diff options
Diffstat (limited to 'usr.bin/xlint/lint1/scan.l')
-rw-r--r-- | usr.bin/xlint/lint1/scan.l | 1426 |
1 files changed, 1426 insertions, 0 deletions
diff --git a/usr.bin/xlint/lint1/scan.l b/usr.bin/xlint/lint1/scan.l new file mode 100644 index 0000000..5e23c6b --- /dev/null +++ b/usr.bin/xlint/lint1/scan.l @@ -0,0 +1,1426 @@ +%{ +/* $NetBSD: scan.l,v 1.8 1995/10/23 13:38:51 jpo Exp $ */ + +/* + * Copyright (c) 1994, 1995 Jochen Pohl + * All Rights Reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by Jochen Pohl for + * The NetBSD Project. + * 4. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef lint +static char rcsid[] = "$NetBSD: scan.l,v 1.8 1995/10/23 13:38:51 jpo Exp $"; +#endif + +#include <stdlib.h> +#include <string.h> +#include <limits.h> +#include <float.h> +#include <ctype.h> +#include <errno.h> +#include <math.h> +#include <err.h> + +#include "lint1.h" +#include "y.tab.h" + +#define CHAR_MASK (~(~0 << CHAR_BIT)) + +/* XXX declaration of strtouq() is missing in stdlib.h ? */ +extern u_quad_t strtouq __P((const char *, char **, int)); + +/* Current position (its also updated when an included file is parsed) */ +pos_t curr_pos = { 1, "" }; + +/* + * Current position in C source (not updated when an included file is + * parsed). + */ +pos_t csrc_pos = { 1, "" }; + +static void incline __P((void)); +static void badchar __P((int)); +static sbuf_t *allocsb __P((void)); +static void freesb __P((sbuf_t *)); +static int inpc __P((void)); +static int hash __P((const char *)); +static sym_t *search __P((sbuf_t *)); +static int name __P((void)); +static int keyw __P((sym_t *)); +static int icon __P((int)); +static int fcon __P((void)); +static int operator __P((int, op_t)); +static int ccon __P((void)); +static int wccon __P((void)); +static int getescc __P((int)); +static void directive __P((void)); +static void comment __P((void)); +static int string __P((void)); +static int wcstrg __P((void)); + +%} + +L [_A-Za-z] +D [0-9] +NZD [1-9] +OD [0-7] +HD [0-9A-Fa-f] +EX ([eE][+-]?[0-9]+) + +%% + +{L}({L}|{D})* return (name()); +0{OD}*[lLuU]* return (icon(8)); +{NZD}{D}*[lLuU]* return (icon(10)); +0[xX]{HD}+[lLuU]* return (icon(16)); +{D}+\.{D}*{EX}?[fFlL]? | +{D}+{EX}[fFlL]? | +\.{D}+{EX}?[fFlL]? return (fcon()); +"=" return (operator(T_ASSIGN, ASSIGN)); +"*=" return (operator(T_OPASS, MULASS)); +"/=" return (operator(T_OPASS, DIVASS)); +"%=" return (operator(T_OPASS, MODASS)); +"+=" return (operator(T_OPASS, ADDASS)); +"-=" return (operator(T_OPASS, SUBASS)); +"<<=" return (operator(T_OPASS, SHLASS)); +">>=" return (operator(T_OPASS, SHRASS)); +"&=" return (operator(T_OPASS, ANDASS)); +"^=" return (operator(T_OPASS, XORASS)); +"|=" return (operator(T_OPASS, ORASS)); +"||" return (operator(T_LOGOR, LOGOR)); +"&&" return (operator(T_LOGAND, LOGAND)); +"|" return (operator(T_OR, OR)); +"&" return (operator(T_AND, AND)); +"^" return (operator(T_XOR, XOR)); +"==" return (operator(T_EQOP, EQ)); +"!=" return (operator(T_EQOP, NE)); +"<" return (operator(T_RELOP, LT)); +">" return (operator(T_RELOP, GT)); +"<=" return (operator(T_RELOP, LE)); +">=" return (operator(T_RELOP, GE)); +"<<" return (operator(T_SHFTOP, SHL)); +">>" return (operator(T_SHFTOP, SHR)); +"++" return (operator(T_INCDEC, INC)); +"--" return (operator(T_INCDEC, DEC)); +"->" return (operator(T_STROP, ARROW)); +"." return (operator(T_STROP, POINT)); +"+" return (operator(T_ADDOP, PLUS)); +"-" return (operator(T_ADDOP, MINUS)); +"*" return (operator(T_MULT, MULT)); +"/" return (operator(T_DIVOP, DIV)); +"%" return (operator(T_DIVOP, MOD)); +"!" return (operator(T_UNOP, NOT)); +"~" return (operator(T_UNOP, COMPL)); +"\"" return (string()); +"L\"" return (wcstrg()); +";" return (T_SEMI); +"{" return (T_LBRACE); +"}" return (T_RBRACE); +"," return (T_COMMA); +":" return (T_COLON); +"?" return (T_QUEST); +"[" return (T_LBRACK); +"]" return (T_RBRACK); +"(" return (T_LPARN); +")" return (T_RPARN); +"..." return (T_ELLIPSE); +"'" return (ccon()); +"L'" return (wccon()); +^#.