scan.l revision 1.127 1 1.1 cgd %{
2 1.127 rillig /* $NetBSD: scan.l,v 1.127 2021/01/18 17:20:15 rillig Exp $ */
3 1.2 cgd
4 1.1 cgd /*
5 1.9 cgd * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved.
6 1.1 cgd * Copyright (c) 1994, 1995 Jochen Pohl
7 1.1 cgd * All Rights Reserved.
8 1.1 cgd *
9 1.1 cgd * Redistribution and use in source and binary forms, with or without
10 1.1 cgd * modification, are permitted provided that the following conditions
11 1.1 cgd * are met:
12 1.1 cgd * 1. Redistributions of source code must retain the above copyright
13 1.1 cgd * notice, this list of conditions and the following disclaimer.
14 1.1 cgd * 2. Redistributions in binary form must reproduce the above copyright
15 1.1 cgd * notice, this list of conditions and the following disclaimer in the
16 1.1 cgd * documentation and/or other materials provided with the distribution.
17 1.1 cgd * 3. All advertising materials mentioning features or use of this software
18 1.1 cgd * must display the following acknowledgement:
19 1.1 cgd * This product includes software developed by Jochen Pohl for
20 1.1 cgd * The NetBSD Project.
21 1.1 cgd * 4. The name of the author may not be used to endorse or promote products
22 1.1 cgd * derived from this software without specific prior written permission.
23 1.1 cgd *
24 1.1 cgd * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
25 1.1 cgd * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
26 1.1 cgd * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
27 1.1 cgd * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
28 1.1 cgd * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
29 1.1 cgd * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
30 1.1 cgd * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
31 1.1 cgd * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
32 1.1 cgd * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
33 1.1 cgd * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
34 1.1 cgd */
35 1.1 cgd
36 1.11 christos #include <sys/cdefs.h>
37 1.25 tv #if defined(__RCSID) && !defined(lint)
38 1.127 rillig __RCSID("$NetBSD: scan.l,v 1.127 2021/01/18 17:20:15 rillig Exp $");
39 1.1 cgd #endif
40 1.1 cgd
41 1.1 cgd #include <ctype.h>
42 1.1 cgd #include <errno.h>
43 1.94 rillig #include <float.h>
44 1.94 rillig #include <limits.h>
45 1.1 cgd #include <math.h>
46 1.94 rillig #include <stdlib.h>
47 1.94 rillig #include <string.h>
48 1.1 cgd
49 1.1 cgd #include "lint1.h"
50 1.12 tv #include "cgram.h"
51 1.1 cgd
52 1.115 rillig #define CHAR_MASK ((int)(~(~0U << CHAR_SIZE)))
53 1.1 cgd
54 1.105 rillig /* Current position (it's also updated when an included file is parsed) */
55 1.9 cgd pos_t curr_pos = { 1, "", 0 };
56 1.1 cgd
57 1.1 cgd /*
58 1.1 cgd * Current position in C source (not updated when an included file is
59 1.1 cgd * parsed).
60 1.1 cgd */
61 1.9 cgd pos_t csrc_pos = { 1, "", 0 };
62 1.1 cgd
63 1.55 christos /* Are we parsing a gcc attribute? */
64 1.118 rillig bool attron;
65 1.55 christos
66 1.121 rillig bool in_system_header = false;
67 1.121 rillig
68 1.19 lukem static void incline(void);
69 1.19 lukem static void badchar(int);
70 1.19 lukem static sbuf_t *allocsb(void);
71 1.19 lukem static void freesb(sbuf_t *);
72 1.19 lukem static int inpc(void);
73 1.19 lukem static int hash(const char *);
74 1.19 lukem static sym_t *search(sbuf_t *);
75 1.19 lukem static int name(void);
76 1.19 lukem static int keyw(sym_t *);
77 1.19 lukem static int icon(int);
78 1.19 lukem static int fcon(void);
79 1.19 lukem static int operator(int, op_t);
80 1.19 lukem static int ccon(void);
81 1.19 lukem static int wccon(void);
82 1.19 lukem static int getescc(int);
83 1.19 lukem static void directive(void);
84 1.19 lukem static void comment(void);
85 1.19 lukem static void slashslashcomment(void);
86 1.19 lukem static int string(void);
87 1.19 lukem static int wcstrg(void);
88 1.1 cgd
89 1.1 cgd %}
90 1.1 cgd
91 1.62 christos
92 1.1 cgd L [_A-Za-z]
93 1.1 cgd D [0-9]
94 1.1 cgd NZD [1-9]
95 1.62 christos BD [0-1]
96 1.1 cgd OD [0-7]
97 1.1 cgd HD [0-9A-Fa-f]
98 1.1 cgd EX ([eE][+-]?[0-9]+)
99 1.47 christos HX (p[+-]?[0-9A-Fa-f]+)
100 1.47 christos TL ([fFlL]?[i]?)
101 1.1 cgd
102 1.46 christos %option nounput
103 1.46 christos
104 1.1 cgd %%
105 1.1 cgd
106 1.97 rillig {L}({L}|{D})* return name();
107 1.97 rillig 0[bB]{BD}+[lLuU]* return icon(2);
108 1.97 rillig 0{OD}*[lLuU]* return icon(8);
109 1.97 rillig {NZD}{D}*[lLuU]* return icon(10);
110 1.97 rillig 0[xX]{HD}+[lLuU]* return icon(16);
111 1.47 christos {D}+\.{D}*{EX}?{TL} |
112 1.47 christos {D}+{EX}{TL} |
113 1.47 christos 0[xX]{HD}+\.{HD}*{HX}{TL} |
114 1.96 rillig 0[xX]{HD}+{HX}{TL} |
115 1.97 rillig \.{D}+{EX}?{TL} return fcon();
116 1.126 rillig "=" return operator(T_ASSIGN, NOOP);
117 1.127 rillig "*=" return operator(T_OPASSIGN, MULASS);
118 1.127 rillig "/=" return operator(T_OPASSIGN, DIVASS);
119 1.127 rillig "%=" return operator(T_OPASSIGN, MODASS);
120 1.127 rillig "+=" return operator(T_OPASSIGN, ADDASS);
121 1.127 rillig "-=" return operator(T_OPASSIGN, SUBASS);
122 1.127 rillig "<<=" return operator(T_OPASSIGN, SHLASS);
123 1.127 rillig ">>=" return operator(T_OPASSIGN, SHRASS);
124 1.127 rillig "&=" return operator(T_OPASSIGN, ANDASS);
125 1.127 rillig "^=" return operator(T_OPASSIGN, XORASS);
126 1.127 rillig "|=" return operator(T_OPASSIGN, ORASS);
127 1.126 rillig "||" return operator(T_LOGOR, NOOP);
128 1.126 rillig "&&" return operator(T_LOGAND, NOOP);
129 1.127 rillig "|" return operator(T_BITOR, NOOP);
130 1.126 rillig "&" return operator(T_AMPER, NOOP);
131 1.126 rillig "^" return operator(T_XOR, NOOP);
132 1.127 rillig "==" return operator(T_EQUALITY, EQ);
133 1.127 rillig "!=" return operator(T_EQUALITY, NE);
134 1.127 rillig "<" return operator(T_RELATIONAL, LT);
135 1.127 rillig ">" return operator(T_RELATIONAL, GT);
136 1.127 rillig "<=" return operator(T_RELATIONAL, LE);
137 1.127 rillig ">=" return operator(T_RELATIONAL, GE);
138 1.127 rillig "<<" return operator(T_SHIFT, SHL);
139 1.127 rillig ">>" return operator(T_SHIFT, SHR);
140 1.97 rillig "++" return operator(T_INCDEC, INC);
141 1.97 rillig "--" return operator(T_INCDEC, DEC);
142 1.122 rillig "->" return operator(T_MEMBACC, ARROW);
143 1.122 rillig "." return operator(T_MEMBACC, POINT);
144 1.127 rillig "+" return operator(T_ADDITIVE, PLUS);
145 1.127 rillig "-" return operator(T_ADDITIVE, MINUS);
146 1.117 rillig "*" return operator(T_ASTERISK, NOOP);
147 1.127 rillig "/" return operator(T_MULTIPLICATIVE, DIV);
148 1.127 rillig "%" return operator(T_MULTIPLICATIVE, MOD);
149 1.127 rillig "!" return operator(T_UNARY, NOT);
150 1.127 rillig "~" return operator(T_UNARY, COMPL);
151 1.97 rillig "\"" return string();
152 1.97 rillig "L\"" return wcstrg();
153 1.97 rillig ";" return T_SEMI;
154 1.97 rillig "{" return T_LBRACE;
155 1.97 rillig "}" return T_RBRACE;
156 1.97 rillig "," return T_COMMA;
157 1.97 rillig ":" return T_COLON;
158 1.97 rillig "?" return T_QUEST;
159 1.97 rillig "[" return T_LBRACK;
160 1.97 rillig "]" return T_RBRACK;
161 1.108 rillig "(" return T_LPAREN;
162 1.108 rillig ")" return T_RPAREN;
163 1.123 rillig "..." return T_ELLIPSIS;
164 1.97 rillig "'" return ccon();
165 1.97 rillig "L'" return wccon();
166 1.1 cgd ^#.*$ directive();
167 1.1 cgd \n incline();
168 1.1 cgd \t|" "|\f|\v ;
169 1.1 cgd "/*" comment();
170 1.18 lukem "//" slashslashcomment();
171 1.1 cgd . badchar(yytext[0]);
172 1.1 cgd
173 1.1 cgd %%
174 1.1 cgd
175 1.1 cgd static void
176 1.19 lukem incline(void)
177 1.1 cgd {
178 1.1 cgd curr_pos.p_line++;
179 1.9 cgd curr_pos.p_uniq = 0;
180 1.113 rillig #ifdef DEBUG
181 1.113 rillig printf("parsing %s:%d\n", curr_pos.p_file, curr_pos.p_line);
182 1.113 rillig #endif
183 1.9 cgd if (curr_pos.p_file == csrc_pos.p_file) {
184 1.1 cgd csrc_pos.p_line++;
185 1.9 cgd csrc_pos.p_uniq = 0;
186 1.9 cgd }
187 1.1 cgd }
188 1.1 cgd
189 1.1 cgd static void
190 1.19 lukem badchar(int c)
191 1.1 cgd {
192 1.19 lukem
193 1.1 cgd /* unknown character \%o */
194 1.1 cgd error(250, c);
195 1.1 cgd }
196 1.1 cgd
197 1.1 cgd /*
198 1.1 cgd * Keywords.
199 1.1 cgd * During initialisation they are written to the symbol table.
