scan.l revision 1.67 1 1.1 cgd %{
2 1.67 christos /* $NetBSD: scan.l,v 1.67 2016/11/05 01:09:30 christos 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 1.67 christos __RCSID("$NetBSD: scan.l,v 1.67 2016/11/05 01:09:30 christos Exp $");
39 1.1 cgd #endif
40 1.1 cgd
41 1.1 cgd #include <stdlib.h>
42 1.1 cgd #include <string.h>
43 1.1 cgd #include <limits.h>
44 1.1 cgd #include <float.h>
45 1.1 cgd #include <ctype.h>
46 1.1 cgd #include <errno.h>
47 1.1 cgd #include <math.h>
48 1.1 cgd
49 1.1 cgd #include "lint1.h"
50 1.12 tv #include "cgram.h"
51 1.1 cgd
52 1.61 joerg #define CHAR_MASK ((int)(~(~0U << CHAR_BIT)))
53 1.1 cgd
54 1.1 cgd /* Current position (its 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 1.55 christos int attron;
65 1.55 christos
66 1.19 lukem static void incline(void);
67 1.19 lukem static void badchar(int);
68 1.19 lukem static sbuf_t *allocsb(void);
69 1.19 lukem static void freesb(sbuf_t *);
70 1.19 lukem static int inpc(void);
71 1.19 lukem static int hash(const char *);
72 1.19 lukem static sym_t *search(sbuf_t *);
73 1.19 lukem static int name(void);
74 1.19 lukem static int keyw(sym_t *);
75 1.19 lukem static int icon(int);
76 1.19 lukem static int fcon(void);
77 1.19 lukem static int operator(int, op_t);
78 1.19 lukem static int ccon(void);
79 1.19 lukem static int wccon(void);
80 1.19 lukem static int getescc(int);
81 1.19 lukem static void directive(void);
82 1.19 lukem static void comment(void);
83 1.19 lukem static void slashslashcomment(void);
84 1.19 lukem static int string(void);
85 1.19 lukem static int wcstrg(void);
86 1.1 cgd
87 1.1 cgd %}
88 1.1 cgd
89 1.62 christos
90 1.1 cgd L [_A-Za-z]
91 1.1 cgd D [0-9]
92 1.1 cgd NZD [1-9]
93 1.62 christos BD [0-1]
94 1.1 cgd OD [0-7]
95 1.1 cgd HD [0-9A-Fa-f]
96 1.1 cgd EX ([eE][+-]?[0-9]+)
97 1.47 christos HX (p[+-]?[0-9A-Fa-f]+)
98 1.47 christos TL ([fFlL]?[i]?)
99 1.1 cgd
100 1.46 christos %option nounput
101 1.46 christos
102 1.1 cgd %%
103 1.1 cgd
104 1.1 cgd {L}({L}|{D})* return (name());
105 1.62 christos 0[bB]{BD}+[lLuU]* return (icon(2));
106 1.1 cgd 0{OD}*[lLuU]* return (icon(8));
107 1.1 cgd {NZD}{D}*[lLuU]* return (icon(10));
108 1.1 cgd 0[xX]{HD}+[lLuU]* return (icon(16));
109 1.47 christos {D}+\.{D}*{EX}?{TL} |
110 1.47 christos {D}+{EX}{TL} |
111 1.47 christos 0[xX]{HD}+\.{HD}*{HX}{TL} |
112 1.47 christos 0[xX]{HD}+{HX}{TL} |
113 1.47 christos \.{D}+{EX}?{TL} return (fcon());
114 1.1 cgd "=" return (operator(T_ASSIGN, ASSIGN));
115 1.1 cgd "*=" return (operator(T_OPASS, MULASS));
116 1.1 cgd "/=" return (operator(T_OPASS, DIVASS));
117 1.1 cgd "%=" return (operator(T_OPASS, MODASS));
118 1.1 cgd "+=" return (operator(T_OPASS, ADDASS));
119 1.1 cgd "-=" return (operator(T_OPASS, SUBASS));
120 1.1 cgd "<<=" return (operator(T_OPASS, SHLASS));
121 1.1 cgd ">>=" return (operator(T_OPASS, SHRASS));
122 1.1 cgd "&=" return (operator(T_OPASS, ANDASS));
123 1.1 cgd "^=" return (operator(T_OPASS, XORASS));
124 1.1 cgd "|=" return (operator(T_OPASS, ORASS));
125 1.1 cgd "||" return (operator(T_LOGOR, LOGOR));
126 1.1 cgd "&&" return (operator(T_LOGAND, LOGAND));
127 1.1 cgd "|" return (operator(T_OR, OR));
128 1.1 cgd "&" return (operator(T_AND, AND));
129 1.1 cgd "^" return (operator(T_XOR, XOR));
130 1.1 cgd "==" return (operator(T_EQOP, EQ));
131 1.1 cgd "!=" return (operator(T_EQOP, NE));
132 1.1 cgd "<" return (operator(T_RELOP, LT));
133 1.1 cgd ">" return (operator(T_RELOP, GT));
134 1.1 cgd "<=" return (operator(T_RELOP, LE));
135 1.1 cgd ">=" return (operator(T_RELOP, GE));
136 1.1 cgd "<<" return (operator(T_SHFTOP, SHL));
137 1.1 cgd ">>" return (operator(T_SHFTOP, SHR));
138 1.1 cgd "++" return (operator(T_INCDEC, INC));
139 1.1 cgd "--" return (operator(T_INCDEC, DEC));
140 1.1 cgd "->" return (operator(T_STROP, ARROW));
141 1.1 cgd "." return (operator(T_STROP, POINT));
142 1.1 cgd "+" return (operator(T_ADDOP, PLUS));
143 1.1 cgd "-" return (operator(T_ADDOP, MINUS));
144 1.1 cgd "*" return (operator(T_MULT, MULT));
145 1.1 cgd "/" return (operator(T_DIVOP, DIV));
146 1.1 cgd "%" return (operator(T_DIVOP, MOD));
147 1.1 cgd "!" return (operator(T_UNOP, NOT));
148 1.1 cgd "~" return (operator(T_UNOP, COMPL));
149 1.1 cgd "\"" return (string());
150 1.1 cgd "L\"" return (wcstrg());
151 1.1 cgd ";" return (T_SEMI);
152 1.1 cgd "{" return (T_LBRACE);
153 1.1 cgd "}" return (T_RBRACE);
154 1.1 cgd "," return (T_COMMA);
155 1.1 cgd ":" return (T_COLON);
156 1.1 cgd "?" return (T_QUEST);
157 1.1 cgd "[" return (T_LBRACK);
158 1.1 cgd "]" return (T_RBRACK);
159 1.1 cgd "(" return (T_LPARN);
160 1.1 cgd ")" return (T_RPARN);
161 1.1 cgd "..." return (T_ELLIPSE);
162 1.1 cgd "'" return (ccon());
163 1.1 cgd "L'" return (wccon());
164 1.1 cgd ^#.*$ directive();
165 1.1 cgd \n incline();
166 1.1 cgd \t|" "|\f|\v ;
167 1.1 cgd "/*" comment();
168 1.18 lukem "//" slashslashcomment();
169 1.1 cgd . badchar(yytext[0]);
170 1.1 cgd
171 1.1 cgd %%
172 1.1 cgd
173 1.1 cgd static void
174 1.19 lukem incline(void)
175 1.1 cgd {
176 1.1 cgd curr_pos.p_line++;
177 1.9 cgd curr_pos.p_uniq = 0;
178 1.9 cgd if (curr_pos.p_file == csrc_pos.p_file) {
179 1.1 cgd csrc_pos.p_line++;
180 1.9 cgd csrc_pos.p_uniq = 0;
181 1.9 cgd }
182 1.1 cgd }
183 1.1 cgd
184 1.1 cgd static void
185 1.19 lukem badchar(int c)
186 1.1 cgd {
187 1.19 lukem
188 1.1 cgd /* unknown character \%o */
189 1.1 cgd error(250, c);
190 1.1 cgd }
191 1.1 cgd
192 1.1 cgd /*
193 1.1 cgd * Keywords.
194 1.1 cgd * During initialisation they are written to the symbol table.
