xref: /original-bsd/lib/libc/db/btree/bt_split.c (revision 438c054e)
1 /*-
2  * Copyright (c) 1990 The Regents of the University of California.
3  * All rights reserved.
4  *
5  * This code is derived from software contributed to Berkeley by
6  * Mike Olson.
7  *
8  * %sccs.include.redist.c%
9  */
10 
11 #if defined(LIBC_SCCS) && !defined(lint)
12 static char sccsid[] = "@(#)bt_split.c	5.15 (Berkeley) 02/19/93";
13 #endif /* LIBC_SCCS and not lint */
14 
15 #include <sys/types.h>
16 
17 #define	__DBINTERFACE_PRIVATE
18 #include <limits.h>
19 #include <stdio.h>
20 #include <stdlib.h>
21 #include <string.h>
22 
23 #include <db.h>
24 #include "btree.h"
25 
26 static int	 bt_broot __P((BTREE *, PAGE *, PAGE *, PAGE *));
27 static PAGE	*bt_page
28 		    __P((BTREE *, PAGE *, PAGE **, PAGE **, u_int *, size_t));
29 static int	 bt_preserve __P((BTREE *, pgno_t));
30 static PAGE	*bt_psplit
31 		    __P((BTREE *, PAGE *, PAGE *, PAGE *, u_int *, size_t));
32 static PAGE	*bt_root
33 		    __P((BTREE *, PAGE *, PAGE **, PAGE **, u_int *, size_t));
34 static int	 bt_rroot __P((BTREE *, PAGE *, PAGE *, PAGE *));
35 static recno_t	 rec_total __P((PAGE *));
36 
37 #ifdef STATISTICS
38 u_long	bt_rootsplit, bt_split, bt_sortsplit, bt_pfxsaved;
39 #endif
40 
41 /*
42  * __BT_SPLIT -- Split the tree.
43  *
44  * Parameters:
45  *	t:	tree
46  *	sp:	page to split
47  *	key:	key to insert
48  *	data:	data to insert
49  *	flags:	BIGKEY/BIGDATA flags
50  *	ilen:	insert length
51  *	skip:	index to leave open
52  *
53  * Returns:
54  *	RET_ERROR, RET_SUCCESS
55  */
56 int
57 __bt_split(t, sp, key, data, flags, ilen, skip)
58 	BTREE *t;
59 	PAGE *sp;
60 	const DBT *key, *data;
61 	u_long flags;
62 	size_t ilen;
63 	u_int skip;
64 {
65 	BINTERNAL *bi;
66 	BLEAF *bl, *tbl;
67 	DBT a, b;
68 	EPGNO *parent;
69 	PAGE *h, *l, *r, *lchild, *rchild;
70 	indx_t nxtindex;
71 	size_t n, nbytes, nksize;
72 	int nosplit;
73 	char *dest;
74 
75 	/*
76 	 * Split the page into two pages, l and r.  The split routines return
77 	 * a pointer to the page into which the key should be inserted and with
78 	 * skip set to the offset which should be used.  Additionally, l and r
79 	 * are pinned.
80 	 */
81 	h = sp->pgno == P_ROOT ?
82 	    bt_root(t, sp, &l, &r, &skip, ilen) :
83 	    bt_page(t, sp, &l, &r, &skip, ilen);
84 	if (h == NULL)
85 		return (RET_ERROR);
86 
87 	/*
88 	 * Insert the new key/data pair into the leaf page.  (Key inserts
89 	 * always cause a leaf page to split first.)
90 	 */
91 	h->linp[skip] = h->upper -= ilen;
92 	dest = (char *)h + h->upper;
93 	if (ISSET(t, BTF_RECNO))
94 		WR_RLEAF(dest, data, flags)
95 	else
96 		WR_BLEAF(dest, key, data, flags)
97 
98 	/* If the root page was split, make it look right. */
99 	if (sp->pgno == P_ROOT &&
100 	    (ISSET(t, BTF_RECNO) ?
101 	    bt_rroot(t, sp, l, r) : bt_broot(t, sp, l, r)) == RET_ERROR)
102 		goto err2;
103 
104 	/*
105 	 * Now we walk the parent page stack -- a LIFO stack of the pages that
106 	 * were traversed when we searched for the page that split.  Each stack
107 	 * entry is a page number and a page index offset.  The offset is for
108 	 * the page traversed on the search.  We've just split a page, so we
109 	 * have to insert a new key into the parent page.
