scan.l revision 1.126 1 %{
2 /* $NetBSD: scan.l,v 1.126 2021/01/18 17:11:14 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 1.126 2021/01/18 17:11:14 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_OPASS, MULASS);
118 "/=" return operator(T_OPASS, DIVASS);
119 "%=" return operator(T_OPASS, MODASS);
120 "+=" return operator(T_OPASS, ADDASS);
121 "-=" return operator(T_OPASS, SUBASS);
122 "<<=" return operator(T_OPASS, SHLASS);
123 ">>=" return operator(T_OPASS, SHRASS);
124 "&=" return operator(T_OPASS, ANDASS);
125 "^=" return operator(T_OPASS, XORASS);
126 "|=" return operator(T_OPASS, ORASS);
127 "||" return operator(T_LOGOR, NOOP);
128 "&&" return operator(T_LOGAND, NOOP);
129 "|" return operator(T_OR, NOOP);
130 "&" return operator(T_AMPER, NOOP);
131 "^" return operator(T_XOR, NOOP);
132 "==" return operator(T_EQOP, EQ);
133 "!=" return operator(T_EQOP, NE);
134 "<" return operator(T_RELOP, LT);
135 ">" return operator(T_RELOP, GT);
136 "<=" return operator(T_RELOP, LE);
137 ">=" return operator(T_RELOP, GE);
138 "<<" return operator(T_SHFTOP, SHL);
139 ">>" return operator(T_SHFTOP, 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_ADDOP, PLUS);
145 "-" return operator(T_ADDOP, MINUS);
146 "*" return operator(T_ASTERISK, NOOP);
147 "/" return operator(T_DIVOP, DIV);
148 "%" return operator(T_DIVOP, MOD);
149 "!" return operator(T_UNOP, NOT);
150 "~" return operator(T_UNOP, 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 T_TYPE or T_SOU */
206 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */
207 bool kw_c89 : 1; /* C89 keyword */
208 bool kw_c99 : 1; /* C99 keyword */
209 bool kw_gcc : 1; /* GCC keyword */
210 bool kw_attr : 1; /* GCC attribute, keyword */
211 u_int kw_deco : 3; /* 1 = name, 2 = __name, 4 = __name__ */
212 } kwtab[] = {
213 #ifdef INT128_SIZE
214 { "__int128_t", T_TYPE, 0, INT128, 0, 0,1,0,0,1 },
215 { "__uint128_t",T_TYPE, 0, UINT128,0, 0,1,0,0,1 },
216 #endif
217 { "__thread", T_QUAL, 0, 0, THREAD, 0,0,1,0,1 },
218 { "_Alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,1 },
219 { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 },
220 { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 },
221 { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 },
222 { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 },
223 { "_Thread_local",T_QUAL, 0, 0, THREAD, 0,1,0,0,1 },
224 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 },
225 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 },
226 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 },
227 { "alloc_size", T_AT_ALLOC_SIZE,0, 0, 0, 0,0,1,1,5 },
228 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 },
229 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 },
230 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 },
231 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 },
232 { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 },
233 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 },
234 { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 },
235 { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 },
236 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 },
237 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 },
238 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 },
239 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 },
240 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 },
241 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 },
242 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 },
243 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 },
244 { "destructor", T_AT_DESTRUCTOR,0, 0, 0, 0,0,1,1,5 },
245 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 },
246 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 },
247 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 },
248 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 },
249 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 },
250 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 },
251 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 },
252 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 },
253 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 },
254 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 },
255 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 },
256 { "gnu_printf", T_AT_FORMAT_GNU_PRINTF,0,0, 0, 0,0,1,1,5 },
257 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 },
258 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 },
259 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 },
260 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 },
261 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 },
262 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 },
263 { "malloc", T_AT_MALLOC, 0, 0, 0, 0,0,1,1,5 },
264 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 },
265 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 },
266 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 },
267 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION,
268 0, 0, 0, 0,0,1,1,5 },
269 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 },
270 { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 },
271 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 },
272 { "nothrow", T_AT_NOTHROW, 0, 0, 0, 0,0,1,1,5 },
273 { "optimize", T_AT_OPTIMIZE, 0, 0, 0, 0,0,1,1,5 },
274 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 },