200 1.1 cgd */
201 1.1 cgd static struct kwtab {
202 1.1 cgd const char *kw_name; /* keyword */
203 1.1 cgd int kw_token; /* token returned by yylex() */
204 1.1 cgd scl_t kw_scl; /* storage class if kw_token T_SCLASS */
205 1.1 cgd tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */
206 1.1 cgd tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */
207 1.110 rillig bool kw_c89 : 1; /* C89 keyword */
208 1.110 rillig bool kw_c99 : 1; /* C99 keyword */
209 1.110 rillig bool kw_gcc : 1; /* GCC keyword */
210 1.110 rillig bool kw_attr : 1; /* GCC attribute, keyword */
211 1.98 rillig u_int kw_deco : 3; /* 1 = name, 2 = __name, 4 = __name__ */
212 1.1 cgd } kwtab[] = {
213 1.83 christos #ifdef INT128_SIZE
214 1.83 christos { "__int128_t", T_TYPE, 0, INT128, 0, 0,1,0,0,1 },
215 1.83 christos { "__uint128_t",T_TYPE, 0, UINT128,0, 0,1,0,0,1 },
216 1.83 christos #endif
217 1.90 christos { "__thread", T_QUAL, 0, 0, THREAD, 0,0,1,0,1 },
218 1.91 christos { "_Alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,1 },
219 1.73 christos { "_Bool", T_TYPE, 0, BOOL, 0, 0,1,0,0,1 },
220 1.73 christos { "_Complex", T_TYPE, 0, COMPLEX,0, 0,1,0,0,1 },
221 1.73 christos { "_Generic", T_GENERIC, 0, 0, 0, 0,1,0,0,1 },
222 1.80 christos { "_Noreturn", T_NORETURN, 0, 0, 0, 0,1,0,0,1 },
223 1.90 christos { "_Thread_local",T_QUAL, 0, 0, THREAD, 0,1,0,0,1 },
224 1.73 christos { "alias", T_AT_ALIAS, 0, 0, 0, 0,0,1,1,5 },
225 1.73 christos { "aligned", T_AT_ALIGNED, 0, 0, 0, 0,0,1,1,5 },
226 1.73 christos { "alignof", T_ALIGNOF, 0, 0, 0, 0,0,0,0,4 },
227 1.87 christos { "alloc_size", T_AT_ALLOC_SIZE,0, 0, 0, 0,0,1,1,5 },
228 1.73 christos { "always_inline", T_AT_ALWAYS_INLINE, 0,0, 0, 0,0,1,1,5 },
229 1.73 christos { "asm", T_ASM, 0, 0, 0, 0,0,1,0,7 },
230 1.73 christos { "attribute", T_ATTRIBUTE, 0, 0, 0, 0,0,1,0,6 },
231 1.73 christos { "auto", T_SCLASS, AUTO, 0, 0, 0,0,0,0,1 },
232 1.75 christos { "bounded", T_AT_BOUNDED, 0, 0, 0, 0,0,1,1,5 },
233 1.73 christos { "break", T_BREAK, 0, 0, 0, 0,0,0,0,1 },
234 1.76 christos { "buffer", T_AT_BUFFER, 0, 0, 0, 0,0,1,1,5 },
235 1.79 christos { "builtin_offsetof", T_BUILTIN_OFFSETOF, 0, 0, 0, 0,0,1,0,2 },
236 1.73 christos { "case", T_CASE, 0, 0, 0, 0,0,0,0,1 },
237 1.73 christos { "char", T_TYPE, 0, CHAR, 0, 0,0,0,0,1 },
238 1.73 christos { "cold", T_AT_COLD, 0, 0, 0, 0,0,1,1,5 },
239 1.73 christos { "const", T_QUAL, 0, 0, CONST, 1,0,0,0,7 },
240 1.73 christos { "constructor",T_AT_CONSTRUCTOR,0, 0, 0, 0,0,1,1,5 },
241 1.73 christos { "continue", T_CONTINUE, 0, 0, 0, 0,0,0,0,1 },
242 1.73 christos { "default", T_DEFAULT, 0, 0, 0, 0,0,0,0,1 },
243 1.73 christos { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0,0,1,1,5 },
244 1.86 christos { "destructor", T_AT_DESTRUCTOR,0, 0, 0, 0,0,1,1,5 },
245 1.73 christos { "do", T_DO, 0, 0, 0, 0,0,0,0,1 },
246 1.73 christos { "double", T_TYPE, 0, DOUBLE, 0, 0,0,0,0,1 },
247 1.73 christos { "else", T_ELSE, 0, 0, 0, 0,0,0,0,1 },
248 1.73 christos { "enum", T_ENUM, 0, 0, 0, 0,0,0,0,1 },
249 1.73 christos { "extension", T_EXTENSION, 0, 0, 0, 0,0,1,0,4 },
250 1.73 christos { "extern", T_SCLASS, EXTERN, 0, 0, 0,0,0,0,1 },
251 1.73 christos { "float", T_TYPE, 0, FLOAT, 0, 0,0,0,0,1 },
252 1.73 christos { "for", T_FOR, 0, 0, 0, 0,0,0,0,1 },
253 1.73 christos { "format", T_AT_FORMAT, 0, 0, 0, 0,0,1,1,5 },
254 1.73 christos { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0,0,1,1,5 },
255 1.73 christos { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0,0,1,1,5 },
256 1.88 christos { "gnu_printf", T_AT_FORMAT_GNU_PRINTF,0,0, 0, 0,0,1,1,5 },
257 1.73 christos { "goto", T_GOTO, 0, 0, 0, 0,0,0,0,1 },
258 1.73 christos { "if", T_IF, 0, 0, 0, 0,0,0,0,1 },
259 1.73 christos { "imag", T_IMAG, 0, 0, 0, 0,1,0,0,4 },
260 1.73 christos { "inline", T_SCLASS, INLINE, 0, 0, 0,1,0,0,7 },
261 1.73 christos { "int", T_TYPE, 0, INT, 0, 0,0,0,0,1 },
262 1.73 christos { "long", T_TYPE, 0, LONG, 0, 0,0,0,0,1 },
263 1.87 christos { "malloc", T_AT_MALLOC, 0, 0, 0, 0,0,1,1,5 },
264 1.73 christos { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0,0,1,1,5 },
265 1.75 christos { "minbytes", T_AT_MINBYTES, 0, 0, 0, 0,0,1,1,5 },
266 1.73 christos { "mode", T_AT_MODE, 0, 0, 0, 0,0,1,1,5 },
267 1.69 christos { "no_instrument_function", T_AT_NO_INSTRUMENT_FUNCTION,
268 1.73 christos 0, 0, 0, 0,0,1,1,5 },
269 1.75 christos { "nonnull", T_AT_NONNULL, 0, 0, 0, 0,0,1,1,5 },
270 1.82 christos { "noinline", T_AT_NOINLINE, 0, 0, 0, 0,0,1,1,5 },
271 1.73 christos { "noreturn", T_AT_NORETURN, 0, 0, 0, 0,0,1,1,5 },
272 1.87 christos { "nothrow", T_AT_NOTHROW, 0, 0, 0, 0,0,1,1,5 },
273 1.92 christos { "optimize", T_AT_OPTIMIZE, 0, 0, 0, 0,0,1,1,5 },
274 1.73 christos { "packed", T_AT_PACKED, 0, 0, 0, 0,0,1,1,5 },
275 1.73 christos { "packed", T_PACKED, 0, 0, 0, 0,0,0,0,2 },
276 1.77 christos { "pcs", T_AT_PCS, 0, 0, 0, 0,0,0,0,5 },
277 1.73 christos { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0,0,1,1,5 },
278 1.73 christos { "pure", T_AT_PURE, 0, 0, 0, 0,0,1,1,5 },
279 1.73 christos { "real", T_REAL, 0, 0, 0, 0,1,0,0,4 },
280 1.73 christos { "register", T_SCLASS, REG, 0, 0, 0,0,0,0,1 },
281 1.73 christos { "restrict", T_QUAL, 0, 0, RESTRICT, 0,1,0,0,5 },
282 1.73 christos { "return", T_RETURN, 0, 0, 0, 0,0,0,0,1 },
283 1.73 christos { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0,0,1,1,5 },
284 1.73 christos { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0,0,1,1,5 },
285 1.73 christos { "section", T_AT_SECTION, 0, 0, 0, 0,0,1,1,7 },
286 1.73 christos { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0,0,1,1,5 },
287 1.73 christos { "short", T_TYPE, 0, SHORT, 0, 0,0,0,0,1 },
288 1.73 christos { "signed", T_TYPE, 0, SIGNED, 0, 1,0,0,0,3 },
289 1.73 christos { "sizeof", T_SIZEOF, 0, 0, 0, 0,0,0,0,1 },
290 1.73 christos { "static", T_SCLASS, STATIC, 0, 0, 0,0,0,0,1 },
291 1.73 christos { "strfmon", T_AT_FORMAT_STRFMON,0, 0, 0, 0,0,1,1,5 },
292 1.73 christos { "strftime", T_AT_FORMAT_STRFTIME,0, 0, 0, 0,0,1,1,5 },
293 1.75 christos { "string", T_AT_STRING, 0, 0, 0, 0,0,1,1,5 },
294 1.73 christos { "struct", T_SOU, 0, STRUCT, 0, 0,0,0,0,1 },
295 1.73 christos { "switch", T_SWITCH, 0, 0, 0, 0,0,0,0,1 },
296 1.89 christos { "symbolrename", T_SYMBOLRENAME,0, 0, 0, 0,0,0,0,2 },
297 1.81 christos { "syslog", T_AT_FORMAT_SYSLOG,0, 0, 0, 0,0,1,1,5 },
298 1.73 christos { "transparent_union",T_AT_TUNION,0, 0, 0, 0,0,1,1,5 },
299 1.89 christos { "tls_model", T_AT_TLS_MODEL, 0, 0, 0, 0,0,1,1,5 },
300 1.73 christos { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0,0,0,0,1 },
301 1.74 christos { "typeof", T_TYPEOF, 0, 0, 0, 0,0,1,0,7 },
302 1.73 christos { "union", T_SOU, 0, UNION, 0, 0,0,0,0,1 },
303 1.73 christos { "unsigned", T_TYPE, 0, UNSIGN, 0, 0,0,0,0,1 },
304 1.73 christos { "unused", T_AT_UNUSED, 0, 0, 0, 0,0,1,1,5 },
305 1.73 christos { "used", T_AT_USED, 0, 0, 0, 0,0,1,1,5 },
306 1.73 christos { "visibility", T_AT_VISIBILITY,0, 0, 0, 0,0,1,1,5 },
307 1.73 christos { "void", T_TYPE, 0, VOID, 0, 0,0,0,0,1 },
308 1.73 christos { "volatile", T_QUAL, 0, 0, VOLATILE, 1,0,0,0,7 },
309 1.85 christos { "warn_unused_result", T_AT_WARN_UNUSED_RESULT, 0, 0, 0, 0,0,1,1,5 },
310 1.73 christos { "weak", T_AT_WEAK, 0, 0, 0, 0,0,1,1,5 },
311 1.73 christos { "while", T_WHILE, 0, 0, 0, 0,0,0,0,1 },
312 1.73 christos { NULL, 0, 0, 0, 0, 0,0,0,0,0 }
313 1.1 cgd };
314 1.1 cgd
315 1.1 cgd /* Symbol table */
316 1.1 cgd static sym_t *symtab[HSHSIZ1];
317 1.1 cgd
318 1.1 cgd /* bit i of the entry with index i is set */
319 1.115 rillig uint64_t qbmasks[64];
320 1.1 cgd
321 1.1 cgd /* least significant i bits are set in the entry with index i */
322 1.115 rillig uint64_t qlmasks[64 + 1];
323 1.1 cgd
324 1.1 cgd /* least significant i bits are not set in the entry with index i */
325 1.115 rillig uint64_t qumasks[64 + 1];
326 1.1 cgd