110 	 *
111 	 * If the insert into the parent page causes it to split, may have to
112 	 * continue splitting all the way up the tree.  We stop if the root
113 	 * splits or the page inserted into didn't have to split to hold the
114 	 * new key.  Some algorithms replace the key for the old page as well
115 	 * as the new page.  We don't, as there's no reason to believe that the
116 	 * first key on the old page is any better than the key we have, and,
117 	 * in the case of a key being placed at index 0 causing the split, the
118 	 * key is unavailable.
119 	 *
120 	 * There are a maximum of 5 pages pinned at any time.  We keep the left
121 	 * and right pages pinned while working on the parent.   The 5 are the
122 	 * two children, left parent and right parent (when the parent splits)
123 	 * and the root page or the overflow key page when calling bt_preserve.
124 	 * This code must make sure that all pins are released other than the
125 	 * root page or overflow page which is unlocked elsewhere.
126 	 */
127 	for (nosplit = 0; (parent = BT_POP(t)) != NULL;) {
128 		lchild = l;
129 		rchild = r;
130 
131 		/* Get the parent page. */
132 		if ((h = mpool_get(t->bt_mp, parent->pgno, 0)) == NULL)
133 			goto err2;
134 
135 	 	/*
136 		 * The new key goes ONE AFTER the index, because the split
137 		 * was to the right.
138 		 */
139 		skip = parent->index + 1;
140 
141 		/*
142 		 * Calculate the space needed on the parent page.
143 		 *
144 		 * Prefix trees: space hack when inserting into BINTERNAL
145 		 * pages.  Retain only what's needed to distinguish between
146 		 * the new entry and the LAST entry on the page to its left.
147 		 * If the keys compare equal, retain the entire key.  Note,
148 		 * we don't touch overflow keys, and the entire key must be
149 		 * retained for the next-to-left most key on the leftmost
150 		 * page of each level, or the search will fail.  Applicable
151 		 * ONLY to internal pages that have leaf pages as children.
152 		 * Further reduction of the key between pairs of internal
153 		 * pages loses too much information.
154 		 */
155 		switch (rchild->flags & P_TYPE) {
156 		case P_BINTERNAL:
157 			bi = GETBINTERNAL(rchild, 0);
158 			nbytes = NBINTERNAL(bi->ksize);
159 			break;
160 		case P_BLEAF:
161 			bl = GETBLEAF(rchild, 0);
162 			nbytes = NBINTERNAL(bl->ksize);
163 			if (t->bt_pfx && !(bl->flags & P_BIGKEY) &&
164 			    (h->prevpg != P_INVALID || skip > 1)) {
165 				tbl = GETBLEAF(lchild, NEXTINDEX(lchild) - 1);
166 				a.size = tbl->ksize;
167 				a.data = tbl->bytes;
168 				b.size = bl->ksize;
169 				b.data = bl->bytes;
170 				nksize = t->bt_pfx(&a, &b);
171 				n = NBINTERNAL(nksize);
172 				if (n < nbytes) {
173 #ifdef STATISTICS
174 					bt_pfxsaved += nbytes - n;
175 #endif
176 					nbytes = n;
177 				} else
178 					nksize = 0;
179 			} else
180 				nksize = 0;
181 			break;
182 		case P_RINTERNAL:
183 		case P_RLEAF:
184 			nbytes = NRINTERNAL;
185 			break;
186 		default:
187 			abort();
188 		}
189 
190 		/* Split the parent page if necessary or shift the indices. */
191 		if (h->upper - h->lower < nbytes + sizeof(indx_t)) {
192 			sp = h;
193 			h = h->pgno == P_ROOT ?
