buf_size: allow strncmp("foo", bar, 100) where 100 is larger than "foo"
[smatch.git] / smatch_flow.c
blobcbdc53d45e78d777c8d302452f07d3d2721941a6
1 /*
2 * Copyright (C) 2006,2008 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 #define _GNU_SOURCE 1
19 #include <unistd.h>
20 #include <stdio.h>
21 #include "token.h"
22 #include "scope.h"
23 #include "smatch.h"
24 #include "smatch_expression_stacks.h"
25 #include "smatch_extra.h"
26 #include "smatch_slist.h"
28 int __in_fake_assign;
29 int final_pass;
30 int __inline_call;
31 struct expression *__inline_fn;
33 static int __smatch_lineno = 0;
35 static char *base_file;
36 static const char *filename;
37 static char *pathname;
38 static char *full_filename;
39 static char *cur_func;
40 static unsigned int loop_count;
41 int __expr_stmt_count;
42 int __in_function_def;
43 static struct expression_list *switch_expr_stack = NULL;
44 static struct expression_list *post_op_stack = NULL;
46 struct expression_list *big_expression_stack;
47 struct statement_list *big_statement_stack;
48 int __in_pre_condition = 0;
49 int __bail_on_rest_of_function = 0;
50 static struct timeval fn_start_time;
51 char *get_function(void) { return cur_func; }
52 int get_lineno(void) { return __smatch_lineno; }
53 int inside_loop(void) { return !!loop_count; }
54 int definitely_inside_loop(void) { return !!(loop_count & ~0x80000000); }
55 struct expression *get_switch_expr(void) { return top_expression(switch_expr_stack); }
56 int in_expression_statement(void) { return !!__expr_stmt_count; }
58 static void split_symlist(struct symbol_list *sym_list);
59 static void split_declaration(struct symbol_list *sym_list);
60 static void split_expr_list(struct expression_list *expr_list);
61 static void add_inline_function(struct symbol *sym);
62 static void parse_inline(struct expression *expr);
64 int option_assume_loops = 0;
65 int option_known_conditions = 0;
66 int option_two_passes = 0;
67 struct symbol *cur_func_sym = NULL;
69 int outside_of_function(void)
71 return cur_func_sym == NULL;
74 const char *get_filename(void)
76 if (option_info)
77 return base_file;
78 if (option_full_path)
79 return full_filename;
80 return filename;
83 const char *get_base_file(void)
85 return base_file;
88 static void set_position(struct position pos)
90 int len;
91 static int prev_stream = -1;
93 if (pos.stream == 0 && pos.line == 0)
94 return;
96 __smatch_lineno = pos.line;
98 if (pos.stream == prev_stream)
99 return;
101 filename = stream_name(pos.stream);
103 free(full_filename);
104 pathname = getcwd(NULL, 0);
105 if (pathname) {
106 len = strlen(pathname) + 1 + strlen(filename) + 1;
107 full_filename = malloc(len);
108 snprintf(full_filename, len, "%s/%s", pathname, filename);
109 } else {
110 full_filename = alloc_string(filename);
112 free(pathname);
115 static int is_inline_func(struct expression *expr)
117 if (expr->type != EXPR_SYMBOL || !expr->symbol)
118 return 0;
119 if (expr->symbol->ctype.modifiers & MOD_INLINE)
120 return 1;
121 return 0;
124 static int is_noreturn_func(struct expression *expr)
126 if (expr->type != EXPR_SYMBOL || !expr->symbol)
127 return 0;
128 if (expr->symbol->ctype.modifiers & MOD_NORETURN)
129 return 1;
130 return 0;
133 int inlinable(struct expression *expr)
135 struct symbol *sym;
137 if (__inline_fn) /* don't nest */
138 return 0;
140 if (expr->type != EXPR_SYMBOL || !expr->symbol)
141 return 0;
142 if (is_no_inline_function(expr->symbol->ident->name))
143 return 0;
144 sym = get_base_type(expr->symbol);
145 if (sym->stmt && sym->stmt->type == STMT_COMPOUND) {
146 if (ptr_list_size((struct ptr_list *)sym->stmt->stmts) <= 10)
147 return 1;
148 return 0;
150 if (sym->inline_stmt && sym->inline_stmt->type == STMT_COMPOUND) {
151 if (ptr_list_size((struct ptr_list *)sym->inline_stmt->stmts) <= 10)
152 return 1;
153 return 0;
155 return 0;
158 void __process_post_op_stack(void)
160 struct expression *expr;
162 FOR_EACH_PTR(post_op_stack, expr) {
163 __pass_to_client(expr, OP_HOOK);
164 } END_FOR_EACH_PTR(expr);
166 __free_ptr_list((struct ptr_list **)&post_op_stack);
169 void __split_expr(struct expression *expr)
171 if (!expr)
172 return;
174 // sm_msg(" Debug expr_type %d %s", expr->type, show_special(expr->op));
176 if (__in_fake_assign && expr->type != EXPR_ASSIGNMENT)
177 return;
178 if (__in_fake_assign >= 4) /* don't allow too much nesting */
179 return;
