scan.l revision 1.68 1 %{
2 /* $NetBSD: scan.l,v 1.68 2016/12/24 17:43:45 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 1.68 2016/12/24 17:43:45 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 } kwtab[] = {
207 { "__alignof__", T_ALIGNOF, 0, 0, 0, 0, 0, 0, 0 },
208 { "__attribute__",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 },
209 { "__attribute",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 },
210 { "__packed__", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 },
211 { "packed", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 },
212 { "__aligned__",T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 },
213 { "aligned", T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 },
214 { "__transparent_union__",T_AT_TUNION,0,0, 0, 0, 0, 1, 1 },
215 { "transparent_union",T_AT_TUNION,0, 0, 0, 0, 0, 1, 1 },
216 { "__unused__", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 },
217 { "unused", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 },
218 { "__gnu_inline__",T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 },
219 { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 },
220 { "__deprecated__",T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 },
221 { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 },
222 { "__may_alias__",T_AT_MAY_ALIAS,0, 0, 0, 0, 0, 1, 1 },
223 { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0, 0, 1, 1 },
224 { "format", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 },
225 { "__format__", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 },
226 { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 },
227 { "__printf__", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 },
228 { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 },
229 { "__scanf__", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 },
230 { "strftime", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 },
231 { "__strftime__",T_AT_FORMAT_STRFTIME,0,0, 0, 0, 0, 1, 1 },
232 { "pure", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 },
233 { "__pure__", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 },
234 { "noreturn", T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 },
235 { "__noreturn__",T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 },
236 { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 },
237 { "__sentinel__",T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 },
238 { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 },
239 { "__format_arg__", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 },
240 { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 },
241 { "__returns_twice__", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 },
242 { "cold", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 },
243 { "__cold__", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 },
244 { "asm", T_ASM, 0, 0, 0, 0, 0, 1, 0 },
245 { "__asm", T_ASM, 0, 0, 0, 0, 0, 0, 0 },
246 { "__asm__", T_ASM, 0, 0, 0, 0, 0, 0, 0 },
247 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0, 0, 0 },
248 { "break", T_BREAK, 0, 0, 0, 0, 0, 0, 0 },
249 { "_Bool", T_TYPE, 0, BOOL, 0, 0, 1, 0, 0 },
250 { "case", T_CASE, 0, 0, 0, 0, 0, 0, 0 },
251 { "char", T_TYPE, 0, CHAR, 0, 0, 0, 0, 0 },
252 { "const", T_QUAL, 0, 0, CONST, 1, 0, 0, 0 },
253 { "_Complex", T_TYPE, 0, COMPLEX,0, 0, 1, 0, 0 },
254 { "__const__", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 },
255 { "__const", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 },
256 { "continue", T_CONTINUE, 0, 0, 0, 0, 0, 0, 0 },
257 { "default", T_DEFAULT, 0, 0, 0, 0, 0, 0, 0 },
258 { "do", T_DO, 0, 0, 0, 0, 0, 0, 0 },
259 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0, 0, 0 },
260 { "else", T_ELSE, 0, 0, 0, 0, 0, 0, 0 },
261 { "enum", T_ENUM, 0, 0, 0, 0, 0, 0, 0 },
262 { "__extension__", T_EXTENSION, 0, 0, 0, 0, 0, 1, 0 },
263 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0, 0, 0 },
264 { "float", T_TYPE, 0, FLOAT, 0, 0, 0, 0, 0 },
265 { "for", T_FOR, 0, 0, 0, 0, 0, 0, 0 },
266 { "_Generic", T_GENERIC, 0, 0, 0, 0, 1, 0, 0 },
267 { "goto", T_GOTO, 0, 0, 0, 0, 0, 0, 0 },
268 { "if", T_IF, 0, 0, 0, 0, 0, 0, 0 },
269 { "__imag__", T_IMAG, 0, 0, 0, 0, 1, 0, 0 },
270 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1, 0, 0 },
271 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 },
272 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 },
273 { "int", T_TYPE, 0, INT, 0, 0, 0, 0, 0 },
274 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0, 0, 0 },
275 { "long", T_TYPE, 0, LONG, 0, 0, 0, 0, 0 },
276 { "__real__", T_REAL, 0, 0, 0, 0, 1, 0, 0 },
277 { "register", T_SCLASS, REG, 0, 0, 0, 0, 0, 0 },
278 { "__restrict__",T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 },
279 { "restrict", T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 },
280 { "return", T_RETURN, 0, 0, 0, 0, 0, 0, 0 },
281 { "__packed", T_PACKED, 0, 0, 0, 0, 0, 0, 0 },
282 { "short", T_TYPE, 0, SHORT, 0, 0, 0, 0, 0 },
283 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0, 0, 0 },
284 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 },
285 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 },
286 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0, 0, 0 },
287 { "static", T_SCLASS, STATIC, 0, 0, 0, 0, 0, 0 },
288 { "struct", T_SOU, 0, STRUCT, 0, 0, 0, 0, 0 },
289 { "switch", T_SWITCH, 0, 0, 0, 0, 0, 0, 0 },
290 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0, 0, 0 },
291 { "typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 },
292 { "__typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 },
293 { "__typeof__", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 },
294 { "union", T_SOU, 0, UNION, 0, 0, 0, 0, 0 },
295 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0, 0, 0 },
296 { "void", T_TYPE, 0, VOID, 0, 0, 0, 0, 0 },
297 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0, 0, 0 },
298 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 },
299 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 },
300 { "visibility", T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 },
301 { "__visibility",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 },
302 { "__visibility__",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 },
303 { "weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 },
304 { "__weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 },
305 { "__weak__", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 },
306 { "while", T_WHILE, 0, 0, 0, 0, 0, 0, 0 },
307 { NULL, 0, 0, 0, 0, 0, 0, 0, 0 }
308 };
309
310 /* Symbol table */
311 static sym_t *symtab[HSHSIZ1];
312
313 /* bit i of the entry with index i is set */
314 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
315
316 /* least significant i bits are set in the entry with index i */
317 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
318
319 /* least significant i bits are not set in the entry with index i */
320 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
321
322 /* free list for sbuf structures */
323 static sbuf_t *sbfrlst;
324
325 /* Typ of next expected symbol */
326 symt_t symtyp;
327
328
329 /*
330 * All keywords are written to the symbol table. This saves us looking
331 * in a extra table for each name we found.