327 1.1 cgd /* free list for sbuf structures */
328 1.1 cgd static sbuf_t *sbfrlst;
329 1.1 cgd
330 1.1 cgd /* Typ of next expected symbol */
331 1.1 cgd symt_t symtyp;
332 1.1 cgd
333 1.1 cgd
334 1.73 christos static void
335 1.98 rillig add_keyword(struct kwtab *kw, int deco)
336 1.73 christos {
337 1.73 christos sym_t *sym;
338 1.73 christos size_t h;
339 1.73 christos char buf[256];
340 1.73 christos const char *name;
341 1.73 christos
342 1.73 christos if (!(kw->kw_deco & deco))
343 1.73 christos return;
344 1.73 christos
345 1.73 christos switch (deco) {
346 1.73 christos case 1:
347 1.73 christos name = kw->kw_name;
348 1.73 christos break;
349 1.73 christos case 2:
350 1.73 christos snprintf(buf, sizeof(buf), "__%s", kw->kw_name);
351 1.73 christos name = strdup(buf);
352 1.73 christos break;
353 1.125 rillig default:
354 1.125 rillig lint_assert(deco == 4);
355 1.73 christos snprintf(buf, sizeof(buf), "__%s__", kw->kw_name);
356 1.73 christos name = strdup(buf);
357 1.73 christos break;
358 1.73 christos }
359 1.73 christos
360 1.73 christos if (name == NULL)
361 1.73 christos err(1, "Can't init symbol table");
362 1.95 rillig
363 1.73 christos sym = getblk(sizeof (sym_t));
364 1.73 christos sym->s_name = name;
365 1.103 rillig sym->s_keyword = kw;
366 1.73 christos sym->s_value.v_quad = kw->kw_token;
367 1.73 christos if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
368 1.73 christos sym->s_tspec = kw->kw_tspec;
369 1.73 christos } else if (kw->kw_token == T_SCLASS) {
370 1.73 christos sym->s_scl = kw->kw_scl;
371 1.73 christos } else if (kw->kw_token == T_QUAL) {
372 1.73 christos sym->s_tqual = kw->kw_tqual;
373 1.73 christos }
374 1.73 christos h = hash(sym->s_name);
375 1.73 christos if ((sym->s_link = symtab[h]) != NULL)
376 1.73 christos symtab[h]->s_rlink = &sym->s_link;
377 1.73 christos sym->s_rlink = &symtab[h];
378 1.73 christos symtab[h] = sym;
379 1.73 christos }
380 1.73 christos
381 1.1 cgd /*
382 1.1 cgd * All keywords are written to the symbol table. This saves us looking
383 1.1 cgd * in a extra table for each name we found.
384 1.1 cgd */
385 1.1 cgd void
386 1.19 lukem initscan(void)
387 1.1 cgd {
388 1.1 cgd struct kwtab *kw;
389 1.73 christos size_t i;
390 1.27 thorpej uint64_t uq;
391 1.1 cgd
392 1.1 cgd for (kw = kwtab; kw->kw_name != NULL; kw++) {
393 1.32 perry if ((kw->kw_c89 || kw->kw_c99) && tflag)
394 1.32 perry continue;
395 1.32 perry if (kw->kw_c99 && !(Sflag || gflag))
396 1.5 jpo continue;
397 1.5 jpo if (kw->kw_gcc && !gflag)
398 1.5 jpo continue;
399 1.98 rillig add_keyword(kw, 1);
400 1.98 rillig add_keyword(kw, 2);
401 1.98 rillig add_keyword(kw, 4);
402 1.1 cgd }
403 1.1 cgd
404 1.1 cgd /* initialize bit-masks for quads */
405 1.115 rillig for (i = 0; i < 64; i++) {
406 1.27 thorpej qbmasks[i] = (uint64_t)1 << i;
407 1.27 thorpej uq = ~(uint64_t)0 << i;
408 1.1 cgd qumasks[i] = uq;
409 1.1 cgd qlmasks[i] = ~uq;
410 1.1 cgd }
411 1.1 cgd qumasks[i] = 0;
412 1.27 thorpej qlmasks[i] = ~(uint64_t)0;
413 1.1 cgd }
414 1.1 cgd
415 1.1 cgd /*
416 1.1 cgd * Get a free sbuf structure, if possible from the free list
417 1.1 cgd */
418 1.1 cgd static sbuf_t *
419 1.19 lukem allocsb(void)
420 1.1 cgd {
421 1.1 cgd sbuf_t *sb;
422 1.1 cgd
423 1.1 cgd if ((sb = sbfrlst) != NULL) {
424 1.100 rillig sbfrlst = sb->sb_next;
425 1.55 christos #ifdef BLKDEBUG
426 1.55 christos (void)memset(sb, 0, sizeof (*sb));
427 1.55 christos #else
428 1.100 rillig sb->sb_next = NULL;
429 1.55 christos #endif
430 1.1 cgd } else {
431 1.1 cgd sb = xmalloc(sizeof (sbuf_t));
432 1.55 christos (void)memset(sb, 0, sizeof (*sb));
433 1.1 cgd }
434 1.97 rillig return sb;
435 1.1 cgd }
436 1.1 cgd
437 1.1 cgd /*
438 1.1 cgd * Put a sbuf structure to the free list
439 1.1 cgd */
440 1.1 cgd static void
441 1.19 lukem freesb(sbuf_t *sb)
442 1.1 cgd {
443 1.19 lukem
444 1.55 christos (void)memset(sb, ZERO, sizeof (*sb));
445 1.100 rillig sb->sb_next = sbfrlst;
446 1.1 cgd sbfrlst = sb;
447 1.1 cgd }
448 1.1 cgd
449 1.1 cgd /*
450 1.1 cgd * Read a character and ensure that it is positive (except EOF).
451 1.1 cgd * Increment line count(s) if necessary.
452 1.1 cgd */
453 1.1 cgd static int
454 1.19 lukem inpc(void)
455 1.1 cgd {
456 1.1 cgd int c;
457 1.1 cgd
458 1.1 cgd if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
459 1.1 cgd incline();
460 1.97 rillig return c;
461 1.1 cgd }
462 1.1 cgd
463 1.1 cgd static int
464 1.19 lukem hash(const char *s)
465 1.1 cgd {
466 1.1 cgd u_int v;
467 1.1 cgd const u_char *us;
468 1.1 cgd
469 1.1 cgd v = 0;
470 1.1 cgd for (us = (const u_char *)s; *us != '\0'; us++) {
471 1.1 cgd v = (v << sizeof (v)) + *us;
472 1.1 cgd v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
473 1.1 cgd }
474 1.97 rillig return v % HSHSIZ1;
475 1.1 cgd }
476 1.1 cgd
477 1.1 cgd /*
478 1.1 cgd * Lex has found a letter followed by zero or more letters or digits.
479 1.1 cgd * It looks for a symbol in the symbol table with the same name. This
480 1.1 cgd * symbol must either be a keyword or a symbol of the type required by
481 1.1 cgd * symtyp (label, member, tag, ...).
482 1.1 cgd *
483 1.1 cgd * If it is a keyword, the token is returned. In some cases it is described
484 1.1 cgd * more deeply by data written to yylval.
485 1.1 cgd *
486 1.1 cgd * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
487 1.60 snj * is stored in yylval. This struct contains the name of the symbol, its
488 1.1 cgd * length and hash value. If there is already a symbol of the same name
489 1.1 cgd * and type in the symbol table, the sbuf struct also contains a pointer
490 1.1 cgd * to the symbol table entry.
491 1.1 cgd */
492 1.1 cgd static int
493 1.19 lukem name(void)
494 1.1 cgd {
495 1.1 cgd char *s;
496 1.1 cgd sbuf_t *sb;
497 1.1 cgd sym_t *sym;
498 1.1 cgd int tok;
499 1.1 cgd
500 1.1 cgd sb = allocsb();
501 1.1 cgd sb->sb_name = yytext;
502 1.1 cgd sb->sb_len = yyleng;
503 1.1 cgd sb->sb_hash = hash(yytext);
504 1.103 rillig if ((sym = search(sb)) != NULL && sym->s_keyword) {
505 1.56 christos freesb(sb);
506 1.97 rillig return keyw(sym);
507 1.1 cgd }
508 1.1 cgd
509 1.1 cgd sb->sb_sym = sym;
510 1.1 cgd
511 1.1 cgd if (sym != NULL) {
512 1.107 rillig lint_assert(blklev >= sym->s_blklev);
513 1.1 cgd sb->sb_name = sym->s_name;
514 1.1 cgd sb->sb_len = strlen(sym->s_name);
515 1.1 cgd tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
516 1.1 cgd } else {
517 1.1 cgd s = getblk(yyleng + 1);
518 1.1 cgd (void)memcpy(s, yytext, yyleng + 1);
519 1.1 cgd sb->sb_name = s;
520 1.1 cgd sb->sb_len = yyleng;
521 1.1 cgd tok = T_NAME;
522 1.1 cgd }
523 1.1 cgd
524 1.1 cgd yylval.y_sb = sb;
525 1.97 rillig return tok;
526 1.1 cgd }
527 1.1 cgd
528 1.1 cgd static sym_t *
529 1.19 lukem search(sbuf_t *sb)
530 1.1 cgd {
531 1.1 cgd sym_t *sym;
532 1.1 cgd
533 1.1 cgd for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
534 1.1 cgd if (strcmp(sym->s_name, sb->sb_name) == 0) {
535 1.103 rillig if (sym->s_keyword) {
536 1.103 rillig struct kwtab *kw = sym->s_keyword;
537 1.56 christos if (!kw->kw_attr || attron)
538 1.97 rillig return sym;
539 1.72 christos } else if (!attron && sym->s_kind == symtyp)
540 1.56 christos return sym;
541 1.1 cgd }
542 1.1 cgd }
543 1.1 cgd
544 1.97 rillig return NULL;
545 1.1 cgd }
546 1.19 lukem
547 1.1 cgd static int
548 1.19 lukem keyw(sym_t *sym)
549 1.1 cgd {
550 1.1 cgd int t;
551 1.1 cgd
552 1.1 cgd if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
553 1.1 cgd yylval.y_scl = sym->s_scl;
554 1.1 cgd } else if (t == T_TYPE || t == T_SOU) {
555 1.1 cgd yylval.y_tspec = sym->s_tspec;
556 1.1 cgd } else if (t == T_QUAL) {
557 1.1 cgd yylval.y_tqual = sym->s_tqual;
558 1.1 cgd }
559 1.97 rillig return t;
560 1.1 cgd }
561 1.1 cgd
562 1.1 cgd /*
563 1.1 cgd * Convert a string representing an integer into internal representation.
