scan.l revision 1.128 1 %{
2 /* $NetBSD: scan.l,v 1.128 2021/01/18 17:54:50 rillig Exp $ */
3
4 /*
5 * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved.
6 * Copyright (c) 1994, 1995 Jochen Pohl
7 * All Rights Reserved.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed by Jochen Pohl for
20 * The NetBSD Project.
21 * 4. The name of the author may not be used to endorse or promote products
22 * derived from this software without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
25 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
26 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
27 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
28 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
29 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
30 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
31 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
32 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
33 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
34 */
35
36 #include <sys/cdefs.h>
37 #if defined(__RCSID) && !defined(lint)
38 __RCSID("$NetBSD: scan.l,v 1.128 2021/01/18 17:54:50 rillig Exp $");
39 #endif
40
41 #include <ctype.h>
42 #include <errno.h>
43 #include <float.h>
44 #include <limits.h>
45 #include <math.h>
46 #include <stdlib.h>
47 #include <string.h>
48
49 #include "lint1.h"
50 #include "cgram.h"
51
52 #define CHAR_MASK ((int)(~(~0U << CHAR_SIZE)))
53
54 /* Current position (it's also updated when an included file is parsed) */
55 pos_t curr_pos = { 1, "", 0 };
56
57 /*
58 * Current position in C source (not updated when an included file is
59 * parsed).
60 */
61 pos_t csrc_pos = { 1, "", 0 };
62
63 /* Are we parsing a gcc attribute? */
64 bool attron;
65
66 bool in_system_header = false;
67
68 static void incline(void);
69 static void badchar(int);
70 static sbuf_t *allocsb(void);
71 static void freesb(sbuf_t *);
72 static int inpc(void);
73 static int hash(const char *);
74 static sym_t *search(sbuf_t *);
75 static int name(void);
76 static int keyw(sym_t *);
77 static int icon(int);
78 static int fcon(void);
79 static int operator(int, op_t);
80 static int ccon(void);
81 static int wccon(void);
82 static int getescc(int);
83 static void directive(void);
84 static void comment(void);
85 static void slashslashcomment(void);
86 static int string(void);
87 static int wcstrg(void);
88
89 %}
90
91
92 L [_A-Za-z]
93 D [0-9]
94 NZD [1-9]
95 BD [0-1]
96 OD [0-7]
97 HD [0-9A-Fa-f]
98 EX ([eE][+-]?[0-9]+)
99 HX (p[+-]?[0-9A-Fa-f]+)
100 TL ([fFlL]?[i]?)
101
102 %option nounput
103
104 %%
105
106 {L}({L}|{D})* return name();
107 0[bB]{BD}+[lLuU]* return icon(2);
108 0{OD}*[lLuU]* return icon(8);
109 {NZD}{D}*[lLuU]* return icon(10);
110 0[xX]{HD}+[lLuU]* return icon(16);
111 {D}+\.{D}*{EX}?{TL} |
112 {D}+{EX}{TL} |
113 0[xX]{HD}+\.{HD}*{HX}{TL} |
114 0[xX]{HD}+{HX}{TL} |
115 \.{D}+{EX}?{TL} return fcon();
116 "=" return operator(T_ASSIGN, NOOP);
117 "*=" return operator(T_OPASSIGN, MULASS);
118 "/=" return operator(T_OPASSIGN, DIVASS);
119 "%=" return operator(T_OPASSIGN, MODASS);
120 "+=" return operator(T_OPASSIGN, ADDASS);
121 "-=" return operator(T_OPASSIGN, SUBASS);
122 "<<=" return operator(T_OPASSIGN, SHLASS);
123 ">>=" return operator(T_OPASSIGN, SHRASS);
124 "&=" return operator(T_OPASSIGN, ANDASS);
125 "^=" return operator(T_OPASSIGN, XORASS);
126 "|=" return operator(T_OPASSIGN, ORASS);
127 "||" return operator(T_LOGOR, NOOP);
128 "&&" return operator(T_LOGAND, NOOP);
129 "|" return operator(T_BITOR, NOOP);
130 "&" return operator(T_AMPER, NOOP);
131 "^" return operator(T_XOR, NOOP);
132 "==" return operator(T_EQUALITY, EQ);
133 "!=" return operator(T_EQUALITY, NE);
134 "<" return operator(T_RELATIONAL, LT);
135 ">" return operator(T_RELATIONAL, GT);
136 "<=" return operator(T_RELATIONAL, LE);
137 ">=" return operator(T_RELATIONAL, GE);
138 "<<" return operator(T_SHIFT, SHL);
139 ">>" return operator(T_SHIFT, SHR);
140 "++" return operator(T_INCDEC, INC);
141 "--" return operator(T_INCDEC, DEC);
142 "->" return operator(T_MEMBACC, ARROW);
143 "." return operator(T_MEMBACC, POINT);
144 "+" return operator(T_ADDITIVE, PLUS);
145 "-" return operator(T_ADDITIVE, MINUS);
146 "*" return operator(T_ASTERISK, NOOP);
147 "/" return operator(T_MULTIPLICATIVE, DIV);
148 "%" return operator(T_MULTIPLICATIVE, MOD);
149 "!" return operator(T_UNARY, NOT);
150 "~" return operator(T_UNARY, COMPL);
151 "\"" return string();
152 "L\"" return wcstrg();
153 ";" return T_SEMI;
154 "{" return T_LBRACE;
155 "}" return T_RBRACE;
156 "," return T_COMMA;
157 ":" return T_COLON;
158 "?" return T_QUEST;
159 "[" return T_LBRACK;
160 "]" return T_RBRACK;
161 "(" return T_LPAREN;
162 ")" return T_RPAREN;
163 "..." return T_ELLIPSIS;
164 "'" return ccon();
165 "L'" return wccon();
166 ^#.*$ directive();
167 \n incline();
168 \t|" "|\f|\v ;
169 "/*" comment();
170 "//" slashslashcomment();
171 . badchar(yytext[0]);
172
173 %%
174
175 static void
176 incline(void)
177 {
178 curr_pos.p_line++;
179 curr_pos.p_uniq = 0;
180 #ifdef DEBUG
181 printf("parsing %s:%d\n", curr_pos.p_file, curr_pos.p_line);
182 #endif
183 if (curr_pos.p_file == csrc_pos.p_file) {
184 csrc_pos.p_line++;
185 csrc_pos.p_uniq = 0;
186 }
187 }
188
189 static void
190 badchar(int c)
191 {
192
193 /* unknown character \%o */
194 error(250, c);
195 }
196
197 /*
198 * Keywords.
