Annotation of src/usr.bin/xlint/lint1/scan.l, Revision 1.85
1.1 cgd 1: %{
1.85 ! christos 2: /* $NetBSD: scan.l,v 1.84 2018/10/07 14:20:01 christos Exp $ */
1.2 cgd 3:
1.1 cgd 4: /*
1.9 cgd 5: * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved.
1.1 cgd 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:
1.11 christos 36: #include <sys/cdefs.h>
1.25 tv 37: #if defined(__RCSID) && !defined(lint)
1.85 ! christos 38: __RCSID("$NetBSD: scan.l,v 1.84 2018/10/07 14:20:01 christos Exp $");
1.1 cgd 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"
1.12 tv 50: #include "cgram.h"
1.1 cgd 51:
1.61 joerg 52: #define CHAR_MASK ((int)(~(~0U << CHAR_BIT)))
1.1 cgd 53:
54: /* Current position (its also updated when an included file is parsed) */
1.9 cgd 55: pos_t curr_pos = { 1, "", 0 };
1.1 cgd 56:
57: /*
58: * Current position in C source (not updated when an included file is
59: * parsed).
60: */
1.9 cgd 61: pos_t csrc_pos = { 1, "", 0 };
1.1 cgd 62:
1.55 christos 63: /* Are we parsing a gcc attribute? */
64: int attron;
65:
1.19 lukem 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);
1.1 cgd 86:
87: %}
88:
1.62 christos 89:
1.1 cgd 90: L [_A-Za-z]
91: D [0-9]
92: NZD [1-9]
1.62 christos 93: BD [0-1]
1.1 cgd 94: OD [0-7]
95: HD [0-9A-Fa-f]
96: EX ([eE][+-]?[0-9]+)
1.47 christos 97: HX (p[+-]?[0-9A-Fa-f]+)
98: TL ([fFlL]?[i]?)
1.1 cgd 99:
1.46 christos 100: %option nounput
101:
1.1 cgd 102: %%
103:
104: {L}({L}|{D})* return (name());
1.62 christos 105: 0[bB]{BD}+[lLuU]* return (icon(2));
1.1 cgd 106: 0{OD}*[lLuU]* return (icon(8));
107: {NZD}{D}*[lLuU]* return (icon(10));
108: 0[xX]{HD}+[lLuU]* return (icon(16));
1.47 christos 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());
1.1 cgd 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();
1.18 lukem 168: "//" slashslashcomment();
1.1 cgd 169: . badchar(yytext[0]);
170:
171: %%
172:
173: static void
1.19 lukem 174: incline(void)
1.1 cgd 175: {
176: curr_pos.p_line++;
1.9 cgd 177: curr_pos.p_uniq = 0;
178: if (curr_pos.p_file == csrc_pos.p_file) {
1.1 cgd 179: csrc_pos.p_line++;
1.9 cgd 180: csrc_pos.p_uniq = 0;
181: }
1.1 cgd 182: }
183:
184: static void
1.19 lukem 185: badchar(int c)
1.1 cgd 186: {
1.19 lukem 187:
1.1 cgd 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 */
1.55 christos 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 */
1.73 christos 206: u_int kw_deco : 3; /* name[1] __name[2] __name__[4] */
1.1 cgd 207: } kwtab[] = {
1.83 christos 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
1.73 christos 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 },
1.80 christos 215: { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 },
1.73 christos 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 },
1.75 christos 223: { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 },
1.73 christos 224: { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 },
1.76 christos 225: { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 },
1.79 christos 226: { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 },
1.73 christos 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 },
1.75 christos 253: { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 },
1.73 christos 254: { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 },
1.69 christos 255: { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION,
1.73 christos 256: 0, 0, 0, 0,0,1,1,5 },
1.75 christos 257: { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 },
1.82 christos 258: { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 },
1.73 christos 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 },
1.77 christos 262: { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 },
1.73 christos 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 },
1.75 christos 279: { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 },
1.73 christos 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 },
1.81 christos 283: { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 },
1.73 christos 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 },
1.74 christos 286: { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 },
1.73 christos 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 },
1.85 ! christos 294: { "warn_unused_result", T_AT_WARN_UNUSED_RESULT, 0, 0, 0, 0,0,1,1,5 },
1.73 christos 295: { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 },
296: { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 },
297: { NULL, 0, 0, 0, 0, 0,0,0,0,0 }
1.1 cgd 298: };
299:
300: /* Symbol table */
301: static sym_t *symtab[HSHSIZ1];
302:
303: /* bit i of the entry with index i is set */
1.27 thorpej 304: uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
1.1 cgd 305:
306: /* least significant i bits are set in the entry with index i */
1.27 thorpej 307: uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
1.1 cgd 308:
309: /* least significant i bits are not set in the entry with index i */
1.27 thorpej 310: uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
1.1 cgd 311:
312: /* free list for sbuf structures */
313: static sbuf_t *sbfrlst;
314:
315: /* Typ of next expected symbol */
316: symt_t symtyp;
317:
318:
1.73 christos 319: static void
320: addkw(struct kwtab *kw, int deco)
321: {
322: sym_t *sym;
323: size_t h;
324: char buf[256];
325: const char *name;
326:
327: if (!(kw->kw_deco & deco))
328: return;
329:
330: switch (deco) {
331: case 1:
332: name = kw->kw_name;
333: break;
334: case 2:
335: snprintf(buf, sizeof(buf), "__%s", kw->kw_name);
336: name = strdup(buf);
337: break;
338: case 4:
339: snprintf(buf, sizeof(buf), "__%s__", kw->kw_name);
340: name = strdup(buf);
341: break;
342: default:
343: abort();
344: }
345:
346: if (name == NULL)
347: err(1, "Can't init symbol table");
348:
349: sym = getblk(sizeof (sym_t));
350: sym->s_name = name;
351: sym->s_keyw = kw;
352: sym->s_value.v_quad = kw->kw_token;
353: if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
354: sym->s_tspec = kw->kw_tspec;
355: } else if (kw->kw_token == T_SCLASS) {
356: sym->s_scl = kw->kw_scl;
357: } else if (kw->kw_token == T_QUAL) {
358: sym->s_tqual = kw->kw_tqual;
359: }
360: h = hash(sym->s_name);
361: if ((sym->s_link = symtab[h]) != NULL)
362: symtab[h]->s_rlink = &sym->s_link;
363: sym->s_rlink = &symtab[h];
364: symtab[h] = sym;
365: }
366:
1.1 cgd 367: /*
368: * All keywords are written to the symbol table. This saves us looking
369: * in a extra table for each name we found.