195 1.1 cgd */
196 1.1 cgd static struct kwtab {
197 1.1 cgd const char *kw_name; /* keyword */
198 1.1 cgd int kw_token; /* token returned by yylex() */
199 1.1 cgd scl_t kw_scl; /* storage class if kw_token T_SCLASS */
200 1.1 cgd tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */
201 1.1 cgd tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */
202 1.55 christos u_int kw_c89 : 1; /* c89 keyword */
203 1.55 christos u_int kw_c99 : 1; /* c99 keyword */
204 1.55 christos u_int kw_gcc : 1; /* GCC keyword */
205 1.55 christos u_int kw_attr : 1; /* GCC attribute, keyword */
206 1.1 cgd } kwtab[] = {
207 1.55 christos { "__alignof__", T_ALIGNOF, 0, 0, 0, 0, 0, 0, 0 },
208 1.55 christos { "__attribute__",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 },
209 1.59 christos { "__attribute",T_ATTRIBUTE, 0, 0, 0, 0, 0, 1, 0 },
210 1.55 christos { "__packed__", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 },
211 1.55 christos { "packed", T_AT_PACKED, 0, 0, 0, 0, 0, 1, 1 },
212 1.55 christos { "__aligned__",T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 },
213 1.55 christos { "aligned", T_AT_ALIGNED, 0, 0, 0, 0, 0, 1, 1 },
214 1.55 christos { "__transparent_union__",T_AT_TUNION,0,0, 0, 0, 0, 1, 1 },
215 1.55 christos { "transparent_union",T_AT_TUNION,0, 0, 0, 0, 0, 1, 1 },
216 1.55 christos { "__unused__", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 },
217 1.55 christos { "unused", T_AT_UNUSED, 0, 0, 0, 0, 0, 1, 1 },
218 1.66 christos { "__gnu_inline__",T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 },
219 1.66 christos { "gnu_inline", T_AT_GNU_INLINE,0, 0, 0, 0, 0, 1, 1 },
220 1.55 christos { "__deprecated__",T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 },
221 1.55 christos { "deprecated", T_AT_DEPRECATED,0, 0, 0, 0, 0, 1, 1 },
222 1.55 christos { "__may_alias__",T_AT_MAY_ALIAS,0, 0, 0, 0, 0, 1, 1 },
223 1.55 christos { "may_alias", T_AT_MAY_ALIAS, 0, 0, 0, 0, 0, 1, 1 },
224 1.55 christos { "format", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 },
225 1.55 christos { "__format__", T_AT_FORMAT, 0, 0, 0, 0, 0, 1, 1 },
226 1.55 christos { "printf", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 },
227 1.55 christos { "__printf__", T_AT_FORMAT_PRINTF,0, 0, 0, 0, 0, 1, 1 },
228 1.55 christos { "scanf", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 },
229 1.55 christos { "__scanf__", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 },
230 1.55 christos { "strftime", T_AT_FORMAT_SCANF,0, 0, 0, 0, 0, 1, 1 },
231 1.55 christos { "__strftime__",T_AT_FORMAT_STRFTIME,0,0, 0, 0, 0, 1, 1 },
232 1.55 christos { "pure", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 },
233 1.55 christos { "__pure__", T_AT_PURE, 0, 0, 0, 0, 0, 1, 1 },
234 1.55 christos { "noreturn", T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 },
235 1.55 christos { "__noreturn__",T_AT_NORETURN, 0, 0, 0, 0, 0, 1, 1 },
236 1.59 christos { "sentinel", T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 },
237 1.59 christos { "__sentinel__",T_AT_SENTINEL, 0, 0, 0, 0, 0, 1, 1 },
238 1.59 christos { "format_arg", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 },
239 1.59 christos { "__format_arg__", T_AT_FORMAT_ARG,0, 0, 0, 0, 0, 1, 1 },
240 1.59 christos { "returns_twice", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 },
241 1.59 christos { "__returns_twice__", T_AT_RETURNS_TWICE,0,0, 0, 0, 0, 1, 1 },
242 1.59 christos { "cold", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 },
243 1.59 christos { "__cold__", T_AT_COLD, 0, 0, 0, 0, 0, 1, 1 },
244 1.58 christos { "asm", T_ASM, 0, 0, 0, 0, 0, 1, 0 },
245 1.55 christos { "__asm", T_ASM, 0, 0, 0, 0, 0, 0, 0 },
246 1.55 christos { "__asm__", T_ASM, 0, 0, 0, 0, 0, 0, 0 },
247 1.55 christos { "auto", T_SCLASS, AUTO, 0, 0, 0, 0, 0, 0 },
248 1.55 christos { "break", T_BREAK, 0, 0, 0, 0, 0, 0, 0 },
249 1.55 christos { "_Bool", T_TYPE, 0, BOOL, 0, 0, 1, 0, 0 },
250 1.55 christos { "case", T_CASE, 0, 0, 0, 0, 0, 0, 0 },
251 1.55 christos { "char", T_TYPE, 0, CHAR, 0, 0, 0, 0, 0 },
252 1.55 christos { "const", T_QUAL, 0, 0, CONST, 1, 0, 0, 0 },
253 1.55 christos { "_Complex", T_TYPE, 0, COMPLEX,0, 0, 1, 0, 0 },
254 1.55 christos { "__const__", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 },
255 1.55 christos { "__const", T_QUAL, 0, 0, CONST, 0, 0, 0, 0 },
256 1.55 christos { "continue", T_CONTINUE, 0, 0, 0, 0, 0, 0, 0 },
257 1.55 christos { "default", T_DEFAULT, 0, 0, 0, 0, 0, 0, 0 },
258 1.55 christos { "do", T_DO, 0, 0, 0, 0, 0, 0, 0 },
259 1.55 christos { "double", T_TYPE, 0, DOUBLE, 0, 0, 0, 0, 0 },
260 1.55 christos { "else", T_ELSE, 0, 0, 0, 0, 0, 0, 0 },
261 1.55 christos { "enum", T_ENUM, 0, 0, 0, 0, 0, 0, 0 },
262 1.55 christos { "__extension__", T_EXTENSION, 0, 0, 0, 0, 0, 1, 0 },
263 1.55 christos { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0, 0, 0 },
264 1.55 christos { "float", T_TYPE, 0, FLOAT, 0, 0, 0, 0, 0 },
265 1.55 christos { "for", T_FOR, 0, 0, 0, 0, 0, 0, 0 },
266 1.67 christos { "_Generic", T_GENERIC, 0, 0, 0, 0, 1, 0, 0 },
267 1.55 christos { "goto", T_GOTO, 0, 0, 0, 0, 0, 0, 0 },
268 1.55 christos { "if", T_IF, 0, 0, 0, 0, 0, 0, 0 },
269 1.55 christos { "__imag__", T_IMAG, 0, 0, 0, 0, 1, 0, 0 },
270 1.55 christos { "inline", T_SCLASS, INLINE, 0, 0, 0, 1, 0, 0 },
271 1.55 christos { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 },
272 1.55 christos { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0, 0, 0 },
273 1.55 christos { "int", T_TYPE, 0, INT, 0, 0, 0, 0, 0 },
274 1.55 christos { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0, 0, 0 },
275 1.55 christos { "long", T_TYPE, 0, LONG, 0, 0, 0, 0, 0 },
276 1.55 christos { "__real__", T_REAL, 0, 0, 0, 0, 1, 0, 0 },
277 1.55 christos { "register", T_SCLASS, REG, 0, 0, 0, 0, 0, 0 },
278 1.55 christos { "__restrict__",T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 },
279 1.55 christos { "restrict", T_QUAL, 0, 0, RESTRICT, 0, 1, 0, 0 },
280 1.55 christos { "return", T_RETURN, 0, 0, 0, 0, 0, 0, 0 },
281 1.55 christos { "__packed", T_PACKED, 0, 0, 0, 0, 0, 0, 0 },
282 1.55 christos { "short", T_TYPE, 0, SHORT, 0, 0, 0, 0, 0 },
283 1.55 christos { "signed", T_TYPE, 0, SIGNED, 0, 1, 0, 0, 0 },
284 1.55 christos { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 },
285 1.55 christos { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0, 0, 0 },
286 1.55 christos { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0, 0, 0 },
287 1.55 christos { "static", T_SCLASS, STATIC, 0, 0, 0, 0, 0, 0 },
288 1.55 christos { "struct", T_SOU, 0, STRUCT, 0, 0, 0, 0, 0 },
289 1.55 christos { "switch", T_SWITCH, 0, 0, 0, 0, 0, 0, 0 },
290 1.55 christos { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0, 0, 0 },
291 1.55 christos { "typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 },
292 1.55 christos { "__typeof", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 },
293 1.55 christos { "__typeof__", T_TYPEOF, 0, 0, 0, 0, 0, 1, 0 },
294 1.55 christos { "union", T_SOU, 0, UNION, 0, 0, 0, 0, 0 },
295 1.55 christos { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0, 0, 0 },
296 1.55 christos { "void", T_TYPE, 0, VOID, 0, 0, 0, 0, 0 },
297 1.55 christos { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0, 0, 0 },
298 1.55 christos { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 },
299 1.55 christos { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0, 0, 0 },
300 1.65 christos { "visibility", T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 },
301 1.65 christos { "__visibility",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 },
302 1.65 christos { "__visibility__",T_AT_VISIBILITY,0, 0, 0, 0, 0, 1, 1 },
303 1.65 christos { "weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 },
304 1.65 christos { "__weak", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 },
305 1.65 christos { "__weak__", T_AT_WEAK, 0, 0, 0, 0, 0, 1, 1 },
306 1.55 christos { "while", T_WHILE, 0, 0, 0, 0, 0, 0, 0 },
307 1.55 christos { NULL, 0, 0, 0, 0, 0, 0, 0, 0 }
308 1.1 cgd };
309 1.1 cgd
310 1.1 cgd /* Symbol table */
311 1.1 cgd static sym_t *symtab[HSHSIZ1];
312 1.1 cgd
313 1.1 cgd /* bit i of the entry with index i is set */
314 1.27 thorpej uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
315 1.1 cgd
316 1.1 cgd /* least significant i bits are set in the entry with index i */
317 1.27 thorpej uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
318 1.1 cgd
319 1.1 cgd /* least significant i bits are not set in the entry with index i */
320 1.27 thorpej uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
321 1.1 cgd
322 1.1 cgd /* free list for sbuf structures */
323 1.1 cgd static sbuf_t *sbfrlst;
324 1.1 cgd
325 1.1 cgd /* Typ of next expected symbol */
326 1.1 cgd symt_t symtyp;
327 1.1 cgd
328 1.1 cgd
329 1.1 cgd /*
330 1.1 cgd * All keywords are written to the symbol table. This saves us looking
331 1.1 cgd * in a extra table for each name we found.
