scan.l revision 1.84 1 %{
2 /* $NetBSD: scan.l,v 1.84 2018/10/07 14:20:01 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 1.84 2018/10/07 14:20:01 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 { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 },
213 { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 },
214 { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 },
215 { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 },
216 { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 },
217 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 },
218 { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 },
219 { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 },
220 { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 },
221 { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 },
222 { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 },
223 { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 },
224 { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 },
225 { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 },
226 { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 },
227 { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 },
228 { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 },
229 { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 },
230 { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 },
231 { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 },
232 { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 },
233 { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 },
234 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 },
235 { "do", T_DO, 0, 0, 0, 0,0,0,0,1 },
236 { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 },
237 { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 },
238 { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 },
239 { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 },
240 { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 },
241 { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 },
242 { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 },
243 { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 },
244 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 },
245 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 },
246 { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 },
247 { "if", T_IF, 0, 0, 0, 0,0,0,0,1 },
248 { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 },
249 { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 },
250 { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 },
251 { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 },
252 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 },
253 { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 },
254 { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 },
255 { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION,
256 0, 0, 0, 0,0,1,1,5 },
257 { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 },
258 { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 },
259 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 },
260 { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 },
261 { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 },
262 { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 },
263 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 },
264 { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 },
265 { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 },
266 { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 },
267 { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 },
268 { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 },
269 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 },
270 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 },
271 { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 },
272 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 },
273 { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 },
274 { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 },
275 { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 },
276 { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 },
277 { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 },
278 { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 },
279 { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 },
280 { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 },
281 { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 },
282 { "symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0,0,0,0,2 },
283 { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 },
284 { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 },
285 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 },
286 { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 },
287 { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 },
288 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 },
289 { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 },
290 { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 },
291 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 },
292 { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 },
293 { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 },
294 { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 },
295 { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 },
296 { NULL, 0, 0, 0, 0, 0,0,0,0,0 }
297 };
298
299 /* Symbol table */
300 static sym_t *symtab[HSHSIZ1];
301
302 /* bit i of the entry with index i is set */
303 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
304
305 /* least significant i bits are set in the entry with index i */
306 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
307
308 /* least significant i bits are not set in the entry with index i */
309 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
310
311 /* free list for sbuf structures */
312 static sbuf_t *sbfrlst;
313
314 /* Typ of next expected symbol */
315 symt_t symtyp;
316
317
318 static void
319 addkw(struct kwtab *kw, int deco)
320 {
321 sym_t *sym;
322 size_t h;
323 char buf[256];
324 const char *name;
325
326 if (!(kw->kw_deco & deco))
327 return;
328
329 switch (deco) {
330 case 1:
331 name = kw->kw_name;
332 break;
333 case 2:
334 snprintf(buf, sizeof(buf), "__%s", kw->kw_name);
335 name = strdup(buf);
336 break;
337 case 4:
338 snprintf(buf, sizeof(buf), "__%s__", kw->kw_name);
339 name = strdup(buf);
340 break;
341 default:
342 abort();
343 }
344
345 if (name == NULL)
346 err(1, "Can't init symbol table");
347
348 sym = getblk(sizeof (sym_t));
349 sym->s_name = name;
350 sym->s_keyw = kw;
351 sym->s_value.v_quad = kw->kw_token;
352 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
353 sym->s_tspec = kw->kw_tspec;
354 } else if (kw->kw_token == T_SCLASS) {
355 sym->s_scl = kw->kw_scl;
356 } else if (kw->kw_token == T_QUAL) {
357 sym->s_tqual = kw->kw_tqual;
358 }
359 h = hash(sym->s_name);
360 if ((sym->s_link = symtab[h]) != NULL)
361 symtab[h]->s_rlink = &sym->s_link;
362 sym->s_rlink = &symtab[h];
363 symtab[h] = sym;
364 }
365
366 /*
367 * All keywords are written to the symbol table. This saves us looking
368 * in a extra table for each name we found.
369 */
370 void
371 initscan(void)
372 {
373 struct kwtab *kw;
374 size_t i;
375 uint64_t uq;
376
377 for (kw = kwtab; kw->kw_name != NULL; kw++) {
378 if ((kw->kw_c89 || kw->kw_c99) && tflag)
379 continue;
380 if (kw->kw_c99 && !(Sflag || gflag))
381 continue;
382 if (kw->kw_gcc && !gflag)
383 continue;
384 addkw(kw, 1);
385 addkw(kw, 2);
386 addkw(kw, 4);
387 }
388
389 /* initialize bit-masks for quads */
390 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
391 qbmasks[i] = (uint64_t)1 << i;
392 uq = ~(uint64_t)0 << i;
393 qumasks[i] = uq;
394 qlmasks[i] = ~uq;
395 }
396 qumasks[i] = 0;
397 qlmasks[i] = ~(uint64_t)0;
398 }
399
400 /*
401 * Get a free sbuf structure, if possible from the free list
402 */
403 static sbuf_t *
404 allocsb(void)
405 {
406 sbuf_t *sb;
407
408 if ((sb = sbfrlst) != NULL) {
409 sbfrlst = sb->sb_nxt;
410 #ifdef BLKDEBUG
411 (void)memset(sb, 0, sizeof (*sb));
412 #else
413 sb->sb_nxt = NULL;
414 #endif
415 } else {
416 sb = xmalloc(sizeof (sbuf_t));
417 (void)memset(sb, 0, sizeof (*sb));
418 }
419 return (sb);
420 }
421
422 /*
423 * Put a sbuf structure to the free list
424 */
425 static void
426 freesb(sbuf_t *sb)
427 {
428
429 (void)memset(sb, ZERO, sizeof (*sb));
430 sb->sb_nxt = sbfrlst;
431 sbfrlst = sb;
432 }
433
434 /*
435 * Read a character and ensure that it is positive (except EOF).
