tinycc/tccpp.c

3913 lines
108 KiB
C
Raw Normal View History

/*
* TCC - Tiny C Compiler
*
* Copyright (c) 2001-2004 Fabrice Bellard
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#define USING_GLOBALS
#include "tcc.h"
/********************************************************/
/* global variables */
ST_DATA int tok_flags;
ST_DATA int parse_flags;
ST_DATA struct BufferedFile *file;
ST_DATA int ch, tok;
ST_DATA CValue tokc;
ST_DATA const int *macro_ptr;
ST_DATA CString tokcstr; /* current parsed string, if any */
/* display benchmark infos */
ST_DATA int tok_ident;
ST_DATA TokenSym **table_ident;
/* ------------------------------------------------------------------------- */
static TokenSym *hash_ident[TOK_HASH_SIZE];
static char token_buf[STRING_MAX_SIZE + 1];
static CString cstr_buf;
static CString macro_equal_buf;
static TokenString tokstr_buf;
static unsigned char isidnum_table[256 - CH_EOF];
static int pp_debug_tok, pp_debug_symv;
static int pp_once;
static int pp_expr;
static int pp_counter;
static void tok_print(const char *msg, const int *str);
static struct TinyAlloc *toksym_alloc;
static struct TinyAlloc *tokstr_alloc;
static TokenString *macro_stack;
static const char tcc_keywords[] =
#define DEF(id, str) str "\0"
#include "tcctok.h"
#undef DEF
;
/* WARNING: the content of this string encodes token numbers */
2009-07-19 00:07:42 +04:00
static const unsigned char tok_two_chars[] =
/* outdated -- gr
2009-07-19 00:07:42 +04:00
"<=\236>=\235!=\225&&\240||\241++\244--\242==\224<<\1>>\2+=\253"
"-=\255*=\252/=\257%=\245&=\246^=\336|=\374->\313..\250##\266";
*/{
'<','=', TOK_LE,
'>','=', TOK_GE,
'!','=', TOK_NE,
'&','&', TOK_LAND,
'|','|', TOK_LOR,
'+','+', TOK_INC,
'-','-', TOK_DEC,
'=','=', TOK_EQ,
'<','<', TOK_SHL,
'>','>', TOK_SAR,
'+','=', TOK_A_ADD,
'-','=', TOK_A_SUB,
'*','=', TOK_A_MUL,
'/','=', TOK_A_DIV,
'%','=', TOK_A_MOD,
'&','=', TOK_A_AND,
'^','=', TOK_A_XOR,
'|','=', TOK_A_OR,
'-','>', TOK_ARROW,
'.','.', TOK_TWODOTS,
'#','#', TOK_TWOSHARPS,
0
};
2009-05-11 20:45:56 +04:00
static void next_nomacro_spc(void);
ST_FUNC void skip(int c)
{
if (tok != c)
tcc_error("'%c' expected (got \"%s\")", c, get_tok_str(tok, &tokc));
next();
}
2009-05-11 20:45:56 +04:00
ST_FUNC void expect(const char *msg)
{
tcc_error("%s expected", msg);
}
/* ------------------------------------------------------------------------- */
/* Custom allocator for tiny objects */
#define USE_TAL
#ifndef USE_TAL
#define tal_free(al, p) tcc_free(p)
#define tal_realloc(al, p, size) tcc_realloc(p, size)
#define tal_new(a,b,c)
#define tal_delete(a)
#else
#if !defined(MEM_DEBUG)
#define tal_free(al, p) tal_free_impl(al, p)
#define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size)
#define TAL_DEBUG_PARAMS
#else
#define TAL_DEBUG 1
//#define TAL_INFO 1 /* collect and dump allocators stats */
#define tal_free(al, p) tal_free_impl(al, p, __FILE__, __LINE__)
#define tal_realloc(al, p, size) tal_realloc_impl(&al, p, size, __FILE__, __LINE__)
#define TAL_DEBUG_PARAMS , const char *file, int line
#define TAL_DEBUG_FILE_LEN 40
#endif
#define TOKSYM_TAL_SIZE (768 * 1024) /* allocator for tiny TokenSym in table_ident */
#define TOKSTR_TAL_SIZE (768 * 1024) /* allocator for tiny TokenString instances */
#define CSTR_TAL_SIZE (256 * 1024) /* allocator for tiny CString instances */
#define TOKSYM_TAL_LIMIT 256 /* prefer unique limits to distinguish allocators debug msgs */
#define TOKSTR_TAL_LIMIT 128 /* 32 * sizeof(int) */
#define CSTR_TAL_LIMIT 1024
typedef struct TinyAlloc {
unsigned limit;
unsigned size;
uint8_t *buffer;
uint8_t *p;
unsigned nb_allocs;
struct TinyAlloc *next, *top;
#ifdef TAL_INFO
unsigned nb_peak;
unsigned nb_total;
unsigned nb_missed;
uint8_t *peak_p;
#endif
} TinyAlloc;
typedef struct tal_header_t {
unsigned size;
#ifdef TAL_DEBUG
int line_num; /* negative line_num used for double free check */
char file_name[TAL_DEBUG_FILE_LEN + 1];
#endif
} tal_header_t;
/* ------------------------------------------------------------------------- */
static TinyAlloc *tal_new(TinyAlloc **pal, unsigned limit, unsigned size)
{
TinyAlloc *al = tcc_mallocz(sizeof(TinyAlloc));
al->p = al->buffer = tcc_malloc(size);
al->limit = limit;
al->size = size;
if (pal) *pal = al;
return al;
}
static void tal_delete(TinyAlloc *al)
{
TinyAlloc *next;
tail_call:
if (!al)
return;
#ifdef TAL_INFO
fprintf(stderr, "limit=%5d, size=%5g MB, nb_peak=%6d, nb_total=%8d, nb_missed=%6d, usage=%5.1f%%\n",
al->limit, al->size / 1024.0 / 1024.0, al->nb_peak, al->nb_total, al->nb_missed,
(al->peak_p - al->buffer) * 100.0 / al->size);
#endif
#ifdef TAL_DEBUG
if (al->nb_allocs > 0) {
uint8_t *p;
fprintf(stderr, "TAL_DEBUG: memory leak %d chunk(s) (limit= %d)\n",
al->nb_allocs, al->limit);
p = al->buffer;
while (p < al->p) {
tal_header_t *header = (tal_header_t *)p;
if (header->line_num > 0) {
fprintf(stderr, "%s:%d: chunk of %d bytes leaked\n",
header->file_name, header->line_num, header->size);
}
p += header->size + sizeof(tal_header_t);
}
#if MEM_DEBUG-0 == 2
exit(2);
#endif
}
#endif
next = al->next;
tcc_free(al->buffer);
tcc_free(al);
al = next;
goto tail_call;
}
static void tal_free_impl(TinyAlloc *al, void *p TAL_DEBUG_PARAMS)
{
if (!p)
return;
tail_call:
if (al->buffer <= (uint8_t *)p && (uint8_t *)p < al->buffer + al->size) {
#ifdef TAL_DEBUG
tal_header_t *header = (((tal_header_t *)p) - 1);
if (header->line_num < 0) {
fprintf(stderr, "%s:%d: TAL_DEBUG: double frees chunk from\n",
file, line);
fprintf(stderr, "%s:%d: %d bytes\n",
header->file_name, (int)-header->line_num, (int)header->size);
} else
header->line_num = -header->line_num;
#endif
al->nb_allocs--;
if (!al->nb_allocs)
al->p = al->buffer;
} else if (al->next) {
al = al->next;
goto tail_call;
}
else
tcc_free(p);
}
static void *tal_realloc_impl(TinyAlloc **pal, void *p, unsigned size TAL_DEBUG_PARAMS)
{
tal_header_t *header;
void *ret;
int is_own;
unsigned adj_size = (size + 3) & -4;
TinyAlloc *al = *pal;
tail_call:
is_own = (al->buffer <= (uint8_t *)p && (uint8_t *)p < al->buffer + al->size);
if ((!p || is_own) && size <= al->limit) {
2020-01-18 00:58:39 +03:00
if (al->p - al->buffer + adj_size + sizeof(tal_header_t) < al->size) {
header = (tal_header_t *)al->p;
header->size = adj_size;
#ifdef TAL_DEBUG
{ int ofs = strlen(file) - TAL_DEBUG_FILE_LEN;
strncpy(header->file_name, file + (ofs > 0 ? ofs : 0), TAL_DEBUG_FILE_LEN);
header->file_name[TAL_DEBUG_FILE_LEN] = 0;
header->line_num = line; }
#endif
ret = al->p + sizeof(tal_header_t);
al->p += adj_size + sizeof(tal_header_t);
if (is_own) {
header = (((tal_header_t *)p) - 1);
memcpy(ret, p, header->size);
#ifdef TAL_DEBUG
header->line_num = -header->line_num;
#endif
} else {
al->nb_allocs++;
}
#ifdef TAL_INFO
if (al->nb_peak < al->nb_allocs)
al->nb_peak = al->nb_allocs;
if (al->peak_p < al->p)
al->peak_p = al->p;
al->nb_total++;
#endif
return ret;
} else if (is_own) {
al->nb_allocs--;
ret = tal_realloc(*pal, 0, size);
header = (((tal_header_t *)p) - 1);
memcpy(ret, p, header->size);
#ifdef TAL_DEBUG
header->line_num = -header->line_num;
#endif
return ret;
}
if (al->next) {
al = al->next;
} else {
TinyAlloc *bottom = al, *next = al->top ? al->top : al;
al = tal_new(pal, next->limit, next->size * 2);
al->next = next;
bottom->top = al;
}
goto tail_call;
}
if (is_own) {
al->nb_allocs--;
ret = tcc_malloc(size);
header = (((tal_header_t *)p) - 1);
memcpy(ret, p, header->size);
#ifdef TAL_DEBUG
header->line_num = -header->line_num;
#endif
} else if (al->next) {
al = al->next;
goto tail_call;
} else
ret = tcc_realloc(p, size);
#ifdef TAL_INFO
al->nb_missed++;
#endif
return ret;
}
#endif /* USE_TAL */
/* ------------------------------------------------------------------------- */
/* CString handling */
static void cstr_realloc(CString *cstr, int new_size)
{
int size;
size = cstr->size_allocated;
if (size < 8)
size = 8; /* no need to allocate a too small first string */
while (size < new_size)
size = size * 2;
cstr->data = tcc_realloc(cstr->data, size);
cstr->size_allocated = size;
}
/* add a byte */
ST_INLN void cstr_ccat(CString *cstr, int ch)
{
int size;
size = cstr->size + 1;
if (size > cstr->size_allocated)
cstr_realloc(cstr, size);
((unsigned char *)cstr->data)[size - 1] = ch;
cstr->size = size;
}
ST_FUNC void cstr_cat(CString *cstr, const char *str, int len)
{
int size;
if (len <= 0)
len = strlen(str) + 1 + len;
size = cstr->size + len;
if (size > cstr->size_allocated)
cstr_realloc(cstr, size);
memmove(((unsigned char *)cstr->data) + cstr->size, str, len);
cstr->size = size;
}
2009-05-11 20:45:56 +04:00
/* add a wide char */
ST_FUNC void cstr_wccat(CString *cstr, int ch)
{
int size;
size = cstr->size + sizeof(nwchar_t);
if (size > cstr->size_allocated)
cstr_realloc(cstr, size);
*(nwchar_t *)(((unsigned char *)cstr->data) + size - sizeof(nwchar_t)) = ch;
cstr->size = size;
}
ST_FUNC void cstr_new(CString *cstr)
{
memset(cstr, 0, sizeof(CString));
}
/* free string and reset it to NULL */
ST_FUNC void cstr_free(CString *cstr)
{
tcc_free(cstr->data);
cstr_new(cstr);
}
