xref: /illumos-gate/usr/src/tools/smatch/src/lib.c (revision bb9475a199514dcace79d04d02c1eff05d65b94f)
1 /*
2  * 'sparse' library helper routines.
3  *
4  * Copyright (C) 2003 Transmeta Corp.
5  *               2003-2004 Linus Torvalds
6  *
7  * Permission is hereby granted, free of charge, to any person obtaining a copy
8  * of this software and associated documentation files (the "Software"), to deal
9  * in the Software without restriction, including without limitation the rights
10  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
11  * copies of the Software, and to permit persons to whom the Software is
12  * furnished to do so, subject to the following conditions:
13  *
14  * The above copyright notice and this permission notice shall be included in
15  * all copies or substantial portions of the Software.
16  *
17  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
18  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
19  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
20  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
21  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
22  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
23  * THE SOFTWARE.
24  */
25 #include <ctype.h>
26 #include <errno.h>
27 #include <fcntl.h>
28 #include <stdarg.h>
29 #include <stddef.h>
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <unistd.h>
34 #include <assert.h>
35 
36 #include <sys/types.h>
37 
38 #include "lib.h"
39 #include "allocate.h"
40 #include "token.h"
41 #include "parse.h"
42 #include "symbol.h"
43 #include "expression.h"
44 #include "evaluate.h"
45 #include "scope.h"
46 #include "linearize.h"
47 #include "target.h"
48 #include "machine.h"
49 #include "version.h"
50 #include "bits.h"
51 
52 int verbose, optimize_level, optimize_size, preprocessing;
53 int die_if_error = 0;
54 int parse_error;
55 int has_error = 0;
56 int do_output = 0;
57 
58 #ifndef __GNUC__
59 # define __GNUC__ 2
60 # define __GNUC_MINOR__ 95
61 # define __GNUC_PATCHLEVEL__ 0
62 #endif
63 
64 int gcc_major = __GNUC__;
65 int gcc_minor = __GNUC_MINOR__;
66 int gcc_patchlevel = __GNUC_PATCHLEVEL__;
67 
68 const char *base_filename;
69 
70 static const char *diag_prefix = "";
71 static const char *gcc_base_dir = GCC_BASE;
72 static const char *multiarch_dir = MULTIARCH_TRIPLET;
73 static const char *outfile = NULL;
74 
skip_to(struct token * token,int op)75 struct token *skip_to(struct token *token, int op)
76 {
77 	while (!match_op(token, op) && !eof_token(token))
78 		token = token->next;
79 	return token;
80 }
81 
82 static struct token bad_token = { .pos.type = TOKEN_BAD };
expect(struct token * token,int op,const char * where)83 struct token *expect(struct token *token, int op, const char *where)
84 {
85 	if (!match_op(token, op)) {
86 		if (token != &bad_token) {
87 			bad_token.next = token;
88 			sparse_error(token->pos, "Expected %s %s", show_special(op), where);
89 			sparse_error(token->pos, "got %s", show_token(token));
90 		}
91 		if (op == ';')
92 			return skip_to(token, op);
93 		return &bad_token;
94 	}
95 	return token->next;
96 }
97 
98 ///
99 // issue an error message on new parsing errors
100 // @token: the current token
101 // @errmsg: the error message
102 // If the current token is from a previous error, an error message
103 // has already been issued, so nothing more is done.
104 // Otherwise, @errmsg is displayed followed by the current token.
unexpected(struct token * token,const char * errmsg)105 void unexpected(struct token *token, const char *errmsg)
106 {
107 	if (token == &bad_token)
108 		return;
109 	sparse_error(token->pos, "%s", errmsg);
110 	sparse_error(token->pos, "got %s", show_token(token));
111 }
112 
hexval(unsigned int c)113 unsigned int hexval(unsigned int c)
114 {
115 	int retval = 256;
116 	switch (c) {
117 	case '0'...'9':
118 		retval = c - '0';
119 		break;
120 	case 'a'...'f':
121 		retval = c - 'a' + 10;
122 		break;
123 	case 'A'...'F':
124 		retval = c - 'A' + 10;
125 		break;
126 	}
127 	return retval;
128 }
129 
do_warn(const char * type,struct position pos,const char * fmt,va_list args)130 static void do_warn(const char *type, struct position pos, const char * fmt, va_list args)
131 {
132 	static char buffer[512];
133 	const char *name;
134 
135 	/* Shut up warnings if position is bad_token.pos */
136 	if (pos.type == TOKEN_BAD)
137 		return;
138 
139 	vsprintf(buffer, fmt, args);
140 	name = stream_name(pos.stream);
141 
142 	fflush(stdout);
143 	fprintf(stderr, "%s: %s:%d:%d: %s%s\n",
144 		diag_prefix, name, pos.line, pos.pos, type, buffer);
145 }
146 
147 unsigned int fmax_warnings = 100;
148 static int show_info = 1;
149 
info(struct position pos,const char * fmt,...)150 void info(struct position pos, const char * fmt, ...)
151 {
152 	va_list args;
153 
154 	if (!show_info)
155 		return;
156 	va_start(args, fmt);
157 	do_warn("", pos, fmt, args);
158 	va_end(args);
159 }
160 
do_error(struct position pos,const char * fmt,va_list args)161 static void do_error(struct position pos, const char * fmt, va_list args)
162 {
163 	static int errors = 0;
164 
165 	parse_error = 1;
166         die_if_error = 1;
167 	show_info = 1;
168 	/* Shut up warnings if position is bad_token.pos */
169 	if (pos.type == TOKEN_BAD)
170 		return;
171 	/* Shut up warnings after an error */
172 	has_error |= ERROR_CURR_PHASE;
173 	if (errors > 100) {
174 		static int once = 0;
175 		show_info = 0;
176 		if (once)
177 			return;
178 		fmt = "too many errors";
179 		once = 1;
180 	}
181 
182 	do_warn("error: ", pos, fmt, args);
183 	errors++;
184 }
185 
warning(struct position pos,const char * fmt,...)186 void warning(struct position pos, const char * fmt, ...)
187 {
188 	va_list args;
189 
190 	if (Wsparse_error) {
191 		va_start(args, fmt);
192 		do_error(pos, fmt, args);
193 		va_end(args);
194 		return;
195 	}
196 
197 	if (!fmax_warnings || has_error) {
198 		show_info = 0;
199 		return;
200 	}
201 
202 	if (!--fmax_warnings) {
203 		show_info = 0;
204 		fmt = "too many warnings";
205 	}
206 
207 	va_start(args, fmt);
208 	do_warn("warning: ", pos, fmt, args);
209 	va_end(args);
210 }
211 
sparse_error(struct position pos,const char * fmt,...)212 void sparse_error(struct position pos, const char * fmt, ...)
213 {
214 	va_list args;
215 	va_start(args, fmt);
216 	do_error(pos, fmt, args);
217 	va_end(args);
218 }
219 
expression_error(struct expression * expr,const char * fmt,...)220 void expression_error(struct expression *expr, const char *fmt, ...)
221 {
222 	va_list args;
223 	va_start(args, fmt);
224 	do_error(expr->pos, fmt, args);
225 	va_end(args);
226 	expr->ctype = &bad_ctype;
227 }
228 
229 NORETURN_ATTR
error_die(struct position pos,const char * fmt,...)230 void error_die(struct position pos, const char * fmt, ...)
231 {
232 	va_list args;
233 	va_start(args, fmt);
234 	do_warn("error: ", pos, fmt, args);
235 	va_end(args);
236 	exit(1);
237 }
238 
239 NORETURN_ATTR
die(const char * fmt,...)240 void die(const char *fmt, ...)
