2 * TCC - Tiny C Compiler
4 * Copyright (c) 2001-2004 Fabrice Bellard
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
23 /********************************************************/
24 /* global variables */
26 ST_DATA
int tok_flags
;
27 ST_DATA
int parse_flags
;
29 ST_DATA
struct BufferedFile
*file
;
32 ST_DATA
const int *macro_ptr
;
33 ST_DATA CString tokcstr
; /* current parsed string, if any */
35 /* display benchmark infos */
36 ST_DATA
int total_lines
;
37 ST_DATA
int total_bytes
;
38 ST_DATA
int tok_ident
;
39 ST_DATA TokenSym
**table_ident
;
41 /* ------------------------------------------------------------------------- */
43 static TokenSym
*hash_ident
[TOK_HASH_SIZE
];
44 static char token_buf
[STRING_MAX_SIZE
+ 1];
45 static CString cstr_buf
;
46 static CString macro_equal_buf
;
47 static TokenString tokstr_buf
;
48 static unsigned char isidnum_table
[256 - CH_EOF
];
49 static int pp_debug_tok
, pp_debug_symv
;
52 static int pp_counter
;
53 static void tok_print(const char *msg
, const int *str
);
55 static struct TinyAlloc
*toksym_alloc
;
56 static struct TinyAlloc
*tokstr_alloc
;
57 static struct TinyAlloc
*cstr_alloc
;
59 static TokenString
*macro_stack
;
61 static const char tcc_keywords
[] =
62 #define DEF(id, str) str "\0"
67 /* WARNING: the content of this string encodes token numbers */
68 static const unsigned char tok_two_chars
[] =
70 "<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
71 "-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";
93 '#','#', TOK_TWOSHARPS
,
97 static void next_nomacro_spc(void);
99 ST_FUNC
void skip(int c
)
102 tcc_error("'%c' expected (got \"%s\")", c
, get_tok_str(tok
, &tokc
));
106 ST_FUNC
void expect(const char *msg
)
108 tcc_error("%s expected", msg
);
111 /* ------------------------------------------------------------------------- */
112 /* Custom allocator for tiny objects */
117 #define tal_free(al, p) tcc_free(p)
118 #define tal_realloc(al, p, size) tcc_realloc(p, size)
119 #define tal_new(a,b,c)
120 #define tal_delete(a)
122 #if !defined(MEM_DEBUG)
123 #define tal_free(al, p) tal_free_impl(al, p)
124 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size)
125 #define TAL_DEBUG_PARAMS
128 //#define TAL_INFO 1 /* collect and dump allocators stats */
129 #define tal_free(al, p) tal_free_impl(al, p, __FILE__, __LINE__)
130 #define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size, __FILE__, __LINE__)
131 #define TAL_DEBUG_PARAMS , const char *file, int line
132 #define TAL_DEBUG_FILE_LEN 40
135 #define TOKSYM_TAL_SIZE (768 * 1024) /* allocator for tiny TokenSym in table_ident */
136 #define TOKSTR_TAL_SIZE (768 * 1024) /* allocator for tiny TokenString instances */
137 #define CSTR_TAL_SIZE (256 * 1024) /* allocator for tiny CString instances */
138 #define TOKSYM_TAL_LIMIT 256 /* prefer unique limits to distinguish allocators debug msgs */
139 #define TOKSTR_TAL_LIMIT 128 /* 32 * sizeof(int) */
140 #define CSTR_TAL_LIMIT 1024
142 typedef struct TinyAlloc
{
148 struct TinyAlloc
*next
, *top
;
157 typedef struct tal_header_t
{
160 int line_num
; /* negative line_num used for double free check */
161 char file_name
[TAL_DEBUG_FILE_LEN
+ 1];
165 /* ------------------------------------------------------------------------- */
167 static TinyAlloc
*tal_new(TinyAlloc
**pal
, unsigned limit
, unsigned size
)
169 TinyAlloc
*al
= tcc_mallocz(sizeof(TinyAlloc
));
170 al
->p
= al
->buffer
= tcc_malloc(size
);
177 static void tal_delete(TinyAlloc
*al
)
185 fprintf(stderr
, "limit=%5d, size=%5g MB, nb_peak=%6d, nb_total=%8d, nb_missed=%6d, usage=%5.1f%%\n",
186 al
->limit
, al
->size
/ 1024.0 / 1024.0, al
->nb_peak
, al
->nb_total
, al
->nb_missed
,
187 (al
->peak_p
- al
->buffer
) * 100.0 / al
->size
);
190 if (al
->nb_allocs
> 0) {
192 fprintf(stderr
, "TAL_DEBUG: memory leak %d chunk(s) (limit= %d)\n",
193 al
->nb_allocs
, al
->limit
);
196 tal_header_t
*header
= (tal_header_t
*)p
;
197 if (header
->line_num
> 0) {
198 fprintf(stderr
, "%s:%d: chunk of %d bytes leaked\n",
199 header
->file_name
, header
->line_num
, header
->size
);
201 p
+= header
->size
+ sizeof(tal_header_t
);
209 tcc_free(al
->buffer
);
215 static void tal_free_impl(TinyAlloc
*al
, void *p TAL_DEBUG_PARAMS
)
220 if (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
) {
222 tal_header_t
*header
= (((tal_header_t
*)p
) - 1);
223 if (header
->line_num
< 0) {
224 fprintf(stderr
, "%s:%d: TAL_DEBUG: double frees chunk from\n",
226 fprintf(stderr
, "%s:%d: %d bytes\n",
227 header
->file_name
, (int)-header
->line_num
, (int)header
->size
);
229 header
->line_num
= -header
->line_num
;
234 } else if (al
->next
) {
242 static void *tal_realloc_impl(TinyAlloc
**pal
, void *p
, unsigned size TAL_DEBUG_PARAMS
)
244 tal_header_t
*header
;
247 unsigned adj_size
= (size
+ 3) & -4;
248 TinyAlloc
*al
= *pal
;
251 is_own
= (al
->buffer
<= (uint8_t *)p
&& (uint8_t *)p
< al
->buffer
+ al
->size
);
252 if ((!p
|| is_own
) && size
<= al
->limit
) {
253 if (al
->p
+ adj_size
+ sizeof(tal_header_t
) < al
->buffer
+ al
->size
) {
254 header
= (tal_header_t
*)al
->p
;
255 header
->size
= adj_size
;
257 { int ofs
= strlen(file
) - TAL_DEBUG_FILE_LEN
;
258 strncpy(header
->file_name
, file
+ (ofs
> 0 ? ofs
: 0), TAL_DEBUG_FILE_LEN
);
259 header
->file_name
[TAL_DEBUG_FILE_LEN
] = 0;
260 header
->line_num
= line
; }
262 ret
= al
->p
+ sizeof(tal_header_t
);
263 al
->p
+= adj_size
+ sizeof(tal_header_t
);
265 header
= (((tal_header_t
*)p
) - 1);
266 memcpy(ret
, p
, header
->size
);
268 header
->line_num
= -header
->line_num
;
274 if (al
->nb_peak
< al
->nb_allocs
)
275 al
->nb_peak
= al
->nb_allocs
;
276 if (al
->peak_p
< al
->p
)
283 ret
= tal_realloc(*pal
, 0, size
);
284 header
= (((tal_header_t
*)p
) - 1);
285 memcpy(ret
, p
, header
->size
);
287 header
->line_num
= -header
->line_num
;
294 TinyAlloc
*bottom
= al
, *next
= al
->top
? al
->top
: al
;
296 al
= tal_new(pal
, next
->limit
, next
->size
* 2);
304 ret
= tcc_malloc(size
);
305 header
= (((tal_header_t
*)p
) - 1);
306 memcpy(ret
, p
, header
->size
);
308 header
->line_num
= -header
->line_num
;
310 } else if (al
->next
) {
314 ret
= tcc_realloc(p
, size
);
323 /* ------------------------------------------------------------------------- */
324 /* CString handling */
325 static void cstr_realloc(CString
*cstr
, int new_size
)
329 size
= cstr
->size_allocated
;
331 size
= 8; /* no need to allocate a too small first string */
332 while (size
< new_size
)
334 cstr
->data
= tal_realloc(cstr_alloc
, cstr
->data
, size
);
335 cstr
->size_allocated
= size
;
339 ST_INLN
void cstr_ccat(CString
*cstr
, int ch
)
342 size
= cstr
->size
+ 1;
343 if (size
> cstr
->size_allocated
)
344 cstr_realloc(cstr
, size
);
345 ((unsigned char *)cstr
->data
)[size
- 1] = ch
;
349 ST_FUNC
void cstr_cat(CString
*cstr
, const char *str
, int len
)
353 len
= strlen(str
) + 1 + len
;
354 size
= cstr
->size
+ len
;
355 if (size
> cstr
->size_allocated
)
356 cstr_realloc(cstr
, size
);
357 memmove(((unsigned char *)cstr
->data
) + cstr
->size
, str
, len
);
361 /* add a wide char */
362 ST_FUNC
void cstr_wccat(CString
*cstr
, int ch
)
365 size
= cstr
->size
+ sizeof(nwchar_t
);
366 if (size
> cstr
->size_allocated
)
367 cstr_realloc(cstr
, size
);
368 *(nwchar_t
*)(((unsigned char *)cstr
->data
) + size
- sizeof(nwchar_t
)) = ch
;
372 ST_FUNC
void cstr_new(CString
*cstr
)
374 memset(cstr
, 0, sizeof(CString
));
377 /* free string and reset it to NULL */
378 ST_FUNC
void cstr_free(CString
*cstr
)
380 tal_free(cstr_alloc
, cstr
->data
);
384 /* reset string to empty */
385 ST_FUNC
void cstr_reset(CString
*cstr
)
391 static void add_char(CString
*cstr
, int c
)
393 if (c
== '\'' || c
== '\"' || c
== '\\') {
394 /* XXX: could be more precise if char or string */
395 cstr_ccat(cstr
, '\\');
397 if (c
>= 32 && c
<= 126) {
400 cstr_ccat(cstr
, '\\');
402 cstr_ccat(cstr
, 'n');
404 cstr_ccat(cstr
, '0' + ((c
>> 6) & 7));
405 cstr_ccat(cstr
, '0' + ((c
>> 3) & 7));