436 * Increment line count(s) if necessary.
437 */
438 static int
439 inpc(void)
440 {
441 int c;
442
443 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
444 incline();
445 return (c);
446 }
447
448 static int
449 hash(const char *s)
450 {
451 u_int v;
452 const u_char *us;
453
454 v = 0;
455 for (us = (const u_char *)s; *us != '\0'; us++) {
456 v = (v << sizeof (v)) + *us;
457 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
458 }
459 return (v % HSHSIZ1);
460 }
461
462 /*
463 * Lex has found a letter followed by zero or more letters or digits.
464 * It looks for a symbol in the symbol table with the same name. This
465 * symbol must either be a keyword or a symbol of the type required by
466 * symtyp (label, member, tag, ...).
467 *
468 * If it is a keyword, the token is returned. In some cases it is described
469 * more deeply by data written to yylval.
470 *
471 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
472 * is stored in yylval. This struct contains the name of the symbol, its
473 * length and hash value. If there is already a symbol of the same name
474 * and type in the symbol table, the sbuf struct also contains a pointer
475 * to the symbol table entry.
476 */
477 static int
478 name(void)
479 {
480 char *s;
481 sbuf_t *sb;
482 sym_t *sym;
483 int tok;
484
485 sb = allocsb();
486 sb->sb_name = yytext;
487 sb->sb_len = yyleng;
488 sb->sb_hash = hash(yytext);
489 if ((sym = search(sb)) != NULL && sym->s_keyw) {
490 freesb(sb);
491 return (keyw(sym));
492 }
493
494 sb->sb_sym = sym;
495
496 if (sym != NULL) {
497 if (blklev < sym->s_blklev)
498 LERROR("name()");
499 sb->sb_name = sym->s_name;
500 sb->sb_len = strlen(sym->s_name);
501 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
502 } else {
503 s = getblk(yyleng + 1);
504 (void)memcpy(s, yytext, yyleng + 1);
505 sb->sb_name = s;
506 sb->sb_len = yyleng;
507 tok = T_NAME;
508 }
509
510 yylval.y_sb = sb;
511 return (tok);
512 }
513
514 static sym_t *
515 search(sbuf_t *sb)
516 {
517 sym_t *sym;
518
519 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
520 if (strcmp(sym->s_name, sb->sb_name) == 0) {
521 if (sym->s_keyw) {
522 struct kwtab *kw = sym->s_keyw;
523 if (!kw->kw_attr || attron)
524 return (sym);
525 } else if (!attron && sym->s_kind == symtyp)
526 return sym;
527 }
528 }
529
530 return (NULL);
531 }
532
533 static int
534 keyw(sym_t *sym)
535 {
536 int t;
537
538 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
539 yylval.y_scl = sym->s_scl;
540 } else if (t == T_TYPE || t == T_SOU) {
541 yylval.y_tspec = sym->s_tspec;
542 } else if (t == T_QUAL) {
543 yylval.y_tqual = sym->s_tqual;
544 }
545 return (t);
546 }
547
548 /*
549 * Convert a string representing an integer into internal representation.
550 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
551 */
552 static int
553 icon(int base)
554 {
555 int l_suffix, u_suffix;
556 int len;
557 const char *cp;
558 char c, *eptr;
559 tspec_t typ;
560 int ansiu;
561 #ifdef TARG_INT128_MAX
562 __uint128_t uq = 0;
563 static tspec_t contypes[2][4] = {
564 { INT, LONG, QUAD, INT128, },
565 { UINT, ULONG, UQUAD, UINT128, }
566 };
567 #else
568 uint64_t uq = 0;
569 static tspec_t contypes[2][3] = {
570 { INT, LONG, QUAD, },
571 { UINT, ULONG, UQUAD, }
572 };
573 #endif
574
575 cp = yytext;
576 len = yyleng;
577
578 /* skip 0[xX] or 0[bB] */
579 if (base == 16 || base == 2) {
580 cp += 2;
581 len -= 2;
582 }
583
584 /* read suffixes */
585 l_suffix = u_suffix = 0;
586 for ( ; ; ) {
587 if ((c = cp[len - 1]) == 'l' || c == 'L') {
588 l_suffix++;
589 } else if (c == 'u' || c == 'U') {
590 u_suffix++;
591 } else {
592 break;
593 }
594 len--;
595 }
596 if (l_suffix > 2 || u_suffix > 1) {
597 /* malformed integer constant */
598 warning(251);
599 if (l_suffix > 2)
600 l_suffix = 2;
601 if (u_suffix > 1)
602 u_suffix = 1;
603 }
604 if (tflag && u_suffix != 0) {
605 /* suffix U is illegal in traditional C */
606 warning(97);
607 }
608 typ = contypes[u_suffix][l_suffix];
609
610 errno = 0;
611
612 uq = strtouq(cp, &eptr, base);
613 if (eptr != cp + len)
614 LERROR("icon()");
615 if (errno != 0)
616 /* integer constant out of range */
617 warning(252);
618
619 /*
620 * If the value is too big for the current type, we must choose
621 * another type.