564 1.1 cgd * The value is returned in yylval. icon() (and yylex()) returns T_CON.
565 1.1 cgd */
566 1.1 cgd static int
567 1.19 lukem icon(int base)
568 1.1 cgd {
569 1.1 cgd int l_suffix, u_suffix;
570 1.1 cgd int len;
571 1.64 christos const char *cp;
572 1.1 cgd char c, *eptr;
573 1.1 cgd tspec_t typ;
574 1.118 rillig bool ansiu;
575 1.84 christos #ifdef TARG_INT128_MAX
576 1.83 christos __uint128_t uq = 0;
577 1.83 christos static tspec_t contypes[2][4] = {
578 1.83 christos { INT, LONG, QUAD, INT128, },
579 1.83 christos { UINT, ULONG, UQUAD, UINT128, }
580 1.83 christos };
581 1.83 christos #else
582 1.27 thorpej uint64_t uq = 0;
583 1.1 cgd static tspec_t contypes[2][3] = {
584 1.83 christos { INT, LONG, QUAD, },
585 1.83 christos { UINT, ULONG, UQUAD, }
586 1.1 cgd };
587 1.83 christos #endif
588 1.1 cgd
589 1.1 cgd cp = yytext;
590 1.1 cgd len = yyleng;
591 1.1 cgd
592 1.62 christos /* skip 0[xX] or 0[bB] */
593 1.62 christos if (base == 16 || base == 2) {
594 1.1 cgd cp += 2;
595 1.1 cgd len -= 2;
596 1.1 cgd }
597 1.1 cgd
598 1.1 cgd /* read suffixes */
599 1.1 cgd l_suffix = u_suffix = 0;
600 1.1 cgd for ( ; ; ) {
601 1.1 cgd if ((c = cp[len - 1]) == 'l' || c == 'L') {
602 1.1 cgd l_suffix++;
603 1.1 cgd } else if (c == 'u' || c == 'U') {
604 1.1 cgd u_suffix++;
605 1.1 cgd } else {
606 1.1 cgd break;
607 1.1 cgd }
608 1.1 cgd len--;
609 1.1 cgd }
610 1.1 cgd if (l_suffix > 2 || u_suffix > 1) {
611 1.1 cgd /* malformed integer constant */
612 1.1 cgd warning(251);
613 1.1 cgd if (l_suffix > 2)
614 1.1 cgd l_suffix = 2;
615 1.1 cgd if (u_suffix > 1)
616 1.1 cgd u_suffix = 1;
617 1.1 cgd }
618 1.1 cgd if (tflag && u_suffix != 0) {
619 1.1 cgd /* suffix U is illegal in traditional C */
620 1.1 cgd warning(97);
621 1.1 cgd }
622 1.1 cgd typ = contypes[u_suffix][l_suffix];
623 1.1 cgd
624 1.1 cgd errno = 0;
625 1.35 he
626 1.64 christos uq = strtouq(cp, &eptr, base);
627 1.107 rillig lint_assert(eptr == cp + len);
628 1.1 cgd if (errno != 0)
629 1.1 cgd /* integer constant out of range */
630 1.1 cgd warning(252);
631 1.1 cgd
632 1.1 cgd /*
633 1.35 he * If the value is too big for the current type, we must choose
634 1.1 cgd * another type.
635 1.1 cgd */
636 1.118 rillig ansiu = false;
637 1.1 cgd switch (typ) {
638 1.1 cgd case INT:
639 1.35 he if (uq <= TARG_INT_MAX) {
640 1.1 cgd /* ok */
641 1.35 he } else if (uq <= TARG_UINT_MAX && base != 10) {
642 1.1 cgd typ = UINT;
643 1.35 he } else if (uq <= TARG_LONG_MAX) {
644 1.1 cgd typ = LONG;
645 1.1 cgd } else {
646 1.1 cgd typ = ULONG;
647 1.35 he if (uq > TARG_ULONG_MAX) {
648 1.35 he /* integer constant out of range */
649 1.35 he warning(252);
650 1.35 he }
651 1.1 cgd }
652 1.1 cgd if (typ == UINT || typ == ULONG) {
653 1.1 cgd if (tflag) {
654 1.1 cgd typ = LONG;
655 1.1 cgd } else if (!sflag) {
656 1.1 cgd /*
657 1.1 cgd * Remember that the constant is unsigned
658 1.1 cgd * only in ANSI C
659 1.1 cgd */
660 1.118 rillig ansiu = true;
661 1.1 cgd }
662 1.1 cgd }
663 1.1 cgd break;
664 1.1 cgd case UINT:
665 1.35 he if (uq > TARG_UINT_MAX) {
666 1.1 cgd typ = ULONG;
667 1.35 he if (uq > TARG_ULONG_MAX) {
668 1.35 he /* integer constant out of range */
669 1.35 he warning(252);
670 1.35 he }
671 1.35 he }
672 1.1 cgd break;
673 1.1 cgd case LONG:
674 1.35 he if (uq > TARG_LONG_MAX && !tflag) {
675 1.1 cgd typ = ULONG;
676 1.1 cgd if (!sflag)
677 1.118 rillig ansiu = true;
678 1.35 he if (uq > TARG_ULONG_MAX) {
679 1.35 he /* integer constant out of range */
680 1.35 he warning(252);
681 1.35 he }
682 1.1 cgd }
683 1.1 cgd break;
684 1.39 joerg case ULONG:
685 1.39 joerg if (uq > TARG_ULONG_MAX) {
686 1.39 joerg /* integer constant out of range */
687 1.39 joerg warning(252);
688 1.39 joerg }
689 1.39 joerg break;
690 1.1 cgd case QUAD:
691 1.35 he if (uq > TARG_QUAD_MAX && !tflag) {
692 1.1 cgd typ = UQUAD;
693 1.1 cgd if (!sflag)
694 1.118 rillig ansiu = true;
695 1.1 cgd }
696 1.1 cgd break;
697 1.39 joerg case UQUAD:
698 1.39 joerg if (uq > TARG_UQUAD_MAX) {
699 1.39 joerg /* integer constant out of range */
700 1.39 joerg warning(252);
701 1.39 joerg }
702 1.39 joerg break;
703 1.83 christos #ifdef INT128_SIZE
704 1.83 christos case INT128:
705 1.84 christos #ifdef TARG_INT128_MAX
706 1.83 christos if (uq > TARG_INT128_MAX && !tflag) {
707 1.83 christos typ = UINT128;
708 1.83 christos if (!sflag)
709 1.118 rillig ansiu = true;
710 1.83 christos }
711 1.84 christos #endif
712 1.83 christos break;
713 1.83 christos case UINT128:
714 1.84 christos #ifdef TARG_INT128_MAX
715 1.83 christos if (uq > TARG_UINT128_MAX) {
716 1.83 christos /* integer constant out of range */
717 1.83 christos warning(252);
718 1.83 christos }
719 1.84 christos #endif
720 1.83 christos break;
721 1.83 christos #endif
722 1.52 christos /* LINTED206: (enumeration values not handled in switch) */
723 1.11 christos case STRUCT:
724 1.11 christos case VOID:
725 1.11 christos case LDOUBLE:
726 1.11 christos case FUNC:
727 1.11 christos case ARRAY:
728 1.11 christos case PTR:
729 1.11 christos case ENUM:
730 1.11 christos case UNION:
731 1.11 christos case SIGNED:
732 1.11 christos case NOTSPEC:
733 1.11 christos case DOUBLE:
734 1.11 christos case FLOAT:
735 1.11 christos case USHORT:
736 1.11 christos case SHORT:
737 1.11 christos case UCHAR:
738 1.11 christos case SCHAR:
739 1.11 christos case CHAR:
740 1.33 yamt case BOOL:
741 1.11 christos case UNSIGN:
742 1.38 christos case FCOMPLEX:
743 1.38 christos case DCOMPLEX:
744 1.40 matt case LCOMPLEX:
745 1.38 christos case COMPLEX:
746 1.23 thorpej break;
747 1.1 cgd }
748 1.1 cgd
749 1.27 thorpej uq = (uint64_t)xsign((int64_t)uq, typ, -1);
750 1.1 cgd
751 1.1 cgd (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
752 1.1 cgd yylval.y_val->v_ansiu = ansiu;
753 1.27 thorpej yylval.y_val->v_quad = (int64_t)uq;
754 1.1 cgd
755 1.97 rillig return T_CON;
756 1.1 cgd }
757 1.1 cgd
758 1.1 cgd /*
759 1.1 cgd * Returns 1 if t is a signed type and the value is negative.
760 1.1 cgd *
761 1.1 cgd * len is the number of significant bits. If len is -1, len is set
762 1.1 cgd * to the width of type t.
763 1.1 cgd */
764 1.1 cgd int
765 1.27 thorpej sign(int64_t q, tspec_t t, int len)
766 1.1 cgd {
767 1.19 lukem
768 1.116 rillig if (t == PTR || is_uinteger(t))
769 1.97 rillig return 0;
770 1.97 rillig return msb(q, t, len);
771 1.1 cgd }
772 1.1 cgd
773 1.1 cgd int
774 1.27 thorpej msb(int64_t q, tspec_t t, int len)
775 1.1 cgd {
776 1.19 lukem
777 1.1 cgd if (len <= 0)
778 1.1 cgd len = size(t);
779 1.97 rillig return (q & qbmasks[len - 1]) != 0;
780 1.1 cgd }
781 1.1 cgd
782 1.1 cgd /*
783 1.1 cgd * Extends the sign of q.
