xref: /freebsd/contrib/tcsh/sh.hist.c (revision d37eb51047221dc3322b34db1038ff3aa533883f)
1 /*
2  * sh.hist.c: Shell history expansions and substitutions
3  */
4 /*-
5  * Copyright (c) 1980, 1991 The Regents of the University of California.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the University nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 #include "sh.h"
33 #include <stdio.h>	/* for rename(2), grr. */
34 #include <assert.h>
35 #include "tc.h"
36 #include "dotlock.h"
37 
38 extern int histvalid;
39 extern struct Strbuf histline;
40 Char HistLit = 0;
41 
42 static	int	heq	(const struct wordent *, const struct wordent *);
43 static	void	hfree	(struct Hist *);
44 
45 #define HIST_ONLY	0x01
46 #define HIST_SAVE	0x02
47 #define HIST_LOAD	0x04
48 #define HIST_REV	0x08
49 #define HIST_CLEAR	0x10
50 #define HIST_MERGE	0x20
51 #define HIST_TIME	0x40
52 
53 /*
54  * C shell
55  */
56 
57 /* Static functions don't show up in gprof summaries.  So eliminate "static"
58  * modifier from some frequently called functions. */
59 #ifdef PROF
60 #define PG_STATIC
61 #else
62 #define PG_STATIC static
63 #endif
64 
65 /* #define DEBUG_HIST 1 */
66 
67 static const int fastMergeErase = 1;
68 static unsigned histCount = 0;		/* number elements on history list */
69 static int histlen = 0;
70 static struct Hist *histTail = NULL;     /* last element on history list */
71 static struct Hist *histMerg = NULL;	 /* last element merged by Htime */
72 
73 static void insertHistHashTable(struct Hist *, unsigned);
74 
75 /* Insert new element (hp) in history list after specified predecessor (pp). */
76 static void
77 hinsert(struct Hist *hp, struct Hist *pp)
78 {
79     struct Hist *fp = pp->Hnext;        /* following element, if any */
80     hp->Hnext = fp, hp->Hprev = pp;
81     pp->Hnext = hp;
82     if (fp)
83         fp->Hprev = hp;
84     else
85         histTail = hp;                  /* meaning hp->Hnext == NULL */
86     histCount++;
87 }
88 
89 /* Remove the entry from the history list. */
90 static void
91 hremove(struct Hist *hp)
92 {
93     struct Hist *pp = hp->Hprev;
94     assert(pp);                         /* elements always have a previous */
95     pp->Hnext = hp->Hnext;
96     if (hp->Hnext)
97         hp->Hnext->Hprev = pp;
98     else
99         histTail = pp;                  /* we must have been last */
100     if (hp == histMerg)			/* deleting this hint from list */
101 	histMerg = NULL;
102     assert(histCount > 0);
103     histCount--;
104 }
105 
106 /* Prune length of history list to specified size by history variable. */
107 PG_STATIC void
108 discardExcess(int hlen)
109 {
110     struct Hist *hp, *np;
111     if (histTail == NULL) {
112         assert(histCount == 0);
113         return;                         /* no entries on history list */
114     }
115     /* Prune dummy entries from the front, then old entries from the back. If
116      * the list is still too long scan the whole list as before.  But only do a
117      * full scan if the list is more than 6% (1/16th) too long. */
118     while (histCount > (unsigned)hlen && (np = Histlist.Hnext)) {
119         if (eventno - np->Href >= hlen || hlen == 0)
120             hremove(np), hfree(np);
121         else
122             break;
123     }
124     while (histCount > (unsigned)hlen && (np = histTail) != &Histlist) {
125         if (eventno - np->Href >= hlen || hlen == 0)
126             hremove(np), hfree(np);
127         else
128             break;
129     }
130     if (histCount - (hlen >> 4) <= (unsigned)hlen)
131 	return;				/* don't bother doing the full scan */
132     for (hp = &Histlist; histCount > (unsigned)hlen &&
133 	(np = hp->Hnext) != NULL;)
134         if (eventno - np->Href >= hlen || hlen == 0)
135             hremove(np), hfree(np);
136         else
137             hp = np;
138 }
139 
140 /* Add the command "sp" to the history list. */
141 void
142 savehist(
143   struct wordent *sp,
144   int mflg)				/* true if -m (merge) specified */
145 {
146     /* throw away null lines */
147     if (sp && sp->next->word[0] == '\n')
148 	return;
149     if (sp)
150         (void) enthist(++eventno, sp, 1, mflg, histlen);
151     discardExcess(histlen);
152 }
153 
154 #define USE_JENKINS_HASH 1
155 /* #define USE_ONE_AT_A_TIME 1 */
156 #undef PRIME_LENGTH			/* no need for good HTL */
157 
158 #ifdef USE_JENKINS_HASH
159 #define hashFcnName "lookup3"
160 /* From:
161    lookup3.c, by Bob Jenkins, May 2006, Public Domain.
162    "...  You can use this free for any purpose.  It's in
163     the public domain.  It has no warranty."