622 */
623 ansiu = 0;
624 switch (typ) {
625 case INT:
626 if (uq <= TARG_INT_MAX) {
627 /* ok */
628 } else if (uq <= TARG_UINT_MAX && base != 10) {
629 typ = UINT;
630 } else if (uq <= TARG_LONG_MAX) {
631 typ = LONG;
632 } else {
633 typ = ULONG;
634 if (uq > TARG_ULONG_MAX) {
635 /* integer constant out of range */
636 warning(252);
637 }
638 }
639 if (typ == UINT || typ == ULONG) {
640 if (tflag) {
641 typ = LONG;
642 } else if (!sflag) {
643 /*
644 * Remember that the constant is unsigned
645 * only in ANSI C
646 */
647 ansiu = 1;
648 }
649 }
650 break;
651 case UINT:
652 if (uq > TARG_UINT_MAX) {
653 typ = ULONG;
654 if (uq > TARG_ULONG_MAX) {
655 /* integer constant out of range */
656 warning(252);
657 }
658 }
659 break;
660 case LONG:
661 if (uq > TARG_LONG_MAX && !tflag) {
662 typ = ULONG;
663 if (!sflag)
664 ansiu = 1;
665 if (uq > TARG_ULONG_MAX) {
666 /* integer constant out of range */
667 warning(252);
668 }
669 }
670 break;
671 case ULONG:
672 if (uq > TARG_ULONG_MAX) {
673 /* integer constant out of range */
674 warning(252);
675 }
676 break;
677 case QUAD:
678 if (uq > TARG_QUAD_MAX && !tflag) {
679 typ = UQUAD;
680 if (!sflag)
681 ansiu = 1;
682 }
683 break;
684 case UQUAD:
685 if (uq > TARG_UQUAD_MAX) {
686 /* integer constant out of range */
687 warning(252);
688 }
689 break;
690 #ifdef INT128_SIZE
691 case INT128:
692 #ifdef TARG_INT128_MAX
693 if (uq > TARG_INT128_MAX && !tflag) {
694 typ = UINT128;
695 if (!sflag)
696 ansiu = 1;
697 }
698 #endif
699 break;
700 case UINT128:
701 #ifdef TARG_INT128_MAX
702 if (uq > TARG_UINT128_MAX) {
703 /* integer constant out of range */
704 warning(252);
705 }
706 #endif
707 break;
708 #endif
709 /* LINTED206: (enumeration values not handled in switch) */
710 case STRUCT:
711 case VOID:
712 case LDOUBLE:
713 case FUNC:
714 case ARRAY:
715 case PTR:
716 case ENUM:
717 case UNION:
718 case SIGNED:
719 case NOTSPEC:
720 case DOUBLE:
721 case FLOAT:
722 case USHORT:
723 case SHORT:
724 case UCHAR:
725 case SCHAR:
726 case CHAR:
727 case BOOL:
728 case UNSIGN:
729 case FCOMPLEX:
730 case DCOMPLEX:
731 case LCOMPLEX:
732 case COMPLEX:
733 break;
734
735 case NTSPEC: /* this value unused */
736 break;
737 }
738
739 uq = (uint64_t)xsign((int64_t)uq, typ, -1);
740
741 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
742 yylval.y_val->v_ansiu = ansiu;
743 yylval.y_val->v_quad = (int64_t)uq;
744
745 return (T_CON);
746 }
747
748 /*
749 * Returns 1 if t is a signed type and the value is negative.
750 *
751 * len is the number of significant bits. If len is -1, len is set
752 * to the width of type t.
753 */
754 int
755 sign(int64_t q, tspec_t t, int len)
756 {
757
758 if (t == PTR || isutyp(t))
759 return (0);
760 return (msb(q, t, len));
761 }
762
763 int
764 msb(int64_t q, tspec_t t, int len)
765 {
766
767 if (len <= 0)
768 len = size(t);
769 return ((q & qbmasks[len - 1]) != 0);
770 }
771
772 /*
773 * Extends the sign of q.
774 */
775 int64_t
776 xsign(int64_t q, tspec_t t, int len)
777 {
778
779 if (len <= 0)
780 len = size(t);
781
782 if (t == PTR || isutyp(t) || !sign(q, t, len)) {
783 q &= qlmasks[len];
784 } else {
785 q |= qumasks[len];
786 }
787 return (q);
788 }
789
790 /*
791 * Convert a string representing a floating point value into its interal
792 * representation. Type and value are returned in yylval. fcon()
793 * (and yylex()) returns T_CON.
794 * XXX Currently it is not possible to convert constants of type
795 * long double which are greater than DBL_MAX.
