (offset > eof), bn = -1;
3107  *
3108  * note: if index > nextindex of the target leaf page,
3109  * start with 1st entry of next leaf page;
3110  */
dtReadNext(struct inode * ip,loff_t * offset,struct btstack * btstack)3111 static int dtReadNext(struct inode *ip, loff_t * offset,
3112 		      struct btstack * btstack)
3113 {
3114 	int rc = 0;
3115 	struct dtoffset {
3116 		s16 pn;
3117 		s16 index;
3118 		s32 unused;
3119 	} *dtoffset = (struct dtoffset *) offset;
3120 	s64 bn;
3121 	struct metapage *mp;
3122 	dtpage_t *p;
3123 	int index;
3124 	int pn;
3125 	s8 *stbl;
3126 	struct btframe *btsp, *parent;
3127 	pxd_t *xd;
3128 
3129 	/*
3130 	 * get leftmost leaf page pinned
3131 	 */
3132 	if ((rc = dtReadFirst(ip, btstack)))
3133 		return rc;
3134 
3135 	/* get leaf page */
3136 	DT_GETSEARCH(ip, btstack->top, bn, mp, p, index);
3137 
3138 	/* get the start offset (pn:index) */
3139 	pn = dtoffset->pn - 1;	/* Now pn = 0 represents leftmost leaf */
3140 	index = dtoffset->index;
3141 
3142 	/* start at leftmost page ? */
3143 	if (pn == 0) {
3144 		/* offset beyond eof ? */
3145 		if (index < p->header.nextindex)
3146 			goto out;
3147 
3148 		if (p->header.flag & BT_ROOT) {
3149 			bn = -1;
3150 			goto out;
3151 		}
3152 
3153 		/* start with 1st entry of next leaf page */
3154 		dtoffset->pn++;
3155 		dtoffset->index = index = 0;
3156 		goto a;
3157 	}
3158 
3159 	/* start at non-leftmost page: scan parent pages for large pn */
3160 	if (p->header.flag & BT_ROOT) {
3161 		bn = -1;
3162 		goto out;
3163 	}
3164 
3165 	/* start after next leaf page ? */
3166 	if (pn > 1)
3167 		goto b;
3168 
3169 	/* get leaf page pn = 1 */
3170       a:
3171 	bn = le64_to_cpu(p->header.next);
3172 
3173 	/* unpin leaf page */
3174 	DT_PUTPAGE(mp);
3175 
3176 	/* offset beyond eof ? */
3177 	if (bn == 0) {
3178 		bn = -1;
3179 		goto out;
3180 	}
3181 
3182 	goto c;
3183 
3184 	/*
3185 	 * scan last internal page level to get target leaf page
3186 	 */
3187       b:
3188 	/* unpin leftmost leaf page */
3189 	DT_PUTPAGE(mp);
3190 
3191 	/* get left most parent page */
3192 	btsp = btstack->top;
3193 	parent = btsp - 1;
3194 	bn = parent->bn;
3195 	DT_GETPAGE(ip, bn, mp, PSIZE, p, rc);
3196 	if (rc)
3197 		return rc;
3198 
3199 	/* scan parent pages at last internal page level */
3200 	while (pn >= p->header.nextindex) {
3201 		pn -= p->header.nextindex;
3202 
3203 		/* get next parent page address */
3204 		bn = le64_to_cpu(p->header.next);
3205 
3206 		/* unpin current parent page */
3207 		DT_PUTPAGE(mp);
3208 
3209 		/* offset beyond eof ? */
3210 		if (bn == 0) {
3211 			bn = -1;
3212 			goto out;
3213 		}
3214 
3215 		/* get next parent page */
3216 		DT_GETPAGE(ip, bn, mp, PSIZE, p, rc);
3217 		if (rc)
3218 			return rc;
3219 
3220 		/* update parent page stack frame */
3221 		parent->bn = bn;
3222 	}
3223 
3224 	/* get leaf page address */
3225 	stbl = DT_GETSTBL(p);
3226 	xd = (pxd_t *) & p->slot[stbl[pn]];
3227 	bn = addressPXD(xd);
3228 
3229 	/* unpin parent page */
3230 	DT_PUTPAGE(mp);
3231 
3232 	/*
3233 	 * get target leaf page
3234 	 */
3235       c:
3236 	DT_GETPAGE(ip, bn, mp, PSIZE, p, rc);
3237 	if (rc)
3238 		return rc;
3239 
3240 	/*
3241 	 * leaf page has been completed:
3242 	 * start with 1st entry of next leaf page
3243 	 */
3244 	if (index >= p->header.nextindex) {
3245 		bn = le64_to_cpu(p->header.next);
3246 
3247 		/* unpin leaf page */
3248 		DT_PUTPAGE(mp);
3249 
3250 		/* offset beyond eof ? */
3251 		if (bn == 0) {
3252 			bn = -1;
3253 			goto out;
3254 		}
3255 
3256 		/* get next leaf page */
3257 		DT_GETPAGE(ip, bn, mp, PSIZE, p, rc);
3258 		if (rc)
3259 			return rc;
3260 
3261 		/* start with 1st entry of next leaf page */
3262 		dtoffset->pn++;
3263 		dtoffset->index = 0;
3264 	}
3265 
3266       out:
3267 	/* return target leaf page pinned */
3268 	btsp = btstack->top;
3269 	btsp->bn = bn;
3270 	btsp->index = dtoffset->index;
3271 	btsp->mp = mp;
3272 
3273 	return 0;
3274 }
3275 
3276 
3277 /*
3278  *	dtCompare()
3279  *
3280  * function: compare search key with an internal entry
3281  *
3282  * return:
3283  *	< 0 if k is < record
3284  *	= 0 if k is = record
3285  *	> 0 if k is > record
3286  */
dtCompare(struct component_name * key,dtpage_t * p,int si)3287 static int dtCompare(struct component_name * key,	/* search key */
3288 		     dtpage_t * p,	/* directory page */
3289 		     int si)
3290 {				/* entry slot index */
3291 	wchar_t *kname;
3292 	__le16 *name;
3293 	int klen, namlen, len, rc;
3294 	struct idtentry *ih;
3295 	struct dtslot *t;
3296 
3297 	/*
3298 	 * force the left-most key on internal pages, at any level of
3299 	 * the tree, to be less than any search key.
