buf_size: allow strncmp("foo", bar, 100) where 100 is larger than "foo"
[smatch.git] / smatch_buf_size.c
blob16559b5dae84564c23978d31377176a90b80e14a
1 /*
2 * Copyright (C) 2010 Dan Carpenter.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License
6 * as published by the Free Software Foundation; either version 2
7 * of the License, or (at your option) any later version.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, see http://www.gnu.org/copyleft/gpl.txt
18 #include <stdlib.h>
19 #include <errno.h>
20 #include "parse.h"
21 #include "smatch.h"
22 #include "smatch_slist.h"
23 #include "smatch_extra.h"
24 #include "smatch_function_hashtable.h"
26 #define UNKNOWN_SIZE (-1)
28 static int my_size_id;
30 struct limiter {
31 int buf_arg;
32 int limit_arg;
34 static struct limiter b0_l2 = {0, 2};
36 static DEFINE_HASHTABLE_INSERT(insert_func, char, int);
37 static DEFINE_HASHTABLE_SEARCH(search_func, char, int);
38 static struct hashtable *allocation_funcs;
40 static char *get_fn_name(struct expression *expr)
42 if (expr->type != EXPR_CALL)
43 return NULL;
44 if (expr->fn->type != EXPR_SYMBOL)
45 return NULL;
46 return expr_to_var(expr->fn);
49 static int is_allocation_function(struct expression *expr)
51 char *func;
52 int ret = 0;
54 func = get_fn_name(expr);
55 if (!func)
56 return 0;
57 if (search_func(allocation_funcs, func))
58 ret = 1;
59 free_string(func);
60 return ret;
63 static void add_allocation_function(const char *func, void *call_back, int param)
65 insert_func(allocation_funcs, (char *)func, (int *)1);
66 add_function_assign_hook(func, call_back, INT_PTR(param));
69 static int estate_to_size(struct smatch_state *state)
71 sval_t sval;
73 if (!state || !estate_rl(state))
74 return 0;
75 sval = estate_max(state);
76 return sval.value;
79 static struct smatch_state *size_to_estate(int size)
81 sval_t sval;
83 sval.type = &int_ctype;
84 sval.value = size;
86 return alloc_estate_sval(sval);
89 static struct range_list *size_to_rl(int size)
91 sval_t sval;
93 sval.type = &int_ctype;
94 sval.value = size;
96 return alloc_rl(sval, sval);
99 static struct smatch_state *unmatched_size_state(struct sm_state *sm)
101 return size_to_estate(UNKNOWN_SIZE);
104 static void set_size_undefined(struct sm_state *sm, struct expression *mod_expr)
106 set_state(sm->owner, sm->name, sm->sym, size_to_estate(UNKNOWN_SIZE));
109 static struct smatch_state *merge_size_func(struct smatch_state *s1, struct smatch_state *s2)
111 return merge_estates(s1, s2);
114 void set_param_buf_size(const char *name, struct symbol *sym, char *key, char *value)
116 struct range_list *rl = NULL;
117 struct smatch_state *state;
118 char fullname[256];
120 if (strncmp(key, "$$", 2) != 0)
121 return;
123 snprintf(fullname, 256, "%s%s", name, key + 2);
125 str_to_rl(&int_ctype, value, &rl);
126 if (!rl || is_whole_rl(rl))
127 return;
128 state = alloc_estate_rl(rl);
129 set_state(my_size_id, fullname, sym, state);
132 static int bytes_per_element(struct expression *expr)
134 struct symbol *type;
135 int bpe;
137 if (expr->type == EXPR_STRING)
138 return 1;
139 type = get_type(expr);
140 if (!type)
141 return 0;
143 if (type->type != SYM_PTR && type->type != SYM_ARRAY)
