scan.l revision 1.91 1 %{
2 /* $NetBSD: scan.l,v 1.91 2019/11/09 19:54:09 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 1.91 2019/11/09 19:54:09 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 { "_Alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,1 },
214 { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 },
215 { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 },
216 { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 },
217 { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 },
218 { "_Thread_local",T_QUAL, 0, 0, THREAD, 0,1,0,0,1 },
219 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 },
220 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 },
221 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 },
222 { "alloc_size", T_AT_ALLOC_SIZE,0, 0, 0, 0,0,1,1,5 },
223 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 },
224 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 },
225 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 },
226 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 },
227 { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 },
228 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 },
229 { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 },
230 { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 },
231 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 },
232 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 },
233 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 },
234 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 },
235 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 },
236 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 },
237 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 },
238 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 },
239 { "destructor", T_AT_DESTRUCTOR,0, 0, 0, 0,0,1,1,5 },
240 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 },
241 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 },
242 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 },
243 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 },
244 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 },
245 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 },
246 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 },
247 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 },
248 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 },
249 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 },
250 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 },
251 { "gnu_printf", T_AT_FORMAT_GNU_PRINTF,0,0, 0, 0,0,1,1,5 },
252 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 },
253 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 },
254 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 },
255 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 },
256 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 },
257 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 },
258 { "malloc", T_AT_MALLOC, 0, 0, 0, 0,0,1,1,5 },
259 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 },
260 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 },
261 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 },
262 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION,
263 0, 0, 0, 0,0,1,1,5 },
264 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 },
265 { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 },
266 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 },
267 { "nothrow", T_AT_NOTHROW, 0, 0, 0, 0,0,1,1,5 },
268 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 },
269 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 },
270 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 },
271 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 },
272 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 },
273 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 },
274 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 },
275 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 },
276 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 },
277 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 },
278 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 },
279 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 },
280 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 },
281 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 },
282 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 },
283 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 },
284 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 },
285 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 },
286 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 },
287 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 },
288 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 },
289 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 },
290 { "symbolrename", T_SYMBOLRENAME,0, 0, 0, 0,0,0,0,2 },
291 { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 },
292 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 },
293 { "tls_model", T_AT_TLS_MODEL, 0, 0, 0, 0,0,1,1,5 },
294 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 },
295 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 },
296 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 },
297 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 },
298 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 },
299 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 },
300 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 },
301 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 },
302 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 },
303 { "warn_unused_result", T_AT_WARN_UNUSED_RESULT, 0, 0, 0, 0,0,1,1,5 },
304 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 },
305 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 },
306 { NULL, 0, 0, 0, 0, 0,0,0,0,0 }
307 };
308
309 /* Symbol table */
310 static sym_t *symtab[HSHSIZ1];
311
312 /* bit i of the entry with index i is set */
313 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
314
315 /* least significant i bits are set in the entry with index i */
316 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
317
318 /* least significant i bits are not set in the entry with index i */
319 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
320
321 /* free list for sbuf structures */
322 static sbuf_t *sbfrlst;
323
324 /* Typ of next expected symbol */
325 symt_t symtyp;
326
327
328 static void
329 addkw(struct kwtab *kw, int deco)
330 {
331 sym_t *sym;
332 size_t h;
333 char buf[256];
334 const char *name;
335
336 if (!(kw->kw_deco & deco))
337 return;
338
339 switch (deco) {
340 case 1:
341 name = kw->kw_name;
342 break;
343 case 2:
344 snprintf(buf, sizeof(buf), "__%s", kw->kw_name);
345 name = strdup(buf);
346 break;
347 case 4:
348 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name);
349 name = strdup(buf);
350 break;
351 default:
352 abort();
353 }
354
355 if (name == NULL)
356 err(1, "Can't init symbol table");
357
358 sym = getblk(sizeof (sym_t));
359 sym->s_name = name;
360 sym->s_keyw = kw;
361 sym->s_value.v_quad = kw->kw_token;
362 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
363 sym->s_tspec = kw->kw_tspec;
364 } else if (kw->kw_token == T_SCLASS) {
365 sym->s_scl = kw->kw_scl;
366 } else if (kw->kw_token == T_QUAL) {
367 sym->s_tqual = kw->kw_tqual;
368 }
369 h = hash(sym->s_name);
370 if ((sym->s_link = symtab[h]) != NULL)
371 symtab[h]->s_rlink = &sym->s_link;
372 sym->s_rlink = &symtab[h];
373 symtab[h] = sym;
374 }
375
376 /*
377 * All keywords are written to the symbol table. This saves us looking
378 * in a extra table for each name we found.