*$ directive(); +\n incline(); +\t|" "|\f|\v ; +"/*" comment(); +. badchar(yytext[0]); + +%% + +static void +incline() +{ + curr_pos.p_line++; + if (curr_pos.p_file == csrc_pos.p_file) + csrc_pos.p_line++; +} + +static void +badchar(c) + int c; +{ + /* unknown character \%o */ + error(250, c); +} + +/* + * Keywords. + * During initialisation they are written to the symbol table. + */ +static struct kwtab { + const char *kw_name; /* keyword */ + int kw_token; /* token returned by yylex() */ + scl_t kw_scl; /* storage class if kw_token T_SCLASS */ + tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */ + tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */ + u_int kw_stdc : 1; /* STDC keyword */ + u_int kw_gcc : 1; /* GCC keyword */ +} kwtab[] = { + { "asm", T_ASM, 0, 0, 0, 0, 1 }, + { "__asm", T_ASM, 0, 0, 0, 0, 0 }, + { "__asm__", T_ASM, 0, 0, 0, 0, 0 }, + { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 }, + { "break", T_BREAK, 0, 0, 0, 0, 0 }, + { "case", T_CASE, 0, 0, 0, 0, 0 }, + { "char", T_TYPE, 0, CHAR, 0, 0, 0 }, + { "const", T_QUAL, 0, 0, CONST, 1, 0 }, + { "__const__", T_QUAL, 0, 0, CONST, 0, 0 }, + { "__const", T_QUAL, 0, 0, CONST, 0, 0 }, + { "continue", T_CONTINUE, 0, 0, 0, 0, 0 }, + { "default", T_DEFAULT, 0, 0, 0, 0, 0 }, + { "do", T_DO, 0, 0, 0, 0, 0 }, + { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 }, + { "else", T_ELSE, 0, 0, 0, 0, 0 }, + { "enum", T_ENUM, 0, 0, 0, 0, 0 }, + { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 }, + { "float", T_TYPE, 0, FLOAT, 0, 0, 0 }, + { "for", T_FOR, 0, 0, 0, 0, 0 }, + { "goto", T_GOTO, 0, 0, 0, 0, 0 }, + { "if", T_IF, 0, 0, 0, 0, 0 }, + { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 }, + { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 }, + { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 }, + { "int", T_TYPE, 0, INT, 0, 0, 0 }, + { "long", T_TYPE, 0, LONG, 0, 0, 0 }, + { "register", T_SCLASS, REG, 0, 0, 0, 0 }, + { "return", T_RETURN, 0, 0, 0, 0, 0 }, + { "short", T_TYPE, 0, SHORT, 0, 0, 0 }, + { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 }, + { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 }, + { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 }, + { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 }, + { "static", T_SCLASS, STATIC, 0, 0, 0, 0 }, + { "struct", T_SOU, 0, STRUCT, 0, 0, 0 }, + { "switch", T_SWITCH, 0, 0, 0, 0, 0 }, + { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 }, + { "union", T_SOU, 0, UNION, 0, 0, 0 }, + { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 }, + { "void", T_TYPE, 0, VOID, 0, 0, 0 }, + { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 }, + { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 }, + { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 }, + { "while", T_WHILE, 0, 0, 0, 0, 0 }, + { NULL, 0, 0, 0, 0, 0, 0 } +}; + +/* Symbol table */ +static sym_t *symtab[HSHSIZ1]; + +/* bit i of the entry with index i is set */ +u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT]; + +/* least significant i bits are set in the entry with index i */ +u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1]; + +/* least significant i bits are not set in the entry with index i */ +u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1]; + +/* free list for sbuf structures */ +static sbuf_t *sbfrlst; + +/* Typ of next expected symbol */ +symt_t symtyp; + + +/* + * All keywords are written to the symbol table. This saves us looking + * in a extra table for each name we found. + */ +void +initscan() +{ + struct kwtab *kw; + sym_t *sym; + int h, i; + u_quad_t uq; + + for (kw = kwtab; kw->kw_name != NULL; kw++) { + if (kw->kw_stdc && tflag) + continue; + if (kw->kw_gcc && !gflag) + continue; + sym = getblk(sizeof (sym_t)); + sym->s_name = kw->kw_name; + sym->s_keyw = 1; + sym->s_value.v_quad = kw->kw_token; + if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) { + sym->s_tspec = kw->kw_tspec; + } else if (kw->kw_token == T_SCLASS) { + sym->s_scl = kw->kw_scl; + } else if (kw->kw_token == T_QUAL) { + sym->s_tqual = kw->kw_tqual; + } + h = hash(sym->s_name); + if ((sym->s_link = symtab[h]) != NULL) + symtab[h]->s_rlink = &sym->s_link; + (symtab[h] = sym)->s_rlink = &symtab[h]; + } + + /* initialize bit-masks for quads */ + for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) { + qbmasks[i] = (u_quad_t)1 << i; + uq = ~(u_quad_t)0 << i; + qumasks[i] = uq; + qlmasks[i] = ~uq; + } + qumasks[i] = 0; + qlmasks[i] = ~(u_quad_t)0; +} + +/* + * Get a free sbuf structure, if possible from the free list + */ +static sbuf_t * +allocsb() +{ + sbuf_t *sb; + + if ((sb = sbfrlst) != NULL) { + sbfrlst = sb->sb_nxt; + } else { + sb = xmalloc(sizeof (sbuf_t)); + } + (void)memset(sb, 0, sizeof (sb)); + return (sb); +} + +/* + * Put a sbuf structure to the free list + */ +static void +freesb(sb) + sbuf_t *sb; +{ + sb->sb_nxt = sbfrlst; + sbfrlst = sb; +} + +/* + * Read a character and ensure that it is positive (except EOF). + * Increment line count(s) if necessary. + */ +static int +inpc() +{ + int c; + + if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n') + incline(); + return (c); +} + +static int +hash(s) + const char *s; +{ + u_int v; + const u_char *us; + + v = 0; + for (us = (const u_char *)s; *us != '\0'; us++) { + v = (v << sizeof (v)) + *us; + v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v)); + } + return (v % HSHSIZ1); +} + +/* + * Lex has found a letter followed by zero or more letters or digits. + * It looks for a symbol in the symbol table with the same name. This + * symbol must either be a keyword or a symbol of the type required by + * symtyp (label, member, tag, ...). + * + * If it is a keyword, the token is returned. In some cases it is described + * more deeply by data written to yylval. + * + * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct + * is stored in yylval. This struct contains the name of the symbol, it's + * length and hash value. If there is already a symbol of the same name + * and type in the symbol table, the sbuf struct also contains a pointer + * to the symbol table entry. + */ +static int +name() +{ + char *s; + sbuf_t *sb; + sym_t *sym; + int tok; + + sb = allocsb(); + sb->sb_name = yytext; + sb->sb_len = yyleng; + sb->sb_hash = hash(yytext); + + if ((sym = search(sb)) != NULL && sym->s_keyw) { + freesb(sb); + return (keyw(sym)); + } + + sb->sb_sym = sym; + + if (sym != NULL) { + if (blklev < sym->s_blklev) + lerror("name() 1"); + sb->sb_name = sym->s_name; + sb->sb_len = strlen(sym->s_name); + tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME; + } else { + s = getblk(yyleng + 1); + (void)memcpy(s, yytext, yyleng + 1); + sb->sb_name = s; + sb->sb_len = yyleng; + tok = T_NAME; + } + + yylval.y_sb = sb; + return (tok); +} + +static sym_t * +search(sb) + sbuf_t *sb; +{ + sym_t *sym; + + for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) { + if (strcmp(sym->s_name, sb->sb_name) == 0) { + if (sym->s_keyw || sym->s_kind == symtyp) + return (sym); + } + } + + return (NULL); +} + +static int +keyw(sym) + sym_t *sym; +{ + int t; + + if ((t = (int)sym->s_value.v_quad) == T_SCLASS) { + yylval.y_scl = sym->s_scl; + } else if (t == T_TYPE || t == T_SOU) { + yylval.y_tspec = sym->s_tspec; + } else if (t == T_QUAL) { + yylval.y_tqual = sym->s_tqual; + } + return (t); +} + +/* + * Convert a string representing an integer into internal representation. + * The value is returned in yylval. icon() (and yylex()) returns T_CON. + */ +static int +icon(base) + int base; +{ + int l_suffix, u_suffix; + int len; + const char *cp; + char