123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444 |
- /*
- * Copyright 1995-2024 The OpenSSL Project Authors. All Rights Reserved.
- *
- * Licensed under the Apache License 2.0 (the "License"). You may not use
- * this file except in compliance with the License. You can obtain a copy
- * in the file LICENSE in the source distribution or at
- * https://www.openssl.org/source/license.html
- */
- #include <stdio.h>
- #include <string.h>
- #include <stdlib.h>
- #include <openssl/crypto.h>
- #include <openssl/lhash.h>
- #include <openssl/err.h>
- #include "crypto/ctype.h"
- #include "crypto/lhash.h"
- #include "lhash_local.h"
- /*
- * A hashing implementation that appears to be based on the linear hashing
- * algorithm:
- * https://en.wikipedia.org/wiki/Linear_hashing
- *
- * Litwin, Witold (1980), "Linear hashing: A new tool for file and table
- * addressing", Proc. 6th Conference on Very Large Databases: 212-223
- * https://hackthology.com/pdfs/Litwin-1980-Linear_Hashing.pdf
- *
- * From the Wikipedia article "Linear hashing is used in the BDB Berkeley
- * database system, which in turn is used by many software systems such as
- * OpenLDAP, using a C implementation derived from the CACM article and first
- * published on the Usenet in 1988 by Esmond Pitt."
- *
- * The CACM paper is available here:
- * https://pdfs.semanticscholar.org/ff4d/1c5deca6269cc316bfd952172284dbf610ee.pdf
- */
- #undef MIN_NODES
- #define MIN_NODES 16
- #define UP_LOAD (2*LH_LOAD_MULT) /* load times 256 (default 2) */
- #define DOWN_LOAD (LH_LOAD_MULT) /* load times 256 (default 1) */
- static int expand(OPENSSL_LHASH *lh);
- static void contract(OPENSSL_LHASH *lh);
- static OPENSSL_LH_NODE **getrn(OPENSSL_LHASH *lh, const void *data, unsigned long *rhash);
- OPENSSL_LHASH *OPENSSL_LH_set_thunks(OPENSSL_LHASH *lh,
- OPENSSL_LH_HASHFUNCTHUNK hw,
- OPENSSL_LH_COMPFUNCTHUNK cw,
- OPENSSL_LH_DOALL_FUNC_THUNK daw,
- OPENSSL_LH_DOALL_FUNCARG_THUNK daaw)
- {
- if (lh == NULL)
- return NULL;
- lh->compw = cw;
- lh->hashw = hw;
- lh->daw = daw;
- lh->daaw = daaw;
- return lh;
- }
- OPENSSL_LHASH *OPENSSL_LH_new(OPENSSL_LH_HASHFUNC h, OPENSSL_LH_COMPFUNC c)
- {
- OPENSSL_LHASH *ret;
- if ((ret = OPENSSL_zalloc(sizeof(*ret))) == NULL)
- return NULL;
- if ((ret->b = OPENSSL_zalloc(sizeof(*ret->b) * MIN_NODES)) == NULL)
- goto err;
- ret->comp = ((c == NULL) ? (OPENSSL_LH_COMPFUNC)strcmp : c);
- ret->hash = ((h == NULL) ? (OPENSSL_LH_HASHFUNC)OPENSSL_LH_strhash : h);
- ret->num_nodes = MIN_NODES / 2;
- ret->num_alloc_nodes = MIN_NODES;
- ret->pmax = MIN_NODES / 2;
- ret->up_load = UP_LOAD;
- ret->down_load = DOWN_LOAD;
- return ret;
- err:
- OPENSSL_free(ret->b);
- OPENSSL_free(ret);
- return NULL;
- }
- void OPENSSL_LH_free(OPENSSL_LHASH *lh)
- {
- if (lh == NULL)
- return;
- OPENSSL_LH_flush(lh);
- OPENSSL_free(lh->b);
- OPENSSL_free(lh);
- }
- void OPENSSL_LH_flush(OPENSSL_LHASH *lh)
- {
- unsigned int i;
- OPENSSL_LH_NODE *n, *nn;
- if (lh == NULL)
- return;
- for (i = 0; i < lh->num_nodes; i++) {
- n = lh->b[i];
- while (n != NULL) {
- nn = n->next;
- OPENSSL_free(n);
- n = nn;
- }
- lh->b[i] = NULL;
- }
- lh->num_items = 0;
- }
- void *OPENSSL_LH_insert(OPENSSL_LHASH *lh, void *data)
- {
- unsigned long hash;
- OPENSSL_LH_NODE *nn, **rn;
- void *ret;
- lh->error = 0;
- if ((lh->up_load <= (lh->num_items * LH_LOAD_MULT / lh->num_nodes)) && !expand(lh))