Optimize cstr_reset() to only reset string to empty, not call free() and later malloc() A CString could be reset to empty just setting its .size to 0. If memory was already allocated, that would be remembered in .data_allocated and .size_allocated and on consequent string manipulations that memory will be used without immediate need to call malloc(). For $ ./tcc -B. -bench -DONE_SOURCE -DCONFIG_MULTIARCHDIR=\"i386-linux-gnu\" -c tcc.c after the patch malloc/free are called less often: (tcc is run in loop; perf record -a sleep 10 && perf report) before: # Overhead Command Shared Object Symbol # ........ ........... .................. .......................................... # 13.89% tcc tcc [.] next_nomacro1 4.73% tcc libc-2.13.so [.] _int_malloc 4.39% tcc tcc [.] next 2.94% tcc tcc [.] tok_str_add2 2.78% tcc tcc [.] macro_subst_tok 2.75% tcc libc-2.13.so [.] free 2.74% tcc tcc [.] macro_subst 2.63% tcc libc-2.13.so [.] _int_free 2.28% tcc tcc [.] vswap 2.24% tcc tcc [.] next_nomacro_spc 2.06% tcc libc-2.13.so [.] realloc 2.00% tcc libc-2.13.so [.] malloc 1.99% tcc tcc [.] unary 1.85% tcc libc-2.13.so [.] __i686.get_pc_thunk.bx 1.76% kworker/0:1 [kernel.kallsyms] [k] delay_tsc 1.70% tcc tcc [.] next_nomacro 1.62% tcc tcc [.] preprocess 1.41% tcc libc-2.13.so [.] __memcmp_ssse3 1.38% tcc [kernel.kallsyms] [k] memset 1.10% tcc tcc [.] g 1.06% tcc tcc [.] parse_btype 1.05% tcc tcc [.] sym_push2 1.04% tcc libc-2.13.so [.] _int_realloc 1.00% tcc libc-2.13.so [.] malloc_consolidate after: # Overhead Command Shared Object Symbol # ........ ........... .................. .............................................. # 15.26% tcc tcc [.] next_nomacro1 5.07% tcc libc-2.13.so [.] _int_malloc 4.62% tcc tcc [.] next 3.22% tcc tcc [.] tok_str_add2 3.03% tcc tcc [.] macro_subst_tok 3.02% tcc tcc [.] macro_subst 2.59% tcc tcc [.] next_nomacro_spc 2.44% tcc tcc [.] vswap 2.39% tcc libc-2.13.so [.] _int_free 2.28% tcc libc-2.13.so [.] free 2.22% tcc tcc [.] unary 2.07% tcc libc-2.13.so [.] realloc 1.97% tcc libc-2.13.so [.] malloc 1.70% tcc tcc [.] preprocess 1.69% tcc libc-2.13.so [.] __i686.get_pc_thunk.bx 1.68% tcc tcc [.] next_nomacro 1.59% tcc [kernel.kallsyms] [k] memset 1.55% tcc libc-2.13.so [.] __memcmp_ssse3 1.22% tcc tcc [.] parse_comment 1.11% tcc tcc [.] g 1.11% tcc tcc [.] sym_push2 1.10% tcc tcc [.] parse_btype 1.10% tcc libc-2.13.so [.] _int_realloc 1.06% tcc tcc [.] vsetc 0.98% tcc libc-2.13.so [.] malloc_consolidate and this gains small speedup for tcc: # best of 5 runs before: 8268 idents, 47191 lines, 1526670 bytes, 0.153 s, 307997 lines/s, 10.0 MB/s after: 8268 idents, 47203 lines, 1526763 bytes, 0.148 s, 319217 lines/s, 10.3 MB/s
2012-12-21 12:21:59 +04:00
/* reset string to empty */
ST_FUNC void cstr_reset(CString *cstr)
Optimize cstr_reset() to only reset string to empty, not call free() and later malloc() A CString could be reset to empty just setting its .size to 0. If memory was already allocated, that would be remembered in .data_allocated and .size_allocated and on consequent string manipulations that memory will be used without immediate need to call malloc(). For $ ./tcc -B. -bench -DONE_SOURCE -DCONFIG_MULTIARCHDIR=\"i386-linux-gnu\" -c tcc.c after the patch malloc/free are called less often: (tcc is run in loop; perf record -a sleep 10 && perf report) before: # Overhead Command Shared Object Symbol # ........ ........... .................. .......................................... # 13.89% tcc tcc [.] next_nomacro1 4.73% tcc libc-2.13.so [.] _int_malloc 4.39% tcc tcc [.] next 2.94% tcc tcc [.] tok_str_add2 2.78% tcc tcc [.] macro_subst_tok 2.75% tcc libc-2.13.so [.] free 2.74% tcc tcc [.] macro_subst 2.63% tcc libc-2.13.so [.] _int_free 2.28% tcc tcc [.] vswap 2.24% tcc tcc [.] next_nomacro_spc 2.06% tcc libc-2.13.so [.] realloc 2.00% tcc libc-2.13.so [.] malloc 1.99% tcc tcc [.] unary 1.85% tcc libc-2.13.so [.] __i686.get_pc_thunk.bx 1.76% kworker/0:1 [kernel.kallsyms] [k] delay_tsc 1.70% tcc tcc [.] next_nomacro 1.62% tcc tcc [.] preprocess 1.41% tcc libc-2.13.so [.] __memcmp_ssse3 1.38% tcc [kernel.kallsyms] [k] memset 1.10% tcc tcc [.] g 1.06% tcc tcc [.] parse_btype 1.05% tcc tcc [.] sym_push2 1.04% tcc libc-2.13.so [.] _int_realloc 1.00% tcc libc-2.13.so [.] malloc_consolidate after: # Overhead Command Shared Object Symbol # ........ ........... .................. .............................................. # 15.26% tcc tcc [.] next_nomacro1 5.07% tcc libc-2.13.so [.] _int_malloc 4.62% tcc tcc [.] next 3.22% tcc tcc [.] tok_str_add2 3.03% tcc tcc [.] macro_subst_tok 3.02% tcc tcc [.] macro_subst 2.59% tcc tcc [.] next_nomacro_spc 2.44% tcc tcc [.] vswap 2.39% tcc libc-2.13.so [.] _int_free 2.28% tcc libc-2.13.so [.] free 2.22% tcc tcc [.] unary 2.07% tcc libc-2.13.so [.] realloc 1.97% tcc libc-2.13.so [.] malloc 1.70% tcc tcc [.] preprocess 1.69% tcc libc-2.13.so [.] __i686.get_pc_thunk.bx 1.68% tcc tcc [.] next_nomacro 1.59% tcc [kernel.kallsyms] [k] memset 1.55% tcc libc-2.13.so [.] __memcmp_ssse3 1.22% tcc tcc [.] parse_comment 1.11% tcc tcc [.] g 1.11% tcc tcc [.] sym_push2 1.10% tcc tcc [.] parse_btype 1.10% tcc libc-2.13.so [.] _int_realloc 1.06% tcc tcc [.] vsetc 0.98% tcc libc-2.13.so [.] malloc_consolidate and this gains small speedup for tcc: # best of 5 runs before: 8268 idents, 47191 lines, 1526670 bytes, 0.153 s, 307997 lines/s, 10.0 MB/s after: 8268 idents, 47203 lines, 1526763 bytes, 0.148 s, 319217 lines/s, 10.3 MB/s
2012-12-21 12:21:59 +04:00
{
cstr->size = 0;
}
ST_FUNC int cstr_printf(CString *cstr, const char *fmt, ...)
{
va_list v;
int len, size;
va_start(v, fmt);
bcheck cleanup - revert Makefiles to state before last bcheck additions Instead, just load bcheck.o explicitly if that is what is wanted. - move tcc_add_bcheck() to the <target>-link.c files and remove revently added arguments. This function is to support tccelf.c with linking, not for tccgen.c to support compilation. - remove -ba option: It said: "-ba Enable better address checking with bounds checker" Okay, if it is better then to have it is not an option. - remove va_copy. It is C99 and we try to stay C89 in tinycc when possible. For example, MS compilers do not have va_copy. - win64: revert any 'fixes' to alloca It was correct as it was before, except for bound_checking where it was not implemented. This should now work too. - remove parasitic filename:linenum features Such feature is already present with rt_printline in tccrun.c. If it doesn't work it can be fixed. - revert changes to gen_bounded_ptr_add() gen_bounded_ptr_add() was working as it should before (mostly). For the sake of simplicity I switched it to CDECL. Anyway, FASTCALL means SLOWCALL with tinycc. In exchange you get one addition which is required for bounds_cnecking function arguments. The important thing is to check them *BEFORE* they are loaded into registers. New function gbound_args() does that. In any case, code instrumentation with the bounds-check functions as such now seems to work flawlessly again, which means when they are inserted as NOPs, any code that tcc can compile, seems to behave just the same as without them. What these functions then do when fully enabled, is a differnt story. I did not touch this.
2019-12-12 17:45:45 +03:00
len = vsnprintf(NULL, 0, fmt, v);
va_end(v);
size = cstr->size + len + 1;
if (size > cstr->size_allocated)
cstr_realloc(cstr, size);
bcheck cleanup - revert Makefiles to state before last bcheck additions Instead, just load bcheck.o explicitly if that is what is wanted. - move tcc_add_bcheck() to the <target>-link.c files and remove revently added arguments. This function is to support tccelf.c with linking, not for tccgen.c to support compilation. - remove -ba option: It said: "-ba Enable better address checking with bounds checker" Okay, if it is better then to have it is not an option. - remove va_copy. It is C99 and we try to stay C89 in tinycc when possible. For example, MS compilers do not have va_copy. - win64: revert any 'fixes' to alloca It was correct as it was before, except for bound_checking where it was not implemented. This should now work too. - remove parasitic filename:linenum features Such feature is already present with rt_printline in tccrun.c. If it doesn't work it can be fixed. - revert changes to gen_bounded_ptr_add() gen_bounded_ptr_add() was working as it should before (mostly). For the sake of simplicity I switched it to CDECL. Anyway, FASTCALL means SLOWCALL with tinycc. In exchange you get one addition which is required for bounds_cnecking function arguments. The important thing is to check them *BEFORE* they are loaded into registers. New function gbound_args() does that. In any case, code instrumentation with the bounds-check functions as such now seems to work flawlessly again, which means when they are inserted as NOPs, any code that tcc can compile, seems to behave just the same as without them. What these functions then do when fully enabled, is a differnt story. I did not touch this.
