?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
user_data: cleanup get_user_rl()
[smatch.git] / check_user_data2.c
blob08f473e85093ff9f033d3ed081490c01bf54276d
1 /*
2 * Copyright (C) 2011 Dan Carpenter.
3 *
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.
8 *
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.
13 *
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
16 */
17
18 /*
19 * There are a couple checks that try to see if a variable
20 * comes from the user. It would be better to unify them
21 * into one place. Also it we should follow the data down
22 * the call paths. Hence this file.
23 */
24
25 #include "smatch.h"
26 #include "smatch_slist.h"
27 #include "smatch_extra.h"
28
29 static int my_id;
30 static int my_call_id;
31
32 STATE(called);
33 static bool func_gets_user_data;
34
35 static const char * kstr_funcs[] = {
36 "kstrtoull", "kstrtoll", "kstrtoul", "kstrtol", "kstrtouint",
37 "kstrtoint", "kstrtou64", "kstrtos64", "kstrtou32", "kstrtos32",
38 "kstrtou16", "kstrtos16", "kstrtou8", "kstrtos8", "kstrtoull_from_user"
39 "kstrtoll_from_user", "kstrtoul_from_user", "kstrtol_from_user",
40 "kstrtouint_from_user", "kstrtoint_from_user", "kstrtou16_from_user",
41 "kstrtos16_from_user", "kstrtou8_from_user", "kstrtos8_from_user",
42 "kstrtou64_from_user", "kstrtos64_from_user", "kstrtou32_from_user",
43 "kstrtos32_from_user",
44 };
45
46 static void set_points_to_user_data(struct expression *expr);
47
48 static struct stree *start_states;
49 static struct stree_stack *saved_stack;
50 static void save_start_states(struct statement *stmt)
51 {
52 start_states = clone_stree(__get_cur_stree());
53 }
54
55 static void free_start_states(void)
56 {
57 free_stree(&start_states);
58 }
59
60 static void match_save_states(struct expression *expr)
61 {
62 push_stree(&saved_stack, start_states);
63 start_states = NULL;
64 }
65
66 static void match_restore_states(struct expression *expr)
67 {
68 free_stree(&start_states);
69 start_states = pop_stree(&saved_stack);
70 }
71
72 static struct smatch_state *empty_state(struct sm_state *sm)
73 {
74 return alloc_estate_empty();
75 }
76
77 static void pre_merge_hook(struct sm_state *sm)
78 {
79 struct smatch_state *user;
80 struct smatch_state *extra;
81 struct range_list *rl;
82 sval_t dummy;
83 sval_t sval_100 = {
84 .type = &int_ctype,
85 .value = 100,
86 };
87
88 user = get_state(my_id, sm->name, sm->sym);
89 if (!user)
90 return;
91 if (!estate_rl(sm->state)) {
92 /*
93 * If the one side is capped and the other side is empty then
94 * let's just mark it as not-user data because the information
95 * isn't going to be useful. How this looks is:
96 *
97 * if (user_var > trusted)
98 * user_var = trusted; <-- empty state
99 * else
100 * <-- capped
101 *
102 * The problem is that sometimes things are capped to a literal
103 * and we'd like to keep the state in that case... Ugh. I've
104 * added a check which assumes that everything less than 100 is
105 * probably capped against a literal.