164    http://burtleburtle.net/bob/hash/index.html
165  */
166 
167 #define rot(x,k) (((x)<<(k)) | ((x)>>(32-(k))))
168 #define mix(a,b,c) \
169 { \
170   a -= c;  a ^= rot(c, 4);  c += b; \
171   b -= a;  b ^= rot(a, 6);  a += c; \
172   c -= b;  c ^= rot(b, 8);  b += a; \
173   a -= c;  a ^= rot(c,16);  c += b; \
174   b -= a;  b ^= rot(a,19);  a += c; \
175   c -= b;  c ^= rot(b, 4);  b += a; \
176 }
177 #define final(a,b,c) \
178 { \
179   c ^= b; c -= rot(b,14); \
180   a ^= c; a -= rot(c,11); \
181   b ^= a; b -= rot(a,25); \
182   c ^= b; c -= rot(b,16); \
183   a ^= c; a -= rot(c, 4); \
184   b ^= a; b -= rot(a,14); \
185   c ^= b; c -= rot(b,24); \
186 }
187 
188 struct hashValue		  /* State used to hash a wordend word list. */
189 {
190     uint32_t a, b, c;
191 };
192 
193 /* Set up the internal state */
194 static void
195 initializeHash(struct hashValue *h)
196 {
197     h->a = h->b = h->c = 0xdeadbeef;
198 }
199 
200 /* This does a partial hash of the Chars in a single word.  For efficiency we
201  * include 3 versions of the code to pack Chars into 32-bit words for the
202  * mixing function. */
203 static void
204 addWordToHash(struct hashValue *h, const Char *word)
205 {
206     uint32_t a = h->a, b = h->b, c = h->c;
207 #ifdef SHORT_STRINGS
208 #define GETK    if ((k = (uChar)*word++) == 0) break
209 #ifdef WIDE_STRINGS
210     assert(sizeof(Char) >= 4);
211     while (1) {
212 	unsigned k;
213 	GETK;
214 	a += k;
215 	GETK;
216 	b += k;
217 	GETK;
218 	c += k;
219 	mix(a, b, c);
220     }
221 #else
222     assert(sizeof(Char) == 2);
223     while (1) {
224 	unsigned k;
225 	GETK;
226 	a += k;
227 	GETK;
228 	a += k << 16;
229 	GETK;
230 	b += k;
231 	GETK;
232 	b += k << 16;
233 	GETK;
234 	c += k;
235 	GETK;
236 	c += k << 16;
237 	mix(a, b, c);
238     }
239 #endif
240 #else
241     assert(sizeof(Char) == 1);
242     while (1) {
243 	unsigned k;
244 	if ((k = *word++) == 0) break; a += k;
245 	if ((k = *word++) == 0) break; a += k << 8;
246 	if ((k = *word++) == 0) break; a += k << 16;
247 	if ((k = *word++) == 0) break; a += k << 24;
248 	if ((k = *word++) == 0) break; b += k;
249 	if ((k = *word++) == 0) break; b += k << 8;
250 	if ((k = *word++) == 0) break; b += k << 16;
251 	if ((k = *word++) == 0) break; b += k << 24;
252 	if ((k = *word++) == 0) break; c += k;
253 	if ((k = *word++) == 0) break; c += k << 8;
254 	if ((k = *word++) == 0) break; c += k << 16;
255 	if ((k = *word++) == 0) break; c += k << 24;
256 	mix(a, b, c);
257     }
258 #endif
259     h->a = a, h->b = b, h->c = c;
260 }
261 
262 static void
263 addCharToHash(struct hashValue *h, Char ch)
264 {
265     /* The compiler (gcc -O2) seems to do a good job optimizing this without
266      * explicitly extracting into local variables. */
267     h->a += (uChar)ch;
268     mix(h->a, h->b, h->c);
269 }
270 
271 static uint32_t
272 finalizeHash(struct hashValue *h)
273 {
274     uint32_t a = h->a, b = h->b, c = h->c;
275     final(a, b, c);
276     return c;
277 }
278 
279 #elif USE_ONE_AT_A_TIME
280 #define hashFcnName "one-at-a-time"
281 /* This one is also from Bob Jenkins, but is slower but simpler than lookup3.
282    "...  The code given here are all public domain."
283    http://burtleburtle.net/bob/hash/doobs.html */
284 
285 #if 0
286 ub4
287 one_at_a_time(char *key, ub4 len)
288 {
289   ub4   hash, i;
290   for (hash=0, i=0; i<len; ++i)
291   {
292     hash += key[i];
293     hash += (hash << 10);
294     hash ^= (hash >> 6);
295   }
296   hash += (hash << 3);
297   hash ^= (hash >> 11);
298   hash += (hash << 15);
299   return (hash & mask);
300 }
301 #endif
302 
303 struct hashValue { uint32_t h; };
304 static void
305 initializeHash(struct hashValue *h)
306 {
307     h->h = 0;
308 }
309 
310 static void
311 addWordToHash(struct hashValue *h, const Char *word)
312 {
313     unsigned k;
314     uint32_t hash = h->h;
315     while (k = (uChar)*word++)
316 	hash += k, hash += hash << 10, hash ^= hash >> 6;
317     h->h = hash;
318 }
319 
320 static void
321 addCharToHash(struct hashValue *h, Char c)
322 {
323     Char b[2] = { c, 0 };
324     addWordToHash(h, b);
325 }
326 
327 static uint32_t
328 finalizeHash(struct hashValue *h)
329 {
330     unsigned hash = h->h;
331     hash += (hash << 3);
332     hash ^= (hash >> 11);
333     hash += (hash << 15);
334     return hash;
335 }
336 
337 #else
338 #define hashFcnName "add-mul"
339 /* Simple multipy and add hash. */
340 #define PRIME_LENGTH 1			/* need "good" HTL */
341 struct hashValue { uint32_t h; };
342 static void
343 initializeHash(struct hashValue *h)
344 {
345     h->h = 0xe13e2345;
346 }
347 
348 static void
349 addWordToHash(struct hashValue *h, const Char *word)
350 {
351     unsigned k;
352     uint32_t hash = h->h;
353     while (k = (uChar)*word++)
354 	hash = hash * 0x9e4167b9 + k;
355     h->h = hash;
356 }
357 
358 static void
359 addCharToHash(struct hashValue *h, Char c)
360 {
361     h->h = h->h * 0x9e4167b9 + (uChar)c;
362 }
363 
364 static uint32_t
365 finalizeHash(struct hashValue *h)
366 {
367     return h->h;
368 }
369 #endif
370 
371 static unsigned
372 hashhist(struct wordent *h0)
373 {
374     struct hashValue s;
375     struct wordent *firstWord = h0->next;
376     struct wordent *h = firstWord;
377     unsigned hash = 0;
378 
379     initializeHash(&s);
380     for (; h != h0; h = h->next) {
