xref: /rk3399_rockchip-uboot/lib/hashtable.c (revision 91a76751a090bf43c166fda0815c9b5b2bfccbe9)
1 /*
2  * This implementation is based on code from uClibc-0.9.30.3 but was
3  * modified and extended for use within U-Boot.
4  *
5  * Copyright (C) 2010 Wolfgang Denk <wd@denx.de>
6  *
7  * Original license header:
8  *
9  * Copyright (C) 1993, 1995, 1996, 1997, 2002 Free Software Foundation, Inc.
10  * This file is part of the GNU C Library.
11  * Contributed by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1993.
12  *
13  * The GNU C Library is free software; you can redistribute it and/or
14  * modify it under the terms of the GNU Lesser General Public
15  * License as published by the Free Software Foundation; either
16  * version 2.1 of the License, or (at your option) any later version.
17  *
18  * The GNU C Library is distributed in the hope that it will be useful,
19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
21  * Lesser General Public License for more details.
22  *
23  * You should have received a copy of the GNU Lesser General Public
24  * License along with the GNU C Library; if not, write to the Free
25  * Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
26  * 02111-1307 USA.
27  */
28 
29 #include <errno.h>
30 #include <malloc.h>
31 
32 #ifdef USE_HOSTCC		/* HOST build */
33 # include <string.h>
34 # include <assert.h>
35 
36 # ifndef debug
37 #  ifdef DEBUG
38 #   define debug(fmt,args...)	printf(fmt ,##args)
39 #  else
40 #   define debug(fmt,args...)
41 #  endif
42 # endif
43 #else				/* U-Boot build */
44 # include <common.h>
45 # include <linux/string.h>
46 #endif
47 
48 #include "search.h"
49 
50 /*
51  * [Aho,Sethi,Ullman] Compilers: Principles, Techniques and Tools, 1986
52  * [Knuth]            The Art of Computer Programming, part 3 (6.4)
53  */
54 
55 /*
56  * The non-reentrant version use a global space for storing the hash table.
57  */
58 static struct hsearch_data htab;
59 
60 /*
61  * The reentrant version has no static variables to maintain the state.
62  * Instead the interface of all functions is extended to take an argument
63  * which describes the current status.
64  */
65 typedef struct _ENTRY {
66 	unsigned int used;
67 	ENTRY entry;
68 } _ENTRY;
69 
70 
71 /*
72  * hcreate()
73  */
74 
75 /*
76  * For the used double hash method the table size has to be a prime. To
77  * correct the user given table size we need a prime test.  This trivial
78  * algorithm is adequate because
79  * a)  the code is (most probably) called a few times per program run and
80  * b)  the number is small because the table must fit in the core
81  * */
82 static int isprime(unsigned int number)
83 {
84 	/* no even number will be passed */
85 	unsigned int div = 3;
86 
87 	while (div * div < number && number % div != 0)
88 		div += 2;
89 
90 	return number % div != 0;
91 }
92 
93 int hcreate(size_t nel)
94 {
95 	return hcreate_r(nel, &htab);
96 }
97 
98 /*
99  * Before using the hash table we must allocate memory for it.
100  * Test for an existing table are done. We allocate one element
101  * more as the found prime number says. This is done for more effective
102  * indexing as explained in the comment for the hsearch function.
103  * The contents of the table is zeroed, especially the field used
104  * becomes zero.