2019-12-12 17:45:45 +03:00
va_start(v, fmt);
vsnprintf((char*)cstr->data + cstr->size, size, fmt, v);
va_end(v);
cstr->size += len;
return len;
}
/* XXX: unicode ? */
static void add_char(CString *cstr, int c)
{
if (c == '\'' || c == '\"' || c == '\\') {
/* XXX: could be more precise if char or string */
cstr_ccat(cstr, '\\');
}
if (c >= 32 && c <= 126) {
cstr_ccat(cstr, c);
} else {
cstr_ccat(cstr, '\\');
if (c == '\n') {
cstr_ccat(cstr, 'n');
} else {
cstr_ccat(cstr, '0' + ((c >> 6) & 7));
cstr_ccat(cstr, '0' + ((c >> 3) & 7));
cstr_ccat(cstr, '0' + (c & 7));
}
}
}
/* ------------------------------------------------------------------------- */
/* allocate a new token */
static TokenSym *tok_alloc_new(TokenSym **pts, const char *str, int len)
{
TokenSym *ts, **ptable;
int i;
if (tok_ident >= SYM_FIRST_ANOM)
tcc_error("memory full (symbols)");
/* expand token table if needed */
i = tok_ident - TOK_IDENT;
if ((i % TOK_ALLOC_INCR) == 0) {
ptable = tcc_realloc(table_ident, (i + TOK_ALLOC_INCR) * sizeof(TokenSym *));
table_ident = ptable;
}
ts = tal_realloc(toksym_alloc, 0, sizeof(TokenSym) + len);
table_ident[i] = ts;
ts->tok = tok_ident++;
ts->sym_define = NULL;
ts->sym_label = NULL;
ts->sym_struct = NULL;
ts->sym_identifier = NULL;
ts->len = len;
ts->hash_next = NULL;
memcpy(ts->str, str, len);
ts->str[len] = '\0';
*pts = ts;
return ts;
}
#define TOK_HASH_INIT 1
#define TOK_HASH_FUNC(h, c) ((h) + ((h) << 5) + ((h) >> 27) + (c))
/* find a token and add it if not found */
ST_FUNC TokenSym *tok_alloc(const char *str, int len)
{
TokenSym *ts, **pts;
int i;
unsigned int h;
h = TOK_HASH_INIT;
for(i=0;i<len;i++)
h = TOK_HASH_FUNC(h, ((unsigned char *)str)[i]);
h &= (TOK_HASH_SIZE - 1);
pts = &hash_ident[h];
for(;;) {
ts = *pts;
if (!ts)
break;
if (ts->len == len && !memcmp(ts->str, str, len))
return ts;
pts = &(ts->hash_next);
}
return tok_alloc_new(pts, str, len);
}
/* XXX: buffer overflow */
/* XXX: float tokens */
ST_FUNC const char *get_tok_str(int v, CValue *cv)
{
char *p;
int i, len;
cstr_reset(&cstr_buf);
p = cstr_buf.data;
switch(v) {
case TOK_CINT:
case TOK_CUINT:
mutiples fix for _Generic * check that _Generic don't match unsigned char * with char * this case is usefull as with -funsigned-char, 'char *' are unsigned * change VT_LONG so it's now a qualifier VT_LONG are never use for code generation, but only durring parsing state, in _Generic we need to be able to make diference between 'long' and 'long long' So VT_LONG is now use as a type qualifier, it's old behaviour is still here, but we can keep trace of what was a long and what wasn't * add TOK_CLONG and TOK_CULONG tcc was directly converting value like '7171L' into TOK_CLLONG or TOK_CINT depending of the machine architecture. because of that, we was unable to make diference between a long and a long long, which doesn't work with _Generic. So now 7171L is a TOK_CLONG, and we can handle _Generic properly * check that _Generic can make diference between long and long long * uncomment "type match twice" as it should now pass tests on any platforms * add inside_generic global the point of this variable is to use VT_LONG in comparaison only when we are evaluating a _Generic. problem is with my lastest patchs tcc can now make the diference between a 'long long' and a 'long', but in 64 bit stddef.h typedef uint64_t as typedef signed long long int int64_t and stdint.h as unsigned long int, so tcc break when stdint.h and stddef.h are include together. Another solution woud be to modifie include/stddef.h so it define uint64_t as unsigned long int when processor is 64 bit, but this could break some legacy code, so for now, VT_LONG are use only inside generc. * check that _Generic parse first argument correctly * check that _Generic evaluate correctly exresion like "f() / 2"
2017-07-10 18:44:53 +03:00
case TOK_CLONG:
case TOK_CULONG:
case TOK_CLLONG:
case TOK_CULLONG:
/* XXX: not quite exact, but only useful for testing */
#ifdef _WIN32
sprintf(p, "%u", (unsigned)cv->i);
#else
sprintf(p, "%llu", (unsigned long long)cv->i);
#endif
break;
case TOK_LCHAR:
cstr_ccat(&cstr_buf, 'L');
case TOK_CCHAR:
cstr_ccat(&cstr_buf, '\'');
add_char(&cstr_buf, cv->i);
cstr_ccat(&cstr_buf, '\'');
cstr_ccat(&cstr_buf, '\0');
break;
case TOK_PPNUM:
case TOK_PPSTR:
return (char*)cv->str.data;
case TOK_LSTR:
cstr_ccat(&cstr_buf, 'L');
case TOK_STR:
cstr_ccat(&cstr_buf, '\"');
if (v == TOK_STR) {
len = cv->str.size - 1;
for(i=0;i<len;i++)
add_char(&cstr_buf, ((unsigned char *)cv->str.data)[i]);
} else {
len = (cv->str.size / sizeof(nwchar_t)) - 1;
for(i=0;i<len;i++)
add_char(&cstr_buf, ((nwchar_t *)cv->str.data)[i]);
}
cstr_ccat(&cstr_buf, '\"');
cstr_ccat(&cstr_buf, '\0');
break;
case TOK_CFLOAT:
cstr_cat(&cstr_buf, "<float>", 0);
break;
case TOK_CDOUBLE:
cstr_cat(&cstr_buf, "<double>", 0);
break;
case TOK_CLDOUBLE:
cstr_cat(&cstr_buf, "<long double>", 0);
break;
case TOK_LINENUM:
cstr_cat(&cstr_buf, "<linenumber>", 0);
break;
/* above tokens have value, the ones below don't */
case TOK_LT:
v = '<';
goto addv;
case TOK_GT:
v = '>';
goto addv;
case TOK_DOTS:
return strcpy(p, "...");
case TOK_A_SHL:
return strcpy(p, "<<=");
case TOK_A_SAR:
return strcpy(p, ">>=");
case TOK_EOF:
return strcpy(p, "<eof>");
default:
if (v < TOK_IDENT) {
/* search in two bytes table */
2009-07-19 00:07:42 +04:00
const unsigned char *q = tok_two_chars;
while (*q) {
if (q[2] == v) {
*p++ = q[0];
*p++ = q[1];
*p = '\0';
return cstr_buf.data;
}
q += 3;
}
if (v >= 127) {
sprintf(cstr_buf.data, "<%02x>", v);
return cstr_buf.data;
}
addv:
*p++ = v;
*p = '\0';
} else if (v < tok_ident) {
return table_ident[v - TOK_IDENT]->str;
} else if (v >= SYM_FIRST_ANOM) {
/* special name for anonymous symbol */
sprintf(p, "L.%u", v - SYM_FIRST_ANOM);
} else {
/* should never happen */
return NULL;
}
break;
}
return cstr_buf.data;
}
/* return the current character, handling end of block if necessary
(but not stray) */
static int handle_eob(void)
2009-05-05 22:17:49 +04:00
{
BufferedFile *bf = file;
2009-05-05 22:17:49 +04:00
int len;
2009-05-05 22:17:49 +04:00
/* only tries to read if really end of buffer */
if (bf->buf_ptr >= bf->buf_end) {
various stuff win32/Makefile ("for cygwin") removed - On cygwin, the normal ./configure && make can be used with either cygwin's "GCC for Win32 Toolchain" ./configure --cross-prefix=i686-w64-mingw32- or with an existing tcc: ./configure --cc=<old-tccdir>/tcc.exe tcctest.c: - exclude test_high_clobbers() on _WIN64 (does not work) tests2/95_bitfield.c: - use 'signed char' for ARM (where default 'char' is unsigned) tests: - remove -I "expr" diff option to allow tests with busybox-diff. libtcc.c, tcc.c: - removed -iwithprefix option. It is supposed to be combined with -iprefix which we don't have either. tccgen.c: - fix assignments and return of 'void', as in void f() { void *p, *q; *p = *q: return *p; } This appears to be allowed but should do nothing. tcc.h, libtcc.c, tccpp.c: - Revert "Introduce VIP sysinclude paths which are always searched first" This reverts commit 1d5e386b0a78393ac6b670c209a185849ec798a1. The patch was giving tcc's system includes priority over -I which is not how it should be. tccelf.c: - add DT_TEXTREL tag only if text relocations are actually used (which is likely not the case on x86_64) - prepare_dynamic_rel(): avoid relocation of unresolved (weak) symbols tccrun.c: - for HAVE_SELINUX, use two mappings to the same (real) file. (it was so once except the RX mapping wasn't used at all). tccpe.c: - fix relocation constant used for x86_64 (by Andrei E. Warentin) - #ifndef _WIN32 do "chmod 755 ..." to get runnable exes on cygwin. tccasm.c: - keep forward asm labels static, otherwise they will endup in dynsym eventually. configure, Makefile: - mingw32: respect ./configure options --bindir --docdir --libdir - allow overriding tcc when building libtcc1.a and libtcc.def with make XTCC=<tcc program to use> - use $(wildcard ...) for install to allow installing just a cross compiler for example make cross-arm make install - use name <target>-libtcc1.a build-tcc.bat: - add options: -clean, -b bindir
2017-10-11 19:13:43 +03:00
if (bf->fd >= 0) {
2009-05-05 22:17:49 +04:00
#if defined(PARSE_DEBUG)
len = 1;
2009-05-05 22:17:49 +04:00
#else
len = IO_BUF_SIZE;
#endif
len = read(bf->fd, bf->buffer, len);
2009-05-05 22:17:49 +04:00
if (len < 0)
len = 0;
} else {
len = 0;
}
total_bytes += len;
bf->buf_ptr = bf->buffer;
bf->buf_end = bf->buffer + len;
*bf->buf_end = CH_EOB;
}
if (bf->buf_ptr < bf->buf_end) {
return bf->buf_ptr[0];
} else {
bf->buf_ptr = bf->buf_end;
return CH_EOF;
}
}
/* read next char from current input file and handle end of input buffer */
static inline void inp(void)
2009-05-05 22:17:49 +04:00
{
ch = *(++(file->buf_ptr));
/* end of buffer/file handling */
if (ch == CH_EOB)
ch = handle_eob();
}
/* handle '\[\r]\n' */
static int handle_stray_noerror(void)
{
while (ch == '\\') {
inp();
if (ch == '\n') {
file->line_num++;
inp();
} else if (ch == '\r') {
inp();
if (ch != '\n')
goto fail;
file->line_num++;
inp();
} else {
fail:
return 1;
}
}
return 0;
}
static void handle_stray(void)
{
if (handle_stray_noerror())
tcc_error("stray '\\' in program");
2009-05-05 22:17:49 +04:00
}
/* skip the stray and handle the \\n case. Output an error if
incorrect char after the stray */
static int handle_stray1(uint8_t *p)
{
int c;
file->buf_ptr = p;
2009-05-05 22:17:49 +04:00
if (p >= file->buf_end) {
c = handle_eob();
if (c != '\\')
return c;
2009-05-05 22:17:49 +04:00
p = file->buf_ptr;
}
ch = *p;
if (handle_stray_noerror()) {
if (!(parse_flags & PARSE_FLAG_ACCEPT_STRAYS))
tcc_error("stray '\\' in program");
*--file->buf_ptr = '\\';
}
p = file->buf_ptr;
c = *p;
2009-05-05 22:17:49 +04:00
return c;
}
/* handle just the EOB case, but not stray */
#define PEEKC_EOB(c, p)\
{\
p++;\
c = *p;\
if (c == '\\') {\
file->buf_ptr = p;\
c = handle_eob();\
p = file->buf_ptr;\
}\
}
/* handle the complicated stray case */
#define PEEKC(c, p)\
{\
p++;\
c = *p;\
if (c == '\\') {\
c = handle_stray1(p);\
p = file->buf_ptr;\
}\
}
/* input with '\[\r]\n' handling. Note that this function cannot
handle other characters after '\', so you cannot call it inside