381         if (h->word[0] == '\n')
382             break;                      /* don't hash newline */
383         if (h != firstWord)
384             addCharToHash(&s, ' ');	/* space between words */
385 	addWordToHash(&s, h->word);
386     }
387     hash = finalizeHash(&s);
388     /* Zero means no hash value, so never return zero as a hash value. */
389     return hash ? hash : 0x7fffffff;	/* prime! */
390 }
391 
392 #if 0
393 unsigned
394 hashStr(Char *str)
395 {
396     struct hashValue s;
397     initializeHash(&s);
398     addWordToHash(&s, str);
399     return finalizeHash(&s);
400 }
401 #endif
402 
403 #ifdef PRIME_LENGTH			/* need good HTL */
404 #define hash2tableIndex(hash, len) ((hash) % len)
405 #else
406 #define hash2tableIndex(hash, len) ((hash) & (len-1))
407 #endif
408 
409 /* This code can be enabled to test the above hash functions for speed and
410  * collision avoidance.  The testing is enabled by "occasional" calls to
411  * displayHistStats(), see which. */
412 #ifdef DEBUG_HIST
413 
414 #ifdef BSDTIMES
415 static double
416 doTiming(int start) {
417     static struct timeval beginTime;
418     if (start) {
419 	gettimeofday(&beginTime, NULL);
420 	return 0.0;
421     } else {
422 	struct timeval now;
423 	gettimeofday(&now, NULL);
424 	return (now.tv_sec-beginTime.tv_sec) +
425 	    (now.tv_usec-beginTime.tv_usec)/1e6;
426     }
427 }
428 #else
429 static double
430 doTiming(int start) {
431     USE(start);
432     return 0.0;
433 }
434 #endif
435 
436 static void
437 generateHashes(int nChars, unsigned nWords, unsigned samples, unsigned *hashes,
438     unsigned length)
439 {
440     if (nChars < 1)
441 	return;
442     nWords = (nWords < 1) ? 1 : (nWords > 4) ? 4 : nWords;
443     Char *number = xmalloc((nChars+nWords)*sizeof(Char));
444     struct wordent word[4];
445     struct wordent base = { NULL, &word[0], &word[0] };
446     word[0].word = number, word[0].next = &base, word[0].prev = &base;
447     unsigned w = 0;			/* word number */
448     /* Generate multiple words of length 2, 3, 5, then all the rest. */
449     unsigned wBoundaries[4] = { 2-1, 2+3-1, 2+3+5-1, 0 };
450     /* Ensure the last word has at least 4 Chars in it. */
451     while (nWords >= 2 && nChars < (wBoundaries[nWords-2]+1) + 4)
452 	nWords--;
453     wBoundaries[nWords-1] = 0xffffffff;	/* don't end word past this point */
454     unsigned i;
455     for (i = 0; i<nChars; i++) {
456 	/* In deference to the gawd awful add-mul hash, we won't use the worse
457 	 * case here (setting all Chars to 1), but assume mostly (or at least
458 	 * initially) ASCII data. */
459 	number[i+w] = '!';		/* 0x21 = 33 */
460 
461 	if (i == wBoundaries[w]) {	/* end a word here and move to next */
462 	    w++;			/* next word */
463 	    number[i+w] = 0;		/* terminate */
464 	    word[w].word = &number[i+w+1];
465 	    word[w].next = &base, word[w].prev = &word[w-1];
466 	    word[w-1].next = &word[w], base.prev = &word[w];
467 	}
468     }
469     /* w is the index of the last word actually created. */
470     number[nChars + w] = 0;		/* terminate last word */
471     unsigned timeLimit = !samples;
472     if (samples == 0)
473 	samples = 1000000000;
474     doTiming(1);
475     double sec;
476     for (i = 0; i < samples; i++) {
477 	/* increment 4 digit base 255 number; last characters vary fastest */
478 	unsigned j = nChars-1 + w;
479 	while (1) {
480 	    if (++number[j] != 0)
481 		break;
482 	    /* else reset this digit and proceed to next one */
483 	    number[j] = 1;
484 	    if (&number[j] <= word[w].word)
485 		break;			/* stop at beginning of last word */
486 	    j--;
487 	}
488 	if (word[w].word[0] == '\n')
489 	    word[w].word[0]++;		/* suppress newline character */
490 	unsigned hash = hashhist(&base);
491 	hashes[hash2tableIndex(hash, length)]++;
492 	if (timeLimit && (i & 0x3ffff) == 0x3ffff) {
493 	    sec = doTiming(0);
494 	    if (sec > 10)
495 		break;
496 	}
497     }
498     if (i >= samples)
499 	sec = doTiming(0);
500     else
501 	samples = i;			/* number we actually did */
502     if (sec > 0.01) {
503 	xprintf("Hash %d (%d Char %u words) with %s: %d nsec/hash, %d mcps\n",
504 		samples, nChars, w+1, hashFcnName, (int)((sec/samples)*1e9),
505 		(int)((double)samples*nChars/sec/1e6));
506     }
507 }
508 #endif /* DEBUG_HIST */
509 
510 #ifdef DEBUG_HIST
511 static void
512 testHash(void)
513 {
514     static const Char STRtestHashTimings[] =
515 	{ 't','e','s','t','H','a','s','h','T','i','m','i','n','g','s', 0 };
516     struct varent *vp = adrof(STRtestHashTimings);
517     if (vp && vp->vec) {
518 	unsigned hashes[4];		/* dummy place to put hashes */
519 	Char **vals = vp->vec;
520 	while (*vals) {
521 	    int length = getn(*vals);
522 	    unsigned words =
523 		(length < 5) ? 1 : (length < 25) ? 2 : (length < 75) ? 3 : 4;
524 	    if (length > 0)
525 		generateHashes(length, words, 0, hashes, 4);
526 	    vals++;
527 	}
528     }
529     unsigned length = 1024;
530 #ifdef PRIME_LENGTH			/* need good HTL */
531     length = 1021;
532 #endif
533     unsigned *hashes = xmalloc(length*sizeof(unsigned));
534     memset(hashes, 0, length*sizeof(unsigned));
535     /* Compute collision statistics for half full hashes modulo "length". */
536     generateHashes(4, 1, length/2, hashes, length);
537     /* Evaluate collisions by comparing occupancy rates (mean value 0.5).