332 */
333 void
334 initscan(void)
335 {
336 struct kwtab *kw;
337 sym_t *sym;
338 size_t h, i;
339 uint64_t uq;
340
341 for (kw = kwtab; kw->kw_name != NULL; kw++) {
342 if ((kw->kw_c89 || kw->kw_c99) && tflag)
343 continue;
344 if (kw->kw_c99 && !(Sflag || gflag))
345 continue;
346 if (kw->kw_gcc && !gflag)
347 continue;
348 sym = getblk(sizeof (sym_t));
349 sym->s_name = kw->kw_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 /* initialize bit-masks for quads */
367 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
368 qbmasks[i] = (uint64_t)1 << i;
369 uq = ~(uint64_t)0 << i;
370 qumasks[i] = uq;
371 qlmasks[i] = ~uq;
372 }
373 qumasks[i] = 0;
374 qlmasks[i] = ~(uint64_t)0;
375 }
376
377 /*
378 * Get a free sbuf structure, if possible from the free list
379 */
380 static sbuf_t *
381 allocsb(void)
382 {
383 sbuf_t *sb;
384
385 if ((sb = sbfrlst) != NULL) {
386 sbfrlst = sb->sb_nxt;
387 #ifdef BLKDEBUG
388 (void)memset(sb, 0, sizeof (*sb));
389 #else
390 sb->sb_nxt = NULL;
391 #endif
392 } else {
393 sb = xmalloc(sizeof (sbuf_t));
394 (void)memset(sb, 0, sizeof (*sb));
395 }
396 return (sb);
397 }
398
399 /*
400 * Put a sbuf structure to the free list
401 */
402 static void
403 freesb(sbuf_t *sb)
404 {
405
406 (void)memset(sb, ZERO, sizeof (*sb));
407 sb->sb_nxt = sbfrlst;
408 sbfrlst = sb;
409 }
410
411 /*
412 * Read a character and ensure that it is positive (except EOF).
413 * Increment line count(s) if necessary.
414 */
415 static int
416 inpc(void)
417 {
418 int c;
419
420 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
421 incline();
422 return (c);
423 }
424
425 static int
426 hash(const char *s)
427 {
428 u_int v;
429 const u_char *us;
430
431 v = 0;
432 for (us = (const u_char *)s; *us != '\0'; us++) {
433 v = (v << sizeof (v)) + *us;
434 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
435 }
436 return (v % HSHSIZ1);
437 }
438
439 /*
440 * Lex has found a letter followed by zero or more letters or digits.
441 * It looks for a symbol in the symbol table with the same name. This
442 * symbol must either be a keyword or a symbol of the type required by
443 * symtyp (label, member, tag, ...).
444 *
445 * If it is a keyword, the token is returned. In some cases it is described
446 * more deeply by data written to yylval.
447 *
448 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
449 * is stored in yylval. This struct contains the name of the symbol, its
450 * length and hash value. If there is already a symbol of the same name
451 * and type in the symbol table, the sbuf struct also contains a pointer
452 * to the symbol table entry.
453 */
454 static int
455 name(void)
456 {
457 char *s;
458 sbuf_t *sb;
459 sym_t *sym;
460 int tok;
461
462 sb = allocsb();
463 sb->sb_name = yytext;
464 sb->sb_len = yyleng;
465 sb->sb_hash = hash(yytext);
466 if ((sym = search(sb)) != NULL && sym->s_keyw) {
467 freesb(sb);
468 return (keyw(sym));
469 }
470
471 sb->sb_sym = sym;
472
473 if (sym != NULL) {
474 if (blklev < sym->s_blklev)
475 LERROR("name()");
476 sb->sb_name = sym->s_name;
477 sb->sb_len = strlen(sym->s_name);
478 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
479 } else {
480 s = getblk(yyleng + 1);
481 (void)memcpy(s, yytext, yyleng + 1);
482 sb->sb_name = s;
483 sb->sb_len = yyleng;
484 tok = T_NAME;
485 }
486
487 yylval.y_sb = sb;
488 return (tok);
489 }
490
491 static sym_t *
492 search(sbuf_t *sb)
493 {
494 sym_t *sym;
495
496 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
497 if (strcmp(sym->s_name, sb->sb_name) == 0) {
498 if (sym->s_keyw) {
499 struct kwtab *kw = sym->s_keyw;
500 if (!kw->kw_attr || attron)
501 return (sym);
502 } else if (sym->s_kind == symtyp)
503 return sym;
504 }
505 }
506
507 return (NULL);
508 }
509
510 static int
511 keyw(sym_t *sym)
512 {
513 int t;
514
515 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
516 yylval.y_scl = sym->s_scl;
517 } else if (t == T_TYPE || t == T_SOU) {
518 yylval.y_tspec = sym->s_tspec;
519 } else if (t == T_QUAL) {
520 yylval.y_tqual = sym->s_tqual;
521 }
522 return (t);
523 }
524
525 /*
526 * Convert a string representing an integer into internal representation.
