scan.l revision 1.90 1 %{
2 /* $NetBSD: scan.l,v 1.90 2019/09/05 20:12:11 christos 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 1.90 2019/09/05 20:12:11 christos Exp $");
39 #endif
40
41 #include <stdlib.h>
42 #include <string.h>
43 #include <limits.h>
44 #include <float.h>
45 #include <ctype.h>
46 #include <errno.h>
47 #include <math.h>
48
49 #include "lint1.h"
50 #include "cgram.h"
51
52 #define CHAR_MASK ((int)(~(~0U << CHAR_BIT)))
53
54 /* Current position (its 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 int attron;
65
66 static void incline(void);
67 static void badchar(int);
68 static sbuf_t *allocsb(void);
69 static void freesb(sbuf_t *);
70 static int inpc(void);
71 static int hash(const char *);
72 static sym_t *search(sbuf_t *);
73 static int name(void);
74 static int keyw(sym_t *);
75 static int icon(int);
76 static int fcon(void);
77 static int operator(int, op_t);
78 static int ccon(void);
79 static int wccon(void);
80 static int getescc(int);
81 static void directive(void);
82 static void comment(void);
83 static void slashslashcomment(void);
84 static int string(void);
85 static int wcstrg(void);
86
87 %}
88
89
90 L [_A-Za-z]
91 D [0-9]
92 NZD [1-9]
93 BD [0-1]
94 OD [0-7]
95 HD [0-9A-Fa-f]
96 EX ([eE][+-]?[0-9]+)
97 HX (p[+-]?[0-9A-Fa-f]+)
98 TL ([fFlL]?[i]?)
99
100 %option nounput
101
102 %%
103
104 {L}({L}|{D})* return (name());
105 0[bB]{BD}+[lLuU]* return (icon(2));
106 0{OD}*[lLuU]* return (icon(8));
107 {NZD}{D}*[lLuU]* return (icon(10));
108 0[xX]{HD}+[lLuU]* return (icon(16));
109 {D}+\.{D}*{EX}?{TL} |
110 {D}+{EX}{TL} |
111 0[xX]{HD}+\.{HD}*{HX}{TL} |
112 0[xX]{HD}+{HX}{TL} |
113 \.{D}+{EX}?{TL} return (fcon());
114 "=" return (operator(T_ASSIGN, ASSIGN));
115 "*=" return (operator(T_OPASS, MULASS));
116 "/=" return (operator(T_OPASS, DIVASS));
117 "%=" return (operator(T_OPASS, MODASS));
118 "+=" return (operator(T_OPASS, ADDASS));
119 "-=" return (operator(T_OPASS, SUBASS));
120 "<<=" return (operator(T_OPASS, SHLASS));
121 ">>=" return (operator(T_OPASS, SHRASS));
122 "&=" return (operator(T_OPASS, ANDASS));
123 "^=" return (operator(T_OPASS, XORASS));
124 "|=" return (operator(T_OPASS, ORASS));
125 "||" return (operator(T_LOGOR, LOGOR));
126 "&&" return (operator(T_LOGAND, LOGAND));
127 "|" return (operator(T_OR, OR));
128 "&" return (operator(T_AND, AND));
129 "^" return (operator(T_XOR, XOR));
130 "==" return (operator(T_EQOP, EQ));
131 "!=" return (operator(T_EQOP, NE));
132 "<" return (operator(T_RELOP, LT));
133 ">" return (operator(T_RELOP, GT));
134 "<=" return (operator(T_RELOP, LE));
135 ">=" return (operator(T_RELOP, GE));
136 "<<" return (operator(T_SHFTOP, SHL));
137 ">>" return (operator(T_SHFTOP, SHR));
138 "++" return (operator(T_INCDEC, INC));
139 "--" return (operator(T_INCDEC, DEC));
140 "->" return (operator(T_STROP, ARROW));
141 "." return (operator(T_STROP, POINT));
142 "+" return (operator(T_ADDOP, PLUS));
143 "-" return (operator(T_ADDOP, MINUS));
144 "*" return (operator(T_MULT, MULT));
145 "/" return (operator(T_DIVOP, DIV));
146 "%" return (operator(T_DIVOP, MOD));
147 "!" return (operator(T_UNOP, NOT));
148 "~" return (operator(T_UNOP, COMPL));
149 "\"" return (string());
150 "L\"" return (wcstrg());
151 ";" return (T_SEMI);
152 "{" return (T_LBRACE);
153 "}" return (T_RBRACE);
154 "," return (T_COMMA);
155 ":" return (T_COLON);
156 "?" return (T_QUEST);
157 "[" return (T_LBRACK);
158 "]" return (T_RBRACK);
159 "(" return (T_LPARN);
160 ")" return (T_RPARN);
161 "..." return (T_ELLIPSE);
162 "'" return (ccon());
163 "L'" return (wccon());
164 ^#.*$ directive();
165 \n incline();
166 \t|" "|\f|\v ;
167 "/*" comment();
168 "//" slashslashcomment();
169 . badchar(yytext[0]);
170
171 %%
172
173 static void
174 incline(void)
175 {
176 curr_pos.p_line++;
177 curr_pos.p_uniq = 0;
178 if (curr_pos.p_file == csrc_pos.p_file) {
179 csrc_pos.p_line++;
180 csrc_pos.p_uniq = 0;
181 }
182 }
183
184 static void
185 badchar(int c)
186 {
187
188 /* unknown character \%o */
189 error(250, c);
190 }
191
192 /*
193 * Keywords.
194 * During initialisation they are written to the symbol table.