c, *eptr; + tspec_t typ; + u_long ul; + u_quad_t uq; + int ansiu; + static tspec_t contypes[2][3] = { + { INT, LONG, QUAD }, + { UINT, ULONG, UQUAD } + }; + + cp = yytext; + len = yyleng; + + /* skip 0x */ + if (base == 16) { + cp += 2; + len -= 2; + } + + /* read suffixes */ + l_suffix = u_suffix = 0; + for ( ; ; ) { + if ((c = cp[len - 1]) == 'l' || c == 'L') { + l_suffix++; + } else if (c == 'u' || c == 'U') { + u_suffix++; + } else { + break; + } + len--; + } + if (l_suffix > 2 || u_suffix > 1) { + /* malformed integer constant */ + warning(251); + if (l_suffix > 2) + l_suffix = 2; + if (u_suffix > 1) + u_suffix = 1; + } + if (tflag && u_suffix != 0) { + /* suffix U is illegal in traditional C */ + warning(97); + } + typ = contypes[u_suffix][l_suffix]; + + errno = 0; + if (l_suffix < 2) { + ul = strtoul(cp, &eptr, base); + } else { + uq = strtouq(cp, &eptr, base); + } + if (eptr != cp + len) + lerror("icon() 1"); + if (errno != 0) + /* integer constant out of range */ + warning(252); + + /* + * If the value is to big for the current type, we must choose + * another type. + */ + ansiu = 0; + switch (typ) { + case INT: + if (ul <= INT_MAX) { + /* ok */ + } else if (ul <= (unsigned)UINT_MAX && base != 10) { + typ = UINT; + } else if (ul <= LONG_MAX) { + typ = LONG; + } else { + typ = ULONG; + } + if (typ == UINT || typ == ULONG) { + if (tflag) { + typ = LONG; + } else if (!sflag) { + /* + * Remember that the constant is unsigned + * only in ANSI C + */ + ansiu = 1; + } + } + break; + case UINT: + if (ul > (u_int)UINT_MAX) + typ = ULONG; + break; + case LONG: + if (ul > LONG_MAX && !tflag) { + typ = ULONG; + if (!sflag) + ansiu = 1; + } + break; + case QUAD: + if (uq > QUAD_MAX && !tflag) { + typ = UQUAD; + if (!sflag) + ansiu = 1; + } + break; + /* LINTED (enumeration values not handled in switch) */ + } + + if (typ != QUAD && typ != UQUAD) { + if (isutyp(typ)) { + uq = ul; + } else { + uq = (quad_t)(long)ul; + } + } + + uq = (u_quad_t)xsign((quad_t)uq, typ, -1); + + (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; + yylval.y_val->v_ansiu = ansiu; + yylval.y_val->v_quad = (quad_t)uq; + + return (T_CON); +} + +/* + * Returns 1 if t is a signed type and the value is negative. + * + * len is the number of significant bits. If len is -1, len is set + * to the width of type t. + */ +int +sign(q, t, len) + quad_t q; + tspec_t t; + int len; +{ + if (t == PTR || isutyp(t)) + return (0); + return (msb(q, t, len)); +} + +int +msb(q, t, len) + quad_t q; + tspec_t t; + int len; +{ + if (len <= 0) + len = size(t); + return ((q & qbmasks[len - 1]) != 0); +} + +/* + * Extends the sign of q. + */ +quad_t +xsign(q, t, len) + quad_t q; + tspec_t t; + int len; +{ + if (len <= 0) + len = size(t); + + if (t == PTR || isutyp(t) || !sign(q, t, len)) { + q &= qlmasks[len]; + } else { + q |= qumasks[len]; + } + return (q); +} + +/* + * Convert a string representing a floating point value into its interal + * representation. Type and value are returned in yylval. fcon() + * (and yylex()) returns T_CON. + * XXX Currently it is not possible to convert constants of type + * long double which are greater then DBL_MAX. + */ +static int +fcon() +{ + const char *cp; + int len; + tspec_t typ; + char c, *eptr; + double d; + float f; + + cp = yytext; + len = yyleng; + + if ((c = cp[len - 1]) == 'f' || c == 'F') { + typ = FLOAT; + len--; + } else if (c == 'l' || c == 'L') { + typ = LDOUBLE; + len--; + } else { + typ = DOUBLE; + } + + if (tflag && typ != DOUBLE) { + /* suffixes F and L are illegal in traditional C */ + warning(98); + } + + errno = 0; + d = strtod(cp, &eptr); + if (eptr != cp + len) + lerror("fcon() 1"); + if (errno != 0) + /* floating-point constant out of range */ + warning(248); + + if (typ == FLOAT) { + f = (float)d; + if (isinf(f)) { + /* floating-point constant out of range */ + warning(248); + f = f > 