3300 	 * this obviates having to update the leftmost key on an internal
3301 	 * page when the user inserts a new key in the tree smaller than
3302 	 * anything that has been stored.
3303 	 *
3304 	 * (? if/when dtSearch() narrows down to 1st entry (index = 0),
3305 	 * at any internal page at any level of the tree,
3306 	 * it descends to child of the entry anyway -
3307 	 * ? make the entry as min size dummy entry)
3308 	 *
3309 	 * if (e->index == 0 && h->prevpg == P_INVALID && !(h->flags & BT_LEAF))
3310 	 * return (1);
3311 	 */
3312 
3313 	kname = key->name;
3314 	klen = key->namlen;
3315 
3316 	ih = (struct idtentry *) & p->slot[si];
3317 	si = ih->next;
3318 	name = ih->name;
3319 	namlen = ih->namlen;
3320 	len = min(namlen, DTIHDRDATALEN);
3321 
3322 	/* compare with head/only segment */
3323 	len = min(klen, len);
3324 	if ((rc = UniStrncmp_le(kname, name, len)))
3325 		return rc;
3326 
3327 	klen -= len;
3328 	namlen -= len;
3329 
3330 	/* compare with additional segment(s) */
3331 	kname += len;
3332 	while (klen > 0 && namlen > 0) {
3333 		/* compare with next name segment */
3334 		t = (struct dtslot *) & p->slot[si];
3335 		len = min(namlen, DTSLOTDATALEN);
3336 		len = min(klen, len);
3337 		name = t->name;
3338 		if ((rc = UniStrncmp_le(kname, name, len)))
3339 			return rc;
3340 
3341 		klen -= len;
3342 		namlen -= len;
3343 		kname += len;
3344 		si = t->next;
3345 	}
3346 
3347 	return (klen - namlen);
3348 }
3349 
3350 
3351 
3352 
3353 /*
3354  *	ciCompare()
3355  *
3356  * function: compare search key with an (leaf/internal) entry
3357  *
3358  * return:
3359  *	< 0 if k is < record
3360  *	= 0 if k is = record
3361  *	> 0 if k is > record
3362  */
ciCompare(struct component_name * key,dtpage_t * p,int si,int flag)3363 static int ciCompare(struct component_name * key,	/* search key */
3364 		     dtpage_t * p,	/* directory page */
3365 		     int si,	/* entry slot index */
3366 		     int flag)
3367 {
3368 	wchar_t *kname, x;
3369 	__le16 *name;
3370 	int klen, namlen, len, rc;
3371 	struct ldtentry *lh;
3372 	struct idtentry *ih;
3373 	struct dtslot *t;
3374 	int i;
3375 
3376 	/*
3377 	 * force the left-most key on internal pages, at any level of
3378 	 * the tree, to be less than any search key.
3379 	 * this obviates having to update the leftmost key on an internal
3380 	 * page when the user inserts a new key in the tree smaller than
3381 	 * anything that has been stored.