106 *
107 */
108 if (is_capped_var_sym(sm->name, sm->sym) &&
109 sval_cmp(estate_max(user), sval_100) > 0)
110 set_state(my_id, sm->name, sm->sym, alloc_estate_empty());
111 return;
112 }
113 extra = get_state(SMATCH_EXTRA, sm->name, sm->sym);
114 if (!extra || !estate_rl(extra))
115 return;
116 rl = rl_intersection(estate_rl(user), estate_rl(extra));
117 if (rl_to_sval(rl, &dummy))
118 rl = NULL;
119 set_state(my_id, sm->name, sm->sym, alloc_estate_rl(clone_rl(rl)));
120 }
121
122 static void extra_nomod_hook(const char *name, struct symbol *sym, struct expression *expr, struct smatch_state *state)
123 {
124 struct smatch_state *user;
125 struct range_list *rl;
126
127 user = get_state(my_id, name, sym);
128 if (!user)
129 return;
130 rl = rl_intersection(estate_rl(user), estate_rl(state));
131 if (rl_equiv(rl, estate_rl(user)))
132 return;
133 set_state(my_id, name, sym, alloc_estate_rl(rl));
134 }
135
136 static void tag_inner_struct_members(struct expression *expr, struct symbol *member)
137 {
138 struct expression *edge_member;
139 struct symbol *base = get_real_base_type(member);
140 struct symbol *tmp;
141
142 if (member->ident)
143 expr = member_expression(expr, '.', member->ident);
144
145 FOR_EACH_PTR(base->symbol_list, tmp) {
146 struct symbol *type;
147
148 type = get_real_base_type(tmp);
149 if (!type)
150 continue;
151
152 if (type->type == SYM_UNION || type->type == SYM_STRUCT) {
153 tag_inner_struct_members(expr, tmp);
154 continue;
155 }
156
157 if (!tmp->ident)
158 continue;
159
160 edge_member = member_expression(expr, '.', tmp->ident);
161 set_state_expr(my_id, edge_member, alloc_estate_whole(type));
162 } END_FOR_EACH_PTR(tmp);
163 }
164
165 static void tag_struct_members(struct symbol *type, struct expression *expr)
166 {
167 struct symbol *tmp;
168 struct expression *member;
169 int op = '*';
170
171 if (expr->type == EXPR_PREOP && expr->op == '&') {
172 expr = strip_expr(expr->unop);
173 op = '.';
174 }
175
176 FOR_EACH_PTR(type->symbol_list, tmp) {
177 type = get_real_base_type(tmp);
178 if (!type)
179 continue;
180
181 if (type->type == SYM_UNION || type->type == SYM_STRUCT) {
182 tag_inner_struct_members(expr, tmp);
183 continue;
184 }
185
186 if (!tmp->ident)
187 continue;
188
189 member = member_expression(expr, op, tmp->ident);
190 set_state_expr(my_id, member, alloc_estate_whole(get_type(member)));
191
192 if (type->type == SYM_ARRAY)
193 set_points_to_user_data(member);
194 } END_FOR_EACH_PTR(tmp);
195 }
196
197 static void tag_base_type(struct expression *expr)
198 {
199 if (expr->type == EXPR_PREOP && expr->op == '&')
200 expr = strip_expr(expr->unop);
201 else
202 expr = deref_expression(expr);
203 set_state_expr(my_id, expr, alloc_estate_whole(get_type(expr)));
204 }
205
206 static void tag_as_user_data(struct expression *expr)
207 {
208 struct symbol *type;
209
210 expr = strip_expr(expr);
211
212 type = get_type(expr);
213 if (!type || type->type != SYM_PTR)
214 return;
215 type = get_real_base_type(type);
216 if (!type)
217 return;
218 if (type == &void_ctype) {
219 set_state_expr(my_id, deref_expression(expr), alloc_estate_whole(&ulong_ctype));
220 return;
221 }
222 if (type->type == SYM_BASETYPE)
223 tag_base_type(expr);
224 if (type->type == SYM_STRUCT || type->type == SYM_UNION) {
225 if (expr->type != EXPR_PREOP || expr->op != '&')
226 expr = deref_expression(expr);
227 else
228 set_state_expr(my_id, deref_expression(expr), alloc_estate_whole(&ulong_ctype));
229 tag_struct_members(type, expr);
230 }
231 }
232
233 static void match_user_copy(const char *fn, struct expression *expr, void *_param)
234 {
235 int param = PTR_INT(_param);
236 struct expression *dest;
237
238 func_gets_user_data = true;
239
240 dest = get_argument_from_call_expr(expr->args, param);
241 dest = strip_expr(dest);
242 if (!dest)
243 return;
244 tag_as_user_data(dest);
245 }
246
247 static void match_sscanf(const char *fn, struct expression *expr, void *unused)
248 {
249 struct expression *arg;
250 int i;
251
252 func_gets_user_data = true;
253
254 i = -1;
255 FOR_EACH_PTR(expr->args, arg) {
256 i++;
257 if (i < 2)
258 continue;
259 tag_as_user_data(arg);
260 } END_FOR_EACH_PTR(arg);
261 }
262
263 static int is_skb_data(struct expression *expr)
264 {
265 struct symbol *sym;
266
267 if (!expr)
268 return 0;
269
270 if (expr->type == EXPR_BINOP && expr->op == '+')
271 return is_skb_data(expr->left);
272
273 expr = strip_expr(expr);
274 if (!expr)
275 return 0;
276 if (expr->type != EXPR_DEREF || expr->op != '.')