241 {
242 	va_list args;
243 	static char buffer[512];
244 
245 	va_start(args, fmt);
246 	vsnprintf(buffer, sizeof(buffer), fmt, args);
247 	va_end(args);
248 
249 	fprintf(stderr, "%s: %s\n", diag_prefix, buffer);
250 	exit(1);
251 }
252 
253 static struct token *pre_buffer_begin = NULL;
254 static struct token *pre_buffer_end = NULL;
255 
256 int Waddress = 0;
257 int Waddress_space = 1;
258 int Wbitwise = 1;
259 int Wbitwise_pointer = 0;
260 int Wcast_from_as = 0;
261 int Wcast_to_as = 0;
262 int Wcast_truncate = 1;
263 int Wconstant_suffix = 0;
264 int Wconstexpr_not_const = 0;
265 int Wcontext = 1;
266 int Wdecl = 1;
267 int Wdeclarationafterstatement = -1;
268 int Wdefault_bitfield_sign = 0;
269 int Wdesignated_init = 1;
270 int Wdo_while = 0;
271 int Wimplicit_int = 1;
272 int Winit_cstring = 0;
273 int Wint_to_pointer_cast = 1;
274 int Wenum_mismatch = 1;
275 int Wexternal_function_has_definition = 1;
276 int Wsparse_error = 0;
277 int Wmemcpy_max_count = 1;
278 int Wnon_pointer_null = 1;
279 int Wold_initializer = 1;
280 int Wold_style_definition = 1;
281 int Wone_bit_signed_bitfield = 1;
282 int Woverride_init = 1;
283 int Woverride_init_all = 0;
284 int Woverride_init_whole_range = 0;
285 int Wparen_string = 0;
286 int Wpointer_arith = 0;
287 int Wpointer_to_int_cast = 1;
288 int Wptr_subtraction_blows = 0;
289 int Wreturn_void = 0;
290 int Wshadow = 0;
291 int Wshift_count_negative = 1;
292 int Wshift_count_overflow = 1;
293 int Wsizeof_bool = 0;
294 int Wstrict_prototypes = 1;
295 int Wtautological_compare = 0;
296 int Wtransparent_union = 0;
297 int Wtypesign = 0;
298 int Wundef = 0;
299 int Wuninitialized = 1;
300 int Wunknown_attribute = 0;
301 int Wvla = 1;
302 
303 int dump_macro_defs = 0;
304 int dump_macros_only = 0;
305 
306 int dbg_compound = 0;
307 int dbg_dead = 0;
308 int dbg_domtree = 0;
309 int dbg_entry = 0;
310 int dbg_ir = 0;
311 int dbg_postorder = 0;
312 
313 unsigned long fdump_ir;
314 int fmem_report = 0;
315 unsigned long long fmemcpy_max_count = 100000;
316 unsigned long fpasses = ~0UL;
317 int funsigned_char = UNSIGNED_CHAR;
318 
319 int preprocess_only;
320 
321 enum standard standard = STANDARD_GNU89;
322 
323 int arch_m64 = ARCH_M64_DEFAULT;
324 int arch_msize_long = 0;
325 int arch_big_endian = ARCH_BIG_ENDIAN;
326 int arch_mach = MACH_NATIVE;
327 
328 
329 #define CMDLINE_INCLUDE 20
330 static int cmdline_include_nr = 0;
331 static char *cmdline_include[CMDLINE_INCLUDE];
332 
333 
add_pre_buffer(const char * fmt,...)334 void add_pre_buffer(const char *fmt, ...)
335 {
336 	va_list args;
337 	unsigned int size;
338 	struct token *begin, *end;
339 	char buffer[4096];
340 
341 	va_start(args, fmt);
342 	size = vsnprintf(buffer, sizeof(buffer), fmt, args);
343 	va_end(args);
344 	begin = tokenize_buffer(buffer, size, &end);
345 	if (!pre_buffer_begin)
346 		pre_buffer_begin = begin;
347 	if (pre_buffer_end)
348 		pre_buffer_end->next = begin;
349 	pre_buffer_end = end;
350 }
351 
handle_switch_D(char * arg,char ** next)352 static char **handle_switch_D(char *arg, char **next)
353 {
354 	const char *name = arg + 1;
355 	const char *value = "1";
356 
357 	if (!*name) {
358 		arg = *++next;
359 		if (!arg)
360 			die("argument to `-D' is missing");
361 		name = arg;
362 	}
363 
364 	for (;;arg++) {
365 		char c;
366 		c = *arg;
367 		if (!c)
368 			break;
369 		if (c == '=') {
370 			*arg = '\0';
371 			value = arg + 1;
372 			break;
373 		}
374 	}
375 	add_pre_buffer("#define %s %s\n", name, value);
376 	return next;
377 }
378 
handle_switch_E(char * arg,char ** next)379 static char **handle_switch_E(char *arg, char **next)
380 {
381 	if (arg[1] == '\0')
382 		preprocess_only = 1;
383 	return next;
384 }
385 
handle_switch_I(char * arg,char ** next)386 static char **handle_switch_I(char *arg, char **next)
387 {
388 	char *path = arg+1;
389 
390 	switch (arg[1]) {
391 	case '-':
392 		add_pre_buffer("#split_include\n");
393 		break;
394 
395 	case '\0':	/* Plain "-I" */
396 		path = *++next;
397 		if (!path)
398 			die("missing argument for -I option");
399 		/* Fall through */
400 	default:
401 		add_pre_buffer("#add_include \"%s/\"\n", path);
402 	}
403 	return next;
404 }
405 
add_cmdline_include(char * filename)406 static void add_cmdline_include(char *filename)
407 {
408 	if (cmdline_include_nr >= CMDLINE_INCLUDE)
409 		die("too many include files for %s\n", filename);
410 	cmdline_include[cmdline_include_nr++] = filename;
411 }
412 
handle_switch_i(char * arg,char ** next)413 static char **handle_switch_i(char *arg, char **next)
414 {
415 	if (*next && !strcmp(arg, "include"))
416 		add_cmdline_include(*++next);
417 	else if (*next && !strcmp(arg, "imacros"))
418 		add_cmdline_include(*++next);
419 	else if (*next && !strcmp(arg, "isystem")) {
420 		char *path = *++next;
421 		if (!path)
422 			die("missing argument for -isystem option");
423 		add_pre_buffer("#add_isystem \"%s/\"\n", path);
424 	} else if (*next && !strcmp(arg, "idirafter")) {
425 		char *path = *++next;
426 		if (!path)
427 			die("missing argument for -idirafter option");
428 		add_pre_buffer("#add_dirafter \"%s/\"\n", path);
429 	}
430 	return next;
431 }
432 
handle_switch_M(char * arg,char ** next)433 static char **handle_switch_M(char *arg, char **next)
434 {
435 	if (!strcmp(arg, "MF") || !strcmp(arg,"MQ") || !strcmp(arg,"MT")) {
436 		if (!*next)
437 			die("missing argument for -%s option", arg);
438 		return next + 1;
439 	}
440 	return next;
441 }
442 
handle_multiarch_dir(char * arg,char ** next)443 static char **handle_multiarch_dir(char *arg, char **next)
444 {
445 	multiarch_dir = *++next;
446 	if (!multiarch_dir)
447 		die("missing argument for -multiarch-dir option");
448 	return next;
449 }
450 
handle_switch_m(char * arg,char ** next)451 static char **handle_switch_m(char *arg, char **next)
452 {
453 	if (!strcmp(arg, "m64")) {
454 		arch_m64 = ARCH_LP64;
455 	} else if (!strcmp(arg, "m32") || !strcmp(arg, "m16")) {
456 		arch_m64 = ARCH_LP32;
457 	} else if (!strcmp(arg, "mx32")) {
458 		arch_m64 = ARCH_X32;
459 	} else if (!strcmp(arg, "msize-llp64")) {
460 		arch_m64 = ARCH_LLP64;
461 	} else if (!strcmp(arg, "msize-long")) {
462 		arch_msize_long = 1;
463 	} else if (!strcmp(arg, "multiarch-dir")) {
464 		return handle_multiarch_dir(arg, next);
465 	} else if (!strcmp(arg, "mbig-endian")) {
466 		arch_big_endian = 1;
467 	} else if (!strcmp(arg, "mlittle-endian")) {
468 		arch_big_endian = 0;
469 	}
470 	return next;
471 }
472 
handle_arch_msize_long_finalize(void)473 static void handle_arch_msize_long_finalize(void)
474 {
475 	if (arch_msize_long) {
476 		size_t_ctype = &ulong_ctype;
477 		ssize_t_ctype = &long_ctype;
478 	}
479 }
480 