181 push_expression(&big_expression_stack, expr);
182 set_position(expr->pos);
183 __pass_to_client(expr, EXPR_HOOK);
185 switch (expr->type) {
186 case EXPR_PREOP:
187 if (expr->op == '*')
188 __pass_to_client(expr, DEREF_HOOK);
189 __split_expr(expr->unop);
190 __pass_to_client(expr, OP_HOOK);
191 break;
192 case EXPR_POSTOP:
193 __split_expr(expr->unop);
194 push_expression(&post_op_stack, expr);
195 break;
196 case EXPR_STATEMENT:
197 __expr_stmt_count++;
198 __split_stmt(expr->statement);
199 __expr_stmt_count--;
200 break;
201 case EXPR_LOGICAL:
202 case EXPR_COMPARE:
203 __pass_to_client(expr, LOGIC_HOOK);
204 __handle_logic(expr);
205 break;
206 case EXPR_BINOP:
207 __pass_to_client(expr, BINOP_HOOK);
208 case EXPR_COMMA:
209 __split_expr(expr->left);
210 __process_post_op_stack();
211 __split_expr(expr->right);
212 break;
213 case EXPR_ASSIGNMENT: {
214 struct expression *tmp;
216 if (!expr->right)
217 break;
219 __pass_to_client(expr, RAW_ASSIGNMENT_HOOK);
221 /* foo = !bar() */
222 if (__handle_condition_assigns(expr))
223 break;
224 /* foo = (x < 5 ? foo : 5); */
225 if (__handle_select_assigns(expr))
226 break;
227 /* foo = ({frob(); frob(); frob(); 1;}) */
228 if (__handle_expr_statement_assigns(expr))
229 break;
231 __split_expr(expr->right);
232 if (outside_of_function())
233 __pass_to_client(expr, GLOBAL_ASSIGNMENT_HOOK);
234 else
235 __pass_to_client(expr, ASSIGNMENT_HOOK);
237 __fake_struct_member_assignments(expr);
239 tmp = strip_expr(expr->right);
240 if (tmp->type == EXPR_CALL)
241 __pass_to_client(expr, CALL_ASSIGNMENT_HOOK);
242 if (get_macro_name(tmp->pos) &&
243 get_macro_name(expr->pos) != get_macro_name(tmp->pos))
244 __pass_to_client(expr, MACRO_ASSIGNMENT_HOOK);
245 __split_expr(expr->left);
246 break;
248 case EXPR_DEREF:
249 __pass_to_client(expr, DEREF_HOOK);
250 __split_expr(expr->deref);
251 break;
252 case EXPR_SLICE:
253 __split_expr(expr->base);
254 break;
255 case EXPR_CAST:
256 case EXPR_FORCE_CAST:
257 __pass_to_client(expr, CAST_HOOK);
258 __split_expr(expr->cast_expression);
259 break;
260 case EXPR_SIZEOF:
261 if (expr->cast_expression)
262 __pass_to_client(strip_parens(expr->cast_expression),
263 SIZEOF_HOOK);
264 break;
265 case EXPR_OFFSETOF:
266 case EXPR_ALIGNOF:
267 evaluate_expression(expr);
268 break;
269 case EXPR_CONDITIONAL:
270 case EXPR_SELECT:
271 if (known_condition_true(expr->conditional)) {
272 __split_expr(expr->cond_true);
273 break;
275 if (known_condition_false(expr->conditional)) {
276 __split_expr(expr->cond_false);
277 break;
279 __pass_to_client(expr, SELECT_HOOK);
280 __split_whole_condition(expr->conditional);
281 __split_expr(expr->cond_true);
282 __push_true_states();
283 __use_false_states();
284 __split_expr(expr->cond_false);
285 __merge_true_states();
286 break;
287 case EXPR_CALL:
288 if (sym_name_is("__builtin_constant_p", expr->fn))
289 break;
290 split_expr_list(expr->args);
291 __split_expr(expr->fn);
292 if (is_inline_func(expr->fn))
293 add_inline_function(expr->fn->symbol);
294 if (inlinable(expr->fn))
295 __inline_call = 1;
296 __process_post_op_stack();
297 __pass_to_client(expr, FUNCTION_CALL_HOOK);
298 __inline_call = 0;
299 if (inlinable(expr->fn)) {
300 parse_inline(expr);
302 __pass_to_client(expr, CALL_HOOK_AFTER_INLINE);
303 if (is_noreturn_func(expr->fn))
304 nullify_path();
305 break;
306 case EXPR_INITIALIZER:
307 split_expr_list(expr->expr_list);
308 break;
309 case EXPR_IDENTIFIER:
310 __split_expr(expr->ident_expression);
311 break;
312 case EXPR_INDEX:
313 __split_expr(expr->idx_expression);
314 break;
315 case EXPR_POS:
316 __split_expr(expr->init_expr);
317 break;
318 case EXPR_SYMBOL:
319 __pass_to_client(expr, SYM_HOOK);
320 break;
321 case EXPR_STRING:
322 __pass_to_client(expr, STRING_HOOK);
323 break;
324 default:
325 break;
327 pop_expression(&big_expression_stack);
330 static int is_forever_loop(struct statement *stmt)
332 struct expression *expr;
334 expr = strip_expr(stmt->iterator_pre_condition);
335 if (!expr)
336 expr = stmt->iterator_post_condition;
337 if (!expr) {
338 /* this is a for(;;) loop... */
339 return 1;
342 if (expr->type == EXPR_VALUE && expr->value == 1)
343 return 1;
345 return 0;
348 static int loop_num;
349 static char *get_loop_name(int num)
351 char buf[256];
353 snprintf(buf, 255, "-loop%d", num);
354 buf[255] = '\0';
355 return alloc_sname(buf);
359 * Pre Loops are while and for loops.