527 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
528 */
529 static int
530 icon(int base)
531 {
532 int l_suffix, u_suffix;
533 int len;
534 const char *cp;
535 char c, *eptr;
536 tspec_t typ;
537 uint64_t uq = 0;
538 int ansiu;
539 static tspec_t contypes[2][3] = {
540 { INT, LONG, QUAD },
541 { UINT, ULONG, UQUAD }
542 };
543
544 cp = yytext;
545 len = yyleng;
546
547 /* skip 0[xX] or 0[bB] */
548 if (base == 16 || base == 2) {
549 cp += 2;
550 len -= 2;
551 }
552
553 /* read suffixes */
554 l_suffix = u_suffix = 0;
555 for ( ; ; ) {
556 if ((c = cp[len - 1]) == 'l' || c == 'L') {
557 l_suffix++;
558 } else if (c == 'u' || c == 'U') {
559 u_suffix++;
560 } else {
561 break;
562 }
563 len--;
564 }
565 if (l_suffix > 2 || u_suffix > 1) {
566 /* malformed integer constant */
567 warning(251);
568 if (l_suffix > 2)
569 l_suffix = 2;
570 if (u_suffix > 1)
571 u_suffix = 1;
572 }
573 if (tflag && u_suffix != 0) {
574 /* suffix U is illegal in traditional C */
575 warning(97);
576 }
577 typ = contypes[u_suffix][l_suffix];
578
579 errno = 0;
580
581 uq = strtouq(cp, &eptr, base);
582 if (eptr != cp + len)
583 LERROR("icon()");
584 if (errno != 0)
585 /* integer constant out of range */
586 warning(252);
587
588 /*
589 * If the value is too big for the current type, we must choose
590 * another type.
591 */
592 ansiu = 0;
593 switch (typ) {
594 case INT:
595 if (uq <= TARG_INT_MAX) {
596 /* ok */
597 } else if (uq <= TARG_UINT_MAX && base != 10) {
598 typ = UINT;
599 } else if (uq <= TARG_LONG_MAX) {
600 typ = LONG;
601 } else {
602 typ = ULONG;
603 if (uq > TARG_ULONG_MAX) {
604 /* integer constant out of range */
605 warning(252);
606 }
607 }
608 if (typ == UINT || typ == ULONG) {
609 if (tflag) {
610 typ = LONG;
611 } else if (!sflag) {
612 /*
613 * Remember that the constant is unsigned
614 * only in ANSI C
615 */
616 ansiu = 1;
617 }
618 }
619 break;
620 case UINT:
621 if (uq > TARG_UINT_MAX) {
622 typ = ULONG;
623 if (uq > TARG_ULONG_MAX) {
624 /* integer constant out of range */
625 warning(252);
626 }
627 }
628 break;
629 case LONG:
630 if (uq > TARG_LONG_MAX && !tflag) {
631 typ = ULONG;
632 if (!sflag)
633 ansiu = 1;
634 if (uq > TARG_ULONG_MAX) {
635 /* integer constant out of range */
636 warning(252);
637 }
638 }
639 break;
640 case ULONG:
641 if (uq > TARG_ULONG_MAX) {
642 /* integer constant out of range */
643 warning(252);
644 }
645 break;
646 case QUAD:
647 if (uq > TARG_QUAD_MAX && !tflag) {
648 typ = UQUAD;
649 if (!sflag)
650 ansiu = 1;
651 }
652 break;
653 case UQUAD:
654 if (uq > TARG_UQUAD_MAX) {
655 /* integer constant out of range */
656 warning(252);
657 }
658 break;
659 /* LINTED206: (enumeration values not handled in switch) */
660 case STRUCT:
661 case VOID:
662 case LDOUBLE:
663 case FUNC:
664 case ARRAY:
665 case PTR:
666 case ENUM:
667 case UNION:
668 case SIGNED:
669 case NOTSPEC:
670 case DOUBLE:
671 case FLOAT:
672 case USHORT:
673 case SHORT:
674 case UCHAR:
675 case SCHAR:
676 case CHAR:
677 case BOOL:
678 case UNSIGN:
679 case FCOMPLEX:
680 case DCOMPLEX:
681 case LCOMPLEX:
682 case COMPLEX:
683 break;
684
685 case NTSPEC: /* this value unused */
686 break;
687 }
688
689 uq = (uint64_t)xsign((int64_t)uq, typ, -1);
690
691 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
692 yylval.y_val->v_ansiu = ansiu;
693 yylval.y_val->v_quad = (int64_t)uq;
694
695 return (T_CON);
696 }
697
698 /*
699 * Returns 1 if t is a signed type and the value is negative.