370: */
371: void
1.19 lukem 372: initscan(void)
1.1 cgd 373: {
374: struct kwtab *kw;
1.73 christos 375: size_t i;
1.27 thorpej 376: uint64_t uq;
1.1 cgd 377:
378: for (kw = kwtab; kw->kw_name != NULL; kw++) {
1.32 perry 379: if ((kw->kw_c89 || kw->kw_c99) && tflag)
380: continue;
381: if (kw->kw_c99 && !(Sflag || gflag))
1.5 jpo 382: continue;
383: if (kw->kw_gcc && !gflag)
384: continue;
1.73 christos 385: addkw(kw, 1);
386: addkw(kw, 2);
387: addkw(kw, 4);
1.1 cgd 388: }
389:
390: /* initialize bit-masks for quads */
1.27 thorpej 391: for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
392: qbmasks[i] = (uint64_t)1 << i;
393: uq = ~(uint64_t)0 << i;
1.1 cgd 394: qumasks[i] = uq;
395: qlmasks[i] = ~uq;
396: }
397: qumasks[i] = 0;
1.27 thorpej 398: qlmasks[i] = ~(uint64_t)0;
1.1 cgd 399: }
400:
401: /*
402: * Get a free sbuf structure, if possible from the free list
403: */
404: static sbuf_t *
1.19 lukem 405: allocsb(void)
1.1 cgd 406: {
407: sbuf_t *sb;
408:
409: if ((sb = sbfrlst) != NULL) {
410: sbfrlst = sb->sb_nxt;
1.55 christos 411: #ifdef BLKDEBUG
412: (void)memset(sb, 0, sizeof (*sb));
413: #else
414: sb->sb_nxt = NULL;
415: #endif
1.1 cgd 416: } else {
417: sb = xmalloc(sizeof (sbuf_t));
1.55 christos 418: (void)memset(sb, 0, sizeof (*sb));
1.1 cgd 419: }
420: return (sb);
421: }
422:
423: /*
424: * Put a sbuf structure to the free list
425: */
426: static void
1.19 lukem 427: freesb(sbuf_t *sb)
1.1 cgd 428: {
1.19 lukem 429:
1.55 christos 430: (void)memset(sb, ZERO, sizeof (*sb));
1.1 cgd 431: sb->sb_nxt = sbfrlst;
432: sbfrlst = sb;
433: }
434:
435: /*
436: * Read a character and ensure that it is positive (except EOF).
437: * Increment line count(s) if necessary.
438: */
439: static int
1.19 lukem 440: inpc(void)
1.1 cgd 441: {
442: int c;
443:
444: if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
445: incline();
446: return (c);
447: }
448:
449: static int
1.19 lukem 450: hash(const char *s)
1.1 cgd 451: {
452: u_int v;
453: const u_char *us;
454:
455: v = 0;
456: for (us = (const u_char *)s; *us != '\0'; us++) {
457: v = (v << sizeof (v)) + *us;
458: v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
459: }
460: return (v % HSHSIZ1);
461: }
462:
463: /*
464: * Lex has found a letter followed by zero or more letters or digits.
465: * It looks for a symbol in the symbol table with the same name. This
466: * symbol must either be a keyword or a symbol of the type required by
467: * symtyp (label, member, tag, ...).
468: *
469: * If it is a keyword, the token is returned. In some cases it is described
470: * more deeply by data written to yylval.
471: *
472: * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
1.60 snj 473: * is stored in yylval. This struct contains the name of the symbol, its
1.1 cgd 474: * length and hash value. If there is already a symbol of the same name
475: * and type in the symbol table, the sbuf struct also contains a pointer
476: * to the symbol table entry.
477: */
478: static int
1.19 lukem 479: name(void)
1.1 cgd 480: {
481: char *s;
482: sbuf_t *sb;
483: sym_t *sym;
484: int tok;
485:
486: sb = allocsb();
487: sb->sb_name = yytext;
488: sb->sb_len = yyleng;
489: sb->sb_hash = hash(yytext);
490: if ((sym = search(sb)) != NULL && sym->s_keyw) {
1.56 christos 491: freesb(sb);
492: return (keyw(sym));
1.1 cgd 493: }
494:
495: sb->sb_sym = sym;
496:
497: if (sym != NULL) {
498: if (blklev < sym->s_blklev)
1.28 christos 499: LERROR("name()");
1.1 cgd 500: sb->sb_name = sym->s_name;
501: sb->sb_len = strlen(sym->s_name);
502: tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
503: } else {
504: s = getblk(yyleng + 1);
505: (void)memcpy(s, yytext, yyleng + 1);
506: sb->sb_name = s;
507: sb->sb_len = yyleng;
508: tok = T_NAME;
509: }
510:
511: yylval.y_sb = sb;
512: return (tok);
513: }
514:
515: static sym_t *
1.19 lukem 516: search(sbuf_t *sb)
1.1 cgd 517: {
518: sym_t *sym;
519:
520: for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
521: if (strcmp(sym->s_name, sb->sb_name) == 0) {
1.56 christos 522: if (sym->s_keyw) {
523: struct kwtab *kw = sym->s_keyw;
524: if (!kw->kw_attr || attron)
525: return (sym);
1.72 christos 526: } else if (!attron && sym->s_kind == symtyp)
1.56 christos 527: return sym;
1.1 cgd 528: }
529: }
530:
531: return (NULL);
532: }
1.19 lukem 533:
1.1 cgd 534: static int
1.19 lukem 535: keyw(sym_t *sym)
1.1 cgd 536: {
537: int t;
538:
539: if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
540: yylval.y_scl = sym->s_scl;
541: } else if (t == T_TYPE || t == T_SOU) {
542: yylval.y_tspec = sym->s_tspec;
543: } else if (t == T_QUAL) {
544: yylval.y_tqual = sym->s_tqual;
545: }
546: return (t);
547: }
548:
549: /*
550: * Convert a string representing an integer into internal representation.