275 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 },
276 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 },
277 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 },
278 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 },
279 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 },
280 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 },
281 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 },
282 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 },
283 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 },
284 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 },
285 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 },
286 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 },
287 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 },
288 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 },
289 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 },
290 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 },
291 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 },
292 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 },
293 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 },
294 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 },
295 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 },
296 { "symbolrename", T_SYMBOLRENAME,0, 0, 0, 0,0,0,0,2 },
297 { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 },
298 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 },
299 { "tls_model", T_AT_TLS_MODEL, 0, 0, 0, 0,0,1,1,5 },
300 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 },
301 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 },
302 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 },
303 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 },
304 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 },
305 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 },
306 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 },
307 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 },
308 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 },
309 { "warn_unused_result", T_AT_WARN_UNUSED_RESULT, 0, 0, 0, 0,0,1,1,5 },
310 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 },
311 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 },
312 { NULL, 0, 0, 0, 0, 0,0,0,0,0 }
313 };
314
315 /* Symbol table */
316 static sym_t *symtab[HSHSIZ1];
317
318 /* bit i of the entry with index i is set */
319 uint64_t qbmasks[64];
320
321 /* least significant i bits are set in the entry with index i */
322 uint64_t qlmasks[64 + 1];
323
324 /* least significant i bits are not set in the entry with index i */
325 uint64_t qumasks[64 + 1];
326
327 /* free list for sbuf structures */
328 static sbuf_t *sbfrlst;
329
330 /* Typ of next expected symbol */
331 symt_t symtyp;
332
333
334 static void
335 add_keyword(struct kwtab *kw, int deco)
336 {
337 sym_t *sym;
338 size_t h;
339 char buf[256];
340 const char *name;
341
342 if (!(kw->kw_deco & deco))
343 return;
344
345 switch (deco) {
346 case 1:
347 name = kw->kw_name;
348 break;
349 case 2:
350 snprintf(buf, sizeof(buf), "__%s", kw->kw_name);
351 name = strdup(buf);
352 break;
353 default:
354 lint_assert(deco == 4);
355 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name);
356 name = strdup(buf);
357 break;
358 }
359
360 if (name == NULL)
361 err(1, "Can't init symbol table");
362
363 sym = getblk(sizeof (sym_t));
364 sym->s_name = name;
365 sym->s_keyword = kw;
366 sym->s_value.v_quad = kw->kw_token;
367 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
368 sym->s_tspec = kw->kw_tspec;
369 } else if (kw->kw_token == T_SCLASS) {
370 sym->s_scl = kw->kw_scl;
371 } else if (kw->kw_token == T_QUAL) {
372 sym->s_tqual = kw->kw_tqual;
373 }
374 h = hash(sym->s_name);
375 if ((sym->s_link = symtab[h]) != NULL)
376 symtab[h]->s_rlink = &sym->s_link;
377 sym->s_rlink = &symtab[h];
378 symtab[h] = sym;
379 }
380
381 /*
382 * All keywords are written to the symbol table. This saves us looking
383 * in a extra table for each name we found.
384 */
385 void
386 initscan(void)
387 {
388 struct kwtab *kw;
389 size_t i;
390 uint64_t uq;
391
392 for (kw = kwtab; kw->kw_name != NULL; kw++) {
393 if ((kw->kw_c89 || kw->kw_c99) && tflag)
394 continue;
395 if (kw->kw_c99 && !(Sflag || gflag))
396 continue;
397 if (kw->kw_gcc && !gflag)
398 continue;
399 add_keyword(kw, 1);
400 add_keyword(kw, 2);
401 add_keyword(kw, 4);
402 }
403
404 /* initialize bit-masks for quads */
405 for (i = 0; i < 64; i++) {
406 qbmasks[i] = (uint64_t)1 << i;
407 uq = ~(uint64_t)0 << i;
408 qumasks[i] = uq;
409 qlmasks[i] = ~uq;
410 }
411 qumasks[i] = 0;
412 qlmasks[i] = ~(uint64_t)0;
413 }
414
415 /*
416 * Get a free sbuf structure, if possible from the free list
417 */
418 static sbuf_t *
419 allocsb(void)
420 {
421 sbuf_t *sb;
422
423 if ((sb = sbfrlst) != NULL) {
424 sbfrlst = sb->sb_next;
425 #ifdef BLKDEBUG
426 (void)memset(sb, 0, sizeof (*sb));
427 #else
428 sb->sb_next = NULL;
429 #endif
430 } else {
431 sb = xmalloc(sizeof (sbuf_t));
432 (void)memset(sb, 0, sizeof (*sb));
433 }
434 return sb;
435 }
436
437 /*
438 * Put a sbuf structure to the free list
439 */
440 static void
441 freesb(sbuf_t *sb)
442 {
443
444 (void)memset(sb, ZERO, sizeof (*sb));
445 sb->sb_next = sbfrlst;
446 sbfrlst = sb;
447 }
448
449 /*
450 * Read a character and ensure that it is positive (except EOF).