700 *
701 * len is the number of significant bits. If len is -1, len is set
702 * to the width of type t.
703 */
704 int
705 sign(int64_t q, tspec_t t, int len)
706 {
707
708 if (t == PTR || isutyp(t))
709 return (0);
710 return (msb(q, t, len));
711 }
712
713 int
714 msb(int64_t q, tspec_t t, int len)
715 {
716
717 if (len <= 0)
718 len = size(t);
719 return ((q & qbmasks[len - 1]) != 0);
720 }
721
722 /*
723 * Extends the sign of q.
724 */
725 int64_t
726 xsign(int64_t q, tspec_t t, int len)
727 {
728
729 if (len <= 0)
730 len = size(t);
731
732 if (t == PTR || isutyp(t) || !sign(q, t, len)) {
733 q &= qlmasks[len];
734 } else {
735 q |= qumasks[len];
736 }
737 return (q);
738 }
739
740 /*
741 * Convert a string representing a floating point value into its interal
742 * representation. Type and value are returned in yylval. fcon()
743 * (and yylex()) returns T_CON.
744 * XXX Currently it is not possible to convert constants of type
745 * long double which are greater than DBL_MAX.
746 */
747 static int
748 fcon(void)
749 {
750 const char *cp;
751 int len;
752 tspec_t typ;
753 char c, *eptr;
754 double d;
755 float f = 0;
756
757 cp = yytext;
758 len = yyleng;
759
760 if (cp[len - 1] == 'i') {
761 /* imaginary, do nothing for now */
762 len--;
763 }
764 if ((c = cp[len - 1]) == 'f' || c == 'F') {
765 typ = FLOAT;
766 len--;
767 } else if (c == 'l' || c == 'L') {
768 typ = LDOUBLE;
769 len--;
770 } else {
771 if (c == 'd' || c == 'D')
772 len--;
773 typ = DOUBLE;
774 }
775
776 if (tflag && typ != DOUBLE) {
777 /* suffixes F and L are illegal in traditional C */
778 warning(98);
779 }
780
781 errno = 0;
782 d = strtod(cp, &eptr);
783 if (eptr != cp + len) {
784 switch (*eptr) {
785 /*
786 * XXX: non-native non-current strtod() may not handle hex
787 * floats, ignore the rest if we find traces of hex float
788 * syntax...
789 */
790 case 'p':
791 case 'P':
792 case 'x':
793 case 'X':
794 d = 0;
795 errno = 0;
796 break;
797 default:
798 LERROR("fcon(%s->%s)", cp, eptr);
799 }
800 }
801 if (errno != 0)
802 /* floating-point constant out of range */
803 warning(248);
804
805 if (typ == FLOAT) {
806 f = (float)d;
807 if (!finite(f)) {
808 /* floating-point constant out of range */
809 warning(248);
810 f = f > 0 ? FLT_MAX : -FLT_MAX;
811 }
812 }
813
814 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
815 if (typ == FLOAT) {
816 yylval.y_val->v_ldbl = f;
817 } else {
818 yylval.y_val->v_ldbl = d;
819 }
820
821 return (T_CON);
822 }
823
824 static int
825 operator(int t, op_t o)
826 {
827
828 yylval.y_op = o;
829 return (t);
830 }
831
832 /*
833 * Called if lex found a leading \'.
834 */
835 static int
836 ccon(void)
837 {
838 size_t n;
839 int val, c;
840 char cv;
841
842 n = 0;
843 val = 0;
844 while ((c = getescc('\'')) >= 0) {
845 val = (val << CHAR_BIT) + c;
846 n++;
847 }
848 if (c == -2) {
849 /* unterminated character constant */
850 error(253);
851 } else {
852 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
853 /* too many characters in character constant */
854 error(71);
855 } else if (n > 1) {
856 /* multi-character character constant */
857 warning(294);
858 } else if (n == 0) {
859 /* empty character constant */
860 error(73);
861 }
862 }
863 if (n == 1) {
864 cv = (char)val;
865 val = cv;
866 }
867
868 yylval.y_val = xcalloc(1, sizeof (val_t));
869 yylval.y_val->v_tspec = INT;
870 yylval.y_val->v_quad = val;
871
872 return (T_CON);
873 }
874
875 /*
876 * Called if lex found a leading L\'
877 */
878 static int
879 wccon(void)
880 {
881 static char buf[MB_LEN_MAX + 1];
882 size_t i;
883 int c;
884 wchar_t wc;
885
886 i = 0;
887 while ((c = getescc('\'')) >= 0) {
888 if (i < MB_CUR_MAX)
889 buf[i] = (char)c;
890 i++;
891 }
892
893 wc = 0;
894
895 if (c == -2) {
896 /* unterminated character constant */
897 error(253);
898 } else if (c == 0) {
899 /* empty character constant */
900 error(73);
901 } else {
902 if (i > MB_CUR_MAX) {
903 i = MB_CUR_MAX;
904 /* too many characters in character constant */
905 error(71);
906 } else {
907 buf[i] = '\0';
908 (void)mbtowc(NULL, NULL, 0);
909 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
910 /* invalid multibyte character */
911 error(291);
912 }
913 }
914
915 yylval.y_val = xcalloc(1, sizeof (val_t));
916 yylval.y_val->v_tspec = WCHAR;
917 yylval.y_val->v_quad = wc;
918
919 return (T_CON);
920 }
921
922 /*
923 * Read a character which is part of a character constant or of a string
924 * and handle escapes.