406 cstr_ccat(cstr
, '0' + (c
& 7));
411 /* ------------------------------------------------------------------------- */
412 /* allocate a new token */
413 static TokenSym
*tok_alloc_new(TokenSym
**pts
, const char *str
, int len
)
415 TokenSym
*ts
, **ptable
;
418 if (tok_ident
>= SYM_FIRST_ANOM
)
419 tcc_error("memory full (symbols)");
421 /* expand token table if needed */
422 i
= tok_ident
- TOK_IDENT
;
423 if ((i
% TOK_ALLOC_INCR
) == 0) {
424 ptable
= tcc_realloc(table_ident
, (i
+ TOK_ALLOC_INCR
) * sizeof(TokenSym
*));
425 table_ident
= ptable
;
428 ts
= tal_realloc(toksym_alloc
, 0, sizeof(TokenSym
) + len
);
430 ts
->tok
= tok_ident
++;
431 ts
->sym_define
= NULL
;
432 ts
->sym_label
= NULL
;
433 ts
->sym_struct
= NULL
;
434 ts
->sym_identifier
= NULL
;
436 ts
->hash_next
= NULL
;
437 memcpy(ts
->str
, str
, len
);
443 #define TOK_HASH_INIT 1
444 #define TOK_HASH_FUNC(h, c) ((h) + ((h) << 5) + ((h) >> 27) + (c))
447 /* find a token and add it if not found */
448 ST_FUNC TokenSym
*tok_alloc(const char *str
, int len
)
456 h
= TOK_HASH_FUNC(h
, ((unsigned char *)str
)[i
]);
457 h
&= (TOK_HASH_SIZE
- 1);
459 pts
= &hash_ident
[h
];
464 if (ts
->len
== len
&& !memcmp(ts
->str
, str
, len
))
466 pts
= &(ts
->hash_next
);
468 return tok_alloc_new(pts
, str
, len
);
471 /* XXX: buffer overflow */
472 /* XXX: float tokens */
473 ST_FUNC
const char *get_tok_str(int v
, CValue
*cv
)
478 cstr_reset(&cstr_buf
);
488 /* XXX: not quite exact, but only useful for testing */
490 sprintf(p
, "%u", (unsigned)cv
->i
);
492 sprintf(p
, "%llu", (unsigned long long)cv
->i
);
496 cstr_ccat(&cstr_buf
, 'L');
498 cstr_ccat(&cstr_buf
, '\'');
499 add_char(&cstr_buf
, cv
->i
);
500 cstr_ccat(&cstr_buf
, '\'');
501 cstr_ccat(&cstr_buf
, '\0');
505 return (char*)cv
->str
.data
;
507 cstr_ccat(&cstr_buf
, 'L');
509 cstr_ccat(&cstr_buf
, '\"');
511 len
= cv
->str
.size
- 1;
513 add_char(&cstr_buf
, ((unsigned char *)cv
->str
.data
)[i
]);
515 len
= (cv
->str
.size
/ sizeof(nwchar_t
)) - 1;
517 add_char(&cstr_buf
, ((nwchar_t
*)cv
->str
.data
)[i
]);
519 cstr_ccat(&cstr_buf
, '\"');
520 cstr_ccat(&cstr_buf
, '\0');
524 cstr_cat(&cstr_buf
, "<float>", 0);
527 cstr_cat(&cstr_buf
, "<double>", 0);
530 cstr_cat(&cstr_buf
, "<long double>", 0);
533 cstr_cat(&cstr_buf
, "<linenumber>", 0);
536 /* above tokens have value, the ones below don't */
544 return strcpy(p
, "...");
546 return strcpy(p
, "<<=");
548 return strcpy(p
, ">>=");
550 return strcpy(p
, "<eof>");
553 /* search in two bytes table */
554 const unsigned char *q
= tok_two_chars
;
560 return cstr_buf
.data
;
565 sprintf(cstr_buf
.data
, "<%02x>", v
);
566 return cstr_buf
.data
;
571 } else if (v
< tok_ident
) {
572 return table_ident
[v
- TOK_IDENT
]->str
;
573 } else if (v
>= SYM_FIRST_ANOM
) {
574 /* special name for anonymous symbol */
575 sprintf(p
, "L.%u", v
- SYM_FIRST_ANOM
);
577 /* should never happen */
582 return cstr_buf
.data
;
585 /* return the current character, handling end of block if necessary
587 ST_FUNC
int handle_eob(void)
589 BufferedFile
*bf
= file
;
592 /* only tries to read if really end of buffer */
593 if (bf
->buf_ptr
>= bf
->buf_end
) {
595 #if defined(PARSE_DEBUG)
600 len
= read(bf
->fd
, bf
->buffer
, len
);
607 bf
->buf_ptr
= bf
->buffer
;
608 bf
->buf_end
= bf
->buffer
+ len
;
609 *bf
->buf_end
= CH_EOB
;
611 if (bf
->buf_ptr
< bf
->buf_end
) {
612 return bf
->buf_ptr
[0];
614 bf
->buf_ptr
= bf
->buf_end
;
619 /* read next char from current input file and handle end of input buffer */
620 ST_INLN
void inp(void)
622 ch
= *(++(file
->buf_ptr
));
623 /* end of buffer/file handling */
628 /* handle '\[\r]\n' */
629 static int handle_stray_noerror(void)
636 } else if (ch
== '\r') {
650 static void handle_stray(void)
652 if (handle_stray_noerror())
653 tcc_error("stray '\\' in program");
656 /* skip the stray and handle the \\n case. Output an error if
657 incorrect char after the stray */
658 static int handle_stray1(uint8_t *p
)
663 if (p
>= file
->buf_end
) {
670 if (handle_stray_noerror()) {
671 if (!(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
672 tcc_error("stray '\\' in program");
673 *--file
->buf_ptr
= '\\';
680 /* handle just the EOB case, but not stray */
681 #define PEEKC_EOB(c, p)\
692 /* handle the complicated stray case */
698 c = handle_stray1(p);\
703 /* input with '\[\r]\n' handling. Note that this function cannot
704 handle other characters after '\', so you cannot call it inside
705 strings or comments */
706 ST_FUNC
void minp(void)
713 /* single line C++ comments */
714 static uint8_t *parse_line_comment(uint8_t *p
)
722 if (c
== '\n' || c
== CH_EOF
) {
724 } else if (c
== '\\') {
733 } else if (c
== '\r') {
751 ST_FUNC
uint8_t *parse_comment(uint8_t *p
)
760 if (c
== '\n' || c
== '*' || c
== '\\')
764 if (c
== '\n' || c
== '*' || c
== '\\')
768 /* now we can handle all the cases */
772 } else if (c
== '*') {
778 } else if (c
== '/') {
780 } else if (c
== '\\') {
785 tcc_error("unexpected end of file in comment");
787 /* skip '\[\r]\n', otherwise just skip the stray */
793 } else if (c
== '\r') {
810 /* stray, eob or eof */
815 tcc_error("unexpected end of file in comment");
816 } else if (c
== '\\') {
826 ST_FUNC
int set_idnum(int c
, int val
)
828 int prev
= isidnum_table
[c
- CH_EOF
];
829 isidnum_table
[c
- CH_EOF
] = val
;
835 static inline void skip_spaces(void)
837 while (isidnum_table
[ch
- CH_EOF
] & IS_SPC
)
841 static inline int check_space(int t
, int *spc
)
843 if (t
< 256 && (isidnum_table
[t
- CH_EOF
] & IS_SPC
)) {
852 /* parse a string without interpreting escapes */
853 static uint8_t *parse_pp_string(uint8_t *p
,
854 int sep
, CString
*str
)
862 } else if (c
== '\\') {
868 /* XXX: indicate line number of start of string */
869 tcc_error("missing terminating %c character", sep
);
870 } else if (c
== '\\') {
871 /* escape : just skip \[\r]\n */
876 } else if (c
== '\r') {
879 expect("'\n' after '\r'");
882 } else if (c
== CH_EOF
) {
883 goto unterminated_string
;
886 cstr_ccat(str
, '\\');
892 } else if (c
== '\n') {
895 } else if (c
== '\r') {
899 cstr_ccat(str
, '\r');
915 /* skip block of text until #else, #elif or #endif. skip also pairs of
917 static void preprocess_skip(void)
919 int a
, start_of_line
, c
, in_warn_or_error
;
926 in_warn_or_error
= 0;
947 } else if (c
== '\\') {
948 ch
= file
->buf_ptr
[0];
949 handle_stray_noerror();
956 if (in_warn_or_error
)
958 p
= parse_pp_string(p
, c
, NULL
);
962 if (in_warn_or_error
)
969 p
= parse_comment(p
);
970 } else if (ch
== '/') {
971 p
= parse_line_comment(p
);
981 (tok
== TOK_ELSE
|| tok
== TOK_ELIF
|| tok
== TOK_ENDIF
))
983 if (tok
== TOK_IF
|| tok
== TOK_IFDEF
|| tok
== TOK_IFNDEF
)
985 else if (tok
== TOK_ENDIF
)
987 else if( tok
== TOK_ERROR
|| tok
== TOK_WARNING
)
988 in_warn_or_error
= 1;
989 else if (tok
== TOK_LINEFEED
)
991 else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
992 p
= parse_line_comment(p
- 1);
993 } else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
994 p
= parse_line_comment(p
- 1);
1008 /* return the number of additional 'ints' necessary to store the
1010 static inline int tok_size(const int *p
)
1025 return 1 + ((sizeof(CString
) + ((CString
*)(p
+1))->size
+ 3) >> 2);
1028 return 1 + LONG_SIZE
/ 4;
1034 return 1 + LDOUBLE_SIZE
/ 4;
1041 /* token string handling */
1042 ST_INLN
void tok_str_new(TokenString
*s
)
1045 s
->len
= s
->lastlen
= 0;
1046 s
->allocated_len
= 0;
1047 s
->last_line_num
= -1;
1050 ST_FUNC TokenString
*tok_str_alloc(void)
1052 TokenString
*str
= tal_realloc(tokstr_alloc
, 0, sizeof *str
);
1057 ST_FUNC
int *tok_str_dup(TokenString
*s
)
1061 str
= tal_realloc(tokstr_alloc
, 0, s
->len
* sizeof(int));
1062 memcpy(str
, s
->str
, s
->len
* sizeof(int));
1066 ST_FUNC
void tok_str_free_str(int *str
)
1068 tal_free(tokstr_alloc
, str
);
1071 ST_FUNC
void tok_str_free(TokenString
*str
)
1073 tok_str_free_str(str
->str
);
1074 tal_free(tokstr_alloc
, str
);
1077 ST_FUNC