195 */
196 static struct kwtab {
197 const char *kw_name; /* keyword */
198 int kw_token; /* token returned by yylex() */
199 scl_t kw_scl; /* storage class if kw_token T_SCLASS */
200 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */
201 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */
202 u_int kw_c89 : 1; /* c89 keyword */
203 u_int kw_c99 : 1; /* c99 keyword */
204 u_int kw_gcc : 1; /* GCC keyword */
205 u_int kw_attr : 1; /* GCC attribute, keyword */
206 u_int kw_deco : 3; /* name[1] __name[2] __name__[4] */
207 } kwtab[] = {
208 #ifdef INT128_SIZE
209 { "__int128_t", T_TYPE, 0, INT128, 0, 0,1,0,0,1 },
210 { "__uint128_t",T_TYPE, 0, UINT128,0, 0,1,0,0,1 },
211 #endif
212 { "__thread", T_QUAL, 0, 0, THREAD, 0,0,1,0,1 },
213 { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 },
214 { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 },
215 { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 },
216 { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 },
217 { "_Thread_local",T_QUAL, 0, 0, THREAD, 0,1,0,0,1 },
218 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 },
219 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 },
220 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 },
221 { "alloc_size", T_AT_ALLOC_SIZE,0, 0, 0, 0,0,1,1,5 },
222 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 },
223 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 },
224 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 },
225 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 },
226 { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 },
227 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 },
228 { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 },
229 { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 },
230 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 },
231 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 },
232 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 },
233 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 },
234 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 },
235 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 },
236 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 },
237 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 },
238 { "destructor", T_AT_DESTRUCTOR,0, 0, 0, 0,0,1,1,5 },
239 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 },
240 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 },
241 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 },
242 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 },
243 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 },
244 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 },
245 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 },
246 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 },
247 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 },
248 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 },
249 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 },
250 { "gnu_printf", T_AT_FORMAT_GNU_PRINTF,0,0, 0, 0,0,1,1,5 },
251 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 },
252 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 },
253 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 },
254 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 },
255 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 },
256 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 },
257 { "malloc", T_AT_MALLOC, 0, 0, 0, 0,0,1,1,5 },
258 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 },
259 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 },
260 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 },
261 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION,
262 0, 0, 0, 0,0,1,1,5 },
263 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 },
264 { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 },
265 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 },
266 { "nothrow", T_AT_NOTHROW, 0, 0, 0, 0,0,1,1,5 },
267 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 },
268 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 },
269 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 },
270 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 },
271 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 },
272 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 },
273 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 },
274 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 },
275 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 },
276 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 },
277 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 },
278 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 },
279 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 },
280 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 },
281 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 },
282 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 },
283 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 },
284 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 },
285 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 },
286 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 },
287 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 },
288 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 },
289 { "symbolrename", T_SYMBOLRENAME,0, 0, 0, 0,0,0,0,2 },
290 { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 },
291 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 },
292 { "tls_model", T_AT_TLS_MODEL, 0, 0, 0, 0,0,1,1,5 },
293 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 },
294 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 },
295 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 },
296 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 },
297 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 },
298 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 },
299 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 },
300 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 },
301 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 },
302 { "warn_unused_result", T_AT_WARN_UNUSED_RESULT, 0, 0, 0, 0,0,1,1,5 },
303 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 },
304 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 },
305 { NULL, 0, 0, 0, 0, 0,0,0,0,0 }
306 };
307
308 /* Symbol table */
309 static sym_t *symtab[HSHSIZ1];
310
311 /* bit i of the entry with index i is set */
312 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
313
314 /* least significant i bits are set in the entry with index i */
315 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
316
317 /* least significant i bits are not set in the entry with index i */
318 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
319
320 /* free list for sbuf structures */
321 static sbuf_t *sbfrlst;
322
323 /* Typ of next expected symbol */
324 symt_t symtyp;
325
326
327 static void
328 addkw(struct kwtab *kw, int deco)
329 {
330 sym_t *sym;
331 size_t h;
332 char buf[256];
333 const char *name;
334
335 if (!(kw->kw_deco & deco))
336 return;
337
338 switch (deco) {
339 case 1:
340 name = kw->kw_name;
341 break;
342 case 2:
343 snprintf(buf, sizeof(buf), "__%s", kw->kw_name);
344 name = strdup(buf);
345 break;
346 case 4:
347 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name);
348 name = strdup(buf);
349 break;
350 default:
351 abort();
352 }
353
354 if (name == NULL)
355 err(1, "Can't init symbol table");
356
357 sym = getblk(sizeof (sym_t));
358 sym->s_name = name;
359 sym->s_keyw = kw;
360 sym->s_value.v_quad = kw->kw_token;
361 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
362 sym->s_tspec = kw->kw_tspec;
363 } else if (kw->kw_token == T_SCLASS) {
364 sym->s_scl = kw->kw_scl;
365 } else if (kw->kw_token == T_QUAL) {
366 sym->s_tqual = kw->kw_tqual;
367 }
368 h = hash(sym->s_name);
369 if ((sym->s_link = symtab[h]) != NULL)
370 symtab[h]->s_rlink = &sym->s_link;
371 sym->s_rlink = &symtab[h];
372 symtab[h] = sym;
373 }
374
375 /*
376 * All keywords are written to the symbol table. This saves us looking
377 * in a extra table for each name we found.
378 */
379 void
380 initscan(void)
381 {
382 struct kwtab *kw;
383 size_t i;
384 uint64_t uq;
385
386 for (kw = kwtab; kw->kw_name != NULL; kw++) {
387 if ((kw->kw_c89 || kw->kw_c99) && tflag)
388 continue;
389 if (kw->kw_c99 && !(Sflag || gflag))
390 continue;
391 if (kw->kw_gcc && !gflag)
392 continue;
393 addkw(kw, 1);
394 addkw(kw, 2);
395 addkw(kw, 4);
396 }
397
398 /* initialize bit-masks for quads */
399 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
400 qbmasks[i] = (uint64_t)1 << i;
401 uq = ~(uint64_t)0 << i;
402 qumasks[i] = uq;
403 qlmasks[i] = ~uq;
404 }
405 qumasks[i] = 0;
406 qlmasks[i] = ~(uint64_t)0;
407 }
408
409 /*
410 * Get a free sbuf structure, if possible from the free list
411 */
412 static sbuf_t *
413 allocsb(void)
414 {
415 sbuf_t *sb;
416
417 if ((sb = sbfrlst) != NULL) {
418 sbfrlst = sb->sb_nxt;
419 #ifdef BLKDEBUG
420 (void)memset(sb, 0, sizeof (*sb));
421 #else
422 sb->sb_nxt = NULL;
423 #endif
424 } else {
425 sb = xmalloc(sizeof (sbuf_t));
426 (void)memset(sb, 0, sizeof (*sb));
427 }
428 return (sb);
429 }
430
431 /*
432 * Put a sbuf structure to the free list
433 */
434 static void
435 freesb(sbuf_t *sb)
436 {
437
438 (void)memset(sb, ZERO, sizeof (*sb));
439 sb->sb_nxt = sbfrlst;
440 sbfrlst = sb;
441 }
442
443 /*
444 * Read a character and ensure that it is positive (except EOF).