strings or comments */
static void minp(void)
2009-05-05 22:17:49 +04:00
{
inp();
if (ch == '\\')
2009-05-05 22:17:49 +04:00
handle_stray();
}
/* single line C++ comments */
static uint8_t *parse_line_comment(uint8_t *p)
2009-05-05 22:17:49 +04:00
{
int c;
p++;
for(;;) {
c = *p;
redo:
if (c == '\n' || c == CH_EOF) {
break;
} else if (c == '\\') {
file->buf_ptr = p;
c = handle_eob();
p = file->buf_ptr;
if (c == '\\') {
PEEKC_EOB(c, p);
if (c == '\n') {
file->line_num++;
PEEKC_EOB(c, p);
} else if (c == '\r') {
PEEKC_EOB(c, p);
if (c == '\n') {
file->line_num++;
PEEKC_EOB(c, p);
}
}
} else {
goto redo;
}
} else {
p++;
}
}
return p;
}
/* C comments */
static uint8_t *parse_comment(uint8_t *p)
2009-05-05 22:17:49 +04:00
{
int c;
2009-05-05 22:17:49 +04:00
p++;
for(;;) {
/* fast skip loop */
for(;;) {
c = *p;
if (c == '\n' || c == '*' || c == '\\')
break;
p++;
c = *p;
if (c == '\n' || c == '*' || c == '\\')
break;
p++;
}
/* now we can handle all the cases */
if (c == '\n') {
file->line_num++;
p++;
} else if (c == '*') {
p++;
for(;;) {
c = *p;
if (c == '*') {
p++;
} else if (c == '/') {
goto end_of_comment;
} else if (c == '\\') {
file->buf_ptr = p;
c = handle_eob();
p = file->buf_ptr;
if (c == CH_EOF)
tcc_error("unexpected end of file in comment");
2009-05-05 22:17:49 +04:00
if (c == '\\') {
/* skip '\[\r]\n', otherwise just skip the stray */
while (c == '\\') {
PEEKC_EOB(c, p);
if (c == '\n') {
file->line_num++;
PEEKC_EOB(c, p);
} else if (c == '\r') {
PEEKC_EOB(c, p);
if (c == '\n') {
file->line_num++;
PEEKC_EOB(c, p);
}
} else {
goto after_star;
}
}
}
} else {
break;
}
}
after_star: ;
} else {
/* stray, eob or eof */
file->buf_ptr = p;
c = handle_eob();
p = file->buf_ptr;
if (c == CH_EOF) {
tcc_error("unexpected end of file in comment");
2009-05-05 22:17:49 +04:00
} else if (c == '\\') {
p++;
}
}
}
end_of_comment:
p++;
return p;
}
ST_FUNC int set_idnum(int c, int val)
{
int prev = isidnum_table[c - CH_EOF];
isidnum_table[c - CH_EOF] = val;
return prev;
}
2009-05-05 22:17:49 +04:00
#define cinp minp
static inline void skip_spaces(void)
{
while (isidnum_table[ch - CH_EOF] & IS_SPC)
2009-05-05 22:17:49 +04:00
cinp();
}
static inline int check_space(int t, int *spc)
2009-05-05 22:17:49 +04:00
{
if (t < 256 && (isidnum_table[t - CH_EOF] & IS_SPC)) {
if (*spc)
2009-05-05 22:17:49 +04:00
return 1;
*spc = 1;
} else
2009-05-05 22:17:49 +04:00
*spc = 0;
return 0;
}
/* parse a string without interpreting escapes */
static uint8_t *parse_pp_string(uint8_t *p,
int sep, CString *str)
{
int c;
p++;
for(;;) {
c = *p;
if (c == sep) {
break;
} else if (c == '\\') {
file->buf_ptr = p;
c = handle_eob();
p = file->buf_ptr;
if (c == CH_EOF) {
unterminated_string:
/* XXX: indicate line number of start of string */
tcc_error("missing terminating %c character", sep);
2009-05-05 22:17:49 +04:00
} else if (c == '\\') {
/* escape : just skip \[\r]\n */
PEEKC_EOB(c, p);
if (c == '\n') {
file->line_num++;
p++;
} else if (c == '\r') {
PEEKC_EOB(c, p);
if (c != '\n')
expect("'\n' after '\r'");
file->line_num++;
p++;
} else if (c == CH_EOF) {
goto unterminated_string;
} else {
if (str) {
cstr_ccat(str, '\\');
cstr_ccat(str, c);
}
p++;
}
}
} else if (c == '\n') {
file->line_num++;
goto add_char;
} else if (c == '\r') {
PEEKC_EOB(c, p);
if (c != '\n') {
if (str)
cstr_ccat(str, '\r');
} else {
file->line_num++;
goto add_char;
}
} else {
add_char:
if (str)
cstr_ccat(str, c);
p++;
}
}
p++;
return p;
}
/* skip block of text until #else, #elif or #endif. skip also pairs of
#if/#endif */
static void preprocess_skip(void)
2009-05-05 22:17:49 +04:00
{
int a, start_of_line, c, in_warn_or_error;
uint8_t *p;
p = file->buf_ptr;
a = 0;
redo_start:
start_of_line = 1;
in_warn_or_error = 0;
for(;;) {
redo_no_start:
c = *p;
switch(c) {
case ' ':
case '\t':
case '\f':
case '\v':
case '\r':
p++;
goto redo_no_start;
case '\n':
file->line_num++;
p++;
goto redo_start;
case '\\':
file->buf_ptr = p;
c = handle_eob();
if (c == CH_EOF) {
expect("#endif");
} else if (c == '\\') {
ch = file->buf_ptr[0];
handle_stray_noerror();
}
p = file->buf_ptr;
goto redo_no_start;
/* skip strings */
case '\"':
case '\'':
if (in_warn_or_error)
goto _default;
p = parse_pp_string(p, c, NULL);
break;
/* skip comments */
case '/':
if (in_warn_or_error)
goto _default;
file->buf_ptr = p;
ch = *p;
minp();
p = file->buf_ptr;
if (ch == '*') {
p = parse_comment(p);
2009-05-05 22:17:49 +04:00
} else if (ch == '/') {
p = parse_line_comment(p);
2009-05-05 22:17:49 +04:00
}
break;
case '#':
p++;
if (start_of_line) {
file->buf_ptr = p;
next_nomacro();
p = file->buf_ptr;
if (a == 0 &&
2009-05-05 22:17:49 +04:00
(tok == TOK_ELSE || tok == TOK_ELIF || tok == TOK_ENDIF))
goto the_end;
if (tok == TOK_IF || tok == TOK_IFDEF || tok == TOK_IFNDEF)
a++;
else if (tok == TOK_ENDIF)
a--;
else if( tok == TOK_ERROR || tok == TOK_WARNING)
in_warn_or_error = 1;
else if (tok == TOK_LINEFEED)
goto redo_start;
else if (parse_flags & PARSE_FLAG_ASM_FILE)
p = parse_line_comment(p - 1);
} else if (parse_flags & PARSE_FLAG_ASM_FILE)
p = parse_line_comment(p - 1);
2009-05-05 22:17:49 +04:00
break;
_default:
default:
p++;
break;
}
start_of_line = 0;
}
the_end: ;
file->buf_ptr = p;
}
#if 0
/* return the number of additional 'ints' necessary to store the
token */
static inline int tok_size(const int *p)
{
switch(*p) {
/* 4 bytes */
case TOK_CINT:
case TOK_CUINT:
case TOK_CCHAR:
case TOK_LCHAR:
case TOK_CFLOAT:
case TOK_LINENUM:
return 1 + 1;
case TOK_STR:
case TOK_LSTR:
case TOK_PPNUM:
case TOK_PPSTR:
return 1 + ((sizeof(CString) + ((CString *)(p+1))->size + 3) >> 2);
case TOK_CLONG:
case TOK_CULONG:
return 1 + LONG_SIZE / 4;
case TOK_CDOUBLE:
case TOK_CLLONG:
case TOK_CULLONG:
return 1 + 2;
case TOK_CLDOUBLE:
return 1 + LDOUBLE_SIZE / 4;
default:
return 1 + 0;
}
}
#endif
2009-05-05 22:17:49 +04:00
/* token string handling */
ST_INLN void tok_str_new(TokenString *s)
2009-05-05 22:17:49 +04:00
{
s->str = NULL;
s->len = s->lastlen = 0;
2009-05-05 22:17:49 +04:00
s->allocated_len = 0;
s->last_line_num = -1;
}
ST_FUNC TokenString *tok_str_alloc(void)
{
TokenString *str = tal_realloc(tokstr_alloc, 0, sizeof *str);
tok_str_new(str);
return str;
}
ST_FUNC int *tok_str_dup(TokenString *s)
{
int *str;
str = tal_realloc(tokstr_alloc, 0, s->len * sizeof(int));
memcpy(str, s->str, s->len * sizeof(int));
return str;
}
ST_FUNC void tok_str_free_str(int *str)
2009-05-05 22:17:49 +04:00
{
tal_free(tokstr_alloc, str);
2009-05-05 22:17:49 +04:00
}
ST_FUNC void tok_str_free(TokenString *str)
{
tok_str_free_str(str->str);
tal_free(tokstr_alloc, str);
}
ST_FUNC int *tok_str_realloc(TokenString *s, int new_size)
2009-05-05 22:17:49 +04:00
{
int *str, size;
2009-05-05 22:17:49 +04:00
size = s->allocated_len;
if (size < 16)
size = 16;
while (size < new_size)
size = size * 2;
if (size > s->allocated_len) {
str = tal_realloc(tokstr_alloc, s->str, size * sizeof(int));
s->allocated_len = size;
s->str = str;
2009-05-05 22:17:49 +04:00
}
return s->str;
2009-05-05 22:17:49 +04:00
}
ST_FUNC void tok_str_add(TokenString *s, int t)
2009-05-05 22:17:49 +04:00
{
int len, *str;
len = s->len;
str = s->str;
if (len >= s->allocated_len)
str = tok_str_realloc(s, len + 1);
2009-05-05 22:17:49 +04:00
str[len++] = t;
s->len = len;
}
ST_FUNC void begin_macro(TokenString *str, int alloc)
{
str->alloc = alloc;
str->prev = macro_stack;
str->prev_ptr = macro_ptr;
str->save_line_num = file->line_num;
macro_ptr = str->str;
macro_stack = str;
}
ST_FUNC void end_macro(void)
{
TokenString *str = macro_stack;
macro_stack = str->prev;
macro_ptr = str->prev_ptr;
file->line_num = str->save_line_num;
if (str->alloc != 0) {
if (str->alloc == 2)
str->str = NULL; /* don't free */
tok_str_free(str);
}
}
2009-05-05 22:17:49 +04:00
static void tok_str_add2(TokenString *s, int t, CValue *cv)
{
int len, *str;
len = s->lastlen = s->len;
2009-05-05 22:17:49 +04:00
str = s->str;
/* allocate space for worst case */
if (len + TOK_MAX_SIZE >= s->allocated_len)
str = tok_str_realloc(s, len + TOK_MAX_SIZE + 1);
2009-05-05 22:17:49 +04:00
str[len++] = t;
switch(t) {
case TOK_CINT:
case TOK_CUINT:
case TOK_CCHAR:
case TOK_LCHAR:
case TOK_CFLOAT:
case TOK_LINENUM:
#if LONG_SIZE == 4
mutiples fix for _Generic * check that _Generic don't match unsigned char * with char * this case is usefull as with -funsigned-char, 'char *' are unsigned * change VT_LONG so it's now a qualifier VT_LONG are never use for code generation, but only durring parsing state, in _Generic we need to be able to make diference between 'long' and 'long long' So VT_LONG is now use as a type qualifier, it's old behaviour is still here, but we can keep trace of what was a long and what wasn't * add TOK_CLONG and TOK_CULONG tcc was directly converting value like '7171L' into TOK_CLLONG or TOK_CINT depending of the machine architecture. because of that, we was unable to make diference between a long and a long long, which doesn't work with _Generic. So now 7171L is a TOK_CLONG, and we can handle _Generic properly * check that _Generic can make diference between long and long long * uncomment "type match twice" as it should now pass tests on any platforms * add inside_generic global the point of this variable is to use VT_LONG in comparaison only when we are evaluating a _Generic. problem is with my lastest patchs tcc can now make the diference between a 'long long' and a 'long', but in 64 bit stddef.h typedef uint64_t as typedef signed long long int int64_t and stdint.h as unsigned long int, so tcc break when stdint.h and stddef.h are include together. Another solution woud be to modifie include/stddef.h so it define uint64_t as unsigned long int when processor is 64 bit, but this could break some legacy code, so for now, VT_LONG are use only inside generc. * check that _Generic parse first argument correctly * check that _Generic evaluate correctly exresion like "f() / 2"
2017-07-10 18:44:53 +03:00
case TOK_CLONG:
case TOK_CULONG:
#endif
2009-05-05 22:17:49 +04:00
str[len++] = cv->tab[0];
break;
case TOK_PPNUM:
case TOK_PPSTR:
2009-05-05 22:17:49 +04:00
case TOK_STR:
case TOK_LSTR:
{
/* Insert the string into the int array. */
size_t nb_words =
1 + (cv->str.size + sizeof(int) - 1) / sizeof(int);