551: * The value is returned in yylval. icon() (and yylex()) returns T_CON.
552: */
553: static int
1.19 lukem 554: icon(int base)
1.1 cgd 555: {
556: int l_suffix, u_suffix;
557: int len;
1.64 christos 558: const char *cp;
1.1 cgd 559: char c, *eptr;
560: tspec_t typ;
1.83 christos 561: int ansiu;
1.84 christos 562: #ifdef TARG_INT128_MAX
1.83 christos 563: __uint128_t uq = 0;
564: static tspec_t contypes[2][4] = {
565: { INT, LONG, QUAD, INT128, },
566: { UINT, ULONG, UQUAD, UINT128, }
567: };
568: #else
1.27 thorpej 569: uint64_t uq = 0;
1.1 cgd 570: static tspec_t contypes[2][3] = {
1.83 christos 571: { INT, LONG, QUAD, },
572: { UINT, ULONG, UQUAD, }
1.1 cgd 573: };
1.83 christos 574: #endif
1.1 cgd 575:
576: cp = yytext;
577: len = yyleng;
578:
1.62 christos 579: /* skip 0[xX] or 0[bB] */
580: if (base == 16 || base == 2) {
1.1 cgd 581: cp += 2;
582: len -= 2;
583: }
584:
585: /* read suffixes */
586: l_suffix = u_suffix = 0;
587: for ( ; ; ) {
588: if ((c = cp[len - 1]) == 'l' || c == 'L') {
589: l_suffix++;
590: } else if (c == 'u' || c == 'U') {
591: u_suffix++;
592: } else {
593: break;
594: }
595: len--;
596: }
597: if (l_suffix > 2 || u_suffix > 1) {
598: /* malformed integer constant */
599: warning(251);
600: if (l_suffix > 2)
601: l_suffix = 2;
602: if (u_suffix > 1)
603: u_suffix = 1;
604: }
605: if (tflag && u_suffix != 0) {
606: /* suffix U is illegal in traditional C */
607: warning(97);
608: }
609: typ = contypes[u_suffix][l_suffix];
610:
611: errno = 0;
1.35 he 612:
1.64 christos 613: uq = strtouq(cp, &eptr, base);
1.1 cgd 614: if (eptr != cp + len)
1.28 christos 615: LERROR("icon()");
1.1 cgd 616: if (errno != 0)
617: /* integer constant out of range */
618: warning(252);
619:
620: /*
1.35 he 621: * If the value is too big for the current type, we must choose
1.1 cgd 622: * another type.
623: */
624: ansiu = 0;
625: switch (typ) {
626: case INT:
1.35 he 627: if (uq <= TARG_INT_MAX) {
1.1 cgd 628: /* ok */
1.35 he 629: } else if (uq <= TARG_UINT_MAX && base != 10) {
1.1 cgd 630: typ = UINT;
1.35 he 631: } else if (uq <= TARG_LONG_MAX) {
1.1 cgd 632: typ = LONG;
633: } else {
634: typ = ULONG;
1.35 he 635: if (uq > TARG_ULONG_MAX) {
636: /* integer constant out of range */
637: warning(252);
638: }
1.1 cgd 639: }
640: if (typ == UINT || typ == ULONG) {
641: if (tflag) {
642: typ = LONG;
643: } else if (!sflag) {
644: /*
645: * Remember that the constant is unsigned
646: * only in ANSI C
647: */
648: ansiu = 1;
649: }
650: }
651: break;
652: case UINT:
1.35 he 653: if (uq > TARG_UINT_MAX) {
1.1 cgd 654: typ = ULONG;
1.35 he 655: if (uq > TARG_ULONG_MAX) {
656: /* integer constant out of range */
657: warning(252);
658: }
659: }
1.1 cgd 660: break;
661: case LONG:
1.35 he 662: if (uq > TARG_LONG_MAX && !tflag) {
1.1 cgd 663: typ = ULONG;
664: if (!sflag)
665: ansiu = 1;
1.35 he 666: if (uq > TARG_ULONG_MAX) {
667: /* integer constant out of range */
668: warning(252);
669: }
1.1 cgd 670: }
671: break;
1.39 joerg 672: case ULONG:
673: if (uq > TARG_ULONG_MAX) {
674: /* integer constant out of range */
675: warning(252);
676: }
677: break;
1.1 cgd 678: case QUAD:
1.35 he 679: if (uq > TARG_QUAD_MAX && !tflag) {
1.1 cgd 680: typ = UQUAD;
681: if (!sflag)
682: ansiu = 1;
683: }
684: break;
1.39 joerg 685: case UQUAD:
686: if (uq > TARG_UQUAD_MAX) {
687: /* integer constant out of range */
688: warning(252);
689: }
690: break;
1.83 christos 691: #ifdef INT128_SIZE
692: case INT128:
1.84 christos 693: #ifdef TARG_INT128_MAX
1.83 christos 694: if (uq > TARG_INT128_MAX && !tflag) {
695: typ = UINT128;
696: if (!sflag)
697: ansiu = 1;
698: }
1.84 christos 699: #endif
1.83 christos 700: break;
701: case UINT128:
1.84 christos 702: #ifdef TARG_INT128_MAX
1.83 christos 703: if (uq > TARG_UINT128_MAX) {
704: /* integer constant out of range */
705: warning(252);
706: }
1.84 christos 707: #endif
1.83 christos 708: break;
709: #endif
1.52 christos 710: /* LINTED206: (enumeration values not handled in switch) */
1.11 christos 711: case STRUCT:
712: case VOID:
713: case LDOUBLE:
714: case FUNC:
715: case ARRAY:
716: case PTR:
717: case ENUM:
718: case UNION:
719: case SIGNED:
720: case NOTSPEC:
721: case DOUBLE:
722: case FLOAT:
723: case USHORT:
724: case SHORT:
725: case UCHAR:
726: case SCHAR:
727: case CHAR:
1.33 yamt 728: case BOOL:
1.11 christos 729: case UNSIGN:
1.38 christos 730: case FCOMPLEX:
731: case DCOMPLEX:
1.40 matt 732: case LCOMPLEX:
1.38 christos 733: case COMPLEX:
1.23 thorpej 734: break;
735:
736: case NTSPEC: /* this value unused */
1.11 christos 737: break;
1.1 cgd 738: }
739:
1.27 thorpej 740: uq = (uint64_t)xsign((int64_t)uq, typ, -1);
1.1 cgd 741:
742: (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
743: yylval.y_val->v_ansiu = ansiu;
1.27 thorpej 744: yylval.y_val->v_quad = (int64_t)uq;
1.1 cgd 745:
746: return (T_CON);
747: }
748:
749: /*
750: * Returns 1 if t is a signed type and the value is negative.