445 * Increment line count(s) if necessary.
446 */
447 static int
448 inpc(void)
449 {
450 int c;
451
452 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
453 incline();
454 return (c);
455 }
456
457 static int
458 hash(const char *s)
459 {
460 u_int v;
461 const u_char *us;
462
463 v = 0;
464 for (us = (const u_char *)s; *us != '\0'; us++) {
465 v = (v << sizeof (v)) + *us;
466 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
467 }
468 return (v % HSHSIZ1);
469 }
470
471 /*
472 * Lex has found a letter followed by zero or more letters or digits.
473 * It looks for a symbol in the symbol table with the same name. This
474 * symbol must either be a keyword or a symbol of the type required by
475 * symtyp (label, member, tag, ...).
476 *
477 * If it is a keyword, the token is returned. In some cases it is described
478 * more deeply by data written to yylval.
479 *
480 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
481 * is stored in yylval. This struct contains the name of the symbol, its
482 * length and hash value. If there is already a symbol of the same name
483 * and type in the symbol table, the sbuf struct also contains a pointer
484 * to the symbol table entry.
485 */
486 static int
487 name(void)
488 {
489 char *s;
490 sbuf_t *sb;
491 sym_t *sym;
492 int tok;
493
494 sb = allocsb();
495 sb->sb_name = yytext;
496 sb->sb_len = yyleng;
497 sb->sb_hash = hash(yytext);
498 if ((sym = search(sb)) != NULL && sym->s_keyw) {
499 freesb(sb);
500 return (keyw(sym));
501 }
502
503 sb->sb_sym = sym;
504
505 if (sym != NULL) {
506 if (blklev < sym->s_blklev)
507 LERROR("name()");
508 sb->sb_name = sym->s_name;
509 sb->sb_len = strlen(sym->s_name);
510 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
511 } else {
512 s = getblk(yyleng + 1);
513 (void)memcpy(s, yytext, yyleng + 1);
514 sb->sb_name = s;
515 sb->sb_len = yyleng;
516 tok = T_NAME;
517 }
518
519 yylval.y_sb = sb;
520 return (tok);
521 }
522
523 static sym_t *
524 search(sbuf_t *sb)
525 {
526 sym_t *sym;
527
528 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
529 if (strcmp(sym->s_name, sb->sb_name) == 0) {
530 if (sym->s_keyw) {
531 struct kwtab *kw = sym->s_keyw;
532 if (!kw->kw_attr || attron)
533 return (sym);
534 } else if (!attron && sym->s_kind == symtyp)
535 return sym;
536 }
537 }
538
539 return (NULL);
540 }
541
542 static int
543 keyw(sym_t *sym)
544 {
545 int t;
546
547 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
548 yylval.y_scl = sym->s_scl;
549 } else if (t == T_TYPE || t == T_SOU) {
550 yylval.y_tspec = sym->s_tspec;
551 } else if (t == T_QUAL) {
552 yylval.y_tqual = sym->s_tqual;
553 }
554 return (t);
555 }
556
557 /*
558 * Convert a string representing an integer into internal representation.
559 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
560 */
561 static int
562 icon(int base)
563 {
564 int l_suffix, u_suffix;
565 int len;
566 const char *cp;
567 char c, *eptr;
568 tspec_t typ;
569 int ansiu;
570 #ifdef TARG_INT128_MAX
571 __uint128_t uq = 0;
572 static tspec_t contypes[2][4] = {
573 { INT, LONG, QUAD, INT128, },
574 { UINT, ULONG, UQUAD, UINT128, }
575 };
576 #else
577 uint64_t uq = 0;
578 static tspec_t contypes[2][3] = {
579 { INT, LONG, QUAD, },
580 { UINT, ULONG, UQUAD, }
581 };
582 #endif
583
584 cp = yytext;
585 len = yyleng;
586
587 /* skip 0[xX] or 0[bB] */
588 if (base == 16 || base == 2) {
589 cp += 2;
590 len -= 2;
591 }
592
593 /* read suffixes */
594 l_suffix = u_suffix = 0;
595 for ( ; ; ) {
596 if ((c = cp[len - 1]) == 'l' || c == 'L') {
597 l_suffix++;
598 } else if (c == 'u' || c == 'U') {
599 u_suffix++;
600 } else {
601 break;
602 }
603 len--;
604 }
605 if (l_suffix > 2 || u_suffix > 1) {
606 /* malformed integer constant */
607 warning(251);
608 if (l_suffix > 2)
609 l_suffix = 2;
610 if (u_suffix > 1)
611 u_suffix = 1;
612 }
613 if (tflag && u_suffix != 0) {
614 /* suffix U is illegal in traditional C */
615 warning(97);
616 }
617 typ = contypes[u_suffix][l_suffix];
618
619 errno = 0;
620
621 uq = strtouq(cp, &eptr, base);
622 if (eptr != cp + len)
623 LERROR("icon()");
624 if (errno != 0)
625 /* integer constant out of range */
626 warning(252);
627
628 /*
629 * If the value is too big for the current type, we must choose
630 * another type.