- return NULL; /* 'lh->error++' already done in 'expand' */
- rn = getrn(lh, data, &hash);
- if (*rn == NULL) {
- if ((nn = OPENSSL_malloc(sizeof(*nn))) == NULL) {
- lh->error++;
- return NULL;
- }
- nn->data = data;
- nn->next = NULL;
- nn->hash = hash;
- *rn = nn;
- ret = NULL;
- lh->num_items++;
- } else { /* replace same key */
- ret = (*rn)->data;
- (*rn)->data = data;
- }
- return ret;
- }
- void *OPENSSL_LH_delete(OPENSSL_LHASH *lh, const void *data)
- {
- unsigned long hash;
- OPENSSL_LH_NODE *nn, **rn;
- void *ret;
- lh->error = 0;
- rn = getrn(lh, data, &hash);
- if (*rn == NULL) {
- return NULL;
- } else {
- nn = *rn;
- *rn = nn->next;
- ret = nn->data;
- OPENSSL_free(nn);
- }
- lh->num_items--;
- if ((lh->num_nodes > MIN_NODES) &&
- (lh->down_load >= (lh->num_items * LH_LOAD_MULT / lh->num_nodes)))
- contract(lh);
- return ret;
- }
- void *OPENSSL_LH_retrieve(OPENSSL_LHASH *lh, const void *data)
- {
- unsigned long hash;
- OPENSSL_LH_NODE **rn;
- if (lh->error != 0)
- lh->error = 0;
- rn = getrn(lh, data, &hash);
- return *rn == NULL ? NULL : (*rn)->data;
- }
- static void doall_util_fn(OPENSSL_LHASH *lh, int use_arg,
- OPENSSL_LH_DOALL_FUNC_THUNK wfunc,
- OPENSSL_LH_DOALL_FUNC func,
- OPENSSL_LH_DOALL_FUNCARG func_arg,
- OPENSSL_LH_DOALL_FUNCARG_THUNK wfunc_arg,
- void *arg)
- {
- int i;
- OPENSSL_LH_NODE *a, *n;
- if (lh == NULL)
- return;
- /*
- * reverse the order so we search from 'top to bottom' We were having
- * memory leaks otherwise
- */
- for (i = lh->num_nodes - 1; i >= 0; i--) {
- a = lh->b[i];
- while (a != NULL) {
- n = a->next;
- if (use_arg)
- wfunc_arg(a->data, arg, func_arg);
- else
- wfunc(a->data, func);
- a = n;
- }
- }
- }
- void OPENSSL_LH_doall(OPENSSL_LHASH *lh, OPENSSL_LH_DOALL_FUNC func)
- {
- if (lh == NULL)
- return;
- doall_util_fn(lh, 0, lh->daw, func, (OPENSSL_LH_DOALL_FUNCARG)NULL,
- (OPENSSL_LH_DOALL_FUNCARG_THUNK)NULL, NULL);
- }
- void OPENSSL_LH_doall_arg(OPENSSL_LHASH *lh,
- OPENSSL_LH_DOALL_FUNCARG func, void *arg)
- {
- if (lh == NULL)
- return;
- doall_util_fn(lh, 1, (OPENSSL_LH_DOALL_FUNC_THUNK)NULL,
- (OPENSSL_LH_DOALL_FUNC)NULL, func, lh->daaw, arg);
- }
- void OPENSSL_LH_doall_arg_thunk(OPENSSL_LHASH *lh,
- OPENSSL_LH_DOALL_FUNCARG_THUNK daaw,
- OPENSSL_LH_DOALL_FUNCARG fn, void *arg)
- {
- doall_util_fn(lh, 1, (OPENSSL_LH_DOALL_FUNC_THUNK)NULL,
- (OPENSSL_LH_DOALL_FUNC)NULL, fn, daaw, arg);
- }
- static int expand(OPENSSL_LHASH *lh)
- {
- OPENSSL_LH_NODE **n, **n1, **n2, *np;
- unsigned int p, pmax, nni, j;
- unsigned long hash;
- nni = lh->num_alloc_nodes;
- p = lh->p;
- pmax = lh->pmax;
- if (p + 1 >= pmax) {
- j = nni * 2;
- n = OPENSSL_realloc(lh->b, sizeof(OPENSSL_LH_NODE *) * j);
- if (n == NULL) {
- lh->error++;
- return 0;
- }
- lh->b = n;
- memset(n + nni, 0, sizeof(*n) * (j - nni));
- lh->pmax = nni;
- lh->num_alloc_nodes = j;
- lh->p = 0;
- } else {
- lh->p++;
- }
- lh->num_nodes++;
- n1 = &(lh->b[p]);
- n2 = &(lh->b[p + pmax]);
- *n2 = NULL;
- for (np = *n1; np != NULL;) {
- hash = np->hash;
- if ((hash % nni) != p) { /* move it */
- *n1 = (*n1)->next;
- np->next = *n2;
- *n2 = np;
- } else
- n1 = &((*n1)->next);
- np = *n1;
- }
- return 1;
- }
- static void contract(OPENSSL_LHASH *lh)
- {
- OPENSSL_LH_NODE **n, *n1, *np;
- np = lh->b[lh->p + lh->pmax - 1];
- lh->b[lh->p + lh->pmax - 1] = NULL; /* 24/07-92 - eay - weird but :-( */
- if (lh->p == 0) {
- n = OPENSSL_realloc(lh->b,