379 */
380 void
381 initscan(void)
382 {
383 struct kwtab *kw;
384 size_t i;
385 uint64_t uq;
386
387 for (kw = kwtab; kw->kw_name != NULL; kw++) {
388 if ((kw->kw_c89 || kw->kw_c99) && tflag)
389 continue;
390 if (kw->kw_c99 && !(Sflag || gflag))
391 continue;
392 if (kw->kw_gcc && !gflag)
393 continue;
394 addkw(kw, 1);
395 addkw(kw, 2);
396 addkw(kw, 4);
397 }
398
399 /* initialize bit-masks for quads */
400 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
401 qbmasks[i] = (uint64_t)1 << i;
402 uq = ~(uint64_t)0 << i;
403 qumasks[i] = uq;
404 qlmasks[i] = ~uq;
405 }
406 qumasks[i] = 0;
407 qlmasks[i] = ~(uint64_t)0;
408 }
409
410 /*
411 * Get a free sbuf structure, if possible from the free list
412 */
413 static sbuf_t *
414 allocsb(void)
415 {
416 sbuf_t *sb;
417
418 if ((sb = sbfrlst) != NULL) {
419 sbfrlst = sb->sb_nxt;
420 #ifdef BLKDEBUG
421 (void)memset(sb, 0, sizeof (*sb));
422 #else
423 sb->sb_nxt = NULL;
424 #endif
425 } else {
426 sb = xmalloc(sizeof (sbuf_t));
427 (void)memset(sb, 0, sizeof (*sb));
428 }
429 return (sb);
430 }
431
432 /*
433 * Put a sbuf structure to the free list
434 */
435 static void
436 freesb(sbuf_t *sb)
437 {
438
439 (void)memset(sb, ZERO, sizeof (*sb));
440 sb->sb_nxt = sbfrlst;
441 sbfrlst = sb;
442 }
443
444 /*
445 * Read a character and ensure that it is positive (except EOF).
446 * Increment line count(s) if necessary.
447 */
448 static int
449 inpc(void)
450 {
451 int c;
452
453 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
454 incline();
455 return (c);
456 }
457
458 static int
459 hash(const char *s)
460 {
461 u_int v;
462 const u_char *us;
463
464 v = 0;
465 for (us = (const u_char *)s; *us != '\0'; us++) {
466 v = (v << sizeof (v)) + *us;
467 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
468 }
469 return (v % HSHSIZ1);
470 }
471
472 /*
473 * Lex has found a letter followed by zero or more letters or digits.
474 * It looks for a symbol in the symbol table with the same name. This
475 * symbol must either be a keyword or a symbol of the type required by
476 * symtyp (label, member, tag, ...).
477 *
478 * If it is a keyword, the token is returned. In some cases it is described
479 * more deeply by data written to yylval.
480 *
481 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
482 * is stored in yylval. This struct contains the name of the symbol, its
483 * length and hash value. If there is already a symbol of the same name
484 * and type in the symbol table, the sbuf struct also contains a pointer
485 * to the symbol table entry.
486 */
487 static int
488 name(void)
489 {
490 char *s;
491 sbuf_t *sb;
492 sym_t *sym;
493 int tok;
494
495 sb = allocsb();
496 sb->sb_name = yytext;
497 sb->sb_len = yyleng;
498 sb->sb_hash = hash(yytext);
499 if ((sym = search(sb)) != NULL && sym->s_keyw) {
500 freesb(sb);
501 return (keyw(sym));
502 }
503
504 sb->sb_sym = sym;
505
506 if (sym != NULL) {
507 if (blklev < sym->s_blklev)
508 LERROR("name()");
509 sb->sb_name = sym->s_name;
510 sb->sb_len = strlen(sym->s_name);
511 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
512 } else {
513 s = getblk(yyleng + 1);
514 (void)memcpy(s, yytext, yyleng + 1);
515 sb->sb_name = s;
516 sb->sb_len = yyleng;
517 tok = T_NAME;
518 }
519
520 yylval.y_sb = sb;
521 return (tok);
522 }
523
524 static sym_t *
525 search(sbuf_t *sb)
526 {
527 sym_t *sym;
528
529 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
530 if (strcmp(sym->s_name, sb->sb_name) == 0) {
531 if (sym->s_keyw) {
532 struct kwtab *kw = sym->s_keyw;
533 if (!kw->kw_attr || attron)
534 return (sym);
535 } else if (!attron && sym->s_kind == symtyp)
536 return sym;
537 }
538 }
539
540 return (NULL);
541 }
542
543 static int
544 keyw(sym_t *sym)
545 {
546 int t;
547
548 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
549 yylval.y_scl = sym->s_scl;
550 } else if (t == T_TYPE || t == T_SOU) {
551 yylval.y_tspec = sym->s_tspec;
552 } else if (t == T_QUAL) {
553 yylval.y_tqual = sym->s_tqual;
554 }
555 return (t);
556 }
557
558 /*
559 * Convert a string representing an integer into internal representation.