751: *
752: * len is the number of significant bits. If len is -1, len is set
753: * to the width of type t.
754: */
755: int
1.27 thorpej 756: sign(int64_t q, tspec_t t, int len)
1.1 cgd 757: {
1.19 lukem 758:
1.1 cgd 759: if (t == PTR || isutyp(t))
760: return (0);
761: return (msb(q, t, len));
762: }
763:
764: int
1.27 thorpej 765: msb(int64_t q, tspec_t t, int len)
1.1 cgd 766: {
1.19 lukem 767:
1.1 cgd 768: if (len <= 0)
769: len = size(t);
770: return ((q & qbmasks[len - 1]) != 0);
771: }
772:
773: /*
774: * Extends the sign of q.
775: */
1.27 thorpej 776: int64_t
777: xsign(int64_t q, tspec_t t, int len)
1.1 cgd 778: {
1.19 lukem 779:
1.1 cgd 780: if (len <= 0)
781: len = size(t);
782:
783: if (t == PTR || isutyp(t) || !sign(q, t, len)) {
784: q &= qlmasks[len];
785: } else {
786: q |= qumasks[len];
787: }
788: return (q);
789: }
790:
791: /*
792: * Convert a string representing a floating point value into its interal
793: * representation. Type and value are returned in yylval. fcon()
794: * (and yylex()) returns T_CON.
795: * XXX Currently it is not possible to convert constants of type
1.20 wiz 796: * long double which are greater than DBL_MAX.
1.1 cgd 797: */
798: static int
1.19 lukem 799: fcon(void)
1.1 cgd 800: {
801: const char *cp;
802: int len;
803: tspec_t typ;
804: char c, *eptr;
805: double d;
1.11 christos 806: float f = 0;
1.1 cgd 807:
808: cp = yytext;
809: len = yyleng;
810:
1.38 christos 811: if (cp[len - 1] == 'i') {
812: /* imaginary, do nothing for now */
813: len--;
814: }
1.1 cgd 815: if ((c = cp[len - 1]) == 'f' || c == 'F') {
816: typ = FLOAT;
817: len--;
818: } else if (c == 'l' || c == 'L') {
819: typ = LDOUBLE;
820: len--;
821: } else {
1.49 christos 822: if (c == 'd' || c == 'D')
823: len--;
1.1 cgd 824: typ = DOUBLE;
825: }
826:
827: if (tflag && typ != DOUBLE) {
828: /* suffixes F and L are illegal in traditional C */
829: warning(98);
830: }
831:
832: errno = 0;
833: d = strtod(cp, &eptr);
1.37 he 834: if (eptr != cp + len) {
835: switch (*eptr) {
836: /*
837: * XXX: non-native non-current strtod() may not handle hex
838: * floats, ignore the rest if we find traces of hex float
839: * syntax...
840: */
841: case 'p':
842: case 'P':
843: case 'x':
844: case 'X':
845: d = 0;
846: errno = 0;
847: break;
848: default:
1.49 christos 849: LERROR("fcon(%s->%s)", cp, eptr);
1.37 he 850: }
851: }
1.1 cgd 852: if (errno != 0)
853: /* floating-point constant out of range */
854: warning(248);
855:
856: if (typ == FLOAT) {
857: f = (float)d;
1.26 tv 858: if (!finite(f)) {
1.1 cgd 859: /* floating-point constant out of range */
860: warning(248);
861: f = f > 0 ? FLT_MAX : -FLT_MAX;
862: }
863: }
864:
865: (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
866: if (typ == FLOAT) {
867: yylval.y_val->v_ldbl = f;
868: } else {
869: yylval.y_val->v_ldbl = d;
870: }
871:
872: return (T_CON);
873: }
874:
875: static int
1.19 lukem 876: operator(int t, op_t o)
1.1 cgd 877: {
1.19 lukem 878:
1.1 cgd 879: yylval.y_op = o;
880: return (t);
881: }
882:
883: /*
884: * Called if lex found a leading \'.
885: */
886: static int
1.19 lukem 887: ccon(void)
1.1 cgd 888: {
1.43 christos 889: size_t n;
890: int val, c;
1.1 cgd 891: char cv;
892:
893: n = 0;
894: val = 0;
895: while ((c = getescc('\'')) >= 0) {
896: val = (val << CHAR_BIT) + c;
897: n++;
898: }
899: if (c == -2) {
900: /* unterminated character constant */
901: error(253);
902: } else {
903: if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
904: /* too many characters in character constant */
905: error(71);
906: } else if (n > 1) {
907: /* multi-character character constant */
908: warning(294);
909: } else if (n == 0) {
910: /* empty character constant */
911: error(73);
912: }
913: }
914: if (n == 1) {
915: cv = (char)val;
916: val = cv;
917: }
1.19 lukem 918:
1.1 cgd 919: yylval.y_val = xcalloc(1, sizeof (val_t));
920: yylval.y_val->v_tspec = INT;
921: yylval.y_val->v_quad = val;
922:
923: return (T_CON);
924: }
925:
926: /*
927: * Called if lex found a leading L\'
928: */
929: static int
1.19 lukem 930: wccon(void)
1.1 cgd 931: {
1.8 jpo 932: static char buf[MB_LEN_MAX + 1];
1.43 christos 933: size_t i;
934: int c;
1.1 cgd 935: wchar_t wc;
936:
937: i = 0;
938: while ((c = getescc('\'')) >= 0) {
939: if (i < MB_CUR_MAX)
940: buf[i] = (char)c;
941: i++;
942: }
943:
944: wc = 0;
945:
946: if (c == -2) {
947: /* unterminated character constant */
948: error(253);
949: } else if (c == 0) {
950: /* empty character constant */
951: error(73);
952: } else {
953: if (i > MB_CUR_MAX) {
954: i = MB_CUR_MAX;
955: /* too many characters in character constant */
956: error(71);
957: } else {
958: buf[i] = '\0';
959: (void)mbtowc(NULL, NULL, 0);
960: if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
961: /* invalid multibyte character */
962: error(291);
963: }
964: }
965:
966: yylval.y_val = xcalloc(1, sizeof (val_t));
967: yylval.y_val->v_tspec = WCHAR;
968: yylval.y_val->v_quad = wc;
969:
970: return (T_CON);
971: }
972:
973: /*
974: * Read a character which is part of a character constant or of a string
975: * and handle escapes.