796 */
797 static int
798 fcon(void)
799 {
800 const char *cp;
801 int len;
802 tspec_t typ;
803 char c, *eptr;
804 double d;
805 float f = 0;
806
807 cp = yytext;
808 len = yyleng;
809
810 if (cp[len - 1] == 'i') {
811 /* imaginary, do nothing for now */
812 len--;
813 }
814 if ((c = cp[len - 1]) == 'f' || c == 'F') {
815 typ = FLOAT;
816 len--;
817 } else if (c == 'l' || c == 'L') {
818 typ = LDOUBLE;
819 len--;
820 } else {
821 if (c == 'd' || c == 'D')
822 len--;
823 typ = DOUBLE;
824 }
825
826 if (tflag && typ != DOUBLE) {
827 /* suffixes F and L are illegal in traditional C */
828 warning(98);
829 }
830
831 errno = 0;
832 d = strtod(cp, &eptr);
833 if (eptr != cp + len) {
834 switch (*eptr) {
835 /*
836 * XXX: non-native non-current strtod() may not handle hex
837 * floats, ignore the rest if we find traces of hex float
838 * syntax...
839 */
840 case 'p':
841 case 'P':
842 case 'x':
843 case 'X':
844 d = 0;
845 errno = 0;
846 break;
847 default:
848 LERROR("fcon(%s->%s)", cp, eptr);
849 }
850 }
851 if (errno != 0)
852 /* floating-point constant out of range */
853 warning(248);
854
855 if (typ == FLOAT) {
856 f = (float)d;
857 if (!finite(f)) {
858 /* floating-point constant out of range */
859 warning(248);
860 f = f > 0 ? FLT_MAX : -FLT_MAX;
861 }
862 }
863
864 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
865 if (typ == FLOAT) {
866 yylval.y_val->v_ldbl = f;
867 } else {
868 yylval.y_val->v_ldbl = d;
869 }
870
871 return (T_CON);
872 }
873
874 static int
875 operator(int t, op_t o)
876 {
877
878 yylval.y_op = o;
879 return (t);
880 }
881
882 /*
883 * Called if lex found a leading \'.
884 */
885 static int
886 ccon(void)
887 {
888 size_t n;
889 int val, c;
890 char cv;
891
892 n = 0;
893 val = 0;
894 while ((c = getescc('\'')) >= 0) {
895 val = (val << CHAR_BIT) + c;
896 n++;
897 }
898 if (c == -2) {
899 /* unterminated character constant */
900 error(253);
901 } else {
902 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
903 /* too many characters in character constant */
904 error(71);
905 } else if (n > 1) {
906 /* multi-character character constant */
907 warning(294);
908 } else if (n == 0) {
909 /* empty character constant */
910 error(73);
911 }
912 }
913 if (n == 1) {
914 cv = (char)val;
915 val = cv;
916 }
917
918 yylval.y_val = xcalloc(1, sizeof (val_t));
919 yylval.y_val->v_tspec = INT;
920 yylval.y_val->v_quad = val;
921
922 return (T_CON);
923 }
924
925 /*
926 * Called if lex found a leading L\'
927 */
928 static int
929 wccon(void)
930 {
931 static char buf[MB_LEN_MAX + 1];
932 size_t i;
933 int c;
934 wchar_t wc;
935
936 i = 0;
937 while ((c = getescc('\'')) >= 0) {
938 if (i < MB_CUR_MAX)
939 buf[i] = (char)c;
940 i++;
941 }
942
943 wc = 0;
944
945 if (c == -2) {
946 /* unterminated character constant */
947 error(253);
948 } else if (c == 0) {
949 /* empty character constant */
950 error(73);
951 } else {
952 if (i > MB_CUR_MAX) {
953 i = MB_CUR_MAX;
954 /* too many characters in character constant */
955 error(71);
956 } else {
957 buf[i] = '\0';
958 (void)mbtowc(NULL, NULL, 0);
959 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
960 /* invalid multibyte character */
961 error(291);
962 }
963 }
964
965 yylval.y_val = xcalloc(1, sizeof (val_t));
966 yylval.y_val->v_tspec = WCHAR;
967 yylval.y_val->v_quad = wc;
968
969 return (T_CON);
970 }
971
972 /*
973 * Read a character which is part of a character constant or of a string
974 * and handle escapes.
975 *
976 * The Argument is the character which delimits the character constant or
977 * string.
978 *
979 * Returns -1 if the end of the character constant or string is reached,
980 * -2 if the EOF is reached, and the character otherwise.