199 * During initialisation they are written to the symbol table.
200 */
201 static struct kwtab {
202 const char *kw_name; /* keyword */
203 int kw_token; /* token returned by yylex() */
204 scl_t kw_scl; /* storage class if kw_token T_SCLASS */
205 tspec_t kw_tspec; /* type spec. if kw_token
206 * T_TYPE or T_STRUCT_OR_UNION */
207 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */
208 bool kw_c89 : 1; /* C89 keyword */
209 bool kw_c99 : 1; /* C99 keyword */
210 bool kw_gcc : 1; /* GCC keyword */
211 bool kw_attr : 1; /* GCC attribute, keyword */
212 u_int kw_deco : 3; /* 1 = name, 2 = __name, 4 = __name__ */
213 } kwtab[] = {
214 #ifdef INT128_SIZE
215 { "__int128_t", T_TYPE, 0, INT128, 0, 0,1,0,0,1 },
216 { "__uint128_t",T_TYPE, 0, UINT128,0, 0,1,0,0,1 },
217 #endif
218 { "__thread", T_QUAL, 0, 0, THREAD, 0,0,1,0,1 },
219 { "_Alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,1 },
220 { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 },
221 { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 },
222 { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 },
223 { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 },
224 { "_Thread_local",T_QUAL, 0, 0, THREAD, 0,1,0,0,1 },
225 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 },
226 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 },
227 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 },
228 { "alloc_size", T_AT_ALLOC_SIZE,0, 0, 0, 0,0,1,1,5 },
229 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 },
230 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 },
231 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 },
232 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 },
233 { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 },
234 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 },
235 { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 },
236 { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 },
237 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 },
238 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 },
239 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 },
240 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 },
241 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 },
242 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 },
243 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 },
244 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 },
245 { "destructor", T_AT_DESTRUCTOR,0, 0, 0, 0,0,1,1,5 },
246 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 },
247 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 },
248 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 },
249 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 },
250 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 },
251 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 },
252 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 },
253 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 },
254 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 },
255 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 },
256 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 },
257 { "gnu_printf", T_AT_FORMAT_GNU_PRINTF,0,0, 0, 0,0,1,1,5 },
258 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 },
259 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 },
260 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 },
261 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 },
262 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 },
263 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 },
264 { "malloc", T_AT_MALLOC, 0, 0, 0, 0,0,1,1,5 },
265 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 },
266 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 },
267 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 },
268 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION,
269 0, 0, 0, 0,0,1,1,5 },
270 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 },
271 { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 },
272 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 },
273 { "nothrow", T_AT_NOTHROW, 0, 0, 0, 0,0,1,1,5 },
274 { "optimize", T_AT_OPTIMIZE, 0, 0, 0, 0,0,1,1,5 },
275 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 },
276 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 },
277 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 },
278 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 },
279 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 },
280 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 },
281 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 },
282 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 },
283 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 },
284 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 },
285 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 },
286 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 },
287 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 },
288 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 },
289 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 },
290 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 },
291 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 },
292 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 },
293 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 },
294 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 },
295 { "struct", T_STRUCT_OR_UNION, 0, STRUCT, 0, 0,0,0,0,1 },
296 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 },
297 { "symbolrename", T_SYMBOLRENAME,0, 0, 0, 0,0,0,0,2 },
298 { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 },
299 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 },
300 { "tls_model", T_AT_TLS_MODEL, 0, 0, 0, 0,0,1,1,5 },
301 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 },
302 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 },
303 { "union", T_STRUCT_OR_UNION, 0, UNION, 0, 0,0,0,0,1 },
304 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 },
305 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 },
306 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 },
307 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 },
308 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 },
309 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 },
310 { "warn_unused_result", T_AT_WARN_UNUSED_RESULT, 0, 0, 0, 0,0,1,1,5 },
311 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 },
312 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 },
313 { NULL, 0, 0, 0, 0, 0,0,0,0,0 }
314 };
315
316 /* Symbol table */
317 static sym_t *symtab[HSHSIZ1];
318
319 /* bit i of the entry with index i is set */
320 uint64_t qbmasks[64];
321
322 /* least significant i bits are set in the entry with index i */
323 uint64_t qlmasks[64 + 1];
324
325 /* least significant i bits are not set in the entry with index i */
326 uint64_t qumasks[64 + 1];
327
328 /* free list for sbuf structures */
329 static sbuf_t *sbfrlst;
330
331 /* Typ of next expected symbol */
332 symt_t symtyp;
333
334
335 static void
336 add_keyword(struct kwtab *kw, int deco)
337 {
338 sym_t *sym;
339 size_t h;
340 char buf[256];
341 const char *name;
342
343 if (!(kw->kw_deco & deco))
344 return;
345
346 switch (deco) {
347 case 1:
348 name = kw->kw_name;
349 break;
350 case 2:
351 snprintf(buf, sizeof(buf), "__%s", kw->kw_name);
352 name = strdup(buf);
353 break;
354 default:
355 lint_assert(deco == 4);
356 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name);
357 name = strdup(buf);
358 break;
359 }
360
361 if (name == NULL)
362 err(1, "Can't init symbol table");
363
364 sym = getblk(sizeof (sym_t));
365 sym->s_name = name;
366 sym->s_keyword = kw;
367 sym->s_value.v_quad = kw->kw_token;
368 if (kw->kw_token == T_TYPE || kw->kw_token == T_STRUCT_OR_UNION) {
369 sym->s_tspec = kw->kw_tspec;
370 } else if (kw->kw_token == T_SCLASS) {
371 sym->s_scl = kw->kw_scl;
372 } else if (kw->kw_token == T_QUAL) {
373 sym->s_tqual = kw->kw_tqual;
374 }
375 h = hash(sym->s_name);
376 if ((sym->s_link = symtab[h]) != NULL)
377 symtab[h]->s_rlink = &sym->s_link;
378 sym->s_rlink = &symtab[h];
379 symtab[h] = sym;
380 }
381
382 /*
383 * All keywords are written to the symbol table. This saves us looking
384 * in a extra table for each name we found.