976: *
977: * The Argument is the character which delimits the character constant or
978: * string.
979: *
980: * Returns -1 if the end of the character constant or string is reached,
1.16 lukem 981: * -2 if the EOF is reached, and the character otherwise.
1.1 cgd 982: */
983: static int
1.19 lukem 984: getescc(int d)
1.1 cgd 985: {
986: static int pbc = -1;
987: int n, c, v;
988:
989: if (pbc == -1) {
990: c = inpc();
991: } else {
992: c = pbc;
993: pbc = -1;
994: }
995: if (c == d)
996: return (-1);
997: switch (c) {
998: case '\n':
1.16 lukem 999: if (tflag) {
1000: /* newline in string or char constant */
1001: error(254);
1002: return (-2);
1003: }
1004: return (c);
1.1 cgd 1005: case EOF:
1006: return (-2);
1007: case '\\':
1008: switch (c = inpc()) {
1009: case '"':
1010: if (tflag && d == '\'')
1011: /* \" inside character constant undef. ... */
1012: warning(262);
1013: return ('"');
1014: case '\'':
1015: return ('\'');
1016: case '?':
1017: if (tflag)
1018: /* \? undefined in traditional C */
1019: warning(263);
1020: return ('?');
1021: case '\\':
1022: return ('\\');
1023: case 'a':
1024: if (tflag)
1025: /* \a undefined in traditional C */
1026: warning(81);
1027: return ('\a');
1028: case 'b':
1029: return ('\b');
1030: case 'f':
1031: return ('\f');
1032: case 'n':
1033: return ('\n');
1034: case 'r':
1035: return ('\r');
1036: case 't':
1037: return ('\t');
1038: case 'v':
1039: if (tflag)
1040: /* \v undefined in traditional C */
1041: warning(264);
1042: return ('\v');
1043: case '8': case '9':
1044: /* bad octal digit %c */
1045: warning(77, c);
1046: /* FALLTHROUGH */
1047: case '0': case '1': case '2': case '3':
1048: case '4': case '5': case '6': case '7':
1049: n = 3;
1050: v = 0;
1051: do {
1052: v = (v << 3) + (c - '0');
1053: c = inpc();
1054: } while (--n && isdigit(c) && (tflag || c <= '7'));
1055: if (tflag && n > 0 && isdigit(c))
1056: /* bad octal digit %c */
1057: warning(77, c);
1058: pbc = c;
1.50 christos 1059: if (v > TARG_UCHAR_MAX) {
1.1 cgd 1060: /* character escape does not fit in char. */
1061: warning(76);
1062: v &= CHAR_MASK;
1063: }
1064: return (v);
1065: case 'x':
1066: if (tflag)
1067: /* \x undefined in traditional C */
1068: warning(82);
1069: v = 0;
1070: n = 0;
1071: while ((c = inpc()) >= 0 && isxdigit(c)) {
1072: c = isdigit(c) ?
1073: c - '0' : toupper(c) - 'A' + 10;
1074: v = (v << 4) + c;
1075: if (n >= 0) {
1076: if ((v & ~CHAR_MASK) != 0) {
1077: /* overflow in hex escape */
1078: warning(75);
1079: n = -1;
1080: } else {
1081: n++;
1082: }
1083: }
1084: }
1085: pbc = c;
1086: if (n == 0) {
1087: /* no hex digits follow \x */
1088: error(74);
1089: } if (n == -1) {
1090: v &= CHAR_MASK;
1091: }
1092: return (v);
1093: case '\n':
1094: return (getescc(d));
1095: case EOF:
1096: return (-2);
1097: default:
1098: if (isprint(c)) {
1099: /* dubious escape \%c */
1100: warning(79, c);
1101: } else {
1102: /* dubious escape \%o */
1103: warning(80, c);
1104: }
1105: }
1106: }
1107: return (c);
1108: }
1109:
1110: /*
1111: * Called for preprocessor directives. Currently implemented are:
1112: * # lineno
1113: * # lineno "filename"
1114: */
1115: static void
1.19 lukem 1116: directive(void)
1.1 cgd 1117: {
1118: const char *cp, *fn;
1119: char c, *eptr;
1120: size_t fnl;
1121: long ln;
1122: static int first = 1;
1123:
1124: /* Go to first non-whitespace after # */
1.19 lukem 1125: for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1126: continue;
1.1 cgd 1127:
1.13 christos 1128: if (!isdigit((unsigned char)c)) {
1.78 christos 1129: if (strncmp(cp, "pragma", 6) == 0
1130: && isspace((unsigned char)cp[6]))
1131: return;
1.1 cgd 1132: error:
1133: /* undefined or invalid # directive */
1134: warning(255);
1135: return;
1136: }
1137: ln = strtol(--cp, &eptr, 10);
1138: if (cp == eptr)
1139: goto error;
1140: if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1141: goto error;
1.17 lukem 1142: while ((c = *cp++) == ' ' || c == '\t')
1143: continue;
1.1 cgd 1144: if (c != '\0') {
1145: if (c != '"')
1146: goto error;
1147: fn = cp;
1148: while ((c = *cp) != '"' && c != '\0')
1149: cp++;
1150: if (c != '"')
1151: goto error;
1152: if ((fnl = cp++ - fn) > PATH_MAX)
1153: goto error;
1.17 lukem 1154: while ((c = *cp++) == ' ' || c == '\t')
1155: continue;
1.1 cgd 1156: #if 0
1157: if (c != '\0')
1158: warning("extra character(s) after directive");
1159: #endif
1.9 cgd 1160:
1161: /* empty string means stdin */
1162: if (fnl == 0) {
1163: fn = "{standard input}";
1164: fnl = 16; /* strlen (fn) */
1165: }
1.1 cgd 1166: curr_pos.p_file = fnnalloc(fn, fnl);
1167: /*
1168: * If this is the first directive, the name is the name
1169: * of the C source file as specified at the command line.