981 */
982 static int
983 getescc(int d)
984 {
985 static int pbc = -1;
986 int n, c, v;
987
988 if (pbc == -1) {
989 c = inpc();
990 } else {
991 c = pbc;
992 pbc = -1;
993 }
994 if (c == d)
995 return (-1);
996 switch (c) {
997 case '\n':
998 if (tflag) {
999 /* newline in string or char constant */
1000 error(254);
1001 return (-2);
1002 }
1003 return (c);
1004 case EOF:
1005 return (-2);
1006 case '\\':
1007 switch (c = inpc()) {
1008 case '"':
1009 if (tflag && d == '\'')
1010 /* \" inside character constant undef. ... */
1011 warning(262);
1012 return ('"');
1013 case '\'':
1014 return ('\'');
1015 case '?':
1016 if (tflag)
1017 /* \? undefined in traditional C */
1018 warning(263);
1019 return ('?');
1020 case '\\':
1021 return ('\\');
1022 case 'a':
1023 if (tflag)
1024 /* \a undefined in traditional C */
1025 warning(81);
1026 return ('\a');
1027 case 'b':
1028 return ('\b');
1029 case 'f':
1030 return ('\f');
1031 case 'n':
1032 return ('\n');
1033 case 'r':
1034 return ('\r');
1035 case 't':
1036 return ('\t');
1037 case 'v':
1038 if (tflag)
1039 /* \v undefined in traditional C */
1040 warning(264);
1041 return ('\v');
1042 case '8': case '9':
1043 /* bad octal digit %c */
1044 warning(77, c);
1045 /* FALLTHROUGH */
1046 case '0': case '1': case '2': case '3':
1047 case '4': case '5': case '6': case '7':
1048 n = 3;
1049 v = 0;
1050 do {
1051 v = (v << 3) + (c - '0');
1052 c = inpc();
1053 } while (--n && isdigit(c) && (tflag || c <= '7'));
1054 if (tflag && n > 0 && isdigit(c))
1055 /* bad octal digit %c */
1056 warning(77, c);
1057 pbc = c;
1058 if (v > TARG_UCHAR_MAX) {
1059 /* character escape does not fit in char. */
1060 warning(76);
1061 v &= CHAR_MASK;
1062 }
1063 return (v);
1064 case 'x':
1065 if (tflag)
1066 /* \x undefined in traditional C */
1067 warning(82);
1068 v = 0;
1069 n = 0;
1070 while ((c = inpc()) >= 0 && isxdigit(c)) {
1071 c = isdigit(c) ?
1072 c - '0' : toupper(c) - 'A' + 10;
1073 v = (v << 4) + c;
1074 if (n >= 0) {
1075 if ((v & ~CHAR_MASK) != 0) {
1076 /* overflow in hex escape */
1077 warning(75);
1078 n = -1;
1079 } else {
1080 n++;
1081 }
1082 }
1083 }
1084 pbc = c;
1085 if (n == 0) {
1086 /* no hex digits follow \x */
1087 error(74);
1088 } if (n == -1) {
1089 v &= CHAR_MASK;
1090 }
1091 return (v);
1092 case '\n':
1093 return (getescc(d));
1094 case EOF:
1095 return (-2);
1096 default:
1097 if (isprint(c)) {
1098 /* dubious escape \%c */
1099 warning(79, c);
1100 } else {
1101 /* dubious escape \%o */
1102 warning(80, c);
1103 }
1104 }
1105 }
1106 return (c);
1107 }
1108
1109 /*
1110 * Called for preprocessor directives. Currently implemented are:
1111 * # lineno
1112 * # lineno "filename"
1113 */
1114 static void
1115 directive(void)
1116 {
1117 const char *cp, *fn;
1118 char c, *eptr;
1119 size_t fnl;
1120 long ln;
1121 static int first = 1;
1122
1123 /* Go to first non-whitespace after # */
1124 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1125 continue;
1126
1127 if (!isdigit((unsigned char)c)) {
1128 if (strncmp(cp, "pragma", 6) == 0
1129 && isspace((unsigned char)cp[6]))
1130 return;
1131 error:
1132 /* undefined or invalid # directive */
1133 warning(255);
1134 return;
1135 }
1136 ln = strtol(--cp, &eptr, 10);
1137 if (cp == eptr)
1138 goto error;
1139 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1140 goto error;
1141 while ((c = *cp++) == ' ' || c == '\t')
1142 continue;
1143 if (c != '\0') {
1144 if (c != '"')
1145 goto error;
1146 fn = cp;
1147 while ((c = *cp) != '"' && c != '\0')
1148 cp++;
1149 if (c != '"')
1150 goto error;
1151 if ((fnl = cp++ - fn) > PATH_MAX)
1152 goto error;
1153 while ((c = *cp++) == ' ' || c == '\t')
1154 continue;
1155 #if 0
1156 if (c != '\0')
1157 warning("extra character(s) after directive");
1158 #endif
1159
1160 /* empty string means stdin */
1161 if (fnl == 0) {
1162 fn = "{standard input}";
1163 fnl = 16; /* strlen (fn) */
1164 }
1165 curr_pos.p_file = fnnalloc(fn, fnl);
1166 /*
1167 * If this is the first directive, the name is the name
1168 * of the C source file as specified at the command line.
1169 * It is written to the output file.