if (len + nb_words >= s->allocated_len)
str = tok_str_realloc(s, len + nb_words + 1);
str[len] = cv->str.size;
memcpy(&str[len + 1], cv->str.data, cv->str.size);
2009-05-05 22:17:49 +04:00
len += nb_words;
}
break;
case TOK_CDOUBLE:
case TOK_CLLONG:
case TOK_CULLONG:
#if LONG_SIZE == 8
mutiples fix for _Generic * check that _Generic don't match unsigned char * with char * this case is usefull as with -funsigned-char, 'char *' are unsigned * change VT_LONG so it's now a qualifier VT_LONG are never use for code generation, but only durring parsing state, in _Generic we need to be able to make diference between 'long' and 'long long' So VT_LONG is now use as a type qualifier, it's old behaviour is still here, but we can keep trace of what was a long and what wasn't * add TOK_CLONG and TOK_CULONG tcc was directly converting value like '7171L' into TOK_CLLONG or TOK_CINT depending of the machine architecture. because of that, we was unable to make diference between a long and a long long, which doesn't work with _Generic. So now 7171L is a TOK_CLONG, and we can handle _Generic properly * check that _Generic can make diference between long and long long * uncomment "type match twice" as it should now pass tests on any platforms * add inside_generic global the point of this variable is to use VT_LONG in comparaison only when we are evaluating a _Generic. problem is with my lastest patchs tcc can now make the diference between a 'long long' and a 'long', but in 64 bit stddef.h typedef uint64_t as typedef signed long long int int64_t and stdint.h as unsigned long int, so tcc break when stdint.h and stddef.h are include together. Another solution woud be to modifie include/stddef.h so it define uint64_t as unsigned long int when processor is 64 bit, but this could break some legacy code, so for now, VT_LONG are use only inside generc. * check that _Generic parse first argument correctly * check that _Generic evaluate correctly exresion like "f() / 2"
2017-07-10 18:44:53 +03:00
case TOK_CLONG:
case TOK_CULONG:
#endif
2009-05-05 22:17:49 +04:00
#if LDOUBLE_SIZE == 8
case TOK_CLDOUBLE:
#endif
str[len++] = cv->tab[0];
str[len++] = cv->tab[1];
break;
#if LDOUBLE_SIZE == 12
case TOK_CLDOUBLE:
str[len++] = cv->tab[0];
str[len++] = cv->tab[1];
str[len++] = cv->tab[2];
#elif LDOUBLE_SIZE == 16
case TOK_CLDOUBLE:
str[len++] = cv->tab[0];
str[len++] = cv->tab[1];
str[len++] = cv->tab[2];
str[len++] = cv->tab[3];
#elif LDOUBLE_SIZE != 8
#error add long double size support
#endif
break;
default:
break;
}
s->len = len;
}
/* add the current parse token in token string 's' */
ST_FUNC void tok_str_add_tok(TokenString *s)
2009-05-05 22:17:49 +04:00
{
CValue cval;
/* save line number info */
if (file->line_num != s->last_line_num) {
s->last_line_num = file->line_num;
cval.i = s->last_line_num;
tok_str_add2(s, TOK_LINENUM, &cval);
}
tok_str_add2(s, tok, &tokc);
}
2019-01-13 20:16:46 +03:00
/* get a token from an integer array and increment pointer. */
static inline void TOK_GET(int *t, const int **pp, CValue *cv)
{
const int *p = *pp;
int n, *tab;
tab = cv->tab;
switch(*t = *p++) {
#if LONG_SIZE == 4
case TOK_CLONG:
#endif
case TOK_CINT:
case TOK_CCHAR:
case TOK_LCHAR:
case TOK_LINENUM:
cv->i = *p++;
break;
#if LONG_SIZE == 4
mutiples fix for _Generic * check that _Generic don't match unsigned char * with char * this case is usefull as with -funsigned-char, 'char *' are unsigned * change VT_LONG so it's now a qualifier VT_LONG are never use for code generation, but only durring parsing state, in _Generic we need to be able to make diference between 'long' and 'long long' So VT_LONG is now use as a type qualifier, it's old behaviour is still here, but we can keep trace of what was a long and what wasn't * add TOK_CLONG and TOK_CULONG tcc was directly converting value like '7171L' into TOK_CLLONG or TOK_CINT depending of the machine architecture. because of that, we was unable to make diference between a long and a long long, which doesn't work with _Generic. So now 7171L is a TOK_CLONG, and we can handle _Generic properly * check that _Generic can make diference between long and long long * uncomment "type match twice" as it should now pass tests on any platforms * add inside_generic global the point of this variable is to use VT_LONG in comparaison only when we are evaluating a _Generic. problem is with my lastest patchs tcc can now make the diference between a 'long long' and a 'long', but in 64 bit stddef.h typedef uint64_t as typedef signed long long int int64_t and stdint.h as unsigned long int, so tcc break when stdint.h and stddef.h are include together. Another solution woud be to modifie include/stddef.h so it define uint64_t as unsigned long int when processor is 64 bit, but this could break some legacy code, so for now, VT_LONG are use only inside generc. * check that _Generic parse first argument correctly * check that _Generic evaluate correctly exresion like "f() / 2"
2017-07-10 18:44:53 +03:00
case TOK_CULONG:
#endif
case TOK_CUINT:
cv->i = (unsigned)*p++;
break;
case TOK_CFLOAT:
tab[0] = *p++;
break;
case TOK_STR:
case TOK_LSTR:
case TOK_PPNUM:
case TOK_PPSTR:
cv->str.size = *p++;
cv->str.data = p;
p += (cv->str.size + sizeof(int) - 1) / sizeof(int);
break;
case TOK_CDOUBLE:
case TOK_CLLONG:
case TOK_CULLONG:
#if LONG_SIZE == 8
mutiples fix for _Generic * check that _Generic don't match unsigned char * with char * this case is usefull as with -funsigned-char, 'char *' are unsigned * change VT_LONG so it's now a qualifier VT_LONG are never use for code generation, but only durring parsing state, in _Generic we need to be able to make diference between 'long' and 'long long' So VT_LONG is now use as a type qualifier, it's old behaviour is still here, but we can keep trace of what was a long and what wasn't * add TOK_CLONG and TOK_CULONG tcc was directly converting value like '7171L' into TOK_CLLONG or TOK_CINT depending of the machine architecture. because of that, we was unable to make diference between a long and a long long, which doesn't work with _Generic. So now 7171L is a TOK_CLONG, and we can handle _Generic properly * check that _Generic can make diference between long and long long * uncomment "type match twice" as it should now pass tests on any platforms * add inside_generic global the point of this variable is to use VT_LONG in comparaison only when we are evaluating a _Generic. problem is with my lastest patchs tcc can now make the diference between a 'long long' and a 'long', but in 64 bit stddef.h typedef uint64_t as typedef signed long long int int64_t and stdint.h as unsigned long int, so tcc break when stdint.h and stddef.h are include together. Another solution woud be to modifie include/stddef.h so it define uint64_t as unsigned long int when processor is 64 bit, but this could break some legacy code, so for now, VT_LONG are use only inside generc. * check that _Generic parse first argument correctly * check that _Generic evaluate correctly exresion like "f() / 2"
2017-07-10 18:44:53 +03:00
case TOK_CLONG:
case TOK_CULONG:
#endif
n = 2;
goto copy;
case TOK_CLDOUBLE:
2009-05-05 22:17:49 +04:00
#if LDOUBLE_SIZE == 16
n = 4;
2009-05-05 22:17:49 +04:00
#elif LDOUBLE_SIZE == 12
n = 3;
2009-05-05 22:17:49 +04:00
#elif LDOUBLE_SIZE == 8
n = 2;
2009-05-05 22:17:49 +04:00
#else
# error add long double size support
2009-05-05 22:17:49 +04:00
#endif
copy:
do
*tab++ = *p++;
while (--n);
break;
default:
break;
}
*pp = p;
2009-05-05 22:17:49 +04:00
}
2010-01-14 22:57:50 +03:00
static int macro_is_equal(const int *a, const int *b)
{
CValue cv;
int t;
if (!a || !b)
return 1;
2010-01-14 22:57:50 +03:00
while (*a && *b) {
/* first time preallocate macro_equal_buf, next time only reset position to start */
cstr_reset(&macro_equal_buf);
TOK_GET(&t, &a, &cv);
cstr_cat(&macro_equal_buf, get_tok_str(t, &cv), 0);
TOK_GET(&t, &b, &cv);
if (strcmp(macro_equal_buf.data, get_tok_str(t, &cv)))
2010-01-14 22:57:50 +03:00
return 0;
}
return !(*a || *b);
}
2009-05-05 22:17:49 +04:00
/* defines handling */
ST_INLN void define_push(int v, int macro_type, int *str, Sym *first_arg)
2009-05-05 22:17:49 +04:00
{
Sym *s, *o;
2010-01-14 22:57:50 +03:00
o = define_find(v);
s = sym_push2(&define_stack, v, macro_type, 0);
s->d = str;
2009-05-05 22:17:49 +04:00
s->next = first_arg;
table_ident[v - TOK_IDENT]->sym_define = s;
if (o && !macro_is_equal(o->d, s->d))
tcc_warning("%s redefined", get_tok_str(v, NULL));
2009-05-05 22:17:49 +04:00
}
/* undefined a define symbol. Its name is just set to zero */
ST_FUNC void define_undef(Sym *s)
2009-05-05 22:17:49 +04:00
{
int v = s->v;
if (v >= TOK_IDENT && v < tok_ident)
2009-05-05 22:17:49 +04:00
table_ident[v - TOK_IDENT]->sym_define = NULL;
}
ST_INLN Sym *define_find(int v)
2009-05-05 22:17:49 +04:00
{
v -= TOK_IDENT;
if ((unsigned)v >= (unsigned)(tok_ident - TOK_IDENT))
return NULL;
return table_ident[v]->sym_define;
}
/* free define stack until top reaches 'b' */
ST_FUNC void free_defines(Sym *b)
2009-05-05 22:17:49 +04:00
{
while (define_stack != b) {
Sym *top = define_stack;
define_stack = top->prev;
tok_str_free_str(top->d);
define_undef(top);
2009-05-05 22:17:49 +04:00
sym_free(top);
}
}
/* label lookup */
ST_FUNC Sym *label_find(int v)
2009-05-05 22:17:49 +04:00
{
v -= TOK_IDENT;
if ((unsigned)v >= (unsigned)(tok_ident - TOK_IDENT))
return NULL;
return table_ident[v]->sym_label;
}
ST_FUNC Sym *label_push(Sym **ptop, int v, int flags)
2009-05-05 22:17:49 +04:00
{
Sym *s, **ps;
s = sym_push2(ptop, v, 0, 0);
s->r = flags;
ps = &table_ident[v - TOK_IDENT]->sym_label;
if (ptop == &global_label_stack) {
/* modify the top most local identifier, so that
sym_identifier will point to 's' when popped */
while (*ps != NULL)
ps = &(*ps)->prev_tok;
}
s->prev_tok = *ps;
*ps = s;
return s;
}
/* pop labels until element last is reached. Look if any labels are
undefined. Define symbols if '&&label' was used. */
ST_FUNC void label_pop(Sym **ptop, Sym *slast, int keep)
2009-05-05 22:17:49 +04:00
{
Sym *s, *s1;
for(s = *ptop; s != slast; s = s1) {
s1 = s->prev;
if (s->r == LABEL_DECLARED) {
tcc_warning("label '%s' declared but not used", get_tok_str(s->v, NULL));