0 ? FLT_MAX : -FLT_MAX; + } + } + + (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ; + if (typ == FLOAT) { + yylval.y_val->v_ldbl = f; + } else { + yylval.y_val->v_ldbl = d; + } + + return (T_CON); +} + +static int +operator(t, o) + int t; + op_t o; +{ + yylval.y_op = o; + return (t); +} + +/* + * Called if lex found a leading \'. + */ +static int +ccon() +{ + int n, val, c; + char cv; + + n = 0; + val = 0; + while ((c = getescc('\'')) >= 0) { + val = (val << CHAR_BIT) + c; + n++; + } + if (c == -2) { + /* unterminated character constant */ + error(253); + } else { + if (n > sizeof (int) || (n > 1 && (pflag || hflag))) { + /* too many characters in character constant */ + error(71); + } else if (n > 1) { + /* multi-character character constant */ + warning(294); + } else if (n == 0) { + /* empty character constant */ + error(73); + } + } + if (n == 1) { + cv = (char)val; + val = cv; + } + + yylval.y_val = xcalloc(1, sizeof (val_t)); + yylval.y_val->v_tspec = INT; + yylval.y_val->v_quad = val; + + return (T_CON); +} + +/* + * Called if lex found a leading L\' + */ +static int +wccon() +{ + static char buf[MB_LEN_MAX + 1]; + int i, c; + wchar_t wc; + + i = 0; + while ((c = getescc('\'')) >= 0) { + if (i < MB_CUR_MAX) + buf[i] = (char)c; + i++; + } + + wc = 0; + + if (c == -2) { + /* unterminated character constant */ + error(253); + } else if (c == 0) { + /* empty character constant */ + error(73); + } else { + if (i > MB_CUR_MAX) { + i = MB_CUR_MAX; + /* too many characters in character constant */ + error(71); + } else { + buf[i] = '\0'; + (void)mbtowc(NULL, NULL, 0); + if (mbtowc(&wc, buf, MB_CUR_MAX) < 0) + /* invalid multibyte character */ + error(291); + } + } + + yylval.y_val = xcalloc(1, sizeof (val_t)); + yylval.y_val->v_tspec = WCHAR; + yylval.y_val->v_quad = wc; + + return (T_CON); +} + +/* + * Read a character which is part of a character constant or of a string + * and handle escapes. + * + * The Argument is the character which delimits the character constant or + * string. + * + * Returns -1 if the end of the character constant or string is reached, + * -2 if the EOF is reached, and the charachter otherwise. + */ +static int +getescc(d) + int d; +{ + static int pbc = -1; + int n, c, v; + + if (pbc == -1) { + c = inpc(); + } else { + c = pbc; + pbc = -1; + } + if (c == d) + return (-1); + switch (c) { + case '\n': + /* newline in string or char constant */ + error(254); + return (-2); + case EOF: + return (-2); + case '\\': + switch (c = inpc()) { + case '"': + if (tflag && d == '\'') + /* \" inside character constant undef. ... */ + warning(262); + return ('"'); + case '\'': + return ('\''); + case '?': + if (tflag) + /* \? undefined in traditional C */ + warning(263); + return ('?'); + case '\\': + return ('\\'); + case 'a': + if (tflag) + /* \a undefined in traditional C */ + warning(81); +#ifdef __STDC__ + return ('\a'); +#else + return ('\007'); +#endif + case 'b': + return ('\b'); + case 'f': + return ('\f'); + case 'n': + return ('\n'); + case 'r': + return ('\r'); + case 't': + return ('\t'); + case 'v': + if (tflag) + /* \v undefined in traditional C */ + warning(264); +#ifdef __STDC__ + return ('\v'); +#else + return ('\013'); +#endif + case '8': case '9': + /* bad octal digit %c */ + warning(77, c); + /* FALLTHROUGH */ + case '0': case '1': case '2': case '3': + case '4': case '5': case '6': case '7': + n = 3; + v = 0; + do { + v = (v << 3) + (c - '0'); + c = inpc(); + } while (--n && isdigit(c) && (tflag || c <= '7')); + if (tflag && n > 0 && isdigit(c)) + /* bad octal digit %c */ + warning(77, c); + pbc = c; + if (v > UCHAR_MAX) { + /* character escape does not fit in char. */ + warning(76); + v &= CHAR_MASK; + } + return (v); + case 'x': + if (tflag) + /* \x undefined in traditional C */ + warning(82); + v = 0; + n = 0; + while ((c = inpc()) >= 0 && isxdigit(c)) { + c = isdigit(c) ? + c - '0' : toupper(c) - 'A' + 10; + v = (v << 4) + c; + if (n >= 0) { + if ((v & ~CHAR_MASK) != 0) { + /* overflow in hex escape */ + warning(75); + n = -1; + } else { + n++; + } + } + } + pbc = c; + if (n == 0) { + /* no hex digits follow \x */ + error(74); + } if (n == -1) { + v &= CHAR_MASK; + } + return (v); + case '\n': + return (getescc(d)); + case EOF: + return (-2); + default: + if (isprint(c)) { + /* dubious escape \%c */ + warning(79, c); + } else { + /* dubious escape \%o */ + warning(80, c); + } + } + } + return (c); +} + +/* + * Called for preprocessor directives. Currently implemented are: + * # lineno + * # lineno "filename" + */ +static void +directive() +{ + const char *cp, *fn; + char c, *eptr; + size_t fnl; + long ln; + static int first = 1; + + /* Go to first non-whitespace after # */ + for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ; + + if (!isdigit(c)) { + error: + /* undefined or invalid # directive */ + warning(255); + return; + } + ln = strtol(--cp, &eptr, 10); + if (cp == eptr) + goto error; + if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0') + goto error; + while ((c = *cp++) == ' ' || c == '\t') ; + if (c != '\0') { + if (c != '"') + goto error; + fn = cp; + while ((c = *cp) != '"' && c != '\0') + cp++; + if (c != '"') + goto error; + if ((fnl = cp++ - fn) > PATH_MAX) + goto error; + while ((c = *cp++) == ' ' || c == '\t') ; +#if 0 + if (c != '\0') + warning("extra character(s) after directive"); +#endif + curr_pos.p_file = fnnalloc(fn, fnl); + /* + * If this is the first directive, the name is the name + * of the C source file as specified at the command line. + * It is written to the output file. + */ + if (first) { + csrc_pos.p_file = curr_pos.p_file; + outsrc(curr_pos.p_file); + first = 0; + } + } + curr_pos.p_line = (int)ln - 1; + if (curr_pos.p_file == csrc_pos.p_file) + csrc_pos.p_line = (int)ln - 1; +} + +/* + * Handle lint comments. Following comments are currently understood: + * ARGSUSEDn + * CONSTCOND CONSTANTCOND CONSTANTCONDITION + * FALLTHRU FALLTHROUGH + * LINTLIBRARY + * LINTED NOSTRICT + * LONGLONG + * NOTREACHED + * PRINTFLIKEn + * PROTOLIB + * SCANFLIKEn + * VARARGSn + * If one of this comments is recognized, the arguments, if any, are + * parsed and a function which handles this comment is called. + */ +static void +comment() +{ + int c, lc; + static struct { + const char *keywd; + int arg; + void (*func) __P((int)); + } keywtab[] = { + { "ARGSUSED", 1, argsused }, + { "CONSTCOND", 0, constcond }, + { "CONSTANTCOND", 0, constcond }, + { "CONSTANTCONDITION", 0, constcond }, + { "FALLTHRU", 0, fallthru }, + { "FALLTHROUGH", 0, fallthru }, + { "LINTLIBRARY", 0, lintlib }, + { "LINTED", 0, linted }, + { "LONGLONG", 0, longlong }, + { "NOSTRICT", 0, linted }, + { "NOTREACHED", 0, notreach }, + { "PRINTFLIKE", 1, printflike }, + { "PROTOLIB", 1, protolib }, + { "SCANFLIKE", 1, scanflike }, + { "VARARGS", 1, varargs }, + }; + char keywd[32]; + char arg[32]; + int l, i, a; + int eoc; + + eoc = 0; + + /* Skip white spaces after the start of the comment */ + while ((c = inpc()) != EOF && isspace(c)) ; + + /* Read the potential keyword to keywd */ + l = 0; + while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) { + keywd[l++] = (char)c; + c = inpc(); + } + keywd[l] = '\0'; + + /* look for the keyword */ + for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) { + if (strcmp(keywtab[i].keywd, keywd) == 0) + break; + } + if (i == sizeof (keywtab) / sizeof (keywtab[0])) + goto skip_rest; + + /* skip white spaces after the keyword */ + while (c != EOF && isspace(c)) + c = inpc(); + + /* read the argument, if the keyword accepts one and there is one */ + l = 0; + if (keywtab[i].arg) { + while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) { + arg[l++] = (char)c; + c = inpc(); + } + } + arg[l] = '\0'; + a = l != 