144 return 0;
146 type = get_base_type(type);
147 bpe = bits_to_bytes(type->bit_size);
149 if (bpe == -1) /* void pointer */
150 bpe = 1;
152 return bpe;
155 static int bytes_to_elements(struct expression *expr, int bytes)
157 int bpe;
159 bpe = bytes_per_element(expr);
160 if (bpe == 0)
161 return 0;
162 return bytes / bpe;
165 static int elements_to_bytes(struct expression *expr, int elements)
167 int bpe;
169 bpe = bytes_per_element(expr);
170 return elements * bpe;
173 static int get_initializer_size(struct expression *expr)
175 switch (expr->type) {
176 case EXPR_STRING:
177 return expr->string->length;
178 case EXPR_INITIALIZER: {
179 struct expression *tmp;
180 int i = 0;
182 FOR_EACH_PTR(expr->expr_list, tmp) {
183 if (tmp->type == EXPR_INDEX) {
184 if (tmp->idx_to >= i)
185 i = tmp->idx_to;
186 else
187 continue;
190 i++;
191 } END_FOR_EACH_PTR(tmp);
192 return i;
194 case EXPR_SYMBOL:
195 return get_array_size(expr);
197 return 0;
200 static struct range_list *db_size_rl;
201 static int db_size_callback(void *unused, int argc, char **argv, char **azColName)
203 struct range_list *tmp = NULL;
205 if (!db_size_rl) {
206 str_to_rl(&int_ctype, argv[0], &db_size_rl);
207 } else {
208 str_to_rl(&int_ctype, argv[0], &tmp);
209 db_size_rl = rl_union(db_size_rl, tmp);
211 return 0;
214 static struct range_list *size_from_db(struct expression *expr)
216 int this_file_only = 0;
217 char *name;
219 name = get_member_name(expr);
220 if (!name && is_static(expr)) {
221 name = expr_to_var(expr);
222 this_file_only = 1;
224 if (!name)
225 return 0;
227 if (this_file_only) {
228 db_size_rl = NULL;
229 run_sql(db_size_callback, "select size from function_type_size where type = '%s' and file = '%s';",
230 name, get_filename());
231 if (db_size_rl)
232 return db_size_rl;
233 return 0;
236 db_size_rl = NULL;
237 run_sql(db_size_callback, "select size from type_size where type = '%s';",
238 name);
239 return db_size_rl;
242 static void db_returns_buf_size(struct expression *expr, int param, char *unused, char *math)
244 struct expression *call;
245 sval_t sval;
247 if (expr->type != EXPR_ASSIGNMENT)
248 return;
249 call = strip_expr(expr->right);
251 if (!parse_call_math(call, math, &sval))
252 return;
253 set_state_expr(my_size_id, expr->left, size_to_estate(sval.value));
256 int get_real_array_size(struct expression *expr)
258 struct symbol *type;
259 sval_t sval;
261 if (expr->type == EXPR_BINOP) /* array elements foo[5] */
262 return 0;
264 type = get_type(expr);
265 if (!type)
266 return 0;
267 if (!type || type->type != SYM_ARRAY)
268 return 0;
270 if (!get_implied_value(type->array_size, &sval))
271 return 0;
273 /* People put one element arrays on the end of structs */
274 if (sval.value == 1)
275 return 0;
277 return sval.value;
280 static int get_size_from_initializer(struct expression *expr)
282 if (expr->type != EXPR_SYMBOL || !expr->symbol || !expr->symbol->initializer)
283 return 0;
284 if (expr->symbol->initializer == expr) /* int a = a; */
285 return 0;
286 return get_initializer_size(expr->symbol->initializer);
289 static struct range_list *get_stored_size_bytes(struct expression *expr)