361 static void handle_pre_loop(struct statement *stmt)
363 int once_through; /* we go through the loop at least once */
364 struct sm_state *extra_sm = NULL;
365 int unchanged = 0;
366 char *loop_name;
367 struct stree *stree = NULL;
368 struct sm_state *sm = NULL;
370 loop_name = get_loop_name(loop_num);
371 loop_num++;
373 __split_stmt(stmt->iterator_pre_statement);
375 once_through = implied_condition_true(stmt->iterator_pre_condition);
377 loop_count++;
378 __push_continues();
379 __push_breaks();
381 __merge_gotos(loop_name);
383 extra_sm = __extra_handle_canonical_loops(stmt, &stree);
384 __in_pre_condition++;
385 __pass_to_client(stmt, PRELOOP_HOOK);
386 __split_whole_condition(stmt->iterator_pre_condition);
387 __in_pre_condition--;
388 FOR_EACH_SM(stree, sm) {
389 set_state(sm->owner, sm->name, sm->sym, sm->state);
390 } END_FOR_EACH_SM(sm);
391 free_stree(&stree);
392 if (extra_sm)
393 extra_sm = get_sm_state(extra_sm->owner, extra_sm->name, extra_sm->sym);
395 if (option_assume_loops)
396 once_through = 1;
398 __split_stmt(stmt->iterator_statement);
399 __warn_on_silly_pre_loops();
400 if (is_forever_loop(stmt)) {
401 __save_gotos(loop_name);
403 __push_fake_cur_stree();
404 __split_stmt(stmt->iterator_post_statement);
405 stree = __pop_fake_cur_stree();
407 __discard_continues();
408 __discard_false_states();
409 __use_breaks();
411 if (!__path_is_null())
412 __merge_stree_into_cur(stree);
413 free_stree(&stree);
414 } else {
415 __merge_continues();
416 unchanged = __iterator_unchanged(extra_sm);
417 __split_stmt(stmt->iterator_post_statement);
418 __save_gotos(loop_name);
419 __split_whole_condition(stmt->iterator_pre_condition);
420 nullify_path();
421 __merge_false_states();
422 if (once_through)
423 __discard_false_states();
424 else
425 __merge_false_states();
427 if (extra_sm && unchanged)
428 __extra_pre_loop_hook_after(extra_sm,
429 stmt->iterator_post_statement,
430 stmt->iterator_pre_condition);
431 __merge_breaks();
433 loop_count--;
437 * Post loops are do {} while();
439 static void handle_post_loop(struct statement *stmt)
441 char *loop_name;
443 loop_name = get_loop_name(loop_num);
444 loop_num++;
445 loop_count++;
447 __push_continues();
448 __push_breaks();
449 __merge_gotos(loop_name);
450 __split_stmt(stmt->iterator_statement);
451 __merge_continues();
452 if (!is_zero(stmt->iterator_post_condition))
453 __save_gotos(loop_name);
455 if (is_forever_loop(stmt)) {
456 __use_breaks();
457 } else {
458 __split_whole_condition(stmt->iterator_post_condition);
459 __use_false_states();
460 __merge_breaks();
462 loop_count--;
465 static int empty_statement(struct statement *stmt)
467 if (!stmt)
468 return 0;
469 if (stmt->type == STMT_EXPRESSION && !stmt->expression)
470 return 1;
471 return 0;
474 static int last_stmt_on_same_line()
476 struct statement *stmt;
477 int i = 0;
479 FOR_EACH_PTR_REVERSE(big_statement_stack, stmt) {
480 if (!i++)
481 continue;
482 if (stmt->pos.line == get_lineno())
483 return 1;
484 return 0;
485 } END_FOR_EACH_PTR_REVERSE(stmt);
486 return 0;
489 static struct statement *last_stmt;
490 static int is_last_stmt(struct statement *stmt)
492 if (stmt == last_stmt)
493 return 1;
494 return 0;
497 static void print_unreached_initializers(struct symbol_list *sym_list)
499 struct symbol *sym;
501 FOR_EACH_PTR(sym_list, sym) {
502 if (sym->initializer)
503 sm_msg("info: '%s' is not actually initialized (unreached code).",
504 (sym->ident ? sym->ident->name : "this variable"));
505 } END_FOR_EACH_PTR(sym);
508 static void print_unreached(struct statement *stmt)
510 static int print = 1;
512 if (__inline_fn)
513 return;
515 if (!__path_is_null()) {
516 print = 1;
517 return;
519 if (!print)
520 return;
522 switch (stmt->type) {
523 case STMT_COMPOUND: /* after a switch before a case stmt */
524 case STMT_RANGE:
525 case STMT_CASE:
526 case STMT_LABEL:
527 return;
528 case STMT_DECLARATION: /* switch (x) { int a; case foo: ... */
529 print_unreached_initializers(stmt->declaration);
530 return;
531 case STMT_RETURN: /* gcc complains if you don't have a return statement */
532 if (is_last_stmt(stmt))
533 return;
534 break;
535 case STMT_GOTO:
536 /* people put extra breaks inside switch statements */
537 if (stmt->goto_label && stmt->goto_label->type == SYM_NODE &&
538 strcmp(stmt->goto_label->ident->name, "break") == 0)
539 return;
540 break;
541 default:
542 break;
544 if (empty_statement(stmt))
545 return;