560 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
561 */
562 static int
563 icon(int base)
564 {
565 int l_suffix, u_suffix;
566 int len;
567 const char *cp;
568 char c, *eptr;
569 tspec_t typ;
570 int ansiu;
571 #ifdef TARG_INT128_MAX
572 __uint128_t uq = 0;
573 static tspec_t contypes[2][4] = {
574 { INT, LONG, QUAD, INT128, },
575 { UINT, ULONG, UQUAD, UINT128, }
576 };
577 #else
578 uint64_t uq = 0;
579 static tspec_t contypes[2][3] = {
580 { INT, LONG, QUAD, },
581 { UINT, ULONG, UQUAD, }
582 };
583 #endif
584
585 cp = yytext;
586 len = yyleng;
587
588 /* skip 0[xX] or 0[bB] */
589 if (base == 16 || base == 2) {
590 cp += 2;
591 len -= 2;
592 }
593
594 /* read suffixes */
595 l_suffix = u_suffix = 0;
596 for ( ; ; ) {
597 if ((c = cp[len - 1]) == 'l' || c == 'L') {
598 l_suffix++;
599 } else if (c == 'u' || c == 'U') {
600 u_suffix++;
601 } else {
602 break;
603 }
604 len--;
605 }
606 if (l_suffix > 2 || u_suffix > 1) {
607 /* malformed integer constant */
608 warning(251);
609 if (l_suffix > 2)
610 l_suffix = 2;
611 if (u_suffix > 1)
612 u_suffix = 1;
613 }
614 if (tflag && u_suffix != 0) {
615 /* suffix U is illegal in traditional C */
616 warning(97);
617 }
618 typ = contypes[u_suffix][l_suffix];
619
620 errno = 0;
621
622 uq = strtouq(cp, &eptr, base);
623 if (eptr != cp + len)
624 LERROR("icon()");
625 if (errno != 0)
626 /* integer constant out of range */
627 warning(252);
628
629 /*
630 * If the value is too big for the current type, we must choose
631 * another type.
632 */
633 ansiu = 0;
634 switch (typ) {
635 case INT:
636 if (uq <= TARG_INT_MAX) {
637 /* ok */
638 } else if (uq <= TARG_UINT_MAX && base != 10) {
639 typ = UINT;
640 } else if (uq <= TARG_LONG_MAX) {
641 typ = LONG;
642 } else {
643 typ = ULONG;
644 if (uq > TARG_ULONG_MAX) {
645 /* integer constant out of range */
646 warning(252);
647 }
648 }
649 if (typ == UINT || typ == ULONG) {
650 if (tflag) {
651 typ = LONG;
652 } else if (!sflag) {
653 /*
654 * Remember that the constant is unsigned
655 * only in ANSI C
656 */
657 ansiu = 1;
658 }
659 }
660 break;
661 case UINT:
662 if (uq > TARG_UINT_MAX) {
663 typ = ULONG;
664 if (uq > TARG_ULONG_MAX) {
665 /* integer constant out of range */
666 warning(252);
667 }
668 }
669 break;
670 case LONG:
671 if (uq > TARG_LONG_MAX && !tflag) {
672 typ = ULONG;
673 if (!sflag)
674 ansiu = 1;
675 if (uq > TARG_ULONG_MAX) {
676 /* integer constant out of range */
677 warning(252);
678 }
679 }
680 break;
681 case ULONG:
682 if (uq > TARG_ULONG_MAX) {
683 /* integer constant out of range */
684 warning(252);
685 }
686 break;
687 case QUAD:
688 if (uq > TARG_QUAD_MAX && !tflag) {
689 typ = UQUAD;
690 if (!sflag)
691 ansiu = 1;
692 }
693 break;
694 case UQUAD:
695 if (uq > TARG_UQUAD_MAX) {
696 /* integer constant out of range */
697 warning(252);
698 }
699 break;
700 #ifdef INT128_SIZE
701 case INT128:
702 #ifdef TARG_INT128_MAX
703 if (uq > TARG_INT128_MAX && !tflag) {
704 typ = UINT128;
705 if (!sflag)
706 ansiu = 1;
707 }
708 #endif
709 break;
710 case UINT128:
711 #ifdef TARG_INT128_MAX
712 if (uq > TARG_UINT128_MAX) {
713 /* integer constant out of range */
714 warning(252);
715 }
716 #endif
717 break;
718 #endif
719 /* LINTED206: (enumeration values not handled in switch) */
720 case STRUCT:
721 case VOID:
722 case LDOUBLE:
723 case FUNC:
724 case ARRAY:
725 case PTR:
726 case ENUM:
727 case UNION:
728 case SIGNED:
729 case NOTSPEC:
730 case DOUBLE:
731 case FLOAT:
732 case USHORT:
733 case SHORT:
734 case UCHAR:
735 case SCHAR:
736 case CHAR:
737 case BOOL:
738 case UNSIGN:
739 case FCOMPLEX:
740 case DCOMPLEX:
741 case LCOMPLEX:
742 case COMPLEX:
743 break;
744
745 case NTSPEC: /* this value unused */
746 break;
747 }
748
749 uq = (uint64_t)xsign((int64_t)uq, typ, -1);
750
751 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
752 yylval.y_val->v_ansiu = ansiu;
753 yylval.y_val->v_quad = (int64_t)uq;
754
755 return (T_CON);
756 }
757
758 /*
759 * Returns 1 if t is a signed type and the value is negative.
760 *
761 * len is the number of significant bits. If len is -1, len is set
762 * to the width of type t.
763 */
764 int
765 sign(int64_t q, tspec_t t, int len)
766 {
767
768 if (t == PTR || isutyp(t))
769 return (0);
770 return (msb(q, t, len));
771 }
772
773 int
774 msb(int64_t q, tspec_t t, int len)
775 {
776
777 if (len <= 0)
778 len = size(t);
779 return ((q & qbmasks[len - 1]) != 0);
780 }
781
782 /*
783 * Extends the sign of q.