291 struct smatch_state *state;
293 state = get_state_expr(my_size_id, expr);
294 if (!state)
295 return NULL;
296 return estate_rl(state);
299 static int get_bytes_from_address(struct expression *expr)
301 struct symbol *type;
302 int ret;
304 if (!option_spammy)
305 return 0;
306 if (expr->type != EXPR_PREOP || expr->op != '&')
307 return 0;
308 type = get_type(expr);
309 if (!type)
310 return 0;
312 if (type->type == SYM_PTR)
313 type = get_base_type(type);
315 ret = bits_to_bytes(type->bit_size);
316 if (ret == -1)
317 return 0;
318 if (ret == 1)
319 return 0; /* ignore char pointers */
321 return ret;
324 static struct expression *remove_addr_fluff(struct expression *expr)
326 struct expression *tmp;
327 sval_t sval;
329 expr = strip_expr(expr);
331 /* remove '&' and '*' operations that cancel */
332 while (expr && expr->type == EXPR_PREOP && expr->op == '&') {
333 tmp = strip_expr(expr->unop);
334 if (tmp->type != EXPR_PREOP)
335 break;
336 if (tmp->op != '*')
337 break;
338 expr = strip_expr(tmp->unop);
341 if (!expr)
342 return NULL;
344 /* "foo + 0" is just "foo" */
345 if (expr->type == EXPR_BINOP && expr->op == '+' &&
346 get_value(expr->right, &sval) && sval.value == 0)
347 return expr->left;
349 return expr;
352 static int is_last_member_of_struct(struct symbol *sym, struct ident *member)
354 struct symbol *tmp;
355 int i;
357 i = 0;
358 FOR_EACH_PTR_REVERSE(sym->symbol_list, tmp) {
359 if (i++ || !tmp->ident)
360 return 0;
361 if (tmp->ident == member)
362 return 1;
363 return 0;
364 } END_FOR_EACH_PTR_REVERSE(tmp);
366 return 0;
369 static int get_stored_size_end_struct_bytes(struct expression *expr)
371 struct symbol *type;
372 char *name;
373 struct symbol *sym;
374 struct smatch_state *state;
375 sval_t sval;
377 if (expr->type == EXPR_BINOP) /* array elements foo[5] */
378 return 0;
380 type = get_type(expr);
381 if (!type || type->type != SYM_ARRAY)
382 return 0;
384 if (!get_implied_value(type->array_size, &sval))
385 return 0;
387 if (sval.value != 0 && sval.value != 1)
388 return 0;
390 name = expr_to_var_sym(expr, &sym);
391 free_string(name);
392 if (!sym || !sym->ident || !sym->ident->name)
393 return 0;
394 if (!sym->bit_size)
395 return 0;
397 if (sym->type != SYM_NODE)
398 return 0;
400 state = get_state(my_size_id, sym->ident->name, sym);
401 if (!estate_to_size(state))
402 return 0;
404 sym = get_real_base_type(sym);
405 if (!sym || sym->type != SYM_PTR)
406 return 0;
407 sym = get_real_base_type(sym);
408 if (!sym || sym->type != SYM_STRUCT)
409 return 0;
410 if (!is_last_member_of_struct(sym, expr->member))
411 return 0;
413 return estate_to_size(state) - bits_to_bytes(sym->bit_size) +
414 bits_to_bytes(type->bit_size);
417 static struct range_list *alloc_int_rl(int value)
419 sval_t sval = {
420 .type = &int_ctype,
421 {.value = value},
424 return alloc_rl(sval, sval);
427 struct range_list *get_array_size_bytes_rl(struct expression *expr)
429 int declared_size = 0;
430 struct range_list *ret = NULL;
431 int size;
433 expr = remove_addr_fluff(expr);
434 if (!expr)
435 return NULL;
437 /* "BAR" */
438 if (expr->type == EXPR_STRING)
439 return alloc_int_rl(expr->string->length);
441 /* buf[4] */