handle_arch_finalize(void)481 static void handle_arch_finalize(void)
482 {
483 	handle_arch_msize_long_finalize();
484 }
485 
match_option(const char * arg,const char * prefix)486 static const char *match_option(const char *arg, const char *prefix)
487 {
488 	unsigned int n = strlen(prefix);
489 	if (strncmp(arg, prefix, n) == 0)
490 		return arg + n;
491 	return NULL;
492 }
493 
494 
495 struct mask_map {
496 	const char *name;
497 	unsigned long mask;
498 };
499 
apply_mask(unsigned long * val,const char * str,unsigned len,const struct mask_map * map,int neg)500 static int apply_mask(unsigned long *val, const char *str, unsigned len, const struct mask_map *map, int neg)
501 {
502 	const char *name;
503 
504 	for (;(name = map->name); map++) {
505 		if (!strncmp(name, str, len) && !name[len]) {
506 			if (neg == 0)
507 				*val |= map->mask;
508 			else
509 				*val &= ~map->mask;
510 			return 0;
511 		}
512 	}
513 	return 1;
514 }
515 
handle_suboption_mask(const char * arg,const char * opt,const struct mask_map * map,unsigned long * flag)516 static int handle_suboption_mask(const char *arg, const char *opt, const struct mask_map *map, unsigned long *flag)
517 {
518 	if (*opt == '\0') {
519 		apply_mask(flag, "", 0, map, 0);
520 		return 1;
521 	}
522 	if (*opt++ != '=')
523 		return 0;
524 	while (1) {
525 		unsigned int len = strcspn(opt, ",+");
526 		int neg = 0;
527 		if (len == 0)
528 			goto end;
529 		if (!strncmp(opt, "no-", 3)) {
530 			opt += 3;
531 			len -= 3;
532 			neg = 1;
533 		}
534 		if (apply_mask(flag, opt, len, map, neg))
535 			die("error: wrong option '%.*s' for \'%s\'", len, opt, arg);
536 
537 end:
538 		opt += len;
539 		if (*opt++ == '\0')
540 			break;
541 	}
542 	return 1;
543 }
544 
545 
546 #define OPT_INVERSE	1
547 struct flag {
548 	const char *name;
549 	int *flag;
550 	int (*fun)(const char *arg, const char *opt, const struct flag *, int options);
551 	unsigned long mask;
552 };
553 
handle_switches(const char * ori,const char * opt,const struct flag * flags)554 static int handle_switches(const char *ori, const char *opt, const struct flag *flags)
555 {
556 	const char *arg = opt;
557 	int val = 1;
558 
559 	// Prefixe "no-" mean to turn flag off.
560 	if (strncmp(arg, "no-", 3) == 0) {
561 		arg += 3;
562 		val = 0;
563 	}
564 
565 	for (; flags->name; flags++) {
566 		const char *opt = match_option(arg, flags->name);
567 		int rc;
568 
569 		if (!opt)
570 			continue;
571 
572 		if (flags->fun) {
573 			int options = 0;
574 			if (!val)
575 				options |= OPT_INVERSE;
576 			if ((rc = flags->fun(ori, opt, flags, options)))
577 				return rc;
578 		}
579 
580 		// boolean flag
581 		if (opt[0] == '\0' && flags->flag) {
582 			if (flags->mask & OPT_INVERSE)
583 				val = !val;
584 			*flags->flag = val;
585 			return 1;
586 		}
587 	}
588 
589 	// not handled
590 	return 0;
591 }
592 
593 
594 #define	OPTNUM_ZERO_IS_INF		1
595 #define	OPTNUM_UNLIMITED		2
596 
597 #define OPT_NUMERIC(NAME, TYPE, FUNCTION)	\
598 static int opt_##NAME(const char *arg, const char *opt, TYPE *ptr, int flag)	\
599 {									\
600 	char *end;							\
601 	TYPE val;							\
602 									\
603 	val = FUNCTION(opt, &end, 0);					\
604 	if (*end != '\0' || end == opt) {				\
605 		if ((flag & OPTNUM_UNLIMITED) && !strcmp(opt, "unlimited"))	\
606 			val = ~val;					\
607 		else							\
608 			die("error: wrong argument to \'%s\'", arg);	\
609 	}								\
610 	if ((flag & OPTNUM_ZERO_IS_INF) && val == 0)			\
611 		val = ~val;						\
612 	*ptr = val;							\
613 	return 1;							\
614 }
615 
OPT_NUMERIC(ullong,unsigned long long,strtoull)616 OPT_NUMERIC(ullong, unsigned long long, strtoull)
617 OPT_NUMERIC(uint, unsigned int, strtoul)
618 
619 
620 static char **handle_switch_o(char *arg, char **next)
621 {
622 	if (!strcmp (arg, "o")) {       // "-o foo"
623 		if (!*++next)
624 			die("argument to '-o' is missing");
625 		outfile = *next;
626 	}
627 	// else "-ofoo"
628 
629 	return next;
630 }
631 
632 static const struct flag warnings[] = {
633 	{ "address", &Waddress },
634 	{ "address-space", &Waddress_space },
635 	{ "bitwise", &Wbitwise },
636 	{ "bitwise-pointer", &Wbitwise_pointer},
637 	{ "cast-from-as", &Wcast_from_as },
638 	{ "cast-to-as", &Wcast_to_as },
639 	{ "cast-truncate", &Wcast_truncate },
640 	{ "constant-suffix", &Wconstant_suffix },
641 	{ "constexpr-not-const", &Wconstexpr_not_const},
642 	{ "context", &Wcontext },
643 	{ "decl", &Wdecl },
644 	{ "declaration-after-statement", &Wdeclarationafterstatement },
645 	{ "default-bitfield-sign", &Wdefault_bitfield_sign },
646 	{ "designated-init", &Wdesignated_init },
647 	{ "do-while", &Wdo_while },
648 	{ "enum-mismatch", &Wenum_mismatch },
649 	{ "external-function-has-definition", &Wexternal_function_has_definition },
650 	{ "implicit-int", &Wimplicit_int },
651 	{ "init-cstring", &Winit_cstring },
652 	{ "int-to-pointer-cast", &Wint_to_pointer_cast },
653 	{ "memcpy-max-count", &Wmemcpy_max_count },
654 	{ "non-pointer-null", &Wnon_pointer_null },
655 	{ "old-initializer", &Wold_initializer },
656 	{ "old-style-definition", &Wold_style_definition },
657 	{ "one-bit-signed-bitfield", &Wone_bit_signed_bitfield },
658 	{ "override-init", &Woverride_init },
659 	{ "override-init-all", &Woverride_init_all },
660 	{ "paren-string", &Wparen_string },
661 	{ "pointer-to-int-cast", &Wpointer_to_int_cast },
662 	{ "ptr-subtraction-blows", &Wptr_subtraction_blows },
663 	{ "return-void", &Wreturn_void },
664 	{ "shadow", &Wshadow },
665 	{ "shift-count-negative", &Wshift_count_negative },
666 	{ "shift-count-overflow", &Wshift_count_overflow },
667 	{ "sizeof-bool", &Wsizeof_bool },
668 	{ "strict-prototypes", &Wstrict_prototypes },
669 	{ "pointer-arith", &Wpointer_arith },
670 	{ "sparse-error", &Wsparse_error },
671 	{ "tautological-compare", &Wtautological_compare },
672 	{ "transparent-union", &Wtransparent_union },
673 	{ "typesign", &Wtypesign },
674 	{ "undef", &Wundef },
675 	{ "uninitialized", &Wuninitialized },
676 	{ "unknown-attribute", &Wunknown_attribute },
677 	{ "vla", &Wvla },
678 };
679 
680 enum {
681 	WARNING_OFF,
682 	WARNING_ON,
683 	WARNING_FORCE_OFF
684 };
685 
686 
handle_onoff_switch(char * arg,char ** next,const struct flag warnings[],int n)687 static char **handle_onoff_switch(char *arg, char **next, const struct flag warnings[], int n)
688 {
689 	int flag = WARNING_ON;
690 	char *p = arg + 1;
691 	unsigned i;
692 
693 	if (!strcmp(p, "sparse-all")) {
694 		for (i = 0; i < n; i++) {
695 			if (*warnings[i].flag != WARNING_FORCE_OFF && warnings[i].flag != &Wsparse_error)
696 				*warnings[i].flag = WARNING_ON;
697 		}
698 		return NULL;
699 	}
700 
701 	// Prefixes "no" and "no-" mean to turn warning off.