2009-05-05 22:17:49 +04:00
} else if (s->r == LABEL_FORWARD) {
tcc_error("label '%s' used but not defined",
2009-05-05 22:17:49 +04:00
get_tok_str(s->v, NULL));
} else {
if (s->c) {
/* define corresponding symbol. A size of
1 is put. */
put_extern_sym(s, cur_text_section, s->jnext, 1);
2009-05-05 22:17:49 +04:00
}
}
/* remove label */
if (s->r != LABEL_GONE)
table_ident[s->v - TOK_IDENT]->sym_label = s->prev_tok;
if (!keep)
sym_free(s);
else
s->r = LABEL_GONE;
2009-05-05 22:17:49 +04:00
}
if (!keep)
*ptop = slast;
2009-05-05 22:17:49 +04:00
}
/* fake the nth "#if defined test_..." for tcc -dt -run */
static void maybe_run_test(TCCState *s)
{
const char *p;
if (s->include_stack_ptr != s->include_stack)
return;
p = get_tok_str(tok, NULL);
if (0 != memcmp(p, "test_", 5))
return;
if (0 != --s->run_test)
return;
fprintf(s->ppfp, "\n[%s]\n" + !(s->dflag & 32), p), fflush(s->ppfp);
define_push(tok, MACRO_OBJ, NULL, NULL);
}
2009-05-05 22:17:49 +04:00
/* eval an expression for #if/#elif */
static int expr_preprocess(void)
{
int c, t;
TokenString *str;
str = tok_str_alloc();
pp_expr = 1;
2009-05-05 22:17:49 +04:00
while (tok != TOK_LINEFEED && tok != TOK_EOF) {
next(); /* do macro subst */
if (tok == TOK_DEFINED) {
next_nomacro();
t = tok;
if (t == '(')
2009-05-05 22:17:49 +04:00
next_nomacro();
if (tok < TOK_IDENT)
expect("identifier");
if (tcc_state->run_test)
maybe_run_test(tcc_state);
c = define_find(tok) != 0;
if (t == '(') {
2009-05-05 22:17:49 +04:00
next_nomacro();
if (tok != ')')
expect("')'");
}
2009-05-05 22:17:49 +04:00
tok = TOK_CINT;
tokc.i = c;
} else if (tok >= TOK_IDENT) {
/* if undefined macro */
tok = TOK_CINT;
tokc.i = 0;
}
tok_str_add_tok(str);
2009-05-05 22:17:49 +04:00
}
pp_expr = 0;
tok_str_add(str, -1); /* simulate end of file */
tok_str_add(str, 0);
2009-05-05 22:17:49 +04:00
/* now evaluate C constant expression */
begin_macro(str, 1);
2009-05-05 22:17:49 +04:00
next();
c = expr_const();
end_macro();
2009-05-05 22:17:49 +04:00
return c != 0;
}
2009-05-05 22:17:49 +04:00
/* parse after #define */
ST_FUNC void parse_define(void)
2009-05-05 22:17:49 +04:00
{
Sym *s, *first, **ps;
int v, t, varg, is_vaargs, spc;
int saved_parse_flags = parse_flags;
2009-05-05 22:17:49 +04:00
v = tok;
if (v < TOK_IDENT || v == TOK_DEFINED)
tcc_error("invalid macro name '%s'", get_tok_str(tok, &tokc));
2009-05-05 22:17:49 +04:00
/* XXX: should check if same macro (ANSI) */
first = NULL;
t = MACRO_OBJ;
/* We have to parse the whole define as if not in asm mode, in particular
no line comment with '#' must be ignored. Also for function
macros the argument list must be parsed without '.' being an ID
character. */
parse_flags = ((parse_flags & ~PARSE_FLAG_ASM_FILE) | PARSE_FLAG_SPACES);
2009-05-05 22:17:49 +04:00
/* '(' must be just after macro definition for MACRO_FUNC */
next_nomacro_spc();
if (tok == '(') {
int dotid = set_idnum('.', 0);
2009-05-05 22:17:49 +04:00
next_nomacro();
ps = &first;
if (tok != ')') for (;;) {
2009-05-05 22:17:49 +04:00
varg = tok;
next_nomacro();
is_vaargs = 0;
if (varg == TOK_DOTS) {
varg = TOK___VA_ARGS__;
is_vaargs = 1;
} else if (tok == TOK_DOTS && gnu_ext) {
is_vaargs = 1;
next_nomacro();
}
if (varg < TOK_IDENT)
bad_list:
tcc_error("bad macro parameter list");
2009-05-05 22:17:49 +04:00
s = sym_push2(&define_stack, varg | SYM_FIELD, is_vaargs, 0);
*ps = s;
ps = &s->next;
if (tok == ')')
break;
if (tok != ',' || is_vaargs)
goto bad_list;
2009-05-05 22:17:49 +04:00
next_nomacro();
}
next_nomacro_spc();
2009-05-05 22:17:49 +04:00
t = MACRO_FUNC;
set_idnum('.', dotid);
2009-05-05 22:17:49 +04:00
}
tokstr_buf.len = 0;
2009-05-05 22:17:49 +04:00
spc = 2;
parse_flags |= PARSE_FLAG_ACCEPT_STRAYS | PARSE_FLAG_SPACES | PARSE_FLAG_LINEFEED;
/* The body of a macro definition should be parsed such that identifiers
are parsed like the file mode determines (i.e. with '.' being an
ID character in asm mode). But '#' should be retained instead of
regarded as line comment leader, so still don't set ASM_FILE
in parse_flags. */
2009-05-05 22:17:49 +04:00
while (tok != TOK_LINEFEED && tok != TOK_EOF) {
/* remove spaces around ## and after '#' */
2009-05-05 22:17:49 +04:00
if (TOK_TWOSHARPS == tok) {
if (2 == spc)
goto bad_twosharp;
2009-05-05 22:17:49 +04:00
if (1 == spc)
--tokstr_buf.len;
spc = 3;
tok = TOK_PPJOIN;
2009-05-05 22:17:49 +04:00
} else if ('#' == tok) {
spc = 4;
2009-05-05 22:17:49 +04:00
} else if (check_space(tok, &spc)) {
goto skip;
}
tok_str_add2(&tokstr_buf, tok, &tokc);
2009-05-05 22:17:49 +04:00
skip:
next_nomacro_spc();
}
parse_flags = saved_parse_flags;
2009-05-05 22:17:49 +04:00
if (spc == 1)
--tokstr_buf.len; /* remove trailing space */
tok_str_add(&tokstr_buf, 0);
if (3 == spc)
bad_twosharp:
tcc_error("'##' cannot appear at either end of macro");
define_push(v, t, tok_str_dup(&tokstr_buf), first);
2009-05-05 22:17:49 +04:00
}
static CachedInclude *search_cached_include(TCCState *s1, const char *filename, int add)
2009-05-05 22:17:49 +04:00
{
const unsigned char *s;
unsigned int h;
CachedInclude *e;
int i;
2009-05-05 22:17:49 +04:00
h = TOK_HASH_INIT;
2014-03-09 18:52:31 +04:00
s = (unsigned char *) filename;
2009-05-05 22:17:49 +04:00
while (*s) {
#ifdef _WIN32
h = TOK_HASH_FUNC(h, toup(*s));
#else
2009-05-05 22:17:49 +04:00
h = TOK_HASH_FUNC(h, *s);
#endif
2009-05-05 22:17:49 +04:00
s++;
}
h &= (CACHED_INCLUDES_HASH_SIZE - 1);
i = s1->cached_includes_hash[h];
for(;;) {
if (i == 0)
break;
e = s1->cached_includes[i - 1];
if (0 == PATHCMP(e->filename, filename))
2009-05-05 22:17:49 +04:00
return e;
i = e->hash_next;
}
if (!add)
return NULL;
2009-05-05 22:17:49 +04:00
e = tcc_malloc(sizeof(CachedInclude) + strlen(filename));
strcpy(e->filename, filename);
e->ifndef_macro = e->once = 0;
dynarray_add(&s1->cached_includes, &s1->nb_cached_includes, e);
2009-05-05 22:17:49 +04:00
/* add in hash table */
e->hash_next = s1->cached_includes_hash[h];
s1->cached_includes_hash[h] = s1->nb_cached_includes;
#ifdef INC_DEBUG
printf("adding cached '%s'\n", filename);
#endif
return e;
2009-05-05 22:17:49 +04:00
}
static void pragma_parse(TCCState *s1)
{
next_nomacro();
if (tok == TOK_push_macro || tok == TOK_pop_macro) {
int t = tok, v;
Sym *s;
if (next(), tok != '(')
goto pragma_err;
if (next(), tok != TOK_STR)
goto pragma_err;
v = tok_alloc(tokc.str.data, tokc.str.size - 1)->tok;
if (next(), tok != ')')
goto pragma_err;
if (t == TOK_push_macro) {
while (NULL == (s = define_find(v)))
define_push(v, 0, NULL, NULL);
s->type.ref = s; /* set push boundary */
} else {
for (s = define_stack; s; s = s->prev)
if (s->v == v && s->type.ref == s) {
s->type.ref = NULL;
break;
}
}
if (s)
table_ident[v - TOK_IDENT]->sym_define = s->d ? s : NULL;
else
tcc_warning("unbalanced #pragma pop_macro");
pp_debug_tok = t, pp_debug_symv = v;
} else if (tok == TOK_once) {
search_cached_include(s1, file->filename, 1)->once = pp_once;
} else if (s1->output_type == TCC_OUTPUT_PREPROCESS) {
/* tcc -E: keep pragmas below unchanged */
unget_tok(' ');
unget_tok(TOK_PRAGMA);
unget_tok('#');
unget_tok(TOK_LINEFEED);
} else if (tok == TOK_pack) {
/* This may be:
#pragma pack(1) // set
#pragma pack() // reset to default
#pragma pack(push,1) // push & set
#pragma pack(pop) // restore previous */
2009-05-05 22:17:49 +04:00
next();
skip('(');
if (tok == TOK_ASM_pop) {
next();
if (s1->pack_stack_ptr <= s1->pack_stack) {
stk_error:
tcc_error("out of pack stack");
2009-05-05 22:17:49 +04:00
}
s1->pack_stack_ptr--;
} else {
int val = 0;
2009-05-05 22:17:49 +04:00
if (tok != ')') {
if (tok == TOK_ASM_push) {
next();
if (s1->pack_stack_ptr >= s1->pack_stack + PACK_STACK_SIZE - 1)
goto stk_error;
s1->pack_stack_ptr++;
skip(',');
}
if (tok != TOK_CINT)
goto pragma_err;
2009-05-05 22:17:49 +04:00
val = tokc.i;
if (val < 1 || val > 16 || (val & (val - 1)) != 0)
goto pragma_err;
2009-05-05 22:17:49 +04:00
next();
}
*s1->pack_stack_ptr = val;
}
if (tok != ')')
goto pragma_err;
} else if (tok == TOK_comment) {
char *p; int t;
next();
skip('(');
t = tok;
next();
skip(',');
if (tok != TOK_STR)
goto pragma_err;
p = tcc_strdup((char *)tokc.str.data);
next();
if (tok != ')')
goto pragma_err;
if (t == TOK_lib) {
dynarray_add(&s1->pragma_libs, &s1->nb_pragma_libs, p);
} else {
if (t == TOK_option)
tcc_set_options(s1, p);
tcc_free(p);
}
} else if (s1->warn_unsupported) {
tcc_warning("#pragma %s is ignored", get_tok_str(tok, &tokc));
2009-05-05 22:17:49 +04:00
}
return;
pragma_err:
tcc_error("malformed #pragma directive");
return;
2009-05-05 22:17:49 +04:00
}
/* is_bof is true if first non space token at beginning of file */
ST_FUNC void preprocess(int is_bof)
2009-05-05 22:17:49 +04:00
{
TCCState *s1 = tcc_state;
int i, c, n, saved_parse_flags;
2009-05-05 22:17:49 +04:00
char buf[1024], *q;
Sym *s;
2009-05-05 22:17:49 +04:00
saved_parse_flags = parse_flags;
parse_flags = PARSE_FLAG_PREPROCESS
| PARSE_FLAG_TOK_NUM
| PARSE_FLAG_TOK_STR
| PARSE_FLAG_LINEFEED
| (parse_flags & PARSE_FLAG_ASM_FILE)
;
2009-05-05 22:17:49 +04:00
next_nomacro();
redo:
switch(tok) {
case TOK_DEFINE:
pp_debug_tok = tok;
2009-05-05 22:17:49 +04:00
next_nomacro();
pp_debug_symv = tok;
2009-05-05 22:17:49 +04:00
parse_define();
break;
case TOK_UNDEF:
pp_debug_tok = tok;
2009-05-05 22:17:49 +04:00
next_nomacro();
pp_debug_symv = tok;
2009-05-05 22:17:49 +04:00
s = define_find(tok);
/* undefine symbol by putting an invalid name */
if (s)
define_undef(s);
break;
case TOK_INCLUDE:
case TOK_INCLUDE_NEXT:
ch = file->buf_ptr[0];
/* XXX: incorrect if comments : use next_nomacro with a special mode */
skip_spaces();
if (ch == '<') {
2009-05-05 22:17:49 +04:00
c = '>';
goto read_name;
} else if (ch == '\"') {
2009-05-05 22:17:49 +04:00
c = ch;
read_name:
2009-05-05 22:17:49 +04:00
inp();
q = buf;
while (ch != c && ch != '\n' && ch != CH_EOF) {
if ((q - buf) < sizeof(buf) - 1)
*q++ = ch;
if (ch == '\\') {
if (handle_stray_noerror() == 0)
--q;
} else
inp();
}
*q = '\0';
minp();
#if 0
/* eat all spaces and comments after include */
/* XXX: slightly incorrect */
while (ch1 != '\n' && ch1 != CH_EOF)
inp();
#endif
2009-05-05 22:17:49 +04:00
} else {
int len;
/* computed #include : concatenate everything up to linefeed,
the result must be one of the two accepted forms.