105  */
106 int hcreate_r(size_t nel, struct hsearch_data *htab)
107 {
108 	/* Test for correct arguments.  */
109 	if (htab == NULL) {
110 		__set_errno(EINVAL);
111 		return 0;
112 	}
113 
114 	/* There is still another table active. Return with error. */
115 	if (htab->table != NULL)
116 		return 0;
117 
118 	/* Change nel to the first prime number not smaller as nel. */
119 	nel |= 1;		/* make odd */
120 	while (!isprime(nel))
121 		nel += 2;
122 
123 	htab->size = nel;
124 	htab->filled = 0;
125 
126 	/* allocate memory and zero out */
127 	htab->table = (_ENTRY *) calloc(htab->size + 1, sizeof(_ENTRY));
128 	if (htab->table == NULL)
129 		return 0;
130 
131 	/* everything went alright */
132 	return 1;
133 }
134 
135 
136 /*
137  * hdestroy()
138  */
139 void hdestroy(void)
140 {
141 	hdestroy_r(&htab);
142 }
143 
144 /*
145  * After using the hash table it has to be destroyed. The used memory can
146  * be freed and the local static variable can be marked as not used.
147  */
148 void hdestroy_r(struct hsearch_data *htab)
149 {
150 	int i;
151 
152 	/* Test for correct arguments.  */
153 	if (htab == NULL) {
154 		__set_errno(EINVAL);
155 		return;
156 	}
157 
158 	/* free used memory */
159 	for (i = 1; i <= htab->size; ++i) {
160 		if (htab->table[i].used) {
161 			ENTRY *ep = &htab->table[i].entry;
162 
163 			free(ep->key);
164 			free(ep->data);
165 		}
166 	}
167 	free(htab->table);
168 
169 	/* the sign for an existing table is an value != NULL in htable */
170 	htab->table = NULL;
171 }
172 
173 /*
174  * hsearch()
175  */
176 
177 /*
178  * This is the search function. It uses double hashing with open addressing.
179  * The argument item.key has to be a pointer to an zero terminated, most
180  * probably strings of chars. The function for generating a number of the
181  * strings is simple but fast. It can be replaced by a more complex function
182  * like ajw (see [Aho,Sethi,Ullman]) if the needs are shown.
183  *
184  * We use an trick to speed up the lookup. The table is created by hcreate
185  * with one more element available. This enables us to use the index zero
186  * special. This index will never be used because we store the first hash
187  * index in the field used where zero means not used. Every other value
188  * means used. The used field can be used as a first fast comparison for
189  * equality of the stored and the parameter value. This helps to prevent
190  * unnecessary expensive calls of strcmp.
191  *
192  * This implementation differs from the standard library version of
193  * this function in a number of ways:
194  *
195  * - While the standard version does not make any assumptions about
196  *   the type of the stored data objects at all, this implementation
197  *   works with NUL terminated strings only.
198  * - Instead of storing just pointers to the original objects, we
199  *   create local copies so the caller does not need to care about the
200  *   data any more.
201  * - The standard implementation does not provide a way to update an
202  *   existing entry.  This version will create a new entry or update an
203  *   existing one when both "action == ENTER" and "item.data != NULL".
204  * - Instead of returning 1 on success, we return the index into the
205  *   internal hash table, which is also guaranteed to be positive.
206  *   This allows us direct access to the found hash table slot for
207  *   example for functions like hdelete().
208  */
209 
210 ENTRY *hsearch(ENTRY item, ACTION action)
211 {
212 	ENTRY *result;
213 
214 	(void) hsearch_r(item, action, &result, &htab);
215 
216 	return result;
217 }
218 
219 int hsearch_r(ENTRY item, ACTION action, ENTRY ** retval,
220 	      struct hsearch_data *htab)
221 {
222 	unsigned int hval;
223 	unsigned int count;
224 	unsigned int len = strlen(item.key);
225 	unsigned int idx;
226 
227 	/* Compute an value for the given string. Perhaps use a better method. */
228 	hval = len;
229 	count = len;
230 	while (count-- > 0) {
231 		hval <<= 4;
232 		hval += item.key[count];
233 	}
234 
235 	/*
236 	 * First hash function:
237 	 * simply take the modul but prevent zero.
238 	 */
239 	hval %= htab->size;
240 	if (hval == 0)
241 		++hval;
242 
243 	/* The first index tried. */
244 	idx = hval;
245 
246 	if (htab->table[idx].used) {
247 		/*
248                  * Further action might be required according to the
249 		 * action value.