784 1.1 cgd */
785 1.27 thorpej int64_t
786 1.27 thorpej xsign(int64_t q, tspec_t t, int len)
787 1.1 cgd {
788 1.19 lukem
789 1.1 cgd if (len <= 0)
790 1.1 cgd len = size(t);
791 1.1 cgd
792 1.116 rillig if (t == PTR || is_uinteger(t) || !sign(q, t, len)) {
793 1.1 cgd q &= qlmasks[len];
794 1.1 cgd } else {
795 1.1 cgd q |= qumasks[len];
796 1.1 cgd }
797 1.97 rillig return q;
798 1.1 cgd }
799 1.1 cgd
800 1.1 cgd /*
801 1.105 rillig * Convert a string representing a floating point value into its integral
802 1.1 cgd * representation. Type and value are returned in yylval. fcon()
803 1.1 cgd * (and yylex()) returns T_CON.
804 1.1 cgd * XXX Currently it is not possible to convert constants of type
805 1.20 wiz * long double which are greater than DBL_MAX.
806 1.1 cgd */
807 1.1 cgd static int
808 1.19 lukem fcon(void)
809 1.1 cgd {
810 1.1 cgd const char *cp;
811 1.1 cgd int len;
812 1.1 cgd tspec_t typ;
813 1.1 cgd char c, *eptr;
814 1.1 cgd double d;
815 1.11 christos float f = 0;
816 1.1 cgd
817 1.1 cgd cp = yytext;
818 1.1 cgd len = yyleng;
819 1.1 cgd
820 1.38 christos if (cp[len - 1] == 'i') {
821 1.38 christos /* imaginary, do nothing for now */
822 1.38 christos len--;
823 1.38 christos }
824 1.1 cgd if ((c = cp[len - 1]) == 'f' || c == 'F') {
825 1.1 cgd typ = FLOAT;
826 1.1 cgd len--;
827 1.1 cgd } else if (c == 'l' || c == 'L') {
828 1.1 cgd typ = LDOUBLE;
829 1.1 cgd len--;
830 1.1 cgd } else {
831 1.95 rillig if (c == 'd' || c == 'D')
832 1.49 christos len--;
833 1.1 cgd typ = DOUBLE;
834 1.1 cgd }
835 1.1 cgd
836 1.1 cgd if (tflag && typ != DOUBLE) {
837 1.1 cgd /* suffixes F and L are illegal in traditional C */
838 1.1 cgd warning(98);
839 1.1 cgd }
840 1.1 cgd
841 1.1 cgd errno = 0;
842 1.1 cgd d = strtod(cp, &eptr);
843 1.37 he if (eptr != cp + len) {
844 1.37 he switch (*eptr) {
845 1.37 he /*
846 1.37 he * XXX: non-native non-current strtod() may not handle hex
847 1.37 he * floats, ignore the rest if we find traces of hex float
848 1.37 he * syntax...
849 1.37 he */
850 1.37 he case 'p':
851 1.37 he case 'P':
852 1.37 he case 'x':
853 1.37 he case 'X':
854 1.37 he d = 0;
855 1.37 he errno = 0;
856 1.37 he break;
857 1.37 he default:
858 1.49 christos LERROR("fcon(%s->%s)", cp, eptr);
859 1.37 he }
860 1.37 he }
861 1.1 cgd if (errno != 0)
862 1.1 cgd /* floating-point constant out of range */
863 1.1 cgd warning(248);
864 1.1 cgd
865 1.1 cgd if (typ == FLOAT) {
866 1.1 cgd f = (float)d;
867 1.26 tv if (!finite(f)) {
868 1.1 cgd /* floating-point constant out of range */
869 1.1 cgd warning(248);
870 1.1 cgd f = f > 0 ? FLT_MAX : -FLT_MAX;
871 1.1 cgd }
872 1.1 cgd }
873 1.1 cgd
874 1.1 cgd (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
875 1.1 cgd if (typ == FLOAT) {
876 1.1 cgd yylval.y_val->v_ldbl = f;
877 1.1 cgd } else {
878 1.1 cgd yylval.y_val->v_ldbl = d;
879 1.1 cgd }
880 1.1 cgd
881 1.97 rillig return T_CON;
882 1.1 cgd }
883 1.1 cgd
884 1.1 cgd static int
885 1.19 lukem operator(int t, op_t o)
886 1.1 cgd {
887 1.19 lukem
888 1.1 cgd yylval.y_op = o;
889 1.97 rillig return t;
890 1.1 cgd }
891 1.1 cgd
892 1.1 cgd /*
893 1.1 cgd * Called if lex found a leading \'.
894 1.1 cgd */
895 1.1 cgd static int
896 1.19 lukem ccon(void)
897 1.1 cgd {
898 1.43 christos size_t n;
899 1.43 christos int val, c;
900 1.1 cgd char cv;
901 1.1 cgd
902 1.1 cgd n = 0;
903 1.1 cgd val = 0;
904 1.1 cgd while ((c = getescc('\'')) >= 0) {
905 1.115 rillig val = (val << CHAR_SIZE) + c;
906 1.1 cgd n++;
907 1.1 cgd }
908 1.1 cgd if (c == -2) {
909 1.1 cgd /* unterminated character constant */
910 1.1 cgd error(253);
911 1.1 cgd } else {
912 1.1 cgd if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
913 1.1 cgd /* too many characters in character constant */
914 1.1 cgd error(71);
915 1.1 cgd } else if (n > 1) {
916 1.1 cgd /* multi-character character constant */
917 1.1 cgd warning(294);
918 1.1 cgd } else if (n == 0) {
919 1.1 cgd /* empty character constant */
920 1.1 cgd error(73);
921 1.1 cgd }
922 1.1 cgd }
923 1.1 cgd if (n == 1) {
924 1.1 cgd cv = (char)val;
925 1.1 cgd val = cv;
926 1.1 cgd }
927 1.19 lukem
928 1.1 cgd yylval.y_val = xcalloc(1, sizeof (val_t));
929 1.1 cgd yylval.y_val->v_tspec = INT;
930 1.1 cgd yylval.y_val->v_quad = val;
931 1.1 cgd
932 1.97 rillig return T_CON;
933 1.1 cgd }
934 1.1 cgd
935 1.1 cgd /*
936 1.1 cgd * Called if lex found a leading L\'
937 1.1 cgd */
938 1.1 cgd static int
939 1.19 lukem wccon(void)
940 1.1 cgd {
941 1.8 jpo static char buf[MB_LEN_MAX + 1];
942 1.43 christos size_t i;
943 1.43 christos int c;
944 1.1 cgd wchar_t wc;
945 1.1 cgd
946 1.1 cgd i = 0;
947 1.1 cgd while ((c = getescc('\'')) >= 0) {
948 1.1 cgd if (i < MB_CUR_MAX)
949 1.1 cgd buf[i] = (char)c;
950 1.1 cgd i++;
951 1.1 cgd }
952 1.1 cgd
953 1.1 cgd wc = 0;
954 1.1 cgd
955 1.1 cgd if (c == -2) {
956 1.1 cgd /* unterminated character constant */
957 1.1 cgd error(253);
958 1.1 cgd } else if (c == 0) {
959 1.1 cgd /* empty character constant */
960 1.1 cgd error(73);
961 1.1 cgd } else {
962 1.1 cgd if (i > MB_CUR_MAX) {
963 1.1 cgd i = MB_CUR_MAX;
964 1.1 cgd /* too many characters in character constant */
965 1.1 cgd error(71);
966 1.1 cgd } else {
967 1.1 cgd buf[i] = '\0';
968 1.1 cgd (void)mbtowc(NULL, NULL, 0);
969 1.1 cgd if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
970 1.1 cgd /* invalid multibyte character */
971 1.1 cgd error(291);
972 1.1 cgd }
973 1.1 cgd }
974 1.1 cgd
975 1.1 cgd yylval.y_val = xcalloc(1, sizeof (val_t));
976 1.1 cgd yylval.y_val->v_tspec = WCHAR;
977 1.1 cgd yylval.y_val->v_quad = wc;
978 1.1 cgd
979 1.97 rillig return T_CON;
980 1.1 cgd }
981 1.1 cgd
982 1.1 cgd /*
983 1.1 cgd * Read a character which is part of a character constant or of a string
984 1.1 cgd * and handle escapes.
985 1.1 cgd *
986 1.1 cgd * The Argument is the character which delimits the character constant or
987 1.1 cgd * string.
988 1.1 cgd *
989 1.1 cgd * Returns -1 if the end of the character constant or string is reached,
990 1.16 lukem * -2 if the EOF is reached, and the character otherwise.