277 return 0;
278
279 if (!expr->member)
280 return 0;
281 if (strcmp(expr->member->name, "data") != 0)
282 return 0;
283
284 sym = expr_to_sym(expr->deref);
285 if (!sym)
286 return 0;
287 sym = get_real_base_type(sym);
288 if (!sym || sym->type != SYM_PTR)
289 return 0;
290 sym = get_real_base_type(sym);
291 if (!sym || sym->type != SYM_STRUCT || !sym->ident)
292 return 0;
293 if (strcmp(sym->ident->name, "sk_buff") != 0)
294 return 0;
295
296 return 1;
297 }
298
299 static int points_to_user_data(struct expression *expr)
300 {
301 struct smatch_state *state;
302 char buf[256];
303 struct symbol *sym;
304 char *name;
305 int ret = 0;
306
307 expr = strip_expr(expr);
308 if (!expr)
309 return 0;
310 if (is_skb_data(expr))
311 return 1;
312
313 if (expr->type == EXPR_BINOP && expr->op == '+') {
314 if (points_to_user_data(expr->left))
315 return 1;
316 if (points_to_user_data(expr->right))
317 return 1;
318 return 0;
319 }
320
321 name = expr_to_var_sym(expr, &sym);
322 if (!name || !sym)
323 goto free;
324 snprintf(buf, sizeof(buf), "*%s", name);
325 state = get_state(my_id, buf, sym);
326 if (state && estate_rl(state))
327 ret = 1;
328 free:
329 free_string(name);
330 return ret;
331 }
332
333 static void set_points_to_user_data(struct expression *expr)
334 {
335 char *name;
336 struct symbol *sym;
337 char buf[256];
338
339 name = expr_to_var_sym(expr, &sym);
340 if (!name || !sym)
341 goto free;
342 snprintf(buf, sizeof(buf), "*%s", name);
343 set_state(my_id, buf, sym, alloc_estate_whole(&llong_ctype));
344 free:
345 free_string(name);
346 }
347
348 static int comes_from_skb_data(struct expression *expr)
349 {
350 expr = strip_expr(expr);
351 if (!expr || expr->type != EXPR_PREOP || expr->op != '*')
352 return 0;
353
354 expr = strip_expr(expr->unop);
355 if (!expr)
356 return 0;
357 if (expr->type == EXPR_BINOP && expr->op == '+')
358 expr = strip_expr(expr->left);
359
360 return is_skb_data(expr);
361 }
362
363 static int handle_struct_assignment(struct expression *expr)
364 {
365 struct expression *right;
366 struct symbol *left_type, *right_type;
367
368 left_type = get_type(expr->left);
369 if (!left_type || left_type->type != SYM_PTR)
370 return 0;
371 left_type = get_real_base_type(left_type);
372 if (!left_type)
373 return 0;
374 if (left_type->type != SYM_STRUCT &&
375 left_type->type != SYM_UNION)
376 return 0;
377
378 /*
379 * Ignore struct to struct assignments because for those we look at the
380 * individual members.
381 */
382 right = strip_expr(expr->right);
383 right_type = get_type(right);
384 if (!right_type || right_type->type != SYM_PTR)
385 return 0;
386
387 /* If we are assigning struct members then normally that is handled
388 * by fake assignments, however if we cast one struct to a different
389 * of struct then we handle that here.