194 			    bt_root(t, h, &l, &r, &skip, nbytes) :
195 			    bt_page(t, h, &l, &r, &skip, nbytes);
196 			if (h == NULL)
197 				goto err1;
198 		} else {
199 			if (skip < (nxtindex = NEXTINDEX(h)))
200 				memmove(h->linp + skip + 1, h->linp + skip,
201 				    (nxtindex - skip) * sizeof(indx_t));
202 			h->lower += sizeof(indx_t);
203 			nosplit = 1;
204 		}
205 
206 		/* Insert the key into the parent page. */
207 		switch(rchild->flags & P_TYPE) {
208 		case P_BINTERNAL:
209 			h->linp[skip] = h->upper -= nbytes;
210 			dest = (char *)h + h->linp[skip];
211 			memmove(dest, bi, nbytes);
212 			((BINTERNAL *)dest)->pgno = rchild->pgno;
213 			break;
214 		case P_BLEAF:
215 			h->linp[skip] = h->upper -= nbytes;
216 			dest = (char *)h + h->linp[skip];
217 			WR_BINTERNAL(dest, nksize ? nksize : bl->ksize,
218 			    rchild->pgno, bl->flags & P_BIGKEY);
219 			memmove(dest, bl->bytes, nksize ? nksize : bl->ksize);
220 			if (bl->flags & P_BIGKEY &&
221 			    bt_preserve(t, *(pgno_t *)bl->bytes) == RET_ERROR)
222 				goto err1;
223 			break;
224 		case P_RINTERNAL:
225 			/* Update both left and right page counts. */
226 			h->linp[skip] = h->upper -= nbytes;
227 			dest = (char *)h + h->linp[skip];
228 			((RINTERNAL *)dest)->nrecs = rec_total(rchild);
229 			((RINTERNAL *)dest)->pgno = rchild->pgno;
230 			dest = (char *)h + h->linp[skip - 1];
231 			((RINTERNAL *)dest)->nrecs = rec_total(lchild);
232 			((RINTERNAL *)dest)->pgno = lchild->pgno;
233 			break;
234 		case P_RLEAF:
235 			/* Update both left and right page counts. */
236 			h->linp[skip] = h->upper -= nbytes;
237 			dest = (char *)h + h->linp[skip];
238 			((RINTERNAL *)dest)->nrecs = NEXTINDEX(rchild);
239 			((RINTERNAL *)dest)->pgno = rchild->pgno;
240 			dest = (char *)h + h->linp[skip - 1];
241 			((RINTERNAL *)dest)->nrecs = NEXTINDEX(lchild);
242 			((RINTERNAL *)dest)->pgno = lchild->pgno;
243 			break;
244 		default:
245 			abort();
246 		}
247 
248 		/* Unpin the held pages. */
249 		if (nosplit) {
250 			mpool_put(t->bt_mp, h, MPOOL_DIRTY);
251 			break;
252 		}
253 
254 		/* If the root page was split, make it look right. */
255 		if (sp->pgno == P_ROOT &&
256 		    (ISSET(t, BTF_RECNO) ?
257 		    bt_rroot(t, sp, l, r) : bt_broot(t, sp, l, r)) == RET_ERROR)
258 			goto err1;
259 
260 		mpool_put(t->bt_mp, lchild, MPOOL_DIRTY);
261 		mpool_put(t->bt_mp, rchild, MPOOL_DIRTY);
262 	}
263 
264 	/* Unpin the held pages. */
265 	mpool_put(t->bt_mp, l, MPOOL_DIRTY);
266 	mpool_put(t->bt_mp, r, MPOOL_DIRTY);
267 
268 	/* Clear any pages left on the stack. */
269 	return (RET_SUCCESS);
270 
271 	/*
272 	 * If something fails in the above loop we were already walking back
273 	 * up the tree and the tree is now inconsistent.  Nothing much we can
274 	 * do about it but release any memory we're holding.
275 	 */
276 err1:	mpool_put(t->bt_mp, lchild, MPOOL_DIRTY);
277 	mpool_put(t->bt_mp, rchild, MPOOL_DIRTY);
278 
279 err2:	mpool_put(t->bt_mp, l, 0);
280 	mpool_put(t->bt_mp, r, 0);
281 	__dbpanic(t->bt_dbp);
282 	return (RET_ERROR);
283 }
284 
285 /*
286  * BT_PAGE -- Split a non-root page of a btree.
287  *
288  * Parameters:
289  *	t:	tree
290  *	h:	root page
291  *	lp:	pointer to left page pointer
292  *	rp:	pointer to right page pointer
293  *	skip:	pointer to index to leave open
294  *	ilen:	insert length
295  *
296  * Returns:
297  *	Pointer to page in which to insert or NULL on error.