int *tok_str_realloc(TokenString
*s
, int new_size
)
1081 size
= s
->allocated_len
;
1084 while (size
< new_size
)
1086 if (size
> s
->allocated_len
) {
1087 str
= tal_realloc(tokstr_alloc
, s
->str
, size
* sizeof(int));
1088 s
->allocated_len
= size
;
1094 ST_FUNC
void tok_str_add(TokenString
*s
, int t
)
1100 if (len
>= s
->allocated_len
)
1101 str
= tok_str_realloc(s
, len
+ 1);
1106 ST_FUNC
void begin_macro(TokenString
*str
, int alloc
)
1109 str
->prev
= macro_stack
;
1110 str
->prev_ptr
= macro_ptr
;
1111 str
->save_line_num
= file
->line_num
;
1112 macro_ptr
= str
->str
;
1116 ST_FUNC
void end_macro(void)
1118 TokenString
*str
= macro_stack
;
1119 macro_stack
= str
->prev
;
1120 macro_ptr
= str
->prev_ptr
;
1121 file
->line_num
= str
->save_line_num
;
1122 if (str
->alloc
== 2) {
1123 str
->alloc
= 3; /* just mark as finished */
1129 static void tok_str_add2(TokenString
*s
, int t
, CValue
*cv
)
1133 len
= s
->lastlen
= s
->len
;
1136 /* allocate space for worst case */
1137 if (len
+ TOK_MAX_SIZE
>= s
->allocated_len
)
1138 str
= tok_str_realloc(s
, len
+ TOK_MAX_SIZE
+ 1);
1151 str
[len
++] = cv
->tab
[0];
1158 /* Insert the string into the int array. */
1160 1 + (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1161 if (len
+ nb_words
>= s
->allocated_len
)
1162 str
= tok_str_realloc(s
, len
+ nb_words
+ 1);
1163 str
[len
] = cv
->str
.size
;
1164 memcpy(&str
[len
+ 1], cv
->str
.data
, cv
->str
.size
);
1175 #if LDOUBLE_SIZE == 8
1178 str
[len
++] = cv
->tab
[0];
1179 str
[len
++] = cv
->tab
[1];
1181 #if LDOUBLE_SIZE == 12
1183 str
[len
++] = cv
->tab
[0];
1184 str
[len
++] = cv
->tab
[1];
1185 str
[len
++] = cv
->tab
[2];
1186 #elif LDOUBLE_SIZE == 16
1188 str
[len
++] = cv
->tab
[0];
1189 str
[len
++] = cv
->tab
[1];
1190 str
[len
++] = cv
->tab
[2];
1191 str
[len
++] = cv
->tab
[3];
1192 #elif LDOUBLE_SIZE != 8
1193 #error add long double size support
1202 /* add the current parse token in token string 's' */
1203 ST_FUNC
void tok_str_add_tok(TokenString
*s
)
1207 /* save line number info */
1208 if (file
->line_num
!= s
->last_line_num
) {
1209 s
->last_line_num
= file
->line_num
;
1210 cval
.i
= s
->last_line_num
;
1211 tok_str_add2(s
, TOK_LINENUM
, &cval
);
1213 tok_str_add2(s
, tok
, &tokc
);
1216 /* get a token from an integer array and increment pointer
1217 accordingly. we code it as a macro to avoid pointer aliasing. */
1218 static inline void TOK_GET(int *t
, const int **pp
, CValue
*cv
)
1238 cv
->i
= (unsigned)*p
++;
1247 cv
->str
.size
= *p
++;
1249 p
+= (cv
->str
.size
+ sizeof(int) - 1) / sizeof(int);
1261 #if LDOUBLE_SIZE == 16
1263 #elif LDOUBLE_SIZE == 12
1265 #elif LDOUBLE_SIZE == 8
1268 # error add long double size support
1281 static int macro_is_equal(const int *a
, const int *b
)
1290 /* first time preallocate macro_equal_buf, next time only reset position to start */
1291 cstr_reset(¯o_equal_buf
);
1292 TOK_GET(&t
, &a
, &cv
);
1293 cstr_cat(¯o_equal_buf
, get_tok_str(t
, &cv
), 0);
1294 TOK_GET(&t
, &b
, &cv
);
1295 if (strcmp(macro_equal_buf
.data
, get_tok_str(t
, &cv
)))
1301 /* defines handling */
1302 ST_INLN
void define_push(int v
, int macro_type
, int *str
, Sym
*first_arg
)
1307 s
= sym_push2(&define_stack
, v
, macro_type
, 0);
1309 s
->next
= first_arg
;
1310 table_ident
[v
- TOK_IDENT
]->sym_define
= s
;
1312 if (o
&& !macro_is_equal(o
->d
, s
->d
))
1313 tcc_warning("%s redefined", get_tok_str(v
, NULL
));
1316 /* undefined a define symbol. Its name is just set to zero */
1317 ST_FUNC
void define_undef(Sym
*s
)
1320 if (v
>= TOK_IDENT
&& v
< tok_ident
)
1321 table_ident
[v
- TOK_IDENT
]->sym_define
= NULL
;
1324 ST_INLN Sym
*define_find(int v
)
1327 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1329 return table_ident
[v
]->sym_define
;
1332 /* free define stack until top reaches 'b' */
1333 ST_FUNC
void free_defines(Sym
*b
)
1335 while (define_stack
!= b
) {
1336 Sym
*top
= define_stack
;
1337 define_stack
= top
->prev
;
1338 tok_str_free_str(top
->d
);
1343 /* restore remaining (-D or predefined) symbols if they were
1344 #undef'd in the file */
1347 if (v
>= TOK_IDENT
&& v
< tok_ident
) {
1348 Sym
**d
= &table_ident
[v
- TOK_IDENT
]->sym_define
;
1357 ST_FUNC Sym
*label_find(int v
)
1360 if ((unsigned)v
>= (unsigned)(tok_ident
- TOK_IDENT
))
1362 return table_ident
[v
]->sym_label
;
1365 ST_FUNC Sym
*label_push(Sym
**ptop
, int v
, int flags
)
1368 s
= sym_push2(ptop
, v
, 0, 0);
1370 ps
= &table_ident
[v
- TOK_IDENT
]->sym_label
;
1371 if (ptop
== &global_label_stack
) {
1372 /* modify the top most local identifier, so that
1373 sym_identifier will point to 's' when popped */
1375 ps
= &(*ps
)->prev_tok
;
1382 /* pop labels until element last is reached. Look if any labels are
1383 undefined. Define symbols if '&&label' was used. */
1384 ST_FUNC
void label_pop(Sym
**ptop
, Sym
*slast
, int keep
)
1387 for(s
= *ptop
; s
!= slast
; s
= s1
) {
1389 if (s
->r
== LABEL_DECLARED
) {
1390 tcc_warning("label '%s' declared but not used", get_tok_str(s
->v
, NULL
));
1391 } else if (s
->r
== LABEL_FORWARD
) {
1392 tcc_error("label '%s' used but not defined",
1393 get_tok_str(s
->v
, NULL
));
1396 /* define corresponding symbol. A size of
1398 put_extern_sym(s
, cur_text_section
, s
->jnext
, 1);
1402 table_ident
[s
->v
- TOK_IDENT
]->sym_label
= s
->prev_tok
;
1410 /* fake the nth "#if defined test_..." for tcc -dt -run */
1411 static void maybe_run_test(TCCState
*s
)
1414 if (s
->include_stack_ptr
!= s
->include_stack
)
1416 p
= get_tok_str(tok
, NULL
);
1417 if (0 != memcmp(p
, "test_", 5))
1419 if (0 != --s
->run_test
)
1421 fprintf(s
->ppfp
, "\n[%s]\n" + !(s
->dflag
& 32), p
), fflush(s
->ppfp
);
1422 define_push(tok
, MACRO_OBJ
, NULL
, NULL
);
1425 /* eval an expression for #if/#elif */
1426 static int expr_preprocess(void)
1431 str
= tok_str_alloc();
1433 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1434 next(); /* do macro subst */
1435 if (tok
== TOK_DEFINED
) {
1440 if (tok
< TOK_IDENT
)
1441 expect("identifier");
1442 if (tcc_state
->run_test
)
1443 maybe_run_test(tcc_state
);
1444 c
= define_find(tok
) != 0;
1452 } else if (tok
>= TOK_IDENT
) {
1453 /* if undefined macro */
1457 tok_str_add_tok(str
);
1460 tok_str_add(str
, -1); /* simulate end of file */
1461 tok_str_add(str
, 0);
1462 /* now evaluate C constant expression */
1463 begin_macro(str
, 1);
1471 /* parse after #define */
1472 ST_FUNC
void parse_define(void)
1474 Sym
*s
, *first
, **ps
;
1475 int v
, t
, varg
, is_vaargs
, spc
;
1476 int saved_parse_flags
= parse_flags
;
1479 if (v
< TOK_IDENT
|| v
== TOK_DEFINED
)
1480 tcc_error("invalid macro name '%s'", get_tok_str(tok
, &tokc
));
1481 /* XXX: should check if same macro (ANSI) */
1484 /* We have to parse the whole define as if not in asm mode, in particular
1485 no line comment with '#' must be ignored. Also for function
1486 macros the argument list must be parsed without '.' being an ID
1488 parse_flags
= ((parse_flags
& ~PARSE_FLAG_ASM_FILE
) | PARSE_FLAG_SPACES
);
1489 /* '(' must be just after macro definition for MACRO_FUNC */
1492 int dotid
= set_idnum('.', 0);
1495 if (tok
!= ')') for (;;) {
1499 if (varg
== TOK_DOTS
) {
1500 varg
= TOK___VA_ARGS__
;
1502 } else if (tok
== TOK_DOTS
&& gnu_ext
) {
1506 if (varg
< TOK_IDENT
)
1508 tcc_error("bad macro parameter list");
1509 s
= sym_push2(&define_stack
, varg
| SYM_FIELD
, is_vaargs
, 0);
1514 if (tok
!= ',' || is_vaargs
)
1520 set_idnum('.', dotid
);
1525 parse_flags
|= PARSE_FLAG_ACCEPT_STRAYS
| PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
;
1526 /* The body of a macro definition should be parsed such that identifiers
1527 are parsed like the file mode determines (i.e. with '.' being an
1528 ID character in asm mode). But '#' should be retained instead of
1529 regarded as line comment leader, so still don't set ASM_FILE
1531 while (tok
!= TOK_LINEFEED
&& tok
!= TOK_EOF
) {
1532 /* remove spaces around ## and after '#' */
1533 if (TOK_TWOSHARPS