390 */
391 right_type = get_real_base_type(right_type);
392 if (right_type == left_type)
393 return 0;
394
395 if (!points_to_user_data(right))
396 return 0;
397
398 tag_as_user_data(expr->left);
399 return 1;
400 }
401
402 static int handle_get_user(struct expression *expr)
403 {
404 char *name;
405 int ret = 0;
406
407 name = get_macro_name(expr->pos);
408 if (!name || strcmp(name, "get_user") != 0)
409 return 0;
410
411 name = expr_to_var(expr->right);
412 if (!name || strcmp(name, "__val_gu") != 0)
413 goto free;
414 set_state_expr(my_id, expr->left, alloc_estate_whole(get_type(expr->left)));
415 ret = 1;
416 free:
417 free_string(name);
418 return ret;
419 }
420
421 static void match_assign(struct expression *expr)
422 {
423 struct range_list *rl;
424
425 if (is_fake_call(expr->right))
426 return;
427 if (handle_get_user(expr))
428 return;
429 if (points_to_user_data(expr->right))
430 set_points_to_user_data(expr->left);
431 if (handle_struct_assignment(expr))
432 return;
433
434 if (!get_user_rl(expr->right, &rl))
435 goto clear_old_state;
436
437 rl = cast_rl(get_type(expr->left), rl);
438 set_state_expr(my_id, expr->left, alloc_estate_rl(rl));
439
440 return;
441
442 clear_old_state:
443 if (get_state_expr(my_id, expr->left))
444 set_state_expr(my_id, expr->left, alloc_estate_empty());
445 }
446
447 static void handle_eq_noteq(struct expression *expr)
448 {
449 struct smatch_state *left_orig, *right_orig;
450
451 left_orig = get_state_expr(my_id, expr->left);
452 right_orig = get_state_expr(my_id, expr->right);
453
454 if (!left_orig && !right_orig)
455 return;
456 if (left_orig && right_orig)
457 return;
458
459 if (left_orig) {
460 set_true_false_states_expr(my_id, expr->left,
461 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
462 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
463 } else {
464 set_true_false_states_expr(my_id, expr->right,
465 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
466 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
467 }
468 }
469
470 static void handle_unsigned_lt_gt(struct expression *expr)
471 {
472 struct symbol *type;
473 struct range_list *left;
474 struct range_list *right;
475 struct range_list *non_negative;
476 sval_t min, minus_one;
477
478 /*
479 * conditions are mostly handled by smatch_extra.c. The special case
480 * here is that say you have if (user_int < unknown_u32) {
481 * In Smatch extra we say that, We have no idea what value
482 * unknown_u32 is so the only thin we can say for sure is that
483 * user_int is not -1 (UINT_MAX). But in check_user_data2.c we should
484 * assume that unless unknown_u32 is user data, it's probably less than
485 * INT_MAX.
486 *
487 */
488
489 type = get_type(expr);
490 if (!type_unsigned(type))
491 return;
492
493 /*
494 * Assume if (user < trusted) { ... because I am lazy and because this
495 * is the correct way to write code.
496 */
497 if (!get_user_rl(expr->left, &left))
498 return;
499 if (get_user_rl(expr->right, &right))
500 return;
501
502 if (!sval_is_negative(rl_min(left)))
503 return;
504 min = rl_min(left);
505 minus_one.type = rl_type(left);
506 minus_one.value = -1;
507 non_negative = remove_range(left, min, minus_one);
508
509 switch (expr->op) {
510 case '<':
511 case SPECIAL_UNSIGNED_LT:
512 case SPECIAL_LTE:
513 case SPECIAL_UNSIGNED_LTE:
514 set_true_false_states_expr(my_id, expr->left,
515 alloc_estate_rl(non_negative), NULL);
516 break;
517 case '>':
518 case SPECIAL_UNSIGNED_GT:
519 case SPECIAL_GTE:
520 case SPECIAL_UNSIGNED_GTE:
521 set_true_false_states_expr(my_id, expr->left,
522 NULL, alloc_estate_rl(non_negative));
523 break;
524 }
525 }
526
527 static void match_condition(struct expression *expr)
528 {
529 if (expr->type != EXPR_COMPARE)
530 return;
531
532 if (expr->op == SPECIAL_EQUAL ||
533 expr->op == SPECIAL_NOTEQUAL) {
534 handle_eq_noteq(expr);
535 return;
536 }
537
538 handle_unsigned_lt_gt(expr);
539 }
540
541 static void match_user_assign_function(const char *fn, struct expression *expr, void *unused)
542 {
543 func_gets_user_data = true;
544
545 tag_as_user_data(expr->left);
546 set_points_to_user_data(expr->left);
547 }
548