784 */
785 int64_t
786 xsign(int64_t q, tspec_t t, int len)
787 {
788
789 if (len <= 0)
790 len = size(t);
791
792 if (t == PTR || isutyp(t) || !sign(q, t, len)) {
793 q &= qlmasks[len];
794 } else {
795 q |= qumasks[len];
796 }
797 return (q);
798 }
799
800 /*
801 * Convert a string representing a floating point value into its interal
802 * representation. Type and value are returned in yylval. fcon()
803 * (and yylex()) returns T_CON.
804 * XXX Currently it is not possible to convert constants of type
805 * long double which are greater than DBL_MAX.
806 */
807 static int
808 fcon(void)
809 {
810 const char *cp;
811 int len;
812 tspec_t typ;
813 char c, *eptr;
814 double d;
815 float f = 0;
816
817 cp = yytext;
818 len = yyleng;
819
820 if (cp[len - 1] == 'i') {
821 /* imaginary, do nothing for now */
822 len--;
823 }
824 if ((c = cp[len - 1]) == 'f' || c == 'F') {
825 typ = FLOAT;
826 len--;
827 } else if (c == 'l' || c == 'L') {
828 typ = LDOUBLE;
829 len--;
830 } else {
831 if (c == 'd' || c == 'D')
832 len--;
833 typ = DOUBLE;
834 }
835
836 if (tflag && typ != DOUBLE) {
837 /* suffixes F and L are illegal in traditional C */
838 warning(98);
839 }
840
841 errno = 0;
842 d = strtod(cp, &eptr);
843 if (eptr != cp + len) {
844 switch (*eptr) {
845 /*
846 * XXX: non-native non-current strtod() may not handle hex
847 * floats, ignore the rest if we find traces of hex float
848 * syntax...
849 */
850 case 'p':
851 case 'P':
852 case 'x':
853 case 'X':
854 d = 0;
855 errno = 0;
856 break;
857 default:
858 LERROR("fcon(%s->%s)", cp, eptr);
859 }
860 }
861 if (errno != 0)
862 /* floating-point constant out of range */
863 warning(248);
864
865 if (typ == FLOAT) {
866 f = (float)d;
867 if (!finite(f)) {
868 /* floating-point constant out of range */
869 warning(248);
870 f = f > 0 ? FLT_MAX : -FLT_MAX;
871 }
872 }
873
874 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
875 if (typ == FLOAT) {
876 yylval.y_val->v_ldbl = f;
877 } else {
878 yylval.y_val->v_ldbl = d;
879 }
880
881 return (T_CON);
882 }
883
884 static int
885 operator(int t, op_t o)
886 {
887
888 yylval.y_op = o;
889 return (t);
890 }
891
892 /*
893 * Called if lex found a leading \'.
894 */
895 static int
896 ccon(void)
897 {
898 size_t n;
899 int val, c;
900 char cv;
901
902 n = 0;
903 val = 0;
904 while ((c = getescc('\'')) >= 0) {
905 val = (val << CHAR_BIT) + c;
906 n++;
907 }
908 if (c == -2) {
909 /* unterminated character constant */
910 error(253);
911 } else {
912 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
913 /* too many characters in character constant */
914 error(71);
915 } else if (n > 1) {
916 /* multi-character character constant */
917 warning(294);
918 } else if (n == 0) {
919 /* empty character constant */
920 error(73);
921 }
922 }
923 if (n == 1) {
924 cv = (char)val;
925 val = cv;
926 }
927
928 yylval.y_val = xcalloc(1, sizeof (val_t));
929 yylval.y_val->v_tspec = INT;
930 yylval.y_val->v_quad = val;
931
932 return (T_CON);
933 }
934
935 /*
936 * Called if lex found a leading L\'
937 */
938 static int
939 wccon(void)
940 {
941 static char buf[MB_LEN_MAX + 1];
942 size_t i;
943 int c;
944 wchar_t wc;
945
946 i = 0;
947 while ((c = getescc('\'')) >= 0) {
948 if (i < MB_CUR_MAX)
949 buf[i] = (char)c;
950 i++;
951 }
952
953 wc = 0;
954
955 if (c == -2) {
956 /* unterminated character constant */
957 error(253);
958 } else if (c == 0) {
959 /* empty character constant */
960 error(73);
961 } else {
962 if (i > MB_CUR_MAX) {
963 i = MB_CUR_MAX;
964 /* too many characters in character constant */
965 error(71);
966 } else {
967 buf[i] = '\0';
968 (void)mbtowc(NULL, NULL, 0);
969 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
970 /* invalid multibyte character */
971 error(291);
972 }
973 }
974
975 yylval.y_val = xcalloc(1, sizeof (val_t));
976 yylval.y_val->v_tspec = WCHAR;
977 yylval.y_val->v_quad = wc;
978
979 return (T_CON);
980 }
981
982 /*
983 * Read a character which is part of a character constant or of a string
984 * and handle escapes.
985 *
986 * The Argument is the character which delimits the character constant or
987 * string.
988 *
989 * Returns -1 if the end of the character constant or string is reached,
990 * -2 if the EOF is reached, and the character otherwise.