== tok
) {
1540 } else if ('#' == tok
) {
1542 } else if (check_space(tok
, &spc
)) {
1545 tok_str_add2(&tokstr_buf
, tok
, &tokc
);
1550 parse_flags
= saved_parse_flags
;
1552 --tokstr_buf
.len
; /* remove trailing space */
1553 tok_str_add(&tokstr_buf
, 0);
1556 tcc_error("'##' cannot appear at either end of macro");
1557 define_push(v
, t
, tok_str_dup(&tokstr_buf
), first
);
1560 static CachedInclude
*search_cached_include(TCCState
*s1
, const char *filename
, int add
)
1562 const unsigned char *s
;
1568 s
= (unsigned char *) filename
;
1571 h
= TOK_HASH_FUNC(h
, toup(*s
));
1573 h
= TOK_HASH_FUNC(h
, *s
);
1577 h
&= (CACHED_INCLUDES_HASH_SIZE
- 1);
1579 i
= s1
->cached_includes_hash
[h
];
1583 e
= s1
->cached_includes
[i
- 1];
1584 if (0 == PATHCMP(e
->filename
, filename
))
1591 e
= tcc_malloc(sizeof(CachedInclude
) + strlen(filename
));
1592 strcpy(e
->filename
, filename
);
1593 e
->ifndef_macro
= e
->once
= 0;
1594 dynarray_add(&s1
->cached_includes
, &s1
->nb_cached_includes
, e
);
1595 /* add in hash table */
1596 e
->hash_next
= s1
->cached_includes_hash
[h
];
1597 s1
->cached_includes_hash
[h
] = s1
->nb_cached_includes
;
1599 printf("adding cached '%s'\n", filename
);
1604 static void pragma_parse(TCCState
*s1
)
1607 if (tok
== TOK_push_macro
|| tok
== TOK_pop_macro
) {
1611 if (next(), tok
!= '(')
1613 if (next(), tok
!= TOK_STR
)
1615 v
= tok_alloc(tokc
.str
.data
, tokc
.str
.size
- 1)->tok
;
1616 if (next(), tok
!= ')')
1618 if (t
== TOK_push_macro
) {
1619 while (NULL
== (s
= define_find(v
)))
1620 define_push(v
, 0, NULL
, NULL
);
1621 s
->type
.ref
= s
; /* set push boundary */
1623 for (s
= define_stack
; s
; s
= s
->prev
)
1624 if (s
->v
== v
&& s
->type
.ref
== s
) {
1630 table_ident
[v
- TOK_IDENT
]->sym_define
= s
->d
? s
: NULL
;
1632 tcc_warning("unbalanced #pragma pop_macro");
1633 pp_debug_tok
= t
, pp_debug_symv
= v
;
1635 } else if (tok
== TOK_once
) {
1636 search_cached_include(s1
, file
->filename
, 1)->once
= pp_once
;
1638 } else if (s1
->output_type
== TCC_OUTPUT_PREPROCESS
) {
1639 /* tcc -E: keep pragmas below unchanged */
1641 unget_tok(TOK_PRAGMA
);
1643 unget_tok(TOK_LINEFEED
);
1645 } else if (tok
== TOK_pack
) {
1647 #pragma pack(1) // set
1648 #pragma pack() // reset to default
1649 #pragma pack(push,1) // push & set
1650 #pragma pack(pop) // restore previous */
1653 if (tok
== TOK_ASM_pop
) {
1655 if (s1
->pack_stack_ptr
<= s1
->pack_stack
) {
1657 tcc_error("out of pack stack");
1659 s1
->pack_stack_ptr
--;
1663 if (tok
== TOK_ASM_push
) {
1665 if (s1
->pack_stack_ptr
>= s1
->pack_stack
+ PACK_STACK_SIZE
- 1)
1667 s1
->pack_stack_ptr
++;
1670 if (tok
!= TOK_CINT
)
1673 if (val
< 1 || val
> 16 || (val
& (val
- 1)) != 0)
1677 *s1
->pack_stack_ptr
= val
;
1682 } else if (tok
== TOK_comment
) {
1691 p
= tcc_strdup((char *)tokc
.str
.data
);
1696 dynarray_add(&s1
->pragma_libs
, &s1
->nb_pragma_libs
, p
);
1698 if (t
== TOK_option
)
1699 tcc_set_options(s1
, p
);
1703 } else if (s1
->warn_unsupported
) {
1704 tcc_warning("#pragma %s is ignored", get_tok_str(tok
, &tokc
));
1709 tcc_error("malformed #pragma directive");
1713 /* is_bof is true if first non space token at beginning of file */
1714 ST_FUNC
void preprocess(int is_bof
)
1716 TCCState
*s1
= tcc_state
;
1717 int i
, c
, n
, saved_parse_flags
;
1721 saved_parse_flags
= parse_flags
;
1722 parse_flags
= PARSE_FLAG_PREPROCESS
1723 | PARSE_FLAG_TOK_NUM
1724 | PARSE_FLAG_TOK_STR
1725 | PARSE_FLAG_LINEFEED
1726 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
1735 pp_debug_symv
= tok
;
1741 pp_debug_symv
= tok
;
1742 s
= define_find(tok
);
1743 /* undefine symbol by putting an invalid name */
1748 case TOK_INCLUDE_NEXT
:
1749 ch
= file
->buf_ptr
[0];
1750 /* XXX: incorrect if comments : use next_nomacro with a special mode */
1755 } else if (ch
== '\"') {
1760 while (ch
!= c
&& ch
!= '\n' && ch
!= CH_EOF
) {
1761 if ((q
- buf
) < sizeof(buf
) - 1)
1764 if (handle_stray_noerror() == 0)
1772 /* eat all spaces and comments after include */
1773 /* XXX: slightly incorrect */
1774 while (ch1
!= '\n' && ch1
!= CH_EOF
)
1779 /* computed #include : concatenate everything up to linefeed,
1780 the result must be one of the two accepted forms.
1781 Don't convert pp-tokens to tokens here. */
1782 parse_flags
= (PARSE_FLAG_PREPROCESS
1783 | PARSE_FLAG_LINEFEED
1784 | (parse_flags
& PARSE_FLAG_ASM_FILE
));
1787 while (tok
!= TOK_LINEFEED
) {
1788 pstrcat(buf
, sizeof(buf
), get_tok_str(tok
, &tokc
));
1792 /* check syntax and remove '<>|""' */
1793 if ((len
< 2 || ((buf
[0] != '"' || buf
[len
-1] != '"') &&
1794 (buf
[0] != '<' || buf
[len
-1] != '>'))))
1795 tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
1797 memmove(buf
, buf
+ 1, len
- 2);
1798 buf
[len
- 2] = '\0';
1801 if (s1
->include_stack_ptr
>= s1
->include_stack
+ INCLUDE_STACK_SIZE
)
1802 tcc_error("#include recursion too deep");
1803 /* store current file in stack, but increment stack later below */
1804 *s1
->include_stack_ptr
= file
;
1805 i
= tok
== TOK_INCLUDE_NEXT
? file
->include_next_index
: 0;
1806 n
= 2 + s1
->nb_include_paths
+ s1
->nb_sysinclude_paths
;
1807 for (; i
< n
; ++i
) {
1808 char buf1
[sizeof file
->filename
];
1813 /* check absolute include path */
1814 if (!IS_ABSPATH(buf
))
1818 } else if (i
== 1) {
1819 /* search in file's dir if "header.h" */
1822 /* https://savannah.nongnu.org/bugs/index.php?50847 */
1823 path
= file
->true_filename
;
1824 pstrncpy(buf1
, path
, tcc_basename(path
) - path
);
1827 /* search in all the include paths */
1828 int j
= i
- 2, k
= j
- s1
->nb_include_paths
;
1829 path
= k
< 0 ? s1
->include_paths
[j
] : s1
->sysinclude_paths
[k
];
1830 pstrcpy(buf1
, sizeof(buf1
), path
);
1831 pstrcat(buf1
, sizeof(buf1
), "/");
1834 pstrcat(buf1
, sizeof(buf1
), buf
);
1835 e
= search_cached_include(s1
, buf1
, 0);
1836 if (e
&& (define_find(e
->ifndef_macro
) || e
->once
== pp_once
)) {
1837 /* no need to parse the include because the 'ifndef macro'
1838 is defined (or had #pragma once) */
1840 printf("%s: skipping cached %s\n", file
->filename
, buf1
);
1845 if (tcc_open(s1
, buf1
) < 0)
1848 file
->include_next_index
= i
+ 1;
1850 printf("%s: including %s\n", file
->prev
->filename
, file
->filename
);
1852 /* update target deps */
1853 dynarray_add(&s1
->target_deps
, &s1
->nb_target_deps
,
1855 /* push current file in stack */
1856 ++s1
->include_stack_ptr
;
1857 /* add include file debug info */
1859 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1860 tok_flags
|= TOK_FLAG_BOF
| TOK_FLAG_BOL
;
1861 ch
= file
->buf_ptr
[0];
1864 tcc_error("include file '%s' not found", buf
);
1871 c
= expr_preprocess();
1877 if (tok
< TOK_IDENT
)
1878 tcc_error("invalid argument for '#if%sdef'", c
? "n" : "");
1882 printf("#ifndef %s\n", get_tok_str(tok
, NULL
));
1884 file
->ifndef_macro
= tok
;
1887 c
= (define_find(tok
) != 0) ^ c
;
1889 if (s1
->ifdef_stack_ptr
>= s1
->ifdef_stack
+ IFDEF_STACK_SIZE
)
1890 tcc_error("memory full (ifdef)");
1891 *s1
->ifdef_stack_ptr
++ = c
;
1894 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1895 tcc_error("#else without matching #if");
1896 if (s1
->ifdef_stack_ptr
[-1] & 2)
1897 tcc_error("#else after #else");
1898 c
= (s1
->ifdef_stack_ptr
[-1] ^= 3);
1901 if (s1
->ifdef_stack_ptr
== s1
->ifdef_stack
)
1902 tcc_error("#elif without matching #if");
1903 c
= s1
->ifdef_stack_ptr
[-1];
1905 tcc_error("#elif after #else");
1906 /* last #if/#elif expression was true: we skip */
1910 c
= expr_preprocess();
1911 s1
->ifdef_stack_ptr
[-1] = c
;
1914 if (s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
+ 1)
1915 file
->ifndef_macro
= 0;
1924 if (s1
->ifdef_stack_ptr
<= file
->ifdef_stack_ptr
)
1925 tcc_error("#endif without matching #if");
1926 s1
->ifdef_stack_ptr
--;
1927 /* '#ifndef macro' was at the start of file. Now we check if
1928 an '#endif' is exactly at the end of file */
1929 if (file
->ifndef_macro
&&
1930 s1
->ifdef_stack_ptr
== file
->ifdef_stack_ptr