631 */
632 ansiu = 0;
633 switch (typ) {
634 case INT:
635 if (uq <= TARG_INT_MAX) {
636 /* ok */
637 } else if (uq <= TARG_UINT_MAX && base != 10) {
638 typ = UINT;
639 } else if (uq <= TARG_LONG_MAX) {
640 typ = LONG;
641 } else {
642 typ = ULONG;
643 if (uq > TARG_ULONG_MAX) {
644 /* integer constant out of range */
645 warning(252);
646 }
647 }
648 if (typ == UINT || typ == ULONG) {
649 if (tflag) {
650 typ = LONG;
651 } else if (!sflag) {
652 /*
653 * Remember that the constant is unsigned
654 * only in ANSI C
655 */
656 ansiu = 1;
657 }
658 }
659 break;
660 case UINT:
661 if (uq > TARG_UINT_MAX) {
662 typ = ULONG;
663 if (uq > TARG_ULONG_MAX) {
664 /* integer constant out of range */
665 warning(252);
666 }
667 }
668 break;
669 case LONG:
670 if (uq > TARG_LONG_MAX && !tflag) {
671 typ = ULONG;
672 if (!sflag)
673 ansiu = 1;
674 if (uq > TARG_ULONG_MAX) {
675 /* integer constant out of range */
676 warning(252);
677 }
678 }
679 break;
680 case ULONG:
681 if (uq > TARG_ULONG_MAX) {
682 /* integer constant out of range */
683 warning(252);
684 }
685 break;
686 case QUAD:
687 if (uq > TARG_QUAD_MAX && !tflag) {
688 typ = UQUAD;
689 if (!sflag)
690 ansiu = 1;
691 }
692 break;
693 case UQUAD:
694 if (uq > TARG_UQUAD_MAX) {
695 /* integer constant out of range */
696 warning(252);
697 }
698 break;
699 #ifdef INT128_SIZE
700 case INT128:
701 #ifdef TARG_INT128_MAX
702 if (uq > TARG_INT128_MAX && !tflag) {
703 typ = UINT128;
704 if (!sflag)
705 ansiu = 1;
706 }
707 #endif
708 break;
709 case UINT128:
710 #ifdef TARG_INT128_MAX
711 if (uq > TARG_UINT128_MAX) {
712 /* integer constant out of range */
713 warning(252);
714 }
715 #endif
716 break;
717 #endif
718 /* LINTED206: (enumeration values not handled in switch) */
719 case STRUCT:
720 case VOID:
721 case LDOUBLE:
722 case FUNC:
723 case ARRAY:
724 case PTR:
725 case ENUM:
726 case UNION:
727 case SIGNED:
728 case NOTSPEC:
729 case DOUBLE:
730 case FLOAT:
731 case USHORT:
732 case SHORT:
733 case UCHAR:
734 case SCHAR:
735 case CHAR:
736 case BOOL:
737 case UNSIGN:
738 case FCOMPLEX:
739 case DCOMPLEX:
740 case LCOMPLEX:
741 case COMPLEX:
742 break;
743
744 case NTSPEC: /* this value unused */
745 break;
746 }
747
748 uq = (uint64_t)xsign((int64_t)uq, typ, -1);
749
750 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
751 yylval.y_val->v_ansiu = ansiu;
752 yylval.y_val->v_quad = (int64_t)uq;
753
754 return (T_CON);
755 }
756
757 /*
758 * Returns 1 if t is a signed type and the value is negative.
759 *
760 * len is the number of significant bits. If len is -1, len is set
761 * to the width of type t.
762 */
763 int
764 sign(int64_t q, tspec_t t, int len)
765 {
766
767 if (t == PTR || isutyp(t))
768 return (0);
769 return (msb(q, t, len));
770 }
771
772 int
773 msb(int64_t q, tspec_t t, int len)
774 {
775
776 if (len <= 0)
777 len = size(t);
778 return ((q & qbmasks[len - 1]) != 0);
779 }
780
781 /*
782 * Extends the sign of q.
783 */
784 int64_t
785 xsign(int64_t q, tspec_t t, int len)
786 {
787
788 if (len <= 0)
789 len = size(t);
790
791 if (t == PTR || isutyp(t) || !sign(q, t, len)) {
792 q &= qlmasks[len];
793 } else {
794 q |= qumasks[len];
795 }
796 return (q);
797 }
798
799 /*
800 * Convert a string representing a floating point value into its interal
801 * representation. Type and value are returned in yylval. fcon()
802 * (and yylex()) returns T_CON.
803 * XXX Currently it is not possible to convert constants of type
804 * long double which are greater than DBL_MAX.
805 */
806 static int
807 fcon(void)
808 {
809 const char *cp;
810 int len;
811 tspec_t typ;
812 char c, *eptr;
813 double d;
814 float f = 0;
815
816 cp = yytext;
817 len = yyleng;
818
819 if (cp[len - 1] == 'i') {
820 /* imaginary, do nothing for now */
821 len--;
822 }
823 if ((c = cp[len - 1]) == 'f' || c == 'F') {
824 typ = FLOAT;
825 len--;
826 } else if (c == 'l' || c == 'L') {
827 typ = LDOUBLE;
828 len--;
829 } else {
830 if (c == 'd' || c == 'D')
831 len--;
832 typ = DOUBLE;
833 }
834
835 if (tflag && typ != DOUBLE) {
836 /* suffixes F and L are illegal in traditional C */
837 warning(98);
838 }
839
840 errno = 0;
841 d = strtod(cp, &eptr);
842 if (eptr != cp + len) {
843 switch (*eptr) {
844 /*
845 * XXX: non-native non-current strtod() may not handle hex
846 * floats, ignore the rest if we find traces of hex float
847 * syntax...
848 */
849 case 'p':
850 case 'P':
851 case 'x':
852 case 'X':
853 d = 0;
854 errno = 0;
855 break;
856 default:
857 LERROR("fcon(%s->%s)", cp, eptr);
858 }
859 }
860 if (errno != 0)
861 /* floating-point constant out of range */
862 warning(248);
863
864 if (typ == FLOAT) {
865 f = (float)d;
866 if (!finite(f)) {
867 /* floating-point constant out of range */
868 warning(248);
869 f = f > 0 ? FLT_MAX : -FLT_MAX;
870 }
871 }
872
873 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
874 if (typ == FLOAT) {
875 yylval.y_val->v_ldbl = f;
876 } else {
877 yylval.y_val->v_ldbl = d;
878 }
879
880 return (T_CON);
881 }
882
883 static int
884 operator(int t, op_t o)
885 {
886
887 yylval.y_op = o;
888 return (t);
889 }
890
891 /*
892 * Called if lex found a leading \'.