991 */
992 static int
993 getescc(int d)
994 {
995 static int pbc = -1;
996 int n, c, v;
997
998 if (pbc == -1) {
999 c = inpc();
1000 } else {
1001 c = pbc;
1002 pbc = -1;
1003 }
1004 if (c == d)
1005 return (-1);
1006 switch (c) {
1007 case '\n':
1008 if (tflag) {
1009 /* newline in string or char constant */
1010 error(254);
1011 return (-2);
1012 }
1013 return (c);
1014 case EOF:
1015 return (-2);
1016 case '\\':
1017 switch (c = inpc()) {
1018 case '"':
1019 if (tflag && d == '\'')
1020 /* \" inside character constant undef. ... */
1021 warning(262);
1022 return ('"');
1023 case '\'':
1024 return ('\'');
1025 case '?':
1026 if (tflag)
1027 /* \? undefined in traditional C */
1028 warning(263);
1029 return ('?');
1030 case '\\':
1031 return ('\\');
1032 case 'a':
1033 if (tflag)
1034 /* \a undefined in traditional C */
1035 warning(81);
1036 return ('\a');
1037 case 'b':
1038 return ('\b');
1039 case 'f':
1040 return ('\f');
1041 case 'n':
1042 return ('\n');
1043 case 'r':
1044 return ('\r');
1045 case 't':
1046 return ('\t');
1047 case 'v':
1048 if (tflag)
1049 /* \v undefined in traditional C */
1050 warning(264);
1051 return ('\v');
1052 case '8': case '9':
1053 /* bad octal digit %c */
1054 warning(77, c);
1055 /* FALLTHROUGH */
1056 case '0': case '1': case '2': case '3':
1057 case '4': case '5': case '6': case '7':
1058 n = 3;
1059 v = 0;
1060 do {
1061 v = (v << 3) + (c - '0');
1062 c = inpc();
1063 } while (--n && isdigit(c) && (tflag || c <= '7'));
1064 if (tflag && n > 0 && isdigit(c))
1065 /* bad octal digit %c */
1066 warning(77, c);
1067 pbc = c;
1068 if (v > TARG_UCHAR_MAX) {
1069 /* character escape does not fit in char. */
1070 warning(76);
1071 v &= CHAR_MASK;
1072 }
1073 return (v);
1074 case 'x':
1075 if (tflag)
1076 /* \x undefined in traditional C */
1077 warning(82);
1078 v = 0;
1079 n = 0;
1080 while ((c = inpc()) >= 0 && isxdigit(c)) {
1081 c = isdigit(c) ?
1082 c - '0' : toupper(c) - 'A' + 10;
1083 v = (v << 4) + c;
1084 if (n >= 0) {
1085 if ((v & ~CHAR_MASK) != 0) {
1086 /* overflow in hex escape */
1087 warning(75);
1088 n = -1;
1089 } else {
1090 n++;
1091 }
1092 }
1093 }
1094 pbc = c;
1095 if (n == 0) {
1096 /* no hex digits follow \x */
1097 error(74);
1098 } if (n == -1) {
1099 v &= CHAR_MASK;
1100 }
1101 return (v);
1102 case '\n':
1103 return (getescc(d));
1104 case EOF:
1105 return (-2);
1106 default:
1107 if (isprint(c)) {
1108 /* dubious escape \%c */
1109 warning(79, c);
1110 } else {
1111 /* dubious escape \%o */
1112 warning(80, c);
1113 }
1114 }
1115 }
1116 return (c);
1117 }
1118
1119 /*
1120 * Called for preprocessor directives. Currently implemented are:
1121 * # lineno
1122 * # lineno "filename"
1123 */
1124 static void
1125 directive(void)
1126 {
1127 const char *cp, *fn;
1128 char c, *eptr;
1129 size_t fnl;
1130 long ln;
1131 static int first = 1;
1132
1133 /* Go to first non-whitespace after # */
1134 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1135 continue;
1136
1137 if (!isdigit((unsigned char)c)) {
1138 if (strncmp(cp, "pragma", 6) == 0
1139 && isspace((unsigned char)cp[6]))
1140 return;
1141 error:
1142 /* undefined or invalid # directive */
1143 warning(255);
1144 return;
1145 }
1146 ln = strtol(--cp, &eptr, 10);
1147 if (cp == eptr)
1148 goto error;
1149 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1150 goto error;
1151 while ((c = *cp++) == ' ' || c == '\t')
1152 continue;
1153 if (c != '\0') {
1154 if (c != '"')
1155 goto error;
1156 fn = cp;
1157 while ((c = *cp) != '"' && c != '\0')
1158 cp++;
1159 if (c != '"')
1160 goto error;
1161 if ((fnl = cp++ - fn) > PATH_MAX)
1162 goto error;
1163 while ((c = *cp++) == ' ' || c == '\t')
1164 continue;
1165 #if 0
1166 if (c != '\0')
1167 warning("extra character(s) after directive");
1168 #endif
1169
1170 /* empty string means stdin */
1171 if (fnl == 0) {
1172 fn = "{standard input}";
1173 fnl = 16; /* strlen (fn) */
1174 }
1175 curr_pos.p_file = fnnalloc(fn, fnl);
1176 /*
1177 * If this is the first directive, the name is the name
1178 * of the C source file as specified at the command line.
1179 * It is written to the output file.