Don't convert pp-tokens to tokens here. */
parse_flags = (PARSE_FLAG_PREPROCESS
| PARSE_FLAG_LINEFEED
| (parse_flags & PARSE_FLAG_ASM_FILE));
2009-05-05 22:17:49 +04:00
next();
buf[0] = '\0';
while (tok != TOK_LINEFEED) {
pstrcat(buf, sizeof(buf), get_tok_str(tok, &tokc));
next();
}
len = strlen(buf);
/* check syntax and remove '<>|""' */
if ((len < 2 || ((buf[0] != '"' || buf[len-1] != '"') &&
(buf[0] != '<' || buf[len-1] != '>'))))
tcc_error("'#include' expects \"FILENAME\" or <FILENAME>");
c = buf[len-1];
memmove(buf, buf + 1, len - 2);
buf[len - 2] = '\0';
2009-05-05 22:17:49 +04:00
}
if (s1->include_stack_ptr >= s1->include_stack + INCLUDE_STACK_SIZE)
tcc_error("#include recursion too deep");
/* push current file on stack */
*s1->include_stack_ptr++ = file;
i = tok == TOK_INCLUDE_NEXT ? file->include_next_index: 0;
various stuff win32/Makefile ("for cygwin") removed - On cygwin, the normal ./configure && make can be used with either cygwin's "GCC for Win32 Toolchain" ./configure --cross-prefix=i686-w64-mingw32- or with an existing tcc: ./configure --cc=<old-tccdir>/tcc.exe tcctest.c: - exclude test_high_clobbers() on _WIN64 (does not work) tests2/95_bitfield.c: - use 'signed char' for ARM (where default 'char' is unsigned) tests: - remove -I "expr" diff option to allow tests with busybox-diff. libtcc.c, tcc.c: - removed -iwithprefix option. It is supposed to be combined with -iprefix which we don't have either. tccgen.c: - fix assignments and return of 'void', as in void f() { void *p, *q; *p = *q: return *p; } This appears to be allowed but should do nothing. tcc.h, libtcc.c, tccpp.c: - Revert "Introduce VIP sysinclude paths which are always searched first" This reverts commit 1d5e386b0a78393ac6b670c209a185849ec798a1. The patch was giving tcc's system includes priority over -I which is not how it should be. tccelf.c: - add DT_TEXTREL tag only if text relocations are actually used (which is likely not the case on x86_64) - prepare_dynamic_rel(): avoid relocation of unresolved (weak) symbols tccrun.c: - for HAVE_SELINUX, use two mappings to the same (real) file. (it was so once except the RX mapping wasn't used at all). tccpe.c: - fix relocation constant used for x86_64 (by Andrei E. Warentin) - #ifndef _WIN32 do "chmod 755 ..." to get runnable exes on cygwin. tccasm.c: - keep forward asm labels static, otherwise they will endup in dynsym eventually. configure, Makefile: - mingw32: respect ./configure options --bindir --docdir --libdir - allow overriding tcc when building libtcc1.a and libtcc.def with make XTCC=<tcc program to use> - use $(wildcard ...) for install to allow installing just a cross compiler for example make cross-arm make install - use name <target>-libtcc1.a build-tcc.bat: - add options: -clean, -b bindir
2017-10-11 19:13:43 +03:00
n = 2 + s1->nb_include_paths + s1->nb_sysinclude_paths;
for (; i < n; ++i) {
char buf1[sizeof file->filename];
CachedInclude *e;
const char *path;
if (i == 0) {
/* check absolute include path */
if (!IS_ABSPATH(buf))
continue;
buf1[0] = 0;
} else if (i == 1) {
/* search in file's dir if "header.h" */
if (c != '\"')
continue;
/* https://savannah.nongnu.org/bugs/index.php?50847 */
path = file->true_filename;
pstrncpy(buf1, path, tcc_basename(path) - path);
} else {
/* search in all the include paths */
various stuff win32/Makefile ("for cygwin") removed - On cygwin, the normal ./configure && make can be used with either cygwin's "GCC for Win32 Toolchain" ./configure --cross-prefix=i686-w64-mingw32- or with an existing tcc: ./configure --cc=<old-tccdir>/tcc.exe tcctest.c: - exclude test_high_clobbers() on _WIN64 (does not work) tests2/95_bitfield.c: - use 'signed char' for ARM (where default 'char' is unsigned) tests: - remove -I "expr" diff option to allow tests with busybox-diff. libtcc.c, tcc.c: - removed -iwithprefix option. It is supposed to be combined with -iprefix which we don't have either. tccgen.c: - fix assignments and return of 'void', as in void f() { void *p, *q; *p = *q: return *p; } This appears to be allowed but should do nothing. tcc.h, libtcc.c, tccpp.c: - Revert "Introduce VIP sysinclude paths which are always searched first" This reverts commit 1d5e386b0a78393ac6b670c209a185849ec798a1. The patch was giving tcc's system includes priority over -I which is not how it should be. tccelf.c: - add DT_TEXTREL tag only if text relocations are actually used (which is likely not the case on x86_64) - prepare_dynamic_rel(): avoid relocation of unresolved (weak) symbols tccrun.c: - for HAVE_SELINUX, use two mappings to the same (real) file. (it was so once except the RX mapping wasn't used at all). tccpe.c: - fix relocation constant used for x86_64 (by Andrei E. Warentin) - #ifndef _WIN32 do "chmod 755 ..." to get runnable exes on cygwin. tccasm.c: - keep forward asm labels static, otherwise they will endup in dynsym eventually. configure, Makefile: - mingw32: respect ./configure options --bindir --docdir --libdir - allow overriding tcc when building libtcc1.a and libtcc.def with make XTCC=<tcc program to use> - use $(wildcard ...) for install to allow installing just a cross compiler for example make cross-arm make install - use name <target>-libtcc1.a build-tcc.bat: - add options: -clean, -b bindir
2017-10-11 19:13:43 +03:00
int j = i - 2, k = j - s1->nb_include_paths;
path = k < 0 ? s1->include_paths[j] : s1->sysinclude_paths[k];
2009-05-05 22:17:49 +04:00
pstrcpy(buf1, sizeof(buf1), path);
pstrcat(buf1, sizeof(buf1), "/");
}
pstrcat(buf1, sizeof(buf1), buf);
e = search_cached_include(s1, buf1, 0);
if (e && (define_find(e->ifndef_macro) || e->once == pp_once)) {
/* no need to parse the include because the 'ifndef macro'
is defined (or had #pragma once) */
#ifdef INC_DEBUG
printf("%s: skipping cached %s\n", file->filename, buf1);
#endif
goto include_done;
}
if (tcc_open(s1, buf1) < 0)
continue;
file->include_next_index = i + 1;
#ifdef INC_DEBUG
printf("%s: including %s\n", file->prev->filename, file->filename);
#endif
tcc: Draft suppoprt for -MD/-MF options In build systems, this is used to automatically collect target dependencies, e.g. ---- 8< (hello.c) ---- #include "hello.h" #include <stdio.h> int main() { printf("Hello World!\n"); return 0; } $ tcc -MD -c hello.c # -> hello.o, hello.d $ cat hello.d hello.o : \ hello.c \ hello.h \ /usr/include/stdio.h \ /usr/include/features.h \ /usr/include/bits/predefs.h \ /usr/include/sys/cdefs.h \ /usr/include/bits/wordsize.h \ /usr/include/gnu/stubs.h \ /usr/include/bits/wordsize.h \ /usr/include/gnu/stubs-32.h \ /home/kirr/local/tcc/lib/tcc/include/stddef.h \ /usr/include/bits/types.h \ /usr/include/bits/wordsize.h \ /usr/include/bits/typesizes.h \ /usr/include/libio.h \ /usr/include/_G_config.h \ /usr/include/wchar.h \ /home/kirr/local/tcc/lib/tcc/include/stdarg.h \ /usr/include/bits/stdio_lim.h \ /usr/include/bits/sys_errlist.h \ NOTE: gcc supports -MD only for .c -> .o, but in tcc, we generate dependencies for whatever action is being taken. E.g. for .c -> exe, the result will be: $ tcc -MD -o hello hello.c # -> hello, hello.d hello: \ /usr/lib/crt1.o \ /usr/lib/crti.o \ hello.c \ hello.h \ /usr/include/stdio.h \ /usr/include/features.h \ /usr/include/bits/predefs.h \ /usr/include/sys/cdefs.h \ /usr/include/bits/wordsize.h \ /usr/include/gnu/stubs.h \ /usr/include/bits/wordsize.h \ /usr/include/gnu/stubs-32.h \ /home/kirr/local/tcc/lib/tcc/include/stddef.h \ /usr/include/bits/types.h \ /usr/include/bits/wordsize.h \ /usr/include/bits/typesizes.h \ /usr/include/libio.h \ /usr/include/_G_config.h \ /usr/include/wchar.h \ /home/kirr/local/tcc/lib/tcc/include/stdarg.h \ /usr/include/bits/stdio_lim.h \ /usr/include/bits/sys_errlist.h \ /usr/lib/libc.so \ /lib/libc.so.6 \ /usr/lib/ld-linux.so.2 \ /lib/ld-linux.so.2 \ /usr/lib/libc_nonshared.a \ /lib/libc.so.6 \ /usr/lib/libc_nonshared.a \ /home/kirr/local/tcc/lib/tcc/libtcc1.a \ /usr/lib/crtn.o \ So tcc dependency generator is a bit more clever than one used in gcc :) Also, I've updated TODO and Changelog (in not-yet-released section). v2: (Taking inputs from grischka and me myself) - put code to generate deps file into a function. - used tcc_fileextension() instead of open-coding - generate deps only when compilation/preprocessing was successful v3: - use pstrcpy instead of snprintf(buf, sizeof(buf), "%s", ...)