) {
1931 file
->ifndef_macro_saved
= file
->ifndef_macro
;
1932 /* need to set to zero to avoid false matches if another
1933 #ifndef at middle of file */
1934 file
->ifndef_macro
= 0;
1935 while (tok
!= TOK_LINEFEED
)
1937 tok_flags
|= TOK_FLAG_ENDIF
;
1942 n
= strtoul((char*)tokc
.str
.data
, &q
, 10);
1946 if (tok
!= TOK_CINT
)
1948 tcc_error("wrong #line format");
1952 if (tok
!= TOK_LINEFEED
) {
1953 if (tok
== TOK_STR
) {
1954 if (file
->true_filename
== file
->filename
)
1955 file
->true_filename
= tcc_strdup(file
->filename
);
1956 pstrcpy(file
->filename
, sizeof(file
->filename
), (char *)tokc
.str
.data
);
1957 } else if (parse_flags
& PARSE_FLAG_ASM_FILE
)
1964 total_lines
+= file
->line_num
- n
;
1967 put_stabs(file
->filename
, N_BINCL
, 0, 0, 0);
1972 ch
= file
->buf_ptr
[0];
1975 while (ch
!= '\n' && ch
!= CH_EOF
) {
1976 if ((q
- buf
) < sizeof(buf
) - 1)
1979 if (handle_stray_noerror() == 0)
1986 tcc_error("#error %s", buf
);
1988 tcc_warning("#warning %s", buf
);
1996 /* ignore gas line comment in an 'S' file. */
1997 if (saved_parse_flags
& PARSE_FLAG_ASM_FILE
)
1999 if (tok
== '!' && is_bof
)
2000 /* '!' is ignored at beginning to allow C scripts. */
2002 tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok
, &tokc
));
2004 file
->buf_ptr
= parse_line_comment(file
->buf_ptr
- 1);
2007 /* ignore other preprocess commands or #! for C scripts */
2008 while (tok
!= TOK_LINEFEED
)
2011 parse_flags
= saved_parse_flags
;
2014 /* evaluate escape codes in a string. */
2015 static void parse_escape_string(CString
*outstr
, const uint8_t *buf
, int is_long
)
2030 case '0': case '1': case '2': case '3':
2031 case '4': case '5': case '6': case '7':
2032 /* at most three octal digits */
2037 n
= n
* 8 + c
- '0';
2041 n
= n
* 8 + c
- '0';
2046 goto add_char_nonext
;
2054 if (c
>= 'a' && c
<= 'f')
2056 else if (c
>= 'A' && c
<= 'F')
2066 goto add_char_nonext
;
2090 goto invalid_escape
;
2100 if (c
>= '!' && c
<= '~')
2101 tcc_warning("unknown escape sequence: \'\\%c\'", c
);
2103 tcc_warning("unknown escape sequence: \'\\x%x\'", c
);
2106 } else if (is_long
&& c
>= 0x80) {
2107 /* assume we are processing UTF-8 sequence */
2108 /* reference: The Unicode Standard, Version 10.0, ch3.9 */
2110 int cont
; /* count of continuation bytes */
2111 int skip
; /* how many bytes should skip when error occurred */
2114 /* decode leading byte */
2116 skip
= 1; goto invalid_utf8_sequence
;
2117 } else if (c
<= 0xDF) {
2118 cont
= 1; n
= c
& 0x1f;
2119 } else if (c
<= 0xEF) {
2120 cont
= 2; n
= c
& 0xf;
2121 } else if (c
<= 0xF4) {
2122 cont
= 3; n
= c
& 0x7;
2124 skip
= 1; goto invalid_utf8_sequence
;
2127 /* decode continuation bytes */
2128 for (i
= 1; i
<= cont
; i
++) {
2129 int l
= 0x80, h
= 0xBF;
2131 /* adjust limit for second byte */
2134 case 0xE0: l
= 0xA0; break;
2135 case 0xED: h
= 0x9F; break;
2136 case 0xF0: l
= 0x90; break;
2137 case 0xF4: h
= 0x8F; break;
2141 if (p
[i
] < l
|| p
[i
] > h
) {
2142 skip
= i
; goto invalid_utf8_sequence
;
2145 n
= (n
<< 6) | (p
[i
] & 0x3f);
2148 /* advance pointer */
2151 goto add_char_nonext
;
2153 /* error handling */
2154 invalid_utf8_sequence
:
2155 tcc_warning("ill-formed UTF-8 subsequence starting with: \'\\x%x\'", c
);
2158 goto add_char_nonext
;
2164 cstr_ccat(outstr
, c
);
2166 #ifdef TCC_TARGET_PE
2167 /* store as UTF-16 */
2169 cstr_wccat(outstr
, c
);
2172 cstr_wccat(outstr
, (c
>> 10) + 0xD800);
2173 cstr_wccat(outstr
, (c
& 0x3FF) + 0xDC00);
2176 cstr_wccat(outstr
, c
);
2180 /* add a trailing '\0' */
2182 cstr_ccat(outstr
, '\0');
2184 cstr_wccat(outstr
, '\0');
2187 static void parse_string(const char *s
, int len
)
2189 uint8_t buf
[1000], *p
= buf
;
2192 if ((is_long
= *s
== 'L'))
2196 if (len
>= sizeof buf
)
2197 p
= tcc_malloc(len
+ 1);
2201 cstr_reset(&tokcstr
);
2202 parse_escape_string(&tokcstr
, p
, is_long
);
2207 int char_size
, i
, n
, c
;
2208 /* XXX: make it portable */
2210 tok
= TOK_CCHAR
, char_size
= 1;
2212 tok
= TOK_LCHAR
, char_size
= sizeof(nwchar_t
);
2213 n
= tokcstr
.size
/ char_size
- 1;
2215 tcc_error("empty character constant");
2217 tcc_warning("multi-character character constant");
2218 for (c
= i
= 0; i
< n
; ++i
) {
2220 c
= ((nwchar_t
*)tokcstr
.data
)[i
];
2222 c
= (c
<< 8) | ((char *)tokcstr
.data
)[i
];
2226 tokc
.str
.size
= tokcstr
.size
;
2227 tokc
.str
.data
= tokcstr
.data
;
2235 /* we use 64 bit numbers */
2238 /* bn = (bn << shift) | or_val */
2239 static void bn_lshift(unsigned int *bn
, int shift
, int or_val
)
2243 for(i
=0;i
<BN_SIZE
;i
++) {
2245 bn
[i
] = (v
<< shift
) | or_val
;
2246 or_val
= v
>> (32 - shift
);
2250 static void bn_zero(unsigned int *bn
)
2253 for(i
=0;i
<BN_SIZE
;i
++) {
2258 /* parse number in null terminated string 'p' and return it in the
2260 static void parse_number(const char *p
)
2262 int b
, t
, shift
, frac_bits
, s
, exp_val
, ch
;
2264 unsigned int bn
[BN_SIZE
];
2275 goto float_frac_parse
;
2276 } else if (t
== '0') {
2277 if (ch
== 'x' || ch
== 'X') {
2281 } else if (tcc_ext
&& (ch
== 'b' || ch
== 'B')) {
2287 /* parse all digits. cannot check octal numbers at this stage
2288 because of floating point constants */
2290 if (ch
>= 'a' && ch
<= 'f')
2292 else if (ch
>= 'A' && ch
<= 'F')
2300 if (q
>= token_buf
+ STRING_MAX_SIZE
) {
2302 tcc_error("number too long");
2308 ((ch
== 'e' || ch
== 'E') && b
== 10) ||
2309 ((ch
== 'p' || ch
== 'P') && (b
== 16 || b
== 2))) {
2311 /* NOTE: strtox should support that for hexa numbers, but
2312 non ISOC99 libcs do not support it, so we prefer to do
2314 /* hexadecimal or binary floats */
2315 /* XXX: handle overflows */
2327 } else if (t
>= 'a') {
2329 } else if (t
>= 'A') {
2334 bn_lshift(bn
, shift
, t
);
2341 if (t
>= 'a' && t
<= 'f') {
2343 } else if (t
>= 'A' && t
<= 'F') {
2345 } else if (t
>= '0' && t
<= '9') {
2351 tcc_error("invalid digit");
2352 bn_lshift(bn
, shift
, t
);
2357 if (ch
!= 'p' && ch
!= 'P')
2364 } else if (ch
== '-') {
2368 if (ch
< '0' || ch
> '9')
2369 expect("exponent digits");
2370 while (ch
>= '0' && ch
<= '9') {
2371 exp_val
= exp_val
* 10 + ch
- '0';
2374 exp_val
= exp_val
* s
;
2376 /* now we can generate the number */
2377 /* XXX: should patch directly float number */
2378 d
= (double)bn
[1] * 4294967296.0 + (double)bn
[0];
2379 d
= ldexp(d
, exp_val
- frac_bits
);
2384 /* float : should handle overflow */
2386 } else if (t
== 'L') {
2388 #ifdef TCC_TARGET_PE
2393 /* XXX: not large enough */
2394 tokc
.ld
= (long double)d
;
2401 /* decimal floats */
2403 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2408 while (ch
>= '0' && ch
<= '9') {
2409 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2415 if (ch
== 'e' || ch
== 'E') {
2416 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2420 if (ch
== '-' || ch
== '+') {
2421 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2426 if (ch
< '0' || ch
> '9')
2427 expect("exponent digits");
2428 while (ch
>= '0' && ch
<= '9') {
2429 if (q
>= token_buf
+ STRING_MAX_SIZE
)
2441 tokc
.f
= strtof(token_buf
, NULL
);
2442 } else if (t
== 'L') {
2444 #ifdef TCC_TARGET_PE
2446 tokc
.d
= strtod(token_buf
, NULL
);
2449 tokc
.ld
= strtold(token_buf
, NULL
);
2453 tokc
.d
= strtod(token_buf
, NULL
);
2457 unsigned long long n
, n1
;
2458 int lcount
, ucount
, ov
= 0;
2461 /* integer number */
2464 if (b
== 10 && *q
== '0') {
2471 /* no need for checks except for base 10 / 8 errors */
2481 tcc_error("invalid digit");
2484 /* detect overflow */
2485 if (n1
>= 0x1000000000000000ULL
&& n
/ b
!= n1
)
2489 /* Determine the characteristics (unsigned and/or 64bit) the type of
2490 the constant must have according to the constant suffix(es) */
2491 lcount
= ucount
= 0;
2497 tcc_error("three 'l's in integer constant");
2498 if (lcount
&& *(p
- 1) != ch
)
2499 tcc_error("incorrect integer suffix: %s", p1
);
2502 } else if (t
== 'U') {
2504 tcc_error("two 'u's in integer constant");
2512 /* Determine if it needs 64 bits and/or unsigned in order to fit */