332 1.1 cgd */
333 1.1 cgd void
334 1.19 lukem initscan(void)
335 1.1 cgd {
336 1.1 cgd struct kwtab *kw;
337 1.1 cgd sym_t *sym;
338 1.43 christos size_t h, i;
339 1.27 thorpej uint64_t uq;
340 1.1 cgd
341 1.1 cgd for (kw = kwtab; kw->kw_name != NULL; kw++) {
342 1.32 perry if ((kw->kw_c89 || kw->kw_c99) && tflag)
343 1.32 perry continue;
344 1.32 perry if (kw->kw_c99 && !(Sflag || gflag))
345 1.5 jpo continue;
346 1.5 jpo if (kw->kw_gcc && !gflag)
347 1.5 jpo continue;
348 1.1 cgd sym = getblk(sizeof (sym_t));
349 1.1 cgd sym->s_name = kw->kw_name;
350 1.55 christos sym->s_keyw = kw;
351 1.1 cgd sym->s_value.v_quad = kw->kw_token;
352 1.1 cgd if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
353 1.1 cgd sym->s_tspec = kw->kw_tspec;
354 1.1 cgd } else if (kw->kw_token == T_SCLASS) {
355 1.1 cgd sym->s_scl = kw->kw_scl;
356 1.1 cgd } else if (kw->kw_token == T_QUAL) {
357 1.1 cgd sym->s_tqual = kw->kw_tqual;
358 1.1 cgd }
359 1.1 cgd h = hash(sym->s_name);
360 1.1 cgd if ((sym->s_link = symtab[h]) != NULL)
361 1.1 cgd symtab[h]->s_rlink = &sym->s_link;
362 1.53 christos sym->s_rlink = &symtab[h];
363 1.53 christos symtab[h] = sym;
364 1.1 cgd }
365 1.1 cgd
366 1.1 cgd /* initialize bit-masks for quads */
367 1.27 thorpej for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
368 1.27 thorpej qbmasks[i] = (uint64_t)1 << i;
369 1.27 thorpej uq = ~(uint64_t)0 << i;
370 1.1 cgd qumasks[i] = uq;
371 1.1 cgd qlmasks[i] = ~uq;
372 1.1 cgd }
373 1.1 cgd qumasks[i] = 0;
374 1.27 thorpej qlmasks[i] = ~(uint64_t)0;
375 1.1 cgd }
376 1.1 cgd
377 1.1 cgd /*
378 1.1 cgd * Get a free sbuf structure, if possible from the free list
379 1.1 cgd */
380 1.1 cgd static sbuf_t *
381 1.19 lukem allocsb(void)
382 1.1 cgd {
383 1.1 cgd sbuf_t *sb;
384 1.1 cgd
385 1.1 cgd if ((sb = sbfrlst) != NULL) {
386 1.1 cgd sbfrlst = sb->sb_nxt;
387 1.55 christos #ifdef BLKDEBUG
388 1.55 christos (void)memset(sb, 0, sizeof (*sb));
389 1.55 christos #else
390 1.55 christos sb->sb_nxt = NULL;
391 1.55 christos #endif
392 1.1 cgd } else {
393 1.1 cgd sb = xmalloc(sizeof (sbuf_t));
394 1.55 christos (void)memset(sb, 0, sizeof (*sb));
395 1.1 cgd }
396 1.1 cgd return (sb);
397 1.1 cgd }
398 1.1 cgd
399 1.1 cgd /*
400 1.1 cgd * Put a sbuf structure to the free list
401 1.1 cgd */
402 1.1 cgd static void
403 1.19 lukem freesb(sbuf_t *sb)
404 1.1 cgd {
405 1.19 lukem
406 1.55 christos (void)memset(sb, ZERO, sizeof (*sb));
407 1.1 cgd sb->sb_nxt = sbfrlst;
408 1.1 cgd sbfrlst = sb;
409 1.1 cgd }
410 1.1 cgd
411 1.1 cgd /*
412 1.1 cgd * Read a character and ensure that it is positive (except EOF).
413 1.1 cgd * Increment line count(s) if necessary.
414 1.1 cgd */
415 1.1 cgd static int
416 1.19 lukem inpc(void)
417 1.1 cgd {
418 1.1 cgd int c;
419 1.1 cgd
420 1.1 cgd if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
421 1.1 cgd incline();
422 1.1 cgd return (c);
423 1.1 cgd }
424 1.1 cgd
425 1.1 cgd static int
426 1.19 lukem hash(const char *s)
427 1.1 cgd {
428 1.1 cgd u_int v;
429 1.1 cgd const u_char *us;
430 1.1 cgd
431 1.1 cgd v = 0;
432 1.1 cgd for (us = (const u_char *)s; *us != '\0'; us++) {
433 1.1 cgd v = (v << sizeof (v)) + *us;
434 1.1 cgd v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
435 1.1 cgd }
436 1.1 cgd return (v % HSHSIZ1);
437 1.1 cgd }
438 1.1 cgd
439 1.1 cgd /*
440 1.1 cgd * Lex has found a letter followed by zero or more letters or digits.
441 1.1 cgd * It looks for a symbol in the symbol table with the same name. This
442 1.1 cgd * symbol must either be a keyword or a symbol of the type required by
443 1.1 cgd * symtyp (label, member, tag, ...).
444 1.1 cgd *
445 1.1 cgd * If it is a keyword, the token is returned. In some cases it is described
446 1.1 cgd * more deeply by data written to yylval.
447 1.1 cgd *
448 1.1 cgd * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
449 1.60 snj * is stored in yylval. This struct contains the name of the symbol, its
450 1.1 cgd * length and hash value. If there is already a symbol of the same name
451 1.1 cgd * and type in the symbol table, the sbuf struct also contains a pointer
452 1.1 cgd * to the symbol table entry.
453 1.1 cgd */
454 1.1 cgd static int
455 1.19 lukem name(void)
456 1.1 cgd {
457 1.1 cgd char *s;
458 1.1 cgd sbuf_t *sb;
459 1.1 cgd sym_t *sym;
460 1.1 cgd int tok;
461 1.1 cgd
462 1.1 cgd sb = allocsb();
463 1.1 cgd sb->sb_name = yytext;
464 1.1 cgd sb->sb_len = yyleng;
465 1.1 cgd sb->sb_hash = hash(yytext);
466 1.1 cgd if ((sym = search(sb)) != NULL && sym->s_keyw) {
467 1.56 christos freesb(sb);
468 1.56 christos return (keyw(sym));
469 1.1 cgd }
470 1.1 cgd
471 1.1 cgd sb->sb_sym = sym;
472 1.1 cgd
473 1.1 cgd if (sym != NULL) {
474 1.1 cgd if (blklev < sym->s_blklev)
475 1.28 christos LERROR("name()");
476 1.1 cgd sb->sb_name = sym->s_name;
477 1.1 cgd sb->sb_len = strlen(sym->s_name);
478 1.1 cgd tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
479 1.1 cgd } else {
480 1.1 cgd s = getblk(yyleng + 1);
481 1.1 cgd (void)memcpy(s, yytext, yyleng + 1);
482 1.1 cgd sb->sb_name = s;
483 1.1 cgd sb->sb_len = yyleng;
484 1.1 cgd tok = T_NAME;
485 1.1 cgd }
486 1.1 cgd
487 1.1 cgd yylval.y_sb = sb;
488 1.1 cgd return (tok);
489 1.1 cgd }
490 1.1 cgd
491 1.1 cgd static sym_t *
492 1.19 lukem search(sbuf_t *sb)
493 1.1 cgd {
494 1.1 cgd sym_t *sym;
495 1.1 cgd
496 1.1 cgd for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
497 1.1 cgd if (strcmp(sym->s_name, sb->sb_name) == 0) {
498 1.56 christos if (sym->s_keyw) {
499 1.56 christos struct kwtab *kw = sym->s_keyw;
500 1.56 christos if (!kw->kw_attr || attron)
501 1.56 christos return (sym);
502 1.56 christos } else if (sym->s_kind == symtyp)
503 1.56 christos return sym;
504 1.1 cgd }
505 1.1 cgd }
506 1.1 cgd
507 1.1 cgd return (NULL);
508 1.1 cgd }
509 1.19 lukem
510 1.1 cgd static int
511 1.19 lukem keyw(sym_t *sym)
512 1.1 cgd {
513 1.1 cgd int t;
514 1.1 cgd
515 1.1 cgd if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
516 1.1 cgd yylval.y_scl = sym->s_scl;
517 1.1 cgd } else if (t == T_TYPE || t == T_SOU) {
518 1.1 cgd yylval.y_tspec = sym->s_tspec;
519 1.1 cgd } else if (t == T_QUAL) {
520 1.1 cgd yylval.y_tqual = sym->s_tqual;
521 1.1 cgd }
522 1.1 cgd return (t);
523 1.1 cgd }
524 1.1 cgd
525 1.1 cgd /*
526 1.1 cgd * Convert a string representing an integer into internal representation.
527 1.1 cgd * The value is returned in yylval. icon() (and yylex()) returns T_CON.