3382 	 *
3383 	 * (? if/when dtSearch() narrows down to 1st entry (index = 0),
3384 	 * at any internal page at any level of the tree,
3385 	 * it descends to child of the entry anyway -
3386 	 * ? make the entry as min size dummy entry)
3387 	 *
3388 	 * if (e->index == 0 && h->prevpg == P_INVALID && !(h->flags & BT_LEAF))
3389 	 * return (1);
3390 	 */
3391 
3392 	kname = key->name;
3393 	klen = key->namlen;
3394 
3395 	/*
3396 	 * leaf page entry
3397 	 */
3398 	if (p->header.flag & BT_LEAF) {
3399 		lh = (struct ldtentry *) & p->slot[si];
3400 		si = lh->next;
3401 		name = lh->name;
3402 		namlen = lh->namlen;
3403 		if (flag & JFS_DIR_INDEX)
3404 			len = min(namlen, DTLHDRDATALEN);
3405 		else
3406 			len = min(namlen, DTLHDRDATALEN_LEGACY);
3407 	}
3408 	/*
3409 	 * internal page entry
3410 	 */
3411 	else {
3412 		ih = (struct idtentry *) & p->slot[si];
3413 		si = ih->next;
3414 		name = ih->name;
3415 		namlen = ih->namlen;
3416 		len = min(namlen, DTIHDRDATALEN);
3417 	}
3418 
3419 	/* compare with head/only segment */
3420 	len = min(klen, len);
3421 	for (i = 0; i < len; i++, kname++, name++) {
3422 		/* only uppercase if case-insensitive support is on */
3423 		if ((flag & JFS_OS2) == JFS_OS2)
3424 			x = UniToupper(le16_to_cpu(*name));
3425 		else
3426 			x = le16_to_cpu(*name);
3427 		if ((rc = *kname - x))
3428 			return rc;
3429 	}
3430 
3431 	klen -= len;
3432 	namlen -= len;
3433 
3434 	/* compare with additional segment(s) */
3435 	while (klen > 0 && namlen > 0) {
3436 		/* compare with next name segment */
3437 		t = (struct dtslot *) & p->slot[si];
3438 		len = min(namlen, DTSLOTDATALEN);
3439 		len = min(klen, len);
3440 		name = t->name;
3441 		for (i = 0; i < len; i++, kname++, name++) {
3442 			/* only uppercase if case-insensitive support is on */
3443 			if ((flag & JFS_OS2) == JFS_OS2)
3444 				x = UniToupper(le16_to_cpu(*name));
3445 			else
3446 				x = le16_to_cpu(*name);
3447 
3448 			if ((rc = *kname - x))
3449 				return rc;
3450 		}
3451 
3452 		klen -= len;
3453 		namlen -= len;
3454 		si = t->next;
3455 	}
3456 
3457 	return (klen - namlen);
3458 }
3459 
3460 
3461 /*
3462  *	ciGetLeafPrefixKey()
3463  *
3464  * function: compute prefix of suffix compression
3465  *	     from two adjacent leaf entries
3466  *	     across page boundary
3467  *
3468  * return: non-zero on error
3469  *
3470  */
ciGetLeafPrefixKey(dtpage_t * lp,int li,dtpage_t * rp,int ri,struct component_name * key,int flag)3471 static int ciGetLeafPrefixKey(dtpage_t * lp, int li, dtpage_t * rp,
3472 			       int ri, struct component_name * key, int flag)
3473 {
3474 	int klen, namlen;
3475 	wchar_t *pl, *pr, *kname;
3476 	struct component_name lkey;
3477 	struct component_name rkey;
3478 
3479 	lkey.name = kmalloc_array(JFS_NAME_MAX + 1, sizeof(wchar_t),
3480 					GFP_KERNEL);
3481 	if (lkey.name == NULL)
3482 		return -ENOMEM;
3483 
3484 	rkey.name = kmalloc_array(JFS_NAME_MAX + 1, sizeof(wchar_t),
3485 					GFP_KERNEL);
3486 	if (rkey.name == NULL) {
3487 		kfree(lkey.name);
3488 		return -ENOMEM;
3489 	}
3490 
3491 	/* get left and right key */
3492 	dtGetKey(lp, li, &lkey, flag);
3493 	lkey.name[lkey.namlen] = 0;
3494 
3495 	if ((flag & JFS_OS2) == JFS_OS2)
3496 		ciToUpper(&lkey);
3497 
3498 	dtGetKey(rp, ri, &rkey, flag);
3499 	rkey.name[rkey.namlen] = 0;
3500 
3501 
3502 	if ((flag & JFS_OS2) == JFS_OS2)
3503 		ciToUpper(&rkey);
3504 
3505 	/* compute prefix */
3506 	klen = 0;
3507 	kname = key->name;
3508 	namlen = min(lkey.namlen, rkey.namlen);
3509 	for (pl = lkey.name, pr = rkey.name;
3510 	     namlen; pl++, pr++, namlen--, klen++, kname++) {
3511 		*kname = *pr;
3512 		if (*pl != *pr) {
3513 			key->namlen = klen + 1;
3514 			goto free_names;
3515 		}
3516 	}
3517 
3518 	/* l->namlen <= r->namlen since l <= r */
3519 	if (lkey.namlen < rkey.namlen) {
3520 		*kname = *pr;
3521 		key->namlen = klen + 1;