1170: * It is written to the output file.
1171: */
1172: if (first) {
1173: csrc_pos.p_file = curr_pos.p_file;
1.68 christos 1174: outsrc(fnxform(curr_pos.p_file,
1175: strlen(curr_pos.p_file)));
1.1 cgd 1176: first = 0;
1177: }
1178: }
1179: curr_pos.p_line = (int)ln - 1;
1.9 cgd 1180: curr_pos.p_uniq = 0;
1181: if (curr_pos.p_file == csrc_pos.p_file) {
1.1 cgd 1182: csrc_pos.p_line = (int)ln - 1;
1.9 cgd 1183: csrc_pos.p_uniq = 0;
1184: }
1.1 cgd 1185: }
1186:
1187: /*
1188: * Handle lint comments. Following comments are currently understood:
1189: * ARGSUSEDn
1.22 thorpej 1190: * BITFIELDTYPE
1.1 cgd 1191: * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1192: * FALLTHRU FALLTHROUGH
1193: * LINTLIBRARY
1.52 christos 1194: * LINTEDn NOSTRICTn
1.7 jpo 1195: * LONGLONG
1.1 cgd 1196: * NOTREACHED
1197: * PRINTFLIKEn
1198: * PROTOLIB
1199: * SCANFLIKEn
1200: * VARARGSn
1201: * If one of this comments is recognized, the arguments, if any, are
1202: * parsed and a function which handles this comment is called.
1203: */
1204: static void
1.19 lukem 1205: comment(void)
1.1 cgd 1206: {
1207: int c, lc;
1208: static struct {
1209: const char *keywd;
1210: int arg;
1.19 lukem 1211: void (*func)(int);
1.1 cgd 1212: } keywtab[] = {
1213: { "ARGSUSED", 1, argsused },
1.22 thorpej 1214: { "BITFIELDTYPE", 0, bitfieldtype },
1.1 cgd 1215: { "CONSTCOND", 0, constcond },
1216: { "CONSTANTCOND", 0, constcond },
1217: { "CONSTANTCONDITION", 0, constcond },
1218: { "FALLTHRU", 0, fallthru },
1219: { "FALLTHROUGH", 0, fallthru },
1220: { "LINTLIBRARY", 0, lintlib },
1.51 christos 1221: { "LINTED", 1, linted },
1.7 jpo 1222: { "LONGLONG", 0, longlong },
1.52 christos 1223: { "NOSTRICT", 1, linted },
1.1 cgd 1224: { "NOTREACHED", 0, notreach },
1225: { "PRINTFLIKE", 1, printflike },
1226: { "PROTOLIB", 1, protolib },
1227: { "SCANFLIKE", 1, scanflike },
1228: { "VARARGS", 1, varargs },
1229: };
1230: char keywd[32];
1231: char arg[32];
1.43 christos 1232: size_t l, i;
1233: int a;
1.1 cgd 1234: int eoc;
1235:
1236: eoc = 0;
1237:
1238: /* Skip white spaces after the start of the comment */
1.17 lukem 1239: while ((c = inpc()) != EOF && isspace(c))
1240: continue;
1.1 cgd 1241:
1242: /* Read the potential keyword to keywd */
1243: l = 0;
1244: while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1245: keywd[l++] = (char)c;
1246: c = inpc();
1247: }
1248: keywd[l] = '\0';
1249:
1250: /* look for the keyword */
1251: for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1252: if (strcmp(keywtab[i].keywd, keywd) == 0)
1253: break;
1254: }
1255: if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1256: goto skip_rest;
1257:
1258: /* skip white spaces after the keyword */
1259: while (c != EOF && isspace(c))
1260: c = inpc();
1261:
1262: /* read the argument, if the keyword accepts one and there is one */
1263: l = 0;
1264: if (keywtab[i].arg) {
1265: while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1266: arg[l++] = (char)c;
1267: c = inpc();
1268: }
1269: }
1270: arg[l] = '\0';
1271: a = l != 0 ? atoi(arg) : -1;
1272:
1273: /* skip white spaces after the argument */
1274: while (c != EOF && isspace(c))
1275: c = inpc();
1276:
1277: if (c != '*' || (c = inpc()) != '/') {
1278: if (keywtab[i].func != linted)
1279: /* extra characters in lint comment */
1280: warning(257);
1281: } else {
1282: /*
1283: * remember that we have already found the end of the
1284: * comment
1285: */
1286: eoc = 1;
1287: }
1288:
1289: if (keywtab[i].func != NULL)
1290: (*keywtab[i].func)(a);
1291:
1292: skip_rest:
1293: while (!eoc) {
1294: lc = c;
1295: if ((c = inpc()) == EOF) {
1296: /* unterminated comment */
1297: error(256);
1298: break;
1299: }
1300: if (lc == '*' && c == '/')
1301: eoc = 1;
1302: }
1.18 lukem 1303: }
1304:
1305: /*
1306: * Handle // style comments
1307: */
1308: static void
1.19 lukem 1309: slashslashcomment(void)
1.18 lukem 1310: {
1311: int c;
1312:
1.32 perry 1313: if (!Sflag && !gflag)
1.21 wiz 1314: /* // comments only supported in C99 */
1.18 lukem 1315: (void)gnuism(312, tflag ? "traditional" : "ANSI");
1316:
1317: while ((c = inpc()) != EOF && c != '\n')
1318: continue;
1.7 jpo 1319: }
1320:
1321: /*
1322: * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1323: * clrwflgs() is called after function definitions and global and
1324: * local declarations and definitions. It is also called between
1325: * the controlling expression and the body of control statements
1326: * (if, switch, for, while).