451 * Increment line count(s) if necessary.
452 */
453 static int
454 inpc(void)
455 {
456 int c;
457
458 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
459 incline();
460 return c;
461 }
462
463 static int
464 hash(const char *s)
465 {
466 u_int v;
467 const u_char *us;
468
469 v = 0;
470 for (us = (const u_char *)s; *us != '\0'; us++) {
471 v = (v << sizeof (v)) + *us;
472 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
473 }
474 return v % HSHSIZ1;
475 }
476
477 /*
478 * Lex has found a letter followed by zero or more letters or digits.
479 * It looks for a symbol in the symbol table with the same name. This
480 * symbol must either be a keyword or a symbol of the type required by
481 * symtyp (label, member, tag, ...).
482 *
483 * If it is a keyword, the token is returned. In some cases it is described
484 * more deeply by data written to yylval.
485 *
486 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
487 * is stored in yylval. This struct contains the name of the symbol, its
488 * length and hash value. If there is already a symbol of the same name
489 * and type in the symbol table, the sbuf struct also contains a pointer
490 * to the symbol table entry.
491 */
492 static int
493 name(void)
494 {
495 char *s;
496 sbuf_t *sb;
497 sym_t *sym;
498 int tok;
499
500 sb = allocsb();
501 sb->sb_name = yytext;
502 sb->sb_len = yyleng;
503 sb->sb_hash = hash(yytext);
504 if ((sym = search(sb)) != NULL && sym->s_keyword) {
505 freesb(sb);
506 return keyw(sym);
507 }
508
509 sb->sb_sym = sym;
510
511 if (sym != NULL) {
512 lint_assert(blklev >= sym->s_blklev);
513 sb->sb_name = sym->s_name;
514 sb->sb_len = strlen(sym->s_name);
515 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
516 } else {
517 s = getblk(yyleng + 1);
518 (void)memcpy(s, yytext, yyleng + 1);
519 sb->sb_name = s;
520 sb->sb_len = yyleng;
521 tok = T_NAME;
522 }
523
524 yylval.y_sb = sb;
525 return tok;
526 }
527
528 static sym_t *
529 search(sbuf_t *sb)
530 {
531 sym_t *sym;
532
533 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
534 if (strcmp(sym->s_name, sb->sb_name) == 0) {
535 if (sym->s_keyword) {
536 struct kwtab *kw = sym->s_keyword;
537 if (!kw->kw_attr || attron)
538 return sym;
539 } else if (!attron && sym->s_kind == symtyp)
540 return sym;
541 }
542 }
543
544 return NULL;
545 }
546
547 static int
548 keyw(sym_t *sym)
549 {
550 int t;
551
552 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
553 yylval.y_scl = sym->s_scl;
554 } else if (t == T_TYPE || t == T_SOU) {
555 yylval.y_tspec = sym->s_tspec;
556 } else if (t == T_QUAL) {
557 yylval.y_tqual = sym->s_tqual;
558 }
559 return t;
560 }
561
562 /*
563 * Convert a string representing an integer into internal representation.
564 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
565 */
566 static int
567 icon(int base)
568 {
569 int l_suffix, u_suffix;
570 int len;
571 const char *cp;
572 char c, *eptr;
573 tspec_t typ;
574 bool ansiu;
575 #ifdef TARG_INT128_MAX
576 __uint128_t uq = 0;
577 static tspec_t contypes[2][4] = {
578 { INT, LONG, QUAD, INT128, },
579 { UINT, ULONG, UQUAD, UINT128, }
580 };
581 #else
582 uint64_t uq = 0;
583 static tspec_t contypes[2][3] = {
584 { INT, LONG, QUAD, },
585 { UINT, ULONG, UQUAD, }
586 };
587 #endif
588
589 cp = yytext;
590 len = yyleng;
591
592 /* skip 0[xX] or 0[bB] */
593 if (base == 16 || base == 2) {
594 cp += 2;
595 len -= 2;
596 }
597
598 /* read suffixes */
599 l_suffix = u_suffix = 0;
600 for ( ; ; ) {
601 if ((c = cp[len - 1]) == 'l' || c == 'L') {
602 l_suffix++;
603 } else if (c == 'u' || c == 'U') {
604 u_suffix++;
605 } else {
606 break;
607 }
608 len--;
609 }
610 if (l_suffix > 2 || u_suffix > 1) {
611 /* malformed integer constant */
612 warning(251);
613 if (l_suffix > 2)
614 l_suffix = 2;
615 if (u_suffix > 1)
616 u_suffix = 1;
617 }
618 if (tflag && u_suffix != 0) {
619 /* suffix U is illegal in traditional C */
620 warning(97);
621 }
622 typ = contypes[u_suffix][l_suffix];
623
624 errno = 0;
625
626 uq = strtouq(cp, &eptr, base);
627 lint_assert(eptr == cp + len);
628 if (errno != 0)
629 /* integer constant out of range */
630 warning(252);
631
632 /*
633 * If the value is too big for the current type, we must choose
634 * another type.