250 		 */
251 		unsigned hval2;
252 
253 		if (htab->table[idx].used == hval
254 		    && strcmp(item.key, htab->table[idx].entry.key) == 0) {
255 			/* Overwrite existing value? */
256 			if ((action == ENTER) && (item.data != NULL)) {
257 				free(htab->table[idx].entry.data);
258 				htab->table[idx].entry.data =
259 					strdup(item.data);
260 				if (!htab->table[idx].entry.data) {
261 					__set_errno(ENOMEM);
262 					*retval = NULL;
263 					return 0;
264 				}
265 			}
266 			/* return found entry */
267 			*retval = &htab->table[idx].entry;
268 			return idx;
269 		}
270 
271 		/*
272 		 * Second hash function:
273 		 * as suggested in [Knuth]
274 		 */
275 		hval2 = 1 + hval % (htab->size - 2);
276 
277 		do {
278 			/*
279                          * Because SIZE is prime this guarantees to
280                          * step through all available indices.
281 			 */
282 			if (idx <= hval2)
283 				idx = htab->size + idx - hval2;
284 			else
285 				idx -= hval2;
286 
287 			/*
288 			 * If we visited all entries leave the loop
289 			 * unsuccessfully.
290 			 */
291 			if (idx == hval)
292 				break;
293 
294 			/* If entry is found use it. */
295 			if ((htab->table[idx].used == hval)
296 			    && strcmp(item.key, htab->table[idx].entry.key) == 0) {
297 				/* Overwrite existing value? */
298 				if ((action == ENTER) && (item.data != NULL)) {
299 					free(htab->table[idx].entry.data);
300 					htab->table[idx].entry.data =
301 						strdup(item.data);
302 					if (!htab->table[idx].entry.data) {
303 						__set_errno(ENOMEM);
304 						*retval = NULL;
305 						return 0;
306 					}
307 				}
308 				/* return found entry */
309 				*retval = &htab->table[idx].entry;
310 				return idx;
311 			}
312 		}
313 		while (htab->table[idx].used);
314 	}
315 
316 	/* An empty bucket has been found. */
317 	if (action == ENTER) {
318 		/*
319                  * If table is full and another entry should be
320                  * entered return with error.
321 		 */
322 		if (htab->filled == htab->size) {
323 			__set_errno(ENOMEM);
324 			*retval = NULL;
325 			return 0;
326 		}
327 
328 		/*
329 		 * Create new entry;
330 		 * create copies of item.key and item.data
331 		 */
332 		htab->table[idx].used = hval;
333 		htab->table[idx].entry.key = strdup(item.key);
334 		htab->table[idx].entry.data = strdup(item.data);
335 		if (!htab->table[idx].entry.key ||
336 		    !htab->table[idx].entry.data) {
337 			__set_errno(ENOMEM);
338 			*retval = NULL;
339 			return 0;
340 		}
341 
342 		++htab->filled;
343 
344 		/* return new entry */
345 		*retval = &htab->table[idx].entry;
346 		return 1;
347 	}
348 
349 	__set_errno(ESRCH);
350 	*retval = NULL;
351 	return 0;
352 }
353 
354 
355 /*
356  * hdelete()
357  */
358 
359 /*
360  * The standard implementation of hsearch(3) does not provide any way
361  * to delete any entries from the hash table.  We extend the code to
362  * do that.