1327: */
1328: void
1.19 lukem 1329: clrwflgs(void)
1.7 jpo 1330: {
1.19 lukem 1331:
1.51 christos 1332: lwarn = LWARN_ALL;
1.7 jpo 1333: quadflg = 0;
1334: ccflg = 0;
1.1 cgd 1335: }
1336:
1337: /*
1338: * Strings are stored in a dynamically alloceted buffer and passed
1339: * in yylval.y_xstrg to the parser. The parser or the routines called
1340: * by the parser are responsible for freeing this buffer.
1341: */
1342: static int
1.19 lukem 1343: string(void)
1.1 cgd 1344: {
1345: u_char *s;
1346: int c;
1347: size_t len, max;
1348: strg_t *strg;
1349:
1350: s = xmalloc(max = 64);
1351:
1352: len = 0;
1353: while ((c = getescc('"')) >= 0) {
1354: /* +1 to reserve space for a trailing NUL character */
1355: if (len + 1 == max)
1356: s = xrealloc(s, max *= 2);
1357: s[len++] = (char)c;
1358: }
1359: s[len] = '\0';
1360: if (c == -2)
1361: /* unterminated string constant */
1362: error(258);
1363:
1364: strg = xcalloc(1, sizeof (strg_t));
1365: strg->st_tspec = CHAR;
1366: strg->st_len = len;
1367: strg->st_cp = s;
1368:
1369: yylval.y_strg = strg;
1370: return (T_STRING);
1371: }
1372:
1373: static int
1.19 lukem 1374: wcstrg(void)
1.1 cgd 1375: {
1376: char *s;
1.43 christos 1377: int c, n;
1378: size_t i, wi;
1.1 cgd 1379: size_t len, max, wlen;
1380: wchar_t *ws;
1381: strg_t *strg;
1382:
1383: s = xmalloc(max = 64);
1384: len = 0;
1385: while ((c = getescc('"')) >= 0) {
1386: /* +1 to save space for a trailing NUL character */
1387: if (len + 1 >= max)
1388: s = xrealloc(s, max *= 2);
1389: s[len++] = (char)c;
1390: }
1391: s[len] = '\0';
1392: if (c == -2)
1393: /* unterminated string constant */
1394: error(258);
1395:
1.48 wiz 1396: /* get length of wide-character string */
1.1 cgd 1397: (void)mblen(NULL, 0);
1398: for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1399: if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1400: /* invalid multibyte character */
1401: error(291);
1402: break;
1403: }
1404: if (n == 0)
1405: n = 1;
1406: }
1407:
1408: ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1409:
1410: /* convert from multibyte to wide char */
1411: (void)mbtowc(NULL, NULL, 0);
1412: for (i = 0, wi = 0; i < len; i += n, wi++) {
1413: if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1414: break;
1415: if (n == 0)
1416: n = 1;
1417: }
1418: ws[wi] = 0;
1419: free(s);
1420:
1421: strg = xcalloc(1, sizeof (strg_t));
1422: strg->st_tspec = WCHAR;
1423: strg->st_len = wlen;
1424: strg->st_wcp = ws;
1425:
1426: yylval.y_strg = strg;
1427: return (T_STRING);
1428: }
1429:
1430: /*
1431: * As noted above the scanner does not create new symbol table entries
1432: * for symbols it cannot find in the symbol table. This is to avoid
1433: * putting undeclared symbols into the symbol table if a syntax error
1434: * occurs.
1435: *
1436: * getsym() is called as soon as it is probably ok to put the symbol to
1437: * the symbol table. This does not mean that it is not possible that
1438: * symbols are put to the symbol table which are than not completely
1439: * declared due to syntax errors. To avoid too many problems in this
1440: * case symbols get type int in getsym().
1441: *
1442: * XXX calls to getsym() should be delayed until decl1*() is called
1443: */
1444: sym_t *
1.19 lukem 1445: getsym(sbuf_t *sb)
1.1 cgd 1446: {
1447: dinfo_t *di;
1448: char *s;
1449: sym_t *sym;
1450:
1451: sym = sb->sb_sym;
1452:
1453: /*
1454: * During member declaration it is possible that name() looked
1455: * for symbols of type FVFT, although it should have looked for
1456: * symbols of type FTAG. Same can happen for labels. Both cases
1457: * are compensated here.
1458: */
1459: if (symtyp == FMOS || symtyp == FLAB) {
1460: if (sym == NULL || sym->s_kind == FVFT)
1461: sym = search(sb);
1462: }
1463:
1464: if (sym != NULL) {
1465: if (sym->s_kind != symtyp)
1.56 christos 1466: LERROR("storesym(%d, %d)", sym->s_kind, symtyp);
1.1 cgd 1467: symtyp = FVFT;
1468: freesb(sb);
1469: return (sym);
1470: }
1471:
1472: /* create a new symbol table entry */
1473:
1474: /* labels must always be allocated at level 1 (outhermost block) */
1475: if (symtyp == FLAB) {
1476: sym = getlblk(1, sizeof (sym_t));
1477: s = getlblk(1, sb->sb_len + 1);
1478: (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1479: sym->s_name = s;
1480: sym->s_blklev = 1;
1481: di = dcs;
1.3 jpo 1482: while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1483: di = di->d_nxt;
1484: if (di->d_ctx != AUTO)
1.28 christos 1485: LERROR("storesym()");
1.1 cgd 1486: } else {
1487: sym = getblk(sizeof (sym_t));
1488: sym->s_name = sb->sb_name;
1489: sym->s_blklev = blklev;
1490: di = dcs;
1491: }
1492:
1.9 cgd 1493: UNIQUE_CURR_POS(sym->s_dpos);
1.1 cgd 1494: if ((sym->s_kind = symtyp) != FLAB)
1495: sym->s_type = gettyp(INT);
1496:
1497: symtyp = FVFT;
1498:
1499: if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1500: symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1.53 christos 1501: sym->s_rlink = &symtab[sb->sb_hash];
1502: symtab[sb->sb_hash] = sym;
1.1 cgd 1503:
1.3 jpo 1504: *di->d_ldlsym = sym;
1505: di->d_ldlsym = &sym->s_dlnxt;
1.1 cgd 1506:
1507: freesb(sb);
1508: return (sym);
1.29 christos 1509: }
1510:
1511: /*
1512: * Construct a temporary symbol. The symbol starts with a digit, so that
1513: * it is illegal.