893 */
894 static int
895 ccon(void)
896 {
897 size_t n;
898 int val, c;
899 char cv;
900
901 n = 0;
902 val = 0;
903 while ((c = getescc('\'')) >= 0) {
904 val = (val << CHAR_BIT) + c;
905 n++;
906 }
907 if (c == -2) {
908 /* unterminated character constant */
909 error(253);
910 } else {
911 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
912 /* too many characters in character constant */
913 error(71);
914 } else if (n > 1) {
915 /* multi-character character constant */
916 warning(294);
917 } else if (n == 0) {
918 /* empty character constant */
919 error(73);
920 }
921 }
922 if (n == 1) {
923 cv = (char)val;
924 val = cv;
925 }
926
927 yylval.y_val = xcalloc(1, sizeof (val_t));
928 yylval.y_val->v_tspec = INT;
929 yylval.y_val->v_quad = val;
930
931 return (T_CON);
932 }
933
934 /*
935 * Called if lex found a leading L\'
936 */
937 static int
938 wccon(void)
939 {
940 static char buf[MB_LEN_MAX + 1];
941 size_t i;
942 int c;
943 wchar_t wc;
944
945 i = 0;
946 while ((c = getescc('\'')) >= 0) {
947 if (i < MB_CUR_MAX)
948 buf[i] = (char)c;
949 i++;
950 }
951
952 wc = 0;
953
954 if (c == -2) {
955 /* unterminated character constant */
956 error(253);
957 } else if (c == 0) {
958 /* empty character constant */
959 error(73);
960 } else {
961 if (i > MB_CUR_MAX) {
962 i = MB_CUR_MAX;
963 /* too many characters in character constant */
964 error(71);
965 } else {
966 buf[i] = '\0';
967 (void)mbtowc(NULL, NULL, 0);
968 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
969 /* invalid multibyte character */
970 error(291);
971 }
972 }
973
974 yylval.y_val = xcalloc(1, sizeof (val_t));
975 yylval.y_val->v_tspec = WCHAR;
976 yylval.y_val->v_quad = wc;
977
978 return (T_CON);
979 }
980
981 /*
982 * Read a character which is part of a character constant or of a string
983 * and handle escapes.
984 *
985 * The Argument is the character which delimits the character constant or
986 * string.
987 *
988 * Returns -1 if the end of the character constant or string is reached,
989 * -2 if the EOF is reached, and the character otherwise.
990 */
991 static int
992 getescc(int d)
993 {
994 static int pbc = -1;
995 int n, c, v;
996
997 if (pbc == -1) {
998 c = inpc();
999 } else {
1000 c = pbc;
1001 pbc = -1;
1002 }
1003 if (c == d)
1004 return (-1);
1005 switch (c) {
1006 case '\n':
1007 if (tflag) {
1008 /* newline in string or char constant */
1009 error(254);
1010 return (-2);
1011 }
1012 return (c);
1013 case EOF:
1014 return (-2);
1015 case '\\':
1016 switch (c = inpc()) {
1017 case '"':
1018 if (tflag && d == '\'')
1019 /* \" inside character constant undef. ... */
1020 warning(262);
1021 return ('"');
1022 case '\'':
1023 return ('\'');
1024 case '?':
1025 if (tflag)
1026 /* \? undefined in traditional C */
1027 warning(263);
1028 return ('?');
1029 case '\\':
1030 return ('\\');
1031 case 'a':
1032 if (tflag)
1033 /* \a undefined in traditional C */
1034 warning(81);
1035 return ('\a');
1036 case 'b':
1037 return ('\b');
1038 case 'f':
1039 return ('\f');
1040 case 'n':
1041 return ('\n');
1042 case 'r':
1043 return ('\r');
1044 case 't':
1045 return ('\t');
1046 case 'v':
1047 if (tflag)
1048 /* \v undefined in traditional C */
1049 warning(264);
1050 return ('\v');
1051 case '8': case '9':
1052 /* bad octal digit %c */
1053 warning(77, c);
1054 /* FALLTHROUGH */
1055 case '0': case '1': case '2': case '3':
1056 case '4': case '5': case '6': case '7':
1057 n = 3;
1058 v = 0;
1059 do {
1060 v = (v << 3) + (c - '0');
1061 c = inpc();
1062 } while (--n && isdigit(c) && (tflag || c <= '7'));
1063 if (tflag && n > 0 && isdigit(c))
1064 /* bad octal digit %c */
1065 warning(77, c);
1066 pbc = c;
1067 if (v > TARG_UCHAR_MAX) {
1068 /* character escape does not fit in char. */
1069 warning(76);
1070 v &= CHAR_MASK;
1071 }
1072 return (v);
1073 case 'x':
1074 if (tflag)
1075 /* \x undefined in traditional C */
1076 warning(82);
1077 v = 0;
1078 n = 0;
1079 while ((c = inpc()) >= 0 && isxdigit(c)) {
1080 c = isdigit(c) ?