3522 	} else			/* l->namelen == r->namelen */
3523 		key->namlen = klen;
3524 
3525 free_names:
3526 	kfree(lkey.name);
3527 	kfree(rkey.name);
3528 	return 0;
3529 }
3530 
3531 
3532 
3533 /*
3534  *	dtGetKey()
3535  *
3536  * function: get key of the entry
3537  */
dtGetKey(dtpage_t * p,int i,struct component_name * key,int flag)3538 static void dtGetKey(dtpage_t * p, int i,	/* entry index */
3539 		     struct component_name * key, int flag)
3540 {
3541 	int si;
3542 	s8 *stbl;
3543 	struct ldtentry *lh;
3544 	struct idtentry *ih;
3545 	struct dtslot *t;
3546 	int namlen, len;
3547 	wchar_t *kname;
3548 	__le16 *name;
3549 
3550 	/* get entry */
3551 	stbl = DT_GETSTBL(p);
3552 	si = stbl[i];
3553 	if (p->header.flag & BT_LEAF) {
3554 		lh = (struct ldtentry *) & p->slot[si];
3555 		si = lh->next;
3556 		namlen = lh->namlen;
3557 		name = lh->name;
3558 		if (flag & JFS_DIR_INDEX)
3559 			len = min(namlen, DTLHDRDATALEN);
3560 		else
3561 			len = min(namlen, DTLHDRDATALEN_LEGACY);
3562 	} else {
3563 		ih = (struct idtentry *) & p->slot[si];
3564 		si = ih->next;
3565 		namlen = ih->namlen;
3566 		name = ih->name;
3567 		len = min(namlen, DTIHDRDATALEN);
3568 	}
3569 
3570 	key->namlen = namlen;
3571 	kname = key->name;
3572 
3573 	/*
3574 	 * move head/only segment
3575 	 */
3576 	UniStrncpy_from_le(kname, name, len);
3577 
3578 	/*
3579 	 * move additional segment(s)
3580 	 */
3581 	while (si >= 0) {
3582 		/* get next segment */
3583 		t = &p->slot[si];
3584 		kname += len;
3585 		namlen -= len;
3586 		len = min(namlen, DTSLOTDATALEN);
3587 		UniStrncpy_from_le(kname, t->name, len);
3588 
3589 		si = t->next;
3590 	}
3591 }
3592 
3593 
3594 /*
3595  *	dtInsertEntry()
3596  *
3597  * function: allocate free slot(s) and
3598  *	     write a leaf/internal entry
3599  *
3600  * return: entry slot index
3601  */
dtInsertEntry(dtpage_t * p,int index,struct component_name * key,ddata_t * data,struct dt_lock ** dtlock)3602 static void dtInsertEntry(dtpage_t * p, int index, struct component_name * key,
3603 			  ddata_t * data, struct dt_lock ** dtlock)
3604 {
3605 	struct dtslot *h, *t;
3606 	struct ldtentry *lh = NULL;
3607 	struct idtentry *ih = NULL;
3608 	int hsi, fsi, klen, len, nextindex;
3609 	wchar_t *kname;
3610 	__le16 *name;
3611 	s8 *stbl;
3612 	pxd_t *xd;
3613 	struct dt_lock *dtlck = *dtlock;
3614 	struct lv *lv;
3615 	int xsi, n;
3616 	s64 bn = 0;
3617 	struct metapage *mp = NULL;
3618 
3619 	klen = key->namlen;
3620 	kname = key->name;
3621 
3622 	/* allocate a free slot */
3623 	hsi = fsi = p->header.freelist;
3624 	h = &p->slot[fsi];
3625 	p->header.freelist = h->next;
3626 	--p->header.freecnt;
3627 
3628 	/* open new linelock */
3629 	if (dtlck->index >= dtlck->maxcnt)
3630 		dtlck = (struct dt_lock *) txLinelock(dtlck);
3631 
3632 	lv = & dtlck->lv[dtlck->index];
3633 	lv->offset = hsi;
3634 
3635 	/* write head/only segment */
3636 	if (p->header.flag & BT_LEAF) {
3637 		lh = (struct ldtentry *) h;
3638 		lh->next = h->next;
3639 		lh->inumber = cpu_to_le32(data->leaf.ino);
3640 		lh->namlen = klen;
3641 		name = lh->name;
3642 		if (data->leaf.ip) {
3643 			len = min(klen, DTLHDRDATALEN);
3644 			if (!(p->header.flag & BT_ROOT))
3645 				bn = addressPXD(&p->header.self);
3646 			lh->index = cpu_to_le32(add_index(data->leaf.tid,
3647 							  data->leaf.ip,
3648 							  bn, index));
3649 		} else
3650 			len = min(klen, DTLHDRDATALEN_LEGACY);
3651 	} else {
3652 		ih = (struct idtentry *) h;
3653 		ih->next = h->next;
3654 		xd = (pxd_t *) ih;
3655 		*xd = data->xd;
3656 		ih->namlen = klen;
3657 		name = ih->name;
3658 		len = min(klen, DTIHDRDATALEN);
3659 	}
3660 
3661 	UniStrncpy_to_le(name, kname, len);
3662 
3663 	n = 1;
3664 	xsi = hsi;
3665 
3666 	/* write additional segment(s) */
3667 	t = h;
3668 	klen -= len;
3669 	while (klen) {
3670 		/* get free slot */