528 1.1 cgd */
529 1.1 cgd static int
530 1.19 lukem icon(int base)
531 1.1 cgd {
532 1.1 cgd int l_suffix, u_suffix;
533 1.1 cgd int len;
534 1.64 christos const char *cp;
535 1.1 cgd char c, *eptr;
536 1.1 cgd tspec_t typ;
537 1.27 thorpej uint64_t uq = 0;
538 1.1 cgd int ansiu;
539 1.1 cgd static tspec_t contypes[2][3] = {
540 1.1 cgd { INT, LONG, QUAD },
541 1.1 cgd { UINT, ULONG, UQUAD }
542 1.1 cgd };
543 1.1 cgd
544 1.1 cgd cp = yytext;
545 1.1 cgd len = yyleng;
546 1.1 cgd
547 1.62 christos /* skip 0[xX] or 0[bB] */
548 1.62 christos if (base == 16 || base == 2) {
549 1.1 cgd cp += 2;
550 1.1 cgd len -= 2;
551 1.1 cgd }
552 1.1 cgd
553 1.1 cgd /* read suffixes */
554 1.1 cgd l_suffix = u_suffix = 0;
555 1.1 cgd for ( ; ; ) {
556 1.1 cgd if ((c = cp[len - 1]) == 'l' || c == 'L') {
557 1.1 cgd l_suffix++;
558 1.1 cgd } else if (c == 'u' || c == 'U') {
559 1.1 cgd u_suffix++;
560 1.1 cgd } else {
561 1.1 cgd break;
562 1.1 cgd }
563 1.1 cgd len--;
564 1.1 cgd }
565 1.1 cgd if (l_suffix > 2 || u_suffix > 1) {
566 1.1 cgd /* malformed integer constant */
567 1.1 cgd warning(251);
568 1.1 cgd if (l_suffix > 2)
569 1.1 cgd l_suffix = 2;
570 1.1 cgd if (u_suffix > 1)
571 1.1 cgd u_suffix = 1;
572 1.1 cgd }
573 1.1 cgd if (tflag && u_suffix != 0) {
574 1.1 cgd /* suffix U is illegal in traditional C */
575 1.1 cgd warning(97);
576 1.1 cgd }
577 1.1 cgd typ = contypes[u_suffix][l_suffix];
578 1.1 cgd
579 1.1 cgd errno = 0;
580 1.35 he
581 1.64 christos uq = strtouq(cp, &eptr, base);
582 1.1 cgd if (eptr != cp + len)
583 1.28 christos LERROR("icon()");
584 1.1 cgd if (errno != 0)
585 1.1 cgd /* integer constant out of range */
586 1.1 cgd warning(252);
587 1.1 cgd
588 1.1 cgd /*
589 1.35 he * If the value is too big for the current type, we must choose
590 1.1 cgd * another type.
591 1.1 cgd */
592 1.1 cgd ansiu = 0;
593 1.1 cgd switch (typ) {
594 1.1 cgd case INT:
595 1.35 he if (uq <= TARG_INT_MAX) {
596 1.1 cgd /* ok */
597 1.35 he } else if (uq <= TARG_UINT_MAX && base != 10) {
598 1.1 cgd typ = UINT;
599 1.35 he } else if (uq <= TARG_LONG_MAX) {
600 1.1 cgd typ = LONG;
601 1.1 cgd } else {
602 1.1 cgd typ = ULONG;
603 1.35 he if (uq > TARG_ULONG_MAX) {
604 1.35 he /* integer constant out of range */
605 1.35 he warning(252);
606 1.35 he }
607 1.1 cgd }
608 1.1 cgd if (typ == UINT || typ == ULONG) {
609 1.1 cgd if (tflag) {
610 1.1 cgd typ = LONG;
611 1.1 cgd } else if (!sflag) {
612 1.1 cgd /*
613 1.1 cgd * Remember that the constant is unsigned
614 1.1 cgd * only in ANSI C
615 1.1 cgd */
616 1.1 cgd ansiu = 1;
617 1.1 cgd }
618 1.1 cgd }
619 1.1 cgd break;
620 1.1 cgd case UINT:
621 1.35 he if (uq > TARG_UINT_MAX) {
622 1.1 cgd typ = ULONG;
623 1.35 he if (uq > TARG_ULONG_MAX) {
624 1.35 he /* integer constant out of range */
625 1.35 he warning(252);
626 1.35 he }
627 1.35 he }
628 1.1 cgd break;
629 1.1 cgd case LONG:
630 1.35 he if (uq > TARG_LONG_MAX && !tflag) {
631 1.1 cgd typ = ULONG;
632 1.1 cgd if (!sflag)
633 1.1 cgd ansiu = 1;
634 1.35 he if (uq > TARG_ULONG_MAX) {
635 1.35 he /* integer constant out of range */
636 1.35 he warning(252);
637 1.35 he }
638 1.1 cgd }
639 1.1 cgd break;
640 1.39 joerg case ULONG:
641 1.39 joerg if (uq > TARG_ULONG_MAX) {
642 1.39 joerg /* integer constant out of range */
643 1.39 joerg warning(252);
644 1.39 joerg }
645 1.39 joerg break;
646 1.1 cgd case QUAD:
647 1.35 he if (uq > TARG_QUAD_MAX && !tflag) {
648 1.1 cgd typ = UQUAD;
649 1.1 cgd if (!sflag)
650 1.1 cgd ansiu = 1;
651 1.1 cgd }
652 1.1 cgd break;
653 1.39 joerg case UQUAD:
654 1.39 joerg if (uq > TARG_UQUAD_MAX) {
655 1.39 joerg /* integer constant out of range */
656 1.39 joerg warning(252);
657 1.39 joerg }
658 1.39 joerg break;
659 1.52 christos /* LINTED206: (enumeration values not handled in switch) */
660 1.11 christos case STRUCT:
661 1.11 christos case VOID:
662 1.11 christos case LDOUBLE:
663 1.11 christos case FUNC:
664 1.11 christos case ARRAY:
665 1.11 christos case PTR:
666 1.11 christos case ENUM:
667 1.11 christos case UNION:
668 1.11 christos case SIGNED:
669 1.11 christos case NOTSPEC:
670 1.11 christos case DOUBLE:
671 1.11 christos case FLOAT:
672 1.11 christos case USHORT:
673 1.11 christos case SHORT:
674 1.11 christos case UCHAR:
675 1.11 christos case SCHAR:
676 1.11 christos case CHAR:
677 1.33 yamt case BOOL:
678 1.11 christos case UNSIGN:
679 1.38 christos case FCOMPLEX:
680 1.38 christos case DCOMPLEX:
681 1.40 matt case LCOMPLEX:
682 1.38 christos case COMPLEX:
683 1.23 thorpej break;
684 1.23 thorpej
685 1.23 thorpej case NTSPEC: /* this value unused */
686 1.11 christos break;
687 1.1 cgd }
688 1.1 cgd
689 1.27 thorpej uq = (uint64_t)xsign((int64_t)uq, typ, -1);
690 1.1 cgd
691 1.1 cgd (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
692 1.1 cgd yylval.y_val->v_ansiu = ansiu;
693 1.27 thorpej yylval.y_val->v_quad = (int64_t)uq;
694 1.1 cgd
695 1.1 cgd return (T_CON);
696 1.1 cgd }
697 1.1 cgd
698 1.1 cgd /*
699 1.1 cgd * Returns 1 if t is a signed type and the value is negative.
700 1.1 cgd *
701 1.1 cgd * len is the number of significant bits. If len is -1, len is set
702 1.1 cgd * to the width of type t.
703 1.1 cgd */
704 1.1 cgd int
705 1.27 thorpej sign(int64_t q, tspec_t t, int len)
706 1.1 cgd {
707 1.19 lukem
708 1.1 cgd if (t == PTR || isutyp(t))
709 1.1 cgd return (0);
710 1.1 cgd return (msb(q, t, len));
711 1.1 cgd }
712 1.1 cgd
713 1.1 cgd int
714 1.27 thorpej msb(int64_t q, tspec_t t, int len)
715 1.1 cgd {
716 1.19 lukem
717 1.1 cgd if (len <= 0)
718 1.1 cgd len = size(t);
719 1.1 cgd return ((q & qbmasks[len - 1]) != 0);
720 1.1 cgd }
721 1.1 cgd
722 1.1 cgd /*
723 1.1 cgd * Extends the sign of q.
724 1.1 cgd */
725 1.27 thorpej int64_t
726 1.27 thorpej xsign(int64_t q, tspec_t t, int len)
727 1.1 cgd {
728 1.19 lukem
729 1.1 cgd if (len <= 0)
730 1.1 cgd len = size(t);
731 1.1 cgd
732 1.1 cgd if (t == PTR || isutyp(t) || !sign(q, t, len)) {
733 1.1 cgd q &= qlmasks[len];
734 1.1 cgd } else {
735 1.1 cgd q |= qumasks[len];
736 1.1 cgd }
737 1.1 cgd return (q);
738 1.1 cgd }
739 1.1 cgd
740 1.1 cgd /*
741 1.1 cgd * Convert a string representing a floating point value into its interal
742 1.1 cgd * representation. Type and value are returned in yylval. fcon()
743 1.1 cgd * (and yylex()) returns T_CON.
744 1.1 cgd * XXX Currently it is not possible to convert constants of type
745 1.20 wiz * long double which are greater than DBL_MAX.
746 1.1 cgd */
747 1.1 cgd static int
748 1.19 lukem fcon(void)
749 1.1 cgd {
750 1.1 cgd const char *cp;
751 1.1 cgd int len;
752 1.1 cgd tspec_t typ;
753 1.1 cgd char c, *eptr;
754 1.1 cgd double d;
755 1.11 christos float f = 0;
756 1.1 cgd
757 1.1 cgd cp = yytext;
758 1.1 cgd len = yyleng;
759 1.1 cgd
760 1.38 christos if (cp[len - 1] == 'i') {
761 1.38 christos /* imaginary, do nothing for now */
762 1.38 christos len--;
763 1.38 christos }
764 1.1 cgd if ((c = cp[len - 1]) == 'f' || c == 'F') {
765 1.1 cgd typ = FLOAT;
766 1.1 cgd len--;
767 1.1 cgd } else if (c == 'l' || c == 'L') {
768 1.1 cgd typ = LDOUBLE;
769 1.1 cgd len--;
770 1.1 cgd } else {
771 1.49 christos if (c == 'd' || c == 'D')
772 1.49 christos len--;
773 1.1 cgd typ = DOUBLE;
774 1.1 cgd }
775 1.1 cgd
776 1.1 cgd if (tflag && typ != DOUBLE) {
777 1.1 cgd /* suffixes F and L are illegal in traditional C */
778 1.1 cgd warning(98);
779 1.1 cgd }
780 1.1 cgd
781 1.1 cgd errno = 0;
782 1.1 cgd d = strtod(cp, &eptr);
783 1.37 he if (eptr != cp + len) {
784 1.37 he switch (*eptr) {
785 1.37 he /*
786 1.37 he * XXX: non-native non-current strtod() may not handle hex
787 1.37 he * floats, ignore the rest if we find traces of hex float
788 1.37 he * syntax...