298  */
299 static PAGE *
300 bt_page(t, h, lp, rp, skip, ilen)
301 	BTREE *t;
302 	PAGE *h, **lp, **rp;
303 	u_int *skip;
304 	size_t ilen;
305 {
306 	PAGE *l, *r, *tp;
307 	pgno_t npg;
308 
309 #ifdef STATISTICS
310 	++bt_split;
311 #endif
312 	/* Put the new right page for the split into place. */
313 	if ((r = __bt_new(t, &npg)) == NULL)
314 		return (NULL);
315 	r->pgno = npg;
316 	r->lower = BTDATAOFF;
317 	r->upper = t->bt_psize;
318 	r->nextpg = h->nextpg;
319 	r->prevpg = h->pgno;
320 	r->flags = h->flags & P_TYPE;
321 
322 	/*
323 	 * If we're splitting the last page on a level because we're appending
324 	 * a key to it (skip is NEXTINDEX()), it's likely that the data is
325 	 * sorted.  Adding an empty page on the side of the level is less work
326 	 * and can push the fill factor much higher than normal.  If we're
327 	 * wrong it's no big deal, we'll just do the split the right way next
328 	 * time.  It may look like it's equally easy to do a similar hack for
329 	 * reverse sorted data, that is, split the tree left, but it's not.
330 	 * Don't even try.
331 	 */
332 	if (h->nextpg == P_INVALID && *skip == NEXTINDEX(h)) {
333 #ifdef STATISTICS
334 		++bt_sortsplit;
335 #endif
336 		h->nextpg = r->pgno;
337 		r->lower = BTDATAOFF + sizeof(indx_t);
338 		*skip = 0;
339 		*lp = h;
340 		*rp = r;
341 		return (r);
342 	}
343 
344 	/* Put the new left page for the split into place. */
345 	if ((l = malloc(t->bt_psize)) == NULL) {
346 		mpool_put(t->bt_mp, r, 0);
347 		return (NULL);
348 	}
349 	l->pgno = h->pgno;
350 	l->nextpg = r->pgno;
351 	l->prevpg = h->prevpg;
352 	l->lower = BTDATAOFF;
353 	l->upper = t->bt_psize;
354 	l->flags = h->flags & P_TYPE;
355 
356 	/* Fix up the previous pointer of the page after the split page. */
357 	if (h->nextpg != P_INVALID) {
358 		if ((tp = mpool_get(t->bt_mp, h->nextpg, 0)) == NULL) {
359 			free(l);
360 			/* XXX mpool_free(t->bt_mp, r->pgno); */
361 			return (NULL);
362 		}
363 		tp->prevpg = r->pgno;
364 		mpool_put(t->bt_mp, tp, 0);
365 	}
366 
367 	/*
368 	 * Split right.  The key/data pairs aren't sorted in the btree page so
369 	 * it's simpler to copy the data from the split page onto two new pages
370 	 * instead of copying half the data to the right page and compacting
371 	 * the left page in place.  Since the left page can't change, we have
372 	 * to swap the original and the allocated left page after the split.
373 	 */
374 	tp = bt_psplit(t, h, l, r, skip, ilen);
375 
376 	/* Move the new left page onto the old left page. */
377 	memmove(h, l, t->bt_psize);
378 	if (tp == l)
379 		tp = h;
380 	free(l);
381 
382 	*lp = h;
383 	*rp = r;
384 	return (tp);
385 }
386 
387 /*
388  * BT_ROOT -- Split the root page of a btree.
389  *
390  * Parameters:
391  *	t:	tree
392  *	h:	root page
393  *	lp:	pointer to left page pointer
394  *	rp:	pointer to right page pointer
395  *	skip:	pointer to index to leave open
396  *	ilen:	insert length
397  *
398  * Returns:
399  *	Pointer to page in which to insert or NULL on error.