3671 		fsi = p->header.freelist;
3672 		t = &p->slot[fsi];
3673 		p->header.freelist = t->next;
3674 		--p->header.freecnt;
3675 
3676 		/* is next slot contiguous ? */
3677 		if (fsi != xsi + 1) {
3678 			/* close current linelock */
3679 			lv->length = n;
3680 			dtlck->index++;
3681 
3682 			/* open new linelock */
3683 			if (dtlck->index < dtlck->maxcnt)
3684 				lv++;
3685 			else {
3686 				dtlck = (struct dt_lock *) txLinelock(dtlck);
3687 				lv = & dtlck->lv[0];
3688 			}
3689 
3690 			lv->offset = fsi;
3691 			n = 0;
3692 		}
3693 
3694 		kname += len;
3695 		len = min(klen, DTSLOTDATALEN);
3696 		UniStrncpy_to_le(t->name, kname, len);
3697 
3698 		n++;
3699 		xsi = fsi;
3700 		klen -= len;
3701 	}
3702 
3703 	/* close current linelock */
3704 	lv->length = n;
3705 	dtlck->index++;
3706 
3707 	*dtlock = dtlck;
3708 
3709 	/* terminate last/only segment */
3710 	if (h == t) {
3711 		/* single segment entry */
3712 		if (p->header.flag & BT_LEAF)
3713 			lh->next = -1;
3714 		else
3715 			ih->next = -1;
3716 	} else
3717 		/* multi-segment entry */
3718 		t->next = -1;
3719 
3720 	/* if insert into middle, shift right succeeding entries in stbl */
3721 	stbl = DT_GETSTBL(p);
3722 	nextindex = p->header.nextindex;
3723 	if (index < nextindex) {
3724 		memmove(stbl + index + 1, stbl + index, nextindex - index);
3725 
3726 		if ((p->header.flag & BT_LEAF) && data->leaf.ip) {
3727 			s64 lblock;
3728 
3729 			/*
3730 			 * Need to update slot number for entries that moved
3731 			 * in the stbl
3732 			 */
3733 			mp = NULL;
3734 			for (n = index + 1; n <= nextindex; n++) {
3735 				lh = (struct ldtentry *) & (p->slot[stbl[n]]);
3736 				modify_index(data->leaf.tid, data->leaf.ip,
3737 					     le32_to_cpu(lh->index), bn, n,
3738 					     &mp, &lblock);
3739 			}
3740 			if (mp)
3741 				release_metapage(mp);
3742 		}
3743 	}
3744 
3745 	stbl[index] = hsi;
3746 
3747 	/* advance next available entry index of stbl */
3748 	++p->header.nextindex;
3749 }
3750 
3751 
3752 /*
3753  *	dtMoveEntry()
3754  *
3755  * function: move entries from split/left page to new/right page
3756  *
3757  *	nextindex of dst page and freelist/freecnt of both pages
3758  *	are updated.
3759  */
dtMoveEntry(dtpage_t * sp,int si,dtpage_t * dp,struct dt_lock ** sdtlock,struct dt_lock ** ddtlock,int do_index)3760 static void dtMoveEntry(dtpage_t * sp, int si, dtpage_t * dp,
3761 			struct dt_lock ** sdtlock, struct dt_lock ** ddtlock,
3762 			int do_index)
3763 {
3764 	int ssi, next;		/* src slot index */
3765 	int di;			/* dst entry index */
3766 	int dsi;		/* dst slot index */
3767 	s8 *sstbl, *dstbl;	/* sorted entry table */
3768 	int snamlen, len;
3769 	struct ldtentry *slh, *dlh = NULL;
3770 	struct idtentry *sih, *dih = NULL;
3771 	struct dtslot *h, *s, *d;
3772 	struct dt_lock *sdtlck = *sdtlock, *ddtlck = *ddtlock;
3773 	struct lv *slv, *dlv;
3774 	int xssi, ns, nd;
3775 	int sfsi;
3776 
3777 	sstbl = (s8 *) & sp->slot[sp->header.stblindex];
3778 	dstbl = (s8 *) & dp->slot[dp->header.stblindex];
3779 
3780 	dsi = dp->header.freelist;	/* first (whole page) free slot */
3781 	sfsi = sp->header.freelist;
3782 
3783 	/* linelock destination entry slot */
3784 	dlv = & ddtlck->lv[ddtlck->index];
3785 	dlv->offset = dsi;
3786 
3787 	/* linelock source entry slot */
3788 	slv = & sdtlck->lv[sdtlck->index];
3789 	slv->offset = sstbl[si];
3790 	xssi = slv->offset - 1;
3791 
3792 	/*
3793 	 * move entries
3794 	 */
3795 	ns = nd = 0;
3796 	for (di = 0; si < sp->header.nextindex; si++, di++) {
3797 		ssi = sstbl[si];
3798 		dstbl[di] = dsi;
3799 
3800 		/* is next slot contiguous ? */
3801 		if (ssi != xssi + 1) {
3802 			/* close current linelock */
3803 			slv->length = ns;
3804 			sdtlck->index++;
3805 
3806 			/* open new linelock */
3807 			if (sdtlck->index < sdtlck->maxcnt)
3808 				slv++;
3809 			else {