546 if (!option_spammy)
547 return;
548 sm_msg("info: ignoring unreachable code.");
549 print = 0;
552 static void split_asm_constraints(struct expression_list *expr_list)
554 struct expression *expr;
555 int state = 0;
557 FOR_EACH_PTR(expr_list, expr) {
558 switch (state) {
559 case 0: /* identifier */
560 case 1: /* constraint */
561 state++;
562 continue;
563 case 2: /* expression */
564 state = 0;
565 __split_expr(expr);
566 continue;
568 } END_FOR_EACH_PTR(expr);
571 static int is_case_val(struct statement *stmt, sval_t sval)
573 sval_t case_sval;
575 if (stmt->type != STMT_CASE)
576 return 0;
577 if (!stmt->case_expression) {
578 __set_default();
579 return 1;
581 if (!get_value(stmt->case_expression, &case_sval))
582 return 0;
583 if (case_sval.value == sval.value)
584 return 1;
585 return 0;
588 static void split_known_switch(struct statement *stmt, sval_t sval)
590 struct statement *tmp;
592 __split_expr(stmt->switch_expression);
594 push_expression(&switch_expr_stack, stmt->switch_expression);
595 __save_switch_states(top_expression(switch_expr_stack));
596 nullify_path();
597 __push_default();
598 __push_breaks();
600 stmt = stmt->switch_statement;
602 if (!last_stmt)
603 last_stmt = last_ptr_list((struct ptr_list *)stmt->stmts);
605 __push_scope_hooks();
606 FOR_EACH_PTR(stmt->stmts, tmp) {
607 __smatch_lineno = tmp->pos.line;
608 if (is_case_val(tmp, sval)) {
609 __merge_switches(top_expression(switch_expr_stack),
610 stmt->case_expression);
611 __pass_case_to_client(top_expression(switch_expr_stack),
612 stmt->case_expression);
614 if (__path_is_null())
615 continue;
616 __split_stmt(tmp);
617 if (__path_is_null()) {
618 __set_default();
619 goto out;
621 } END_FOR_EACH_PTR(tmp);
622 out:
623 __call_scope_hooks();
624 if (!__pop_default())
625 __merge_switches(top_expression(switch_expr_stack),
626 NULL);
627 __discard_switches();
628 __merge_breaks();
629 pop_expression(&switch_expr_stack);
632 static int taking_too_long(void)
634 int ms;
636 ms = ms_since(&fn_start_time);
637 if (ms > 1000 * 60 * 5) /* five minutes */
638 return 1;
639 return 0;
642 void __split_stmt(struct statement *stmt)
644 sval_t sval;
646 if (!stmt)
647 goto out;
649 if (__bail_on_rest_of_function || out_of_memory() || taking_too_long()) {
650 static char *printed = NULL;
652 __bail_on_rest_of_function = 1;
653 if (printed != cur_func)
654 sm_msg("Function too hairy. Giving up.");
655 final_pass = 0; /* turn off sm_msg() from here */
656 printed = cur_func;
657 return;
660 add_ptr_list(&big_statement_stack, stmt);
661 free_expression_stack(&big_expression_stack);
662 set_position(stmt->pos);
663 print_unreached(stmt);
664 __pass_to_client(stmt, STMT_HOOK);
666 switch (stmt->type) {
667 case STMT_DECLARATION:
668 split_declaration(stmt->declaration);
669 break;
670 case STMT_RETURN:
671 __split_expr(stmt->ret_value);
672 __pass_to_client(stmt->ret_value, RETURN_HOOK);
673 nullify_path();
674 break;
675 case STMT_EXPRESSION:
676 __split_expr(stmt->expression);
677 break;
678 case STMT_COMPOUND: {
679 struct statement *tmp;
681 if (!last_stmt)
682 last_stmt = last_ptr_list((struct ptr_list *)stmt->stmts);
683 __push_scope_hooks();
684 FOR_EACH_PTR(stmt->stmts, tmp) {
685 __split_stmt(tmp);
686 } END_FOR_EACH_PTR(tmp);
687 __call_scope_hooks();
688 break;
690 case STMT_IF:
691 if (known_condition_true(stmt->if_conditional)) {
692 __split_stmt(stmt->if_true);
693 break;
695 if (known_condition_false(stmt->if_conditional)) {
696 __split_stmt(stmt->if_false);
697 break;
699 if (option_known_conditions &&
700 implied_condition_true(stmt->if_conditional)) {
701 sm_info("this condition is true.");
702 __split_stmt(stmt->if_true);
703 break;
705 if (option_known_conditions &&
706 implied_condition_false(stmt->if_conditional)) {
707 sm_info("this condition is false.");
708 __split_stmt(stmt->if_false);
709 break;
711 __split_whole_condition(stmt->if_conditional);
712 __split_stmt(stmt->if_true);
713 if (empty_statement(stmt->if_true) &&
714 last_stmt_on_same_line() &&
715 !get_macro_name(stmt->if_true->pos))
716 sm_msg("warn: if();");
717 __push_true_states();
718 __use_false_states();
719 __split_stmt(stmt->if_false);
720 __merge_true_states();
721 break;
722 case STMT_ITERATOR:
723 if (stmt->iterator_pre_condition)
724 handle_pre_loop(stmt);
725 else if (stmt->iterator_post_condition)
726 handle_post_loop(stmt);
727 else {
728 // these are for(;;) type loops.