789 1.37 he */
790 1.37 he case 'p':
791 1.37 he case 'P':
792 1.37 he case 'x':
793 1.37 he case 'X':
794 1.37 he d = 0;
795 1.37 he errno = 0;
796 1.37 he break;
797 1.37 he default:
798 1.49 christos LERROR("fcon(%s->%s)", cp, eptr);
799 1.37 he }
800 1.37 he }
801 1.1 cgd if (errno != 0)
802 1.1 cgd /* floating-point constant out of range */
803 1.1 cgd warning(248);
804 1.1 cgd
805 1.1 cgd if (typ == FLOAT) {
806 1.1 cgd f = (float)d;
807 1.26 tv if (!finite(f)) {
808 1.1 cgd /* floating-point constant out of range */
809 1.1 cgd warning(248);
810 1.1 cgd f = f > 0 ? FLT_MAX : -FLT_MAX;
811 1.1 cgd }
812 1.1 cgd }
813 1.1 cgd
814 1.1 cgd (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
815 1.1 cgd if (typ == FLOAT) {
816 1.1 cgd yylval.y_val->v_ldbl = f;
817 1.1 cgd } else {
818 1.1 cgd yylval.y_val->v_ldbl = d;
819 1.1 cgd }
820 1.1 cgd
821 1.1 cgd return (T_CON);
822 1.1 cgd }
823 1.1 cgd
824 1.1 cgd static int
825 1.19 lukem operator(int t, op_t o)
826 1.1 cgd {
827 1.19 lukem
828 1.1 cgd yylval.y_op = o;
829 1.1 cgd return (t);
830 1.1 cgd }
831 1.1 cgd
832 1.1 cgd /*
833 1.1 cgd * Called if lex found a leading \'.
834 1.1 cgd */
835 1.1 cgd static int
836 1.19 lukem ccon(void)
837 1.1 cgd {
838 1.43 christos size_t n;
839 1.43 christos int val, c;
840 1.1 cgd char cv;
841 1.1 cgd
842 1.1 cgd n = 0;
843 1.1 cgd val = 0;
844 1.1 cgd while ((c = getescc('\'')) >= 0) {
845 1.1 cgd val = (val << CHAR_BIT) + c;
846 1.1 cgd n++;
847 1.1 cgd }
848 1.1 cgd if (c == -2) {
849 1.1 cgd /* unterminated character constant */
850 1.1 cgd error(253);
851 1.1 cgd } else {
852 1.1 cgd if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
853 1.1 cgd /* too many characters in character constant */
854 1.1 cgd error(71);
855 1.1 cgd } else if (n > 1) {
856 1.1 cgd /* multi-character character constant */
857 1.1 cgd warning(294);
858 1.1 cgd } else if (n == 0) {
859 1.1 cgd /* empty character constant */
860 1.1 cgd error(73);
861 1.1 cgd }
862 1.1 cgd }
863 1.1 cgd if (n == 1) {
864 1.1 cgd cv = (char)val;
865 1.1 cgd val = cv;
866 1.1 cgd }
867 1.19 lukem
868 1.1 cgd yylval.y_val = xcalloc(1, sizeof (val_t));
869 1.1 cgd yylval.y_val->v_tspec = INT;
870 1.1 cgd yylval.y_val->v_quad = val;
871 1.1 cgd
872 1.1 cgd return (T_CON);
873 1.1 cgd }
874 1.1 cgd
875 1.1 cgd /*
876 1.1 cgd * Called if lex found a leading L\'
877 1.1 cgd */
878 1.1 cgd static int
879 1.19 lukem wccon(void)
880 1.1 cgd {
881 1.8 jpo static char buf[MB_LEN_MAX + 1];
882 1.43 christos size_t i;
883 1.43 christos int c;
884 1.1 cgd wchar_t wc;
885 1.1 cgd
886 1.1 cgd i = 0;
887 1.1 cgd while ((c = getescc('\'')) >= 0) {
888 1.1 cgd if (i < MB_CUR_MAX)
889 1.1 cgd buf[i] = (char)c;
890 1.1 cgd i++;
891 1.1 cgd }
892 1.1 cgd
893 1.1 cgd wc = 0;
894 1.1 cgd
895 1.1 cgd if (c == -2) {
896 1.1 cgd /* unterminated character constant */
897 1.1 cgd error(253);
898 1.1 cgd } else if (c == 0) {
899 1.1 cgd /* empty character constant */
900 1.1 cgd error(73);
901 1.1 cgd } else {
902 1.1 cgd if (i > MB_CUR_MAX) {
903 1.1 cgd i = MB_CUR_MAX;
904 1.1 cgd /* too many characters in character constant */
905 1.1 cgd error(71);
906 1.1 cgd } else {
907 1.1 cgd buf[i] = '\0';
908 1.1 cgd (void)mbtowc(NULL, NULL, 0);
909 1.1 cgd if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
910 1.1 cgd /* invalid multibyte character */
911 1.1 cgd error(291);
912 1.1 cgd }
913 1.1 cgd }
914 1.1 cgd
915 1.1 cgd yylval.y_val = xcalloc(1, sizeof (val_t));
916 1.1 cgd yylval.y_val->v_tspec = WCHAR;
917 1.1 cgd yylval.y_val->v_quad = wc;
918 1.1 cgd
919 1.1 cgd return (T_CON);
920 1.1 cgd }
921 1.1 cgd
922 1.1 cgd /*
923 1.1 cgd * Read a character which is part of a character constant or of a string
924 1.1 cgd * and handle escapes.
925 1.1 cgd *
926 1.1 cgd * The Argument is the character which delimits the character constant or
927 1.1 cgd * string.
928 1.1 cgd *
929 1.1 cgd * Returns -1 if the end of the character constant or string is reached,
930 1.16 lukem * -2 if the EOF is reached, and the character otherwise.
931 1.1 cgd */
932 1.1 cgd static int
933 1.19 lukem getescc(int d)
934 1.1 cgd {
935 1.1 cgd static int pbc = -1;
936 1.1 cgd int n, c, v;
937 1.1 cgd
938 1.1 cgd if (pbc == -1) {
939 1.1 cgd c = inpc();
940 1.1 cgd } else {
941 1.1 cgd c = pbc;
942 1.1 cgd pbc = -1;
943 1.1 cgd }
944 1.1 cgd if (c == d)
945 1.1 cgd return (-1);
946 1.1 cgd switch (c) {
947 1.1 cgd case '\n':
948 1.16 lukem if (tflag) {
949 1.16 lukem /* newline in string or char constant */
950 1.16 lukem error(254);
951 1.16 lukem return (-2);
952 1.16 lukem }
953 1.16 lukem return (c);
954 1.1 cgd case EOF:
955 1.1 cgd return (-2);
956 1.1 cgd case '\\':
957 1.1 cgd switch (c = inpc()) {
958 1.1 cgd case '"':
959 1.1 cgd if (tflag && d == '\'')
960 1.1 cgd /* \" inside character constant undef. ... */
961 1.1 cgd warning(262);
962 1.1 cgd return ('"');
963 1.1 cgd case '\'':
964 1.1 cgd return ('\'');
965 1.1 cgd case '?':
966 1.1 cgd if (tflag)
967 1.1 cgd /* \? undefined in traditional C */
968 1.1 cgd warning(263);
969 1.1 cgd return ('?');
970 1.1 cgd case '\\':
971 1.1 cgd return ('\\');
972 1.1 cgd case 'a':
973 1.1 cgd if (tflag)
974 1.1 cgd /* \a undefined in traditional C */
975 1.1 cgd warning(81);
976 1.1 cgd return ('\a');
977 1.1 cgd case 'b':
978 1.1 cgd return ('\b');
979 1.1 cgd case 'f':
980 1.1 cgd return ('\f');
981 1.1 cgd case 'n':
982 1.1 cgd return ('\n');
983 1.1 cgd case 'r':
984 1.1 cgd return ('\r');
985 1.1 cgd case 't':
986 1.1 cgd return ('\t');
987 1.1 cgd case 'v':
988 1.1 cgd if (tflag)
989 1.1 cgd /* \v undefined in traditional C */
990 1.1 cgd warning(264);
991 1.1 cgd return ('\v');
992 1.1 cgd case '8': case '9':
993 1.1 cgd /* bad octal digit %c */
994 1.1 cgd warning(77, c);
995 1.1 cgd /* FALLTHROUGH */
996 1.1 cgd case '0': case '1': case '2': case '3':
997 1.1 cgd case '4': case '5': case '6': case '7':
998 1.1 cgd n = 3;
999 1.1 cgd v = 0;
1000 1.1 cgd do {
1001 1.1 cgd v = (v << 3) + (c - '0');
1002 1.1 cgd c = inpc();
1003 1.1 cgd } while (--n && isdigit(c) && (tflag || c <= '7'));
1004 1.1 cgd if (tflag && n > 0 && isdigit(c))
1005 1.1 cgd /* bad octal digit %c */
1006 1.1 cgd warning(77, c);
1007 1.1 cgd pbc = c;
1008 1.50 christos if (v > TARG_UCHAR_MAX) {
1009 1.1 cgd /* character escape does not fit in char. */
1010 1.1 cgd warning(76);
1011 1.1 cgd v &= CHAR_MASK;
1012 1.1 cgd }
1013 1.1 cgd return (v);
1014 1.1 cgd case 'x':
1015 1.1 cgd if (tflag)
1016 1.1 cgd /* \x undefined in traditional C */
1017 1.1 cgd warning(82);
1018 1.1 cgd v = 0;
1019 1.1 cgd n = 0;
1020 1.1 cgd while ((c = inpc()) >= 0 && isxdigit(c)) {
1021 1.1 cgd c = isdigit(c) ?