1081 c - '0' : toupper(c) - 'A' + 10;
1082 v = (v << 4) + c;
1083 if (n >= 0) {
1084 if ((v & ~CHAR_MASK) != 0) {
1085 /* overflow in hex escape */
1086 warning(75);
1087 n = -1;
1088 } else {
1089 n++;
1090 }
1091 }
1092 }
1093 pbc = c;
1094 if (n == 0) {
1095 /* no hex digits follow \x */
1096 error(74);
1097 } if (n == -1) {
1098 v &= CHAR_MASK;
1099 }
1100 return (v);
1101 case '\n':
1102 return (getescc(d));
1103 case EOF:
1104 return (-2);
1105 default:
1106 if (isprint(c)) {
1107 /* dubious escape \%c */
1108 warning(79, c);
1109 } else {
1110 /* dubious escape \%o */
1111 warning(80, c);
1112 }
1113 }
1114 }
1115 return (c);
1116 }
1117
1118 /*
1119 * Called for preprocessor directives. Currently implemented are:
1120 * # lineno
1121 * # lineno "filename"
1122 */
1123 static void
1124 directive(void)
1125 {
1126 const char *cp, *fn;
1127 char c, *eptr;
1128 size_t fnl;
1129 long ln;
1130 static int first = 1;
1131
1132 /* Go to first non-whitespace after # */
1133 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1134 continue;
1135
1136 if (!isdigit((unsigned char)c)) {
1137 if (strncmp(cp, "pragma", 6) == 0
1138 && isspace((unsigned char)cp[6]))
1139 return;
1140 error:
1141 /* undefined or invalid # directive */
1142 warning(255);
1143 return;
1144 }
1145 ln = strtol(--cp, &eptr, 10);
1146 if (cp == eptr)
1147 goto error;
1148 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1149 goto error;
1150 while ((c = *cp++) == ' ' || c == '\t')
1151 continue;
1152 if (c != '\0') {
1153 if (c != '"')
1154 goto error;
1155 fn = cp;
1156 while ((c = *cp) != '"' && c != '\0')
1157 cp++;
1158 if (c != '"')
1159 goto error;
1160 if ((fnl = cp++ - fn) > PATH_MAX)
1161 goto error;
1162 while ((c = *cp++) == ' ' || c == '\t')
1163 continue;
1164 #if 0
1165 if (c != '\0')
1166 warning("extra character(s) after directive");
1167 #endif
1168
1169 /* empty string means stdin */
1170 if (fnl == 0) {
1171 fn = "{standard input}";
1172 fnl = 16; /* strlen (fn) */
1173 }
1174 curr_pos.p_file = fnnalloc(fn, fnl);
1175 /*
1176 * If this is the first directive, the name is the name
1177 * of the C source file as specified at the command line.
1178 * It is written to the output file.
1179 */
1180 if (first) {
1181 csrc_pos.p_file = curr_pos.p_file;
1182 outsrc(fnxform(curr_pos.p_file,
1183 strlen(curr_pos.p_file)));
1184 first = 0;
1185 }
1186 }
1187 curr_pos.p_line = (int)ln - 1;
1188 curr_pos.p_uniq = 0;
1189 if (curr_pos.p_file == csrc_pos.p_file) {
1190 csrc_pos.p_line = (int)ln - 1;
1191 csrc_pos.p_uniq = 0;
1192 }
1193 }
1194
1195 /*
1196 * Handle lint comments. Following comments are currently understood:
1197 * ARGSUSEDn
1198 * BITFIELDTYPE
1199 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1200 * FALLTHRU FALLTHROUGH
1201 * LINTLIBRARY
1202 * LINTEDn NOSTRICTn
1203 * LONGLONG
1204 * NOTREACHED
1205 * PRINTFLIKEn
1206 * PROTOLIB
1207 * SCANFLIKEn
1208 * VARARGSn
1209 * If one of this comments is recognized, the arguments, if any, are
1210 * parsed and a function which handles this comment is called.
1211 */
1212 static void
1213 comment(void)
1214 {
1215 int c, lc;
1216 static struct {
1217 const char *keywd;
1218 int arg;
1219 void (*func)(int);
1220 } keywtab[] = {
1221 { "ARGSUSED", 1, argsused },
1222 { "BITFIELDTYPE", 0, bitfieldtype },
1223 { "CONSTCOND", 0, constcond },
1224 { "CONSTANTCOND", 0, constcond },
1225 { "CONSTANTCONDITION", 0, constcond },
1226 { "FALLTHRU", 0, fallthru },
1227 { "FALLTHROUGH", 0, fallthru },
1228 { "LINTLIBRARY", 0, lintlib },
1229 { "LINTED", 1, linted },
1230 { "LONGLONG", 0, longlong },
1231 { "NOSTRICT", 1, linted },
1232 { "NOTREACHED", 0, notreach },
1233 { "PRINTFLIKE", 1, printflike },
1234 { "PROTOLIB", 1, protolib },
1235 { "SCANFLIKE", 1, scanflike },
1236 { "VARARGS", 1, varargs },
1237 };
1238 char keywd[32];
1239 char arg[32];
1240 size_t l, i;
1241 int a;
1242 int eoc;
1243
1244 eoc = 0;
1245
1246 /* Skip white spaces after the start of the comment */
1247 while ((c = inpc()) != EOF && isspace(c))
1248 continue;
1249
1250 /* Read the potential keyword to keywd */
1251 l = 0;
1252 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1253 keywd[l++] = (char)c;
1254 c = inpc();
1255 }
1256 keywd[l] = '\0';
1257
1258 /* look for the keyword */
1259 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1260 if (strcmp(keywtab[i].keywd, keywd) == 0)
1261 break;
1262 }
1263 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1264 goto skip_rest;
1265
1266 /* skip white spaces after the keyword */
1267 while (c != EOF && isspace(c))
1268 c = inpc();
1269
1270 /* read the argument, if the keyword accepts one and there is one */
1271 l = 0;
1272 if (keywtab[i].arg) {
1273 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1274 arg[l++] = (char)c;
1275 c = inpc();
1276 }
1277 }
1278 arg[l] = '\0';
1279 a = l != 0 ? atoi(arg) : -1;
1280
1281 /* skip white spaces after the argument */
1282 while (c != EOF && isspace(c))
1283 c = inpc();
1284
1285 if (c != '*' || (c = inpc()) != '/') {
1286 if (keywtab[i].func != linted)
1287 /* extra characters in lint comment */
1288 warning(257);
1289 } else {
1290 /*
1291 * remember that we have already found the end of the
1292 * comment
1293 */
1294 eoc = 1;
1295 }
1296
1297 if (keywtab[i].func != NULL)
1298 (*keywtab[i].func)(a);
1299
1300 skip_rest:
1301 while (!eoc) {
1302 lc = c;
1303 if ((c = inpc()) == EOF) {
1304 /* unterminated comment */
1305 error(256);
1306 break;
1307 }
1308 if (lc == '*' && c == '/')
1309 eoc = 1;
1310 }
1311 }
1312
1313 /*
1314 * Handle // style comments
1315 */
1316 static void
1317 slashslashcomment(void)
1318 {
1319 int c;
1320
1321 if (!Sflag && !gflag)
1322 /* // comments only supported in C99 */
1323 (void)gnuism(312, tflag ? "traditional" : "ANSI");
1324
1325 while ((c = inpc()) != EOF && c != '\n')
1326 continue;
1327 }
1328
1329 /*
1330 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1331 * clrwflgs() is called after function definitions and global and
1332 * local declarations and definitions. It is also called between
1333 * the controlling expression and the body of control statements
1334 * (if, switch, for, while).