702 	if (p[0] == 'n' && p[1] == 'o') {
703 		p += 2;
704 		if (p[0] == '-')
705 			p++;
706 		flag = WARNING_FORCE_OFF;
707 	}
708 
709 	for (i = 0; i < n; i++) {
710 		if (!strcmp(p,warnings[i].name)) {
711 			*warnings[i].flag = flag;
712 			return next;
713 		}
714 	}
715 
716 	// Unknown.
717 	return NULL;
718 }
719 
handle_switch_W(char * arg,char ** next)720 static char **handle_switch_W(char *arg, char **next)
721 {
722 	char ** ret = handle_onoff_switch(arg, next, warnings, ARRAY_SIZE(warnings));
723 	if (ret)
724 		return ret;
725 
726 	// Unknown.
727 	return next;
728 }
729 
730 static struct flag debugs[] = {
731 	{ "compound", &dbg_compound},
732 	{ "dead", &dbg_dead},
733 	{ "domtree", &dbg_domtree},
734 	{ "entry", &dbg_entry},
735 	{ "ir", &dbg_ir},
736 	{ "postorder", &dbg_postorder},
737 };
738 
739 
handle_switch_v(char * arg,char ** next)740 static char **handle_switch_v(char *arg, char **next)
741 {
742 	char ** ret = handle_onoff_switch(arg, next, debugs, ARRAY_SIZE(debugs));
743 	if (ret)
744 		return ret;
745 
746 	// Unknown.
747 	do {
748 		verbose++;
749 	} while (*++arg == 'v');
750 	return next;
751 }
752 
handle_switch_d(char * arg,char ** next)753 static char **handle_switch_d(char *arg, char **next)
754 {
755 	char *arg_char = arg + 1;
756 
757 	/*
758 	 * -d<CHARS>, where <CHARS> is a sequence of characters, not preceded
759 	 * by a space. If you specify characters whose behaviour conflicts,
760 	 * the result is undefined.
761 	 */
762 	while (*arg_char) {
763 		switch (*arg_char) {
764 		case 'M': /* dump just the macro definitions */
765 			dump_macros_only = 1;
766 			dump_macro_defs = 0;
767 			break;
768 		case 'D': /* like 'M', but also output pre-processed text */
769 			dump_macro_defs = 1;
770 			dump_macros_only = 0;
771 			break;
772 		case 'N': /* like 'D', but only output macro names not bodies */
773 			break;
774 		case 'I': /* like 'D', but also output #include directives */
775 			break;
776 		case 'U': /* like 'D', but only output expanded macros */
777 			break;
778 		}
779 		arg_char++;
780 	}
781 	return next;
782 }
783 
784 
handle_onoff_switch_finalize(const struct flag warnings[],int n)785 static void handle_onoff_switch_finalize(const struct flag warnings[], int n)
786 {
787 	unsigned i;
788 
789 	for (i = 0; i < n; i++) {
790 		if (*warnings[i].flag == WARNING_FORCE_OFF)
791 			*warnings[i].flag = WARNING_OFF;
792 	}
793 }
794 
handle_switch_W_finalize(void)795 static void handle_switch_W_finalize(void)
796 {
797 	handle_onoff_switch_finalize(warnings, ARRAY_SIZE(warnings));
798 
799 	/* default Wdeclarationafterstatement based on the C dialect */
800 	if (-1 == Wdeclarationafterstatement)
801 	{
802 		switch (standard)
803 		{
804 			case STANDARD_C89:
805 			case STANDARD_C94:
806 				Wdeclarationafterstatement = 1;
807 				break;
808 
809 			case STANDARD_C99:
810 			case STANDARD_GNU89:
811 			case STANDARD_GNU99:
812 			case STANDARD_C11:
813 			case STANDARD_GNU11:
814 				Wdeclarationafterstatement = 0;
815 				break;
816 
817 			default:
818 				assert (0);
819 		}
820 
821 	}
822 }
823 
handle_switch_v_finalize(void)824 static void handle_switch_v_finalize(void)
825 {
826 	handle_onoff_switch_finalize(debugs, ARRAY_SIZE(debugs));
827 }
828 
handle_switch_U(char * arg,char ** next)829 static char **handle_switch_U(char *arg, char **next)
830 {
831 	const char *name = arg + 1;
832 	if (*name == '\0')
833 		name = *++next;
834 	add_pre_buffer ("#undef %s\n", name);
835 	return next;
836 }
837 
handle_switch_O(char * arg,char ** next)838 static char **handle_switch_O(char *arg, char **next)
839 {
840 	int level = 1;
841 	if (arg[1] >= '0' && arg[1] <= '9')
842 		level = arg[1] - '0';
843 	optimize_level = level;
844 	optimize_size = arg[1] == 's';
845 	return next;
846 }
847 
handle_ftabstop(const char * arg,const char * opt,const struct flag * flag,int options)848 static int handle_ftabstop(const char *arg, const char *opt, const struct flag *flag, int options)
849 {
850 	unsigned long val;
851 	char *end;
852 
853 	if (*opt == '\0')
854 		die("error: missing argument to \"%s\"", arg);
855 
856 	/* we silently ignore silly values */
857 	val = strtoul(opt, &end, 10);
858 	if (*end == '\0' && 1 <= val && val <= 100)
859 		tabstop = val;
860 
861 	return 1;
862 }
863 
handle_fpasses(const char * arg,const char * opt,const struct flag * flag,int options)864 static int handle_fpasses(const char *arg, const char *opt, const struct flag *flag, int options)
865 {
866 	unsigned long mask;
867 
868 	mask = flag->mask;
869 	if (*opt == '\0') {
870 		if (options & OPT_INVERSE)
871 			fpasses &= ~mask;
872 		else
873 			fpasses |=  mask;
874 		return 1;
875 	}
876 	if (options & OPT_INVERSE)
877 		return 0;
878 	if (!strcmp(opt, "-enable")) {
879 		fpasses |= mask;
880 		return 1;
881 	}
882 	if (!strcmp(opt, "-disable")) {
883 		fpasses &= ~mask;
884 		return 1;
885 	}
886 	if (!strcmp(opt, "=last")) {
887 		// clear everything above
888 		mask |= mask - 1;
889 		fpasses &= mask;
890 		return 1;
891 	}
892 	return 0;
893 }
894 
handle_fdiagnostic_prefix(const char * arg,const char * opt,const struct flag * flag,int options)895 static int handle_fdiagnostic_prefix(const char *arg, const char *opt, const struct flag *flag, int options)
896 {
897 	switch (*opt) {
898 	case '\0':
899 		diag_prefix = "sparse";
900 		return 1;
901 	case '=':
902 		diag_prefix = xasprintf("%s", opt+1);
903 		return 1;
904 	default:
905 		return 0;
906 	}
907 }
908 
handle_fdump_ir(const char * arg,const char * opt,const struct flag * flag,int options)909 static int handle_fdump_ir(const char *arg, const char *opt, const struct flag *flag, int options)
910 {
911 	static const struct mask_map dump_ir_options[] = {
912 		{ "",			PASS_LINEARIZE },
913 		{ "linearize",		PASS_LINEARIZE },
914 		{ "mem2reg",		PASS_MEM2REG },
915 		{ "final",		PASS_FINAL },
916 		{ },
917 	};
918 
919 	return handle_suboption_mask(arg, opt, dump_ir_options, &fdump_ir);
920 }
921 
handle_fmemcpy_max_count(const char * arg,const char * opt,const struct flag * flag,int options)922 static int handle_fmemcpy_max_count(const char *arg, const char *opt, const struct flag *flag, int options)