549 static void match_simple_strtoul(const char *fn, struct expression *expr, void *unused)
550 {
551 func_gets_user_data = true;
552
553 set_state_expr(my_id, expr->left, alloc_estate_whole(get_type(expr->left)));
554 }
555
556 static int get_user_macro_rl(struct expression *expr, struct range_list **rl)
557 {
558 struct expression *parent;
559 char *macro;
560
561 if (!expr)
562 return 0;
563
564 macro = get_macro_name(expr->pos);
565 if (!macro)
566 return 0;
567
568 /* handle ntohl(foo[i]) where "i" is trusted */
569 parent = expr_get_parent_expr(expr);
570 if (parent && parent->type == EXPR_BINOP) {
571 char *parent_macro = get_macro_name(parent->pos);
572
573 if (parent_macro && strcmp(macro, parent_macro) == 0)
574 return 0;
575 }
576
577 if (strcmp(macro, "ntohl") == 0) {
578 *rl = alloc_whole_rl(&uint_ctype);
579 return 1;
580 }
581 if (strcmp(macro, "ntohs") == 0) {
582 *rl = alloc_whole_rl(&ushort_ctype);
583 return 1;
584 }
585 return 0;
586 }
587
588 struct db_info {
589 struct range_list *rl;
590 struct expression *call;
591 };
592 static int returned_rl_callback(void *_info, int argc, char **argv, char **azColName)
593 {
594 struct db_info *db_info = _info;
595 struct range_list *rl;
596 char *return_ranges = argv[0];
597 char *user_ranges = argv[1];
598 struct expression *arg;
599 int comparison;
600
601 if (argc != 2)
602 return 0;
603
604 call_results_to_rl(db_info->call, get_type(db_info->call), user_ranges, &rl);
605 if (str_to_comparison_arg(return_ranges, db_info->call, &comparison, &arg) &&
606 comparison == SPECIAL_EQUAL) {
607 struct range_list *orig_rl;
608
609 if (!get_user_rl(arg, &orig_rl))
610 return 0;
611 rl = rl_intersection(rl, orig_rl);
612 if (!rl)
613 return 0;
614 }
615 db_info->rl = rl_union(db_info->rl, rl);
616
617 return 0;
618 }
619
620 static int has_user_data(struct symbol *sym)
621 {
622 struct sm_state *tmp;
623
624 FOR_EACH_MY_SM(my_id, __get_cur_stree(), tmp) {
625 if (tmp->sym == sym)
626 return 1;
627 } END_FOR_EACH_SM(tmp);
628 return 0;
629 }
630
631 static int we_pass_user_data(struct expression *call)
632 {
633 struct expression *arg;
634 struct symbol *sym;
635
636 FOR_EACH_PTR(call->args, arg) {
637 sym = expr_to_sym(arg);
638 if (!sym)
639 continue;
640 if (has_user_data(sym))
641 return 1;
642 } END_FOR_EACH_PTR(arg);
643
644 return 0;
645 }
646
647 static int db_returned_user_rl(struct expression *call, struct range_list **rl)
648 {
649 struct db_info db_info = {};
650
651 /* for function pointers assume everything is used */
652 if (call->fn->type != EXPR_SYMBOL)
653 return 0;
654 if (is_fake_call(call))
655 return 0;
656
657 db_info.call = call;
658 run_sql(&returned_rl_callback, &db_info,
659 "select return, value from return_states where %s and type = %d and parameter = -1 and key = '$';",
660 get_static_filter(call->fn->symbol), USER_DATA3_SET);
661 if (db_info.rl) {
662 func_gets_user_data = true;
663 *rl = db_info.rl;
664 return 1;
665 }
666
667 run_sql(&returned_rl_callback, &db_info,
668 "select return, value from return_states where %s and type = %d and parameter = -1 and key = '$';",
669 get_static_filter(call->fn->symbol), USER_DATA3);
670 if (db_info.rl) {
671 if (!we_pass_user_data(call))
672 return 0;
673 *rl = db_info.rl;
674 return 1;
675 }
676
677 return 0;
678 }
679
680 static int user_data_flag;
681 static int no_user_data_flag;
682 static struct range_list *var_user_rl(struct expression *expr)
683 {
684 struct smatch_state *state;
685 struct range_list *rl;
686 struct range_list *absolute_rl;
687
688 if (expr->type == EXPR_BINOP && expr->op == '%') {
689 struct range_list *left, *right;
690
691 if (!get_user_rl(expr->right, &right))
692 return NULL;
693 get_absolute_rl(expr->left, &left);
694 rl = rl_binop(left, '%', right);
695 goto found;
696 }
697
698 if (expr->type == EXPR_BINOP && expr->op == '/') {
699 struct range_list *left = NULL;
700 struct range_list *right = NULL;
701 struct range_list *abs_right;
702
703 /*
704 * The specific bug I'm dealing with is:
705 *
706 * foo = capped_user / unknown;
707 *
708 * Instead of just saying foo is now entirely user_rl we should
709 * probably say instead that it is not at all user data.