0 ? atoi(arg) : -1; + + /* skip white spaces after the argument */ + while (c != EOF && isspace(c)) + c = inpc(); + + if (c != '*' || (c = inpc()) != '/') { + if (keywtab[i].func != linted) + /* extra characters in lint comment */ + warning(257); + } else { + /* + * remember that we have already found the end of the + * comment + */ + eoc = 1; + } + + if (keywtab[i].func != NULL) + (*keywtab[i].func)(a); + + skip_rest: + while (!eoc) { + lc = c; + if ((c = inpc()) == EOF) { + /* unterminated comment */ + error(256); + break; + } + if (lc == '*' && c == '/') + eoc = 1; + } +} + +/* + * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND. + * clrwflgs() is called after function definitions and global and + * local declarations and definitions. It is also called between + * the controlling expression and the body of control statements + * (if, switch, for, while). + */ +void +clrwflgs() +{ + nowarn = 0; + quadflg = 0; + ccflg = 0; +} + +/* + * Strings are stored in a dynamically alloceted buffer and passed + * in yylval.y_xstrg to the parser. The parser or the routines called + * by the parser are responsible for freeing this buffer. + */ +static int +string() +{ + u_char *s; + int c; + size_t len, max; + strg_t *strg; + + s = xmalloc(max = 64); + + len = 0; + while ((c = getescc('"')) >= 0) { + /* +1 to reserve space for a trailing NUL character */ + if (len + 1 == max) + s = xrealloc(s, max *= 2); + s[len++] = (char)c; + } + s[len] = '\0'; + if (c == -2) + /* unterminated string constant */ + error(258); + + strg = xcalloc(1, sizeof (strg_t)); + strg->st_tspec = CHAR; + strg->st_len = len; + strg->st_cp = s; + + yylval.y_strg = strg; + return (T_STRING); +} + +static int +wcstrg() +{ + char *s; + int c, i, n, wi; + size_t len, max, wlen; + wchar_t *ws; + strg_t *strg; + + s = xmalloc(max = 64); + len = 0; + while ((c = getescc('"')) >= 0) { + /* +1 to save space for a trailing NUL character */ + if (len + 1 >= max) + s = xrealloc(s, max *= 2); + s[len++] = (char)c; + } + s[len] = '\0'; + if (c == -2) + /* unterminated string constant */ + error(258); + + /* get length of wide character string */ + (void)mblen(NULL, 0); + for (i = 0, wlen = 0; i < len; i += n, wlen++) { + if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) { + /* invalid multibyte character */ + error(291); + break; + } + if (n == 0) + n = 1; + } + + ws = xmalloc((wlen + 1) * sizeof (wchar_t)); + + /* convert from multibyte to wide char */ + (void)mbtowc(NULL, NULL, 0); + for (i = 0, wi = 0; i < len; i += n, wi++) { + if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1) + break; + if (n == 0) + n = 1; + } + ws[wi] = 0; + free(s); + + strg = xcalloc(1, sizeof (strg_t)); + strg->st_tspec = WCHAR; + strg->st_len = wlen; + strg->st_wcp = ws; + + yylval.y_strg = strg; + return (T_STRING); +} + +/* + * As noted above the scanner does not create new symbol table entries + * for symbols it cannot find in the symbol table. This is to avoid + * putting undeclared symbols into the symbol table if a syntax error + * occurs. + * + * getsym() is called as soon as it is probably ok to put the symbol to + * the symbol table. This does not mean that it is not possible that + * symbols are put to the symbol table which are than not completely + * declared due to syntax errors. To avoid too many problems in this + * case symbols get type int in getsym(). + * + * XXX calls to getsym() should be delayed until decl1*() is called + */ +sym_t * +getsym(sb) + sbuf_t *sb; +{ + dinfo_t *di; + char *s; + sym_t *sym; + + sym = sb->sb_sym; + + /* + * During member declaration it is possible that name() looked + * for symbols of type FVFT, although it should have looked for + * symbols of type FTAG. Same can happen for labels. Both cases + * are compensated here. + */ + if (symtyp == FMOS || symtyp == FLAB) { + if (sym == NULL || sym->s_kind == FVFT) + sym = search(sb); + } + + if (sym != NULL) { + if (sym->s_kind != symtyp) + lerror("storesym() 1"); + symtyp = FVFT; + freesb(sb); + return (sym); + } + + /* create a new symbol table entry */ + + /* labels must always be allocated at level 1 (outhermost block) */ + if (symtyp == FLAB) { + sym = getlblk(1, sizeof (sym_t)); + s = getlblk(1, sb->sb_len + 1); + (void)memcpy(s, sb->sb_name, sb->sb_len + 1); + sym->s_name = s; + sym->s_blklev = 1; + di = dcs; + while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL) + di = di->d_nxt; + if (di->d_ctx != AUTO) + lerror("storesym() 2"); + } else { + sym = getblk(sizeof (sym_t)); + sym->s_name = sb->sb_name; + sym->s_blklev = blklev; + di = dcs; + } + + STRUCT_ASSIGN(sym->s_dpos, curr_pos); + if ((sym->s_kind = symtyp) != FLAB) + sym->s_type = gettyp(INT); + + symtyp = FVFT; + + if ((sym->s_link = symtab[sb->sb_hash]) != NULL) + symtab[sb->sb_hash]->s_rlink = &sym->s_link; + (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash]; + + *di->d_ldlsym = sym; + di->d_ldlsym = &sym->s_dlnxt; + + freesb(sb); + return (sym); +} + +/* + * Remove a symbol forever from the symbol table. s_blklev + * is set to -1 to avoid that the symbol will later be put + * back to the symbol table. + */ +void +rmsym(sym) + sym_t *sym; +{ + if ((*sym->s_rlink = sym->s_link) != NULL) + sym->s_link->s_rlink = sym->s_rlink; + sym->s_blklev = -1; + sym->s_link = NULL; +} + +/* + * Remove a list of symbols declared at one level from the symbol + * table. + */ +void +rmsyms(syms) + sym_t *syms; +{ + sym_t *sym; + + for (sym = syms; sym != NULL; sym = sym->s_dlnxt) { + if (sym->s_blklev != -1) { + if ((*sym->s_rlink = sym->s_link) != NULL) + sym->s_link->s_rlink = sym->s_rlink; + sym->s_link = NULL; + sym->s_rlink = NULL; + } + } +} + +/* + * Put a symbol into the symbol table + */ +void +inssym(bl, sym) + int bl; + sym_t *sym; +{ + int h; + + h = hash(sym->s_name); + if ((sym->s_link = symtab[h]) != NULL) + symtab[h]->s_rlink = &sym->s_link; + (symtab[h] = sym)->s_rlink = &symtab[h]; + sym->s_blklev = bl; + if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev) + lerror("inssym()"); +} + +/* + * Called at level 0 after syntax errors + * Removes all symbols which are not declared at level 0 from the + * symbol table. Also frees all memory which is not associated with + * level 0. + */ +void +cleanup() +{ + sym_t *sym, *nsym; + int i; + + for (i = 0; i < HSHSIZ1; i++) { + for (sym = symtab[i]; sym != NULL; sym = nsym) { + nsym = sym->s_link; + if (sym->s_blklev >= 1) { + if ((*sym->s_rlink = nsym) != NULL) + nsym->s_rlink = sym->s_rlink; + } + } + } + + for (i = mblklev; i > 0; i--) + freelblk(i); +} + +/* + * Create a new symbol with the name of an existing symbol. + */ +sym_t * +pushdown(sym) + sym_t *sym; +{ + int h; + sym_t *nsym; + + h = hash(sym->s_name); + nsym = getblk(sizeof (sym_t)); + if (sym->s_blklev > blklev) + lerror("pushdown()"); + nsym->s_name = sym->s_name; + STRUCT_ASSIGN(nsym->s_dpos, curr_pos); + nsym->s_kind = sym->s_kind; + nsym->s_blklev = blklev; + + if ((nsym->s_link = symtab[h]) != NULL) + symtab[h]->s_rlink = &nsym->s_link; + (symtab[h] = nsym)->s_rlink = &symtab[h]; + + *dcs->d_ldlsym = nsym; + dcs->d_ldlsym = &nsym->s_dlnxt; + + return (nsym); +} + +/* + * Free any dynamically allocated memory referenced by + * the value stack or yylval. + * The type of information in yylval is described by tok. + */ +void +freeyyv(sp, tok) + void *sp; + int tok; +{ + if (tok == T_NAME || tok == T_TYPENAME) { + sbuf_t *sb = *(sbuf_t **)sp; + freesb(sb); + } else if (tok == T_CON) { + val_t *val = *(val_t **)sp; + free(val); + } else if (tok == T_STRING) { + strg_t *strg = *(strg_t **)sp; + if (strg->st_tspec == CHAR) { + free(strg->st_cp); + } else if (strg->st_tspec == WCHAR) { + free(strg->st_wcp); + } else { + lerror("fryylv() 1"); + } + free(strg); + } +} |