923 {
924 	opt_ullong(arg, opt, &fmemcpy_max_count, OPTNUM_ZERO_IS_INF|OPTNUM_UNLIMITED);
925 	return 1;
926 }
927 
handle_fmax_warnings(const char * arg,const char * opt,const struct flag * flag,int options)928 static int handle_fmax_warnings(const char *arg, const char *opt, const struct flag *flag, int options)
929 {
930 	opt_uint(arg, opt, &fmax_warnings, OPTNUM_UNLIMITED);
931 	return 1;
932 }
933 
934 static struct flag fflags[] = {
935 	{ "diagnostic-prefix",	NULL,	handle_fdiagnostic_prefix },
936 	{ "dump-ir",		NULL,	handle_fdump_ir },
937 	{ "linearize",		NULL,	handle_fpasses,	PASS_LINEARIZE },
938 	{ "max-warnings=",	NULL,	handle_fmax_warnings },
939 	{ "mem-report",		&fmem_report },
940 	{ "memcpy-max-count=",	NULL,	handle_fmemcpy_max_count },
941 	{ "tabstop=",		NULL,	handle_ftabstop },
942 	{ "mem2reg",		NULL,	handle_fpasses,	PASS_MEM2REG },
943 	{ "optim",		NULL,	handle_fpasses,	PASS_OPTIM },
944 	{ "signed-char",	&funsigned_char, NULL,	OPT_INVERSE },
945 	{ "unsigned-char",	&funsigned_char, NULL, },
946 	{ },
947 };
948 
handle_switch_f(char * arg,char ** next)949 static char **handle_switch_f(char *arg, char **next)
950 {
951 	if (handle_switches(arg-1, arg+1, fflags))
952 		return next;
953 
954 	return next;
955 }
956 
handle_switch_G(char * arg,char ** next)957 static char **handle_switch_G(char *arg, char **next)
958 {
959 	if (!strcmp (arg, "G") && *next)
960 		return next + 1; // "-G 0"
961 	else
962 		return next;     // "-G0" or (bogus) terminal "-G"
963 }
964 
handle_switch_a(char * arg,char ** next)965 static char **handle_switch_a(char *arg, char **next)
966 {
967 	if (!strcmp (arg, "ansi"))
968 		standard = STANDARD_C89;
969 
970 	return next;
971 }
972 
handle_switch_s(const char * arg,char ** next)973 static char **handle_switch_s(const char *arg, char **next)
974 {
975 	if ((arg = match_option(arg, "std="))) {
976 		if (!strcmp (arg, "c89") ||
977 		    !strcmp (arg, "iso9899:1990"))
978 			standard = STANDARD_C89;
979 
980 		else if (!strcmp (arg, "iso9899:199409"))
981 			standard = STANDARD_C94;
982 
983 		else if (!strcmp (arg, "c99") ||
984 			 !strcmp (arg, "c9x") ||
985 			 !strcmp (arg, "iso9899:1999") ||
986 			 !strcmp (arg, "iso9899:199x"))
987 			standard = STANDARD_C99;
988 
989 		else if (!strcmp (arg, "gnu89"))
990 			standard = STANDARD_GNU89;
991 
992 		else if (!strcmp (arg, "gnu99") || !strcmp (arg, "gnu9x"))
993 			standard = STANDARD_GNU99;
994 
995 		else if (!strcmp(arg, "c11") ||
996 			 !strcmp(arg, "c1x") ||
997 			 !strcmp(arg, "iso9899:2011"))
998 			standard = STANDARD_C11;
999 
1000 		/*
1001 		 * For the interim, allow GNU17 to be treated as GNU11 as C17 is
1002 		 * mostly just a clean up of C11 and is not supposed to add any
1003 		 * new features.
1004 		 */
1005 		else if (!strcmp(arg, "gnu11") ||
1006 		         !strcmp(arg, "gnu17"))
1007 			standard = STANDARD_GNU11;
1008 
1009 		else
1010 			die ("Unsupported C dialect");
1011 	}
1012 
1013 	return next;
1014 }
1015 
handle_nostdinc(char * arg,char ** next)1016 static char **handle_nostdinc(char *arg, char **next)
1017 {
1018 	add_pre_buffer("#nostdinc\n");
1019 	return next;
1020 }
1021 
handle_switch_n(char * arg,char ** next)1022 static char **handle_switch_n(char *arg, char **next)
1023 {
1024 	if (!strcmp (arg, "nostdinc"))
1025 		return handle_nostdinc(arg, next);
1026 
1027 	return next;
1028 }
1029 
handle_base_dir(char * arg,char ** next)1030 static char **handle_base_dir(char *arg, char **next)
1031 {
1032 	gcc_base_dir = *++next;
1033 	if (!gcc_base_dir)
1034 		die("missing argument for -gcc-base-dir option");
1035 	return next;
1036 }
1037 
handle_no_lineno(char * arg,char ** next)1038 static char **handle_no_lineno(char *arg, char **next)
1039 {
1040 	no_lineno = 1;
1041 	return next;
1042 }
1043 
handle_switch_g(char * arg,char ** next)1044 static char **handle_switch_g(char *arg, char **next)
1045 {
1046 	if (!strcmp (arg, "gcc-base-dir"))
1047 		return handle_base_dir(arg, next);
1048 
1049 	return next;
1050 }
1051 
handle_switch_x(char * arg,char ** next)1052 static char **handle_switch_x(char *arg, char **next)
1053 {
1054 	if (!*++next)
1055 		die("missing argument for -x option");
1056 	return next;
1057 }
1058 
handle_version(char * arg,char ** next)1059 static char **handle_version(char *arg, char **next)
1060 {
1061 	printf("%s\n", SPARSE_VERSION);
1062 	exit(0);
1063 }
1064 
handle_param(char * arg,char ** next)1065 static char **handle_param(char *arg, char **next)
1066 {
1067 	char *value = NULL;
1068 
1069 	/* Ignore smatch's --param-mapper */
1070 	if (strcmp(arg, "-mapper") == 0)
1071 		return next;
1072 
1073 	/* For now just skip any '--param=*' or '--param *' */
1074 	if (*arg == '\0') {
1075 		value = *++next;
1076 	} else if (isspace((unsigned char)*arg) || *arg == '=') {
1077 		value = ++arg;
1078 	}
1079 
1080 	if (!value)
1081 		die("missing argument for --param option");
1082 
1083 	return next;
1084 }
1085 
1086 struct switches {
1087 	const char *name;
1088 	char **(*fn)(char *, char **);
1089 	unsigned int prefix:1;
1090 };
1091 
handle_long_options(char * arg,char ** next)1092 static char **handle_long_options(char *arg, char **next)
1093 {
1094 	static struct switches cmd[] = {
1095 		{ "param", handle_param, 1 },
1096 		{ "version", handle_version },
1097 		{ "nostdinc", handle_nostdinc },
1098 		{ "gcc-base-dir", handle_base_dir},
1099 		{ "no-lineno", handle_no_lineno},
1100 		{ NULL, NULL }
1101 	};
1102 	struct switches *s = cmd;
1103 
1104 	while (s->name) {
1105 		int optlen = strlen(s->name);
1106 		if (!strncmp(s->name, arg, optlen + !s->prefix))
1107 			return s->fn(arg + optlen, next);
1108 		s++;
1109 	}
1110 	return next;
1111 }
1112 
handle_switch(char * arg,char ** next)1113 static char **handle_switch(char *arg, char **next)
1114 {
1115 	switch (*arg) {
1116 	case 'a': return handle_switch_a(arg, next);
1117 	case 'D': return handle_switch_D(arg, next);
1118 	case 'd': return handle_switch_d(arg, next);
1119 	case 'E': return handle_switch_E(arg, next);
1120 	case 'f': return handle_switch_f(arg, next);
1121 	case 'g': return handle_switch_g(arg, next);