1335 */
1336 void
1337 clrwflgs(void)
1338 {
1339
1340 lwarn = LWARN_ALL;
1341 quadflg = 0;
1342 ccflg = 0;
1343 }
1344
1345 /*
1346 * Strings are stored in a dynamically alloceted buffer and passed
1347 * in yylval.y_xstrg to the parser. The parser or the routines called
1348 * by the parser are responsible for freeing this buffer.
1349 */
1350 static int
1351 string(void)
1352 {
1353 u_char *s;
1354 int c;
1355 size_t len, max;
1356 strg_t *strg;
1357
1358 s = xmalloc(max = 64);
1359
1360 len = 0;
1361 while ((c = getescc('"')) >= 0) {
1362 /* +1 to reserve space for a trailing NUL character */
1363 if (len + 1 == max)
1364 s = xrealloc(s, max *= 2);
1365 s[len++] = (char)c;
1366 }
1367 s[len] = '\0';
1368 if (c == -2)
1369 /* unterminated string constant */
1370 error(258);
1371
1372 strg = xcalloc(1, sizeof (strg_t));
1373 strg->st_tspec = CHAR;
1374 strg->st_len = len;
1375 strg->st_cp = s;
1376
1377 yylval.y_strg = strg;
1378 return (T_STRING);
1379 }
1380
1381 static int
1382 wcstrg(void)
1383 {
1384 char *s;
1385 int c, n;
1386 size_t i, wi;
1387 size_t len, max, wlen;
1388 wchar_t *ws;
1389 strg_t *strg;
1390
1391 s = xmalloc(max = 64);
1392 len = 0;
1393 while ((c = getescc('"')) >= 0) {
1394 /* +1 to save space for a trailing NUL character */
1395 if (len + 1 >= max)
1396 s = xrealloc(s, max *= 2);
1397 s[len++] = (char)c;
1398 }
1399 s[len] = '\0';
1400 if (c == -2)
1401 /* unterminated string constant */
1402 error(258);
1403
1404 /* get length of wide-character string */
1405 (void)mblen(NULL, 0);
1406 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1407 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1408 /* invalid multibyte character */
1409 error(291);
1410 break;
1411 }
1412 if (n == 0)
1413 n = 1;
1414 }
1415
1416 ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1417
1418 /* convert from multibyte to wide char */
1419 (void)mbtowc(NULL, NULL, 0);
1420 for (i = 0, wi = 0; i < len; i += n, wi++) {
1421 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1422 break;
1423 if (n == 0)
1424 n = 1;
1425 }
1426 ws[wi] = 0;
1427 free(s);
1428
1429 strg = xcalloc(1, sizeof (strg_t));
1430 strg->st_tspec = WCHAR;
1431 strg->st_len = wlen;
1432 strg->st_wcp = ws;
1433
1434 yylval.y_strg = strg;
1435 return (T_STRING);
1436 }
1437
1438 /*
1439 * As noted above the scanner does not create new symbol table entries
1440 * for symbols it cannot find in the symbol table. This is to avoid
1441 * putting undeclared symbols into the symbol table if a syntax error
1442 * occurs.
1443 *
1444 * getsym() is called as soon as it is probably ok to put the symbol to
1445 * the symbol table. This does not mean that it is not possible that
1446 * symbols are put to the symbol table which are than not completely
1447 * declared due to syntax errors. To avoid too many problems in this
1448 * case symbols get type int in getsym().
1449 *
1450 * XXX calls to getsym() should be delayed until decl1*() is called
1451 */
1452 sym_t *
1453 getsym(sbuf_t *sb)
1454 {
1455 dinfo_t *di;
1456 char *s;
1457 sym_t *sym;
1458
1459 sym = sb->sb_sym;
1460
1461 /*
1462 * During member declaration it is possible that name() looked
1463 * for symbols of type FVFT, although it should have looked for
1464 * symbols of type FTAG. Same can happen for labels. Both cases
1465 * are compensated here.
1466 */
1467 if (symtyp == FMOS || symtyp == FLAB) {
1468 if (sym == NULL || sym->s_kind == FVFT)
1469 sym = search(sb);
1470 }
1471
1472 if (sym != NULL) {
1473 if (sym->s_kind != symtyp)
1474 LERROR("storesym(%d, %d)", sym->s_kind, symtyp);
1475 symtyp = FVFT;
1476 freesb(sb);
1477 return (sym);
1478 }
1479
1480 /* create a new symbol table entry */
1481
1482 /* labels must always be allocated at level 1 (outhermost block) */
1483 if (symtyp == FLAB) {
1484 sym = getlblk(1, sizeof (sym_t));
1485 s = getlblk(1, sb->sb_len + 1);
1486 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1487 sym->s_name = s;
1488 sym->s_blklev = 1;
1489 di = dcs;
1490 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1491 di = di->d_nxt;
1492 if (di->d_ctx != AUTO)
1493 LERROR("storesym()");
1494 } else {
1495 sym = getblk(sizeof (sym_t));
1496 sym->s_name = sb->sb_name;
1497 sym->s_blklev = blklev;
1498 di = dcs;
1499 }
1500
1501 UNIQUE_CURR_POS(sym->s_dpos);
1502 if ((sym->s_kind = symtyp) != FLAB)
1503 sym->s_type = gettyp(INT);
1504
1505 symtyp = FVFT;
1506
1507 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1508 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1509 sym->s_rlink = &symtab[sb->sb_hash];
1510 symtab[sb->sb_hash] = sym;
1511
1512 *di->d_ldlsym = sym;
1513 di->d_ldlsym = &sym->s_dlnxt;
1514
1515 freesb(sb);
1516 return (sym);
1517 }
1518
1519 /*
1520 * Construct a temporary symbol. The symbol starts with a digit, so that
1521 * it is illegal.