1170 */
1171 if (first) {
1172 csrc_pos.p_file = curr_pos.p_file;
1173 outsrc(fnxform(curr_pos.p_file,
1174 strlen(curr_pos.p_file)));
1175 first = 0;
1176 }
1177 }
1178 curr_pos.p_line = (int)ln - 1;
1179 curr_pos.p_uniq = 0;
1180 if (curr_pos.p_file == csrc_pos.p_file) {
1181 csrc_pos.p_line = (int)ln - 1;
1182 csrc_pos.p_uniq = 0;
1183 }
1184 }
1185
1186 /*
1187 * Handle lint comments. Following comments are currently understood:
1188 * ARGSUSEDn
1189 * BITFIELDTYPE
1190 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1191 * FALLTHRU FALLTHROUGH
1192 * LINTLIBRARY
1193 * LINTEDn NOSTRICTn
1194 * LONGLONG
1195 * NOTREACHED
1196 * PRINTFLIKEn
1197 * PROTOLIB
1198 * SCANFLIKEn
1199 * VARARGSn
1200 * If one of this comments is recognized, the arguments, if any, are
1201 * parsed and a function which handles this comment is called.
1202 */
1203 static void
1204 comment(void)
1205 {
1206 int c, lc;
1207 static struct {
1208 const char *keywd;
1209 int arg;
1210 void (*func)(int);
1211 } keywtab[] = {
1212 { "ARGSUSED", 1, argsused },
1213 { "BITFIELDTYPE", 0, bitfieldtype },
1214 { "CONSTCOND", 0, constcond },
1215 { "CONSTANTCOND", 0, constcond },
1216 { "CONSTANTCONDITION", 0, constcond },
1217 { "FALLTHRU", 0, fallthru },
1218 { "FALLTHROUGH", 0, fallthru },
1219 { "LINTLIBRARY", 0, lintlib },
1220 { "LINTED", 1, linted },
1221 { "LONGLONG", 0, longlong },
1222 { "NOSTRICT", 1, linted },
1223 { "NOTREACHED", 0, notreach },
1224 { "PRINTFLIKE", 1, printflike },
1225 { "PROTOLIB", 1, protolib },
1226 { "SCANFLIKE", 1, scanflike },
1227 { "VARARGS", 1, varargs },
1228 };
1229 char keywd[32];
1230 char arg[32];
1231 size_t l, i;
1232 int a;
1233 int eoc;
1234
1235 eoc = 0;
1236
1237 /* Skip white spaces after the start of the comment */
1238 while ((c = inpc()) != EOF && isspace(c))
1239 continue;
1240
1241 /* Read the potential keyword to keywd */
1242 l = 0;
1243 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1244 keywd[l++] = (char)c;
1245 c = inpc();
1246 }
1247 keywd[l] = '\0';
1248
1249 /* look for the keyword */
1250 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1251 if (strcmp(keywtab[i].keywd, keywd) == 0)
1252 break;
1253 }
1254 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1255 goto skip_rest;
1256
1257 /* skip white spaces after the keyword */
1258 while (c != EOF && isspace(c))
1259 c = inpc();
1260
1261 /* read the argument, if the keyword accepts one and there is one */
1262 l = 0;
1263 if (keywtab[i].arg) {
1264 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1265 arg[l++] = (char)c;
1266 c = inpc();
1267 }
1268 }
1269 arg[l] = '\0';
1270 a = l != 0 ? atoi(arg) : -1;
1271
1272 /* skip white spaces after the argument */
1273 while (c != EOF && isspace(c))
1274 c = inpc();
1275
1276 if (c != '*' || (c = inpc()) != '/') {
1277 if (keywtab[i].func != linted)
1278 /* extra characters in lint comment */
1279 warning(257);
1280 } else {
1281 /*
1282 * remember that we have already found the end of the
1283 * comment
1284 */
1285 eoc = 1;
1286 }
1287
1288 if (keywtab[i].func != NULL)
1289 (*keywtab[i].func)(a);
1290
1291 skip_rest:
1292 while (!eoc) {
1293 lc = c;
1294 if ((c = inpc()) == EOF) {
1295 /* unterminated comment */
1296 error(256);
1297 break;
1298 }
1299 if (lc == '*' && c == '/')
1300 eoc = 1;
1301 }
1302 }
1303
1304 /*
1305 * Handle // style comments
1306 */
1307 static void
1308 slashslashcomment(void)
1309 {
1310 int c;
1311
1312 if (!Sflag && !gflag)
1313 /* // comments only supported in C99 */
1314 (void)gnuism(312, tflag ? "traditional" : "ANSI");
1315
1316 while ((c = inpc()) != EOF && c != '\n')
1317 continue;
1318 }
1319
1320 /*
1321 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1322 * clrwflgs() is called after function definitions and global and
1323 * local declarations and definitions. It is also called between
1324 * the controlling expression and the body of control statements
1325 * (if, switch, for, while).
1326 */
1327 void
1328 clrwflgs(void)
1329 {
1330
1331 lwarn = LWARN_ALL;
1332 quadflg = 0;
1333 ccflg = 0;
1334 }
1335
1336 /*
1337 * Strings are stored in a dynamically alloceted buffer and passed
1338 * in yylval.y_xstrg to the parser. The parser or the routines called
1339 * by the parser are responsible for freeing this buffer.