1122 	case 'G': return handle_switch_G(arg, next);
1123 	case 'I': return handle_switch_I(arg, next);
1124 	case 'i': return handle_switch_i(arg, next);
1125 	case 'M': return handle_switch_M(arg, next);
1126 	case 'm': return handle_switch_m(arg, next);
1127 	case 'n': return handle_switch_n(arg, next);
1128 	case 'o': return handle_switch_o(arg, next);
1129 	case 'O': return handle_switch_O(arg, next);
1130 	case 's': return handle_switch_s(arg, next);
1131 	case 'U': return handle_switch_U(arg, next);
1132 	case 'v': return handle_switch_v(arg, next);
1133 	case 'W': return handle_switch_W(arg, next);
1134 	case 'x': return handle_switch_x(arg, next);
1135 	case '-': return handle_long_options(arg + 1, next);
1136 	default:
1137 		break;
1138 	}
1139 
1140 	/*
1141 	 * Ignore unknown command line options:
1142 	 * they're probably gcc switches
1143 	 */
1144 	return next;
1145 }
1146 
1147 #define	PTYPE_SIZEOF	(1U << 0)
1148 #define	PTYPE_T		(1U << 1)
1149 #define	PTYPE_MAX	(1U << 2)
1150 #define	PTYPE_MIN	(1U << 3)
1151 #define	PTYPE_WIDTH	(1U << 4)
1152 #define	PTYPE_TYPE	(1U << 5)
1153 #define	PTYPE_ALL	(PTYPE_MAX|PTYPE_SIZEOF|PTYPE_WIDTH)
1154 #define	PTYPE_ALL_T	(PTYPE_MAX|PTYPE_SIZEOF|PTYPE_WIDTH|PTYPE_T)
1155 
predefined_sizeof(const char * name,const char * suffix,unsigned bits)1156 static void predefined_sizeof(const char *name, const char *suffix, unsigned bits)
1157 {
1158 	char buf[32];
1159 
1160 	snprintf(buf, sizeof(buf), "__SIZEOF_%s%s__", name, suffix);
1161 	predefine(buf, 1, "%d", bits/8);
1162 }
1163 
predefined_width(const char * name,unsigned bits)1164 static void predefined_width(const char *name, unsigned bits)
1165 {
1166 	char buf[32];
1167 
1168 	snprintf(buf, sizeof(buf), "__%s_WIDTH__", name);
1169 	predefine(buf, 1, "%d", bits);
1170 }
1171 
predefined_max(const char * name,struct symbol * type)1172 static void predefined_max(const char *name, struct symbol *type)
1173 {
1174 	const char *suffix = builtin_type_suffix(type);
1175 	unsigned bits = type->bit_size - is_signed_type(type);
1176 	unsigned long long max = bits_mask(bits);
1177 	char buf[32];
1178 
1179 	snprintf(buf, sizeof(buf), "__%s_MAX__", name);
1180 	predefine(buf, 1, "%#llx%s", max, suffix);
1181 }
1182 
predefined_min(const char * name,struct symbol * type)1183 static void predefined_min(const char *name, struct symbol *type)
1184 {
1185 	const char *suffix = builtin_type_suffix(type);
1186 	char buf[32];
1187 
1188 	snprintf(buf, sizeof(buf), "__%s_MIN__", name);
1189 
1190 	if (is_signed_type(type))
1191 		predefine(buf, 1, "(-__%s_MAX__ - 1)", name);
1192 	else
1193 		predefine(buf, 1, "0%s", suffix);
1194 }
1195 
predefined_type(const char * name,struct symbol * type)1196 static void predefined_type(const char *name, struct symbol *type)
1197 {
1198 	const char *typename = builtin_typename(type);
1199 	add_pre_buffer("#weak_define __%s_TYPE__ %s\n", name, typename);
1200 }
1201 
predefined_ctype(const char * name,struct symbol * type,int flags)1202 static void predefined_ctype(const char *name, struct symbol *type, int flags)
1203 {
1204 	unsigned bits = type->bit_size;
1205 
1206 	if (flags & PTYPE_SIZEOF) {
1207 		const char *suffix = (flags & PTYPE_T) ? "_T" : "";
1208 		predefined_sizeof(name, suffix, bits);
1209 	}
1210 	if (flags & PTYPE_MAX)
1211 		predefined_max(name, type);
1212 	if (flags & PTYPE_MIN)
1213 		predefined_min(name, type);
1214 	if (flags & PTYPE_TYPE)
1215 		predefined_type(name, type);
1216 	if (flags & PTYPE_WIDTH)
1217 		predefined_width(name, bits);
1218 }
1219 
predefined_macros(void)1220 static void predefined_macros(void)
1221 {
1222 	predefine("__CHECKER__", 0, "1");
1223 	predefine("__GNUC__", 1, "%d", gcc_major);
1224 	predefine("__GNUC_MINOR__", 1, "%d", gcc_minor);
1225 	predefine("__GNUC_PATCHLEVEL__", 1, "%d", gcc_patchlevel);
1226 
1227 	predefine("__STDC__", 1, "1");
1228 	switch (standard) {
1229 	case STANDARD_C89:
1230 		predefine("__STRICT_ANSI__", 1, "1");
1231 		break;
1232 
1233 	case STANDARD_C94:
1234 		predefine("__STDC_VERSION__", 1, "199409L");
1235 		predefine("__STRICT_ANSI__", 1, "1");
1236 		break;
1237 
1238 	case STANDARD_C99:
1239 		predefine("__STDC_VERSION__", 1, "199901L");
1240 		predefine("__STRICT_ANSI__", 1, "1");
1241 		break;
1242 
1243 	case STANDARD_GNU89:
1244 	default:
1245 		break;
1246 
1247 	case STANDARD_GNU99:
1248 		predefine("__STDC_VERSION__", 1, "199901L");
1249 		break;
1250 
1251 	case STANDARD_C11:
1252 		predefine("__STRICT_ANSI__", 1, "1");
1253 	case STANDARD_GNU11:
1254 		predefine("__STDC_NO_ATOMICS__", 1, "1");
1255 		predefine("__STDC_NO_COMPLEX__", 1, "1");
1256 		predefine("__STDC_NO_THREADS__", 1, "1");
1257 		predefine("__STDC_VERSION__", 1, "201112L");
1258 		break;
1259 	}
1260 
1261 	predefine("__CHAR_BIT__", 1, "%d", bits_in_char);
1262 	if (funsigned_char)
1263 		predefine("__CHAR_UNSIGNED__", 1, "1");
1264 
1265 	predefined_ctype("SHORT",     &short_ctype, PTYPE_SIZEOF);
1266 	predefined_ctype("SHRT",      &short_ctype, PTYPE_MAX|PTYPE_WIDTH);
1267 	predefined_ctype("SCHAR",     &schar_ctype, PTYPE_MAX|PTYPE_WIDTH);
1268 	predefined_ctype("WCHAR",      wchar_ctype, PTYPE_ALL_T|PTYPE_MIN|PTYPE_TYPE);
1269 	predefined_ctype("WINT",        wint_ctype, PTYPE_ALL_T|PTYPE_MIN|PTYPE_TYPE);
1270 	predefined_ctype("CHAR16",   &ushort_ctype, PTYPE_TYPE);
1271 	predefined_ctype("CHAR32",     &uint_ctype, PTYPE_TYPE);
1272 
1273 	predefined_ctype("INT",         &int_ctype, PTYPE_ALL);
1274 	predefined_ctype("LONG",       &long_ctype, PTYPE_ALL);
1275 	predefined_ctype("LONG_LONG", &llong_ctype, PTYPE_ALL);
1276 
1277 	predefined_ctype("INT8",      &schar_ctype, PTYPE_MAX|PTYPE_TYPE);
1278 	predefined_ctype("UINT8",     &uchar_ctype, PTYPE_MAX|PTYPE_TYPE);
1279 	predefined_ctype("INT16",     &short_ctype, PTYPE_MAX|PTYPE_TYPE);
1280 	predefined_ctype("UINT16",   &ushort_ctype, PTYPE_MAX|PTYPE_TYPE);
1281 	predefined_ctype("INT32",      int32_ctype, PTYPE_MAX|PTYPE_TYPE);
1282 	predefined_ctype("UINT32",    uint32_ctype, PTYPE_MAX|PTYPE_TYPE);
1283 	predefined_ctype("INT64",      int64_ctype, PTYPE_MAX|PTYPE_TYPE);
1284 	predefined_ctype("UINT64",    uint64_ctype, PTYPE_MAX|PTYPE_TYPE);
1285 