1522 */
1523 sym_t *
1524 mktempsym(type_t *t)
1525 {
1526 static int n = 0;
1527 int h;
1528 char *s = getlblk(blklev, 64);
1529 sym_t *sym = getblk(sizeof (sym_t));
1530
1531 (void)snprintf(s, 64, "%.8d_tmp", n++);
1532 h = hash(s);
1533
1534 sym->s_name = s;
1535 sym->s_type = t;
1536 sym->s_blklev = blklev;
1537 sym->s_scl = AUTO;
1538 sym->s_kind = FVFT;
1539 sym->s_used = 1;
1540 sym->s_set = 1;
1541
1542 if ((sym->s_link = symtab[h]) != NULL)
1543 symtab[h]->s_rlink = &sym->s_link;
1544 sym->s_rlink = &symtab[h];
1545 symtab[h] = sym;
1546
1547 *dcs->d_ldlsym = sym;
1548 dcs->d_ldlsym = &sym->s_dlnxt;
1549
1550 return sym;
1551 }
1552
1553 /*
1554 * Remove a symbol forever from the symbol table. s_blklev
1555 * is set to -1 to avoid that the symbol will later be put
1556 * back to the symbol table.
1557 */
1558 void
1559 rmsym(sym_t *sym)
1560 {
1561
1562 if ((*sym->s_rlink = sym->s_link) != NULL)
1563 sym->s_link->s_rlink = sym->s_rlink;
1564 sym->s_blklev = -1;
1565 sym->s_link = NULL;
1566 }
1567
1568 /*
1569 * Remove a list of symbols declared at one level from the symbol
1570 * table.
1571 */
1572 void
1573 rmsyms(sym_t *syms)
1574 {
1575 sym_t *sym;
1576
1577 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1578 if (sym->s_blklev != -1) {
1579 if ((*sym->s_rlink = sym->s_link) != NULL)
1580 sym->s_link->s_rlink = sym->s_rlink;
1581 sym->s_link = NULL;
1582 sym->s_rlink = NULL;
1583 }
1584 }
1585 }
1586
1587 /*
1588 * Put a symbol into the symbol table
1589 */
1590 void
1591 inssym(int bl, sym_t *sym)
1592 {
1593 int h;
1594
1595 h = hash(sym->s_name);
1596 if ((sym->s_link = symtab[h]) != NULL)
1597 symtab[h]->s_rlink = &sym->s_link;
1598 sym->s_rlink = &symtab[h];
1599 symtab[h] = sym;
1600 sym->s_blklev = bl;
1601 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1602 LERROR("inssym()");
1603 }
1604
1605 /*
1606 * Called at level 0 after syntax errors
1607 * Removes all symbols which are not declared at level 0 from the
1608 * symbol table. Also frees all memory which is not associated with
1609 * level 0.
1610 */
1611 void
1612 cleanup(void)
1613 {
1614 sym_t *sym, *nsym;
1615 int i;
1616
1617 for (i = 0; i < HSHSIZ1; i++) {
1618 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1619 nsym = sym->s_link;
1620 if (sym->s_blklev >= 1) {
1621 if ((*sym->s_rlink = nsym) != NULL)
1622 nsym->s_rlink = sym->s_rlink;
1623 }
1624 }
1625 }
1626
1627 for (i = mblklev; i > 0; i--)
1628 freelblk(i);
1629 }
1630
1631 /*
1632 * Create a new symbol with the name of an existing symbol.
1633 */
1634 sym_t *
1635 pushdown(sym_t *sym)
1636 {
1637 int h;
1638 sym_t *nsym;
1639
1640 h = hash(sym->s_name);
1641 nsym = getblk(sizeof (sym_t));
1642 if (sym->s_blklev > blklev)
1643 LERROR("pushdown()");
1644 nsym->s_name = sym->s_name;
1645 UNIQUE_CURR_POS(nsym->s_dpos);
1646 nsym->s_kind = sym->s_kind;
1647 nsym->s_blklev = blklev;
1648
1649 if ((nsym->s_link = symtab[h]) != NULL)
1650 symtab[h]->s_rlink = &nsym->s_link;
1651 nsym->s_rlink = &symtab[h];
1652 symtab[h] = nsym;
1653
1654 *dcs->d_ldlsym = nsym;
1655 dcs->d_ldlsym = &nsym->s_dlnxt;
1656
1657 return (nsym);
1658 }
1659
1660 /*
1661 * Free any dynamically allocated memory referenced by
1662 * the value stack or yylval.
1663 * The type of information in yylval is described by tok.
1664 */
1665 void
1666 freeyyv(void *sp, int tok)
1667 {
1668 if (tok == T_NAME || tok == T_TYPENAME) {
1669 sbuf_t *sb = *(sbuf_t **)sp;
1670 freesb(sb);
1671 } else if (tok == T_CON) {
1672 val_t *val = *(val_t **)sp;
1673 free(val);
1674 } else if (tok == T_STRING) {
1675 strg_t *strg = *(strg_t **)sp;
1676 if (strg->st_tspec == CHAR) {
1677 free(strg->st_cp);
1678 } else if (strg->st_tspec == WCHAR) {
1679 free(strg->st_wcp);
1680 } else {
1681 LERROR("fryylv()");
1682 }
1683 free(strg);
1684 }
1685 }
1686