3810 				sdtlck = (struct dt_lock *) txLinelock(sdtlck);
3811 				slv = & sdtlck->lv[0];
3812 			}
3813 
3814 			slv->offset = ssi;
3815 			ns = 0;
3816 		}
3817 
3818 		/*
3819 		 * move head/only segment of an entry
3820 		 */
3821 		/* get dst slot */
3822 		h = d = &dp->slot[dsi];
3823 
3824 		/* get src slot and move */
3825 		s = &sp->slot[ssi];
3826 		if (sp->header.flag & BT_LEAF) {
3827 			/* get source entry */
3828 			slh = (struct ldtentry *) s;
3829 			dlh = (struct ldtentry *) h;
3830 			snamlen = slh->namlen;
3831 
3832 			if (do_index) {
3833 				len = min(snamlen, DTLHDRDATALEN);
3834 				dlh->index = slh->index; /* little-endian */
3835 			} else
3836 				len = min(snamlen, DTLHDRDATALEN_LEGACY);
3837 
3838 			memcpy(dlh, slh, 6 + len * 2);
3839 
3840 			next = slh->next;
3841 
3842 			/* update dst head/only segment next field */
3843 			dsi++;
3844 			dlh->next = dsi;
3845 		} else {
3846 			sih = (struct idtentry *) s;
3847 			snamlen = sih->namlen;
3848 
3849 			len = min(snamlen, DTIHDRDATALEN);
3850 			dih = (struct idtentry *) h;
3851 			memcpy(dih, sih, 10 + len * 2);
3852 			next = sih->next;
3853 
3854 			dsi++;
3855 			dih->next = dsi;
3856 		}
3857 
3858 		/* free src head/only segment */
3859 		s->next = sfsi;
3860 		s->cnt = 1;
3861 		sfsi = ssi;
3862 
3863 		ns++;
3864 		nd++;
3865 		xssi = ssi;
3866 
3867 		/*
3868 		 * move additional segment(s) of the entry
3869 		 */
3870 		snamlen -= len;
3871 		while ((ssi = next) >= 0) {
3872 			/* is next slot contiguous ? */
3873 			if (ssi != xssi + 1) {
3874 				/* close current linelock */
3875 				slv->length = ns;
3876 				sdtlck->index++;
3877 
3878 				/* open new linelock */
3879 				if (sdtlck->index < sdtlck->maxcnt)
3880 					slv++;
3881 				else {
3882 					sdtlck =
3883 					    (struct dt_lock *)
3884 					    txLinelock(sdtlck);
3885 					slv = & sdtlck->lv[0];
3886 				}
3887 
3888 				slv->offset = ssi;
3889 				ns = 0;
3890 			}
3891 
3892 			/* get next source segment */
3893 			s = &sp->slot[ssi];
3894 
3895 			/* get next destination free slot */
3896 			d++;
3897 
3898 			len = min(snamlen, DTSLOTDATALEN);
3899 			UniStrncpy_le(d->name, s->name, len);
3900 
3901 			ns++;
3902 			nd++;
3903 			xssi = ssi;
3904 
3905 			dsi++;
3906 			d->next = dsi;
3907 
3908 			/* free source segment */
3909 			next = s->next;
3910 			s->next = sfsi;
3911 			s->cnt = 1;
3912 			sfsi = ssi;
3913 
3914 			snamlen -= len;
3915 		}		/* end while */
3916 
3917 		/* terminate dst last/only segment */
3918 		if (h == d) {
3919 			/* single segment entry */
3920 			if (dp->header.flag & BT_LEAF)
3921 				dlh->next = -1;
3922 			else
3923 				dih->next = -1;
3924 		} else
3925 			/* multi-segment entry */
3926 			d->next = -1;
3927 	}			/* end for */
3928 
3929 	/* close current linelock */
3930 	slv->length = ns;
3931 	sdtlck->index++;
3932 	*sdtlock = sdtlck;
3933 
3934 	dlv->length = nd;
3935 	ddtlck->index++;
3936 	*ddtlock = ddtlck;
3937 
3938 	/* update source header */
3939 	sp->header.freelist = sfsi;
3940 	sp->header.freecnt += nd;
3941 
3942 	/* update destination header */
3943 	dp->header.nextindex = di;
3944 
3945 	dp->header.freelist = dsi;
3946 	dp->header.freecnt -= nd;
3947 }
3948 
3949 
3950 /*
3951  *	dtDeleteEntry()
3952  *
3953  * function: free a (leaf/internal) entry
3954  *
3955  * log freelist header, stbl, and each segment slot of entry
3956  * (even though last/only segment next field is modified,
3957  * physical image logging requires all segment slots of
3958  * the entry logged to avoid applying previous updates
3959  * to the same slots)
3960  */
dtDeleteEntry(dtpage_t * p,int fi,struct dt_lock ** dtlock)3961 static void dtDeleteEntry(dtpage_t * p, int fi, struct dt_lock ** dtlock)
3962 {
3963 	int fsi;		/* free entry slot index */
3964 	s8 *stbl;
3965 	struct dtslot *t;
3966 	int si, freecnt;
3967 	struct dt_lock *dtlck = *dtlock;