710 *
711 */
712
713 get_user_rl(expr->left, &left);
714 get_user_rl(expr->right, &right);
715 get_absolute_rl(expr->right, &abs_right);
716
717 if (left && !right) {
718 rl = rl_binop(left, '/', abs_right);
719 if (sval_cmp(rl_max(left), rl_max(rl)) < 0)
720 no_user_data_flag = 1;
721 }
722
723 return NULL;
724 }
725
726 if (get_user_macro_rl(expr, &rl))
727 goto found;
728
729 if (comes_from_skb_data(expr)) {
730 rl = alloc_whole_rl(get_type(expr));
731 goto found;
732 }
733
734 state = get_state_expr(my_id, expr);
735 if (state && estate_rl(state)) {
736 rl = estate_rl(state);
737 goto found;
738 }
739
740 if (expr->type == EXPR_CALL && db_returned_user_rl(expr, &rl))
741 goto found;
742
743 if (is_array(expr)) {
744 struct expression *array = get_array_base(expr);
745
746 if (!get_state_expr(my_id, array)) {
747 no_user_data_flag = 1;
748 return NULL;
749 }
750 }
751
752 return NULL;
753 found:
754 user_data_flag = 1;
755 absolute_rl = var_to_absolute_rl(expr);
756 return clone_rl(rl_intersection(rl, absolute_rl));
757 }
758
759 int get_user_rl(struct expression *expr, struct range_list **rl)
760 {
761 user_data_flag = 0;
762 no_user_data_flag = 0;
763 custom_get_absolute_rl(expr, &var_user_rl, rl);
764 if (!user_data_flag || no_user_data_flag)
765 *rl = NULL;
766
767 return !!*rl;
768 }
769
770 int get_user_rl_var_sym(const char *name, struct symbol *sym, struct range_list **rl)
771 {
772 struct smatch_state *state;
773
774 state = get_state(my_id, name, sym);
775 if (state && estate_rl(state)) {
776 *rl = estate_rl(state);
777 return 1;
778 }
779 return 0;
780 }
781
782 static void match_call_info(struct expression *expr)
783 {
784 struct range_list *rl;
785 struct expression *arg;
786 int i = 0;
787
788 i = -1;
789 FOR_EACH_PTR(expr->args, arg) {
790 i++;
791
792 if (!get_user_rl(arg, &rl))
793 continue;
794
795 sql_insert_caller_info(expr, USER_DATA3, i, "$", show_rl(rl));
796 } END_FOR_EACH_PTR(arg);
797 }
798
799 static void struct_member_callback(struct expression *call, int param, char *printed_name, struct sm_state *sm)
800 {
801 struct smatch_state *state;
802 struct range_list *rl;
803
804 if (strcmp(sm->state->name, "") == 0)
805 return;
806
807 state = get_state(SMATCH_EXTRA, sm->name, sm->sym);
808 if (!state || !estate_rl(state))
809 rl = estate_rl(sm->state);
810 else
811 rl = rl_intersection(estate_rl(sm->state), estate_rl(state));
812
813 sql_insert_caller_info(call, USER_DATA3, param, printed_name, show_rl(rl));
814 }
815
816 static void set_param_user_data(const char *name, struct symbol *sym, char *key, char *value)
817 {
818 struct range_list *rl = NULL;
819 struct smatch_state *state;
820 struct symbol *type;
821 char fullname[256];
822
823 if (strcmp(key, "*$") == 0)
824 snprintf(fullname, sizeof(fullname), "*%s", name);
825 else if (strncmp(key, "$", 1) == 0)
826 snprintf(fullname, 256, "%s%s", name, key + 1);
827 else
828 return;
829
830 type = get_member_type_from_key(symbol_expression(sym), key);
831
832 /* if the caller passes a void pointer with user data */
833 if (strcmp(key, "*$") == 0 && type && type != &void_ctype) {
834 struct expression *expr = symbol_expression(sym);
835
836 tag_as_user_data(expr);
837 set_points_to_user_data(expr);
838 return;
839 }
840 str_to_rl(type, value, &rl);
841 state = alloc_estate_rl(rl);
842 set_state(my_id, fullname, sym, state);
843 }
844
845 static void set_called(const char *name, struct symbol *sym, char *key, char *value)
846 {
847 set_state(my_call_id, "this_function", NULL, &called);
848 }
849
850 static void match_syscall_definition(struct symbol *sym)
851 {
852 struct symbol *arg;
853 char *macro;
854 char *name;
855 int is_syscall = 0;
856
857 macro = get_macro_name(sym->pos);
858 if (macro &&
859 (strncmp("SYSCALL_DEFINE", macro, strlen("SYSCALL_DEFINE")) == 0 ||
860 strncmp("COMPAT_SYSCALL_DEFINE", macro, strlen("COMPAT_SYSCALL_DEFINE")) == 0))
861 is_syscall = 1;
862
863 name = get_function();
864 if (!option_no_db && get_state(my_call_id, "this_function", NULL) != &called) {
865 if (name && strncmp(name, "sys_", 4) == 0)
866 is_syscall = 1;
867 }