635 */
636 ansiu = false;
637 switch (typ) {
638 case INT:
639 if (uq <= TARG_INT_MAX) {
640 /* ok */
641 } else if (uq <= TARG_UINT_MAX && base != 10) {
642 typ = UINT;
643 } else if (uq <= TARG_LONG_MAX) {
644 typ = LONG;
645 } else {
646 typ = ULONG;
647 if (uq > TARG_ULONG_MAX) {
648 /* integer constant out of range */
649 warning(252);
650 }
651 }
652 if (typ == UINT || typ == ULONG) {
653 if (tflag) {
654 typ = LONG;
655 } else if (!sflag) {
656 /*
657 * Remember that the constant is unsigned
658 * only in ANSI C
659 */
660 ansiu = true;
661 }
662 }
663 break;
664 case UINT:
665 if (uq > TARG_UINT_MAX) {
666 typ = ULONG;
667 if (uq > TARG_ULONG_MAX) {
668 /* integer constant out of range */
669 warning(252);
670 }
671 }
672 break;
673 case LONG:
674 if (uq > TARG_LONG_MAX && !tflag) {
675 typ = ULONG;
676 if (!sflag)
677 ansiu = true;
678 if (uq > TARG_ULONG_MAX) {
679 /* integer constant out of range */
680 warning(252);
681 }
682 }
683 break;
684 case ULONG:
685 if (uq > TARG_ULONG_MAX) {
686 /* integer constant out of range */
687 warning(252);
688 }
689 break;
690 case QUAD:
691 if (uq > TARG_QUAD_MAX && !tflag) {
692 typ = UQUAD;
693 if (!sflag)
694 ansiu = true;
695 }
696 break;
697 case UQUAD:
698 if (uq > TARG_UQUAD_MAX) {
699 /* integer constant out of range */
700 warning(252);
701 }
702 break;
703 #ifdef INT128_SIZE
704 case INT128:
705 #ifdef TARG_INT128_MAX
706 if (uq > TARG_INT128_MAX && !tflag) {
707 typ = UINT128;
708 if (!sflag)
709 ansiu = true;
710 }
711 #endif
712 break;
713 case UINT128:
714 #ifdef TARG_INT128_MAX
715 if (uq > TARG_UINT128_MAX) {
716 /* integer constant out of range */
717 warning(252);
718 }
719 #endif
720 break;
721 #endif
722 /* LINTED206: (enumeration values not handled in switch) */
723 case STRUCT:
724 case VOID:
725 case LDOUBLE:
726 case FUNC:
727 case ARRAY:
728 case PTR:
729 case ENUM:
730 case UNION:
731 case SIGNED:
732 case NOTSPEC:
733 case DOUBLE:
734 case FLOAT:
735 case USHORT:
736 case SHORT:
737 case UCHAR:
738 case SCHAR:
739 case CHAR:
740 case BOOL:
741 case UNSIGN:
742 case FCOMPLEX:
743 case DCOMPLEX:
744 case LCOMPLEX:
745 case COMPLEX:
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 || is_uinteger(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 || is_uinteger(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 integral
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_SIZE) + 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 constants 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 character */
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 /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */
1120 static void
1121 parse_line_directive_flags(const char *p)
1122 {
1123
1124 in_system_header = false;
1125
1126 while (ch_isspace(*p))
1127 p++;
1128 while (ch_isdigit(*p)) {
1129 if (*p == '3' && !ch_isdigit(p[1]))
1130 in_system_header = true;
1131 p++;
1132 while (ch_isspace(*p))
1133 p++;
1134 }
1135
1136 #if 0
1137 if (c != '\0')
1138 warning("extra character(s) after directive");
1139 #endif
1140 }
1141
1142 /*
1143 * Called for preprocessor directives. Currently implemented are:
1144 * # lineno
1145 * # lineno "filename"
1146 */
1147 static void
1148 directive(void)
1149 {
1150 const char *cp, *fn;
1151 char c, *eptr;
1152 size_t fnl;
1153 long ln;
1154 static bool first = true;
1155
1156 /* Go to first non-whitespace after # */
1157 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1158 continue;
1159
1160 if (!ch_isdigit(c)) {
1161 if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6]))
1162 return;
1163 error:
1164 /* undefined or invalid # directive */
1165 warning(255);
1166 return;