400  */
401 static PAGE *
402 bt_root(t, h, lp, rp, skip, ilen)
403 	BTREE *t;
404 	PAGE *h, **lp, **rp;
405 	u_int *skip;
406 	size_t ilen;
407 {
408 	PAGE *l, *r, *tp;
409 	pgno_t lnpg, rnpg;
410 
411 #ifdef STATISTICS
412 	++bt_split;
413 	++bt_rootsplit;
414 #endif
415 	/* Put the new left and right pages for the split into place. */
416 	if ((l = __bt_new(t, &lnpg)) == NULL ||
417 	    (r = __bt_new(t, &rnpg)) == NULL)
418 		return (NULL);
419 	l->pgno = lnpg;
420 	r->pgno = rnpg;
421 	l->nextpg = r->pgno;
422 	r->prevpg = l->pgno;
423 	l->prevpg = r->nextpg = P_INVALID;
424 	l->lower = r->lower = BTDATAOFF;
425 	l->upper = r->upper = t->bt_psize;
426 	l->flags = r->flags = h->flags & P_TYPE;
427 
428 	/* Split the root page. */
429 	tp = bt_psplit(t, h, l, r, skip, ilen);
430 
431 	*lp = l;
432 	*rp = r;
433 	return (tp);
434 }
435 
436 /*
437  * BT_RROOT -- Fix up the recno root page after it has been split.
438  *
439  * Parameters:
440  *	t:	tree
441  *	h:	root page
442  *	l:	left page
443  *	r:	right page
444  *
445  * Returns:
446  *	RET_ERROR, RET_SUCCESS
447  */
448 static int
449 bt_rroot(t, h, l, r)
450 	BTREE *t;
451 	PAGE *h, *l, *r;
452 {
453 	char *dest;
454 
455 	/* Insert the left and right keys, set the header information. */
456 	h->linp[0] = h->upper = t->bt_psize - NRINTERNAL;
457 	dest = (char *)h + h->upper;
458 	WR_RINTERNAL(dest,
459 	    l->flags & P_RLEAF ? NEXTINDEX(l) : rec_total(l), l->pgno);
460 
461 	h->linp[1] = h->upper -= NRINTERNAL;
462 	dest = (char *)h + h->upper;
463 	WR_RINTERNAL(dest,
464 	    r->flags & P_RLEAF ? NEXTINDEX(r) : rec_total(r), r->pgno);
465 
466 	h->lower = BTDATAOFF + 2 * sizeof(indx_t);
467 
468 	/* Unpin the root page, set to recno internal page. */
469 	h->flags &= ~P_TYPE;
470 	h->flags |= P_RINTERNAL;
471 	mpool_put(t->bt_mp, h, MPOOL_DIRTY);
472 
473 	return (RET_SUCCESS);
474 }
475 
476 /*
477  * BT_BROOT -- Fix up the btree root page after it has been split.
478  *
479  * Parameters:
480  *	t:	tree
481  *	h:	root page
482  *	l:	left page
483  *	r:	right page
484  *
485  * Returns:
486  *	RET_ERROR, RET_SUCCESS
487  */
488 static int
489 bt_broot(t, h, l, r)
490 	BTREE *t;
491 	PAGE *h, *l, *r;
492 {
493 	BINTERNAL *bi;
494 	BLEAF *bl;
495 	size_t nbytes;
496 	char *dest;
497 
498 	/*
499 	 * If the root page was a leaf page, change it into an internal page.
500 	 * We copy the key we split on (but not the key's data, in the case of
501 	 * a leaf page) to the new root page.
502 	 *
503 	 * The btree comparison code guarantees that the left-most key on any
504 	 * level of the tree is never used, so it doesn't need to be filled in.
505 	 */
506 	nbytes = NBINTERNAL(0);
507 	h->linp[0] = h->upper = t->bt_psize - nbytes;
508 	dest = (char *)h + h->upper;
509 	WR_BINTERNAL(dest, 0, l->pgno, 0);
510 
511 	switch(h->flags & P_TYPE) {
512 	case P_BLEAF:
513 		bl = GETBLEAF(r, 0);
514 		nbytes = NBINTERNAL(bl->ksize);
515 		h->linp[1] = h->upper -= nbytes;
516 		dest = (char *)h + h->upper;
517 		WR_BINTERNAL(dest, bl->ksize, r->pgno, 0);
518 		memmove(dest, bl->bytes, bl->ksize);
519 
520 		/*
521 		 * If the key is on an overflow page, mark the overflow chain
522 		 * so it isn't deleted when the leaf copy of the key is deleted.