538      * One bin for each number of hits. */
539     unsigned bins[155];
540     memset(bins, 0, sizeof(bins));
541     unsigned highest = 0;
542     unsigned i;
543     for (i = 0; i<length; i++) {
544 	unsigned hits = hashes[i];
545 	if (hits >= sizeof(bins)/sizeof(bins[0])) /* clip */
546 	    hits = highest = sizeof(bins)/sizeof(bins[0]) - 1;
547 	if (hits > highest)
548 	    highest = hits;
549 	bins[hits]++;
550     }
551     xprintf("Occupancy of %d buckets by %d hashes %d Chars %d word with %s\n",
552 	    length, length/2, 4, 1, hashFcnName);
553     for (i = 0; i <= highest; i++) {
554 	xprintf(" %d buckets (%d%%) with %d hits\n",
555 		bins[i], bins[i]*100/length, i);
556     }
557     /* Count run lengths to evaluate linear rehashing effectiveness.  Estimate
558      * a little corrupted by edge effects. */
559     memset(bins, 0, sizeof(bins));
560     highest = 0;
561     for (i = 0; hashes[i] == 0; i++);	/* find first occupied bucket */
562     unsigned run = 0;
563     unsigned rehashed = 0;
564     for (; i<length; i++) {
565 	unsigned hits = hashes[i];
566 	if (hits == 0 && rehashed > 0)
567 	    hits = 1 && rehashed--;
568 	else if (hits > 1)
569 	    rehashed += hits-1;
570 	if (hits)
571 	    run++;
572 	else {
573 	    /* a real free slot, count it */
574 	    if (run >= sizeof(bins)/sizeof(bins[0])) /* clip */
575 		run = highest = sizeof(bins)/sizeof(bins[0]) - 1;
576 	    if (run > highest)
577 		highest = run;
578 	    bins[run]++;
579 	    run = 0;
580 	}
581     }
582     /* Ignore the partial run at end as we ignored the beginning. */
583     double merit = 0.0, entries = 0;
584     for (i = 0; i <= highest; i++) {
585 	entries += bins[i]*i;		/* total hashed objects */
586 	merit += bins[i]*i*i;
587     }
588     xprintf("Rehash collision figure of merit %u (ideal=100), run lengths:\n",
589 	    (int)(100.0*merit/entries));
590     for (i = 0; i <= highest; i++) {
591 	if (bins[i] != 0)
592 	    xprintf(" %d runs of length %d buckets\n", bins[i], i);
593     }
594     xfree(hashes);
595 }
596 #endif /* DEBUG_HIST */
597 
598 /* Compares two word lists for equality. */
599 static int
600 heq(const struct wordent *a0, const struct wordent *b0)
601 {
602     const struct wordent *a = a0->next, *b = b0->next;
603 
604     for (;;) {
605 	if (Strcmp(a->word, b->word) != 0)
606 	    return 0;
607 	a = a->next;
608 	b = b->next;
609 	if (a == a0)
610 	    return (b == b0) ? 1 : 0;
611 	if (b == b0)
612 	    return 0;
613     }
614 }
615 
616 /* Renumber entries following p, which we will be deleting. */
617 PG_STATIC void
618 renumberHist(struct Hist *p)
619 {
620     int n = p->Href;
621     while ((p = p->Hnext))
622         p->Href = n--;
623 }
624 
625 /* The hash table is implemented as an array of pointers to Hist entries.  Each
626  * entry is located in the table using hash2tableIndex() and checking the
627  * following entries in case of a collision (linear rehash).  Free entries in
628  * the table are zero (0, NULL, emptyHTE).  Deleted entries that cannot yet be
629  * freed are set to one (deletedHTE).  The Hist.Hhash member is non-zero iff
630  * the entry is in the hash table.  When the hash table get too full, it is
631  * reallocated to be approximately twice the history length (see
632  * getHashTableSize). */
633 static struct Hist **histHashTable = NULL;
634 static unsigned histHashTableLength = 0; /* number of Hist pointers in table */
635 
636 static struct Hist * const emptyHTE = NULL;
637 static struct Hist * const deletedHTE = (struct Hist *)1;
638 
639 static struct {
640     unsigned insertCount;
641     unsigned removeCount;
642     unsigned rehashes;
643     int deleted;
644 } hashStats;
645 
646 #ifdef DEBUG_HIST
647 void
648 checkHistHashTable(int print)
649 {
650     unsigned occupied = 0;
651     unsigned deleted = 0;
652     unsigned i;
653     for (i = 0; i<histHashTableLength; i++)
654 	if (histHashTable[i] == emptyHTE)
655 	    continue;
656 	else if (histHashTable[i] == deletedHTE)
657 	    deleted++;
658 	else
659 	    occupied++;
660     if (print)
661 	xprintf("  found len %u occupied %u deleted %u\n",
662 		histHashTableLength, occupied, deleted);
663     assert(deleted == hashStats.deleted);
664 }
665 
666 static int doneTest = 0;
667 
668 /* Main entry point for displaying history statistics and hash function
669  * behavior. */
670 void
671 displayHistStats(const char *reason)
672 {
673     /* Just hash statistics for now. */
674     xprintf("%s history hash table len %u count %u (deleted %d)\n", reason,
675 	    histHashTableLength, histCount, hashStats.deleted);
676     xprintf("  inserts %u rehashes %u%% each\n",
677 	    hashStats.insertCount,
678 	    (hashStats.insertCount
679 	     ? 100*hashStats.rehashes/hashStats.insertCount : 0));
680     xprintf("  removes %u net %u\n",
681 	    hashStats.removeCount,
682 	    hashStats.insertCount - hashStats.removeCount);
683     assert(hashStats.insertCount >= hashStats.removeCount);
684     checkHistHashTable(1);
685     memset(&hashStats, 0, sizeof(hashStats));
686     if (!doneTest) {
687 	testHash();
688 	doneTest = 1;
689     }
690 }
691 #else
692 void
693 displayHistStats(const char *reason)
694 {
695     USE(reason);
696 }
697 #endif
698 
699 static void
700 discardHistHashTable(void)
701 {
702     if (histHashTable == NULL)
703         return;
704     displayHistStats("Discarding");
705     xfree(histHashTable);
706     histHashTable = NULL;
707 }
708 
709 /* Computes a new hash table size, when the current one is too small. */
710 static unsigned
711 getHashTableSize(int hlen)
712 {
713     unsigned target = hlen * 2;
714     unsigned e = 5;
715     unsigned size;
716     while ((size = 1<<e) < target)
717 	e++;
718 #ifdef PRIME_LENGTH			/* need good HTL */
719     /* Not all prime, but most are and none have factors smaller than 11. */
720     return size+15;
721 #else
722     assert((size & (size-1)) == 0);	/* must be a power of two */
723     return size;
724 #endif
725 }
726 
727 /* Create the hash table or resize, if necessary. */
728 static void
729 createHistHashTable(int hlen)
730 {
731     if (hlen == 0) {