1167 }
1168 ln = strtol(--cp, &eptr, 10);
1169 if (cp == eptr)
1170 goto error;
1171 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1172 goto error;
1173 while ((c = *cp++) == ' ' || c == '\t')
1174 continue;
1175 if (c != '\0') {
1176 if (c != '"')
1177 goto error;
1178 fn = cp;
1179 while ((c = *cp) != '"' && c != '\0')
1180 cp++;
1181 if (c != '"')
1182 goto error;
1183 if ((fnl = cp++ - fn) > PATH_MAX)
1184 goto error;
1185 parse_line_directive_flags(cp);
1186
1187 /* empty string means stdin */
1188 if (fnl == 0) {
1189 fn = "{standard input}";
1190 fnl = 16; /* strlen (fn) */
1191 }
1192 curr_pos.p_file = fnnalloc(fn, fnl);
1193 /*
1194 * If this is the first directive, the name is the name
1195 * of the C source file as specified at the command line.
1196 * It is written to the output file.
1197 */
1198 if (first) {
1199 csrc_pos.p_file = curr_pos.p_file;
1200 outsrc(fnxform(curr_pos.p_file,
1201 strlen(curr_pos.p_file)));
1202 first = false;
1203 }
1204 }
1205 curr_pos.p_line = (int)ln - 1;
1206 curr_pos.p_uniq = 0;
1207 if (curr_pos.p_file == csrc_pos.p_file) {
1208 csrc_pos.p_line = (int)ln - 1;
1209 csrc_pos.p_uniq = 0;
1210 }
1211 }
1212
1213 /*
1214 * Handle lint comments. Following comments are currently understood:
1215 * ARGSUSEDn
1216 * BITFIELDTYPE
1217 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1218 * FALLTHRU FALLTHROUGH
1219 * LINTLIBRARY
1220 * LINTEDn NOSTRICTn
1221 * LONGLONG
1222 * NOTREACHED
1223 * PRINTFLIKEn
1224 * PROTOLIB
1225 * SCANFLIKEn
1226 * VARARGSn
1227 * If one of this comments is recognized, the arguments, if any, are
1228 * parsed and a function which handles this comment is called.
1229 */
1230 static void
1231 comment(void)
1232 {
1233 int c, lc;
1234 static struct {
1235 const char *keywd;
1236 int arg;
1237 void (*func)(int);
1238 } keywtab[] = {
1239 { "ARGSUSED", 1, argsused },
1240 { "BITFIELDTYPE", 0, bitfieldtype },
1241 { "CONSTCOND", 0, constcond },
1242 { "CONSTANTCOND", 0, constcond },
1243 { "CONSTANTCONDITION", 0, constcond },
1244 { "FALLTHRU", 0, fallthru },
1245 { "FALLTHROUGH", 0, fallthru },
1246 { "LINTLIBRARY", 0, lintlib },
1247 { "LINTED", 1, linted },
1248 { "LONGLONG", 0, longlong },
1249 { "NOSTRICT", 1, linted },
1250 { "NOTREACHED", 0, notreach },
1251 { "PRINTFLIKE", 1, printflike },
1252 { "PROTOLIB", 1, protolib },
1253 { "SCANFLIKE", 1, scanflike },
1254 { "VARARGS", 1, varargs },
1255 };
1256 char keywd[32];
1257 char arg[32];
1258 size_t l, i;
1259 int a;
1260 bool eoc;
1261
1262 eoc = false;
1263
1264 /* Skip white spaces after the start of the comment */
1265 while ((c = inpc()) != EOF && isspace(c))
1266 continue;
1267
1268 /* Read the potential keyword to keywd */
1269 l = 0;
1270 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1271 keywd[l++] = (char)c;
1272 c = inpc();
1273 }
1274 keywd[l] = '\0';
1275
1276 /* look for the keyword */
1277 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1278 if (strcmp(keywtab[i].keywd, keywd) == 0)
1279 break;
1280 }
1281 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1282 goto skip_rest;
1283
1284 /* skip white spaces after the keyword */
1285 while (c != EOF && isspace(c))
1286 c = inpc();
1287
1288 /* read the argument, if the keyword accepts one and there is one */
1289 l = 0;
1290 if (keywtab[i].arg) {
1291 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1292 arg[l++] = (char)c;
1293 c = inpc();
1294 }
1295 }
1296 arg[l] = '\0';
1297 a = l != 0 ? atoi(arg) : -1;
1298
1299 /* skip white spaces after the argument */
1300 while (c != EOF && isspace(c))
1301 c = inpc();
1302
1303 if (c != '*' || (c = inpc()) != '/') {
1304 if (keywtab[i].func != linted)