385 */
386 void
387 initscan(void)
388 {
389 struct kwtab *kw;
390 size_t i;
391 uint64_t uq;
392
393 for (kw = kwtab; kw->kw_name != NULL; kw++) {
394 if ((kw->kw_c89 || kw->kw_c99) && tflag)
395 continue;
396 if (kw->kw_c99 && !(Sflag || gflag))
397 continue;
398 if (kw->kw_gcc && !gflag)
399 continue;
400 add_keyword(kw, 1);
401 add_keyword(kw, 2);
402 add_keyword(kw, 4);
403 }
404
405 /* initialize bit-masks for quads */
406 for (i = 0; i < 64; i++) {
407 qbmasks[i] = (uint64_t)1 << i;
408 uq = ~(uint64_t)0 << i;
409 qumasks[i] = uq;
410 qlmasks[i] = ~uq;
411 }
412 qumasks[i] = 0;
413 qlmasks[i] = ~(uint64_t)0;
414 }
415
416 /*
417 * Get a free sbuf structure, if possible from the free list
418 */
419 static sbuf_t *
420 allocsb(void)
421 {
422 sbuf_t *sb;
423
424 if ((sb = sbfrlst) != NULL) {
425 sbfrlst = sb->sb_next;
426 #ifdef BLKDEBUG
427 (void)memset(sb, 0, sizeof (*sb));
428 #else
429 sb->sb_next = NULL;
430 #endif
431 } else {
432 sb = xmalloc(sizeof (sbuf_t));
433 (void)memset(sb, 0, sizeof (*sb));
434 }
435 return sb;
436 }
437
438 /*
439 * Put a sbuf structure to the free list
440 */
441 static void
442 freesb(sbuf_t *sb)
443 {
444
445 (void)memset(sb, ZERO, sizeof (*sb));
446 sb->sb_next = sbfrlst;
447 sbfrlst = sb;
448 }
449
450 /*
451 * Read a character and ensure that it is positive (except EOF).
452 * Increment line count(s) if necessary.
453 */
454 static int
455 inpc(void)
456 {
457 int c;
458
459 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
460 incline();
461 return c;
462 }
463
464 static int
465 hash(const char *s)
466 {
467 u_int v;
468 const u_char *us;
469
470 v = 0;
471 for (us = (const u_char *)s; *us != '\0'; us++) {
472 v = (v << sizeof (v)) + *us;
473 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
474 }
475 return v % HSHSIZ1;
476 }
477
478 /*
479 * Lex has found a letter followed by zero or more letters or digits.
480 * It looks for a symbol in the symbol table with the same name. This
481 * symbol must either be a keyword or a symbol of the type required by
482 * symtyp (label, member, tag, ...).
483 *
484 * If it is a keyword, the token is returned. In some cases it is described
485 * more deeply by data written to yylval.
486 *
487 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
488 * is stored in yylval. This struct contains the name of the symbol, its
489 * length and hash value. If there is already a symbol of the same name
490 * and type in the symbol table, the sbuf struct also contains a pointer
491 * to the symbol table entry.
492 */
493 static int
494 name(void)
495 {
496 char *s;
497 sbuf_t *sb;
498 sym_t *sym;
499 int tok;
500
501 sb = allocsb();
502 sb->sb_name = yytext;
503 sb->sb_len = yyleng;
504 sb->sb_hash = hash(yytext);
505 if ((sym = search(sb)) != NULL && sym->s_keyword) {
506 freesb(sb);
507 return keyw(sym);
508 }
509
510 sb->sb_sym = sym;
511
512 if (sym != NULL) {
513 lint_assert(blklev >= sym->s_blklev);
514 sb->sb_name = sym->s_name;
515 sb->sb_len = strlen(sym->s_name);
516 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
517 } else {
518 s = getblk(yyleng + 1);
519 (void)memcpy(s, yytext, yyleng + 1);
520 sb->sb_name = s;
521 sb->sb_len = yyleng;
522 tok = T_NAME;
523 }
524
525 yylval.y_sb = sb;
526 return tok;
527 }
528
529 static sym_t *
530 search(sbuf_t *sb)
531 {
532 sym_t *sym;
533
534 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
535 if (strcmp(sym->s_name, sb->sb_name) == 0) {
536 if (sym->s_keyword) {
537 struct kwtab *kw = sym->s_keyword;
538 if (!kw->kw_attr || attron)
539 return sym;
540 } else if (!attron && sym->s_kind == symtyp)
541 return sym;
542 }
543 }
544
545 return NULL;
546 }
547
548 static int
549 keyw(sym_t *sym)
550 {
551 int t;
552
553 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
554 yylval.y_scl = sym->s_scl;
555 } else if (t == T_TYPE || t == T_STRUCT_OR_UNION) {
556 yylval.y_tspec = sym->s_tspec;
557 } else if (t == T_QUAL) {
558 yylval.y_tqual = sym->s_tqual;
559 }
560 return t;
561 }
562
563 /*
564 * Convert a string representing an integer into internal representation.
565 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
566 */
567 static int
568 icon(int base)
569 {
570 int l_suffix, u_suffix;
571 int len;
572 const char *cp;
573 char c, *eptr;
574 tspec_t typ;
575 bool ansiu;
576 #ifdef TARG_INT128_MAX
577 __uint128_t uq = 0;
578 static tspec_t contypes[2][4] = {
579 { INT, LONG, QUAD, INT128, },
580 { UINT, ULONG, UQUAD, UINT128, }
581 };
582 #else
583 uint64_t uq = 0;
584 static tspec_t contypes[2][3] = {
585 { INT, LONG, QUAD, },
586 { UINT, ULONG, UQUAD, }
587 };
588 #endif
589
590 cp = yytext;
591 len = yyleng;
592
593 /* skip 0[xX] or 0[bB] */
594 if (base == 16 || base == 2) {
595 cp += 2;
596 len -= 2;
597 }
598
599 /* read suffixes */
600 l_suffix = u_suffix = 0;
601 for ( ; ; ) {
602 if ((c = cp[len - 1]) == 'l' || c == 'L') {
603 l_suffix++;
604 } else if (c == 'u' || c == 'U') {
605 u_suffix++;
606 } else {
607 break;
608 }
609 len--;
610 }
611 if (l_suffix > 2 || u_suffix > 1) {
612 /* malformed integer constant */
613 warning(251);
614 if (l_suffix > 2)
615 l_suffix = 2;
616 if (u_suffix > 1)
617 u_suffix = 1;
618 }
619 if (tflag && u_suffix != 0) {
620 /* suffix U is illegal in traditional C */
621 warning(97);
622 }
623 typ = contypes[u_suffix][l_suffix];
624
625 errno = 0;
626
627 uq = strtouq(cp, &eptr, base);
628 lint_assert(eptr == cp + len);
629 if (errno != 0)
630 /* integer constant out of range */
631 warning(252);
632
633 /*
634 * If the value is too big for the current type, we must choose
635 * another type.