3968 	struct lv *lv;
3969 	int xsi, n;
3970 
3971 	/* get free entry slot index */
3972 	stbl = DT_GETSTBL(p);
3973 	fsi = stbl[fi];
3974 
3975 	/* open new linelock */
3976 	if (dtlck->index >= dtlck->maxcnt)
3977 		dtlck = (struct dt_lock *) txLinelock(dtlck);
3978 	lv = & dtlck->lv[dtlck->index];
3979 
3980 	lv->offset = fsi;
3981 
3982 	/* get the head/only segment */
3983 	t = &p->slot[fsi];
3984 	if (p->header.flag & BT_LEAF)
3985 		si = ((struct ldtentry *) t)->next;
3986 	else
3987 		si = ((struct idtentry *) t)->next;
3988 	t->next = si;
3989 	t->cnt = 1;
3990 
3991 	n = freecnt = 1;
3992 	xsi = fsi;
3993 
3994 	/* find the last/only segment */
3995 	while (si >= 0) {
3996 		/* is next slot contiguous ? */
3997 		if (si != xsi + 1) {
3998 			/* close current linelock */
3999 			lv->length = n;
4000 			dtlck->index++;
4001 
4002 			/* open new linelock */
4003 			if (dtlck->index < dtlck->maxcnt)
4004 				lv++;
4005 			else {
4006 				dtlck = (struct dt_lock *) txLinelock(dtlck);
4007 				lv = & dtlck->lv[0];
4008 			}
4009 
4010 			lv->offset = si;
4011 			n = 0;
4012 		}
4013 
4014 		n++;
4015 		xsi = si;
4016 		freecnt++;
4017 
4018 		t = &p->slot[si];
4019 		t->cnt = 1;
4020 		si = t->next;
4021 	}
4022 
4023 	/* close current linelock */
4024 	lv->length = n;
4025 	dtlck->index++;
4026 
4027 	*dtlock = dtlck;
4028 
4029 	/* update freelist */
4030 	t->next = p->header.freelist;
4031 	p->header.freelist = fsi;
4032 	p->header.freecnt += freecnt;
4033 
4034 	/* if delete from middle,
4035 	 * shift left the succedding entries in the stbl
4036 	 */
4037 	si = p->header.nextindex;
4038 	if (fi < si - 1)
4039 		memmove(&stbl[fi], &stbl[fi + 1], si - fi - 1);
4040 
4041 	p->header.nextindex--;
4042 }
4043 
4044 
4045 /*
4046  *	dtTruncateEntry()
4047  *
4048  * function: truncate a (leaf/internal) entry
4049  *
4050  * log freelist header, stbl, and each segment slot of entry
4051  * (even though last/only segment next field is modified,
4052  * physical image logging requires all segment slots of
4053  * the entry logged to avoid applying previous updates
4054  * to the same slots)
4055  */
dtTruncateEntry(dtpage_t * p,int ti,struct dt_lock ** dtlock)4056 static void dtTruncateEntry(dtpage_t * p, int ti, struct dt_lock ** dtlock)
4057 {
4058 	int tsi;		/* truncate entry slot index */
4059 	s8 *stbl;
4060 	struct dtslot *t;
4061 	int si, freecnt;
4062 	struct dt_lock *dtlck = *dtlock;
4063 	struct lv *lv;
4064 	int fsi, xsi, n;
4065 
4066 	/* get free entry slot index */
4067 	stbl = DT_GETSTBL(p);
4068 	tsi = stbl[ti];
4069 
4070 	/* open new linelock */
4071 	if (dtlck->index >= dtlck->maxcnt)
4072 		dtlck = (struct dt_lock *) txLinelock(dtlck);
4073 	lv = & dtlck->lv[dtlck->index];
4074 
4075 	lv->offset = tsi;
4076 
4077 	/* get the head/only segment */
4078 	t = &p->slot[tsi];
4079 	ASSERT(p->header.flag & BT_INTERNAL);
4080 	((struct idtentry *) t)->namlen = 0;
4081 	si = ((struct idtentry *) t)->next;
4082 	((struct idtentry *) t)->next = -1;
4083 
4084 	n = 1;
4085 	freecnt = 0;
4086 	fsi = si;
4087 	xsi = tsi;
4088 
4089 	/* find the last/only segment */
4090 	while (si >= 0) {
4091 		/* is next slot contiguous ? */
4092 		if (si != xsi + 1) {
4093 			/* close current linelock */
4094 			lv->length = n;
4095 			dtlck->index++;
4096 
4097 			/* open new linelock */
4098 			if (dtlck->index < dtlck->maxcnt)
4099 				lv++;
4100 			else {
4101 				dtlck = (struct dt_lock *) txLinelock(dtlck);
4102 				lv = & dtlck->lv[0];
4103 			}
4104 
4105 			lv->offset = si;
4106 			n = 0;
4107 		}
4108 
4109 		n++;
4110 		xsi = si;
4111 		freecnt++;
4112 
4113 		t = &p->slot[si];
4114 		t->cnt = 1;
4115 		si = t->next;
4116 	}
4117 
4118 	/* close current linelock */
4119 	lv->length = n;
4120 	dtlck->index++;
4121 
4122 	*dtlock = dtlck;
4123 
4124 	/* update freelist */
4125 	if (freecnt == 0)