729 handle_pre_loop(stmt);
731 break;
732 case STMT_SWITCH:
733 if (get_value(stmt->switch_expression, &sval)) {
734 split_known_switch(stmt, sval);
735 break;
737 __split_expr(stmt->switch_expression);
738 push_expression(&switch_expr_stack, stmt->switch_expression);
739 __save_switch_states(top_expression(switch_expr_stack));
740 nullify_path();
741 __push_default();
742 __push_breaks();
743 __split_stmt(stmt->switch_statement);
744 if (!__pop_default())
745 __merge_switches(top_expression(switch_expr_stack),
746 NULL);
747 __discard_switches();
748 __merge_breaks();
749 pop_expression(&switch_expr_stack);
750 break;
751 case STMT_CASE:
752 __merge_switches(top_expression(switch_expr_stack),
753 stmt->case_expression);
754 __pass_case_to_client(top_expression(switch_expr_stack),
755 stmt->case_expression);
756 if (!stmt->case_expression)
757 __set_default();
758 __split_expr(stmt->case_expression);
759 __split_expr(stmt->case_to);
760 __split_stmt(stmt->case_statement);
761 break;
762 case STMT_LABEL:
763 if (stmt->label_identifier &&
764 stmt->label_identifier->type == SYM_LABEL &&
765 stmt->label_identifier->ident) {
766 loop_count |= 0x80000000;
767 __merge_gotos(stmt->label_identifier->ident->name);
769 __split_stmt(stmt->label_statement);
770 break;
771 case STMT_GOTO:
772 __split_expr(stmt->goto_expression);
773 if (stmt->goto_label && stmt->goto_label->type == SYM_NODE) {
774 if (!strcmp(stmt->goto_label->ident->name, "break")) {
775 __process_breaks();
776 } else if (!strcmp(stmt->goto_label->ident->name,
777 "continue")) {
778 __process_continues();
780 } else if (stmt->goto_label &&
781 stmt->goto_label->type == SYM_LABEL &&
782 stmt->goto_label->ident) {
783 __save_gotos(stmt->goto_label->ident->name);
785 nullify_path();
786 break;
787 case STMT_NONE:
788 break;
789 case STMT_ASM:
790 __pass_to_client(stmt, ASM_HOOK);
791 __split_expr(stmt->asm_string);
792 split_asm_constraints(stmt->asm_outputs);
793 split_asm_constraints(stmt->asm_inputs);
794 split_asm_constraints(stmt->asm_clobbers);
795 break;
796 case STMT_CONTEXT:
797 break;
798 case STMT_RANGE:
799 __split_expr(stmt->range_expression);
800 __split_expr(stmt->range_low);
801 __split_expr(stmt->range_high);
802 break;
804 __pass_to_client(stmt, STMT_HOOK_AFTER);
805 out:
806 __process_post_op_stack();
809 static void split_expr_list(struct expression_list *expr_list)
811 struct expression *expr;
813 FOR_EACH_PTR(expr_list, expr) {
814 __split_expr(expr);
815 __process_post_op_stack();
816 } END_FOR_EACH_PTR(expr);
819 static void split_sym(struct symbol *sym)
821 if (!sym)
822 return;
823 if (!(sym->namespace & NS_SYMBOL))
824 return;
826 __split_stmt(sym->stmt);
827 __split_expr(sym->array_size);
828 split_symlist(sym->arguments);
829 split_symlist(sym->symbol_list);
830 __split_stmt(sym->inline_stmt);
831 split_symlist(sym->inline_symbol_list);
834 static void split_symlist(struct symbol_list *sym_list)
836 struct symbol *sym;
838 FOR_EACH_PTR(sym_list, sym) {
839 split_sym(sym);
840 } END_FOR_EACH_PTR(sym);
843 typedef void (fake_cb)(struct expression *expr);
845 static int member_to_number(struct expression *expr, struct ident *member)
847 struct symbol *type, *tmp;
848 char *name;
849 int i;
851 if (!member)
852 return -1;
853 name = member->name;
855 type = get_type(expr);
856 if (!type || type->type != SYM_STRUCT)
857 return -1;
859 i = -1;
860 FOR_EACH_PTR(type->symbol_list, tmp) {
861 i++;
862 if (!tmp->ident)
863 continue;
864 if (strcmp(name, tmp->ident->name) == 0)
865 return i;
866 } END_FOR_EACH_PTR(tmp);
867 return -1;
870 static struct ident *number_to_member(struct expression *expr, int num)
872 struct symbol *type, *member;
873 int i = 0;
875 type = get_type(expr);
876 if (!type || type->type != SYM_STRUCT)
877 return NULL;
879 FOR_EACH_PTR(type->symbol_list, member) {
880 if (i == num)
881 return member->ident;
882 i++;
883 } END_FOR_EACH_PTR(member);
884 return NULL;
887 static void fake_element_assigns_helper(struct expression *array, struct expression_list *expr_list, fake_cb *fake_cb);
889 struct member_set {
890 struct ident *ident;
891 int set;
894 static struct member_set *alloc_member_set(struct symbol *type)
896 struct member_set *member_set;
897 struct symbol *member;
898 int member_count;
899 int member_idx;
901 member_count = ptr_list_size((struct ptr_list *)type->symbol_list);
902 member_set = malloc(member_count * sizeof(*member_set));
903 member_idx = 0;
904 FOR_EACH_PTR(type->symbol_list, member) {
905 member_set[member_idx].ident = member->ident;
906 member_set[member_idx].set = 0;
907 member_idx++;
908 } END_FOR_EACH_PTR(member);
910 return member_set;
913 static void mark_member_as_set(struct symbol *type, struct member_set *member_set, struct ident *ident)
915 int member_count = ptr_list_size((struct ptr_list *)type->symbol_list);
916 int i;
918 for (i = 0; i < member_count; i++) {
919 if (member_set[i].ident == ident) {
920 member_set[i].set = 1;
921 return;
924 // crap. this is buggy.