523 		 */
524 		if (bl->flags & P_BIGKEY &&
525 		    bt_preserve(t, *(pgno_t *)bl->bytes) == RET_ERROR)
526 			return (RET_ERROR);
527 		break;
528 	case P_BINTERNAL:
529 		bi = GETBINTERNAL(r, 0);
530 		nbytes = NBINTERNAL(bi->ksize);
531 		h->linp[1] = h->upper -= nbytes;
532 		dest = (char *)h + h->upper;
533 		memmove(dest, bi, nbytes);
534 		((BINTERNAL *)dest)->pgno = r->pgno;
535 		break;
536 	default:
537 		abort();
538 	}
539 
540 	/* There are two keys on the page. */
541 	h->lower = BTDATAOFF + 2 * sizeof(indx_t);
542 
543 	/* Unpin the root page, set to btree internal page. */
544 	h->flags &= ~P_TYPE;
545 	h->flags |= P_BINTERNAL;
546 	mpool_put(t->bt_mp, h, MPOOL_DIRTY);
547 
548 	return (RET_SUCCESS);
549 }
550 
551 /*
552  * BT_PSPLIT -- Do the real work of splitting the page.
553  *
554  * Parameters:
555  *	t:	tree
556  *	h:	page to be split
557  *	l:	page to put lower half of data
558  *	r:	page to put upper half of data
559  *	pskip:	pointer to index to leave open
560  *	ilen:	insert length
561  *
562  * Returns:
563  *	Pointer to page in which to insert.
564  */
565 static PAGE *
566 bt_psplit(t, h, l, r, pskip, ilen)
567 	BTREE *t;
568 	PAGE *h, *l, *r;
569 	u_int *pskip;
570 	size_t ilen;
571 {
572 	BINTERNAL *bi;
573 	BLEAF *bl;
574 	RLEAF *rl;
575 	EPGNO *c;
576 	PAGE *rval;
577 	void *src;
578 	indx_t full, half, nxt, off, skip, top, used;
579 	size_t nbytes;
580 	int bigkeycnt, isbigkey;
581 
582 	/*
583 	 * Split the data to the left and right pages.  Leave the skip index
584 	 * open.  Additionally, make some effort not to split on an overflow
585 	 * key.  This makes internal page processing faster and can save
586 	 * space as overflow keys used by internal pages are never deleted.
587 	 */
588 	bigkeycnt = 0;
589 	skip = *pskip;
590 	full = t->bt_psize - BTDATAOFF;
591 	half = full / 2;
592 	used = 0;
593 	for (nxt = off = 0, top = NEXTINDEX(h); nxt < top; ++off) {
594 		if (skip == off) {
595 			nbytes = ilen;
596 			isbigkey = 0;		/* XXX: not really known. */
597 		} else
598 			switch (h->flags & P_TYPE) {
599 			case P_BINTERNAL:
600 				src = bi = GETBINTERNAL(h, nxt);
601 				nbytes = NBINTERNAL(bi->ksize);
602 				isbigkey = bi->flags & P_BIGKEY;
603 				break;
604 			case P_BLEAF:
605 				src = bl = GETBLEAF(h, nxt);
606 				nbytes = NBLEAF(bl);
607 				isbigkey = bl->flags & P_BIGKEY;
608 				break;
609 			case P_RINTERNAL:
610 				src = GETRINTERNAL(h, nxt);
611 				nbytes = NRINTERNAL;
612 				isbigkey = 0;
613 				break;
614 			case P_RLEAF:
615 				src = rl = GETRLEAF(h, nxt);
616 				nbytes = NRLEAF(rl);
617 				isbigkey = 0;
618 				break;
619 			default:
620 				abort();
621 			}
622 
623 		/*
624 		 * If the key/data pairs are substantial fractions of the max
625 		 * possible size for the page, it's possible to get situations
626 		 * where we decide to try and copy too much onto the left page.
627 		 * Make sure that doesn't happen.
628 		 */
629 		if (skip <= off && used + nbytes >= full) {
630 			--off;
631 			break;
632 		}
633 
634 		/* Copy the key/data pair, if not the skipped index. */
635 		if (skip != off) {
636 			++nxt;
637 
638 			l->linp[off] = l->upper -= nbytes;
639 			memmove((char *)l + l->upper, src, nbytes);
640 		}
641 
642 		used += nbytes;
643 		if (used >= half) {
644 			if (!isbigkey || bigkeycnt == 3)
645 				break;
646 			else
647 				++bigkeycnt;
648 		}
649 	}
650 
651 	/*
652 	 * Off is the last offset that's valid for the left page.
653 	 * Nxt is the first offset to be placed on the right page.