636 */
637 ansiu = false;
638 switch (typ) {
639 case INT:
640 if (uq <= TARG_INT_MAX) {
641 /* ok */
642 } else if (uq <= TARG_UINT_MAX && base != 10) {
643 typ = UINT;
644 } else if (uq <= TARG_LONG_MAX) {
645 typ = LONG;
646 } else {
647 typ = ULONG;
648 if (uq > TARG_ULONG_MAX) {
649 /* integer constant out of range */
650 warning(252);
651 }
652 }
653 if (typ == UINT || typ == ULONG) {
654 if (tflag) {
655 typ = LONG;
656 } else if (!sflag) {
657 /*
658 * Remember that the constant is unsigned
659 * only in ANSI C
660 */
661 ansiu = true;
662 }
663 }
664 break;
665 case UINT:
666 if (uq > TARG_UINT_MAX) {
667 typ = ULONG;
668 if (uq > TARG_ULONG_MAX) {
669 /* integer constant out of range */
670 warning(252);
671 }
672 }
673 break;
674 case LONG:
675 if (uq > TARG_LONG_MAX && !tflag) {
676 typ = ULONG;
677 if (!sflag)
678 ansiu = true;
679 if (uq > TARG_ULONG_MAX) {
680 /* integer constant out of range */
681 warning(252);
682 }
683 }
684 break;
685 case ULONG:
686 if (uq > TARG_ULONG_MAX) {
687 /* integer constant out of range */
688 warning(252);
689 }
690 break;
691 case QUAD:
692 if (uq > TARG_QUAD_MAX && !tflag) {
693 typ = UQUAD;
694 if (!sflag)
695 ansiu = true;
696 }
697 break;
698 case UQUAD:
699 if (uq > TARG_UQUAD_MAX) {
700 /* integer constant out of range */
701 warning(252);
702 }
703 break;
704 #ifdef INT128_SIZE
705 case INT128:
706 #ifdef TARG_INT128_MAX
707 if (uq > TARG_INT128_MAX && !tflag) {
708 typ = UINT128;
709 if (!sflag)
710 ansiu = true;
711 }
712 #endif
713 break;
714 case UINT128:
715 #ifdef TARG_INT128_MAX
716 if (uq > TARG_UINT128_MAX) {
717 /* integer constant out of range */
718 warning(252);
719 }
720 #endif
721 break;
722 #endif
723 /* LINTED206: (enumeration values not handled in switch) */
724 case STRUCT:
725 case VOID:
726 case LDOUBLE:
727 case FUNC:
728 case ARRAY:
729 case PTR:
730 case ENUM:
731 case UNION:
732 case SIGNED:
733 case NOTSPEC:
734 case DOUBLE:
735 case FLOAT:
736 case USHORT:
737 case SHORT:
738 case UCHAR:
739 case SCHAR:
740 case CHAR:
741 case BOOL:
742 case UNSIGN:
743 case FCOMPLEX:
744 case DCOMPLEX:
745 case LCOMPLEX:
746 case COMPLEX:
747 break;
748 }
749
750 uq = (uint64_t)xsign((int64_t)uq, typ, -1);
751
752 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
753 yylval.y_val->v_ansiu = ansiu;
754 yylval.y_val->v_quad = (int64_t)uq;
755
756 return T_CON;
757 }
758
759 /*
760 * Returns 1 if t is a signed type and the value is negative.
761 *
762 * len is the number of significant bits. If len is -1, len is set
763 * to the width of type t.
764 */
765 int
766 sign(int64_t q, tspec_t t, int len)
767 {
768
769 if (t == PTR || is_uinteger(t))
770 return 0;
771 return msb(q, t, len);
772 }
773
774 int
775 msb(int64_t q, tspec_t t, int len)
776 {
777
778 if (len <= 0)
779 len = size(t);
780 return (q & qbmasks[len - 1]) != 0;
781 }
782
783 /*
784 * Extends the sign of q.
785 */
786 int64_t
787 xsign(int64_t q, tspec_t t, int len)
788 {
789
790 if (len <= 0)
791 len = size(t);
792
793 if (t == PTR || is_uinteger(t) || !sign(q, t, len)) {
794 q &= qlmasks[len];
795 } else {
796 q |= qumasks[len];
797 }
798 return q;
799 }
800
801 /*
802 * Convert a string representing a floating point value into its integral
803 * representation. Type and value are returned in yylval. fcon()
804 * (and yylex()) returns T_CON.
805 * XXX Currently it is not possible to convert constants of type
806 * long double which are greater than DBL_MAX.
807 */
808 static int
809 fcon(void)
810 {
811 const char *cp;
812 int len;
813 tspec_t typ;
814 char c, *eptr;
815 double d;
816 float f = 0;
817
818 cp = yytext;
819 len = yyleng;
820
821 if (cp[len - 1] == 'i') {
822 /* imaginary, do nothing for now */
823 len--;
824 }
825 if ((c = cp[len - 1]) == 'f' || c == 'F') {
826 typ = FLOAT;
827 len--;
828 } else if (c == 'l' || c == 'L') {
829 typ = LDOUBLE;
830 len--;
831 } else {
832 if (c == 'd' || c == 'D')
833 len--;
834 typ = DOUBLE;
835 }
836
837 if (tflag && typ != DOUBLE) {
838 /* suffixes F and L are illegal in traditional C */
839 warning(98);
840 }
841
842 errno = 0;
843 d = strtod(cp, &eptr);
844 if (eptr != cp + len) {
845 switch (*eptr) {
846 /*
847 * XXX: non-native non-current strtod() may not handle hex
848 * floats, ignore the rest if we find traces of hex float
849 * syntax...