925 // sm_msg("internal smatch error in initializer %s.%s", type->ident->name, ident->name);
928 static void set_unset_to_zero(struct expression *symbol, struct symbol *type, struct member_set *member_set)
930 struct expression *deref, *assign;
931 struct symbol *member, *member_type;
932 int member_idx;
934 member_idx = 0;
935 FOR_EACH_PTR(type->symbol_list, member) {
936 if (!member->ident || member_set[member_idx].set) {
937 member_idx++;
938 continue;
940 member_type = get_real_base_type(member);
941 if (!member_type || member_type->type == SYM_ARRAY) {
942 member_idx++;
943 continue;
945 /* TODO: this should be handled recursively and not ignored */
946 if (member_type->type == SYM_STRUCT || member_type->type == SYM_UNION) {
947 member_idx++;
948 continue;
950 deref = member_expression(symbol, '.', member->ident);
951 assign = assign_expression(deref, zero_expr());
952 __split_expr(assign);
953 member_idx++;
954 } END_FOR_EACH_PTR(member);
958 static void fake_member_assigns_helper(struct expression *symbol, struct expression_list *members, fake_cb *fake_cb)
960 struct expression *deref, *assign, *tmp;
961 struct symbol *struct_type, *type;
962 struct ident *member;
963 int member_idx;
964 struct member_set *member_set;
966 struct_type = get_type(symbol);
967 if (!struct_type ||
968 (struct_type->type != SYM_STRUCT && struct_type->type != SYM_UNION))
969 return;
971 member_set = alloc_member_set(struct_type);
973 member_idx = 0;
974 FOR_EACH_PTR(members, tmp) {
975 member = number_to_member(symbol, member_idx);
976 while (tmp->type == EXPR_IDENTIFIER) {
977 member = tmp->expr_ident;
978 member_idx = member_to_number(symbol, member);
979 tmp = tmp->ident_expression;
981 mark_member_as_set(struct_type, member_set, member);
982 member_idx++;
983 deref = member_expression(symbol, '.', member);
984 if (tmp->type == EXPR_INITIALIZER) {
985 type = get_type(deref);
986 if (type && type->type == SYM_ARRAY)
987 fake_element_assigns_helper(deref, tmp->expr_list, fake_cb);
988 else
989 fake_member_assigns_helper(deref, tmp->expr_list, fake_cb);
990 } else {
991 assign = assign_expression(deref, tmp);
992 fake_cb(assign);
994 } END_FOR_EACH_PTR(tmp);
996 set_unset_to_zero(symbol, struct_type, member_set);
999 static void fake_member_assigns(struct symbol *sym, fake_cb *fake_cb)
1001 fake_member_assigns_helper(symbol_expression(sym),
1002 sym->initializer->expr_list, fake_cb);
1005 static void fake_element_assigns_helper(struct expression *array, struct expression_list *expr_list, fake_cb *fake_cb)
1007 struct expression *offset, *binop, *assign, *tmp;
1008 struct symbol *type;
1009 int idx;
1011 idx = 0;
1012 FOR_EACH_PTR(expr_list, tmp) {
1013 if (tmp->type == EXPR_INDEX) {
1014 if (tmp->idx_from != tmp->idx_to)
1015 return;
1016 idx = tmp->idx_from;
1017 if (!tmp->idx_expression)
1018 goto next;
1019 tmp = tmp->idx_expression;
1021 offset = value_expr(idx);
1022 binop = array_element_expression(array, offset);
1023 if (tmp->type == EXPR_INITIALIZER) {
1024 type = get_type(binop);
1025 if (type && type->type == SYM_ARRAY)
1026 fake_element_assigns_helper(binop, tmp->expr_list, fake_cb);
1027 else
1028 fake_member_assigns_helper(binop, tmp->expr_list, fake_cb);
1029 } else {
1030 assign = assign_expression(binop, tmp);
1031 fake_cb(assign);
1033 next:
1034 idx++;
1035 } END_FOR_EACH_PTR(tmp);
1038 static void fake_element_assigns(struct symbol *sym, fake_cb *fake_cb)
1040 fake_element_assigns_helper(symbol_expression(sym), sym->initializer->expr_list, fake_cb);
1043 static void fake_assign_expr(struct symbol *sym)
1045 struct expression *assign, *symbol;
1047 symbol = symbol_expression(sym);
1048 assign = assign_expression(symbol, sym->initializer);
1049 __split_expr(assign);
1052 static void call_split_expr(struct expression *expr)
1054 __split_expr(expr);
1057 static void do_initializer_stuff(struct symbol *sym)
1059 if (!sym->initializer)
1060 return;
1062 if (sym->initializer->type == EXPR_INITIALIZER) {
1063 if (get_real_base_type(sym)->type == SYM_ARRAY)