1180 */
1181 if (first) {
1182 csrc_pos.p_file = curr_pos.p_file;
1183 outsrc(fnxform(curr_pos.p_file,
1184 strlen(curr_pos.p_file)));
1185 first = 0;
1186 }
1187 }
1188 curr_pos.p_line = (int)ln - 1;
1189 curr_pos.p_uniq = 0;
1190 if (curr_pos.p_file == csrc_pos.p_file) {
1191 csrc_pos.p_line = (int)ln - 1;
1192 csrc_pos.p_uniq = 0;
1193 }
1194 }
1195
1196 /*
1197 * Handle lint comments. Following comments are currently understood:
1198 * ARGSUSEDn
1199 * BITFIELDTYPE
1200 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1201 * FALLTHRU FALLTHROUGH
1202 * LINTLIBRARY
1203 * LINTEDn NOSTRICTn
1204 * LONGLONG
1205 * NOTREACHED
1206 * PRINTFLIKEn
1207 * PROTOLIB
1208 * SCANFLIKEn
1209 * VARARGSn
1210 * If one of this comments is recognized, the arguments, if any, are
1211 * parsed and a function which handles this comment is called.
1212 */
1213 static void
1214 comment(void)
1215 {
1216 int c, lc;
1217 static struct {
1218 const char *keywd;
1219 int arg;
1220 void (*func)(int);
1221 } keywtab[] = {
1222 { "ARGSUSED", 1, argsused },
1223 { "BITFIELDTYPE", 0, bitfieldtype },
1224 { "CONSTCOND", 0, constcond },
1225 { "CONSTANTCOND", 0, constcond },
1226 { "CONSTANTCONDITION", 0, constcond },
1227 { "FALLTHRU", 0, fallthru },
1228 { "FALLTHROUGH", 0, fallthru },
1229 { "LINTLIBRARY", 0, lintlib },
1230 { "LINTED", 1, linted },
1231 { "LONGLONG", 0, longlong },
1232 { "NOSTRICT", 1, linted },
1233 { "NOTREACHED", 0, notreach },
1234 { "PRINTFLIKE", 1, printflike },
1235 { "PROTOLIB", 1, protolib },
1236 { "SCANFLIKE", 1, scanflike },
1237 { "VARARGS", 1, varargs },
1238 };
1239 char keywd[32];
1240 char arg[32];
1241 size_t l, i;
1242 int a;
1243 int eoc;
1244
1245 eoc = 0;
1246
1247 /* Skip white spaces after the start of the comment */
1248 while ((c = inpc()) != EOF && isspace(c))
1249 continue;
1250
1251 /* Read the potential keyword to keywd */
1252 l = 0;
1253 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1254 keywd[l++] = (char)c;
1255 c = inpc();
1256 }
1257 keywd[l] = '\0';
1258
1259 /* look for the keyword */
1260 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1261 if (strcmp(keywtab[i].keywd, keywd) == 0)
1262 break;
1263 }
1264 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1265 goto skip_rest;
1266
1267 /* skip white spaces after the keyword */
1268 while (c != EOF && isspace(c))
1269 c = inpc();
1270
1271 /* read the argument, if the keyword accepts one and there is one */
1272 l = 0;
1273 if (keywtab[i].arg) {
1274 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1275 arg[l++] = (char)c;
1276 c = inpc();
1277 }
1278 }
1279 arg[l] = '\0';
1280 a = l != 0 ? atoi(arg) : -1;
1281
1282 /* skip white spaces after the argument */
1283 while (c != EOF && isspace(c))
1284 c = inpc();
1285
1286 if (c != '*' || (c = inpc()) != '/') {
1287 if (keywtab[i].func != linted)
1288 /* extra characters in lint comment */
1289 warning(257);
1290 } else {
1291 /*
1292 * remember that we have already found the end of the
1293 * comment
1294 */
1295 eoc = 1;
1296 }
1297
1298 if (keywtab[i].func != NULL)
1299 (*keywtab[i].func)(a);
1300
1301 skip_rest:
1302 while (!eoc) {
1303 lc = c;
1304 if ((c = inpc()) == EOF) {
1305 /* unterminated comment */
1306 error(256);
1307 break;
1308 }
1309 if (lc == '*' && c == '/')
1310 eoc = 1;
1311 }
1312 }
1313
1314 /*
1315 * Handle // style comments
1316 */
1317 static void
1318 slashslashcomment(void)
1319 {
1320 int c;
1321
1322 if (!Sflag && !gflag)
1323 /* // comments only supported in C99 */
1324 (void)gnuism(312, tflag ? "traditional" : "ANSI");
1325
1326 while ((c = inpc()) != EOF && c != '\n')
1327 continue;
1328 }
1329
1330 /*
1331 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1332 * clrwflgs() is called after function definitions and global and
1333 * local declarations and definitions. It is also called between
1334 * the controlling expression and the body of control statements
1335 * (if, switch, for, while).
1336 */
1337 void
1338 clrwflgs(void)
1339 {
1340
1341 lwarn = LWARN_ALL;
1342 quadflg = 0;
1343 ccflg = 0;
1344 }
1345
1346 /*
1347 * Strings are stored in a dynamically alloceted buffer and passed
1348 * in yylval.y_xstrg to the parser. The parser or the routines called
1349 * by the parser are responsible for freeing this buffer.