850 */
851 case 'p':
852 case 'P':
853 case 'x':
854 case 'X':
855 d = 0;
856 errno = 0;
857 break;
858 default:
859 LERROR("fcon(%s->%s)", cp, eptr);
860 }
861 }
862 if (errno != 0)
863 /* floating-point constant out of range */
864 warning(248);
865
866 if (typ == FLOAT) {
867 f = (float)d;
868 if (!finite(f)) {
869 /* floating-point constant out of range */
870 warning(248);
871 f = f > 0 ? FLT_MAX : -FLT_MAX;
872 }
873 }
874
875 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
876 if (typ == FLOAT) {
877 yylval.y_val->v_ldbl = f;
878 } else {
879 yylval.y_val->v_ldbl = d;
880 }
881
882 return T_CON;
883 }
884
885 static int
886 operator(int t, op_t o)
887 {
888
889 yylval.y_op = o;
890 return t;
891 }
892
893 /*
894 * Called if lex found a leading \'.
895 */
896 static int
897 ccon(void)
898 {
899 size_t n;
900 int val, c;
901 char cv;
902
903 n = 0;
904 val = 0;
905 while ((c = getescc('\'')) >= 0) {
906 val = (val << CHAR_SIZE) + c;
907 n++;
908 }
909 if (c == -2) {
910 /* unterminated character constant */
911 error(253);
912 } else {
913 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
914 /* too many characters in character constant */
915 error(71);
916 } else if (n > 1) {
917 /* multi-character character constant */
918 warning(294);
919 } else if (n == 0) {
920 /* empty character constant */
921 error(73);
922 }
923 }
924 if (n == 1) {
925 cv = (char)val;
926 val = cv;
927 }
928
929 yylval.y_val = xcalloc(1, sizeof (val_t));
930 yylval.y_val->v_tspec = INT;
931 yylval.y_val->v_quad = val;
932
933 return T_CON;
934 }
935
936 /*
937 * Called if lex found a leading L\'
938 */
939 static int
940 wccon(void)
941 {
942 static char buf[MB_LEN_MAX + 1];
943 size_t i;
944 int c;
945 wchar_t wc;
946
947 i = 0;
948 while ((c = getescc('\'')) >= 0) {
949 if (i < MB_CUR_MAX)
950 buf[i] = (char)c;
951 i++;
952 }
953
954 wc = 0;
955
956 if (c == -2) {
957 /* unterminated character constant */
958 error(253);
959 } else if (c == 0) {
960 /* empty character constant */
961 error(73);
962 } else {
963 if (i > MB_CUR_MAX) {
964 i = MB_CUR_MAX;
965 /* too many characters in character constant */
966 error(71);
967 } else {
968 buf[i] = '\0';
969 (void)mbtowc(NULL, NULL, 0);
970 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
971 /* invalid multibyte character */
972 error(291);
973 }
974 }
975
976 yylval.y_val = xcalloc(1, sizeof (val_t));
977 yylval.y_val->v_tspec = WCHAR;
978 yylval.y_val->v_quad = wc;
979
980 return T_CON;
981 }
982
983 /*
984 * Read a character which is part of a character constant or of a string
985 * and handle escapes.
986 *
987 * The Argument is the character which delimits the character constant or
988 * string.
989 *
990 * Returns -1 if the end of the character constant or string is reached,
991 * -2 if the EOF is reached, and the character otherwise.
992 */
993 static int
994 getescc(int d)
995 {
996 static int pbc = -1;
997 int n, c, v;
998
999 if (pbc == -1) {
1000 c = inpc();
1001 } else {
1002 c = pbc;
1003 pbc = -1;
1004 }
1005 if (c == d)
1006 return -1;
1007 switch (c) {
1008 case '\n':
1009 if (tflag) {
1010 /* newline in string or char constant */
1011 error(254);
1012 return -2;
1013 }
1014 return c;
1015 case EOF:
1016 return -2;
1017 case '\\':
1018 switch (c = inpc()) {
1019 case '"':
1020 if (tflag && d == '\'')
1021 /* \" inside character constants undef... */
1022 warning(262);
1023 return '"';
1024 case '\'':
1025 return '\'';
1026 case '?':
1027 if (tflag)
1028 /* \? undefined in traditional C */
1029 warning(263);
1030 return '?';
1031 case '\\':
1032 return '\\';
1033 case 'a':
1034 if (tflag)
1035 /* \a undefined in traditional C */
1036 warning(81);
1037 return '\a';
1038 case 'b':
1039 return '\b';
1040 case 'f':
1041 return '\f';
1042 case 'n':
1043 return '\n';
1044 case 'r':
1045 return '\r';
1046 case 't':
1047 return '\t';
1048 case 'v':
1049 if (tflag)
1050 /* \v undefined in traditional C */
1051 warning(264);
1052 return '\v';
1053 case '8': case '9':
1054 /* bad octal digit %c */
1055 warning(77, c);
1056 /* FALLTHROUGH */
1057 case '0': case '1': case '2': case '3':
1058 case '4': case '5': case '6': case '7':
1059 n = 3;
1060 v = 0;
1061 do {
1062 v = (v << 3) + (c - '0');
1063 c = inpc();
1064 } while (--n && isdigit(c) && (tflag || c <= '7'));
1065 if (tflag && n > 0 && isdigit(c))
1066 /* bad octal digit %c */
1067 warning(77, c);
1068 pbc = c;
1069 if (v > TARG_UCHAR_MAX) {
1070 /* character escape does not fit in character */
1071 warning(76);
1072 v &= CHAR_MASK;
1073 }
1074 return v;
1075 case 'x':
1076 if (tflag)
1077 /* \x undefined in traditional C */
1078 warning(82);
1079 v = 0;
1080 n = 0;
1081 while ((c = inpc()) >= 0 && isxdigit(c)) {
1082 c = isdigit(c) ?
