Home | History | Annotate | Line # | Download | only in lint1
emit1.c revision 1.84
      1 /* $NetBSD: emit1.c,v 1.84 2024/02/03 12:57:12 rillig Exp $ */
      2 
      3 /*
      4  * Copyright (c) 1996 Christopher G. Demetriou.  All Rights Reserved.
      5  * Copyright (c) 1994, 1995 Jochen Pohl
      6  * All Rights Reserved.
      7  *
      8  * Redistribution and use in source and binary forms, with or without
      9  * modification, are permitted provided that the following conditions
     10  * are met:
     11  * 1. Redistributions of source code must retain the above copyright
     12  *    notice, this list of conditions and the following disclaimer.
     13  * 2. Redistributions in binary form must reproduce the above copyright
     14  *    notice, this list of conditions and the following disclaimer in the
     15  *    documentation and/or other materials provided with the distribution.
     16  * 3. All advertising materials mentioning features or use of this software
     17  *    must display the following acknowledgement:
     18  *	This product includes software developed by Jochen Pohl for
     19  *	The NetBSD Project.
     20  * 4. The name of the author may not be used to endorse or promote products
     21  *    derived from this software without specific prior written permission.
     22  *
     23  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
     24  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
     25  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
     26  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
     27  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
     28  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
     29  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
     30  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
     31  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
     32  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     33  */
     34 
     35 #if HAVE_NBTOOL_CONFIG_H
     36 #include "nbtool_config.h"
     37 #endif
     38 
     39 #include <sys/cdefs.h>
     40 #if defined(__RCSID)
     41 __RCSID("$NetBSD: emit1.c,v 1.84 2024/02/03 12:57:12 rillig Exp $");
     42 #endif
     43 
     44 #include "lint1.h"
     45 
     46 static void outtt(sym_t *, sym_t *);
     47 static void outfstrg(const char *);
     48 
     49 /*
     50  * Write type into the output file, encoded as follows:
     51  *	const			c
     52  *	volatile		v
     53  *	_Bool			B
     54  *	_Complex float		s X
     55  *	_Complex double		X
     56  *	_Complex long double	l X
     57  *	char			C
     58  *	signed char		s C
     59  *	unsigned char		u C
     60  *	short			S
     61  *	unsigned short		u S
     62  *	int			I
     63  *	unsigned int		u I
     64  *	long			L
     65  *	unsigned long		u L
     66  *	long long		Q
     67  *	unsigned long long	u Q
     68  *	float			s D
     69  *	double			D
     70  *	long double		l D
     71  *	void			V
     72  *	*			P
     73  *	[n]			A n
     74  *	()			F
     75  *	(void)			F 0
     76  *	(n parameters)		F n arg1 arg2 ... argn
     77  *	(n parameters, ...)	F n arg1 arg2 ... argn E
     78  *	enum tag		e T tag_or_typename
     79  *	struct tag		s T tag_or_typename
     80  *	union tag		u T tag_or_typename
     81  *
     82  *	tag_or_typename		0 (obsolete)		no tag or type name
     83  *				1 n tag			tagged type
     84  *				2 n typename		only typedef name
     85  *				3 line.file.uniq	anonymous types
     86  */
     87 void
     88 outtype(const type_t *tp)
     89 {
     90 	/* Available letters: ------GH--K-MNO--R--U-W-YZ */
     91 #ifdef INT128_SIZE
     92 	static const char tt[NTSPEC] = "???BCCCSSIILLQQJJDDD?XXXVTTTPAF";
     93 	static const char ss[NTSPEC] = "???  su u u u u us l?s l sue   ";
     94 #else
     95 	static const char tt[NTSPEC] = "???BCCCSSIILLQQDDD?XXXVTTTPAF";
     96 	static const char ss[NTSPEC] = "???  su u u u us l?s l sue   ";
     97 #endif
     98 	int na;
     99 	tspec_t ts;
    100 