363  */
364 
365 int hdelete(const char *key)
366 {
367 	return hdelete_r(key, &htab);
368 }
369 
370 int hdelete_r(const char *key, struct hsearch_data *htab)
371 {
372 	ENTRY e, *ep;
373 	int idx;
374 
375 	debug("hdelete: DELETE key \"%s\"\n", key);
376 
377 	e.key = (char *)key;
378 
379 	if ((idx = hsearch_r(e, FIND, &ep, htab)) == 0) {
380 		__set_errno(ESRCH);
381 		return 0;	/* not found */
382 	}
383 
384 	/* free used ENTRY */
385 	debug("hdelete: DELETING key \"%s\"\n", key);
386 
387 	free(ep->key);
388 	free(ep->data);
389 	htab->table[idx].used = 0;
390 
391 	--htab->filled;
392 
393 	return 1;
394 }
395 
396 /*
397  * hexport()
398  */
399 
400 /*
401  * Export the data stored in the hash table in linearized form.
402  *
403  * Entries are exported as "name=value" strings, separated by an
404  * arbitrary (non-NUL, of course) separator character. This allows to
405  * use this function both when formatting the U-Boot environment for
406  * external storage (using '\0' as separator), but also when using it
407  * for the "printenv" command to print all variables, simply by using
408  * as '\n" as separator. This can also be used for new features like
409  * exporting the environment data as text file, including the option
410  * for later re-import.
411  *
412  * The entries in the result list will be sorted by ascending key
413  * values.
414  *
415  * If the separator character is different from NUL, then any
416  * separator characters and backslash characters in the values will
417  * be escaped by a preceeding backslash in output. This is needed for
418  * example to enable multi-line values, especially when the output
419  * shall later be parsed (for example, for re-import).
420  *
421  * There are several options how the result buffer is handled:
422  *
423  * *resp  size
424  * -----------
425  *  NULL    0	A string of sufficient length will be allocated.
426  *  NULL   >0	A string of the size given will be
427  *		allocated. An error will be returned if the size is
428  *		not sufficient.  Any unused bytes in the string will
429  *		be '\0'-padded.
430  * !NULL    0	The user-supplied buffer will be used. No length
431  *		checking will be performed, i. e. it is assumed that
432  *		the buffer size will always be big enough. DANGEROUS.
433  * !NULL   >0	The user-supplied buffer will be used. An error will
434  *		be returned if the size is not sufficient.  Any unused
435  *		bytes in the string will be '\0'-padded.
436  */
437 
438 ssize_t hexport(const char sep, char **resp, size_t size)
439 {
440 	return hexport_r(&htab, sep, resp, size);
441 }
442 
443 static int cmpkey(const void *p1, const void *p2)
444 {
445 	ENTRY *e1 = *(ENTRY **) p1;
446 	ENTRY *e2 = *(ENTRY **) p2;
447 
448 	return (strcmp(e1->key, e2->key));
449 }
450 
451 ssize_t hexport_r(struct hsearch_data *htab, const char sep,
452 		 char **resp, size_t size)
453 {
454 	ENTRY *list[htab->size];
455 	char *res, *p;
456 	size_t totlen;
457 	int i, n;
458 
459 	/* Test for correct arguments.  */
460 	if ((resp == NULL) || (htab == NULL)) {
461 		__set_errno(EINVAL);
462 		return (-1);
463 	}
464 
465 	debug("EXPORT  table = %p, htab.size = %d, htab.filled = %d, size = %d\n",
466 		htab, htab->size, htab->filled, size);
467 	/*
468 	 * Pass 1:
469 	 * search used entries,
470 	 * save addresses and compute total length
471 	 */
472 	for (i = 1, n = 0, totlen = 0; i <= htab->size; ++i) {
473 
474 		if (htab->table[i].used) {
475 			ENTRY *ep = &htab->table[i].entry;
476 