991 1.1 cgd */
992 1.1 cgd static int
993 1.19 lukem getescc(int d)
994 1.1 cgd {
995 1.1 cgd static int pbc = -1;
996 1.1 cgd int n, c, v;
997 1.1 cgd
998 1.1 cgd if (pbc == -1) {
999 1.1 cgd c = inpc();
1000 1.1 cgd } else {
1001 1.1 cgd c = pbc;
1002 1.1 cgd pbc = -1;
1003 1.1 cgd }
1004 1.1 cgd if (c == d)
1005 1.97 rillig return -1;
1006 1.1 cgd switch (c) {
1007 1.1 cgd case '\n':
1008 1.16 lukem if (tflag) {
1009 1.16 lukem /* newline in string or char constant */
1010 1.16 lukem error(254);
1011 1.97 rillig return -2;
1012 1.16 lukem }
1013 1.97 rillig return c;
1014 1.1 cgd case EOF:
1015 1.97 rillig return -2;
1016 1.1 cgd case '\\':
1017 1.1 cgd switch (c = inpc()) {
1018 1.1 cgd case '"':
1019 1.1 cgd if (tflag && d == '\'')
1020 1.106 rillig /* \" inside character constants undef... */
1021 1.1 cgd warning(262);
1022 1.97 rillig return '"';
1023 1.1 cgd case '\'':
1024 1.97 rillig return '\'';
1025 1.1 cgd case '?':
1026 1.1 cgd if (tflag)
1027 1.1 cgd /* \? undefined in traditional C */
1028 1.1 cgd warning(263);
1029 1.97 rillig return '?';
1030 1.1 cgd case '\\':
1031 1.97 rillig return '\\';
1032 1.1 cgd case 'a':
1033 1.1 cgd if (tflag)
1034 1.1 cgd /* \a undefined in traditional C */
1035 1.1 cgd warning(81);
1036 1.97 rillig return '\a';
1037 1.1 cgd case 'b':
1038 1.97 rillig return '\b';
1039 1.1 cgd case 'f':
1040 1.97 rillig return '\f';
1041 1.1 cgd case 'n':
1042 1.97 rillig return '\n';
1043 1.1 cgd case 'r':
1044 1.97 rillig return '\r';
1045 1.1 cgd case 't':
1046 1.97 rillig return '\t';
1047 1.1 cgd case 'v':
1048 1.1 cgd if (tflag)
1049 1.1 cgd /* \v undefined in traditional C */
1050 1.1 cgd warning(264);
1051 1.97 rillig return '\v';
1052 1.1 cgd case '8': case '9':
1053 1.1 cgd /* bad octal digit %c */
1054 1.1 cgd warning(77, c);
1055 1.1 cgd /* FALLTHROUGH */
1056 1.1 cgd case '0': case '1': case '2': case '3':
1057 1.1 cgd case '4': case '5': case '6': case '7':
1058 1.1 cgd n = 3;
1059 1.1 cgd v = 0;
1060 1.1 cgd do {
1061 1.1 cgd v = (v << 3) + (c - '0');
1062 1.1 cgd c = inpc();
1063 1.1 cgd } while (--n && isdigit(c) && (tflag || c <= '7'));
1064 1.1 cgd if (tflag && n > 0 && isdigit(c))
1065 1.1 cgd /* bad octal digit %c */
1066 1.1 cgd warning(77, c);
1067 1.1 cgd pbc = c;
1068 1.50 christos if (v > TARG_UCHAR_MAX) {
1069 1.106 rillig /* character escape does not fit in character */
1070 1.1 cgd warning(76);
1071 1.1 cgd v &= CHAR_MASK;
1072 1.1 cgd }
1073 1.97 rillig return v;
1074 1.1 cgd case 'x':
1075 1.1 cgd if (tflag)
1076 1.1 cgd /* \x undefined in traditional C */
1077 1.1 cgd warning(82);
1078 1.1 cgd v = 0;
1079 1.1 cgd n = 0;
1080 1.1 cgd while ((c = inpc()) >= 0 && isxdigit(c)) {
1081 1.1 cgd c = isdigit(c) ?
1082 1.1 cgd c - '0' : toupper(c) - 'A' + 10;
1083 1.1 cgd v = (v << 4) + c;
1084 1.1 cgd if (n >= 0) {
1085 1.1 cgd if ((v & ~CHAR_MASK) != 0) {
1086 1.1 cgd /* overflow in hex escape */
1087 1.1 cgd warning(75);
1088 1.1 cgd n = -1;
1089 1.1 cgd } else {
1090 1.1 cgd n++;
1091 1.1 cgd }
1092 1.1 cgd }
1093 1.1 cgd }
1094 1.1 cgd pbc = c;
1095 1.1 cgd if (n == 0) {
1096 1.1 cgd /* no hex digits follow \x */
1097 1.1 cgd error(74);
1098 1.1 cgd } if (n == -1) {
1099 1.1 cgd v &= CHAR_MASK;
1100 1.1 cgd }
1101 1.97 rillig return v;
1102 1.1 cgd case '\n':
1103 1.97 rillig return getescc(d);
1104 1.1 cgd case EOF:
1105 1.97 rillig return -2;
1106 1.1 cgd default:
1107 1.1 cgd if (isprint(c)) {
1108 1.1 cgd /* dubious escape \%c */
1109 1.1 cgd warning(79, c);
1110 1.1 cgd } else {
1111 1.1 cgd /* dubious escape \%o */
1112 1.1 cgd warning(80, c);
1113 1.1 cgd }
1114 1.1 cgd }
1115 1.1 cgd }
1116 1.97 rillig return c;
1117 1.1 cgd }
1118 1.1 cgd
1119 1.121 rillig /* See https://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html */
1120 1.121 rillig static void
1121 1.121 rillig parse_line_directive_flags(const char *p)
1122 1.121 rillig {
1123 1.121 rillig
1124 1.121 rillig in_system_header = false;
1125 1.121 rillig
1126 1.121 rillig while (ch_isspace(*p))
1127 1.121 rillig p++;
1128 1.121 rillig while (ch_isdigit(*p)) {
1129 1.121 rillig if (*p == '3' && !ch_isdigit(p[1]))
1130 1.121 rillig in_system_header = true;
1131 1.121 rillig p++;
1132 1.121 rillig while (ch_isspace(*p))
1133 1.121 rillig p++;
1134 1.121 rillig }
1135 1.121 rillig
1136 1.121 rillig #if 0
1137 1.121 rillig if (c != '\0')
1138 1.121 rillig warning("extra character(s) after directive");
1139 1.121 rillig #endif
1140 1.121 rillig }
1141 1.121 rillig
1142 1.1 cgd /*
1143 1.1 cgd * Called for preprocessor directives. Currently implemented are:
1144 1.1 cgd * # lineno
1145 1.1 cgd * # lineno "filename"
1146 1.1 cgd */
1147 1.1 cgd static void
1148 1.19 lukem directive(void)
1149 1.1 cgd {
1150 1.1 cgd const char *cp, *fn;
1151 1.1 cgd char c, *eptr;
1152 1.1 cgd size_t fnl;
1153 1.1 cgd long ln;
1154 1.118 rillig static bool first = true;
1155 1.1 cgd
1156 1.1 cgd /* Go to first non-whitespace after # */
1157 1.19 lukem for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1158 1.19 lukem continue;
1159 1.1 cgd
1160 1.118 rillig if (!ch_isdigit(c)) {
1161 1.118 rillig if (strncmp(cp, "pragma", 6) == 0 && ch_isspace(cp[6]))
1162 1.78 christos return;
1163 1.1 cgd error:
1164 1.1 cgd /* undefined or invalid # directive */
1165 1.1 cgd warning(255);
1166 1.1 cgd return;
1167 1.1 cgd }
1168 1.1 cgd ln = strtol(--cp, &eptr, 10);
1169 1.1 cgd if (cp == eptr)
1170 1.1 cgd goto error;
1171 1.1 cgd if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1172 1.1 cgd goto error;
1173 1.17 lukem while ((c = *cp++) == ' ' || c == '\t')
1174 1.17 lukem continue;
1175 1.1 cgd if (c != '\0') {
1176 1.1 cgd if (c != '"')
1177 1.1 cgd goto error;
1178 1.1 cgd fn = cp;
1179 1.1 cgd while ((c = *cp) != '"' && c != '\0')
1180 1.1 cgd cp++;
1181 1.1 cgd if (c != '"')
1182 1.1 cgd goto error;
1183 1.1 cgd if ((fnl = cp++ - fn) > PATH_MAX)
1184 1.1 cgd goto error;
1185 1.121 rillig parse_line_directive_flags(cp);
1186 1.9 cgd
1187 1.9 cgd /* empty string means stdin */
1188 1.9 cgd if (fnl == 0) {
1189 1.9 cgd fn = "{standard input}";
1190 1.9 cgd fnl = 16; /* strlen (fn) */
1191 1.9 cgd }
1192 1.1 cgd curr_pos.p_file = fnnalloc(fn, fnl);
1193 1.1 cgd /*
1194 1.1 cgd * If this is the first directive, the name is the name
1195 1.1 cgd * of the C source file as specified at the command line.
1196 1.1 cgd * It is written to the output file.
1197 1.1 cgd */
1198 1.1 cgd if (first) {
1199 1.1 cgd csrc_pos.p_file = curr_pos.p_file;
1200 1.68 christos outsrc(fnxform(curr_pos.p_file,
1201 1.68 christos strlen(curr_pos.p_file)));
1202 1.118 rillig first = false;
1203 1.1 cgd }
1204 1.1 cgd }
1205 1.1 cgd curr_pos.p_line = (int)ln - 1;
1206 1.9 cgd curr_pos.p_uniq = 0;
1207 1.9 cgd if (curr_pos.p_file == csrc_pos.p_file) {
1208 1.1 cgd csrc_pos.p_line = (int)ln - 1;
1209 1.9 cgd csrc_pos.p_uniq = 0;
1210 1.9 cgd }
1211 1.1 cgd }
1212 1.1 cgd
1213 1.1 cgd /*
1214 1.1 cgd * Handle lint comments. Following comments are currently understood:
1215 1.1 cgd * ARGSUSEDn
1216 1.22 thorpej * BITFIELDTYPE
1217 1.1 cgd * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1218 1.1 cgd * FALLTHRU FALLTHROUGH
1219 1.1 cgd * LINTLIBRARY
1220 1.52 christos * LINTEDn NOSTRICTn
1221 1.7 jpo * LONGLONG
1222 1.1 cgd * NOTREACHED
1223 1.1 cgd * PRINTFLIKEn
1224 1.1 cgd * PROTOLIB
1225 1.1 cgd * SCANFLIKEn
1226 1.1 cgd * VARARGSn
1227 1.1 cgd * If one of this comments is recognized, the arguments, if any, are
1228 1.1 cgd * parsed and a function which handles this comment is called.