    101 	while (tp != NULL) {
    102 		if ((ts = tp->t_tspec) == INT && tp->t_is_enum)
    103 			ts = ENUM;
    104 		lint_assert(tt[ts] != '?' && ss[ts] != '?');
    105 		if (tp->t_const)
    106 			outchar('c');
    107 		if (tp->t_volatile)
    108 			outchar('v');
    109 		if (ss[ts] != ' ')
    110 			outchar(ss[ts]);
    111 		outchar(tt[ts]);
    112 
    113 		if (ts == ARRAY) {
    114 			outint(tp->t_dim);
    115 		} else if (ts == ENUM) {
    116 			outtt(tp->t_enum->en_tag, tp->t_enum->en_first_typedef);
    117 		} else if (is_struct_or_union(ts)) {
    118 			outtt(tp->t_sou->sou_tag, tp->t_sou->sou_first_typedef);
    119 		} else if (ts == FUNC && tp->t_proto) {
    120 			na = 0;
    121 			for (const sym_t *param = tp->t_params;
    122 			    param != NULL; param = param->s_next)
    123 				na++;
    124 			if (tp->t_vararg)
    125 				na++;
    126 			outint(na);
    127 			for (const sym_t *param = tp->t_params;
    128 			    param != NULL; param = param->s_next)
    129 				outtype(param->s_type);
    130 			if (tp->t_vararg)
    131 				outchar('E');
    132 		}
    133 		tp = tp->t_subt;
    134 	}
    135 }
    136 
    137 /*
    138  * write the name of a tag or typename
    139  *
    140  * if the tag is named, the name of the tag is written,
    141  * otherwise, if a typename exists which refers to this tag,
    142  * this typename is written
    143  */
    144 static void
    145 outtt(sym_t *tag, sym_t *tdef)
    146 {
    147 
    148 	/* 0 is no longer used. */
    149 
    150 	if (tag->s_name != unnamed) {
    151 		outint(1);
    152 		outname(tag->s_name);
    153 	} else if (tdef != NULL) {
    154 		outint(2);
    155 		outname(tdef->s_name);
    156 	} else {
    157 		outint(3);
    158 		outint(tag->s_def_pos.p_line);
    159 		outchar('.');
    160 		outint(get_filename_id(tag->s_def_pos.p_file));
    161 		outchar('.');
    162 		outint(tag->s_def_pos.p_uniq);
    163 	}
    164 }
    165 
    166 /*
    167  * write information about a globally declared/defined symbol
    168  * with storage class extern
    169  *
    170  * information about function definitions are written in outfdef(),
    171  * not here
    172  */
    173 void
    174 outsym(const sym_t *sym, scl_t sc, def_t def)
    175 {
    176 
    177 	/*
    178 	 * Static function declarations must also be written to the output
    179 	 * file. Compatibility of function declarations (for both static and
    180 	 * extern functions) must be checked in lint2. Lint1 can't do this,
    181 	 * especially not if functions are declared at block level before their
    182 	 * first declaration at level 0.
    183 	 */
    184 	if (sc != EXTERN && !(sc == STATIC && sym->s_type->t_tspec == FUNC))
    185 		return;
    186 	if (ch_isdigit(sym->s_name[0]))	/* 00000000_tmp */
    187 		return;
    188 
    189 	outint(csrc_pos.p_line);
    190 	outchar('d');		/* declaration */
    191 	outint(get_filename_id(sym->s_def_pos.p_file));
    192 	outchar('.');
    193 	outint(sym->s_def_pos.p_line);
    194 
    195 	/* flags */
    196 
    197 	if (def == DEF)
    198 		outchar('d');	/* defined */
    199 	else if (def == TDEF)
    200 		outchar('t');	/* tentative defined */
    201 	else {
    202 		lint_assert(def == DECL);
    203 		outchar('e');	/* declared */
    204 	}
    205 
    206 	if (llibflg && def != DECL) {
    207 		/*
    208 		 * mark it as used so lint2 does not complain about unused
    209 		 * symbols in libraries
    210 		 */
    211 		outchar('u');
    212 	}
    213 
    214 	if (sc == STATIC)
    215 		outchar('s');
    216 
    217 	outname(sym->s_name);
    218 
    219 	if (sym->s_rename != NULL) {
    220 		outchar('r');
    221 		outname(sym->s_rename);
    222 	}
    223 
    224 	outtype(sym->s_type);
    225 	outchar('\n');
    226 }
    227 
    228 /*
    229  * Write information about a function definition. This is also done for static
    230  * functions, to later check if they are called with proper argument types.