1022 1.1 cgd c - '0' : toupper(c) - 'A' + 10;
1023 1.1 cgd v = (v << 4) + c;
1024 1.1 cgd if (n >= 0) {
1025 1.1 cgd if ((v & ~CHAR_MASK) != 0) {
1026 1.1 cgd /* overflow in hex escape */
1027 1.1 cgd warning(75);
1028 1.1 cgd n = -1;
1029 1.1 cgd } else {
1030 1.1 cgd n++;
1031 1.1 cgd }
1032 1.1 cgd }
1033 1.1 cgd }
1034 1.1 cgd pbc = c;
1035 1.1 cgd if (n == 0) {
1036 1.1 cgd /* no hex digits follow \x */
1037 1.1 cgd error(74);
1038 1.1 cgd } if (n == -1) {
1039 1.1 cgd v &= CHAR_MASK;
1040 1.1 cgd }
1041 1.1 cgd return (v);
1042 1.1 cgd case '\n':
1043 1.1 cgd return (getescc(d));
1044 1.1 cgd case EOF:
1045 1.1 cgd return (-2);
1046 1.1 cgd default:
1047 1.1 cgd if (isprint(c)) {
1048 1.1 cgd /* dubious escape \%c */
1049 1.1 cgd warning(79, c);
1050 1.1 cgd } else {
1051 1.1 cgd /* dubious escape \%o */
1052 1.1 cgd warning(80, c);
1053 1.1 cgd }
1054 1.1 cgd }
1055 1.1 cgd }
1056 1.1 cgd return (c);
1057 1.1 cgd }
1058 1.1 cgd
1059 1.1 cgd /*
1060 1.1 cgd * Called for preprocessor directives. Currently implemented are:
1061 1.1 cgd * # lineno
1062 1.1 cgd * # lineno "filename"
1063 1.1 cgd */
1064 1.1 cgd static void
1065 1.19 lukem directive(void)
1066 1.1 cgd {
1067 1.1 cgd const char *cp, *fn;
1068 1.1 cgd char c, *eptr;
1069 1.1 cgd size_t fnl;
1070 1.1 cgd long ln;
1071 1.1 cgd static int first = 1;
1072 1.1 cgd
1073 1.1 cgd /* Go to first non-whitespace after # */
1074 1.19 lukem for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
1075 1.19 lukem continue;
1076 1.1 cgd
1077 1.13 christos if (!isdigit((unsigned char)c)) {
1078 1.1 cgd error:
1079 1.1 cgd /* undefined or invalid # directive */
1080 1.1 cgd warning(255);
1081 1.1 cgd return;
1082 1.1 cgd }
1083 1.1 cgd ln = strtol(--cp, &eptr, 10);
1084 1.1 cgd if (cp == eptr)
1085 1.1 cgd goto error;
1086 1.1 cgd if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
1087 1.1 cgd goto error;
1088 1.17 lukem while ((c = *cp++) == ' ' || c == '\t')
1089 1.17 lukem continue;
1090 1.1 cgd if (c != '\0') {
1091 1.1 cgd if (c != '"')
1092 1.1 cgd goto error;
1093 1.1 cgd fn = cp;
1094 1.1 cgd while ((c = *cp) != '"' && c != '\0')
1095 1.1 cgd cp++;
1096 1.1 cgd if (c != '"')
1097 1.1 cgd goto error;
1098 1.1 cgd if ((fnl = cp++ - fn) > PATH_MAX)
1099 1.1 cgd goto error;
1100 1.17 lukem while ((c = *cp++) == ' ' || c == '\t')
1101 1.17 lukem continue;
1102 1.1 cgd #if 0
1103 1.1 cgd if (c != '\0')
1104 1.1 cgd warning("extra character(s) after directive");
1105 1.1 cgd #endif
1106 1.9 cgd
1107 1.9 cgd /* empty string means stdin */
1108 1.9 cgd if (fnl == 0) {
1109 1.9 cgd fn = "{standard input}";
1110 1.9 cgd fnl = 16; /* strlen (fn) */
1111 1.9 cgd }
1112 1.1 cgd curr_pos.p_file = fnnalloc(fn, fnl);
1113 1.1 cgd /*
1114 1.1 cgd * If this is the first directive, the name is the name
1115 1.1 cgd * of the C source file as specified at the command line.
1116 1.1 cgd * It is written to the output file.
1117 1.1 cgd */
1118 1.1 cgd if (first) {
1119 1.1 cgd csrc_pos.p_file = curr_pos.p_file;
1120 1.1 cgd outsrc(curr_pos.p_file);
1121 1.1 cgd first = 0;
1122 1.1 cgd }
1123 1.1 cgd }
1124 1.1 cgd curr_pos.p_line = (int)ln - 1;
1125 1.9 cgd curr_pos.p_uniq = 0;
1126 1.9 cgd if (curr_pos.p_file == csrc_pos.p_file) {
1127 1.1 cgd csrc_pos.p_line = (int)ln - 1;
1128 1.9 cgd csrc_pos.p_uniq = 0;
1129 1.9 cgd }
1130 1.1 cgd }
1131 1.1 cgd
1132 1.1 cgd /*
1133 1.1 cgd * Handle lint comments. Following comments are currently understood:
1134 1.1 cgd * ARGSUSEDn
1135 1.22 thorpej * BITFIELDTYPE
1136 1.1 cgd * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1137 1.1 cgd * FALLTHRU FALLTHROUGH
1138 1.1 cgd * LINTLIBRARY
1139 1.52 christos * LINTEDn NOSTRICTn
1140 1.7 jpo * LONGLONG
1141 1.1 cgd * NOTREACHED
1142 1.1 cgd * PRINTFLIKEn
1143 1.1 cgd * PROTOLIB
1144 1.1 cgd * SCANFLIKEn
1145 1.1 cgd * VARARGSn
1146 1.1 cgd * If one of this comments is recognized, the arguments, if any, are
1147 1.1 cgd * parsed and a function which handles this comment is called.
1148 1.1 cgd */
1149 1.1 cgd static void
1150 1.19 lukem comment(void)
1151 1.1 cgd {
1152 1.1 cgd int c, lc;
1153 1.1 cgd static struct {
1154 1.1 cgd const char *keywd;
1155 1.1 cgd int arg;
1156 1.19 lukem void (*func)(int);
1157 1.1 cgd } keywtab[] = {
1158 1.1 cgd { "ARGSUSED", 1, argsused },
1159 1.22 thorpej { "BITFIELDTYPE", 0, bitfieldtype },
1160 1.1 cgd { "CONSTCOND", 0, constcond },
1161 1.1 cgd { "CONSTANTCOND", 0, constcond },
1162 1.1 cgd { "CONSTANTCONDITION", 0, constcond },
1163 1.1 cgd { "FALLTHRU", 0, fallthru },
1164 1.1 cgd { "FALLTHROUGH", 0, fallthru },
1165 1.1 cgd { "LINTLIBRARY", 0, lintlib },
1166 1.51 christos { "LINTED", 1, linted },
1167 1.7 jpo { "LONGLONG", 0, longlong },
1168 1.52 christos { "NOSTRICT", 1, linted },
1169 1.1 cgd { "NOTREACHED", 0, notreach },
1170 1.1 cgd { "PRINTFLIKE", 1, printflike },
1171 1.1 cgd { "PROTOLIB", 1, protolib },
1172 1.1 cgd { "SCANFLIKE", 1, scanflike },
1173 1.1 cgd { "VARARGS", 1, varargs },
1174 1.1 cgd };
1175 1.1 cgd char keywd[32];
1176 1.1 cgd char arg[32];
1177 1.43 christos size_t l, i;
1178 1.43 christos int a;
1179 1.1 cgd int eoc;
1180 1.1 cgd
1181 1.1 cgd eoc = 0;
1182 1.1 cgd
1183 1.1 cgd /* Skip white spaces after the start of the comment */
1184 1.17 lukem while ((c = inpc()) != EOF && isspace(c))
1185 1.17 lukem continue;
1186 1.1 cgd
1187 1.1 cgd /* Read the potential keyword to keywd */
1188 1.1 cgd l = 0;
1189 1.1 cgd while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1190 1.1 cgd keywd[l++] = (char)c;
1191 1.1 cgd c = inpc();
1192 1.1 cgd }
1193 1.1 cgd keywd[l] = '\0';
1194 1.1 cgd
1195 1.1 cgd /* look for the keyword */
1196 1.1 cgd for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1197 1.1 cgd if (strcmp(keywtab[i].keywd, keywd) == 0)
1198 1.1 cgd break;
1199 1.1 cgd }
1200 1.1 cgd if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1201 1.1 cgd goto skip_rest;
1202 1.1 cgd
1203 1.1 cgd /* skip white spaces after the keyword */
1204 1.1 cgd while (c != EOF && isspace(c))
1205 1.1 cgd c = inpc();
1206 1.1 cgd
1207 1.1 cgd /* read the argument, if the keyword accepts one and there is one */
1208 1.1 cgd l = 0;
1209 1.1 cgd if (keywtab[i].arg) {
1210 1.1 cgd while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1211 1.1 cgd arg[l++] = (char)c;
1212 1.1 cgd c = inpc();
1213 1.1 cgd }
1214 1.1 cgd }
1215 1.1 cgd arg[l] = '\0';
1216 1.1 cgd a = l != 0 ? atoi(arg) : -1;
1217 1.1 cgd
1218 1.1 cgd /* skip white spaces after the argument */
1219 1.1 cgd while (c != EOF && isspace(c))
1220 1.1 cgd c = inpc();
1221 1.1 cgd
1222 1.1 cgd if (c != '*' || (c = inpc()) != '/') {
1223 1.1 cgd if (keywtab[i].func != linted)
1224 1.1 cgd /* extra characters in lint comment */
1225 1.1 cgd warning(257);
1226 1.1 cgd } else {
1227 1.1 cgd /*
1228 1.1 cgd * remember that we have already found the end of the
1229 1.1 cgd * comment
1230 1.1 cgd */
1231 1.1 cgd eoc = 1;
1232 1.1 cgd }
1233 1.1 cgd
1234 1.1 cgd if (keywtab[i].func != NULL)
1235 1.1 cgd (*keywtab[i].func)(a);
1236 1.1 cgd
1237 1.1 cgd skip_rest:
1238 1.1 cgd while (!eoc) {
1239 1.1 cgd lc = c;
1240 1.1 cgd if ((c = inpc()) == EOF) {
1241 1.1 cgd /* unterminated comment */
1242 1.1 cgd error(256);
1243 1.1 cgd break;
1244 1.1 cgd }
1245 1.1 cgd if (lc == '*' && c == '/')
1246 1.1 cgd eoc = 1;
1247 1.1 cgd }
1248 1.18 lukem }
1249 1.18 lukem
1250 1.18 lukem /*
1251 1.18 lukem * Handle // style comments
1252 1.18 lukem */
1253 1.18 lukem static void
1254 1.19 lukem slashslashcomment(void)
1255 1.18 lukem {
1256 1.18 lukem int c;
1257 1.18 lukem
1258 1.32 perry if (!Sflag && !gflag)
1259 1.21 wiz /* // comments only supported in C99 */
1260 1.18 lukem (void)gnuism(312, tflag ? "traditional" : "ANSI");
1261 1.18 lukem
1262 1.18 lukem while ((c = inpc()) != EOF && c != '\n')
1263 1.18 lukem continue;
1264 1.7 jpo }
1265 1.7 jpo
1266 1.7 jpo /*
1267 1.7 jpo * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1268 1.7 jpo * clrwflgs() is called after function definitions and global and
1269 1.7 jpo * local declarations and definitions. It is also called between
1270 1.7 jpo * the controlling expression and the body of control statements
1271 1.7 jpo * (if, switch, for, while).