868
869 if (name && strncmp(name, "compat_sys_", 11) == 0)
870 is_syscall = 1;
871
872 if (!is_syscall)
873 return;
874
875 FOR_EACH_PTR(sym->ctype.base_type->arguments, arg) {
876 set_state(my_id, arg->ident->name, arg, alloc_estate_whole(get_real_base_type(arg)));
877 } END_FOR_EACH_PTR(arg);
878 }
879
880 static void set_to_user_data(struct expression *expr, char *key, char *value)
881 {
882 char *name;
883 struct symbol *sym;
884 struct symbol *type;
885 struct range_list *rl = NULL;
886
887 type = get_member_type_from_key(expr, key);
888 name = get_variable_from_key(expr, key, &sym);
889 if (!name || !sym)
890 goto free;
891
892 call_results_to_rl(expr, type, value, &rl);
893
894 set_state(my_id, name, sym, alloc_estate_rl(rl));
895 free:
896 free_string(name);
897
898 }
899
900 static void returns_param_user_data(struct expression *expr, int param, char *key, char *value)
901 {
902 struct expression *arg;
903 struct expression *call;
904
905 call = expr;
906 while (call->type == EXPR_ASSIGNMENT)
907 call = strip_expr(call->right);
908 if (call->type != EXPR_CALL)
909 return;
910
911 if (!we_pass_user_data(call))
912 return;
913
914 if (param == -1) {
915 if (expr->type != EXPR_ASSIGNMENT)
916 return;
917 set_to_user_data(expr->left, key, value);
918 return;
919 }
920
921 arg = get_argument_from_call_expr(call->args, param);
922 if (!arg)
923 return;
924 set_to_user_data(arg, key, value);
925 }
926
927 static void returns_param_user_data_set(struct expression *expr, int param, char *key, char *value)
928 {
929 struct expression *arg;
930
931 func_gets_user_data = true;
932
933 if (param == -1) {
934 if (expr->type != EXPR_ASSIGNMENT)
935 return;
936 if (strcmp(key, "*$") == 0) {
937 set_points_to_user_data(expr->left);
938 tag_as_user_data(expr->left);
939 } else {
940 set_to_user_data(expr->left, key, value);
941 }
942 return;
943 }
944
945 while (expr->type == EXPR_ASSIGNMENT)
946 expr = strip_expr(expr->right);
947 if (expr->type != EXPR_CALL)
948 return;
949
950 arg = get_argument_from_call_expr(expr->args, param);
951 if (!arg)
952 return;
953 set_to_user_data(arg, key, value);
954 }
955
956 static int has_empty_state(struct sm_state *sm)
957 {
958 struct sm_state *tmp;
959
960 FOR_EACH_PTR(sm->possible, tmp) {
961 if (!estate_rl(tmp->state))
962 return 1;
963 } END_FOR_EACH_PTR(tmp);
964
965 return 0;
966 }
967
968 static void param_set_to_user_data(int return_id, char *return_ranges, struct expression *expr)
969 {
970 struct sm_state *sm;
971 struct smatch_state *start_state;
972 struct range_list *rl;
973 int param;
974 char *return_str;
975 const char *param_name;
976
977 expr = strip_expr(expr);
978 return_str = expr_to_str(expr);
979
980 FOR_EACH_MY_SM(my_id, __get_cur_stree(), sm) {
981 if (has_empty_state(sm))
982 continue;
983
984 param = get_param_num_from_sym(sm->sym);
985 if (param < 0) {
986 if (expr_to_sym(expr) == sm->sym)
987 param = -1;
988 else
989 continue;
990 }
991
992 /* The logic here was that if we were passed in a user data then
993 * we don't record that. It's like the difference between
994 * param_filter and param_set. When I think about it, I'm not
995 * sure it actually works. It's probably harmless because we
996 * checked earlier that we're not returning a parameter...
997 * Let's mark this as a TODO.
998 */
999 start_state = get_state_stree(start_states, my_id, sm->name, sm->sym);
1000 if (start_state && rl_equiv(estate_rl(sm->state), estate_rl(start_state)))
1001 continue;
1002
1003 if (param == -1)
1004 param_name = state_name_to_param_name(sm->name, return_str);
1005 else
1006 param_name = get_param_name(sm);
1007 if (!param_name)
1008 continue;
1009 if (strcmp(param_name, "$") == 0) /* The -1 param is handled after the loop */
1010 continue;
1011
1012 sql_insert_return_states(return_id, return_ranges,
1013 func_gets_user_data ? USER_DATA3_SET : USER_DATA3,
1014 param, param_name, show_rl(estate_rl(sm->state)));
1015 } END_FOR_EACH_SM(sm);
1016
1017 if (points_to_user_data(expr)) {
1018 sql_insert_return_states(return_id, return_ranges,
1019 (is_skb_data(expr) || !func_gets_user_data) ?