    231  */
    232 void
    233 outfdef(const sym_t *fsym, const pos_t *posp, bool rval, bool osdef,
    234 	const sym_t *args)
    235 {
    236 	int narg;
    237 
    238 	if (posp->p_file == csrc_pos.p_file) {
    239 		outint(posp->p_line);
    240 	} else {
    241 		outint(csrc_pos.p_line);
    242 	}
    243 	outchar('d');		/* declaration */
    244 	outint(get_filename_id(posp->p_file));
    245 	outchar('.');
    246 	outint(posp->p_line);
    247 
    248 	/* both SCANFLIKE and PRINTFLIKE imply VARARGS */
    249 	if (printflike_argnum != -1) {
    250 		nvararg = printflike_argnum;
    251 	} else if (scanflike_argnum != -1) {
    252 		nvararg = scanflike_argnum;
    253 	}
    254 
    255 	if (nvararg != -1) {
    256 		outchar('v');
    257 		outint(nvararg);
    258 	}
    259 	if (scanflike_argnum != -1) {
    260 		outchar('S');
    261 		outint(scanflike_argnum);
    262 	}
    263 	if (printflike_argnum != -1) {
    264 		outchar('P');
    265 		outint(printflike_argnum);
    266 	}
    267 	nvararg = printflike_argnum = scanflike_argnum = -1;
    268 
    269 	outchar('d');
    270 
    271 	if (rval)
    272 		outchar('r');	/* has return value */
    273 
    274 	if (llibflg)
    275 		/*
    276 		 * mark it as used so lint2 does not complain about unused
    277 		 * symbols in libraries
    278 		 */
    279 		outchar('u');
    280 
    281 	if (osdef)
    282 		outchar('o');	/* old-style function definition */
    283 
    284 	if (fsym->s_inline)
    285 		outchar('i');
    286 
    287 	if (fsym->s_scl == STATIC)
    288 		outchar('s');
    289 
    290 	outname(fsym->s_name);
    291 
    292 	if (fsym->s_rename != NULL) {
    293 		outchar('r');
    294 		outname(fsym->s_rename);
    295 	}
    296 
    297 	/* parameter types and return value */
    298 	if (osdef) {
    299 		narg = 0;
    300 		for (const sym_t *arg = args; arg != NULL; arg = arg->s_next)
    301 			narg++;
    302 		outchar('f');
    303 		outint(narg);
    304 		for (const sym_t *arg = args; arg != NULL; arg = arg->s_next)
    305 			outtype(arg->s_type);
    306 		outtype(fsym->s_type->t_subt);
    307 	} else {
    308 		outtype(fsym->s_type);
    309 	}
    310 	outchar('\n');
    311 }
    312 
    313 /*
    314  * write out all information necessary for lint2 to check function
    315  * calls
    316  *
    317  * retval_used is set if the return value is used (assigned to a variable)
    318  * retval_discarded is set if the return value is neither used nor ignored
    319  * (that is, cast to void)
    320  */
    321 void
    322 outcall(const tnode_t *tn, bool retval_used, bool retval_discarded)
    323 {
    324 	tnode_t *args, *arg;
    325 	int narg, i;
    326 
    327 	outint(csrc_pos.p_line);
    328 	outchar('c');		/* function call */
    329 	outint(get_filename_id(curr_pos.p_file));
    330 	outchar('.');
    331 	outint(curr_pos.p_line);
    332 
    333 	/*
    334 	 * flags; 'u' and 'i' must be last to make sure a letter is between the
    335 	 * numeric argument of a flag and the name of the function
    336 	 */
    337 	narg = 0;
    338 	args = tn_ck_right(tn);
    339 	for (arg = args; arg != NULL; arg = tn_ck_right(arg))
    340 		narg++;
    341 	/* information about arguments */
    342 	for (int n = 1; n <= narg; n++) {
    343 		/* the last argument is the top one in the tree */
    344 		for (i = narg, arg = args; i > n; i--, arg = arg->tn_right)
    345 			continue;
    346 		arg = arg->tn_left;
    347 		if (arg->tn_op == CON) {
    348 			tspec_t t = arg->tn_type->t_tspec;
    349 			if (is_integer(t)) {
    350 				/*
    351 				 * XXX it would probably be better to
    352 				 * explicitly test the sign
    353 				 */
    354 				int64_t si = arg->tn_val.u.integer;
    355 				if (si == 0) {
    356 					/* zero constant */
    357 					outchar('z');
    358 				} else if (!msb(si, t)) {
    359 					/* positive if cast to signed */
    360 					outchar('p');
    361 				} else {
    362 					/* negative if cast to signed */
    363 					outchar('n');
    364 				}
    365 				outint(n);
    366 			}
    367 		} else if (arg->tn_op == ADDR &&
    368 		    arg->tn_left->tn_op == STRING &&
    369 		    arg->tn_left->tn_string->data != NULL) {
    370 			/* constant string, write all format specifiers */
    371 			outchar('s');
    372 			outint(n);
    373 			outfstrg(arg->tn_left->tn_string->data);
    374 		}
    375 	}
    376 	outchar((char)(retval_discarded ? 'd' : retval_used ? 'u' : 'i'));
    377 