1305 /* extra characters in lint comment */
1306 warning(257);
1307 } else {
1308 /*
1309 * remember that we have already found the end of the
1310 * comment
1311 */
1312 eoc = true;
1313 }
1314
1315 if (keywtab[i].func != NULL)
1316 (*keywtab[i].func)(a);
1317
1318 skip_rest:
1319 while (!eoc) {
1320 lc = c;
1321 if ((c = inpc()) == EOF) {
1322 /* unterminated comment */
1323 error(256);
1324 break;
1325 }
1326 if (lc == '*' && c == '/')
1327 eoc = true;
1328 }
1329 }
1330
1331 /*
1332 * Handle // style comments
1333 */
1334 static void
1335 slashslashcomment(void)
1336 {
1337 int c;
1338
1339 if (!Sflag && !gflag)
1340 /* %s C does not support // comments */
1341 gnuism(312, tflag ? "traditional" : "ANSI");
1342
1343 while ((c = inpc()) != EOF && c != '\n')
1344 continue;
1345 }
1346
1347 /*
1348 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1349 * clear_warn_flags() is called after function definitions and global and
1350 * local declarations and definitions. It is also called between
1351 * the controlling expression and the body of control statements
1352 * (if, switch, for, while).
1353 */
1354 void
1355 clear_warn_flags(void)
1356 {
1357
1358 lwarn = LWARN_ALL;
1359 quadflg = false;
1360 constcond_flag = false;
1361 }
1362
1363 /*
1364 * Strings are stored in a dynamically allocated buffer and passed
1365 * in yylval.y_xstrg to the parser. The parser or the routines called
1366 * by the parser are responsible for freeing this buffer.
1367 */
1368 static int
1369 string(void)
1370 {
1371 u_char *s;
1372 int c;
1373 size_t len, max;
1374 strg_t *strg;
1375
1376 s = xmalloc(max = 64);
1377
1378 len = 0;
1379 while ((c = getescc('"')) >= 0) {
1380 /* +1 to reserve space for a trailing NUL character */
1381 if (len + 1 == max)
1382 s = xrealloc(s, max *= 2);
1383 s[len++] = (char)c;
1384 }
1385 s[len] = '\0';
1386 if (c == -2)
1387 /* unterminated string constant */
1388 error(258);
1389
1390 strg = xcalloc(1, sizeof (strg_t));
1391 strg->st_tspec = CHAR;
1392 strg->st_len = len;
1393 strg->st_cp = s;
1394
1395 yylval.y_string = strg;
1396 return T_STRING;
1397 }
1398
1399 static int
1400 wcstrg(void)
1401 {
1402 char *s;
1403 int c, n;
1404 size_t i, wi;
1405 size_t len, max, wlen;
1406 wchar_t *ws;
1407 strg_t *strg;
1408
1409 s = xmalloc(max = 64);
1410 len = 0;
1411 while ((c = getescc('"')) >= 0) {
1412 /* +1 to save space for a trailing NUL character */
1413 if (len + 1 >= max)
1414 s = xrealloc(s, max *= 2);
1415 s[len++] = (char)c;
1416 }
1417 s[len] = '\0';
1418 if (c == -2)
1419 /* unterminated string constant */
1420 error(258);
1421
1422 /* get length of wide-character string */
1423 (void)mblen(NULL, 0);
1424 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1425 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1426 /* invalid multibyte character */
1427 error(291);
1428 break;
1429 }
1430 if (n == 0)
1431 n = 1;
1432 }
1433
1434 ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1435
1436 /* convert from multibyte to wide char */
1437 (void)mbtowc(NULL, NULL, 0);
1438 for (i = 0, wi = 0; i < len; i += n, wi++) {
1439 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1440 break;
1441 if (n == 0)
1442 n = 1;
1443 }
1444 ws[wi] = 0;
1445 free(s);
1446
1447 strg = xcalloc(1, sizeof (strg_t));
1448 strg->st_tspec = WCHAR;
1449 strg->st_len = wlen;
1450 strg->st_wcp = ws;
1451
1452 yylval.y_string = strg;
1453 return T_STRING;
1454 }
1455
1456 /*
1457 * As noted above the scanner does not create new symbol table entries
1458 * for symbols it cannot find in the symbol table. This is to avoid
1459 * putting undeclared symbols into the symbol table if a syntax error
1460 * occurs.