4126 		return;
4127 	t->next = p->header.freelist;
4128 	p->header.freelist = fsi;
4129 	p->header.freecnt += freecnt;
4130 }
4131 
4132 
4133 /*
4134  *	dtLinelockFreelist()
4135  */
dtLinelockFreelist(dtpage_t * p,int m,struct dt_lock ** dtlock)4136 static void dtLinelockFreelist(dtpage_t * p,	/* directory page */
4137 			       int m,	/* max slot index */
4138 			       struct dt_lock ** dtlock)
4139 {
4140 	int fsi;		/* free entry slot index */
4141 	struct dtslot *t;
4142 	int si;
4143 	struct dt_lock *dtlck = *dtlock;
4144 	struct lv *lv;
4145 	int xsi, n;
4146 
4147 	/* get free entry slot index */
4148 	fsi = p->header.freelist;
4149 
4150 	/* open new linelock */
4151 	if (dtlck->index >= dtlck->maxcnt)
4152 		dtlck = (struct dt_lock *) txLinelock(dtlck);
4153 	lv = & dtlck->lv[dtlck->index];
4154 
4155 	lv->offset = fsi;
4156 
4157 	n = 1;
4158 	xsi = fsi;
4159 
4160 	t = &p->slot[fsi];
4161 	si = t->next;
4162 
4163 	/* find the last/only segment */
4164 	while (si < m && si >= 0) {
4165 		/* is next slot contiguous ? */
4166 		if (si != xsi + 1) {
4167 			/* close current linelock */
4168 			lv->length = n;
4169 			dtlck->index++;
4170 
4171 			/* open new linelock */
4172 			if (dtlck->index < dtlck->maxcnt)
4173 				lv++;
4174 			else {
4175 				dtlck = (struct dt_lock *) txLinelock(dtlck);
4176 				lv = & dtlck->lv[0];
4177 			}
4178 
4179 			lv->offset = si;
4180 			n = 0;
4181 		}
4182 
4183 		n++;
4184 		xsi = si;
4185 
4186 		t = &p->slot[si];
4187 		si = t->next;
4188 	}
4189 
4190 	/* close current linelock */
4191 	lv->length = n;
4192 	dtlck->index++;
4193 
4194 	*dtlock = dtlck;
4195 }
4196 
4197 
4198 /*
4199  * NAME: dtModify
4200  *
4201  * FUNCTION: Modify the inode number part of a directory entry
4202  *
4203  * PARAMETERS:
4204  *	tid	- Transaction id
4205  *	ip	- Inode of parent directory
4206  *	key	- Name of entry to be modified
4207  *	orig_ino	- Original inode number expected in entry
4208  *	new_ino	- New inode number to put into entry
4209  *	flag	- JFS_RENAME
4210  *
4211  * RETURNS:
4212  *	-ESTALE	- If entry found does not match orig_ino passed in
4213  *	-ENOENT	- If no entry can be found to match key
4214  *	0	- If successfully modified entry
4215  */
dtModify(tid_t tid,struct inode * ip,struct component_name * key,ino_t * orig_ino,ino_t new_ino,int flag)4216 int dtModify(tid_t tid, struct inode *ip,
4217 	 struct component_name * key, ino_t * orig_ino, ino_t new_ino, int flag)
4218 {
4219 	int rc;
4220 	s64 bn;
4221 	struct metapage *mp;
4222 	dtpage_t *p;
4223 	int index;
4224 	struct btstack btstack;
4225 	struct tlock *tlck;
4226 	struct dt_lock *dtlck;
4227 	struct lv *lv;
4228 	s8 *stbl;
4229 	int entry_si;		/* entry slot index */
4230 	struct ldtentry *entry;
4231 
4232 	/*
4233 	 *	search for the entry to modify:
4234 	 *
4235 	 * dtSearch() returns (leaf page pinned, index at which to modify).
4236 	 */
4237 	if ((rc = dtSearch(ip, key, orig_ino, &btstack, flag)))
4238 		return rc;
4239 
4240 	/* retrieve search result */
4241 	DT_GETSEARCH(ip, btstack.top, bn, mp, p, index);
4242 
4243 	BT_MARK_DIRTY(mp, ip);
4244 	/*
4245 	 * acquire a transaction lock on the leaf page of named entry
4246 	 */
4247 	tlck = txLock(tid, ip, mp, tlckDTREE | tlckENTRY);
4248 	dtlck = (struct dt_lock *) & tlck->lock;
4249 
4250 	/* get slot index of the entry */
4251 	stbl = DT_GETSTBL(p);
4252 	entry_si = stbl[index];
4253 
4254 	/* linelock entry */
4255 	ASSERT(dtlck->index == 0);
4256 	lv = & dtlck->lv[0];
4257 	lv->offset = entry_si;
4258 	lv->length = 1;
4259 	dtlck->index++;
4260 
4261 	/* get the head/only segment */
4262 	entry = (struct ldtentry *) & p->slot[entry_si];
4263 
4264 	/* substitute the inode number of the entry */
4265 	entry->inumber = cpu_to_le32(new_ino);
4266 
4267 	/* unpin the leaf page */
4268 	DT_PUTPAGE(mp);
4269 
4270 	return 0;
4271 }
4272