925 *
926 * The Argument is the character which delimits the character constant or
927 * string.
928 *
929 * Returns -1 if the end of the character constant or string is reached,
930 * -2 if the EOF is reached, and the character otherwise.
931 */
932 static int
933 getescc(int d)
934 {
935 static int pbc = -1;
936 int n, c, v;
937
938 if (pbc == -1) {
939 c = inpc();
940 } else {
941 c = pbc;
942 pbc = -1;
943 }
944 if (c == d)
945 return (-1);
946 switch (c) {
947 case '\n':
948 if (tflag) {
949 /* newline in string or char constant */
950 error(254);
951 return (-2);
952 }
953 return (c);
954 case EOF:
955 return (-2);
956 case '\\':
957 switch (c = inpc()) {
958 case '"':
959 if (tflag && d == '\'')
960 /* \" inside character constant undef. ... */
961 warning(262);
962 return ('"');
963 case '\'':
964 return ('\'');
965 case '?':
966 if (tflag)
967 /* \? undefined in traditional C */
968 warning(263);
969 return ('?');
970 case '\\':
971 return ('\\');
972 case 'a':
973 if (tflag)
974 /* \a undefined in traditional C */
975 warning(81);
976 return ('\a');
977 case 'b':
978 return ('\b');
979 case 'f':
980 return ('\f');
981 case 'n':
982 return ('\n');
983 case 'r':
984 return ('\r');
985 case 't':
986 return ('\t');
987 case 'v':
988 if (tflag)
989 /* \v undefined in traditional C */
990 warning(264);
991 return ('\v');
992 case '8': case '9':
993 /* bad octal digit %c */
994 warning(77, c);
995 /* FALLTHROUGH */
996 case '0': case '1': case '2': case '3':
997 case '4': case '5': case '6': case '7':
998 n = 3;
999 v = 0;
1000 do {
1001 v = (v << 3) + (c - '0');
1002 c = inpc();
1003 } while (--n && isdigit(c) && (tflag || c <= '7'));
1004 if (tflag && n > 0 && isdigit(c))
1005 /* bad octal digit %c */
1006 warning(77, c);
1007 pbc = c;
1008 if (v > TARG_UCHAR_MAX) {
1009 /* character escape does not fit in char. */
1010 warning(76);
1011 v &= CHAR_MASK;
1012 }
1013 return (v);
1014 case 'x':
1015 if (tflag)
1016 /* \x undefined in traditional C */
1017 warning(82);
1018 v = 0;
1019 n = 0;
1020 while ((c = inpc()) >= 0 && isxdigit(c)) {
1021 c = isdigit(c) ?
1022 c - '0' : toupper(c) - 'A' + 10;
1023 v = (v << 4) + c;
1024 if (n >= 0) {
1025 if ((v & ~CHAR_MASK) != 0) {
1026 /* overflow in hex escape */
1027 warning(75);
1028 n = -1;
1029 } else {
1030 n++;
1031 }
1032 }
1033 }
1034 pbc = c;
1035 if (n == 0) {
1036 /* no hex digits follow \x */
1037 error(74);
1038 } if (n == -1) {
1039 v &= CHAR_MASK;
1040 }
1041 return (v);
1042 case '\n':
1043 return (getescc(d));
1044 case EOF:
1045 return (-2);
1046 default:
1047 if (isprint(c)) {
1048 /* dubious escape \%c */
1049 warning(79, c);
1050 } else {
1051 /* dubious escape \%o */
1052 warning(80, c);
1053 }
1054 }
1055 }
1056 return (c);
1057 }
1058
1059 /*
1060 * Called for preprocessor directives. Currently implemented are:
1061 * # lineno
1062 * # lineno "filename"
1063 */
1064 static void
1065 directive(void)
1066 {
1067 const char *cp, *fn;
1068 char c, *eptr;
1069 size_t fnl;
1070 long ln;
1071 static int first = 1;
1072
1073 /* Go to first non-whitespace after # */
1074 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1075 continue;
1076
1077 if (!isdigit((unsigned char)c)) {
1078 error:
1079 /* undefined or invalid # directive */
1080 warning(255);
1081 return;
1082 }
1083 ln = strtol(--cp, &eptr, 10);
1084 if (cp == eptr)
1085 goto error;
1086 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1087 goto error;
1088 while ((c = *cp++) == ' ' || c == '\t')
1089 continue;
1090 if (c != '\0') {
1091 if (c != '"')
1092 goto error;
1093 fn = cp;
1094 while ((c = *cp) != '"' && c != '\0')
1095 cp++;
1096 if (c != '"')
1097 goto error;
1098 if ((fnl = cp++ - fn) > PATH_MAX)
1099 goto error;
1100 while ((c = *cp++) == ' ' || c == '\t')
1101 continue;
1102 #if 0
1103 if (c != '\0')
1104 warning("extra character(s) after directive");
1105 #endif
1106
1107 /* empty string means stdin */
1108 if (fnl == 0) {
1109 fn = "{standard input}";
1110 fnl = 16; /* strlen (fn) */
1111 }
1112 curr_pos.p_file = fnnalloc(fn, fnl);
1113 /*
1114 * If this is the first directive, the name is the name
1115 * of the C source file as specified at the command line.