1064 fake_element_assigns(sym, call_split_expr);
1065 else
1066 fake_member_assigns(sym, call_split_expr);
1067 } else {
1068 fake_assign_expr(sym);
1072 static void split_declaration(struct symbol_list *sym_list)
1074 struct symbol *sym;
1076 FOR_EACH_PTR(sym_list, sym) {
1077 __pass_to_client(sym, DECLARATION_HOOK);
1078 do_initializer_stuff(sym);
1079 split_sym(sym);
1080 } END_FOR_EACH_PTR(sym);
1083 static void call_global_assign_hooks(struct expression *assign)
1085 __pass_to_client(assign, GLOBAL_ASSIGNMENT_HOOK);
1088 static void fake_global_assign(struct symbol *sym)
1090 struct expression *assign, *symbol;
1092 if (get_real_base_type(sym)->type == SYM_ARRAY) {
1093 if (sym->initializer && sym->initializer->type == EXPR_INITIALIZER) {
1094 fake_element_assigns(sym, call_global_assign_hooks);
1095 } else if (sym->initializer) {
1096 symbol = symbol_expression(sym);
1097 assign = assign_expression(symbol, sym->initializer);
1098 __pass_to_client(assign, GLOBAL_ASSIGNMENT_HOOK);
1099 } else {
1100 fake_element_assigns_helper(symbol_expression(sym), NULL, call_global_assign_hooks);
1102 } else if (get_real_base_type(sym)->type == SYM_STRUCT) {
1103 if (sym->initializer && sym->initializer->type == EXPR_INITIALIZER) {
1104 fake_member_assigns(sym, call_global_assign_hooks);
1105 } else if (sym->initializer) {
1106 symbol = symbol_expression(sym);
1107 assign = assign_expression(symbol, sym->initializer);
1108 __pass_to_client(assign, GLOBAL_ASSIGNMENT_HOOK);
1109 } else {
1110 fake_member_assigns_helper(symbol_expression(sym), NULL, call_global_assign_hooks);
1112 } else {
1113 symbol = symbol_expression(sym);
1114 if (sym->initializer)
1115 assign = assign_expression(symbol, sym->initializer);
1116 else
1117 assign = assign_expression(symbol, zero_expr());
1118 __pass_to_client(assign, GLOBAL_ASSIGNMENT_HOOK);
1122 static void start_function_definition(struct symbol *sym)
1124 __in_function_def = 1;
1125 __pass_to_client(sym, FUNC_DEF_HOOK);
1126 __in_function_def = 0;
1127 __pass_to_client(sym, AFTER_DEF_HOOK);
1131 static void split_function(struct symbol *sym)
1133 struct symbol *base_type = get_base_type(sym);
1135 gettimeofday(&fn_start_time, NULL);
1136 cur_func_sym = sym;
1137 if (sym->ident)
1138 cur_func = sym->ident->name;
1139 __smatch_lineno = sym->pos.line;
1140 last_stmt = NULL;
1141 loop_count = 0;
1142 sm_debug("new function: %s\n", cur_func);
1143 __stree_id = 0;
1144 if (option_two_passes) {
1145 __unnullify_path();
1146 loop_num = 0;
1147 final_pass = 0;
1148 start_function_definition(sym);
1149 __split_stmt(base_type->stmt);
1150 __split_stmt(base_type->inline_stmt);
1151 nullify_path();
1153 __unnullify_path();
1154 loop_num = 0;
1155 final_pass = 1;
1156 start_function_definition(sym);
1157 __split_stmt(base_type->stmt);
1158 __split_stmt(base_type->inline_stmt);
1159 __pass_to_client(sym, END_FUNC_HOOK);
1160 __pass_to_client(sym, AFTER_FUNC_HOOK);
1162 cur_func_sym = NULL;
1163 cur_func = NULL;
1164 clear_all_states();
1165 free_data_info_allocs();
1166 free_expression_stack(&switch_expr_stack);
1167 __free_ptr_list((struct ptr_list **)&big_statement_stack);
1168 __bail_on_rest_of_function = 0;
1171 static void parse_inline(struct expression *call)
1173 struct symbol *base_type;
1174 int loop_num_bak = loop_num;
1175 int final_pass_bak = final_pass;
1176 char *cur_func_bak = cur_func;
1177 struct statement_list *big_statement_stack_bak = big_statement_stack;
1178 struct expression_list *big_expression_stack_bak = big_expression_stack;
1179 struct expression_list *switch_expr_stack_bak = switch_expr_stack;
1180 struct symbol *cur_func_sym_bak = cur_func_sym;
1182 __pass_to_client(call, INLINE_FN_START);
1183 final_pass = 0; /* don't print anything */
1184 __inline_fn = call;
1186 base_type = get_base_type(call->fn->symbol);
1187 cur_func_sym = call->fn->symbol;
1188 if (call->fn->symbol->ident)
1189 cur_func = call->fn->symbol->ident->name;
1190 else
1191 cur_func = NULL;
1192 set_position(call->fn->symbol->pos);