1286 	predefined_sizeof("INT128", "", 128);
1287 
1288 	predefined_ctype("INTMAX",    intmax_ctype, PTYPE_MAX|PTYPE_TYPE|PTYPE_WIDTH);
1289 	predefined_ctype("UINTMAX",  uintmax_ctype, PTYPE_MAX|PTYPE_TYPE);
1290 	predefined_ctype("INTPTR",   ssize_t_ctype, PTYPE_MAX|PTYPE_TYPE|PTYPE_WIDTH);
1291 	predefined_ctype("UINTPTR",   size_t_ctype, PTYPE_MAX|PTYPE_TYPE);
1292 	predefined_ctype("PTRDIFF",  ssize_t_ctype, PTYPE_ALL_T|PTYPE_TYPE);
1293 	predefined_ctype("SIZE",      size_t_ctype, PTYPE_ALL_T|PTYPE_TYPE);
1294 	predefined_ctype("POINTER",     &ptr_ctype, PTYPE_SIZEOF);
1295 
1296 	predefined_sizeof("FLOAT", "", bits_in_float);
1297 	predefined_sizeof("DOUBLE", "", bits_in_double);
1298 	predefined_sizeof("LONG_DOUBLE", "", bits_in_longdouble);
1299 
1300 	predefine("__ORDER_LITTLE_ENDIAN__", 1, "1234");
1301 	predefine("__ORDER_BIG_ENDIAN__", 1, "4321");
1302 	predefine("__ORDER_PDP_ENDIAN__", 1, "3412");
1303 	if (arch_big_endian) {
1304 		predefine("__BIG_ENDIAN__", 1, "1");
1305 		predefine("__BYTE_ORDER__", 1, "__ORDER_BIG_ENDIAN__");
1306 	} else {
1307 		predefine("__LITTLE_ENDIAN__", 1, "1");
1308 		predefine("__BYTE_ORDER__", 1, "__ORDER_LITTLE_ENDIAN__");
1309 	}
1310 
1311 	if (optimize_level)
1312 		predefine("__OPTIMIZE__", 0, "1");
1313 	if (optimize_size)
1314 		predefine("__OPTIMIZE_SIZE__", 0, "1");
1315 
1316 	predefine("__PRAGMA_REDEFINE_EXTNAME", 1, "1");
1317 
1318 	// Temporary hacks
1319 	predefine("__extension__", 0, NULL);
1320 	predefine("__pragma__", 0, NULL);
1321 
1322 	switch (arch_m64) {
1323 	case ARCH_LP32:
1324 		break;
1325 	case ARCH_X32:
1326 		predefine("__ILP32__", 1, "1");
1327 		predefine("_ILP32", 1, "1");
1328 		break;
1329 	case ARCH_LP64:
1330 		predefine("__LP64__", 1, "1");
1331 		predefine("_LP64", 1, "1");
1332 		break;
1333 	case ARCH_LLP64:
1334 		predefine("__LLP64__", 1, "1");
1335 		break;
1336 	}
1337 
1338 	switch (arch_mach) {
1339 	case MACH_ARM64:
1340 		predefine("__aarch64__", 1, "1");
1341 		break;
1342 	case MACH_ARM:
1343 		predefine("__arm__", 1, "1");
1344 		break;
1345 	case MACH_M68K:
1346 		predefine("__m68k__", 1, "1");
1347 		break;
1348 	case MACH_MIPS64:
1349 		if (arch_m64 == ARCH_LP64)
1350 			predefine("__mips64", 1, "64");
1351 		/* fall-through */
1352 	case MACH_MIPS32:
1353 		predefine("__mips", 1, "%d", ptr_ctype.bit_size);
1354 		predefine("_MIPS_SZINT", 1, "%d", int_ctype.bit_size);
1355 		predefine("_MIPS_SZLONG", 1, "%d", long_ctype.bit_size);
1356 		predefine("_MIPS_SZPTR", 1, "%d", ptr_ctype.bit_size);
1357 		break;
1358 	case MACH_PPC64:
1359 		if (arch_m64 == ARCH_LP64) {
1360 			predefine("__powerpc64__", 1, "1");
1361 			predefine("__ppc64__", 1, "1");
1362 			predefine("__PPC64__", 1, "1");
1363 		}
1364 		/* fall-through */
1365 	case MACH_PPC32:
1366 		predefine("__powerpc__", 1, "1");
1367 		predefine("__powerpc", 1, "1");
1368 		predefine("__ppc__", 1, "1");
1369 		predefine("__PPC__", 1, "1");
1370 		break;
1371 	case MACH_RISCV64:
1372 	case MACH_RISCV32:
1373 		predefine("__riscv", 1, "1");
1374 		predefine("__riscv_xlen", 1, "%d", ptr_ctype.bit_size);
1375 		break;
1376 	case MACH_S390X:
1377 		predefine("__zarch__", 1, "1");
1378 		predefine("__s390x__", 1, "1");
1379 		predefine("__s390__", 1, "1");
1380 		break;
1381 	case MACH_SPARC64:
1382 		if (arch_m64 == ARCH_LP64) {
1383 			predefine("__sparc_v9__", 1, "1");
1384 			predefine("__sparcv9__", 1, "1");
1385 			predefine("__sparcv9", 1, "1");
1386 			predefine("__sparc64__", 1, "1");
1387 			predefine("__arch64__", 1, "1");
1388 		}
1389 		/* fall-through */
1390 	case MACH_SPARC32:
1391 		predefine("__sparc__", 1, "1");
1392 		predefine("__sparc", 1, "1");
1393 		predefine_nostd("sparc");
1394 		break;
1395 	case MACH_X86_64:
1396 		if (arch_m64 != ARCH_LP32) {
1397 			predefine("__x86_64__", 1, "1");
1398 			predefine("__x86_64", 1, "1");
1399 			predefine("__amd64__", 1, "1");
1400 			predefine("__amd64", 1, "1");
1401 			break;
1402 		}
1403 		/* fall-through */
1404 	case MACH_I386:
1405 		predefine("__i386__", 1, "1");
1406 		predefine("__i386", 1, "1");
1407 		predefine_nostd("i386");
1408 		break;
1409 	}
1410 
1411 #if defined(__unix__)
1412 	predefine("__unix__", 1, "1");
1413 	predefine("__unix", 1, "1");
1414 	predefine_nostd("unix");
1415 #endif
1416 
1417 
1418 #if defined(__sun__) || defined(__sun)
1419 	predefine("__sun__", 1, "1");
1420 	predefine("__sun", 1, "1");
1421 	predefine_nostd("sun");
1422 	predefine("__svr4__", 1, "1");
1423 #endif
1424 
1425 }
1426 
create_builtin_stream(void)1427 static void create_builtin_stream(void)
1428 {
1429 	// Temporary hack
1430 	add_pre_buffer("#define _Pragma(x)\n");
1431 
1432 	/* add the multiarch include directories, if any */
1433 	if (multiarch_dir && *multiarch_dir) {
1434 		add_pre_buffer("#add_system \"/usr/include/%s\"\n", multiarch_dir);
1435 		add_pre_buffer("#add_system \"/usr/local/include/%s\"\n", multiarch_dir);
1436 	}
1437 
1438 	/* We add compiler headers path here because we have to parse
1439 	 * the arguments to get it, falling back to default. */
1440 	add_pre_buffer("#add_system \"%s/include\"\n", gcc_base_dir);
1441 	add_pre_buffer("#add_system \"%s/include-fixed\"\n", gcc_base_dir);
1442 
1443 	add_pre_buffer("#define __has_builtin(x) 0\n");
1444 	add_pre_buffer("#define __has_attribute(x) 0\n");
1445 	add_pre_buffer("#define __builtin_stdarg_start(a,b) ((a) = (__builtin_va_list)(&(b)))\n");
1446 	add_pre_buffer("#define __builtin_va_start(a,b) ((a) = (__builtin_va_list)(&(b)))\n");
1447 	add_pre_buffer("#define __builtin_ms_va_start(a,b) ((a) = (__builtin_ms_va_list)(&(b)))\n");
1448 	add_pre_buffer("#define __builtin_va_arg(arg,type)  ({ type __va_arg_ret = *(type *)(arg); arg += sizeof(type); __va_arg_ret; })\n");
1449 	add_pre_buffer("#define __builtin_va_alist (*(void *)0)\n");
1450 	add_pre_buffer("#define __builtin_va_arg_incr(x) ((x) + 1)\n");
1451 	add_pre_buffer("#define __builtin_va_copy(dest, src) ({ dest = src; (void)0; })\n");
1452 	add_pre_buffer("#define __builtin_ms_va_copy(dest, src) ({ dest = src; (void)0; })\n");
1453 	add_pre_buffer("#define __builtin_va_end(arg)\n");
1454 	add_pre_buffer("#define __builtin_ms_va_end(arg)\n");