442 size = get_real_array_size(expr);
443 if (size)
444 declared_size = elements_to_bytes(expr, size);
446 /* buf = malloc(1024); */
447 ret = get_stored_size_bytes(expr);
448 if (ret) {
449 if (declared_size)
450 return rl_union(ret, alloc_int_rl(size));
451 return ret;
453 if (declared_size)
454 return alloc_int_rl(declared_size);
456 size = get_stored_size_end_struct_bytes(expr);
457 if (size)
458 return alloc_int_rl(size);
460 /* char *foo = "BAR" */
461 size = get_size_from_initializer(expr);
462 if (size)
463 return alloc_int_rl(elements_to_bytes(expr, size));
465 size = get_bytes_from_address(expr);
466 if (size)
467 return alloc_int_rl(size);
469 /* if (strlen(foo) > 4) */
470 size = get_size_from_strlen(expr);
471 if (size)
472 return alloc_int_rl(size);
474 ret = size_from_db(expr);
475 if (ret)
476 return ret;
478 return NULL;
481 int get_array_size_bytes(struct expression *expr)
483 struct range_list *rl;
484 sval_t sval;
486 rl = get_array_size_bytes_rl(expr);
487 if (!rl_to_sval(rl, &sval))
488 return 0;
489 if (sval.uvalue >= INT_MAX)
490 return 0;
491 return sval.value;
494 int get_array_size_bytes_max(struct expression *expr)
496 struct range_list *rl;
497 sval_t bytes;
499 rl = get_array_size_bytes_rl(expr);
500 if (!rl)
501 return 0;
502 bytes = rl_min(rl);
503 if (bytes.value < 0)
504 return 0;
505 bytes = rl_max(rl);
506 if (bytes.uvalue >= INT_MAX)
507 return 0;
508 return bytes.value;
511 int get_array_size_bytes_min(struct expression *expr)
513 struct range_list *rl;
514 struct data_range *range;
516 rl = get_array_size_bytes_rl(expr);
517 if (!rl)
518 return 0;
520 FOR_EACH_PTR(rl, range) {
521 if (range->min.value <= 0)
522 return 0;
523 if (range->max.value <= 0)
524 return 0;
525 if (range->min.uvalue >= INT_MAX)
526 return 0;
527 return range->min.value;
528 } END_FOR_EACH_PTR(range);
530 return 0;
533 int get_array_size(struct expression *expr)
535 return bytes_to_elements(expr, get_array_size_bytes_max(expr));
538 static void match_strlen_condition(struct expression *expr)
540 struct expression *left;
541 struct expression *right;
542 struct expression *str = NULL;
543 int strlen_left = 0;
544 int strlen_right = 0;
545 sval_t sval;
546 struct smatch_state *true_state = NULL;
547 struct smatch_state *false_state = NULL;
549 if (expr->type != EXPR_COMPARE)
550 return;
551 left = strip_expr(expr->left);
552 right = strip_expr(expr->right);
554 if (left->type == EXPR_CALL && sym_name_is("strlen", left->fn)) {
555 str = get_argument_from_call_expr(left->args, 0);
556 strlen_left = 1;
558 if (right->type == EXPR_CALL && sym_name_is("strlen", right->fn)) {
559 str = get_argument_from_call_expr(right->args, 0);
560 strlen_right = 1;
563 if (!strlen_left && !strlen_right)
564 return;
565 if (strlen_left && strlen_right)
566 return;
568 if (strlen_left) {
569 if (!get_value(right, &sval))
570 return;
572 if (strlen_right) {
573 if (!get_value(left, &sval))
574 return;
577 /* FIXME: why are we using my_size_id here instead of my_strlen_id */
579 if (expr->op == SPECIAL_EQUAL) {
580 set_true_false_states_expr(my_size_id, str, size_to_estate(sval.value + 1), NULL);
581 return;
583 if (expr->op == SPECIAL_NOTEQUAL) {