1116 * It is written to the output file.
1117 */
1118 if (first) {
1119 csrc_pos.p_file = curr_pos.p_file;
1120 outsrc(fnxform(curr_pos.p_file,
1121 strlen(curr_pos.p_file)));
1122 first = 0;
1123 }
1124 }
1125 curr_pos.p_line = (int)ln - 1;
1126 curr_pos.p_uniq = 0;
1127 if (curr_pos.p_file == csrc_pos.p_file) {
1128 csrc_pos.p_line = (int)ln - 1;
1129 csrc_pos.p_uniq = 0;
1130 }
1131 }
1132
1133 /*
1134 * Handle lint comments. Following comments are currently understood:
1135 * ARGSUSEDn
1136 * BITFIELDTYPE
1137 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1138 * FALLTHRU FALLTHROUGH
1139 * LINTLIBRARY
1140 * LINTEDn NOSTRICTn
1141 * LONGLONG
1142 * NOTREACHED
1143 * PRINTFLIKEn
1144 * PROTOLIB
1145 * SCANFLIKEn
1146 * VARARGSn
1147 * If one of this comments is recognized, the arguments, if any, are
1148 * parsed and a function which handles this comment is called.
1149 */
1150 static void
1151 comment(void)
1152 {
1153 int c, lc;
1154 static struct {
1155 const char *keywd;
1156 int arg;
1157 void (*func)(int);
1158 } keywtab[] = {
1159 { "ARGSUSED", 1, argsused },
1160 { "BITFIELDTYPE", 0, bitfieldtype },
1161 { "CONSTCOND", 0, constcond },
1162 { "CONSTANTCOND", 0, constcond },
1163 { "CONSTANTCONDITION", 0, constcond },
1164 { "FALLTHRU", 0, fallthru },
1165 { "FALLTHROUGH", 0, fallthru },
1166 { "LINTLIBRARY", 0, lintlib },
1167 { "LINTED", 1, linted },
1168 { "LONGLONG", 0, longlong },
1169 { "NOSTRICT", 1, linted },
1170 { "NOTREACHED", 0, notreach },
1171 { "PRINTFLIKE", 1, printflike },
1172 { "PROTOLIB", 1, protolib },
1173 { "SCANFLIKE", 1, scanflike },
1174 { "VARARGS", 1, varargs },
1175 };
1176 char keywd[32];
1177 char arg[32];
1178 size_t l, i;
1179 int a;
1180 int eoc;
1181
1182 eoc = 0;
1183
1184 /* Skip white spaces after the start of the comment */
1185 while ((c = inpc()) != EOF && isspace(c))
1186 continue;
1187
1188 /* Read the potential keyword to keywd */
1189 l = 0;
1190 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1191 keywd[l++] = (char)c;
1192 c = inpc();
1193 }
1194 keywd[l] = '\0';
1195
1196 /* look for the keyword */
1197 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1198 if (strcmp(keywtab[i].keywd, keywd) == 0)
1199 break;
1200 }
1201 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1202 goto skip_rest;
1203
1204 /* skip white spaces after the keyword */
1205 while (c != EOF && isspace(c))
1206 c = inpc();
1207
1208 /* read the argument, if the keyword accepts one and there is one */
1209 l = 0;
1210 if (keywtab[i].arg) {
1211 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1212 arg[l++] = (char)c;
1213 c = inpc();
1214 }
1215 }
1216 arg[l] = '\0';
1217 a = l != 0 ? atoi(arg) : -1;
1218
1219 /* skip white spaces after the argument */
1220 while (c != EOF && isspace(c))
1221 c = inpc();
1222
1223 if (c != '*' || (c = inpc()) != '/') {
1224 if (keywtab[i].func != linted)
1225 /* extra characters in lint comment */
1226 warning(257);
1227 } else {
1228 /*
1229 * remember that we have already found the end of the
1230 * comment
1231 */
1232 eoc = 1;
1233 }
1234
1235 if (keywtab[i].func != NULL)
1236 (*keywtab[i].func)(a);
1237
1238 skip_rest:
1239 while (!eoc) {
1240 lc = c;
1241 if ((c = inpc()) == EOF) {
1242 /* unterminated comment */
1243 error(256);
1244 break;
1245 }
1246 if (lc == '*' && c == '/')
1247 eoc = 1;
1248 }
1249 }
1250
1251 /*
1252 * Handle // style comments
1253 */
1254 static void
1255 slashslashcomment(void)
1256 {
1257 int c;
1258
1259 if (!Sflag && !gflag)
1260 /* // comments only supported in C99 */
1261 (void)gnuism(312, tflag ? "traditional" : "ANSI");
1262
1263 while ((c = inpc()) != EOF && c != '\n')
1264 continue;
1265 }
1266
1267 /*
1268 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1269 * clrwflgs() is called after function definitions and global and
1270 * local declarations and definitions. It is also called between
1271 * the controlling expression and the body of control statements
1272 * (if, switch, for, while).
1273 */
1274 void
1275 clrwflgs(void)
1276 {
1277
1278 lwarn = LWARN_ALL;
1279 quadflg = 0;
1280 ccflg = 0;
1281 }
1282
1283 /*
1284 * Strings are stored in a dynamically alloceted buffer and passed
1285 * in yylval.y_xstrg to the parser. The parser or the routines called
1286 * by the parser are responsible for freeing this buffer.