1455 	add_pre_buffer("#define __builtin_va_arg_pack()\n");
1456 }
1457 
sparse_tokenstream(struct token * token)1458 static struct symbol_list *sparse_tokenstream(struct token *token)
1459 {
1460 	int builtin = token && !token->pos.stream;
1461 
1462 	// Preprocess the stream
1463 	token = preprocess(token);
1464 
1465 	if (dump_macro_defs || dump_macros_only) {
1466 		if (!builtin)
1467 			dump_macro_definitions();
1468 		if (dump_macros_only)
1469 			return NULL;
1470 	}
1471 
1472 	if (preprocess_only) {
1473 		while (!eof_token(token)) {
1474 			int prec = 1;
1475 			struct token *next = token->next;
1476 			const char *separator = "";
1477 			if (next->pos.whitespace)
1478 				separator = " ";
1479 			if (next->pos.newline) {
1480 				separator = "\n\t\t\t\t\t";
1481 				prec = next->pos.pos;
1482 				if (prec > 4)
1483 					prec = 4;
1484 			}
1485 			printf("%s%.*s", show_token(token), prec, separator);
1486 			token = next;
1487 		}
1488 		putchar('\n');
1489 
1490 		return NULL;
1491 	}
1492 
1493 	// Parse the resulting C code
1494 	while (!eof_token(token))
1495 		token = external_declaration(token, &translation_unit_used_list, NULL);
1496 	return translation_unit_used_list;
1497 }
1498 
sparse_file(const char * filename)1499 static struct symbol_list *sparse_file(const char *filename)
1500 {
1501 	int fd;
1502 	struct token *token;
1503 
1504 	if (strcmp (filename, "-") == 0) {
1505 		fd = 0;
1506 	} else {
1507 		fd = open(filename, O_RDONLY);
1508 		if (fd < 0)
1509 			die("No such file: %s", filename);
1510 	}
1511 	base_filename = filename;
1512 
1513 	// Tokenize the input stream
1514 	token = tokenize(filename, fd, NULL, includepath);
1515 	store_all_tokens(token);
1516 	close(fd);
1517 
1518 	return sparse_tokenstream(token);
1519 }
1520 
1521 /*
1522  * This handles the "-include" directive etc: we're in global
1523  * scope, and all types/macros etc will affect all the following
1524  * files.
1525  *
1526  * NOTE NOTE NOTE! "#undef" of anything in this stage will
1527  * affect all subsequent files too, i.e. we can have non-local
1528  * behaviour between files!
1529  */
sparse_initial(void)1530 static struct symbol_list *sparse_initial(void)
1531 {
1532 	int i;
1533 
1534 	// Prepend any "include" file to the stream.
1535 	// We're in global scope, it will affect all files!
1536 	for (i = 0; i < cmdline_include_nr; i++)
1537 		add_pre_buffer("#argv_include \"%s\"\n", cmdline_include[i]);
1538 
1539 	return sparse_tokenstream(pre_buffer_begin);
1540 }
1541 
endswith(const char * str,const char * suffix)1542 static int endswith(const char *str, const char *suffix)
1543 {
1544 	const char *found = strstr(str, suffix);
1545 	return (found && strcmp(found, suffix) == 0);
1546 }
1547 
sparse_initialize(int argc,char ** argv,struct string_list ** filelist)1548 struct symbol_list *sparse_initialize(int argc, char **argv, struct string_list **filelist)
1549 {
1550 	char **args;
1551 	struct symbol_list *list;
1552 
1553 	// Initialize symbol stream first, so that we can add defines etc
1554 	init_symbols();
1555 	init_include_path();
1556 
1557 	diag_prefix = argv[0];
1558 
1559 	args = argv;
1560 	for (;;) {
1561 		char *arg = *++args;
1562 		if (!arg)
1563 			break;
1564 
1565 		if (arg[0] == '-' && arg[1]) {
1566 			args = handle_switch(arg+1, args);
1567 			continue;
1568 		}
1569 
1570 		if (endswith(arg, ".a") || endswith(arg, ".so") ||
1571 		    endswith(arg, ".so.1") || endswith(arg, ".o"))
1572 			continue;
1573 
1574 		add_ptr_list(filelist, arg);
1575 	}
1576 	handle_switch_W_finalize();
1577 	handle_switch_v_finalize();
1578 
1579 	// Redirect stdout if needed
1580 	if (dump_macro_defs || preprocess_only)
1581 		do_output = 1;
1582 	if (do_output && outfile && strcmp(outfile, "-")) {
1583 		if (!freopen(outfile, "w", stdout))
1584 			die("error: cannot open %s: %s", outfile, strerror(errno));
1585 	}
1586 
1587 	if (fdump_ir == 0)
1588 		fdump_ir = PASS_FINAL;
1589 
1590 	list = NULL;
1591 	if (filelist) {
1592 		// Initialize type system
1593 		init_target();
1594 		handle_arch_finalize();
1595 		init_ctype();
1596 
1597 		predefined_macros();
1598 		create_builtin_stream();
1599 		declare_builtins();
1600 
1601 		list = sparse_initial();
1602 
1603 		/*
1604 		 * Protect the initial token allocations, since
1605 		 * they need to survive all the others
1606 		 */
1607 		protect_token_alloc();
1608 	}
1609 	/*
1610 	 * Evaluate the complete symbol list
1611 	 * Note: This is not needed for normal cases.
1612 	 *	 These symbols should only be predefined defines and
1613 	 *	 declaratons which will be evaluated later, when needed.
1614 	 *	 This is also the case when a file is directly included via
1615 	 *	 '-include <file>' on the command line *AND* the file only
1616 	 *	 contains defines, declarations and inline definitions.
1617 	 *	 However, in the rare cases where the given file should
1618 	 *	 contain some definitions, these will never be evaluated
1619 	 *	 and thus won't be able to be linearized correctly.
1620 	 *	 Hence the evaluate_symbol_list() here under.
1621 	 */
1622 	evaluate_symbol_list(list);
1623 	return list;
1624 }
1625 
sparse_keep_tokens(char * filename)1626 struct symbol_list * sparse_keep_tokens(char *filename)
1627 {
1628 	struct symbol_list *res;
1629 
1630 	/* Clear previous symbol list */
1631 	translation_unit_used_list = NULL;
1632 
1633 	new_file_scope();
1634 	res = sparse_file(filename);
1635 
1636 	/* And return it */
1637 	return res;
1638 }
1639 
1640 
__sparse(char * filename)1641 struct symbol_list * __sparse(char *filename)
1642 {
1643 	struct symbol_list *res;
1644 
1645 	res = sparse_keep_tokens(filename);
1646 
1647 	/* Drop the tokens for this file after parsing */
1648 	clear_token_alloc();
1649 
1650 	/* And return it */
1651 	return res;
1652 }
1653 
sparse(char * filename)1654 struct symbol_list * sparse(char *filename)
1655 {
1656 	struct symbol_list *res = __sparse(filename);
1657 
1658 	if (has_error & ERROR_CURR_PHASE)
1659 		has_error = ERROR_PREV_PHASE;
1660 	/* Evaluate the complete symbol list */
1661 	evaluate_symbol_list(res);
1662 
1663 	return res;
1664 }
1665