732 	discardHistHashTable();
733         return;
734     }
735     if (hlen < 0) {
736 	if (histlen <= 0)
737 	    return;			/* no need for hash table */
738 	hlen = histlen;
739     }
740     if (histHashTable != NULL) {
741 	if (histCount < histHashTableLength * 3 / 4)
742 	    return;			/* good enough for now */
743 	discardHistHashTable();		/* too small */
744     }
745     histHashTableLength = getHashTableSize(
746 	hlen > (int)histCount ? hlen : (int)histCount);
747     histHashTable = xmalloc(histHashTableLength * sizeof(struct Hist *));
748     memset(histHashTable, 0, histHashTableLength * sizeof(struct Hist *));
749     assert(histHashTable[0] == emptyHTE);
750 
751     /* Now insert all the entries on the history list into the hash table. */
752     {
753         struct Hist *hp;
754         for (hp = &Histlist; (hp = hp->Hnext) != NULL;) {
755             unsigned lpHash = hashhist(&hp->Hlex);
756             assert(!hp->Hhash || hp->Hhash == lpHash);
757             hp->Hhash = 0;              /* force insert to new hash table */
758             insertHistHashTable(hp, lpHash);
759         }
760     }
761 }
762 
763 /* Insert np into the hash table.  We assume that np is already on the
764  * Histlist.  The specified hashval matches the new Hist entry but has not yet
765  * been assigned to Hhash (or the element is already on the hash table). */
766 static void
767 insertHistHashTable(struct Hist *np, unsigned hashval)
768 {
769     unsigned rehashes = 0;
770     unsigned hi = 0;
771     if (!histHashTable)
772 	return;
773     if (np->Hhash != 0) {
774         /* already in hash table */
775         assert(hashval == np->Hhash);
776         return;
777     }
778     assert(np != deletedHTE);
779     /* Find a free (empty or deleted) slot, using linear rehash. */
780     assert(histHashTable);
781     for (rehashes = 0;
782          ((hi = hash2tableIndex(hashval + rehashes, histHashTableLength)),
783           histHashTable[hi] != emptyHTE && histHashTable[hi] != deletedHTE);
784          rehashes++) {
785         assert(np != histHashTable[hi]);
786         if (rehashes >= histHashTableLength / 10) {
787             /* Hash table is full, so grow it.  We assume the create function
788              * will roughly double the size we give it.  Create initializes the
789              * new table with everything on the Histlist, so we are done when
790              * it returns.  */
791 #ifdef DEBUG_HIST
792 	    xprintf("Growing history hash table from %d ...",
793 		histHashTableLength);
794 	    flush();
795 #endif
796             discardHistHashTable();
797             createHistHashTable(histHashTableLength);
798 #ifdef DEBUG_HIST
799 	    xprintf("to %d.\n", histHashTableLength);
800 #endif
801             return;
802         }
803     }
804     /* Might be sensible to grow hash table if rehashes is "too big" here. */
805     if (histHashTable[hi] == deletedHTE)
806 	hashStats.deleted--;
807     histHashTable[hi] = np;
808     np->Hhash = hashval;
809     hashStats.insertCount++;
810     hashStats.rehashes += rehashes;
811 }
812 
813 /* Remove the 'np' entry from the hash table. */
814 static void
815 removeHistHashTable(struct Hist *np)
816 {
817     unsigned hi = np->Hhash;
818     if (!histHashTable || !hi)
819         return;                         /* no hash table or not on it */
820     /* find desired entry */
821     while ((hi = hash2tableIndex(hi, histHashTableLength)),
822            histHashTable[hi] != emptyHTE) {
823         if (np == histHashTable[hi]) {
824 	    unsigned i;
825 	    unsigned deletes = 0;
826 	    histHashTable[hi] = deletedHTE; /* dummy, but non-zero entry */
827 	    /* now peek ahead to see if the dummies are really necessary. */
828 	    i = 1;
829 	    while (histHashTable[hash2tableIndex(hi+i, histHashTableLength)] ==
830 		   deletedHTE)
831 		i++;
832 	    if (histHashTable[hash2tableIndex(hi+i, histHashTableLength)] ==
833 		emptyHTE) {
834 		/* dummies are no longer necessary placeholders. */
835 		deletes = i;
836 		while (i-- > 0) {
837 		    histHashTable[hash2tableIndex(hi+i, histHashTableLength)] =
838 			emptyHTE;
839 		}
840 	    }
841 	    hashStats.deleted += 1 - deletes; /* delta deleted entries */
842 	    hashStats.removeCount++;
843             return;
844         }
845         hi++;                           /* linear rehash */
846     }
847     assert(!"Hist entry not found in hash table");
848 }
849 
850 /* Search the history hash table for a command matching lp, using hashval as
851  * its hash value. */
852 static struct Hist *
853 findHistHashTable(struct wordent *lp, unsigned hashval)
854 {
855     unsigned deleted = 0;		/* number of deleted entries skipped */
856     unsigned hi = hashval;
857     struct Hist *hp;
858     if (!histHashTable)
859 	return NULL;
860     while ((hi = hash2tableIndex(hi, histHashTableLength)),
861            (hp = histHashTable[hi]) != emptyHTE) {
862         if (hp == deletedHTE)
863 	    deleted++;
864 	else if (hp->Hhash == hashval && heq(lp, &(hp->Hlex)))
865             return hp;
866 	if (deleted > (histHashTableLength>>4)) {
867 	    /* lots of deletes, so we need a sparser table. */
868             discardHistHashTable();
869             createHistHashTable(histHashTableLength);
870 	    return findHistHashTable(lp, hashval);
871 	}
872         hi++;                           /* linear rehash */
873     }
874     return NULL;
875 }
876 
877 /* When merge semantics are in use, find the approximate predecessor for the
878  * new entry, so that the Htime entries are decreasing.  Return the entry just
879  * before the first entry with equal times, so the caller can check for
880  * duplicates.  When pTime is not NULL, use it as a starting point for search,
881  * otherwise search from beginning (largest time value) of history list. */
882 PG_STATIC struct Hist *
883 mergeInsertionPoint(
884     struct Hist *np,                      /* new entry to be inserted */
885     struct Hist *pTime)                   /* hint about where to insert */
886 {
887     struct Hist *pp, *p;
888     if (histTail && histTail->Htime >= np->Htime)