1287 */
1288 static int
1289 string(void)
1290 {
1291 u_char *s;
1292 int c;
1293 size_t len, max;
1294 strg_t *strg;
1295
1296 s = xmalloc(max = 64);
1297
1298 len = 0;
1299 while ((c = getescc('"')) >= 0) {
1300 /* +1 to reserve space for a trailing NUL character */
1301 if (len + 1 == max)
1302 s = xrealloc(s, max *= 2);
1303 s[len++] = (char)c;
1304 }
1305 s[len] = '\0';
1306 if (c == -2)
1307 /* unterminated string constant */
1308 error(258);
1309
1310 strg = xcalloc(1, sizeof (strg_t));
1311 strg->st_tspec = CHAR;
1312 strg->st_len = len;
1313 strg->st_cp = s;
1314
1315 yylval.y_strg = strg;
1316 return (T_STRING);
1317 }
1318
1319 static int
1320 wcstrg(void)
1321 {
1322 char *s;
1323 int c, n;
1324 size_t i, wi;
1325 size_t len, max, wlen;
1326 wchar_t *ws;
1327 strg_t *strg;
1328
1329 s = xmalloc(max = 64);
1330 len = 0;
1331 while ((c = getescc('"')) >= 0) {
1332 /* +1 to save space for a trailing NUL character */
1333 if (len + 1 >= max)
1334 s = xrealloc(s, max *= 2);
1335 s[len++] = (char)c;
1336 }
1337 s[len] = '\0';
1338 if (c == -2)
1339 /* unterminated string constant */
1340 error(258);
1341
1342 /* get length of wide-character string */
1343 (void)mblen(NULL, 0);
1344 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1345 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1346 /* invalid multibyte character */
1347 error(291);
1348 break;
1349 }
1350 if (n == 0)
1351 n = 1;
1352 }
1353
1354 ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1355
1356 /* convert from multibyte to wide char */
1357 (void)mbtowc(NULL, NULL, 0);
1358 for (i = 0, wi = 0; i < len; i += n, wi++) {
1359 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1360 break;
1361 if (n == 0)
1362 n = 1;
1363 }
1364 ws[wi] = 0;
1365 free(s);
1366
1367 strg = xcalloc(1, sizeof (strg_t));
1368 strg->st_tspec = WCHAR;
1369 strg->st_len = wlen;
1370 strg->st_wcp = ws;
1371
1372 yylval.y_strg = strg;
1373 return (T_STRING);
1374 }
1375
1376 /*
1377 * As noted above the scanner does not create new symbol table entries
1378 * for symbols it cannot find in the symbol table. This is to avoid
1379 * putting undeclared symbols into the symbol table if a syntax error
1380 * occurs.
1381 *
1382 * getsym() is called as soon as it is probably ok to put the symbol to
1383 * the symbol table. This does not mean that it is not possible that
1384 * symbols are put to the symbol table which are than not completely
1385 * declared due to syntax errors. To avoid too many problems in this
1386 * case symbols get type int in getsym().
1387 *
1388 * XXX calls to getsym() should be delayed until decl1*() is called
1389 */
1390 sym_t *
1391 getsym(sbuf_t *sb)
1392 {
1393 dinfo_t *di;
1394 char *s;
1395 sym_t *sym;
1396
1397 sym = sb->sb_sym;
1398
1399 /*
1400 * During member declaration it is possible that name() looked
1401 * for symbols of type FVFT, although it should have looked for
1402 * symbols of type FTAG. Same can happen for labels. Both cases
1403 * are compensated here.
1404 */
1405 if (symtyp == FMOS || symtyp == FLAB) {
1406 if (sym == NULL || sym->s_kind == FVFT)
1407 sym = search(sb);
1408 }
1409
1410 if (sym != NULL) {
1411 if (sym->s_kind != symtyp)
1412 LERROR("storesym(%d, %d)", sym->s_kind, symtyp);
1413 symtyp = FVFT;
1414 freesb(sb);
1415 return (sym);
1416 }
1417
1418 /* create a new symbol table entry */
1419
1420 /* labels must always be allocated at level 1 (outhermost block) */
1421 if (symtyp == FLAB) {
1422 sym = getlblk(1, sizeof (sym_t));
1423 s = getlblk(1, sb->sb_len + 1);
1424 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1425 sym->s_name = s;
1426 sym->s_blklev = 1;
1427 di = dcs;
1428 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1429 di = di->d_nxt;
1430 if (di->d_ctx != AUTO)
1431 LERROR("storesym()");
1432 } else {
1433 sym = getblk(sizeof (sym_t));
1434 sym->s_name = sb->sb_name;
1435 sym->s_blklev = blklev;
1436 di = dcs;
1437 }
1438
1439 UNIQUE_CURR_POS(sym->s_dpos);
1440 if ((sym->s_kind = symtyp) != FLAB)
1441 sym->s_type = gettyp(INT);
1442
1443 symtyp = FVFT;
1444
1445 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1446 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1447 sym->s_rlink = &symtab[sb->sb_hash];
1448 symtab[sb->sb_hash] = sym;
1449
1450 *di->d_ldlsym = sym;
1451 di->d_ldlsym = &sym->s_dlnxt;
1452
1453 freesb(sb);
1454 return (sym);
1455 }
1456
1457 /*
1458 * Construct a temporary symbol. The symbol starts with a digit, so that
1459 * it is illegal.