1229 1.1 cgd */
1230 1.1 cgd static void
1231 1.19 lukem comment(void)
1232 1.1 cgd {
1233 1.1 cgd int c, lc;
1234 1.1 cgd static struct {
1235 1.1 cgd const char *keywd;
1236 1.1 cgd int arg;
1237 1.19 lukem void (*func)(int);
1238 1.1 cgd } keywtab[] = {
1239 1.1 cgd { "ARGSUSED", 1, argsused },
1240 1.22 thorpej { "BITFIELDTYPE", 0, bitfieldtype },
1241 1.1 cgd { "CONSTCOND", 0, constcond },
1242 1.1 cgd { "CONSTANTCOND", 0, constcond },
1243 1.1 cgd { "CONSTANTCONDITION", 0, constcond },
1244 1.1 cgd { "FALLTHRU", 0, fallthru },
1245 1.1 cgd { "FALLTHROUGH", 0, fallthru },
1246 1.1 cgd { "LINTLIBRARY", 0, lintlib },
1247 1.51 christos { "LINTED", 1, linted },
1248 1.7 jpo { "LONGLONG", 0, longlong },
1249 1.52 christos { "NOSTRICT", 1, linted },
1250 1.1 cgd { "NOTREACHED", 0, notreach },
1251 1.1 cgd { "PRINTFLIKE", 1, printflike },
1252 1.1 cgd { "PROTOLIB", 1, protolib },
1253 1.1 cgd { "SCANFLIKE", 1, scanflike },
1254 1.1 cgd { "VARARGS", 1, varargs },
1255 1.1 cgd };
1256 1.1 cgd char keywd[32];
1257 1.1 cgd char arg[32];
1258 1.43 christos size_t l, i;
1259 1.43 christos int a;
1260 1.118 rillig bool eoc;
1261 1.1 cgd
1262 1.118 rillig eoc = false;
1263 1.1 cgd
1264 1.1 cgd /* Skip white spaces after the start of the comment */
1265 1.17 lukem while ((c = inpc()) != EOF && isspace(c))
1266 1.17 lukem continue;
1267 1.1 cgd
1268 1.1 cgd /* Read the potential keyword to keywd */
1269 1.1 cgd l = 0;
1270 1.1 cgd while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1271 1.1 cgd keywd[l++] = (char)c;
1272 1.1 cgd c = inpc();
1273 1.1 cgd }
1274 1.1 cgd keywd[l] = '\0';
1275 1.1 cgd
1276 1.1 cgd /* look for the keyword */
1277 1.1 cgd for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1278 1.1 cgd if (strcmp(keywtab[i].keywd, keywd) == 0)
1279 1.1 cgd break;
1280 1.1 cgd }
1281 1.1 cgd if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1282 1.1 cgd goto skip_rest;
1283 1.1 cgd
1284 1.1 cgd /* skip white spaces after the keyword */
1285 1.1 cgd while (c != EOF && isspace(c))
1286 1.1 cgd c = inpc();
1287 1.1 cgd
1288 1.1 cgd /* read the argument, if the keyword accepts one and there is one */
1289 1.1 cgd l = 0;
1290 1.1 cgd if (keywtab[i].arg) {
1291 1.1 cgd while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1292 1.1 cgd arg[l++] = (char)c;
1293 1.1 cgd c = inpc();
1294 1.1 cgd }
1295 1.1 cgd }
1296 1.1 cgd arg[l] = '\0';
1297 1.1 cgd a = l != 0 ? atoi(arg) : -1;
1298 1.1 cgd
1299 1.1 cgd /* skip white spaces after the argument */
1300 1.1 cgd while (c != EOF && isspace(c))
1301 1.1 cgd c = inpc();
1302 1.1 cgd
1303 1.1 cgd if (c != '*' || (c = inpc()) != '/') {
1304 1.1 cgd if (keywtab[i].func != linted)
1305 1.1 cgd /* extra characters in lint comment */
1306 1.1 cgd warning(257);
1307 1.1 cgd } else {
1308 1.1 cgd /*
1309 1.1 cgd * remember that we have already found the end of the
1310 1.1 cgd * comment
1311 1.1 cgd */
1312 1.118 rillig eoc = true;
1313 1.1 cgd }
1314 1.1 cgd
1315 1.1 cgd if (keywtab[i].func != NULL)
1316 1.1 cgd (*keywtab[i].func)(a);
1317 1.1 cgd
1318 1.1 cgd skip_rest:
1319 1.1 cgd while (!eoc) {
1320 1.1 cgd lc = c;
1321 1.1 cgd if ((c = inpc()) == EOF) {
1322 1.1 cgd /* unterminated comment */
1323 1.1 cgd error(256);
1324 1.1 cgd break;
1325 1.1 cgd }
1326 1.1 cgd if (lc == '*' && c == '/')
1327 1.118 rillig eoc = true;
1328 1.1 cgd }
1329 1.18 lukem }
1330 1.18 lukem
1331 1.18 lukem /*
1332 1.18 lukem * Handle // style comments
1333 1.18 lukem */
1334 1.18 lukem static void
1335 1.19 lukem slashslashcomment(void)
1336 1.18 lukem {
1337 1.18 lukem int c;
1338 1.18 lukem
1339 1.32 perry if (!Sflag && !gflag)
1340 1.111 rillig /* %s C does not support // comments */
1341 1.111 rillig gnuism(312, tflag ? "traditional" : "ANSI");
1342 1.18 lukem
1343 1.18 lukem while ((c = inpc()) != EOF && c != '\n')
1344 1.18 lukem continue;
1345 1.7 jpo }
1346 1.7 jpo
1347 1.7 jpo /*
1348 1.7 jpo * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1349 1.124 rillig * clear_warn_flags() is called after function definitions and global and
1350 1.7 jpo * local declarations and definitions. It is also called between
1351 1.7 jpo * the controlling expression and the body of control statements
1352 1.7 jpo * (if, switch, for, while).
1353 1.7 jpo */
1354 1.7 jpo void
1355 1.124 rillig clear_warn_flags(void)
1356 1.7 jpo {
1357 1.19 lukem
1358 1.51 christos lwarn = LWARN_ALL;
1359 1.118 rillig quadflg = false;
1360 1.118 rillig constcond_flag = false;
1361 1.1 cgd }
1362 1.1 cgd
1363 1.1 cgd /*
1364 1.112 rillig * Strings are stored in a dynamically allocated buffer and passed
1365 1.1 cgd * in yylval.y_xstrg to the parser. The parser or the routines called
1366 1.1 cgd * by the parser are responsible for freeing this buffer.
1367 1.1 cgd */
1368 1.1 cgd static int
1369 1.19 lukem string(void)
1370 1.1 cgd {
1371 1.1 cgd u_char *s;
1372 1.1 cgd int c;
1373 1.1 cgd size_t len, max;
1374 1.1 cgd strg_t *strg;
1375 1.1 cgd
1376 1.1 cgd s = xmalloc(max = 64);
1377 1.1 cgd
1378 1.1 cgd len = 0;
1379 1.1 cgd while ((c = getescc('"')) >= 0) {
1380 1.1 cgd /* +1 to reserve space for a trailing NUL character */
1381 1.1 cgd if (len + 1 == max)
1382 1.1 cgd s = xrealloc(s, max *= 2);
1383 1.1 cgd s[len++] = (char)c;
1384 1.1 cgd }
1385 1.1 cgd s[len] = '\0';
1386 1.1 cgd if (c == -2)
1387 1.1 cgd /* unterminated string constant */
1388 1.1 cgd error(258);
1389 1.1 cgd
1390 1.1 cgd strg = xcalloc(1, sizeof (strg_t));
1391 1.1 cgd strg->st_tspec = CHAR;
1392 1.1 cgd strg->st_len = len;
1393 1.1 cgd strg->st_cp = s;
1394 1.1 cgd
1395 1.102 rillig yylval.y_string = strg;
1396 1.97 rillig return T_STRING;
1397 1.1 cgd }
1398 1.1 cgd
1399 1.1 cgd static int
1400 1.19 lukem wcstrg(void)
1401 1.1 cgd {
1402 1.1 cgd char *s;
1403 1.43 christos int c, n;
1404 1.43 christos size_t i, wi;
1405 1.1 cgd size_t len, max, wlen;
1406 1.1 cgd wchar_t *ws;
1407 1.1 cgd strg_t *strg;
1408 1.1 cgd
1409 1.1 cgd s = xmalloc(max = 64);
1410 1.1 cgd len = 0;
1411 1.1 cgd while ((c = getescc('"')) >= 0) {
1412 1.1 cgd /* +1 to save space for a trailing NUL character */
1413 1.1 cgd if (len + 1 >= max)
1414 1.1 cgd s = xrealloc(s, max *= 2);
1415 1.1 cgd s[len++] = (char)c;
1416 1.1 cgd }
1417 1.1 cgd s[len] = '\0';
1418 1.1 cgd if (c == -2)
1419 1.1 cgd /* unterminated string constant */
1420 1.1 cgd error(258);
1421 1.1 cgd
1422 1.48 wiz /* get length of wide-character string */
1423 1.1 cgd (void)mblen(NULL, 0);
1424 1.1 cgd for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1425 1.1 cgd if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1426 1.1 cgd /* invalid multibyte character */
1427 1.1 cgd error(291);
1428 1.1 cgd break;
1429 1.1 cgd }
1430 1.1 cgd if (n == 0)
1431 1.1 cgd n = 1;
1432 1.1 cgd }
1433 1.1 cgd
1434 1.1 cgd ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1435 1.1 cgd
1436 1.1 cgd /* convert from multibyte to wide char */
1437 1.1 cgd (void)mbtowc(NULL, NULL, 0);
1438 1.1 cgd for (i = 0, wi = 0; i < len; i += n, wi++) {
1439 1.1 cgd if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1440 1.1 cgd break;
1441 1.1 cgd if (n == 0)
1442 1.1 cgd n = 1;
1443 1.1 cgd }
1444 1.1 cgd ws[wi] = 0;
1445 1.1 cgd free(s);
1446 1.1 cgd
1447 1.1 cgd strg = xcalloc(1, sizeof (strg_t));
1448 1.1 cgd strg->st_tspec = WCHAR;
1449 1.1 cgd strg->st_len = wlen;
1450 1.1 cgd strg->st_wcp = ws;
1451 1.1 cgd
1452 1.102 rillig yylval.y_string = strg;
1453 1.97 rillig return T_STRING;
1454 1.1 cgd }
1455 1.1 cgd
1456 1.1 cgd /*
1457 1.1 cgd * As noted above the scanner does not create new symbol table entries
1458 1.1 cgd * for symbols it cannot find in the symbol table. This is to avoid
1459 1.1 cgd * putting undeclared symbols into the symbol table if a syntax error
1460 1.1 cgd * occurs.
1461 1.1 cgd *
1462 1.1 cgd * getsym() is called as soon as it is probably ok to put the symbol to
1463 1.1 cgd * the symbol table. This does not mean that it is not possible that
1464 1.1 cgd * symbols are put to the symbol table which are than not completely
1465 1.1 cgd * declared due to syntax errors. To avoid too many problems in this
1466 1.1 cgd * case symbols get type int in getsym().
1467 1.1 cgd *
1468 1.1 cgd * XXX calls to getsym() should be delayed until decl1*() is called
1469 1.1 cgd */
1470 1.1 cgd sym_t *
1471 1.19 lukem getsym(sbuf_t *sb)
1472 1.1 cgd {
1473 1.1 cgd dinfo_t *di;
1474 1.1 cgd char *s;
1475 1.1 cgd sym_t *sym;
1476 1.1 cgd
1477 1.1 cgd sym = sb->sb_sym;
1478 1.1 cgd
1479 1.1 cgd /*
1480 1.1 cgd * During member declaration it is possible that name() looked
1481 1.1 cgd * for symbols of type FVFT, although it should have looked for
1482 1.1 cgd * symbols of type FTAG. Same can happen for labels. Both cases
1483 1.1 cgd * are compensated here.