1461 *
1462 * getsym() is called as soon as it is probably ok to put the symbol to
1463 * the symbol table. This does not mean that it is not possible that
1464 * symbols are put to the symbol table which are than not completely
1465 * declared due to syntax errors. To avoid too many problems in this
1466 * case symbols get type int in getsym().
1467 *
1468 * XXX calls to getsym() should be delayed until decl1*() is called
1469 */
1470 sym_t *
1471 getsym(sbuf_t *sb)
1472 {
1473 dinfo_t *di;
1474 char *s;
1475 sym_t *sym;
1476
1477 sym = sb->sb_sym;
1478
1479 /*
1480 * During member declaration it is possible that name() looked
1481 * for symbols of type FVFT, although it should have looked for
1482 * symbols of type FTAG. Same can happen for labels. Both cases
1483 * are compensated here.
1484 */
1485 if (symtyp == FMEMBER || symtyp == FLABEL) {
1486 if (sym == NULL || sym->s_kind == FVFT)
1487 sym = search(sb);
1488 }
1489
1490 if (sym != NULL) {
1491 if (sym->s_kind != symtyp)
1492 LERROR("getsym(%d, %d)", sym->s_kind, symtyp);
1493 symtyp = FVFT;
1494 freesb(sb);
1495 return sym;
1496 }
1497
1498 /* create a new symbol table entry */
1499
1500 /* labels must always be allocated at level 1 (outhermost block) */
1501 if (symtyp == FLABEL) {
1502 sym = getlblk(1, sizeof (sym_t));
1503 s = getlblk(1, sb->sb_len + 1);
1504 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1505 sym->s_name = s;
1506 sym->s_blklev = 1;
1507 di = dcs;
1508 while (di->d_next != NULL && di->d_next->d_next != NULL)
1509 di = di->d_next;
1510 lint_assert(di->d_ctx == AUTO);
1511 } else {
1512 sym = getblk(sizeof (sym_t));
1513 sym->s_name = sb->sb_name;
1514 sym->s_blklev = blklev;
1515 di = dcs;
1516 }
1517
1518 UNIQUE_CURR_POS(sym->s_def_pos);
1519 if ((sym->s_kind = symtyp) != FLABEL)
1520 sym->s_type = gettyp(INT);
1521
1522 symtyp = FVFT;
1523
1524 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1525 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1526 sym->s_rlink = &symtab[sb->sb_hash];
1527 symtab[sb->sb_hash] = sym;
1528
1529 *di->d_ldlsym = sym;
1530 di->d_ldlsym = &sym->s_dlnxt;
1531
1532 freesb(sb);
1533 return sym;
1534 }
1535
1536 /*
1537 * Construct a temporary symbol. The symbol starts with a digit, so that
1538 * it is illegal.
1539 */
1540 sym_t *
1541 mktempsym(type_t *t)
1542 {
1543 static int n = 0;
1544 int h;
1545 char *s = getlblk(blklev, 64);
1546 sym_t *sym = getblk(sizeof (sym_t));
1547
1548 (void)snprintf(s, 64, "%.8d_tmp", n++);
1549 h = hash(s);
1550
1551 sym->s_name = s;
1552 sym->s_type = t;
1553 sym->s_blklev = blklev;
1554 sym->s_scl = AUTO;
1555 sym->s_kind = FVFT;
1556 sym->s_used = true;
1557 sym->s_set = true;
1558
1559 if ((sym->s_link = symtab[h]) != NULL)
1560 symtab[h]->s_rlink = &sym->s_link;
1561 sym->s_rlink = &symtab[h];
1562 symtab[h] = sym;
1563
1564 *dcs->d_ldlsym = sym;
1565 dcs->d_ldlsym = &sym->s_dlnxt;
1566
1567 return sym;
1568 }
1569
1570 /*
1571 * Remove a symbol forever from the symbol table. s_blklev
1572 * is set to -1 to avoid that the symbol will later be put
1573 * back to the symbol table.