1194 save_all_states();
1195 big_statement_stack = NULL;
1196 big_expression_stack = NULL;
1197 switch_expr_stack = NULL;
1199 sm_debug("inline function: %s\n", cur_func);
1200 __unnullify_path();
1201 loop_num = 0;
1202 start_function_definition(call->fn->symbol);
1203 __split_stmt(base_type->stmt);
1204 __split_stmt(base_type->inline_stmt);
1205 __pass_to_client(call->fn->symbol, END_FUNC_HOOK);
1206 __pass_to_client(call->fn->symbol, AFTER_FUNC_HOOK);
1208 free_expression_stack(&switch_expr_stack);
1209 __free_ptr_list((struct ptr_list **)&big_statement_stack);
1210 nullify_path();
1211 free_goto_stack();
1213 loop_num = loop_num_bak;
1214 final_pass = final_pass_bak;
1215 cur_func_sym = cur_func_sym_bak;
1216 cur_func = cur_func_bak;
1217 big_statement_stack = big_statement_stack_bak;
1218 big_expression_stack = big_expression_stack_bak;
1219 switch_expr_stack = switch_expr_stack_bak;
1221 restore_all_states();
1222 set_position(call->pos);
1223 __inline_fn = NULL;
1224 __pass_to_client(call, INLINE_FN_END);
1227 static struct symbol_list *inlines_called;
1228 static void add_inline_function(struct symbol *sym)
1230 static struct symbol_list *already_added;
1231 struct symbol *tmp;
1233 FOR_EACH_PTR(already_added, tmp) {
1234 if (tmp == sym)
1235 return;
1236 } END_FOR_EACH_PTR(tmp);
1238 add_ptr_list(&already_added, sym);
1239 add_ptr_list(&inlines_called, sym);
1242 static void process_inlines()
1244 struct symbol *tmp;
1246 FOR_EACH_PTR(inlines_called, tmp) {
1247 split_function(tmp);
1248 } END_FOR_EACH_PTR(tmp);
1249 free_ptr_list(&inlines_called);
1252 static struct symbol *get_last_scoped_symbol(struct symbol_list *big_list, int use_static)
1254 struct symbol *sym;
1256 FOR_EACH_PTR_REVERSE(big_list, sym) {
1257 if (!sym->scope)
1258 continue;
1259 if (use_static && sym->ctype.modifiers & MOD_STATIC)
1260 return sym;
1261 if (!use_static && !(sym->ctype.modifiers & MOD_STATIC))
1262 return sym;
1263 } END_FOR_EACH_PTR_REVERSE(sym);
1265 return NULL;
1268 static void split_inlines_in_scope(struct symbol *sym)
1270 struct symbol *base;
1271 struct symbol_list *scope_list;
1272 int stream;
1274 scope_list = sym->scope->symbols;
1275 stream = sym->pos.stream;
1277 /* find the last static symbol in the file */
1278 FOR_EACH_PTR_REVERSE(scope_list, sym) {
1279 if (sym->pos.stream != stream)
1280 continue;
1281 if (sym->type != SYM_NODE)
1282 continue;
1283 base = get_base_type(sym);
1284 if (!base)
1285 continue;
1286 if (base->type != SYM_FN)
1287 continue;
1288 if (!base->inline_stmt)
1289 continue;
1290 add_inline_function(sym);
1291 } END_FOR_EACH_PTR_REVERSE(sym);
1293 process_inlines();
1296 static void split_inlines(struct symbol_list *sym_list)
1298 struct symbol *sym;
1300 sym = get_last_scoped_symbol(sym_list, 0);
1301 if (sym)
1302 split_inlines_in_scope(sym);
1303 sym = get_last_scoped_symbol(sym_list, 1);
1304 if (sym)
1305 split_inlines_in_scope(sym);
1308 static void split_functions(struct symbol_list *sym_list)
1310 struct symbol *sym;
1312 FOR_EACH_PTR(sym_list, sym) {
1313 set_position(sym->pos);
1314 if (sym->type == SYM_NODE && get_base_type(sym)->type == SYM_FN) {
1315 split_function(sym);
1316 process_inlines();
1317 } else {
1318 __pass_to_client(sym, BASE_HOOK);
1319 fake_global_assign(sym);
1321 } END_FOR_EACH_PTR(sym);
1322 split_inlines(sym_list);
1323 __pass_to_client(sym_list, END_FILE_HOOK);
1326 void smatch(int argc, char **argv)
1329 struct string_list *filelist = NULL;
1330 struct symbol_list *sym_list;
1332 if (argc < 2) {
1333 printf("Usage: smatch [--debug] <filename.c>\n");
1334 exit(1);
1336 sparse_initialize(argc, argv, &filelist);
1337 FOR_EACH_PTR_NOTAG(filelist, base_file) {
1338 if (option_file_output) {
1339 char buf[256];
1341 snprintf(buf, sizeof(buf), "%s.smatch", base_file);
1342 sm_outfd = fopen(buf, "w");
1343 if (!sm_outfd) {
1344 printf("Error: Cannot open %s\n", base_file);
1345 exit(1);
1348 sym_list = sparse_keep_tokens(base_file);
1349 split_functions(sym_list);
1350 } END_FOR_EACH_PTR_NOTAG(base_file);