1460 */
1461 sym_t *
1462 mktempsym(type_t *t)
1463 {
1464 static int n = 0;
1465 int h;
1466 char *s = getlblk(blklev, 64);
1467 sym_t *sym = getblk(sizeof (sym_t));
1468
1469 (void)snprintf(s, 64, "%.8d_tmp", n++);
1470 h = hash(s);
1471
1472 sym->s_name = s;
1473 sym->s_type = t;
1474 sym->s_blklev = blklev;
1475 sym->s_scl = AUTO;
1476 sym->s_kind = FVFT;
1477 sym->s_used = 1;
1478 sym->s_set = 1;
1479
1480 if ((sym->s_link = symtab[h]) != NULL)
1481 symtab[h]->s_rlink = &sym->s_link;
1482 sym->s_rlink = &symtab[h];
1483 symtab[h] = sym;
1484
1485 *dcs->d_ldlsym = sym;
1486 dcs->d_ldlsym = &sym->s_dlnxt;
1487
1488 return sym;
1489 }
1490
1491 /*
1492 * Remove a symbol forever from the symbol table. s_blklev
1493 * is set to -1 to avoid that the symbol will later be put
1494 * back to the symbol table.
1495 */
1496 void
1497 rmsym(sym_t *sym)
1498 {
1499
1500 if ((*sym->s_rlink = sym->s_link) != NULL)
1501 sym->s_link->s_rlink = sym->s_rlink;
1502 sym->s_blklev = -1;
1503 sym->s_link = NULL;
1504 }
1505
1506 /*
1507 * Remove a list of symbols declared at one level from the symbol
1508 * table.
1509 */
1510 void
1511 rmsyms(sym_t *syms)
1512 {
1513 sym_t *sym;
1514
1515 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1516 if (sym->s_blklev != -1) {
1517 if ((*sym->s_rlink = sym->s_link) != NULL)
1518 sym->s_link->s_rlink = sym->s_rlink;
1519 sym->s_link = NULL;
1520 sym->s_rlink = NULL;
1521 }
1522 }
1523 }
1524
1525 /*
1526 * Put a symbol into the symbol table
1527 */
1528 void
1529 inssym(int bl, sym_t *sym)
1530 {
1531 int h;
1532
1533 h = hash(sym->s_name);
1534 if ((sym->s_link = symtab[h]) != NULL)
1535 symtab[h]->s_rlink = &sym->s_link;
1536 sym->s_rlink = &symtab[h];
1537 symtab[h] = sym;
1538 sym->s_blklev = bl;
1539 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1540 LERROR("inssym()");
1541 }
1542
1543 /*
1544 * Called at level 0 after syntax errors
1545 * Removes all symbols which are not declared at level 0 from the
1546 * symbol table. Also frees all memory which is not associated with
1547 * level 0.
1548 */
1549 void
1550 cleanup(void)
1551 {
1552 sym_t *sym, *nsym;
1553 int i;
1554
1555 for (i = 0; i < HSHSIZ1; i++) {
1556 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1557 nsym = sym->s_link;
1558 if (sym->s_blklev >= 1) {
1559 if ((*sym->s_rlink = nsym) != NULL)
1560 nsym->s_rlink = sym->s_rlink;
1561 }
1562 }
1563 }
1564
1565 for (i = mblklev; i > 0; i--)
1566 freelblk(i);
1567 }
1568
1569 /*
1570 * Create a new symbol with the name of an existing symbol.
1571 */
1572 sym_t *
1573 pushdown(sym_t *sym)
1574 {
1575 int h;
1576 sym_t *nsym;
1577
1578 h = hash(sym->s_name);
1579 nsym = getblk(sizeof (sym_t));
1580 if (sym->s_blklev > blklev)
1581 LERROR("pushdown()");
1582 nsym->s_name = sym->s_name;
1583 UNIQUE_CURR_POS(nsym->s_dpos);
1584 nsym->s_kind = sym->s_kind;
1585 nsym->s_blklev = blklev;
1586
1587 if ((nsym->s_link = symtab[h]) != NULL)
1588 symtab[h]->s_rlink = &nsym->s_link;
1589 nsym->s_rlink = &symtab[h];
1590 symtab[h] = nsym;
1591
1592 *dcs->d_ldlsym = nsym;
1593 dcs->d_ldlsym = &nsym->s_dlnxt;
1594
1595 return (nsym);
1596 }
1597
1598 /*
1599 * Free any dynamically allocated memory referenced by
1600 * the value stack or yylval.
1601 * The type of information in yylval is described by tok.
1602 */
1603 void
1604 freeyyv(void *sp, int tok)
1605 {
1606 if (tok == T_NAME || tok == T_TYPENAME) {
1607 sbuf_t *sb = *(sbuf_t **)sp;
1608 freesb(sb);
1609 } else if (tok == T_CON) {
1610 val_t *val = *(val_t **)sp;
1611 free(val);
1612 } else if (tok == T_STRING) {
1613 strg_t *strg = *(strg_t **)sp;
1614 if (strg->st_tspec == CHAR) {
1615 free(strg->st_cp);
1616 } else if (strg->st_tspec == WCHAR) {
1617 free(strg->st_wcp);
1618 } else {
1619 LERROR("fryylv()");
1620 }
1621 free(strg);
1622 }
1623 }
1624