1272 1.7 jpo */
1273 1.7 jpo void
1274 1.19 lukem clrwflgs(void)
1275 1.7 jpo {
1276 1.19 lukem
1277 1.51 christos lwarn = LWARN_ALL;
1278 1.7 jpo quadflg = 0;
1279 1.7 jpo ccflg = 0;
1280 1.1 cgd }
1281 1.1 cgd
1282 1.1 cgd /*
1283 1.1 cgd * Strings are stored in a dynamically alloceted buffer and passed
1284 1.1 cgd * in yylval.y_xstrg to the parser. The parser or the routines called
1285 1.1 cgd * by the parser are responsible for freeing this buffer.
1286 1.1 cgd */
1287 1.1 cgd static int
1288 1.19 lukem string(void)
1289 1.1 cgd {
1290 1.1 cgd u_char *s;
1291 1.1 cgd int c;
1292 1.1 cgd size_t len, max;
1293 1.1 cgd strg_t *strg;
1294 1.1 cgd
1295 1.1 cgd s = xmalloc(max = 64);
1296 1.1 cgd
1297 1.1 cgd len = 0;
1298 1.1 cgd while ((c = getescc('"')) >= 0) {
1299 1.1 cgd /* +1 to reserve space for a trailing NUL character */
1300 1.1 cgd if (len + 1 == max)
1301 1.1 cgd s = xrealloc(s, max *= 2);
1302 1.1 cgd s[len++] = (char)c;
1303 1.1 cgd }
1304 1.1 cgd s[len] = '\0';
1305 1.1 cgd if (c == -2)
1306 1.1 cgd /* unterminated string constant */
1307 1.1 cgd error(258);
1308 1.1 cgd
1309 1.1 cgd strg = xcalloc(1, sizeof (strg_t));
1310 1.1 cgd strg->st_tspec = CHAR;
1311 1.1 cgd strg->st_len = len;
1312 1.1 cgd strg->st_cp = s;
1313 1.1 cgd
1314 1.1 cgd yylval.y_strg = strg;
1315 1.1 cgd return (T_STRING);
1316 1.1 cgd }
1317 1.1 cgd
1318 1.1 cgd static int
1319 1.19 lukem wcstrg(void)
1320 1.1 cgd {
1321 1.1 cgd char *s;
1322 1.43 christos int c, n;
1323 1.43 christos size_t i, wi;
1324 1.1 cgd size_t len, max, wlen;
1325 1.1 cgd wchar_t *ws;
1326 1.1 cgd strg_t *strg;
1327 1.1 cgd
1328 1.1 cgd s = xmalloc(max = 64);
1329 1.1 cgd len = 0;
1330 1.1 cgd while ((c = getescc('"')) >= 0) {
1331 1.1 cgd /* +1 to save space for a trailing NUL character */
1332 1.1 cgd if (len + 1 >= max)
1333 1.1 cgd s = xrealloc(s, max *= 2);
1334 1.1 cgd s[len++] = (char)c;
1335 1.1 cgd }
1336 1.1 cgd s[len] = '\0';
1337 1.1 cgd if (c == -2)
1338 1.1 cgd /* unterminated string constant */
1339 1.1 cgd error(258);
1340 1.1 cgd
1341 1.48 wiz /* get length of wide-character string */
1342 1.1 cgd (void)mblen(NULL, 0);
1343 1.1 cgd for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1344 1.1 cgd if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1345 1.1 cgd /* invalid multibyte character */
1346 1.1 cgd error(291);
1347 1.1 cgd break;
1348 1.1 cgd }
1349 1.1 cgd if (n == 0)
1350 1.1 cgd n = 1;
1351 1.1 cgd }
1352 1.1 cgd
1353 1.1 cgd ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1354 1.1 cgd
1355 1.1 cgd /* convert from multibyte to wide char */
1356 1.1 cgd (void)mbtowc(NULL, NULL, 0);
1357 1.1 cgd for (i = 0, wi = 0; i < len; i += n, wi++) {
1358 1.1 cgd if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1359 1.1 cgd break;
1360 1.1 cgd if (n == 0)
1361 1.1 cgd n = 1;
1362 1.1 cgd }
1363 1.1 cgd ws[wi] = 0;
1364 1.1 cgd free(s);
1365 1.1 cgd
1366 1.1 cgd strg = xcalloc(1, sizeof (strg_t));
1367 1.1 cgd strg->st_tspec = WCHAR;
1368 1.1 cgd strg->st_len = wlen;
1369 1.1 cgd strg->st_wcp = ws;
1370 1.1 cgd
1371 1.1 cgd yylval.y_strg = strg;
1372 1.1 cgd return (T_STRING);
1373 1.1 cgd }
1374 1.1 cgd
1375 1.1 cgd /*
1376 1.1 cgd * As noted above the scanner does not create new symbol table entries
1377 1.1 cgd * for symbols it cannot find in the symbol table. This is to avoid
1378 1.1 cgd * putting undeclared symbols into the symbol table if a syntax error
1379 1.1 cgd * occurs.
1380 1.1 cgd *
1381 1.1 cgd * getsym() is called as soon as it is probably ok to put the symbol to
1382 1.1 cgd * the symbol table. This does not mean that it is not possible that
1383 1.1 cgd * symbols are put to the symbol table which are than not completely
1384 1.1 cgd * declared due to syntax errors. To avoid too many problems in this
1385 1.1 cgd * case symbols get type int in getsym().
1386 1.1 cgd *
1387 1.1 cgd * XXX calls to getsym() should be delayed until decl1*() is called
1388 1.1 cgd */
1389 1.1 cgd sym_t *
1390 1.19 lukem getsym(sbuf_t *sb)
1391 1.1 cgd {
1392 1.1 cgd dinfo_t *di;
1393 1.1 cgd char *s;
1394 1.1 cgd sym_t *sym;
1395 1.1 cgd
1396 1.1 cgd sym = sb->sb_sym;
1397 1.1 cgd
1398 1.1 cgd /*
1399 1.1 cgd * During member declaration it is possible that name() looked
1400 1.1 cgd * for symbols of type FVFT, although it should have looked for
1401 1.1 cgd * symbols of type FTAG. Same can happen for labels. Both cases
1402 1.1 cgd * are compensated here.