889 	pTime = histTail;		/* new entry goes at the end */
890     if (histMerg && histMerg != &Histlist && histMerg != Histlist.Hnext) {
891 	/* Check above and below previous insertion point, in case we're adding
892 	 * sequential times in the middle of the list (e.g. history -M). */
893 	if (histMerg->Htime >= np->Htime)
894 	    pTime = histMerg;
895 	else if (histMerg->Hprev->Htime >= np->Htime)
896 	    pTime = histMerg->Hprev;
897     }
898     if (pTime) {
899         /* With hint, search up the list until Htime is greater.  We skip past
900          * the equal ones, too, so our caller can elide duplicates. */
901         pp = pTime;
902         while (pp != &Histlist && pp->Htime <= np->Htime)
903             pp = pp->Hprev;
904     } else
905         pp = &Histlist;
906     /* Search down the list while current entry's time is too large. */
907     while ((p = pp->Hnext) && (p->Htime > np->Htime))
908             pp = p;                     /* advance insertion point */
909     /* Remember recent position as hint for next time */
910     histMerg = pp;
911     return pp;
912 }
913 
914 /* Bubble Hnum & Href in new entry down to pp through earlier part of list. */
915 PG_STATIC void bubbleHnumHrefDown(struct Hist *np, struct Hist *pp)
916 {
917     struct Hist *p;
918     for (p = Histlist.Hnext; p != pp->Hnext; p = p->Hnext) {
919         /* swap Hnum & Href values of p and np. */
920         int n = p->Hnum, r = p->Href;
921         p->Hnum = np->Hnum; p->Href = np->Href;
922         np->Hnum = n; np->Href = r;
923     }
924 }
925 
926 /* Enter new command into the history list according to current settings. */
927 struct Hist *
928 enthist(
929   int event,				/* newly incremented global eventno */
930   struct wordent *lp,
931   int docopy,
932   int mflg,				/* true if merge requested */
933   int hlen)				/* -1 if unknown */
934 {
935     struct Hist *p = NULL, *pp = &Histlist, *pTime = NULL;
936     struct Hist *np;
937     const Char *dp;
938     unsigned lpHash = 0;                /* non-zero if hashing entries */
939 
940     if ((dp = varval(STRhistdup)) != STRNULL) {
941 	if (eq(dp, STRerase)) {
942 	    /* masaoki@akebono.tky.hp.com (Kobayashi Masaoki) */
943             createHistHashTable(hlen);
944             lpHash = hashhist(lp);
945             assert(lpHash != 0);
946             p = findHistHashTable(lp, lpHash);
947             if (p) {
948 		if (Htime != 0 && p->Htime > Htime)
949 		    Htime = p->Htime;
950                 /* If we are merging, and the old entry is at the place we want
951                  * to insert the new entry, then remember the place. */
952                 if (mflg && Htime != 0 && p->Hprev->Htime >= Htime)
953                     pTime = p->Hprev;
954 		if (!fastMergeErase)
955 		    renumberHist(p);	/* Reset Href of subsequent entries */
956                 hremove(p);
957 		hfree(p);
958                 p = NULL;               /* so new entry is allocated below */
959 	    }
960 	}
961 	else if (eq(dp, STRall)) {
962             createHistHashTable(hlen);
963             lpHash = hashhist(lp);
964             assert(lpHash != 0);
965             p = findHistHashTable(lp, lpHash);
966 	    if (p)   /* p!=NULL, only update this entry's Htime below */
967 		eventno--;		/* not adding a new event */
968 	}
969 	else if (eq(dp, STRprev)) {
970 	    if (pp->Hnext && heq(lp, &(pp->Hnext->Hlex))) {
971 		p = pp->Hnext;
972 		eventno--;
973 	    }
974 	}
975     }
976 
977     np = p ? p : xmalloc(sizeof(*np));
978 
979     /* Pick up timestamp set by lex() in Htime if reading saved history */
980     if (Htime != 0) {
981 	np->Htime = Htime;
982 	Htime = 0;
983     }
984     else
985         (void) time(&(np->Htime));
986 
987     if (p == np)
988         return np;                      /* reused existing entry */
989 
990     /* Initialize the new entry. */
991     np->Hnum = np->Href = event;
992     if (docopy) {
993         copylex(&np->Hlex, lp);
994 	if (histvalid)
995 	    np->histline = Strsave(histline.s);
996 	else
997 	    np->histline = NULL;
998     }
999     else {
1000 	np->Hlex.next = lp->next;
1001 	lp->next->prev = &np->Hlex;
1002 	np->Hlex.prev = lp->prev;
1003         lp->prev->next = &np->Hlex;
1004         np->histline = NULL;
1005     }
1006     np->Hhash = 0;
1007 
1008     /* The head of history list is the default insertion point.
1009        If merging, advance insertion point, in pp, according to Htime. */
1010     /* XXX -- In histdup=all, Htime values can be non-monotonic. */
1011     if (mflg) {                         /* merge according to np->Htime */
1012         pp = mergeInsertionPoint(np, pTime);
1013         for (p = pp->Hnext; p && p->Htime == np->Htime; pp = p, p = p->Hnext) {
1014             if (heq(&p->Hlex, &np->Hlex)) {
1015                 eventno--;              /* duplicate, so don't add new event */
1016                 hfree(np);
1017                 return (p);
1018               }
1019           }
1020         /* pp is now the last entry with time >= to np. */
1021 	if (!fastMergeErase) {		/* renumber at end of loadhist */
1022 	    /* Before inserting np after pp, bubble its Hnum & Href values down
1023 	     * through the earlier part of list. */
1024 	    bubbleHnumHrefDown(np, pp);
1025 	}
1026     }
1027     else
1028         pp = &Histlist;                 /* insert at beginning of history */
1029     hinsert(np, pp);
1030     if (lpHash && hlen != 0)		/* erase & all modes use hash table */
1031         insertHistHashTable(np, lpHash);
1032     else
1033         discardHistHashTable();
1034     return (np);
1035 }
1036 
1037 static void
1038 hfree(struct Hist *hp)
1039 {
1040     assert(hp != histMerg);
1041     if (hp->Hhash)
1042         removeHistHashTable(hp);
1043     freelex(&hp->Hlex);
1044     if (hp->histline)
1045         xfree(hp->histline);
1046     xfree(hp);
1047 }
1048 
1049 PG_STATIC void
1050 phist(struct Hist *hp, int hflg)
1051 {
1052     if (hp->Href < 0)
1053 	return;
1054     if (hflg & HIST_ONLY) {
1055 	int old_output_raw;
1056 
1057        /*
1058         * Control characters have to be written as is (output_raw).