    378 	/* name of the called function */
    379 	outname(tn_ck_left(tn->tn_left)->tn_sym->s_name);
    380 
    381 	/* types of arguments */
    382 	outchar('f');
    383 	outint(narg);
    384 	for (int n = 1; n <= narg; n++) {
    385 		/* the last argument is the top one in the tree */
    386 		for (i = narg, arg = args; i > n; i--, arg = arg->tn_right)
    387 			continue;
    388 		outtype(arg->tn_left->tn_type);
    389 	}
    390 	/* expected type of return value */
    391 	outtype(tn->tn_type);
    392 	outchar('\n');
    393 }
    394 
    395 /* write a character to the output file, quoted if necessary */
    396 static void
    397 outqchar(char c)
    398 {
    399 
    400 	if (ch_isprint(c) && c != '\\' && c != '"' && c != '\'') {
    401 		outchar(c);
    402 		return;
    403 	}
    404 
    405 	outchar('\\');
    406 	switch (c) {
    407 	case '\\':
    408 		outchar('\\');
    409 		break;
    410 	case '"':
    411 		outchar('"');
    412 		break;
    413 	case '\'':
    414 		outchar('\'');
    415 		break;
    416 	case '\b':
    417 		outchar('b');
    418 		break;
    419 	case '\t':
    420 		outchar('t');
    421 		break;
    422 	case '\n':
    423 		outchar('n');
    424 		break;
    425 	case '\f':
    426 		outchar('f');
    427 		break;
    428 	case '\r':
    429 		outchar('r');
    430 		break;
    431 	case '\v':
    432 		outchar('v');
    433 		break;
    434 	case '\a':
    435 		outchar('a');
    436 		break;
    437 	default:
    438 		outchar((char)((((unsigned char)c >> 6) & 07) + '0'));
    439 		outchar((char)((((unsigned char)c >> 3) & 07) + '0'));
    440 		outchar((char)((c & 07) + '0'));
    441 		break;
    442 	}
    443 }
    444 
    445 /*
    446  * extracts potential format specifiers for printf() and scanf() and
    447  * writes them, enclosed in "" and quoted if necessary, to the output file
    448  */
    449 static void
    450 outfstrg(const char *cp)
    451 {
    452 
    453 	outchar('"');
    454 
    455 	char c = *cp++;
    456 	while (c != '\0') {
    457 
    458 		if (c != '%') {
    459 			c = *cp++;
    460 			continue;
    461 		}
    462 
    463 		outchar('%');
    464 		c = *cp++;
    465 
    466 		/* flags for printf and scanf and *-fieldwidth for printf */
    467 		while (c == '-' || c == '+' || c == ' ' ||
    468 		    c == '#' || c == '0' || c == '*') {
    469 			outchar(c);
    470 			c = *cp++;
    471 		}
    472 
    473 		/* numeric field width */
    474 		while (ch_isdigit(c)) {
    475 			outchar(c);
    476 			c = *cp++;
    477 		}
    478 
    479 		/* precision for printf */
    480 		if (c == '.') {
    481 			outchar(c);
    482 			c = *cp++;
    483 			if (c == '*') {
    484 				outchar(c);
    485 				c = *cp++;
    486 			} else {
    487 				while (ch_isdigit(c)) {
    488 					outchar(c);
    489 					c = *cp++;
    490 				}
    491 			}
    492 		}
    493 
    494 		/* h, l, L and q flags for printf and scanf */
    495 		if (c == 'h' || c == 'l' || c == 'L' || c == 'q') {
    496 			outchar(c);
    497 			c = *cp++;
    498 		}
    499 
    500 		/*
    501 		 * The last character. It is always written, so we can detect
    502 		 * invalid format specifiers.
    503 		 */
    504 		if (c != '\0') {
    505 			outqchar(c);
    506 			char oc = c;
    507 			c = *cp++;
    508 			/*
    509 			 * handle [ for scanf. [-] means that a minus sign was
    510 			 * found at an undefined position.
    511 			 */
    512 			if (oc == '[') {
    513 				if (c == '^')
    514 					c = *cp++;
    515 				if (c == ']')
    516 					c = *cp++;
    517 				bool first = true;
    518 				while (c != '\0' && c != ']') {
    519 					if (c == '-') {
    520 						if (!first && *cp != ']')
    521 							outchar(c);
    522 					}
    523 					first = false;
    524 					c = *cp++;
    525 				}
    526 				if (c == ']') {
    527 					outchar(c);
    528 					c = *cp++;
    529 				}
    530 			}
    531 		}
    532 	}
    533 
    534 	outchar('"');
    535 }
    536 
    537 /* writes a record if sym was used */
    538 void
    539 outusg(const sym_t *sym)
    540 {
    541 	if (ch_isdigit(sym->s_name[0]))	/* 00000000_tmp, from mktempsym */
    542 		return;
    543 
    544 	outint(csrc_pos.p_line);
    545 	outchar('u');		/* used */
    546 	outint(get_filename_id(curr_pos.p_file));
    547 	outchar('.');
    548 	outint(curr_pos.p_line);
    549 	outchar('x');		/* separate the two numbers */
    550 	outname(sym->s_name);
    551 	outchar('\n');
    552 }
    553