1083 c - '0' : toupper(c) - 'A' + 10;
1084 v = (v << 4) + c;
1085 if (n >= 0) {
1086 if ((v & ~CHAR_MASK) != 0) {
1087 /* overflow in hex escape */
1088 warning(75);
1089 n = -1;
1090 } else {
1091 n++;
1092 }
1093 }
1094 }
1095 pbc = c;
1096 if (n == 0) {
1097 /* no hex digits follow \x */
1098 error(74);
1099 } if (n == -1) {
1100 v &= CHAR_MASK;
1101 }
1102 return v;
1103 case '\n':
1104 return getescc(d);
1105 case EOF:
1106 return -2;
1107 default:
1108 if (isprint(c)) {
1109 /* dubious escape \%c */
1110 warning(79, c);
1111 } else {
1112 /* dubious escape \%o */
1113 warning(80, c);
1114 }
1115 }
1116 }
1117 return c;
1118 }
1119
1120 /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */
1121 static void
1122 parse_line_directive_flags(const char *p)
1123 {
1124
1125 in_system_header = false;
1126
1127 while (ch_isspace(*p))
1128 p++;
1129 while (ch_isdigit(*p)) {
1130 if (*p == '3' && !ch_isdigit(p[1]))
1131 in_system_header = true;
1132 p++;
1133 while (ch_isspace(*p))
1134 p++;
1135 }
1136
1137 #if 0
1138 if (c != '\0')
1139 warning("extra character(s) after directive");
1140 #endif
1141 }
1142
1143 /*
1144 * Called for preprocessor directives. Currently implemented are:
1145 * # lineno
1146 * # lineno "filename"
1147 */
1148 static void
1149 directive(void)
1150 {
1151 const char *cp, *fn;
1152 char c, *eptr;
1153 size_t fnl;
1154 long ln;
1155 static bool first = true;
1156
1157 /* Go to first non-whitespace after # */
1158 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1159 continue;
1160
1161 if (!ch_isdigit(c)) {
1162 if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6]))
1163 return;
1164 error:
1165 /* undefined or invalid # directive */
1166 warning(255);
1167 return;
1168 }
1169 ln = strtol(--cp, &eptr, 10);
1170 if (cp == eptr)
1171 goto error;
1172 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1173 goto error;
1174 while ((c = *cp++) == ' ' || c == '\t')
1175 continue;
1176 if (c != '\0') {
1177 if (c != '"')
1178 goto error;
1179 fn = cp;
1180 while ((c = *cp) != '"' && c != '\0')
1181 cp++;
1182 if (c != '"')
1183 goto error;
1184 if ((fnl = cp++ - fn) > PATH_MAX)
1185 goto error;
1186 parse_line_directive_flags(cp);
1187
1188 /* empty string means stdin */
1189 if (fnl == 0) {
1190 fn = "{standard input}";
1191 fnl = 16; /* strlen (fn) */
1192 }
1193 curr_pos.p_file = fnnalloc(fn, fnl);
1194 /*
1195 * If this is the first directive, the name is the name
1196 * of the C source file as specified at the command line.
1197 * It is written to the output file.
1198 */
1199 if (first) {
1200 csrc_pos.p_file = curr_pos.p_file;
1201 outsrc(fnxform(curr_pos.p_file,
1202 strlen(curr_pos.p_file)));
1203 first = false;
1204 }
1205 }
1206 curr_pos.p_line = (int)ln - 1;
1207 curr_pos.p_uniq = 0;
1208 if (curr_pos.p_file == csrc_pos.p_file) {
1209 csrc_pos.p_line = (int)ln - 1;
1210 csrc_pos.p_uniq = 0;
1211 }
1212 }
1213
1214 /*
1215 * Handle lint comments. Following comments are currently understood:
1216 * ARGSUSEDn
1217 * BITFIELDTYPE
1218 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1219 * FALLTHRU FALLTHROUGH
1220 * LINTLIBRARY
1221 * LINTEDn NOSTRICTn
1222 * LONGLONG
1223 * NOTREACHED
1224 * PRINTFLIKEn
1225 * PROTOLIB
1226 * SCANFLIKEn
1227 * VARARGSn
1228 * If one of this comments is recognized, the arguments, if any, are
1229 * parsed and a function which handles this comment is called.
1230 */
1231 static void
1232 comment(void)
1233 {
1234 int c, lc;
1235 static struct {
1236 const char *keywd;
1237 int arg;
1238 void (*func)(int);
1239 } keywtab[] = {
1240 { "ARGSUSED", 1, argsused },
1241 { "BITFIELDTYPE", 0, bitfieldtype },
1242 { "CONSTCOND", 0, constcond },
1243 { "CONSTANTCOND", 0, constcond },
1244 { "CONSTANTCONDITION", 0, constcond },
1245 { "FALLTHRU", 0, fallthru },
1246 { "FALLTHROUGH", 0, fallthru },
1247 { "LINTLIBRARY", 0, lintlib },
1248 { "LINTED", 1, linted },
1249 { "LONGLONG", 0, longlong },
1250 { "NOSTRICT", 1, linted },
1251 { "NOTREACHED", 0, notreach },
1252 { "PRINTFLIKE", 1, printflike },
1253 { "PROTOLIB", 1, protolib },
1254 { "SCANFLIKE", 1, scanflike },
1255 { "VARARGS", 1, varargs },
1256 };
1257 char keywd[32];
1258 char arg[32];
1259 size_t l, i;
1260 int a;
1261 bool eoc;
1262
1263 eoc = false;
1264
1265 /* Skip white spaces after the start of the comment */
1266 while ((c = inpc()) != EOF && isspace(c))
1267 continue;
1268
1269 /* Read the potential keyword to keywd */
1270 l = 0;
1271 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1272 keywd[l++] = (char)c;
1273 c = inpc();
1274 }
1275 keywd[l] = '\0';
1276
1277 /* look for the keyword */
1278 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1279 if (strcmp(keywtab[i].keywd, keywd) == 0)
1280 break;
1281 }
1282 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1283 goto skip_rest;
1284
1285 /* skip white spaces after the keyword */
1286 while (c != EOF && isspace(c))
1287 c = inpc();
1288
1289 /* read the argument, if the keyword accepts one and there is one */
1290 l = 0;
1291 if (keywtab[i].arg) {
1292 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1293 arg[l++] = (char)c;
1294 c = inpc();
1295 }
1296 }
1297 arg[l] = '\0';
1298 a = l != 0 ? atoi(arg) : -1;
1299
1300 /* skip white spaces after the argument */
1301 while (c != EOF && isspace(c))
1302 c = inpc();
1303
1304 if (c != '*' || (c = inpc()) != '/') {
1305 if (keywtab[i].func != linted)
1306 /* extra characters in lint comment */
1307 warning(257);
1308 } else {
1309 /*
1310 * remember that we have already found the end of the
1311 * comment
1312 */
1313 eoc = true;
1314 }
1315
1316 if (keywtab[i].func != NULL)
1317 (*keywtab[i].func)(a);
1318
1319 skip_rest:
1320 while (!eoc) {
1321 lc = c;
1322 if ((c = inpc()) == EOF) {
1323 /* unterminated comment */
1324 error(256);
1325 break;
1326 }
1327 if (lc == '*' && c == '/')
1328 eoc = true;
1329 }
1330 }
1331
1332 /*
1333 * Handle // style comments
1334 */
1335 static void
1336 slashslashcomment(void)
1337 {
1338 int c;
1339
1340 if (!Sflag && !gflag)
1341 /* %s C does not support // comments */
1342 gnuism(312, tflag ? "traditional" : "ANSI");
1343
1344 while ((c = inpc()) != EOF && c != '\n')
1345 continue;
1346 }
1347
1348 /*
1349 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1350 * clear_warn_flags() is called after function definitions and global and
1351 * local declarations and definitions. It is also called between
1352 * the controlling expression and the body of control statements
1353 * (if, switch, for, while).