1059         * This way one can preserve special characters (like tab) in
1060         * the history file.
1061         * From: mveksler@vnet.ibm.com (Veksler Michael)
1062         */
1063 	old_output_raw = output_raw;
1064         output_raw = 1;
1065 	cleanup_push(&old_output_raw, output_raw_restore);
1066 	if (hflg & HIST_TIME)
1067 	    /*
1068 	     * Make file entry with history time in format:
1069 	     * "+NNNNNNNNNN" (10 digits, left padded with ascii '0')
1070 	     */
1071 
1072 	    xprintf("#+%010lu\n", (unsigned long)hp->Htime);
1073 
1074 	if (HistLit && hp->histline)
1075 	    xprintf("%S\n", hp->histline);
1076 	else
1077 	    prlex(&hp->Hlex);
1078         cleanup_until(&old_output_raw);
1079     }
1080     else {
1081 	Char   *cp = str2short("%h\t%T\t%R\n");
1082 	Char *p;
1083 	struct varent *vp = adrof(STRhistory);
1084 
1085 	if (vp && vp->vec != NULL && vp->vec[0] && vp->vec[1])
1086 	    cp = vp->vec[1];
1087 
1088 	p = tprintf(FMT_HISTORY, cp, NULL, hp->Htime, hp);
1089 	cleanup_push(p, xfree);
1090 	for (cp = p; *cp;)
1091 	    xputwchar(*cp++);
1092 	cleanup_until(p);
1093     }
1094 }
1095 
1096 PG_STATIC void
1097 dophist(int n, int hflg)
1098 {
1099     struct Hist *hp;
1100     if (setintr) {
1101 	int old_pintr_disabled;
1102 
1103 	pintr_push_enable(&old_pintr_disabled);
1104 	cleanup_until(&old_pintr_disabled);
1105     }
1106     if ((hflg & HIST_REV) == 0) {
1107 	/* Since the history list is stored most recent first, non-reversing
1108 	 * print needs to print (backwards) up the list. */
1109 	if ((unsigned)n >= histCount)
1110 	    hp = histTail;
1111 	else {
1112 	    for (hp = Histlist.Hnext;
1113 		 --n > 0 && hp->Hnext != NULL;
1114 		 hp = hp->Hnext)
1115 		;
1116 	}
1117 	if (hp == NULL)
1118 	    return;			/* nothing to print */
1119 	for (; hp != &Histlist; hp = hp->Hprev)
1120 	    phist(hp, hflg);
1121     } else {
1122 	for (hp = Histlist.Hnext; n-- > 0 && hp != NULL; hp = hp->Hnext)
1123 	    phist(hp, hflg);
1124     }
1125 }
1126 
1127 /*ARGSUSED*/
1128 void
1129 dohist(Char **vp, struct command *c)
1130 {
1131     int     n, hflg = 0;
1132 
1133     USE(c);
1134     if (getn(varval(STRhistory)) == 0)
1135 	return;
1136     while (*++vp && **vp == '-') {
1137 	Char   *vp2 = *vp;
1138 
1139 	while (*++vp2)
1140 	    switch (*vp2) {
1141 	    case 'c':
1142 		hflg |= HIST_CLEAR;
1143 		break;
1144 	    case 'h':
1145 		hflg |= HIST_ONLY;
1146 		break;
1147 	    case 'r':
1148 		hflg |= HIST_REV;
1149 		break;
1150 	    case 'S':
1151 		hflg |= HIST_SAVE;
1152 		break;
1153 	    case 'L':
1154 		hflg |= HIST_LOAD;
1155 		break;
1156 	    case 'M':
1157 	    	hflg |= HIST_MERGE;
1158 		break;
1159 	    case 'T':
1160 	    	hflg |= HIST_TIME;
1161 		break;
1162 	    default:
1163 		stderror(ERR_HISTUS, "chrSLMT");
1164 		break;
1165 	    }
1166     }
1167     if (hflg & HIST_CLEAR) {
1168         struct Hist *np, *hp;
1169         for (hp = &Histlist; (np = hp->Hnext) != NULL;)
1170             hremove(np), hfree(np);
1171     }
1172 
1173     if (hflg & (HIST_LOAD | HIST_MERGE))
1174 	loadhist(*vp, (hflg & HIST_MERGE) ? 1 : 0);
1175     else if (hflg & HIST_SAVE)
1176 	rechist(*vp, 1);
1177     else {
1178 	if (*vp)
1179 	    n = getn(*vp);
1180 	else {
1181 	    n = getn(varval(STRhistory));
1182 	}
1183 	dophist(n, hflg);
1184     }
1185 }
1186 
1187 
1188 char *
1189 fmthist(int fmt, ptr_t ptr)
1190 {
1191     struct Hist *hp = ptr;
1192     char *buf;
1193 
1194     switch (fmt) {
1195     case 'h':
1196 	return xasprintf("%6d", hp->Hnum);
1197     case 'R':
1198 	if (HistLit && hp->histline)
1199 	    return xasprintf("%S", hp->histline);
1200 	else {
1201 	    Char *istr, *ip;
1202 	    char *p;
1203 
1204 	    istr = sprlex(&hp->Hlex);
1205 	    buf = xmalloc(Strlen(istr) * MB_LEN_MAX + 1);
1206 
1207 	    for (p = buf, ip = istr; *ip != '\0'; ip++)
1208 		p += one_wctomb(p, *ip);
1209 
1210 	    *p = '\0';
1211 	    xfree(istr);
1212 	    return buf;
1213 	}
1214     default:
1215 	buf = xmalloc(1);
1216 	buf[0] = '\0';
1217 	return buf;
1218     }
1219 }
1220 
1221 static void
1222 dotlock_cleanup(void* lockpath)
1223 {
1224 	dot_unlock((char*)lockpath);
1225 }
1226 
1227 /* Save history before exiting the shell. */
1228 void
1229 rechist(Char *xfname, int ref)
1230 {
1231     Char    *snum, *rs;
1232     int     fp, ftmp, oldidfds, ophup_disabled;
1233     struct varent *shist;
1234     char path[MAXPATHLEN];
1235     struct stat st;
1236     static Char *fname;
1237     static Char   *dumphist[] = {STRhistory, STRmhT, 0, 0};
1238 
1239     if (fname == NULL && !ref)
1240 	return;
1241 
1242     fname = xfname;
1243     ophup_disabled = phup_disabled;
1244     phup_disabled = 1;
1245 
1246     /*
1247      * If $savehist is just set, we use the value of $history
1248      * else we use the value in $savehist
1249      */
1250     if (((snum = varval(STRsavehist)) == STRNULL) &&
1251 	((snum = varval(STRhistory)) == STRNULL))
1252 	snum = STRmaxint;
1253 
1254 
1255     if (fname == NULL) {
1256 	if ((fname = varval(STRhistfile)) == STRNULL)
1257 	    fname = Strspl(varval(STRhome), &STRtildothist[1]);
1258 	else
1259 	    fname = Strsave(fname);
1260     }
1261     else
1262 	fname = globone(fname, G_ERROR);
1263     cleanup_push(fname, xfree);
1264 
1265     /*
1266      * The 'savehist merge' feature is intended for an environment
1267      * with numerous shells being in simultaneous use. Imagine
1268      * any kind of window system. All these shells 'share' the same
1269      * ~/.history file for recording their command line history.