1350 */
1351 static int
1352 string(void)
1353 {
1354 u_char *s;
1355 int c;
1356 size_t len, max;
1357 strg_t *strg;
1358
1359 s = xmalloc(max = 64);
1360
1361 len = 0;
1362 while ((c = getescc('"')) >= 0) {
1363 /* +1 to reserve space for a trailing NUL character */
1364 if (len + 1 == max)
1365 s = xrealloc(s, max *= 2);
1366 s[len++] = (char)c;
1367 }
1368 s[len] = '\0';
1369 if (c == -2)
1370 /* unterminated string constant */
1371 error(258);
1372
1373 strg = xcalloc(1, sizeof (strg_t));
1374 strg->st_tspec = CHAR;
1375 strg->st_len = len;
1376 strg->st_cp = s;
1377
1378 yylval.y_strg = strg;
1379 return (T_STRING);
1380 }
1381
1382 static int
1383 wcstrg(void)
1384 {
1385 char *s;
1386 int c, n;
1387 size_t i, wi;
1388 size_t len, max, wlen;
1389 wchar_t *ws;
1390 strg_t *strg;
1391
1392 s = xmalloc(max = 64);
1393 len = 0;
1394 while ((c = getescc('"')) >= 0) {
1395 /* +1 to save space for a trailing NUL character */
1396 if (len + 1 >= max)
1397 s = xrealloc(s, max *= 2);
1398 s[len++] = (char)c;
1399 }
1400 s[len] = '\0';
1401 if (c == -2)
1402 /* unterminated string constant */
1403 error(258);
1404
1405 /* get length of wide-character string */
1406 (void)mblen(NULL, 0);
1407 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1408 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1409 /* invalid multibyte character */
1410 error(291);
1411 break;
1412 }
1413 if (n == 0)
1414 n = 1;
1415 }
1416
1417 ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1418
1419 /* convert from multibyte to wide char */
1420 (void)mbtowc(NULL, NULL, 0);
1421 for (i = 0, wi = 0; i < len; i += n, wi++) {
1422 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1423 break;
1424 if (n == 0)
1425 n = 1;
1426 }
1427 ws[wi] = 0;
1428 free(s);
1429
1430 strg = xcalloc(1, sizeof (strg_t));
1431 strg->st_tspec = WCHAR;
1432 strg->st_len = wlen;
1433 strg->st_wcp = ws;
1434
1435 yylval.y_strg = strg;
1436 return (T_STRING);
1437 }
1438
1439 /*
1440 * As noted above the scanner does not create new symbol table entries
1441 * for symbols it cannot find in the symbol table. This is to avoid
1442 * putting undeclared symbols into the symbol table if a syntax error
1443 * occurs.
1444 *
1445 * getsym() is called as soon as it is probably ok to put the symbol to
1446 * the symbol table. This does not mean that it is not possible that
1447 * symbols are put to the symbol table which are than not completely
1448 * declared due to syntax errors. To avoid too many problems in this
1449 * case symbols get type int in getsym().
1450 *
1451 * XXX calls to getsym() should be delayed until decl1*() is called
1452 */
1453 sym_t *
1454 getsym(sbuf_t *sb)
1455 {
1456 dinfo_t *di;
1457 char *s;
1458 sym_t *sym;
1459
1460 sym = sb->sb_sym;
1461
1462 /*
1463 * During member declaration it is possible that name() looked
1464 * for symbols of type FVFT, although it should have looked for
1465 * symbols of type FTAG. Same can happen for labels. Both cases
1466 * are compensated here.
1467 */
1468 if (symtyp == FMOS || symtyp == FLAB) {
1469 if (sym == NULL || sym->s_kind == FVFT)
1470 sym = search(sb);
1471 }
1472
1473 if (sym != NULL) {
1474 if (sym->s_kind != symtyp)
1475 LERROR("storesym(%d, %d)", sym->s_kind, symtyp);
1476 symtyp = FVFT;
1477 freesb(sb);
1478 return (sym);
1479 }
1480
1481 /* create a new symbol table entry */
1482
1483 /* labels must always be allocated at level 1 (outhermost block) */
1484 if (symtyp == FLAB) {
1485 sym = getlblk(1, sizeof (sym_t));
1486 s = getlblk(1, sb->sb_len + 1);
1487 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1488 sym->s_name = s;
1489 sym->s_blklev = 1;
1490 di = dcs;
1491 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1492 di = di->d_nxt;
1493 if (di->d_ctx != AUTO)
1494 LERROR("storesym()");
1495 } else {
1496 sym = getblk(sizeof (sym_t));
1497 sym->s_name = sb->sb_name;
1498 sym->s_blklev = blklev;
1499 di = dcs;
1500 }
1501
1502 UNIQUE_CURR_POS(sym->s_dpos);
1503 if ((sym->s_kind = symtyp) != FLAB)
1504 sym->s_type = gettyp(INT);
1505
1506 symtyp = FVFT;
1507
1508 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1509 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1510 sym->s_rlink = &symtab[sb->sb_hash];
1511 symtab[sb->sb_hash] = sym;
1512
1513 *di->d_ldlsym = sym;
1514 di->d_ldlsym = &sym->s_dlnxt;
1515
1516 freesb(sb);
1517 return (sym);
1518 }
1519
1520 /*
1521 * Construct a temporary symbol. The symbol starts with a digit, so that
1522 * it is illegal.