2513 if (ucount
== 0 && b
== 10) {
2514 if (lcount
<= (LONG_SIZE
== 4)) {
2515 if (n
>= 0x80000000U
)
2516 lcount
= (LONG_SIZE
== 4) + 1;
2518 if (n
>= 0x8000000000000000ULL
)
2521 if (lcount
<= (LONG_SIZE
== 4)) {
2522 if (n
>= 0x100000000ULL
)
2523 lcount
= (LONG_SIZE
== 4) + 1;
2524 else if (n
>= 0x80000000U
)
2527 if (n
>= 0x8000000000000000ULL
)
2532 tcc_warning("integer constant overflow");
2541 ++tok
; /* TOK_CU... */
2545 tcc_error("invalid number\n");
2549 #define PARSE2(c1, tok1, c2, tok2) \
2560 /* return next token without macro substitution */
2561 static inline void next_nomacro1(void)
2563 int t
, c
, is_long
, len
;
2576 if (parse_flags
& PARSE_FLAG_SPACES
)
2577 goto keep_tok_flags
;
2578 while (isidnum_table
[*p
- CH_EOF
] & IS_SPC
)
2587 /* first look if it is in fact an end of buffer */
2588 c
= handle_stray1(p
);
2595 TCCState
*s1
= tcc_state
;
2596 if ((parse_flags
& PARSE_FLAG_LINEFEED
)
2597 && !(tok_flags
& TOK_FLAG_EOF
)) {
2598 tok_flags
|= TOK_FLAG_EOF
;
2600 goto keep_tok_flags
;
2601 } else if (!(parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2603 } else if (s1
->ifdef_stack_ptr
!= file
->ifdef_stack_ptr
) {
2604 tcc_error("missing #endif");
2605 } else if (s1
->include_stack_ptr
== s1
->include_stack
) {
2606 /* no include left : end of file. */
2609 tok_flags
&= ~TOK_FLAG_EOF
;
2610 /* pop include file */
2612 /* test if previous '#endif' was after a #ifdef at
2614 if (tok_flags
& TOK_FLAG_ENDIF
) {
2616 printf("#endif %s\n", get_tok_str(file
->ifndef_macro_saved
, NULL
));
2618 search_cached_include(s1
, file
->filename
, 1)
2619 ->ifndef_macro
= file
->ifndef_macro_saved
;
2620 tok_flags
&= ~TOK_FLAG_ENDIF
;
2623 /* add end of include file debug info */
2624 if (tcc_state
->do_debug
) {
2625 put_stabd(N_EINCL
, 0, 0);
2627 /* pop include stack */
2629 s1
->include_stack_ptr
--;
2631 if (p
== file
->buffer
)
2632 tok_flags
= TOK_FLAG_BOF
|TOK_FLAG_BOL
;
2640 tok_flags
|= TOK_FLAG_BOL
;
2643 if (0 == (parse_flags
& PARSE_FLAG_LINEFEED
))
2646 goto keep_tok_flags
;
2651 if ((tok_flags
& TOK_FLAG_BOL
) &&
2652 (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
2654 preprocess(tok_flags
& TOK_FLAG_BOF
);
2660 tok
= TOK_TWOSHARPS
;
2662 if (parse_flags
& PARSE_FLAG_ASM_FILE
) {
2663 p
= parse_line_comment(p
- 1);
2672 /* dollar is allowed to start identifiers when not parsing asm */
2674 if (!(isidnum_table
[c
- CH_EOF
] & IS_ID
)
2675 || (parse_flags
& PARSE_FLAG_ASM_FILE
))
2678 case 'a': case 'b': case 'c': case 'd':
2679 case 'e': case 'f': case 'g': case 'h':
2680 case 'i': case 'j': case 'k': case 'l':
2681 case 'm': case 'n': case 'o': case 'p':
2682 case 'q': case 'r': case 's': case 't':
2683 case 'u': case 'v': case 'w': case 'x':
2685 case 'A': case 'B': case 'C': case 'D':
2686 case 'E': case 'F': case 'G': case 'H':
2687 case 'I': case 'J': case 'K':
2688 case 'M': case 'N': case 'O': case 'P':
2689 case 'Q': case 'R': case 'S': case 'T':
2690 case 'U': case 'V': case 'W': case 'X':
2696 h
= TOK_HASH_FUNC(h
, c
);
2697 while (c
= *++p
, isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2698 h
= TOK_HASH_FUNC(h
, c
);
2703 /* fast case : no stray found, so we have the full token
2704 and we have already hashed it */
2705 h
&= (TOK_HASH_SIZE
- 1);
2706 pts
= &hash_ident
[h
];
2711 if (ts
->len
== len
&& !memcmp(ts
->str
, p1
, len
))
2713 pts
= &(ts
->hash_next
);
2715 ts
= tok_alloc_new(pts
, (char *) p1
, len
);
2719 cstr_reset(&tokcstr
);
2720 cstr_cat(&tokcstr
, (char *) p1
, len
);
2724 while (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2726 cstr_ccat(&tokcstr
, c
);
2729 ts
= tok_alloc(tokcstr
.data
, tokcstr
.size
);
2735 if (t
!= '\\' && t
!= '\'' && t
!= '\"') {
2737 goto parse_ident_fast
;
2740 if (c
== '\'' || c
== '\"') {
2744 cstr_reset(&tokcstr
);
2745 cstr_ccat(&tokcstr
, 'L');
2746 goto parse_ident_slow
;
2751 case '0': case '1': case '2': case '3':
2752 case '4': case '5': case '6': case '7':
2756 /* after the first digit, accept digits, alpha, '.' or sign if
2757 prefixed by 'eEpP' */
2759 cstr_reset(&tokcstr
);
2761 cstr_ccat(&tokcstr
, t
);
2762 if (!((isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))
2764 || ((c
== '+' || c
== '-')
2765 && (((t
== 'e' || t
== 'E')
2766 && !(parse_flags
& PARSE_FLAG_ASM_FILE
2767 /* 0xe+1 is 3 tokens in asm */
2768 && ((char*)tokcstr
.data
)[0] == '0'
2769 && toup(((char*)tokcstr
.data
)[1]) == 'X'))
2770 || t
== 'p' || t
== 'P'))))
2775 /* We add a trailing '\0' to ease parsing */
2776 cstr_ccat(&tokcstr
, '\0');
2777 tokc
.str
.size
= tokcstr
.size
;
2778 tokc
.str
.data
= tokcstr
.data
;
2783 /* special dot handling because it can also start a number */
2788 } else if ((isidnum_table
['.' - CH_EOF
] & IS_ID
)
2789 && (isidnum_table
[c
- CH_EOF
] & (IS_ID
|IS_NUM
))) {
2791 goto parse_ident_fast
;
2792 } else if (c
== '.') {
2798 *--p
= '.'; /* may underflow into file->unget[] */
2809 cstr_reset(&tokcstr
);
2811 cstr_ccat(&tokcstr
, 'L');
2812 cstr_ccat(&tokcstr
, c
);
2813 p
= parse_pp_string(p
, c
, &tokcstr
);
2814 cstr_ccat(&tokcstr
, c
);
2815 cstr_ccat(&tokcstr
, '\0');
2816 tokc
.str
.size
= tokcstr
.size
;
2817 tokc
.str
.data
= tokcstr
.data
;
2826 } else if (c
== '<') {
2843 } else if (c
== '>') {
2861 } else if (c
== '=') {
2874 } else if (c
== '=') {
2887 } else if (c
== '=') {
2900 } else if (c
== '=') {
2903 } else if (c
== '>') {
2911 PARSE2('!', '!', '=', TOK_NE
)
2912 PARSE2('=', '=', '=', TOK_EQ
)
2913 PARSE2('*', '*', '=', TOK_A_MUL
)
2914 PARSE2('%', '%', '=', TOK_A_MOD
)
2915 PARSE2('^', '^', '=', TOK_A_XOR
)
2917 /* comments or operator */
2921 p
= parse_comment(p
);
2922 /* comments replaced by a blank */
2924 goto keep_tok_flags
;
2925 } else if (c
== '/') {
2926 p
= parse_line_comment(p
);
2928 goto keep_tok_flags
;
2929 } else if (c
== '=') {
2949 case '@': /* only used in assembler */
2955 if (c
>= 0x80 && c
<= 0xFF) /* utf8 identifiers */
2956 goto parse_ident_fast
;
2957 if (parse_flags
& PARSE_FLAG_ASM_FILE
)
2959 tcc_error("unrecognized character \\x%02x", c
);
2965 #if defined(PARSE_DEBUG)
2966 printf("token = %d %s\n", tok
, get_tok_str(tok
, &tokc
));
2970 /* return next token without macro substitution. Can read input from
2972 static void next_nomacro_spc(void)
2978 TOK_GET(&tok
, ¯o_ptr
, &tokc
);
2979 if (tok
== TOK_LINENUM
) {
2980 file
->line_num
= tokc
.i
;
2987 //printf("token = %s\n", get_tok_str(tok, &tokc));
2990 ST_FUNC
void next_nomacro(void)
2994 } while (tok
< 256 && (isidnum_table
[tok
- CH_EOF
] & IS_SPC
));
2998 static void macro_subst(
2999 TokenString
*tok_str
,
3001 const int *macro_str
3004 /* substitute arguments in replacement lists in macro_str by the values in
3005 args (field d) and return allocated string */
3006 static int *macro_arg_subst(Sym
**nested_list
, const int *macro_str
, Sym
*args
)
3018 TOK_GET(&t
, ¯o_str
, &cval
);
3023 TOK_GET(&t
, ¯o_str
, &cval
);
3026 s
= sym_find2(args
, t
);
3029 cstr_ccat(&cstr
, '\"');
3033 TOK_GET(&t
, &st
, &cval
);
3034 if (t
!= TOK_PLCHLDR
3036 && 0 == check_space(t
, &spc
)) {
3037 const char *s
= get_tok_str(t
, &cval
);
3039 if (t
== TOK_PPSTR
&& *s
!= '\'')
3040 add_char(&cstr
, *s
);
3042 cstr_ccat(&cstr
, *s
);
3048 cstr_ccat(&cstr
, '\"');
3049 cstr_ccat(&cstr
, '\0');
3051 printf("\nstringize: <%s>\n", (char *)cstr
.data
);
3054 cval
.str
.size
= cstr
.size
;
3055 cval
.str
.data
= cstr
.data
;
3056 tok_str_add2(&str
, TOK_PPSTR
, &cval
);
3060 expect("macro parameter after '#'");
3062 } else if (t
>= TOK_IDENT
) {
3063 s
= sym_find2(args
, t
);
3067 /* if '##' is present before or after, no arg substitution */
3068 if (*macro_str
== TOK_PPJOIN
|| t1
== TOK_PPJOIN
) {
3069 /* special case for var arg macros : ## eats the ','
3070 if empty VA_ARGS variable. */
3071 if (t1
== TOK_PPJOIN
&& t0
== ',' && gnu_ext
&& s
->type
.t
) {
3073 /* suppress ',' '##' */
3076 /* suppress '##' and add variable */
3084 /* Expand arguments tokens and store them. In most