- (unsigned int)(sizeof(OPENSSL_LH_NODE *) * lh->pmax));
- if (n == NULL) {
- /* fputs("realloc error in lhash", stderr); */
- lh->error++;
- } else {
- lh->b = n;
- }
- lh->num_alloc_nodes /= 2;
- lh->pmax /= 2;
- lh->p = lh->pmax - 1;
- } else
- lh->p--;
- lh->num_nodes--;
- n1 = lh->b[(int)lh->p];
- if (n1 == NULL)
- lh->b[(int)lh->p] = np;
- else {
- while (n1->next != NULL)
- n1 = n1->next;
- n1->next = np;
- }
- }
- static OPENSSL_LH_NODE **getrn(OPENSSL_LHASH *lh,
- const void *data, unsigned long *rhash)
- {
- OPENSSL_LH_NODE **ret, *n1;
- unsigned long hash, nn;
- if (lh->hashw != NULL)
- hash = lh->hashw(data, lh->hash);
- else
- hash = lh->hash(data);
- *rhash = hash;
- nn = hash % lh->pmax;
- if (nn < lh->p)
- nn = hash % lh->num_alloc_nodes;
- ret = &(lh->b[(int)nn]);
- for (n1 = *ret; n1 != NULL; n1 = n1->next) {
- if (n1->hash != hash) {
- ret = &(n1->next);
- continue;
- }
- if (lh->compw != NULL) {
- if (lh->compw(n1->data, data, lh->comp) == 0)
- break;
- } else {
- if (lh->comp(n1->data, data) == 0)
- break;
- }
- ret = &(n1->next);
- }
- return ret;
- }
- /*
- * The following hash seems to work very well on normal text strings no
- * collisions on /usr/dict/words and it distributes on %2^n quite well, not
- * as good as MD5, but still good.
- */
- unsigned long OPENSSL_LH_strhash(const char *c)
- {
- unsigned long ret = 0;
- long n;
- unsigned long v;
- int r;
- if ((c == NULL) || (*c == '\0'))
- return ret;
- n = 0x100;
- while (*c) {
- v = n | (*c);
- n += 0x100;
- r = (int)((v >> 2) ^ v) & 0x0f;
- /* cast to uint64_t to avoid 32 bit shift of 32 bit value */
- ret = (ret << r) | (unsigned long)((uint64_t)ret >> (32 - r));
- ret &= 0xFFFFFFFFL;
- ret ^= v * v;
- c++;
- }
- return (ret >> 16) ^ ret;
- }
- /*
- * Case insensitive string hashing.
- *
- * The lower/upper case bit is masked out (forcing all letters to be capitals).
- * The major side effect on non-alpha characters is mapping the symbols and
- * digits into the control character range (which should be harmless).
- * The duplication (with respect to the hash value) of printable characters
- * are that '`', '{', '|', '}' and '~' map to '@', '[', '\', ']' and '^'
- * respectively (which seems tolerable).
- *
- * For EBCDIC, the alpha mapping is to lower case, most symbols go to control
- * characters. The only duplication is '0' mapping to '^', which is better
- * than for ASCII.
- */
- unsigned long ossl_lh_strcasehash(const char *c)
- {
- unsigned long ret = 0;
- long n;
- unsigned long v;
- int r;
- #if defined(CHARSET_EBCDIC) && !defined(CHARSET_EBCDIC_TEST)
- const long int case_adjust = ~0x40;
- #else
- const long int case_adjust = ~0x20;
- #endif
- if (c == NULL || *c == '\0')
- return ret;
- for (n = 0x100; *c != '\0'; n += 0x100) {
- v = n | (case_adjust & *c);
- r = (int)((v >> 2) ^ v) & 0x0f;
- /* cast to uint64_t to avoid 32 bit shift of 32 bit value */
- ret = (ret << r) | (unsigned long)((uint64_t)ret >> (32 - r));
- ret &= 0xFFFFFFFFL;
- ret ^= v * v;
- c++;
- }
- return (ret >> 16) ^ ret;
- }
- unsigned long OPENSSL_LH_num_items(const OPENSSL_LHASH *lh)
- {
- return lh ? lh->num_items : 0;
- }
- unsigned long OPENSSL_LH_get_down_load(const OPENSSL_LHASH *lh)
- {
- return lh->down_load;
- }
- void OPENSSL_LH_set_down_load(OPENSSL_LHASH *lh, unsigned long down_load)
- {
- lh->down_load = down_load;
- }
- int OPENSSL_LH_error(OPENSSL_LHASH *lh)
- {
- return lh->error;
- }
|