654 	 */
655 	l->lower += (off + 1) * sizeof(indx_t);
656 
657 	/*
658 	 * If splitting the page that the cursor was on, the cursor has to be
659 	 * adjusted to point to the same record as before the split.  If the
660 	 * cursor is at or past the skipped slot, the cursor is incremented by
661 	 * one.  If the cursor is on the right page, it is decremented by the
662 	 * number of records split to the left page.
663 	 *
664 	 * Don't bother checking for the BTF_SEQINIT flag, the page number will
665 	 * be P_INVALID.
666 	 */
667 	c = &t->bt_bcursor;
668 	if (c->pgno == h->pgno) {
669 		if (c->index >= skip)
670 			++c->index;
671 		if (c->index < nxt)			/* Left page. */
672 			c->pgno = l->pgno;
673 		else {					/* Right page. */
674 			c->pgno = r->pgno;
675 			c->index -= nxt;
676 		}
677 	}
678 
679 	/*
680 	 * If the skipped index was on the left page, just return that page.
681 	 * Otherwise, adjust the skip index to reflect the new position on
682 	 * the right page.
683 	 */
684 	if (skip <= off) {
685 		skip = 0;
686 		rval = l;
687 	} else {
688 		rval = r;
689 		*pskip -= nxt;
690 	}
691 
692 	for (off = 0; nxt < top; ++off) {
693 		if (skip == nxt) {
694 			++off;
695 			skip = 0;
696 		}
697 		switch (h->flags & P_TYPE) {
698 		case P_BINTERNAL:
699 			src = bi = GETBINTERNAL(h, nxt);
700 			nbytes = NBINTERNAL(bi->ksize);
701 			break;
702 		case P_BLEAF:
703 			src = bl = GETBLEAF(h, nxt);
704 			nbytes = NBLEAF(bl);
705 			break;
706 		case P_RINTERNAL:
707 			src = GETRINTERNAL(h, nxt);
708 			nbytes = NRINTERNAL;
709 			break;
710 		case P_RLEAF:
711 			src = rl = GETRLEAF(h, nxt);
712 			nbytes = NRLEAF(rl);
713 			break;
714 		default:
715 			abort();
716 		}
717 		++nxt;
718 		r->linp[off] = r->upper -= nbytes;
719 		memmove((char *)r + r->upper, src, nbytes);
720 	}
721 	r->lower += off * sizeof(indx_t);
722 
723 	/* If the key is being appended to the page, adjust the index. */
724 	if (skip == top)
725 		r->lower += sizeof(indx_t);
726 
727 	return (rval);
728 }
729 
730 /*
731  * BT_PRESERVE -- Mark a chain of pages as used by an internal node.
732  *
733  * Chains of indirect blocks pointed to by leaf nodes get reclaimed when the
734  * record that references them gets deleted.  Chains pointed to by internal
735  * pages never get deleted.  This routine marks a chain as pointed to by an
736  * internal page.
737  *
738  * Parameters:
739  *	t:	tree
740  *	pg:	page number of first page in the chain.
741  *
742  * Returns:
743  *	RET_SUCCESS, RET_ERROR.
744  */
745 static int
746 bt_preserve(t, pg)
747 	BTREE *t;
748 	pgno_t pg;
749 {
750 	PAGE *h;
751 
752 	if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
753 		return (RET_ERROR);
754 	h->flags |= P_PRESERVE;
755 	mpool_put(t->bt_mp, h, MPOOL_DIRTY);
756 	return (RET_SUCCESS);
757 }
758 
759 /*
760  * REC_TOTAL -- Return the number of recno entries below a page.
761  *
762  * Parameters:
763  *	h:	page
764  *
765  * Returns:
766  *	The number of recno entries below a page.
767  *
768  * XXX
769  * These values could be set by the bt_psplit routine.  The problem is that the
770  * entry has to be popped off of the stack etc. or the values have to be passed
771  * all the way back to bt_split/bt_rroot and it's not very clean.
772  */
773 static recno_t
774 rec_total(h)
775 	PAGE *h;
776 {
777 	recno_t recs;
778 	indx_t nxt, top;
779 
780 	for (recs = 0, nxt = 0, top = NEXTINDEX(h); nxt < top; ++nxt)
781 		recs += GETRINTERNAL(h, nxt)->nrecs;
782 	return (recs);
783 }
784