584 set_true_false_states_expr(my_size_id, str, NULL, size_to_estate(sval.value + 1));
585 return;
588 switch (expr->op) {
589 case '<':
590 case SPECIAL_UNSIGNED_LT:
591 if (strlen_left)
592 true_state = size_to_estate(sval.value);
593 else
594 false_state = size_to_estate(sval.value + 1);
595 break;
596 case SPECIAL_LTE:
597 case SPECIAL_UNSIGNED_LTE:
598 if (strlen_left)
599 true_state = size_to_estate(sval.value + 1);
600 else
601 false_state = size_to_estate(sval.value);
602 break;
603 case SPECIAL_GTE:
604 case SPECIAL_UNSIGNED_GTE:
605 if (strlen_left)
606 false_state = size_to_estate(sval.value);
607 else
608 true_state = size_to_estate(sval.value + 1);
609 break;
610 case '>':
611 case SPECIAL_UNSIGNED_GT:
612 if (strlen_left)
613 false_state = size_to_estate(sval.value + 1);
614 else
615 true_state = size_to_estate(sval.value);
616 break;
618 set_true_false_states_expr(my_size_id, str, true_state, false_state);
621 static struct expression *strip_ampersands(struct expression *expr)
623 struct symbol *type;
625 if (expr->type != EXPR_PREOP)
626 return expr;
627 if (expr->op != '&')
628 return expr;
629 type = get_type(expr->unop);
630 if (!type || type->type != SYM_ARRAY)
631 return expr;
632 return expr->unop;
635 static void info_record_alloction(struct expression *buffer, struct range_list *rl)
637 char *name;
639 if (!option_info)
640 return;
642 name = get_member_name(buffer);
643 if (!name && is_static(buffer))
644 name = expr_to_var(buffer);
645 if (!name)
646 return;
647 if (rl && !is_whole_rl(rl))
648 sql_insert_function_type_size(name, show_rl(rl));
649 else
650 sql_insert_function_type_size(name, "(-1)");
652 free_string(name);
655 static void store_alloc(struct expression *expr, struct range_list *rl)
657 rl = clone_rl(rl); // FIXME!!!
658 info_record_alloction(expr, rl);
659 set_state_expr(my_size_id, expr, alloc_estate_rl(rl));
662 static void match_array_assignment(struct expression *expr)
664 struct expression *left;
665 struct expression *right;
666 struct range_list *rl;
667 sval_t sval;
669 if (expr->op != '=')
670 return;
671 left = strip_expr(expr->left);
672 right = strip_expr(expr->right);
673 right = strip_ampersands(right);
675 if (is_allocation_function(right))
676 return;
678 if (get_implied_value(right, &sval) && sval.value == 0) {
679 rl = alloc_int_rl(0);
680 goto store;
683 rl = get_array_size_bytes_rl(right);
685 store:
686 store_alloc(left, rl);
689 static void match_alloc(const char *fn, struct expression *expr, void *_size_arg)
691 int size_arg = PTR_INT(_size_arg);
692 struct expression *right;
693 struct expression *arg;
694 struct range_list *rl;
696 right = strip_expr(expr->right);
697 arg = get_argument_from_call_expr(right->args, size_arg);
698 get_absolute_rl(arg, &rl);
699 rl = cast_rl(&int_ctype, rl);
700 store_alloc(expr->left, rl);
703 static void match_calloc(const char *fn, struct expression *expr, void *unused)
705 struct expression *right;
706 struct expression *arg;
707 sval_t elements;
708 sval_t size;
710 right = strip_expr(expr->right);
711 arg = get_argument_from_call_expr(right->args, 0);
712 if (!get_implied_value(arg, &elements))
713 return; // FIXME!!!