1340 */
1341 static int
1342 string(void)
1343 {
1344 u_char *s;
1345 int c;
1346 size_t len, max;
1347 strg_t *strg;
1348
1349 s = xmalloc(max = 64);
1350
1351 len = 0;
1352 while ((c = getescc('"')) >= 0) {
1353 /* +1 to reserve space for a trailing NUL character */
1354 if (len + 1 == max)
1355 s = xrealloc(s, max *= 2);
1356 s[len++] = (char)c;
1357 }
1358 s[len] = '\0';
1359 if (c == -2)
1360 /* unterminated string constant */
1361 error(258);
1362
1363 strg = xcalloc(1, sizeof (strg_t));
1364 strg->st_tspec = CHAR;
1365 strg->st_len = len;
1366 strg->st_cp = s;
1367
1368 yylval.y_strg = strg;
1369 return (T_STRING);
1370 }
1371
1372 static int
1373 wcstrg(void)
1374 {
1375 char *s;
1376 int c, n;
1377 size_t i, wi;
1378 size_t len, max, wlen;
1379 wchar_t *ws;
1380 strg_t *strg;
1381
1382 s = xmalloc(max = 64);
1383 len = 0;
1384 while ((c = getescc('"')) >= 0) {
1385 /* +1 to save space for a trailing NUL character */
1386 if (len + 1 >= max)
1387 s = xrealloc(s, max *= 2);
1388 s[len++] = (char)c;
1389 }
1390 s[len] = '\0';
1391 if (c == -2)
1392 /* unterminated string constant */
1393 error(258);
1394
1395 /* get length of wide-character string */
1396 (void)mblen(NULL, 0);
1397 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1398 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1399 /* invalid multibyte character */
1400 error(291);
1401 break;
1402 }
1403 if (n == 0)
1404 n = 1;
1405 }
1406
1407 ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1408
1409 /* convert from multibyte to wide char */
1410 (void)mbtowc(NULL, NULL, 0);
1411 for (i = 0, wi = 0; i < len; i += n, wi++) {
1412 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1413 break;
1414 if (n == 0)
1415 n = 1;
1416 }
1417 ws[wi] = 0;
1418 free(s);
1419
1420 strg = xcalloc(1, sizeof (strg_t));
1421 strg->st_tspec = WCHAR;
1422 strg->st_len = wlen;
1423 strg->st_wcp = ws;
1424
1425 yylval.y_strg = strg;
1426 return (T_STRING);
1427 }
1428
1429 /*
1430 * As noted above the scanner does not create new symbol table entries
1431 * for symbols it cannot find in the symbol table. This is to avoid
1432 * putting undeclared symbols into the symbol table if a syntax error
1433 * occurs.
1434 *
1435 * getsym() is called as soon as it is probably ok to put the symbol to
1436 * the symbol table. This does not mean that it is not possible that
1437 * symbols are put to the symbol table which are than not completely
1438 * declared due to syntax errors. To avoid too many problems in this
1439 * case symbols get type int in getsym().
1440 *
1441 * XXX calls to getsym() should be delayed until decl1*() is called
1442 */
1443 sym_t *
1444 getsym(sbuf_t *sb)
1445 {
1446 dinfo_t *di;
1447 char *s;
1448 sym_t *sym;
1449
1450 sym = sb->sb_sym;
1451
1452 /*
1453 * During member declaration it is possible that name() looked
1454 * for symbols of type FVFT, although it should have looked for
1455 * symbols of type FTAG. Same can happen for labels. Both cases
1456 * are compensated here.
1457 */
1458 if (symtyp == FMOS || symtyp == FLAB) {
1459 if (sym == NULL || sym->s_kind == FVFT)
1460 sym = search(sb);
1461 }
1462
1463 if (sym != NULL) {
1464 if (sym->s_kind != symtyp)
1465 LERROR("storesym(%d, %d)", sym->s_kind, symtyp);
1466 symtyp = FVFT;
1467 freesb(sb);
1468 return (sym);
1469 }
1470
1471 /* create a new symbol table entry */
1472
1473 /* labels must always be allocated at level 1 (outhermost block) */
1474 if (symtyp == FLAB) {
1475 sym = getlblk(1, sizeof (sym_t));
1476 s = getlblk(1, sb->sb_len + 1);
1477 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1478 sym->s_name = s;
1479 sym->s_blklev = 1;
1480 di = dcs;
1481 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1482 di = di->d_nxt;
1483 if (di->d_ctx != AUTO)
1484 LERROR("storesym()");
1485 } else {
1486 sym = getblk(sizeof (sym_t));
1487 sym->s_name = sb->sb_name;
1488 sym->s_blklev = blklev;
1489 di = dcs;
1490 }
1491
1492 UNIQUE_CURR_POS(sym->s_dpos);
1493 if ((sym->s_kind = symtyp) != FLAB)
1494 sym->s_type = gettyp(INT);
1495
1496 symtyp = FVFT;
1497
1498 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1499 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1500 sym->s_rlink = &symtab[sb->sb_hash];
1501 symtab[sb->sb_hash] = sym;
1502
1503 *di->d_ldlsym = sym;
1504 di->d_ldlsym = &sym->s_dlnxt;
1505
1506 freesb(sb);
1507 return (sym);
1508 }
1509
1510 /*
1511 * Construct a temporary symbol. The symbol starts with a digit, so that
1512 * it is illegal.