1514: */
1515: sym_t *
1516: mktempsym(type_t *t)
1517: {
1518: static int n = 0;
1519: int h;
1520: char *s = getlblk(blklev, 64);
1521: sym_t *sym = getblk(sizeof (sym_t));
1522:
1523: (void)snprintf(s, 64, "%.8d_tmp", n++);
1524: h = hash(s);
1.30 christos 1525:
1.29 christos 1526: sym->s_name = s;
1527: sym->s_type = t;
1528: sym->s_blklev = blklev;
1529: sym->s_scl = AUTO;
1.30 christos 1530: sym->s_kind = FVFT;
1.31 christos 1531: sym->s_used = 1;
1532: sym->s_set = 1;
1.30 christos 1533:
1.29 christos 1534: if ((sym->s_link = symtab[h]) != NULL)
1535: symtab[h]->s_rlink = &sym->s_link;
1.53 christos 1536: sym->s_rlink = &symtab[h];
1537: symtab[h] = sym;
1.30 christos 1538:
1539: *dcs->d_ldlsym = sym;
1540: dcs->d_ldlsym = &sym->s_dlnxt;
1541:
1.29 christos 1542: return sym;
1.1 cgd 1543: }
1544:
1545: /*
1546: * Remove a symbol forever from the symbol table. s_blklev
1547: * is set to -1 to avoid that the symbol will later be put
1548: * back to the symbol table.
1549: */
1550: void
1.19 lukem 1551: rmsym(sym_t *sym)
1.1 cgd 1552: {
1.19 lukem 1553:
1.1 cgd 1554: if ((*sym->s_rlink = sym->s_link) != NULL)
1555: sym->s_link->s_rlink = sym->s_rlink;
1556: sym->s_blklev = -1;
1557: sym->s_link = NULL;
1558: }
1559:
1560: /*
1561: * Remove a list of symbols declared at one level from the symbol
1562: * table.
1563: */
1564: void
1.19 lukem 1565: rmsyms(sym_t *syms)
1.1 cgd 1566: {
1567: sym_t *sym;
1568:
1569: for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1570: if (sym->s_blklev != -1) {
1571: if ((*sym->s_rlink = sym->s_link) != NULL)
1572: sym->s_link->s_rlink = sym->s_rlink;
1573: sym->s_link = NULL;
1574: sym->s_rlink = NULL;
1575: }
1576: }
1577: }
1578:
1579: /*
1580: * Put a symbol into the symbol table
1581: */
1582: void
1.19 lukem 1583: inssym(int bl, sym_t *sym)
1.1 cgd 1584: {
1585: int h;
1586:
1587: h = hash(sym->s_name);
1588: if ((sym->s_link = symtab[h]) != NULL)
1589: symtab[h]->s_rlink = &sym->s_link;
1.53 christos 1590: sym->s_rlink = &symtab[h];
1591: symtab[h] = sym;
1.1 cgd 1592: sym->s_blklev = bl;
1593: if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1.28 christos 1594: LERROR("inssym()");
1.1 cgd 1595: }
1596:
1597: /*
1598: * Called at level 0 after syntax errors
1599: * Removes all symbols which are not declared at level 0 from the
1600: * symbol table. Also frees all memory which is not associated with
1601: * level 0.
1602: */
1603: void
1.19 lukem 1604: cleanup(void)
1.1 cgd 1605: {
1606: sym_t *sym, *nsym;
1607: int i;
1608:
1609: for (i = 0; i < HSHSIZ1; i++) {
1610: for (sym = symtab[i]; sym != NULL; sym = nsym) {
1611: nsym = sym->s_link;
1612: if (sym->s_blklev >= 1) {
1613: if ((*sym->s_rlink = nsym) != NULL)
1614: nsym->s_rlink = sym->s_rlink;
1615: }
1616: }
1617: }
1618:
1619: for (i = mblklev; i > 0; i--)
1620: freelblk(i);
1621: }
1622:
1623: /*
1624: * Create a new symbol with the name of an existing symbol.
1625: */
1626: sym_t *
1.19 lukem 1627: pushdown(sym_t *sym)
1.1 cgd 1628: {
1629: int h;
1630: sym_t *nsym;
1631:
1632: h = hash(sym->s_name);
1633: nsym = getblk(sizeof (sym_t));
1634: if (sym->s_blklev > blklev)
1.28 christos 1635: LERROR("pushdown()");
1.1 cgd 1636: nsym->s_name = sym->s_name;
1.9 cgd 1637: UNIQUE_CURR_POS(nsym->s_dpos);
1.1 cgd 1638: nsym->s_kind = sym->s_kind;
1639: nsym->s_blklev = blklev;
1640:
1641: if ((nsym->s_link = symtab[h]) != NULL)
1642: symtab[h]->s_rlink = &nsym->s_link;
1.53 christos 1643: nsym->s_rlink = &symtab[h];
1644: symtab[h] = nsym;
1.1 cgd 1645:
1.3 jpo 1646: *dcs->d_ldlsym = nsym;
1647: dcs->d_ldlsym = &nsym->s_dlnxt;
1.1 cgd 1648:
1649: return (nsym);
1.6 jpo 1650: }
1651:
1652: /*
1653: * Free any dynamically allocated memory referenced by
1654: * the value stack or yylval.
1655: * The type of information in yylval is described by tok.
1656: */
1657: void
1.19 lukem 1658: freeyyv(void *sp, int tok)
1.6 jpo 1659: {
1660: if (tok == T_NAME || tok == T_TYPENAME) {
1661: sbuf_t *sb = *(sbuf_t **)sp;
1662: freesb(sb);
1663: } else if (tok == T_CON) {
1664: val_t *val = *(val_t **)sp;
1665: free(val);
1666: } else if (tok == T_STRING) {
1667: strg_t *strg = *(strg_t **)sp;
1668: if (strg->st_tspec == CHAR) {
1669: free(strg->st_cp);
1670: } else if (strg->st_tspec == WCHAR) {
1671: free(strg->st_wcp);
1672: } else {
1.28 christos 1673: LERROR("fryylv()");
1.6 jpo 1674: }
1675: free(strg);
1.19 lukem 1676: }
1.1 cgd 1677: }
CVSweb <webmaster@jp.NetBSD.org>