1523 */
1524 sym_t *
1525 mktempsym(type_t *t)
1526 {
1527 static int n = 0;
1528 int h;
1529 char *s = getlblk(blklev, 64);
1530 sym_t *sym = getblk(sizeof (sym_t));
1531
1532 (void)snprintf(s, 64, "%.8d_tmp", n++);
1533 h = hash(s);
1534
1535 sym->s_name = s;
1536 sym->s_type = t;
1537 sym->s_blklev = blklev;
1538 sym->s_scl = AUTO;
1539 sym->s_kind = FVFT;
1540 sym->s_used = 1;
1541 sym->s_set = 1;
1542
1543 if ((sym->s_link = symtab[h]) != NULL)
1544 symtab[h]->s_rlink = &sym->s_link;
1545 sym->s_rlink = &symtab[h];
1546 symtab[h] = sym;
1547
1548 *dcs->d_ldlsym = sym;
1549 dcs->d_ldlsym = &sym->s_dlnxt;
1550
1551 return sym;
1552 }
1553
1554 /*
1555 * Remove a symbol forever from the symbol table. s_blklev
1556 * is set to -1 to avoid that the symbol will later be put
1557 * back to the symbol table.
1558 */
1559 void
1560 rmsym(sym_t *sym)
1561 {
1562
1563 if ((*sym->s_rlink = sym->s_link) != NULL)
1564 sym->s_link->s_rlink = sym->s_rlink;
1565 sym->s_blklev = -1;
1566 sym->s_link = NULL;
1567 }
1568
1569 /*
1570 * Remove a list of symbols declared at one level from the symbol
1571 * table.
1572 */
1573 void
1574 rmsyms(sym_t *syms)
1575 {
1576 sym_t *sym;
1577
1578 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1579 if (sym->s_blklev != -1) {
1580 if ((*sym->s_rlink = sym->s_link) != NULL)
1581 sym->s_link->s_rlink = sym->s_rlink;
1582 sym->s_link = NULL;
1583 sym->s_rlink = NULL;
1584 }
1585 }
1586 }
1587
1588 /*
1589 * Put a symbol into the symbol table
1590 */
1591 void
1592 inssym(int bl, sym_t *sym)
1593 {
1594 int h;
1595
1596 h = hash(sym->s_name);
1597 if ((sym->s_link = symtab[h]) != NULL)
1598 symtab[h]->s_rlink = &sym->s_link;
1599 sym->s_rlink = &symtab[h];
1600 symtab[h] = sym;
1601 sym->s_blklev = bl;
1602 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1603 LERROR("inssym()");
1604 }
1605
1606 /*
1607 * Called at level 0 after syntax errors
1608 * Removes all symbols which are not declared at level 0 from the
1609 * symbol table. Also frees all memory which is not associated with
1610 * level 0.
1611 */
1612 void
1613 cleanup(void)
1614 {
1615 sym_t *sym, *nsym;
1616 int i;
1617
1618 for (i = 0; i < HSHSIZ1; i++) {
1619 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1620 nsym = sym->s_link;
1621 if (sym->s_blklev >= 1) {
1622 if ((*sym->s_rlink = nsym) != NULL)
1623 nsym->s_rlink = sym->s_rlink;
1624 }
1625 }
1626 }
1627
1628 for (i = mblklev; i > 0; i--)
1629 freelblk(i);
1630 }
1631
1632 /*
1633 * Create a new symbol with the name of an existing symbol.
1634 */
1635 sym_t *
1636 pushdown(sym_t *sym)
1637 {
1638 int h;
1639 sym_t *nsym;
1640
1641 h = hash(sym->s_name);
1642 nsym = getblk(sizeof (sym_t));
1643 if (sym->s_blklev > blklev)
1644 LERROR("pushdown()");
1645 nsym->s_name = sym->s_name;
1646 UNIQUE_CURR_POS(nsym->s_dpos);
1647 nsym->s_kind = sym->s_kind;
1648 nsym->s_blklev = blklev;
1649
1650 if ((nsym->s_link = symtab[h]) != NULL)
1651 symtab[h]->s_rlink = &nsym->s_link;
1652 nsym->s_rlink = &symtab[h];
1653 symtab[h] = nsym;
1654
1655 *dcs->d_ldlsym = nsym;
1656 dcs->d_ldlsym = &nsym->s_dlnxt;
1657
1658 return (nsym);
1659 }
1660
1661 /*
1662 * Free any dynamically allocated memory referenced by
1663 * the value stack or yylval.
1664 * The type of information in yylval is described by tok.
1665 */
1666 void
1667 freeyyv(void *sp, int tok)
1668 {
1669 if (tok == T_NAME || tok == T_TYPENAME) {
1670 sbuf_t *sb = *(sbuf_t **)sp;
1671 freesb(sb);
1672 } else if (tok == T_CON) {
1673 val_t *val = *(val_t **)sp;
1674 free(val);
1675 } else if (tok == T_STRING) {
1676 strg_t *strg = *(strg_t **)sp;
1677 if (strg->st_tspec == CHAR) {
1678 free(strg->st_cp);
1679 } else if (strg->st_tspec == WCHAR) {
1680 free(strg->st_wcp);
1681 } else {
1682 LERROR("fryylv()");
1683 }
1684 free(strg);
1685 }
1686 }
1687