477 			list[n++] = ep;
478 
479 			totlen += strlen(ep->key) + 2;
480 
481 			if (sep == '\0') {
482 				totlen += strlen(ep->data);
483 			} else {	/* check if escapes are needed */
484 				char *s = ep->data;
485 
486 				while (*s) {
487 					++totlen;
488 					/* add room for needed escape chars */
489 					if ((*s == sep) || (*s == '\\'))
490 						++totlen;
491 					++s;
492 				}
493 			}
494 			totlen += 2;	/* for '=' and 'sep' char */
495 		}
496 	}
497 
498 #ifdef DEBUG
499 	/* Pass 1a: print unsorted list */
500 	printf("Unsorted: n=%d\n", n);
501 	for (i = 0; i < n; ++i) {
502 		printf("\t%3d: %p ==> %-10s => %s\n",
503 		       i, list[i], list[i]->key, list[i]->data);
504 	}
505 #endif
506 
507 	/* Sort list by keys */
508 	qsort(list, n, sizeof(ENTRY *), cmpkey);
509 
510 	/* Check if the user supplied buffer size is sufficient */
511 	if (size) {
512 		if (size < totlen + 1) {	/* provided buffer too small */
513 			debug("### buffer too small: %d, but need %d\n",
514 				size, totlen + 1);
515 			__set_errno(ENOMEM);
516 			return (-1);
517 		}
518 	} else {
519 		size = totlen + 1;
520 	}
521 
522 	/* Check if the user provided a buffer */
523 	if (*resp) {
524 		/* yes; clear it */
525 		res = *resp;
526 		memset(res, '\0', size);
527 	} else {
528 		/* no, allocate and clear one */
529 		*resp = res = calloc(1, size);
530 		if (res == NULL) {
531 			__set_errno(ENOMEM);
532 			return (-1);
533 		}
534 	}
535 	/*
536 	 * Pass 2:
537 	 * export sorted list of result data
538 	 */
539 	for (i = 0, p = res; i < n; ++i) {
540 		char *s;
541 
542 		s = list[i]->key;
543 		while (*s)
544 			*p++ = *s++;
545 		*p++ = '=';
546 
547 		s = list[i]->data;
548 
549 		while (*s) {
550 			if ((*s == sep) || (*s == '\\'))
551 				*p++ = '\\';	/* escape */
552 			*p++ = *s++;
553 		}
554 		*p++ = sep;
555 	}
556 	*p = '\0';		/* terminate result */
557 
558 	return size;
559 }
560 
561 
562 /*
563  * himport()
564  */
565 
566 /*
567  * Import linearized data into hash table.
568  *
569  * This is the inverse function to hexport(): it takes a linear list
570  * of "name=value" pairs and creates hash table entries from it.
571  *
572  * Entries without "value", i. e. consisting of only "name" or
573  * "name=", will cause this entry to be deleted from the hash table.
574  *
575  * The "flag" argument can be used to control the behaviour: when the
576  * H_NOCLEAR bit is set, then an existing hash table will kept, i. e.
577  * new data will be added to an existing hash table; otherwise, old
578  * data will be discarded and a new hash table will be created.
579  *
580  * The separator character for the "name=value" pairs can be selected,
581  * so we both support importing from externally stored environment
582  * data (separated by NUL characters) and from plain text files
583  * (entries separated by newline characters).
584  *
585  * To allow for nicely formatted text input, leading white space
586  * (sequences of SPACE and TAB chars) is ignored, and entries starting
587  * (after removal of any leading white space) with a '#' character are
588  * considered comments and ignored.
589  *
590  * [NOTE: this means that a variable name cannot start with a '#'
591  * character.]
592  *
593  * When using a non-NUL separator character, backslash is used as
594  * escape character in the value part, allowing for example for
595  * multi-line values.
596  *
597  * In theory, arbitrary separator characters can be used, but only
598  * '\0' and '\n' have really been tested.