1270      * We try to handle the case of multiple shells trying to merge
1271      * histories at the same time, by creating semi-unique filenames
1272      * and saving the history there first and then trying to rename
1273      * them in the proper history file.
1274      *
1275      * Users that like to nuke their environment require here an atomic
1276      * loadhist-creat-dohist(dumphist)-close sequence which is given
1277 		 * by optional lock parameter to savehist.
1278      *
1279      * jw.
1280      */
1281     /*
1282      * We need the didfds stuff before loadhist otherwise
1283      * exec in a script will fail to print if merge is set.
1284      * From: mveksler@iil.intel.com (Veksler Michael)
1285      */
1286     oldidfds = didfds;
1287     didfds = 0;
1288     if ((shist = adrof(STRsavehist)) != NULL && shist->vec != NULL) {
1289 	size_t i;
1290 	int merge = 0, lock = 0;
1291 
1292 	for (i = 1; shist->vec[i]; i++) {
1293 	    if (eq(shist->vec[i], STRmerge))
1294 		merge++;
1295 	    if (eq(shist->vec[i], STRlock))
1296 		lock++;
1297 	}
1298 
1299 	if (merge) {
1300 	    jmp_buf_t osetexit;
1301 	    if (lock) {
1302 #ifndef WINNT_NATIVE
1303 		char *lockpath = strsave(short2str(fname));
1304 		cleanup_push(lockpath, xfree);
1305 		/* Poll in 100 miliseconds interval to obtain the lock. */
1306 		if ((dot_lock(lockpath, 100) == 0))
1307 		    cleanup_push(lockpath, dotlock_cleanup);
1308 #endif
1309 	    }
1310 	    getexit(osetexit);
1311 	    if (setexit() == 0)
1312 		loadhist(fname, 1);
1313 	    resexit(osetexit);
1314 	}
1315     }
1316     rs = randsuf();
1317     xsnprintf(path, sizeof(path), "%S.%S", fname, rs);
1318     xfree(rs);
1319 
1320     fp = xcreat(path, 0600);
1321     if (fp == -1) {
1322 	didfds = oldidfds;
1323 	cleanup_until(fname);
1324 	phup_disabled = ophup_disabled;
1325 	return;
1326     }
1327     /* Try to preserve ownership and permissions of the original history file */
1328 #ifndef WINNT_NATIVE
1329     if (stat(short2str(fname), &st) != -1) {
1330 	TCSH_IGNORE(fchown(fp, st.st_uid, st.st_gid));
1331 	TCSH_IGNORE(fchmod(fp, st.st_mode));
1332     }
1333 #else
1334     UNREFERENCED_PARAMETER(st);
1335 #endif
1336     ftmp = SHOUT;
1337     SHOUT = fp;
1338     dumphist[2] = snum;
1339     dohist(dumphist, NULL);
1340     xclose(fp);
1341     SHOUT = ftmp;
1342     didfds = oldidfds;
1343 #ifndef WINNT_NATIVE
1344     (void)rename(path, short2str(fname));
1345 #else
1346     (void)ReplaceFile(short2str(fname), path, NULL, 0, NULL, NULL);
1347 #endif
1348     cleanup_until(fname);
1349     phup_disabled = ophup_disabled;
1350 }
1351 
1352 
1353 /* This is the entry point for loading history data from a file. */
1354 void
1355 loadhist(Char *fname, int mflg)
1356 {
1357     static Char   *loadhist_cmd[] = {STRsource, NULL, NULL, NULL};
1358     loadhist_cmd[1] = mflg ? STRmm : STRmh;
1359 
1360     if (fname != NULL)
1361 	loadhist_cmd[2] = fname;
1362     else if ((fname = varval(STRhistfile)) != STRNULL)
1363 	loadhist_cmd[2] = fname;
1364     else
1365 	loadhist_cmd[2] = STRtildothist;
1366 
1367     dosource(loadhist_cmd, NULL);
1368 
1369     /* During history merging (enthist sees mflg set), we disable management of
1370      * Hnum and Href (because fastMergeErase is true).  So now reset all the
1371      * values based on the final ordering of the history list. */
1372     if (mflg) {
1373 	int n = eventno;
1374         struct Hist *hp = &Histlist;
1375         while ((hp = hp->Hnext))
1376 	    hp->Hnum = hp->Href = n--;
1377     }
1378 }
1379 
1380 void
1381 sethistory(int n)
1382 {
1383     histlen = n;
1384     discardExcess(histlen);
1385 }
1386