3085 cases we could also re-expand each argument if
3086 used multiple times, but not if the argument
3087 contains the __COUNTER__ macro. */
3089 sym_push2(&s
->next
, s
->v
, s
->type
.t
, 0);
3091 macro_subst(&str2
, nested_list
, st
);
3092 tok_str_add(&str2
, 0);
3093 s
->next
->d
= str2
.str
;
3099 TOK_GET(&t2
, &st
, &cval
);
3102 tok_str_add2(&str
, t2
, &cval
);
3104 if (str
.len
== l0
) /* expanded to empty string */
3105 tok_str_add(&str
, TOK_PLCHLDR
);
3107 tok_str_add(&str
, t
);
3110 tok_str_add2(&str
, t
, &cval
);
3114 tok_str_add(&str
, 0);
3118 static char const ab_month_name
[12][4] =
3120 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
3121 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
3124 static int paste_tokens(int t1
, CValue
*v1
, int t2
, CValue
*v2
)
3130 if (t1
!= TOK_PLCHLDR
)
3131 cstr_cat(&cstr
, get_tok_str(t1
, v1
), -1);
3133 if (t2
!= TOK_PLCHLDR
)
3134 cstr_cat(&cstr
, get_tok_str(t2
, v2
), -1);
3135 cstr_ccat(&cstr
, '\0');
3137 tcc_open_bf(tcc_state
, ":paste:", cstr
.size
);
3138 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3142 if (0 == *file
->buf_ptr
)
3146 tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid"
3147 " preprocessing token", n
, cstr
.data
, (char*)cstr
.data
+ n
);
3152 //printf("paste <%s>\n", (char*)cstr.data);
3157 /* handle the '##' operator. Return NULL if no '##' seen. Otherwise
3158 return the resulting string (which must be freed). */
3159 static inline int *macro_twosharps(const int *ptr0
)
3163 TokenString macro_str1
;
3164 int start_of_nosubsts
= -1;
3167 /* we search the first '##' */
3168 for (ptr
= ptr0
;;) {
3169 TOK_GET(&t
, &ptr
, &cval
);
3170 if (t
== TOK_PPJOIN
)
3176 tok_str_new(¯o_str1
);
3178 //tok_print(" $$$", ptr0);
3179 for (ptr
= ptr0
;;) {
3180 TOK_GET(&t
, &ptr
, &cval
);
3183 if (t
== TOK_PPJOIN
)
3185 while (*ptr
== TOK_PPJOIN
) {
3187 /* given 'a##b', remove nosubsts preceding 'a' */
3188 if (start_of_nosubsts
>= 0)
3189 macro_str1
.len
= start_of_nosubsts
;
3190 /* given 'a##b', remove nosubsts preceding 'b' */
3191 while ((t1
= *++ptr
) == TOK_NOSUBST
)
3193 if (t1
&& t1
!= TOK_PPJOIN
) {
3194 TOK_GET(&t1
, &ptr
, &cv1
);
3195 if (t
!= TOK_PLCHLDR
|| t1
!= TOK_PLCHLDR
) {
3196 if (paste_tokens(t
, &cval
, t1
, &cv1
)) {
3197 t
= tok
, cval
= tokc
;
3199 tok_str_add2(¯o_str1
, t
, &cval
);
3205 if (t
== TOK_NOSUBST
) {
3206 if (start_of_nosubsts
< 0)
3207 start_of_nosubsts
= macro_str1
.len
;
3209 start_of_nosubsts
= -1;
3211 tok_str_add2(¯o_str1
, t
, &cval
);
3213 tok_str_add(¯o_str1
, 0);
3214 //tok_print(" ###", macro_str1.str);
3215 return macro_str1
.str
;
3218 /* peek or read [ws_str == NULL] next token from function macro call,
3219 walking up macro levels up to the file if necessary */
3220 static int next_argstream(Sym
**nested_list
, TokenString
*ws_str
)
3228 p
= macro_ptr
, t
= *p
;
3230 while (is_space(t
) || TOK_LINEFEED
== t
|| TOK_PLCHLDR
== t
)
3231 tok_str_add(ws_str
, t
), t
= *++p
;
3235 /* also, end of scope for nested defined symbol */
3237 while (sa
&& sa
->v
== 0)
3246 while (is_space(ch
) || ch
== '\n' || ch
== '/') {
3249 uint8_t *p
= file
->buf_ptr
;
3252 p
= parse_comment(p
);
3253 file
->buf_ptr
= p
- 1;
3254 } else if (c
== '/') {
3255 p
= parse_line_comment(p
);
3256 file
->buf_ptr
= p
- 1;
3263 if (!(ch
== '\f' || ch
== '\v' || ch
== '\r'))
3264 tok_str_add(ws_str
, ch
);
3278 /* do macro substitution of current token with macro 's' and add
3279 result to (tok_str,tok_len). 'nested_list' is the list of all
3280 macros we got inside to avoid recursing. Return non zero if no
3281 substitution needs to be done */
3282 static int macro_subst_tok(
3283 TokenString
*tok_str
,
3287 Sym
*args
, *sa
, *sa1
;
3288 int parlevel
, t
, t1
, spc
;
3295 /* if symbol is a macro, prepare substitution */
3296 /* special macros */
3297 if (tok
== TOK___LINE__
|| tok
== TOK___COUNTER__
) {
3298 t
= tok
== TOK___LINE__
? file
->line_num
: pp_counter
++;
3299 snprintf(buf
, sizeof(buf
), "%d", t
);
3303 } else if (tok
== TOK___FILE__
) {
3304 cstrval
= file
->filename
;
3306 } else if (tok
== TOK___DATE__
|| tok
== TOK___TIME__
) {
3311 tm
= localtime(&ti
);
3312 if (tok
== TOK___DATE__
) {
3313 snprintf(buf
, sizeof(buf
), "%s %2d %d",
3314 ab_month_name
[tm
->tm_mon
], tm
->tm_mday
, tm
->tm_year
+ 1900);
3316 snprintf(buf
, sizeof(buf
), "%02d:%02d:%02d",
3317 tm
->tm_hour
, tm
->tm_min
, tm
->tm_sec
);
3324 cstr_cat(&cstr
, cstrval
, 0);
3325 cval
.str
.size
= cstr
.size
;
3326 cval
.str
.data
= cstr
.data
;
3327 tok_str_add2(tok_str
, t1
, &cval
);
3330 int saved_parse_flags
= parse_flags
;
3331 int *joined_str
= NULL
;
3334 if (s
->type
.t
== MACRO_FUNC
) {
3335 /* whitespace between macro name and argument list */
3337 tok_str_new(&ws_str
);
3340 parse_flags
|= PARSE_FLAG_SPACES
| PARSE_FLAG_LINEFEED
3341 | PARSE_FLAG_ACCEPT_STRAYS
;
3343 /* get next token from argument stream */
3344 t
= next_argstream(nested_list
, &ws_str
);
3346 /* not a macro substitution after all, restore the
3347 * macro token plus all whitespace we've read.
3348 * whitespace is intentionally not merged to preserve
3350 parse_flags
= saved_parse_flags
;
3351 tok_str_add(tok_str
, tok
);
3352 if (parse_flags
& PARSE_FLAG_SPACES
) {
3354 for (i
= 0; i
< ws_str
.len
; i
++)
3355 tok_str_add(tok_str
, ws_str
.str
[i
]);
3357 tok_str_free_str(ws_str
.str
);
3360 tok_str_free_str(ws_str
.str
);
3363 next_nomacro(); /* eat '(' */
3364 } while (tok
== TOK_PLCHLDR
);
3366 /* argument macro */
3369 /* NOTE: empty args are allowed, except if no args */
3372 next_argstream(nested_list
, NULL
);
3373 } while (is_space(tok
) || TOK_LINEFEED
== tok
);
3375 /* handle '()' case */
3376 if (!args
&& !sa
&& tok
== ')')
3379 tcc_error("macro '%s' used with too many args",
3380 get_tok_str(s
->v
, 0));
3383 /* NOTE: non zero sa->t indicates VA_ARGS */
3384 while ((parlevel
> 0 ||
3386 (tok
!= ',' || sa
->type
.t
)))) {
3387 if (tok
== TOK_EOF
|| tok
== 0)
3391 else if (tok
== ')')
3393 if (tok
== TOK_LINEFEED
)
3395 if (!check_space(tok
, &spc
))
3396 tok_str_add2(&str
, tok
, &tokc
);
3397 next_argstream(nested_list
, NULL
);
3402 tok_str_add(&str
, -1);
3403 tok_str_add(&str
, 0);
3404 sa1
= sym_push2(&args
, sa
->v
& ~SYM_FIELD
, sa
->type
.t
, 0);
3408 /* special case for gcc var args: add an empty
3409 var arg argument if it is omitted */
3410 if (sa
&& sa
->type
.t
&& gnu_ext
)
3418 tcc_error("macro '%s' used with too few args",
3419 get_tok_str(s
->v
, 0));
3422 parse_flags
= saved_parse_flags
;
3424 /* now subst each arg */
3425 mstr
= macro_arg_subst(nested_list
, mstr
, args
);
3430 tok_str_free_str(sa
->d
);
3432 tok_str_free_str(sa
->next
->d
);
3440 sym_push2(nested_list
, s
->v
, 0, 0);
3441 parse_flags
= saved_parse_flags
;
3442 joined_str
= macro_twosharps(mstr
);
3443 macro_subst(tok_str
, nested_list
, joined_str
? joined_str
: mstr
);
3445 /* pop nested defined symbol */
3447 *nested_list
= sa1
->prev
;
3450 tok_str_free_str(joined_str
);
3452 tok_str_free_str(mstr
);
3457 /* do macro substitution of macro_str and add result to
3458 (tok_str,tok_len). 'nested_list' is the list of all macros we got
3459 inside to avoid recursing. */
3460 static void macro_subst(
3461 TokenString
*tok_str
,
3463 const int *macro_str
3467 int t
, spc
, nosubst
;
3473 TOK_GET(&t
, ¯o_str
, &cval
);
3477 if (t
>= TOK_IDENT
&& 0 == nosubst
) {
3482 /* if nested substitution, do nothing */
3483 if (sym_find2(*nested_list
, t
)) {
3484 /* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
3485 tok_str_add2(tok_str
, TOK_NOSUBST
, NULL
);
3491 str
.str
= (int*)macro_str
;
3492 begin_macro(&str
, 2);
3495 macro_subst_tok(tok_str
, nested_list
, s
);
3497 if (str
.alloc
== 3) {
3498 /* already finished by reading function macro arguments */
3502 macro_str
= macro_ptr
;
3506 spc
= is_space(t
= tok_str
->str
[tok_str
->lastlen
]);
3508 if (t
== '\\' && !(parse_flags
& PARSE_FLAG_ACCEPT_STRAYS
))
3509 tcc_error("stray '\\' in program");