1020 USER_DATA3_SET : USER_DATA3,
1021 -1, "*$", "");
1022 } else if (get_user_rl(expr, &rl)) {
1023 sql_insert_return_states(return_id, return_ranges,
1024 func_gets_user_data ? USER_DATA3_SET : USER_DATA3,
1025 -1, "$", show_rl(rl));
1026 }
1027
1028 free_string(return_str);
1029 }
1030
1031 static struct int_stack *gets_data_stack;
1032 static void match_function_def(struct symbol *sym)
1033 {
1034 func_gets_user_data = false;
1035 }
1036
1037 static void match_inline_start(struct expression *expr)
1038 {
1039 push_int(&gets_data_stack, func_gets_user_data);
1040 }
1041
1042 static void match_inline_end(struct expression *expr)
1043 {
1044 func_gets_user_data = pop_int(&gets_data_stack);
1045 }
1046
1047 void check_user_data2(int id)
1048 {
1049 int i;
1050
1051 my_id = id;
1052
1053 if (option_project != PROJ_KERNEL)
1054 return;
1055
1056 add_hook(&match_function_def, FUNC_DEF_HOOK);
1057 add_hook(&match_inline_start, INLINE_FN_START);
1058 add_hook(&match_inline_end, INLINE_FN_END);
1059
1060 add_hook(&save_start_states, AFTER_DEF_HOOK);
1061 add_hook(&free_start_states, AFTER_FUNC_HOOK);
1062 add_hook(&match_save_states, INLINE_FN_START);
1063 add_hook(&match_restore_states, INLINE_FN_END);
1064
1065 add_unmatched_state_hook(my_id, &empty_state);
1066 add_extra_nomod_hook(&extra_nomod_hook);
1067 add_pre_merge_hook(my_id, &pre_merge_hook);
1068 add_merge_hook(my_id, &merge_estates);
1069
1070 add_function_hook("copy_from_user", &match_user_copy, INT_PTR(0));
1071 add_function_hook("__copy_from_user", &match_user_copy, INT_PTR(0));
1072 add_function_hook("memcpy_fromiovec", &match_user_copy, INT_PTR(0));
1073 for (i = 0; i < ARRAY_SIZE(kstr_funcs); i++)
1074 add_function_hook(kstr_funcs[i], &match_user_copy, INT_PTR(2));
1075
1076 add_function_assign_hook("simple_strtol", &match_simple_strtoul, NULL);
1077 add_function_assign_hook("simple_strtoll", &match_simple_strtoul, NULL);
1078 add_function_assign_hook("simple_strtoul", &match_simple_strtoul, NULL);
1079 add_function_assign_hook("simple_strtoull", &match_simple_strtoul, NULL);
1080
1081 add_function_hook("sscanf", &match_sscanf, NULL);
1082
1083 add_function_assign_hook("memdup_user", &match_user_assign_function, NULL);
1084 add_function_assign_hook("kmap_atomic", &match_user_assign_function, NULL);
1085 add_function_assign_hook("skb_network_header", &match_user_assign_function, NULL);
1086
1087 add_hook(&match_syscall_definition, AFTER_DEF_HOOK);
1088
1089 add_hook(&match_assign, ASSIGNMENT_HOOK);
1090 add_hook(&match_condition, CONDITION_HOOK);
1091
1092 add_hook(&match_call_info, FUNCTION_CALL_HOOK);
1093 add_member_info_callback(my_id, struct_member_callback);
1094 select_caller_info_hook(set_param_user_data, USER_DATA3);
1095 select_return_states_hook(USER_DATA3, &returns_param_user_data);
1096 select_return_states_hook(USER_DATA3_SET, &returns_param_user_data_set);
1097 add_split_return_callback(&param_set_to_user_data);
1098 }
1099
1100 void check_user_data3(int id)
1101 {
1102 my_call_id = id;
1103
1104 if (option_project != PROJ_KERNEL)
1105 return;
1106 select_caller_info_hook(set_called, INTERNAL);
1107 }
1108
Static analysis for C