1574 */
1575 void
1576 rmsym(sym_t *sym)
1577 {
1578
1579 if ((*sym->s_rlink = sym->s_link) != NULL)
1580 sym->s_link->s_rlink = sym->s_rlink;
1581 sym->s_blklev = -1;
1582 sym->s_link = NULL;
1583 }
1584
1585 /*
1586 * Remove a list of symbols declared at one level from the symbol
1587 * table.
1588 */
1589 void
1590 rmsyms(sym_t *syms)
1591 {
1592 sym_t *sym;
1593
1594 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1595 if (sym->s_blklev != -1) {
1596 if ((*sym->s_rlink = sym->s_link) != NULL)
1597 sym->s_link->s_rlink = sym->s_rlink;
1598 sym->s_link = NULL;
1599 sym->s_rlink = NULL;
1600 }
1601 }
1602 }
1603
1604 /*
1605 * Put a symbol into the symbol table
1606 */
1607 void
1608 inssym(int bl, sym_t *sym)
1609 {
1610 int h;
1611
1612 h = hash(sym->s_name);
1613 if ((sym->s_link = symtab[h]) != NULL)
1614 symtab[h]->s_rlink = &sym->s_link;
1615 sym->s_rlink = &symtab[h];
1616 symtab[h] = sym;
1617 sym->s_blklev = bl;
1618 lint_assert(sym->s_link == NULL ||
1619 sym->s_blklev >= sym->s_link->s_blklev);
1620 }
1621
1622 /*
1623 * Called at level 0 after syntax errors
1624 * Removes all symbols which are not declared at level 0 from the
1625 * symbol table. Also frees all memory which is not associated with
1626 * level 0.
1627 */
1628 void
1629 cleanup(void)
1630 {
1631 sym_t *sym, *nsym;
1632 int i;
1633
1634 for (i = 0; i < HSHSIZ1; i++) {
1635 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1636 nsym = sym->s_link;
1637 if (sym->s_blklev >= 1) {
1638 if ((*sym->s_rlink = nsym) != NULL)
1639 nsym->s_rlink = sym->s_rlink;
1640 }
1641 }
1642 }
1643
1644 for (i = mblklev; i > 0; i--)
1645 freelblk(i);
1646 }
1647
1648 /*
1649 * Create a new symbol with the name of an existing symbol.
1650 */
1651 sym_t *
1652 pushdown(sym_t *sym)
1653 {
1654 int h;
1655 sym_t *nsym;
1656
1657 h = hash(sym->s_name);
1658 nsym = getblk(sizeof (sym_t));
1659 lint_assert(sym->s_blklev <= blklev);
1660 nsym->s_name = sym->s_name;
1661 UNIQUE_CURR_POS(nsym->s_def_pos);
1662 nsym->s_kind = sym->s_kind;
1663 nsym->s_blklev = blklev;
1664
1665 if ((nsym->s_link = symtab[h]) != NULL)
1666 symtab[h]->s_rlink = &nsym->s_link;
1667 nsym->s_rlink = &symtab[h];
1668 symtab[h] = nsym;
1669
1670 *dcs->d_ldlsym = nsym;
1671 dcs->d_ldlsym = &nsym->s_dlnxt;
1672
1673 return nsym;
1674 }
1675
1676 /*
1677 * Free any dynamically allocated memory referenced by
1678 * the value stack or yylval.
1679 * The type of information in yylval is described by tok.
1680 */
1681 void
1682 freeyyv(void *sp, int tok)
1683 {
1684 if (tok == T_NAME || tok == T_TYPENAME) {
1685 sbuf_t *sb = *(sbuf_t **)sp;
1686 freesb(sb);
1687 } else if (tok == T_CON) {
1688 val_t *val = *(val_t **)sp;
1689 free(val);
1690 } else if (tok == T_STRING) {
1691 strg_t *strg = *(strg_t **)sp;
1692 if (strg->st_tspec == CHAR) {
1693 free(strg->st_cp);
1694 } else {
1695 lint_assert(strg->st_tspec == WCHAR);
1696 free(strg->st_wcp);
1697 }
1698 free(strg);
1699 }
1700 }
1701