1403 1.1 cgd */
1404 1.1 cgd if (symtyp == FMOS || symtyp == FLAB) {
1405 1.1 cgd if (sym == NULL || sym->s_kind == FVFT)
1406 1.1 cgd sym = search(sb);
1407 1.1 cgd }
1408 1.1 cgd
1409 1.1 cgd if (sym != NULL) {
1410 1.1 cgd if (sym->s_kind != symtyp)
1411 1.56 christos LERROR("storesym(%d, %d)", sym->s_kind, symtyp);
1412 1.1 cgd symtyp = FVFT;
1413 1.1 cgd freesb(sb);
1414 1.1 cgd return (sym);
1415 1.1 cgd }
1416 1.1 cgd
1417 1.1 cgd /* create a new symbol table entry */
1418 1.1 cgd
1419 1.1 cgd /* labels must always be allocated at level 1 (outhermost block) */
1420 1.1 cgd if (symtyp == FLAB) {
1421 1.1 cgd sym = getlblk(1, sizeof (sym_t));
1422 1.1 cgd s = getlblk(1, sb->sb_len + 1);
1423 1.1 cgd (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1424 1.1 cgd sym->s_name = s;
1425 1.1 cgd sym->s_blklev = 1;
1426 1.1 cgd di = dcs;
1427 1.3 jpo while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1428 1.3 jpo di = di->d_nxt;
1429 1.3 jpo if (di->d_ctx != AUTO)
1430 1.28 christos LERROR("storesym()");
1431 1.1 cgd } else {
1432 1.1 cgd sym = getblk(sizeof (sym_t));
1433 1.1 cgd sym->s_name = sb->sb_name;
1434 1.1 cgd sym->s_blklev = blklev;
1435 1.1 cgd di = dcs;
1436 1.1 cgd }
1437 1.1 cgd
1438 1.9 cgd UNIQUE_CURR_POS(sym->s_dpos);
1439 1.1 cgd if ((sym->s_kind = symtyp) != FLAB)
1440 1.1 cgd sym->s_type = gettyp(INT);
1441 1.1 cgd
1442 1.1 cgd symtyp = FVFT;
1443 1.1 cgd
1444 1.1 cgd if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1445 1.1 cgd symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1446 1.53 christos sym->s_rlink = &symtab[sb->sb_hash];
1447 1.53 christos symtab[sb->sb_hash] = sym;
1448 1.1 cgd
1449 1.3 jpo *di->d_ldlsym = sym;
1450 1.3 jpo di->d_ldlsym = &sym->s_dlnxt;
1451 1.1 cgd
1452 1.1 cgd freesb(sb);
1453 1.1 cgd return (sym);
1454 1.29 christos }
1455 1.29 christos
1456 1.29 christos /*
1457 1.29 christos * Construct a temporary symbol. The symbol starts with a digit, so that
1458 1.29 christos * it is illegal.
1459 1.29 christos */
1460 1.29 christos sym_t *
1461 1.29 christos mktempsym(type_t *t)
1462 1.29 christos {
1463 1.29 christos static int n = 0;
1464 1.29 christos int h;
1465 1.29 christos char *s = getlblk(blklev, 64);
1466 1.29 christos sym_t *sym = getblk(sizeof (sym_t));
1467 1.29 christos
1468 1.29 christos (void)snprintf(s, 64, "%.8d_tmp", n++);
1469 1.29 christos h = hash(s);
1470 1.30 christos
1471 1.29 christos sym->s_name = s;
1472 1.29 christos sym->s_type = t;
1473 1.29 christos sym->s_blklev = blklev;
1474 1.29 christos sym->s_scl = AUTO;
1475 1.30 christos sym->s_kind = FVFT;
1476 1.31 christos sym->s_used = 1;
1477 1.31 christos sym->s_set = 1;
1478 1.30 christos
1479 1.29 christos if ((sym->s_link = symtab[h]) != NULL)
1480 1.29 christos symtab[h]->s_rlink = &sym->s_link;
1481 1.53 christos sym->s_rlink = &symtab[h];
1482 1.53 christos symtab[h] = sym;
1483 1.30 christos
1484 1.30 christos *dcs->d_ldlsym = sym;
1485 1.30 christos dcs->d_ldlsym = &sym->s_dlnxt;
1486 1.30 christos
1487 1.29 christos return sym;
1488 1.1 cgd }
1489 1.1 cgd
1490 1.1 cgd /*
1491 1.1 cgd * Remove a symbol forever from the symbol table. s_blklev
1492 1.1 cgd * is set to -1 to avoid that the symbol will later be put
1493 1.1 cgd * back to the symbol table.
1494 1.1 cgd */
1495 1.1 cgd void
1496 1.19 lukem rmsym(sym_t *sym)
1497 1.1 cgd {
1498 1.19 lukem
1499 1.1 cgd if ((*sym->s_rlink = sym->s_link) != NULL)
1500 1.1 cgd sym->s_link->s_rlink = sym->s_rlink;
1501 1.1 cgd sym->s_blklev = -1;
1502 1.1 cgd sym->s_link = NULL;
1503 1.1 cgd }
1504 1.1 cgd
1505 1.1 cgd /*
1506 1.1 cgd * Remove a list of symbols declared at one level from the symbol
1507 1.1 cgd * table.
1508 1.1 cgd */
1509 1.1 cgd void
1510 1.19 lukem rmsyms(sym_t *syms)
1511 1.1 cgd {
1512 1.1 cgd sym_t *sym;
1513 1.1 cgd
1514 1.1 cgd for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1515 1.1 cgd if (sym->s_blklev != -1) {
1516 1.1 cgd if ((*sym->s_rlink = sym->s_link) != NULL)
1517 1.1 cgd sym->s_link->s_rlink = sym->s_rlink;
1518 1.1 cgd sym->s_link = NULL;
1519 1.1 cgd sym->s_rlink = NULL;
1520 1.1 cgd }
1521 1.1 cgd }
1522 1.1 cgd }
1523 1.1 cgd
1524 1.1 cgd /*
1525 1.1 cgd * Put a symbol into the symbol table
1526 1.1 cgd */
1527 1.1 cgd void
1528 1.19 lukem inssym(int bl, sym_t *sym)
1529 1.1 cgd {
1530 1.1 cgd int h;
1531 1.1 cgd
1532 1.1 cgd h = hash(sym->s_name);
1533 1.1 cgd if ((sym->s_link = symtab[h]) != NULL)
1534 1.1 cgd symtab[h]->s_rlink = &sym->s_link;
1535 1.53 christos sym->s_rlink = &symtab[h];
1536 1.53 christos symtab[h] = sym;
1537 1.1 cgd sym->s_blklev = bl;
1538 1.1 cgd if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1539 1.28 christos LERROR("inssym()");
1540 1.1 cgd }
1541 1.1 cgd
1542 1.1 cgd /*
1543 1.1 cgd * Called at level 0 after syntax errors
1544 1.1 cgd * Removes all symbols which are not declared at level 0 from the
1545 1.1 cgd * symbol table. Also frees all memory which is not associated with
1546 1.1 cgd * level 0.
1547 1.1 cgd */
1548 1.1 cgd void
1549 1.19 lukem cleanup(void)
1550 1.1 cgd {
1551 1.1 cgd sym_t *sym, *nsym;
1552 1.1 cgd int i;
1553 1.1 cgd
1554 1.1 cgd for (i = 0; i < HSHSIZ1; i++) {
1555 1.1 cgd for (sym = symtab[i]; sym != NULL; sym = nsym) {
1556 1.1 cgd nsym = sym->s_link;
1557 1.1 cgd if (sym->s_blklev >= 1) {
1558 1.1 cgd if ((*sym->s_rlink = nsym) != NULL)
1559 1.1 cgd nsym->s_rlink = sym->s_rlink;
1560 1.1 cgd }
1561 1.1 cgd }
1562 1.1 cgd }
1563 1.1 cgd
1564 1.1 cgd for (i = mblklev; i > 0; i--)
1565 1.1 cgd freelblk(i);
1566 1.1 cgd }
1567 1.1 cgd
1568 1.1 cgd /*
1569 1.1 cgd * Create a new symbol with the name of an existing symbol.
1570 1.1 cgd */
1571 1.1 cgd sym_t *
1572 1.19 lukem pushdown(sym_t *sym)
1573 1.1 cgd {
1574 1.1 cgd int h;
1575 1.1 cgd sym_t *nsym;
1576 1.1 cgd
1577 1.1 cgd h = hash(sym->s_name);
1578 1.1 cgd nsym = getblk(sizeof (sym_t));
1579 1.1 cgd if (sym->s_blklev > blklev)
1580 1.28 christos LERROR("pushdown()");
1581 1.1 cgd nsym->s_name = sym->s_name;
1582 1.9 cgd UNIQUE_CURR_POS(nsym->s_dpos);
1583 1.1 cgd nsym->s_kind = sym->s_kind;
1584 1.1 cgd nsym->s_blklev = blklev;
1585 1.1 cgd
1586 1.1 cgd if ((nsym->s_link = symtab[h]) != NULL)
1587 1.1 cgd symtab[h]->s_rlink = &nsym->s_link;
1588 1.53 christos nsym->s_rlink = &symtab[h];
1589 1.53 christos symtab[h] = nsym;
1590 1.1 cgd
1591 1.3 jpo *dcs->d_ldlsym = nsym;
1592 1.3 jpo dcs->d_ldlsym = &nsym->s_dlnxt;
1593 1.1 cgd
1594 1.1 cgd return (nsym);
1595 1.6 jpo }
1596 1.6 jpo
1597 1.6 jpo /*
1598 1.6 jpo * Free any dynamically allocated memory referenced by
1599 1.6 jpo * the value stack or yylval.
1600 1.6 jpo * The type of information in yylval is described by tok.
1601 1.6 jpo */
1602 1.6 jpo void
1603 1.19 lukem freeyyv(void *sp, int tok)
1604 1.6 jpo {
1605 1.6 jpo if (tok == T_NAME || tok == T_TYPENAME) {
1606 1.6 jpo sbuf_t *sb = *(sbuf_t **)sp;
1607 1.6 jpo freesb(sb);
1608 1.6 jpo } else if (tok == T_CON) {
1609 1.6 jpo val_t *val = *(val_t **)sp;
1610 1.6 jpo free(val);
1611 1.6 jpo } else if (tok == T_STRING) {
1612 1.6 jpo strg_t *strg = *(strg_t **)sp;
1613 1.6 jpo if (strg->st_tspec == CHAR) {
1614 1.6 jpo free(strg->st_cp);
1615 1.6 jpo } else if (strg->st_tspec == WCHAR) {
1616 1.6 jpo free(strg->st_wcp);
1617 1.6 jpo } else {
1618 1.28 christos LERROR("fryylv()");
1619 1.6 jpo }
1620 1.6 jpo free(strg);
1621 1.19 lukem }
1622 1.1 cgd }
1623