599  */
600 
601 int himport(const char *env, size_t size, const char sep, int flag)
602 {
603 	return himport_r(&htab, env, size, sep, flag);
604 }
605 
606 int himport_r(struct hsearch_data *htab,
607 	      const char *env, size_t size, const char sep, int flag)
608 {
609 	char *data, *sp, *dp, *name, *value;
610 
611 	/* Test for correct arguments.  */
612 	if (htab == NULL) {
613 		__set_errno(EINVAL);
614 		return 0;
615 	}
616 
617 	/* we allocate new space to make sure we can write to the array */
618 	if ((data = malloc(size)) == NULL) {
619 		debug("himport_r: can't malloc %d bytes\n", size);
620 		__set_errno(ENOMEM);
621 		return 0;
622 	}
623 	memcpy(data, env, size);
624 	dp = data;
625 
626 	if ((flag & H_NOCLEAR) == 0) {
627 		/* Destroy old hash table if one exists */
628 		debug("Destroy Hash Table: %p table = %p\n", htab,
629 		       htab->table);
630 		if (htab->table)
631 			hdestroy_r(htab);
632 	}
633 
634 	/*
635 	 * Create new hash table (if needed).  The computation of the hash
636 	 * table size is based on heuristics: in a sample of some 70+
637 	 * existing systems we found an average size of 39+ bytes per entry
638 	 * in the environment (for the whole key=value pair). Assuming a
639 	 * size of 7 per entry (= safety factor of >5) should provide enough
640 	 * safety margin for any existing environment definitons and still
641 	 * allow for more than enough dynamic additions. Note that the
642 	 * "size" argument is supposed to give the maximum enviroment size
643 	 * (CONFIG_ENV_SIZE).
644 	 */
645 
646 	if (!htab->table) {
647 		int nent = size / 7;
648 
649 		debug("Create Hash Table: N=%d\n", nent);
650 
651 		if (hcreate_r(nent, htab) == 0) {
652 			free(data);
653 			return 0;
654 		}
655 	}
656 
657 	/* Parse environment; allow for '\0' and 'sep' as separators */
658 	do {
659 		ENTRY e, *rv;
660 
661 		/* skip leading white space */
662 		while ((*dp == ' ') || (*dp == '\t'))
663 			++dp;
664 
665 		/* skip comment lines */
666 		if (*dp == '#') {
667 			while (*dp && (*dp != sep))
668 				++dp;
669 			++dp;
670 			continue;
671 		}
672 
673 		/* parse name */
674 		for (name = dp; *dp != '=' && *dp && *dp != sep; ++dp)
675 			;
676 
677 		/* deal with "name" and "name=" entries (delete var) */
678 		if (*dp == '\0' || *(dp + 1) == '\0' ||
679 		    *dp == sep || *(dp + 1) == sep) {
680 			if (*dp == '=')
681 				*dp++ = '\0';
682 			*dp++ = '\0';	/* terminate name */
683 
684 			debug("DELETE CANDIDATE: \"%s\"\n", name);
685 
686 			if (hdelete_r(name, htab) == 0)
687 				debug("DELETE ERROR ##############################\n");
688 
689 			continue;
690 		}
691 		*dp++ = '\0';	/* terminate name */
692 
693 		/* parse value; deal with escapes */
694 		for (value = sp = dp; *dp && (*dp != sep); ++dp) {
695 			if ((*dp == '\\') && *(dp + 1))
696 				++dp;
697 			*sp++ = *dp;
698 		}
699 		*sp++ = '\0';	/* terminate value */
700 		++dp;
701 
702 		/* enter into hash table */
703 		e.key = name;
704 		e.data = value;
705 
706 		hsearch_r(e, ENTER, &rv, htab);
707 		if (rv == NULL) {
708 			printf("himport_r: can't insert \"%s=%s\" into hash table\n", name, value);
709 			return 0;
710 		}
711 
712 		debug("INSERT: %p ==> name=\"%s\" value=\"%s\"\n", rv, name,
713 		       value);
714 		debug("        table = %p, size = %d, filled = %d\n", htab,
715 		       htab->size, htab->filled);
716 	} while ((dp < data + size) && *dp);	/* size check needed for text */
717 						/* without '\0' termination */
718 	free(data);
719 
720 	return 1;		/* everything OK */
721 }
722