714 arg = get_argument_from_call_expr(right->args, 1);
715 if (get_implied_value(arg, &size))
716 store_alloc(expr->left, size_to_rl(elements.value * size.value));
717 else
718 store_alloc(expr->left, size_to_rl(-1));
721 static void match_limited(const char *fn, struct expression *expr, void *_limiter)
723 struct limiter *limiter = (struct limiter *)_limiter;
724 struct expression *dest;
725 struct expression *size_expr;
726 sval_t size;
728 dest = get_argument_from_call_expr(expr->args, limiter->buf_arg);
729 size_expr = get_argument_from_call_expr(expr->args, limiter->limit_arg);
730 if (!get_implied_max(size_expr, &size))
731 return;
732 set_state_expr(my_size_id, dest, size_to_estate(size.value));
735 static void match_strcpy(const char *fn, struct expression *expr, void *unused)
737 struct expression fake_assign;
739 fake_assign.op = '=';
740 fake_assign.left = get_argument_from_call_expr(expr->args, 0);
741 fake_assign.right = get_argument_from_call_expr(expr->args, 1);
742 match_array_assignment(&fake_assign);
745 static void match_strndup(const char *fn, struct expression *expr, void *unused)
747 struct expression *fn_expr;
748 struct expression *size_expr;
749 sval_t size;
751 fn_expr = strip_expr(expr->right);
752 size_expr = get_argument_from_call_expr(fn_expr->args, 1);
753 if (get_implied_max(size_expr, &size)) {
754 size.value++;
755 store_alloc(expr->left, size_to_rl(size.value));
756 } else {
757 store_alloc(expr->left, size_to_rl(-1));
762 static void match_call(struct expression *expr)
764 struct expression *arg;
765 struct range_list *rl;
766 int i;
768 i = 0;
769 FOR_EACH_PTR(expr->args, arg) {
770 rl = get_array_size_bytes_rl(arg);
771 if (rl && !is_whole_rl(rl))
772 sql_insert_caller_info(expr, BUF_SIZE, i, "$$", show_rl(rl));
773 i++;
774 } END_FOR_EACH_PTR(arg);
777 static void struct_member_callback(struct expression *call, int param, char *printed_name, struct smatch_state *state)
779 if (state == &merged)
780 return;
781 sql_insert_caller_info(call, BUF_SIZE, param, printed_name, state->name);
784 void register_buf_size(int id)
786 my_size_id = id;
788 add_unmatched_state_hook(my_size_id, &unmatched_size_state);
790 select_caller_info_hook(set_param_buf_size, BUF_SIZE);
791 select_return_states_hook(BUF_SIZE, &db_returns_buf_size);
793 allocation_funcs = create_function_hashtable(100);
794 add_allocation_function("malloc", &match_alloc, 0);
795 add_allocation_function("calloc", &match_calloc, 0);
796 add_allocation_function("memdup", &match_alloc, 1);
797 if (option_project == PROJ_KERNEL) {
798 add_allocation_function("kmalloc", &match_alloc, 0);
799 add_allocation_function("kzalloc", &match_alloc, 0);
800 add_allocation_function("vmalloc", &match_alloc, 0);
801 add_allocation_function("__vmalloc", &match_alloc, 0);
802 add_allocation_function("kcalloc", &match_calloc, 0);
803 add_allocation_function("kmalloc_array", &match_calloc, 0);
804 add_allocation_function("drm_malloc_ab", &match_calloc, 0);
805 add_allocation_function("drm_calloc_large", &match_calloc, 0);
806 add_allocation_function("sock_kmalloc", &match_alloc, 1);
807 add_allocation_function("kmemdup", &match_alloc, 1);
808 add_allocation_function("kmemdup_user", &match_alloc, 1);
809 add_allocation_function("dma_alloc_attrs", &match_alloc, 1);
810 add_allocation_function("pci_alloc_consistent", &match_alloc, 1);
811 add_allocation_function("pci_alloc_coherent", &match_alloc, 1);
812 add_allocation_function("devm_kmalloc", &match_alloc, 1);
813 add_allocation_function("devm_kzalloc", &match_alloc, 1);
815 add_hook(&match_strlen_condition, CONDITION_HOOK);
817 add_allocation_function("strndup", match_strndup, 0);
818 if (option_project == PROJ_KERNEL)
819 add_allocation_function("kstrndup", match_strndup, 0);
821 add_modification_hook(my_size_id, &set_size_undefined);
823 add_merge_hook(my_size_id, &merge_size_func);
826 void register_buf_size_late(int id)
828 /* has to happen after match_alloc() */
829 add_hook(&match_array_assignment, ASSIGNMENT_HOOK);
831 add_function_hook("strlcpy", &match_limited, &b0_l2);
832 add_function_hook("strlcat", &match_limited, &b0_l2);
833 add_function_hook("memscan", &match_limited, &b0_l2);
835 add_function_hook("strcpy", &match_strcpy, NULL);
837 add_hook(&match_call, FUNCTION_CALL_HOOK);
838 add_member_info_callback(my_size_id, struct_member_callback);