1484 1.1 cgd */
1485 1.99 rillig if (symtyp == FMEMBER || symtyp == FLABEL) {
1486 1.1 cgd if (sym == NULL || sym->s_kind == FVFT)
1487 1.1 cgd sym = search(sb);
1488 1.1 cgd }
1489 1.1 cgd
1490 1.1 cgd if (sym != NULL) {
1491 1.1 cgd if (sym->s_kind != symtyp)
1492 1.107 rillig LERROR("getsym(%d, %d)", sym->s_kind, symtyp);
1493 1.1 cgd symtyp = FVFT;
1494 1.1 cgd freesb(sb);
1495 1.97 rillig return sym;
1496 1.1 cgd }
1497 1.1 cgd
1498 1.1 cgd /* create a new symbol table entry */
1499 1.1 cgd
1500 1.1 cgd /* labels must always be allocated at level 1 (outhermost block) */
1501 1.99 rillig if (symtyp == FLABEL) {
1502 1.1 cgd sym = getlblk(1, sizeof (sym_t));
1503 1.1 cgd s = getlblk(1, sb->sb_len + 1);
1504 1.1 cgd (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1505 1.1 cgd sym->s_name = s;
1506 1.1 cgd sym->s_blklev = 1;
1507 1.1 cgd di = dcs;
1508 1.101 rillig while (di->d_next != NULL && di->d_next->d_next != NULL)
1509 1.101 rillig di = di->d_next;
1510 1.107 rillig lint_assert(di->d_ctx == AUTO);
1511 1.1 cgd } else {
1512 1.1 cgd sym = getblk(sizeof (sym_t));
1513 1.1 cgd sym->s_name = sb->sb_name;
1514 1.1 cgd sym->s_blklev = blklev;
1515 1.1 cgd di = dcs;
1516 1.1 cgd }
1517 1.1 cgd
1518 1.104 rillig UNIQUE_CURR_POS(sym->s_def_pos);
1519 1.99 rillig if ((sym->s_kind = symtyp) != FLABEL)
1520 1.1 cgd sym->s_type = gettyp(INT);
1521 1.1 cgd
1522 1.1 cgd symtyp = FVFT;
1523 1.1 cgd
1524 1.1 cgd if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1525 1.1 cgd symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1526 1.53 christos sym->s_rlink = &symtab[sb->sb_hash];
1527 1.53 christos symtab[sb->sb_hash] = sym;
1528 1.1 cgd
1529 1.3 jpo *di->d_ldlsym = sym;
1530 1.3 jpo di->d_ldlsym = &sym->s_dlnxt;
1531 1.1 cgd
1532 1.1 cgd freesb(sb);
1533 1.97 rillig return sym;
1534 1.29 christos }
1535 1.29 christos
1536 1.29 christos /*
1537 1.29 christos * Construct a temporary symbol. The symbol starts with a digit, so that
1538 1.29 christos * it is illegal.
1539 1.29 christos */
1540 1.29 christos sym_t *
1541 1.29 christos mktempsym(type_t *t)
1542 1.29 christos {
1543 1.29 christos static int n = 0;
1544 1.29 christos int h;
1545 1.29 christos char *s = getlblk(blklev, 64);
1546 1.29 christos sym_t *sym = getblk(sizeof (sym_t));
1547 1.29 christos
1548 1.29 christos (void)snprintf(s, 64, "%.8d_tmp", n++);
1549 1.29 christos h = hash(s);
1550 1.30 christos
1551 1.29 christos sym->s_name = s;
1552 1.29 christos sym->s_type = t;
1553 1.29 christos sym->s_blklev = blklev;
1554 1.29 christos sym->s_scl = AUTO;
1555 1.30 christos sym->s_kind = FVFT;
1556 1.118 rillig sym->s_used = true;
1557 1.118 rillig sym->s_set = true;
1558 1.30 christos
1559 1.29 christos if ((sym->s_link = symtab[h]) != NULL)
1560 1.29 christos symtab[h]->s_rlink = &sym->s_link;
1561 1.53 christos sym->s_rlink = &symtab[h];
1562 1.53 christos symtab[h] = sym;
1563 1.30 christos
1564 1.30 christos *dcs->d_ldlsym = sym;
1565 1.30 christos dcs->d_ldlsym = &sym->s_dlnxt;
1566 1.30 christos
1567 1.29 christos return sym;
1568 1.1 cgd }
1569 1.1 cgd
1570 1.1 cgd /*
1571 1.1 cgd * Remove a symbol forever from the symbol table. s_blklev
1572 1.1 cgd * is set to -1 to avoid that the symbol will later be put
1573 1.1 cgd * back to the symbol table.
1574 1.1 cgd */
1575 1.1 cgd void
1576 1.19 lukem rmsym(sym_t *sym)
1577 1.1 cgd {
1578 1.19 lukem
1579 1.1 cgd if ((*sym->s_rlink = sym->s_link) != NULL)
1580 1.1 cgd sym->s_link->s_rlink = sym->s_rlink;
1581 1.1 cgd sym->s_blklev = -1;
1582 1.1 cgd sym->s_link = NULL;
1583 1.1 cgd }
1584 1.1 cgd
1585 1.1 cgd /*
1586 1.1 cgd * Remove a list of symbols declared at one level from the symbol
1587 1.1 cgd * table.
1588 1.1 cgd */
1589 1.1 cgd void
1590 1.19 lukem rmsyms(sym_t *syms)
1591 1.1 cgd {
1592 1.1 cgd sym_t *sym;
1593 1.1 cgd
1594 1.1 cgd for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1595 1.1 cgd if (sym->s_blklev != -1) {
1596 1.1 cgd if ((*sym->s_rlink = sym->s_link) != NULL)
1597 1.1 cgd sym->s_link->s_rlink = sym->s_rlink;
1598 1.1 cgd sym->s_link = NULL;
1599 1.1 cgd sym->s_rlink = NULL;
1600 1.1 cgd }
1601 1.1 cgd }
1602 1.1 cgd }
1603 1.1 cgd
1604 1.1 cgd /*
1605 1.1 cgd * Put a symbol into the symbol table
1606 1.1 cgd */
1607 1.1 cgd void
1608 1.19 lukem inssym(int bl, sym_t *sym)
1609 1.1 cgd {
1610 1.1 cgd int h;
1611 1.1 cgd
1612 1.1 cgd h = hash(sym->s_name);
1613 1.1 cgd if ((sym->s_link = symtab[h]) != NULL)
1614 1.1 cgd symtab[h]->s_rlink = &sym->s_link;
1615 1.53 christos sym->s_rlink = &symtab[h];
1616 1.53 christos symtab[h] = sym;
1617 1.1 cgd sym->s_blklev = bl;
1618 1.107 rillig lint_assert(sym->s_link == NULL ||
1619 1.107 rillig sym->s_blklev >= sym->s_link->s_blklev);
1620 1.1 cgd }
1621 1.1 cgd
1622 1.1 cgd /*
1623 1.1 cgd * Called at level 0 after syntax errors
1624 1.1 cgd * Removes all symbols which are not declared at level 0 from the
1625 1.1 cgd * symbol table. Also frees all memory which is not associated with
1626 1.1 cgd * level 0.
1627 1.1 cgd */
1628 1.1 cgd void
1629 1.19 lukem cleanup(void)
1630 1.1 cgd {
1631 1.1 cgd sym_t *sym, *nsym;
1632 1.1 cgd int i;
1633 1.1 cgd
1634 1.1 cgd for (i = 0; i < HSHSIZ1; i++) {
1635 1.1 cgd for (sym = symtab[i]; sym != NULL; sym = nsym) {
1636 1.1 cgd nsym = sym->s_link;
1637 1.1 cgd if (sym->s_blklev >= 1) {
1638 1.1 cgd if ((*sym->s_rlink = nsym) != NULL)
1639 1.1 cgd nsym->s_rlink = sym->s_rlink;
1640 1.1 cgd }
1641 1.1 cgd }
1642 1.1 cgd }
1643 1.1 cgd
1644 1.1 cgd for (i = mblklev; i > 0; i--)
1645 1.1 cgd freelblk(i);
1646 1.1 cgd }
1647 1.1 cgd
1648 1.1 cgd /*
1649 1.1 cgd * Create a new symbol with the name of an existing symbol.
1650 1.1 cgd */
1651 1.1 cgd sym_t *
1652 1.19 lukem pushdown(sym_t *sym)
1653 1.1 cgd {
1654 1.1 cgd int h;
1655 1.1 cgd sym_t *nsym;
1656 1.1 cgd
1657 1.1 cgd h = hash(sym->s_name);
1658 1.1 cgd nsym = getblk(sizeof (sym_t));
1659 1.107 rillig lint_assert(sym->s_blklev <= blklev);
1660 1.1 cgd nsym->s_name = sym->s_name;
1661 1.104 rillig UNIQUE_CURR_POS(nsym->s_def_pos);
1662 1.1 cgd nsym->s_kind = sym->s_kind;
1663 1.1 cgd nsym->s_blklev = blklev;
1664 1.1 cgd
1665 1.1 cgd if ((nsym->s_link = symtab[h]) != NULL)
1666 1.1 cgd symtab[h]->s_rlink = &nsym->s_link;
1667 1.53 christos nsym->s_rlink = &symtab[h];
1668 1.53 christos symtab[h] = nsym;
1669 1.1 cgd
1670 1.3 jpo *dcs->d_ldlsym = nsym;
1671 1.3 jpo dcs->d_ldlsym = &nsym->s_dlnxt;
1672 1.1 cgd
1673 1.97 rillig return nsym;
1674 1.6 jpo }
1675 1.6 jpo
1676 1.6 jpo /*
1677 1.6 jpo * Free any dynamically allocated memory referenced by
1678 1.6 jpo * the value stack or yylval.
1679 1.6 jpo * The type of information in yylval is described by tok.
1680 1.6 jpo */
1681 1.6 jpo void
1682 1.19 lukem freeyyv(void *sp, int tok)
1683 1.6 jpo {
1684 1.6 jpo if (tok == T_NAME || tok == T_TYPENAME) {
1685 1.6 jpo sbuf_t *sb = *(sbuf_t **)sp;
1686 1.6 jpo freesb(sb);
1687 1.6 jpo } else if (tok == T_CON) {
1688 1.6 jpo val_t *val = *(val_t **)sp;
1689 1.6 jpo free(val);
1690 1.6 jpo } else if (tok == T_STRING) {
1691 1.6 jpo strg_t *strg = *(strg_t **)sp;
1692 1.6 jpo if (strg->st_tspec == CHAR) {
1693 1.6 jpo free(strg->st_cp);
1694 1.107 rillig } else {
1695 1.107 rillig lint_assert(strg->st_tspec == WCHAR);
1696 1.6 jpo free(strg->st_wcp);
1697 1.6 jpo }
1698 1.6 jpo free(strg);
1699 1.19 lukem }
1700 1.1 cgd }
1701