3511 if (!check_space(t
, &spc
))
3512 tok_str_add2(tok_str
, t
, &cval
);
3515 if (nosubst
> 1 && (spc
|| (++nosubst
== 3 && t
== '(')))
3519 if (t
== TOK_NOSUBST
)
3522 /* GCC supports 'defined' as result of a macro substitution */
3523 if (t
== TOK_DEFINED
&& pp_expr
)
3528 /* return next token with macro substitution */
3529 ST_FUNC
void next(void)
3532 if (parse_flags
& PARSE_FLAG_SPACES
)
3538 if (tok
== TOK_NOSUBST
|| tok
== TOK_PLCHLDR
) {
3539 /* discard preprocessor markers */
3541 } else if (tok
== 0) {
3542 /* end of macro or unget token string */
3546 } else if (tok
>= TOK_IDENT
&& (parse_flags
& PARSE_FLAG_PREPROCESS
)) {
3548 /* if reading from file, try to substitute macros */
3549 s
= define_find(tok
);
3551 Sym
*nested_list
= NULL
;
3553 macro_subst_tok(&tokstr_buf
, &nested_list
, s
);
3554 tok_str_add(&tokstr_buf
, 0);
3555 begin_macro(&tokstr_buf
, 2);
3559 /* convert preprocessor tokens into C tokens */
3560 if (tok
== TOK_PPNUM
) {
3561 if (parse_flags
& PARSE_FLAG_TOK_NUM
)
3562 parse_number((char *)tokc
.str
.data
);
3563 } else if (tok
== TOK_PPSTR
) {
3564 if (parse_flags
& PARSE_FLAG_TOK_STR
)
3565 parse_string((char *)tokc
.str
.data
, tokc
.str
.size
- 1);
3569 /* push back current token and set current token to 'last_tok'. Only
3570 identifier case handled for labels. */
3571 ST_INLN
void unget_tok(int last_tok
)
3574 TokenString
*str
= tok_str_alloc();
3575 tok_str_add2(str
, tok
, &tokc
);
3576 tok_str_add(str
, 0);
3577 begin_macro(str
, 1);
3581 ST_FUNC
void preprocess_start(TCCState
*s1
, int is_asm
)
3586 s1
->include_stack_ptr
= s1
->include_stack
;
3587 s1
->ifdef_stack_ptr
= s1
->ifdef_stack
;
3588 file
->ifdef_stack_ptr
= s1
->ifdef_stack_ptr
;
3591 pp_debug_tok
= pp_debug_symv
= 0;
3593 pvtop
= vtop
= vstack
- 1;
3594 s1
->pack_stack
[0] = 0;
3595 s1
->pack_stack_ptr
= s1
->pack_stack
;
3597 set_idnum('$', s1
->dollars_in_identifiers
? IS_ID
: 0);
3598 set_idnum('.', is_asm
? IS_ID
: 0);
3601 cstr_cat(&cstr
, "\"", -1);
3602 cstr_cat(&cstr
, file
->filename
, -1);
3603 cstr_cat(&cstr
, "\"", 0);
3604 tcc_define_symbol(s1
, "__BASE_FILE__", cstr
.data
);
3607 for (i
= 0; i
< s1
->nb_cmd_include_files
; i
++) {
3608 cstr_cat(&cstr
, "#include \"", -1);
3609 cstr_cat(&cstr
, s1
->cmd_include_files
[i
], -1);
3610 cstr_cat(&cstr
, "\"\n", -1);
3613 *s1
->include_stack_ptr
++ = file
;
3614 tcc_open_bf(s1
, "<command line>", cstr
.size
);
3615 memcpy(file
->buffer
, cstr
.data
, cstr
.size
);
3620 tcc_define_symbol(s1
, "__ASSEMBLER__", NULL
);
3622 parse_flags
= is_asm
? PARSE_FLAG_ASM_FILE
: 0;
3623 tok_flags
= TOK_FLAG_BOL
| TOK_FLAG_BOF
;
3626 /* cleanup from error/setjmp */
3627 ST_FUNC
void preprocess_end(TCCState
*s1
)
3634 ST_FUNC
void tccpp_new(TCCState
*s
)
3639 /* might be used in error() before preprocess_start() */
3640 s
->include_stack_ptr
= s
->include_stack
;
3643 /* init isid table */
3644 for(i
= CH_EOF
; i
<128; i
++)
3646 is_space(i
) ? IS_SPC
3651 for(i
= 128; i
<256; i
++)
3652 set_idnum(i
, IS_ID
);
3654 /* init allocators */
3655 tal_new(&toksym_alloc
, TOKSYM_TAL_LIMIT
, TOKSYM_TAL_SIZE
);
3656 tal_new(&tokstr_alloc
, TOKSTR_TAL_LIMIT
, TOKSTR_TAL_SIZE
);
3657 tal_new(&cstr_alloc
, CSTR_TAL_LIMIT
, CSTR_TAL_SIZE
);
3659 memset(hash_ident
, 0, TOK_HASH_SIZE
* sizeof(TokenSym
*));
3660 cstr_new(&cstr_buf
);
3661 cstr_realloc(&cstr_buf
, STRING_MAX_SIZE
);
3662 tok_str_new(&tokstr_buf
);
3663 tok_str_realloc(&tokstr_buf
, TOKSTR_MAX_SIZE
);
3665 tok_ident
= TOK_IDENT
;
3674 tok_alloc(p
, r
- p
- 1);
3679 ST_FUNC
void tccpp_delete(TCCState
*s
)
3683 /* free -D and compiler defines */
3687 n
= tok_ident
- TOK_IDENT
;
3688 for(i
= 0; i
< n
; i
++)
3689 tal_free(toksym_alloc
, table_ident
[i
]);
3690 tcc_free(table_ident
);
3693 /* free static buffers */
3694 cstr_free(&tokcstr
);
3695 cstr_free(&cstr_buf
);
3696 cstr_free(¯o_equal_buf
);
3697 tok_str_free_str(tokstr_buf
.str
);
3699 /* free allocators */
3700 tal_delete(toksym_alloc
);
3701 toksym_alloc
= NULL
;
3702 tal_delete(tokstr_alloc
);
3703 tokstr_alloc
= NULL
;
3704 tal_delete(cstr_alloc
);
3708 /* ------------------------------------------------------------------------- */
3709 /* tcc -E [-P[1]] [-dD} support */
3711 static void tok_print(const char *msg
, const int *str
)
3717 fp
= tcc_state
->ppfp
;
3718 fprintf(fp
, "%s", msg
);
3720 TOK_GET(&t
, &str
, &cval
);
3723 fprintf(fp
, " %s" + s
, get_tok_str(t
, &cval
)), s
= 1;
3728 static void pp_line(TCCState
*s1
, BufferedFile
*f
, int level
)
3730 int d
= f
->line_num
- f
->line_ref
;
3735 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_NONE
) {
3737 } else if (level
== 0 && f
->line_ref
&& d
< 8) {
3739 fputs("\n", s1
->ppfp
), --d
;
3740 } else if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_STD
) {
3741 fprintf(s1
->ppfp
, "#line %d \"%s\"\n", f
->line_num
, f
->filename
);
3743 fprintf(s1
->ppfp
, "# %d \"%s\"%s\n", f
->line_num
, f
->filename
,
3744 level
> 0 ? " 1" : level
< 0 ? " 2" : "");
3746 f
->line_ref
= f
->line_num
;
3749 static void define_print(TCCState
*s1
, int v
)
3755 if (NULL
== s
|| NULL
== s
->d
)
3759 fprintf(fp
, "#define %s", get_tok_str(v
, NULL
));
3760 if (s
->type
.t
== MACRO_FUNC
) {
3765 fprintf(fp
,"%s", get_tok_str(a
->v
& ~SYM_FIELD
, NULL
));
3772 tok_print("", s
->d
);
3775 static void pp_debug_defines(TCCState
*s1
)
3786 pp_line(s1
, file
, 0);
3787 file
->line_ref
= ++file
->line_num
;
3791 vs
= get_tok_str(v
, NULL
);
3792 if (t
== TOK_DEFINE
) {
3793 define_print(s1
, v
);
3794 } else if (t
== TOK_UNDEF
) {
3795 fprintf(fp
, "#undef %s\n", vs
);
3796 } else if (t
== TOK_push_macro
) {
3797 fprintf(fp
, "#pragma push_macro(\"%s\")\n", vs
);
3798 } else if (t
== TOK_pop_macro
) {
3799 fprintf(fp
, "#pragma pop_macro(\"%s\")\n", vs
);
3804 static void pp_debug_builtins(TCCState
*s1
)
3807 for (v
= TOK_IDENT
; v
< tok_ident
; ++v
)
3808 define_print(s1
, v
);
3811 /* Add a space between tokens a and b to avoid unwanted textual pasting */
3812 static int pp_need_space(int a
, int b
)
3814 return 'E' == a
? '+' == b
|| '-' == b
3815 : '+' == a
? TOK_INC
== b
|| '+' == b
3816 : '-' == a
? TOK_DEC
== b
|| '-' == b
3817 : a
>= TOK_IDENT
? b
>= TOK_IDENT
3818 : a
== TOK_PPNUM
? b
>= TOK_IDENT
3822 /* maybe hex like 0x1e */
3823 static int pp_check_he0xE(int t
, const char *p
)
3825 if (t
== TOK_PPNUM
&& toup(strchr(p
, 0)[-1]) == 'E')
3830 /* Preprocess the current file */
3831 ST_FUNC
int tcc_preprocess(TCCState
*s1
)
3833 BufferedFile
**iptr
;
3834 int token_seen
, spcs
, level
;
3838 parse_flags
= PARSE_FLAG_PREPROCESS
3839 | (parse_flags
& PARSE_FLAG_ASM_FILE
)
3840 | PARSE_FLAG_LINEFEED
3842 | PARSE_FLAG_ACCEPT_STRAYS
3844 /* Credits to Fabrice Bellard's initial revision to demonstrate its
3845 capability to compile and run itself, provided all numbers are
3846 given as decimals. tcc -E -P10 will do. */
3847 if (s1
->Pflag
== LINE_MACRO_OUTPUT_FORMAT_P10
)
3848 parse_flags
|= PARSE_FLAG_TOK_NUM
, s1
->Pflag
= 1;
3851 /* for PP benchmarks */
3852 do next(); while (tok
!= TOK_EOF
);
3856 if (s1
->dflag
& 1) {
3857 pp_debug_builtins(s1
);
3861 token_seen
= TOK_LINEFEED
, spcs
= 0;
3862 pp_line(s1
, file
, 0);
3864 iptr
= s1
->include_stack_ptr
;
3869 level
= s1
->include_stack_ptr
- iptr
;
3872 pp_line(s1
, *iptr
, 0);
3873 pp_line(s1
, file
, level
);
3875 if (s1
->dflag
& 7) {
3876 pp_debug_defines(s1
);
3881 if (is_space(tok
)) {
3882 if (spcs
< sizeof white
- 1)
3883 white
[spcs
++] = tok
;
3885 } else if (tok
== TOK_LINEFEED
) {
3887 if (token_seen
== TOK_LINEFEED
)
3890 } else if (token_seen
== TOK_LINEFEED
) {
3891 pp_line(s1
, file
, 0);
3892 } else if (spcs
== 0 && pp_need_space(token_seen
, tok
)) {
3893 white
[spcs
++] = ' ';
3896 white
[spcs
] = 0, fputs(white
, s1
->ppfp
), spcs
= 0;
3897 fputs(p
= get_tok_str(tok
, &tokc
), s1
->ppfp
);
3898 token_seen
= pp_check_he0xE(tok
, p
);
3903 /* ------------------------------------------------------------------------- */