1354 */
1355 void
1356 clear_warn_flags(void)
1357 {
1358
1359 lwarn = LWARN_ALL;
1360 quadflg = false;
1361 constcond_flag = false;
1362 }
1363
1364 /*
1365 * Strings are stored in a dynamically allocated buffer and passed
1366 * in yylval.y_xstrg to the parser. The parser or the routines called
1367 * by the parser are responsible for freeing this buffer.
1368 */
1369 static int
1370 string(void)
1371 {
1372 u_char *s;
1373 int c;
1374 size_t len, max;
1375 strg_t *strg;
1376
1377 s = xmalloc(max = 64);
1378
1379 len = 0;
1380 while ((c = getescc('"')) >= 0) {
1381 /* +1 to reserve space for a trailing NUL character */
1382 if (len + 1 == max)
1383 s = xrealloc(s, max *= 2);
1384 s[len++] = (char)c;
1385 }
1386 s[len] = '\0';
1387 if (c == -2)
1388 /* unterminated string constant */
1389 error(258);
1390
1391 strg = xcalloc(1, sizeof (strg_t));
1392 strg->st_tspec = CHAR;
1393 strg->st_len = len;
1394 strg->st_cp = s;
1395
1396 yylval.y_string = strg;
1397 return T_STRING;
1398 }
1399
1400 static int
1401 wcstrg(void)
1402 {
1403 char *s;
1404 int c, n;
1405 size_t i, wi;
1406 size_t len, max, wlen;
1407 wchar_t *ws;
1408 strg_t *strg;
1409
1410 s = xmalloc(max = 64);
1411 len = 0;
1412 while ((c = getescc('"')) >= 0) {
1413 /* +1 to save space for a trailing NUL character */
1414 if (len + 1 >= max)
1415 s = xrealloc(s, max *= 2);
1416 s[len++] = (char)c;
1417 }
1418 s[len] = '\0';
1419 if (c == -2)
1420 /* unterminated string constant */
1421 error(258);
1422
1423 /* get length of wide-character string */
1424 (void)mblen(NULL, 0);
1425 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1426 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1427 /* invalid multibyte character */
1428 error(291);
1429 break;
1430 }
1431 if (n == 0)
1432 n = 1;
1433 }
1434
1435 ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1436
1437 /* convert from multibyte to wide char */
1438 (void)mbtowc(NULL, NULL, 0);
1439 for (i = 0, wi = 0; i < len; i += n, wi++) {
1440 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1441 break;
1442 if (n == 0)
1443 n = 1;
1444 }
1445 ws[wi] = 0;
1446 free(s);
1447
1448 strg = xcalloc(1, sizeof (strg_t));
1449 strg->st_tspec = WCHAR;
1450 strg->st_len = wlen;
1451 strg->st_wcp = ws;
1452
1453 yylval.y_string = strg;
1454 return T_STRING;
1455 }
1456
1457 /*
1458 * As noted above the scanner does not create new symbol table entries
1459 * for symbols it cannot find in the symbol table. This is to avoid
1460 * putting undeclared symbols into the symbol table if a syntax error
1461 * occurs.
1462 *
1463 * getsym() is called as soon as it is probably ok to put the symbol to
1464 * the symbol table. This does not mean that it is not possible that
1465 * symbols are put to the symbol table which are than not completely
1466 * declared due to syntax errors. To avoid too many problems in this
1467 * case symbols get type int in getsym().
1468 *
1469 * XXX calls to getsym() should be delayed until decl1*() is called
1470 */
1471 sym_t *
1472 getsym(sbuf_t *sb)
1473 {
1474 dinfo_t *di;
1475 char *s;
1476 sym_t *sym;
1477
1478 sym = sb->sb_sym;
1479
1480 /*
1481 * During member declaration it is possible that name() looked
1482 * for symbols of type FVFT, although it should have looked for
1483 * symbols of type FTAG. Same can happen for labels. Both cases
1484 * are compensated here.