1513 */
1514 sym_t *
1515 mktempsym(type_t *t)
1516 {
1517 static int n = 0;
1518 int h;
1519 char *s = getlblk(blklev, 64);
1520 sym_t *sym = getblk(sizeof (sym_t));
1521
1522 (void)snprintf(s, 64, "%.8d_tmp", n++);
1523 h = hash(s);
1524
1525 sym->s_name = s;
1526 sym->s_type = t;
1527 sym->s_blklev = blklev;
1528 sym->s_scl = AUTO;
1529 sym->s_kind = FVFT;
1530 sym->s_used = 1;
1531 sym->s_set = 1;
1532
1533 if ((sym->s_link = symtab[h]) != NULL)
1534 symtab[h]->s_rlink = &sym->s_link;
1535 sym->s_rlink = &symtab[h];
1536 symtab[h] = sym;
1537
1538 *dcs->d_ldlsym = sym;
1539 dcs->d_ldlsym = &sym->s_dlnxt;
1540
1541 return sym;
1542 }
1543
1544 /*
1545 * Remove a symbol forever from the symbol table. s_blklev
1546 * is set to -1 to avoid that the symbol will later be put
1547 * back to the symbol table.
1548 */
1549 void
1550 rmsym(sym_t *sym)
1551 {
1552
1553 if ((*sym->s_rlink = sym->s_link) != NULL)
1554 sym->s_link->s_rlink = sym->s_rlink;
1555 sym->s_blklev = -1;
1556 sym->s_link = NULL;
1557 }
1558
1559 /*
1560 * Remove a list of symbols declared at one level from the symbol
1561 * table.
1562 */
1563 void
1564 rmsyms(sym_t *syms)
1565 {
1566 sym_t *sym;
1567
1568 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1569 if (sym->s_blklev != -1) {
1570 if ((*sym->s_rlink = sym->s_link) != NULL)
1571 sym->s_link->s_rlink = sym->s_rlink;
1572 sym->s_link = NULL;
1573 sym->s_rlink = NULL;
1574 }
1575 }
1576 }
1577
1578 /*
1579 * Put a symbol into the symbol table
1580 */
1581 void
1582 inssym(int bl, sym_t *sym)
1583 {
1584 int h;
1585
1586 h = hash(sym->s_name);
1587 if ((sym->s_link = symtab[h]) != NULL)
1588 symtab[h]->s_rlink = &sym->s_link;
1589 sym->s_rlink = &symtab[h];
1590 symtab[h] = sym;
1591 sym->s_blklev = bl;
1592 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1593 LERROR("inssym()");
1594 }
1595
1596 /*
1597 * Called at level 0 after syntax errors
1598 * Removes all symbols which are not declared at level 0 from the
1599 * symbol table. Also frees all memory which is not associated with
1600 * level 0.
1601 */
1602 void
1603 cleanup(void)
1604 {
1605 sym_t *sym, *nsym;
1606 int i;
1607
1608 for (i = 0; i < HSHSIZ1; i++) {
1609 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1610 nsym = sym->s_link;
1611 if (sym->s_blklev >= 1) {
1612 if ((*sym->s_rlink = nsym) != NULL)
1613 nsym->s_rlink = sym->s_rlink;
1614 }
1615 }
1616 }
1617
1618 for (i = mblklev; i > 0; i--)
1619 freelblk(i);
1620 }
1621
1622 /*
1623 * Create a new symbol with the name of an existing symbol.
1624 */
1625 sym_t *
1626 pushdown(sym_t *sym)
1627 {
1628 int h;
1629 sym_t *nsym;
1630
1631 h = hash(sym->s_name);
1632 nsym = getblk(sizeof (sym_t));
1633 if (sym->s_blklev > blklev)
1634 LERROR("pushdown()");
1635 nsym->s_name = sym->s_name;
1636 UNIQUE_CURR_POS(nsym->s_dpos);
1637 nsym->s_kind = sym->s_kind;
1638 nsym->s_blklev = blklev;
1639
1640 if ((nsym->s_link = symtab[h]) != NULL)
1641 symtab[h]->s_rlink = &nsym->s_link;
1642 nsym->s_rlink = &symtab[h];
1643 symtab[h] = nsym;
1644
1645 *dcs->d_ldlsym = nsym;
1646 dcs->d_ldlsym = &nsym->s_dlnxt;
1647
1648 return (nsym);
1649 }
1650
1651 /*
1652 * Free any dynamically allocated memory referenced by
1653 * the value stack or yylval.
1654 * The type of information in yylval is described by tok.
1655 */
1656 void
1657 freeyyv(void *sp, int tok)
1658 {
1659 if (tok == T_NAME || tok == T_TYPENAME) {
1660 sbuf_t *sb = *(sbuf_t **)sp;
1661 freesb(sb);
1662 } else if (tok == T_CON) {
1663 val_t *val = *(val_t **)sp;
1664 free(val);
1665 } else if (tok == T_STRING) {
1666 strg_t *strg = *(strg_t **)sp;
1667 if (strg->st_tspec == CHAR) {
1668 free(strg->st_cp);
1669 } else if (strg->st_tspec == WCHAR) {
1670 free(strg->st_wcp);
1671 } else {
1672 LERROR("fryylv()");
1673 }
1674 free(strg);
1675 }
1676 }
1677