1485 */
1486 if (symtyp == FMEMBER || symtyp == FLABEL) {
1487 if (sym == NULL || sym->s_kind == FVFT)
1488 sym = search(sb);
1489 }
1490
1491 if (sym != NULL) {
1492 if (sym->s_kind != symtyp)
1493 LERROR("getsym(%d, %d)", sym->s_kind, symtyp);
1494 symtyp = FVFT;
1495 freesb(sb);
1496 return sym;
1497 }
1498
1499 /* create a new symbol table entry */
1500
1501 /* labels must always be allocated at level 1 (outhermost block) */
1502 if (symtyp == FLABEL) {
1503 sym = getlblk(1, sizeof (sym_t));
1504 s = getlblk(1, sb->sb_len + 1);
1505 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1506 sym->s_name = s;
1507 sym->s_blklev = 1;
1508 di = dcs;
1509 while (di->d_next != NULL && di->d_next->d_next != NULL)
1510 di = di->d_next;
1511 lint_assert(di->d_ctx == AUTO);
1512 } else {
1513 sym = getblk(sizeof (sym_t));
1514 sym->s_name = sb->sb_name;
1515 sym->s_blklev = blklev;
1516 di = dcs;
1517 }
1518
1519 UNIQUE_CURR_POS(sym->s_def_pos);
1520 if ((sym->s_kind = symtyp) != FLABEL)
1521 sym->s_type = gettyp(INT);
1522
1523 symtyp = FVFT;
1524
1525 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1526 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1527 sym->s_rlink = &symtab[sb->sb_hash];
1528 symtab[sb->sb_hash] = sym;
1529
1530 *di->d_ldlsym = sym;
1531 di->d_ldlsym = &sym->s_dlnxt;
1532
1533 freesb(sb);
1534 return sym;
1535 }
1536
1537 /*
1538 * Construct a temporary symbol. The symbol starts with a digit, so that
1539 * it is illegal.
1540 */
1541 sym_t *
1542 mktempsym(type_t *t)
1543 {
1544 static int n = 0;
1545 int h;
1546 char *s = getlblk(blklev, 64);
1547 sym_t *sym = getblk(sizeof (sym_t));
1548
1549 (void)snprintf(s, 64, "%.8d_tmp", n++);
1550 h = hash(s);
1551
1552 sym->s_name = s;
1553 sym->s_type = t;
1554 sym->s_blklev = blklev;
1555 sym->s_scl = AUTO;
1556 sym->s_kind = FVFT;
1557 sym->s_used = true;
1558 sym->s_set = true;
1559
1560 if ((sym->s_link = symtab[h]) != NULL)
1561 symtab[h]->s_rlink = &sym->s_link;
1562 sym->s_rlink = &symtab[h];
1563 symtab[h] = sym;
1564
1565 *dcs->d_ldlsym = sym;
1566 dcs->d_ldlsym = &sym->s_dlnxt;
1567
1568 return sym;
1569 }
1570
1571 /*
1572 * Remove a symbol forever from the symbol table. s_blklev
1573 * is set to -1 to avoid that the symbol will later be put
1574 * back to the symbol table.
1575 */
1576 void
1577 rmsym(sym_t *sym)
1578 {
1579
1580 if ((*sym->s_rlink = sym->s_link) != NULL)
1581 sym->s_link->s_rlink = sym->s_rlink;
1582 sym->s_blklev = -1;
1583 sym->s_link = NULL;
1584 }
1585
1586 /*
1587 * Remove a list of symbols declared at one level from the symbol
1588 * table.
1589 */
1590 void
1591 rmsyms(sym_t *syms)
1592 {
1593 sym_t *sym;
1594
1595 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1596 if (sym->s_blklev != -1) {
1597 if ((*sym->s_rlink = sym->s_link) != NULL)
1598 sym->s_link->s_rlink = sym->s_rlink;
1599 sym->s_link = NULL;
1600 sym->s_rlink = NULL;
1601 }
1602 }
1603 }
1604
1605 /*
1606 * Put a symbol into the symbol table
1607 */
1608 void
1609 inssym(int bl, sym_t *sym)
1610 {
1611 int h;
1612
1613 h = hash(sym->s_name);
1614 if ((sym->s_link = symtab[h]) != NULL)
1615 symtab[h]->s_rlink = &sym->s_link;
1616 sym->s_rlink = &symtab[h];
1617 symtab[h] = sym;
1618 sym->s_blklev = bl;
1619 lint_assert(sym->s_link == NULL ||
1620 sym->s_blklev >= sym->s_link->s_blklev);
1621 }
1622
1623 /*
1624 * Called at level 0 after syntax errors
1625 * Removes all symbols which are not declared at level 0 from the
1626 * symbol table. Also frees all memory which is not associated with
1627 * level 0.
1628 */
1629 void
1630 cleanup(void)
1631 {
1632 sym_t *sym, *nsym;
1633 int i;
1634
1635 for (i = 0; i < HSHSIZ1; i++) {
1636 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1637 nsym = sym->s_link;
1638 if (sym->s_blklev >= 1) {
1639 if ((*sym->s_rlink = nsym) != NULL)
1640 nsym->s_rlink = sym->s_rlink;
1641 }
1642 }
1643 }
1644
1645 for (i = mblklev; i > 0; i--)
1646 freelblk(i);
1647 }
1648
1649 /*
1650 * Create a new symbol with the name of an existing symbol.
1651 */
1652 sym_t *
1653 pushdown(sym_t *sym)
1654 {
1655 int h;
1656 sym_t *nsym;
1657
1658 h = hash(sym->s_name);
1659 nsym = getblk(sizeof (sym_t));
1660 lint_assert(sym->s_blklev <= blklev);
1661 nsym->s_name = sym->s_name;
1662 UNIQUE_CURR_POS(nsym->s_def_pos);
1663 nsym->s_kind = sym->s_kind;
1664 nsym->s_blklev = blklev;
1665
1666 if ((nsym->s_link = symtab[h]) != NULL)
1667 symtab[h]->s_rlink = &nsym->s_link;
1668 nsym->s_rlink = &symtab[h];
1669 symtab[h] = nsym;
1670
1671 *dcs->d_ldlsym = nsym;
1672 dcs->d_ldlsym = &nsym->s_dlnxt;
1673
1674 return nsym;
1675 }
1676
1677 /*
1678 * Free any dynamically allocated memory referenced by
1679 * the value stack or yylval.
1680 * The type of information in yylval is described by tok.
1681 */
1682 void
1683 freeyyv(void *sp, int tok)
1684 {
1685 if (tok == T_NAME || tok == T_TYPENAME) {
1686 sbuf_t *sb = *(sbuf_t **)sp;
1687 freesb(sb);
1688 } else if (tok == T_CON) {
1689 val_t *val = *(val_t **)sp;
1690 free(val);
1691 } else if (tok == T_STRING) {
1692 strg_t *strg = *(strg_t **)sp;
1693 if (strg->st_tspec == CHAR) {
1694 free(strg->st_cp);
1695 } else {
1696 lint_assert(strg->st_tspec == WCHAR);
1697 free(strg->st_wcp);
1698 }
1699 free(strg);
1700 }
1701 }
1702