2010-06-20 20:08:12 +04:00
/* update target deps */
if (s1->gen_deps) {
dynarray_add(&s1->target_deps, &s1->nb_target_deps,
tcc_strdup(buf1));
}
2009-05-05 22:17:49 +04:00
/* add include file debug info */
tcc_debug_bincl(tcc_state);
2009-05-05 22:17:49 +04:00
tok_flags |= TOK_FLAG_BOF | TOK_FLAG_BOL;
ch = file->buf_ptr[0];
goto the_end;
2009-05-05 22:17:49 +04:00
}
tcc_error("include file '%s' not found", buf);
include_done:
--s1->include_stack_ptr;
break;
2009-05-05 22:17:49 +04:00
case TOK_IFNDEF:
c = 1;
goto do_ifdef;
case TOK_IF:
c = expr_preprocess();
goto do_if;
case TOK_IFDEF:
c = 0;
do_ifdef:
next_nomacro();
if (tok < TOK_IDENT)
tcc_error("invalid argument for '#if%sdef'", c ? "n" : "");
2009-05-05 22:17:49 +04:00
if (is_bof) {
if (c) {
#ifdef INC_DEBUG
printf("#ifndef %s\n", get_tok_str(tok, NULL));
#endif
2009-05-05 22:17:49 +04:00
file->ifndef_macro = tok;
}
}
c = (define_find(tok) != 0) ^ c;
do_if:
if (s1->ifdef_stack_ptr >= s1->ifdef_stack + IFDEF_STACK_SIZE)
tcc_error("memory full (ifdef)");
2009-05-05 22:17:49 +04:00
*s1->ifdef_stack_ptr++ = c;
goto test_skip;
case TOK_ELSE:
if (s1->ifdef_stack_ptr == s1->ifdef_stack)
tcc_error("#else without matching #if");
2009-05-05 22:17:49 +04:00
if (s1->ifdef_stack_ptr[-1] & 2)
tcc_error("#else after #else");
2009-05-05 22:17:49 +04:00
c = (s1->ifdef_stack_ptr[-1] ^= 3);
goto test_else;
2009-05-05 22:17:49 +04:00
case TOK_ELIF:
if (s1->ifdef_stack_ptr == s1->ifdef_stack)
tcc_error("#elif without matching #if");
2009-05-05 22:17:49 +04:00
c = s1->ifdef_stack_ptr[-1];
if (c > 1)
tcc_error("#elif after #else");
2009-05-05 22:17:49 +04:00
/* last #if/#elif expression was true: we skip */
if (c == 1) {
c = 0;
} else {
c = expr_preprocess();
s1->ifdef_stack_ptr[-1] = c;
}
test_else:
if (s1->ifdef_stack_ptr == file->ifdef_stack_ptr + 1)
file->ifndef_macro = 0;
2009-05-05 22:17:49 +04:00
test_skip:
if (!(c & 1)) {
preprocess_skip();
is_bof = 0;
goto redo;
}
break;
case TOK_ENDIF:
if (s1->ifdef_stack_ptr <= file->ifdef_stack_ptr)
tcc_error("#endif without matching #if");
2009-05-05 22:17:49 +04:00
s1->ifdef_stack_ptr--;
/* '#ifndef macro' was at the start of file. Now we check if
an '#endif' is exactly at the end of file */
if (file->ifndef_macro &&
s1->ifdef_stack_ptr == file->ifdef_stack_ptr) {
file->ifndef_macro_saved = file->ifndef_macro;
/* need to set to zero to avoid false matches if another
#ifndef at middle of file */
file->ifndef_macro = 0;
while (tok != TOK_LINEFEED)
next_nomacro();
tok_flags |= TOK_FLAG_ENDIF;
goto the_end;
}
break;
case TOK_PPNUM:
n = strtoul((char*)tokc.str.data, &q, 10);
goto _line_num;
2009-05-05 22:17:49 +04:00
case TOK_LINE:
next();
if (tok != TOK_CINT)
_line_err:
tcc_error("wrong #line format");
n = tokc.i;
_line_num:
2009-05-05 22:17:49 +04:00
next();
if (tok != TOK_LINEFEED) {
if (tok == TOK_STR) {
if (file->true_filename == file->filename)
file->true_filename = tcc_strdup(file->filename);
2020-01-18 00:58:39 +03:00
/* prepend directory from real file */
pstrcpy(buf, sizeof buf, file->true_filename);
*tcc_basename(buf) = 0;
pstrcat(buf, sizeof buf, (char *)tokc.str.data);
tcc_debug_putfile(s1, buf);
} else if (parse_flags & PARSE_FLAG_ASM_FILE)
break;
else
goto _line_err;
--n;
2009-05-05 22:17:49 +04:00
}
if (file->fd > 0)
total_lines += file->line_num - n;
file->line_num = n;
2009-05-05 22:17:49 +04:00
break;
case TOK_ERROR:
case TOK_WARNING:
c = tok;
ch = file->buf_ptr[0];
skip_spaces();
q = buf;
while (ch != '\n' && ch != CH_EOF) {
if ((q - buf) < sizeof(buf) - 1)
*q++ = ch;
if (ch == '\\') {
if (handle_stray_noerror() == 0)
--q;
} else
inp();
}
*q = '\0';
if (c == TOK_ERROR)
tcc_error("#error %s", buf);
2009-05-05 22:17:49 +04:00
else
tcc_warning("#warning %s", buf);
2009-05-05 22:17:49 +04:00
break;
case TOK_PRAGMA:
pragma_parse(s1);
break;
case TOK_LINEFEED:
goto the_end;
2009-05-05 22:17:49 +04:00
default:
/* ignore gas line comment in an 'S' file. */
if (saved_parse_flags & PARSE_FLAG_ASM_FILE)
goto ignore;
if (tok == '!' && is_bof)
/* '!' is ignored at beginning to allow C scripts. */
goto ignore;
tcc_warning("Ignoring unknown preprocessing directive #%s", get_tok_str(tok, &tokc));
ignore:
file->buf_ptr = parse_line_comment(file->buf_ptr - 1);
goto the_end;
2009-05-05 22:17:49 +04:00
}
/* ignore other preprocess commands or #! for C scripts */
while (tok != TOK_LINEFEED)
next_nomacro();
the_end:
parse_flags = saved_parse_flags;
}
/* evaluate escape codes in a string. */
static void parse_escape_string(CString *outstr, const uint8_t *buf, int is_long)
{
int c, n;
const uint8_t *p;
p = buf;
for(;;) {
c = *p;
if (c == '\0')
break;
if (c == '\\') {
p++;
/* escape */
c = *p;
switch(c) {
case '0': case '1': case '2': case '3':
case '4': case '5': case '6': case '7':
/* at most three octal digits */
n = c - '0';
p++;
c = *p;
if (isoct(c)) {
n = n * 8 + c - '0';
p++;
c = *p;
if (isoct(c)) {
n = n * 8 + c - '0';
p++;
}
}
c = n;
goto add_char_nonext;
case 'x':
case 'u':
case 'U':
p++;
n = 0;
for(;;) {
c = *p;
if (c >= 'a' && c <= 'f')
c = c - 'a' + 10;
else if (c >= 'A' && c <= 'F')
c = c - 'A' + 10;
else if (isnum(c))
c = c - '0';
else
break;
n = n * 16 + c;
p++;
}
c = n;
goto add_char_nonext;
case 'a':
c = '\a';
break;
case 'b':
c = '\b';
break;
case 'f':
c = '\f';
break;
case 'n':
c = '\n';
break;
case 'r':
c = '\r';
break;
case 't':
c = '\t';
break;
case 'v':
c = '\v';
break;
case 'e':
if (!gnu_ext)
goto invalid_escape;
c = 27;
break;
case '\'':
case '\"':
case '\\':
2009-05-05 22:17:49 +04:00
case '?':
break;
default:
invalid_escape:
if (c >= '!' && c <= '~')
tcc_warning("unknown escape sequence: \'\\%c\'", c);
2009-05-05 22:17:49 +04:00
else
tcc_warning("unknown escape sequence: \'\\x%x\'", c);
2009-05-05 22:17:49 +04:00
break;
}
} else if (is_long && c >= 0x80) {
/* assume we are processing UTF-8 sequence */
/* reference: The Unicode Standard, Version 10.0, ch3.9 */
int cont; /* count of continuation bytes */
int skip; /* how many bytes should skip when error occurred */
int i;
/* decode leading byte */
if (c < 0xC2) {
skip = 1; goto invalid_utf8_sequence;
} else if (c <= 0xDF) {
cont = 1; n = c & 0x1f;
} else if (c <= 0xEF) {
cont = 2; n = c & 0xf;
} else if (c <= 0xF4) {
cont = 3; n = c & 0x7;
} else {
skip = 1; goto invalid_utf8_sequence;
}
/* decode continuation bytes */
for (i = 1; i <= cont; i++) {
int l = 0x80, h = 0xBF;
/* adjust limit for second byte */
if (i == 1) {
switch (c) {
case 0xE0: l = 0xA0; break;
case 0xED: h = 0x9F; break;
case 0xF0: l = 0x90; break;
case 0xF4: h = 0x8F; break;
}
}
if (p[i] < l || p[i] > h) {
skip = i; goto invalid_utf8_sequence;
}
n = (n << 6) | (p[i] & 0x3f);
}
/* advance pointer */
p += 1 + cont;
c = n;
goto add_char_nonext;
/* error handling */
invalid_utf8_sequence:
tcc_warning("ill-formed UTF-8 subsequence starting with: \'\\x%x\'", c);
c = 0xFFFD;
p += skip;
goto add_char_nonext;
2009-05-05 22:17:49 +04:00
}
p++;
add_char_nonext:
if (!is_long)
cstr_ccat(outstr, c);
else {
#ifdef TCC_TARGET_PE
/* store as UTF-16 */
if (c < 0x10000) {
cstr_wccat(outstr, c);
} else {
c -= 0x10000;
cstr_wccat(outstr, (c >> 10) + 0xD800);
cstr_wccat(outstr, (c & 0x3FF) + 0xDC00);
}
#else
2009-05-05 22:17:49 +04:00
cstr_wccat(outstr, c);
#endif
}
2009-05-05 22:17:49 +04:00
}
/* add a trailing '\0' */
if (!is_long)
cstr_ccat(outstr, '\0');
else
cstr_wccat(outstr, '\0');
}
2016-09-26 05:32:41 +03:00
static void parse_string(const char *s, int len)
{
uint8_t buf[1000], *p = buf;
int is_long, sep;
if ((is_long = *s == 'L'))
++s, --len;
sep = *s++;
len -= 2;
if (len >= sizeof buf)
p = tcc_malloc(len + 1);
memcpy(p, s, len);
p[len] = 0;
cstr_reset(&tokcstr);
parse_escape_string(&tokcstr, p, is_long);
if (p != buf)
tcc_free(p);
if (sep == '\'') {
int char_size, i, n, c;
/* XXX: make it portable */
if (!is_long)
tok = TOK_CCHAR, char_size = 1;
else
tok = TOK_LCHAR, char_size = sizeof(nwchar_t);
n = tokcstr.size / char_size - 1;
if (n < 1)
tcc_error("empty character constant");
if (n > 1)
tcc_warning("multi-character character constant");
for (c = i = 0; i < n; ++i) {
if (is_long)
c = ((nwchar_t *)tokcstr.data)[i];
else
c = (c << 8) | ((char *)tokcstr.data)[i];
}
tokc.i = c;
} else {
tokc.str.size = tokcstr.size;
tokc.str.data = tokcstr.data;
if (!is_long)
tok = TOK_STR;
else
tok = TOK_LSTR;
}
}
2009-05-05 22:17:49 +04:00
/* we use 64 bit numbers */
#define BN_SIZE 2
/* bn = (bn << shift) | or_val */
static void bn_lshift(unsigned int *bn, int shift, int or_val)
2009-05-05 22:17:49 +04:00
{
int i;
unsigned int v;
for(i=0;i<BN_SIZE;i++) {
v = bn[i];
bn[i] = (v << shift) | or_val;
or_val = v >> (32 - shift);
}
}
static void bn_zero(unsigned int *bn)
2009-05-05 22:17:49 +04:00
{
int i;
for(i=0;i<BN_SIZE;i++) {
bn[i] = 0;
}
}
/* parse number in null terminated string 'p' and return it in the
current token */
static void parse_number(const char *p)
2009-05-05 22:17:49 +04:00
{
int b, t, shift, frac_bits, s, exp_val, ch;
char *q;
unsigned int bn[BN_SIZE];
double d;
/* number */
q = token_buf;
ch = *p++;
t = ch;
ch = *p++;
*q++ = t;
b = 10;
if (t == '.') {
goto float_frac_parse;
} else if (t == '0') {
if (ch == 'x' || ch == 'X') {
q--;
ch = *p++;
b = 16;
} else if (tcc_state->tcc_ext && (ch == 'b' || ch == 'B')) {
2009-05-05 22:17:49 +04:00
q--;
ch = *p++;
b = 2;
}
}
/* parse all digits. cannot check octal numbers at this stage
because of floating point constants */
while (1) {
if (ch >= 'a' && ch <= 'f')
t = ch - 'a' + 10;
else if (ch >= 'A' && ch <= 'F')
t = ch - 'A' + 10;
else if (isnum(ch))
t = ch - '0';
else
break;
if (t >= b)
break;
if (q >= token_buf + STRING_MAX_SIZE) {
num_too_long:
tcc_error("number too long");
2009-05-05 22:17:49 +04:00
}
*q++ = ch;
ch = *p++;
}
if (ch == '.' ||
((ch == 'e' || ch == 'E') && b == 10) ||
((ch == 'p' || ch == 'P') && (b == 16 || b == 2))) {
if (b != 10) {
/* NOTE: strtox should support that for hexa numbers, but
non ISOC99 libcs do not support it, so we prefer to do
it by hand */
/* hexadecimal or binary floats */
/* XXX: handle overflows */
*q = '\0';
if (b == 16)
shift = 4;
else
shift = 1;
2009-05-05 22:17:49 +04:00
bn_zero(bn);
q = token_buf;
while (1) {
t = *q++;
if (t == '\0') {
break;
} else if (t >= 'a') {
t = t - 'a' + 10;
} else if (t >= 'A') {
t = t - 'A' + 10;
} else {
t = t - '0';
}
bn_lshift(bn, shift, t);
}
frac_bits = 0;
if (ch == '.') {
ch = *p++;
while (1) {
t = ch;
if (t >= 'a' && t <= 'f') {
t = t - 'a' + 10;
} else if (t >= 'A' && t <= 'F') {
t = t - 'A' + 10;
} else if (t >= '0' && t <= '9') {
t = t - '0';
} else {
break;
}
if (t >= b)
tcc_error("invalid digit");
2009-05-05 22:17:49 +04:00
bn_lshift(bn, shift, t);
frac_bits += shift;
ch = *p++;
}
}
if (ch != 'p' && ch != 'P')
expect("exponent");
ch = *p++;
s = 1;
exp_val = 0;
if (ch == '+') {
ch = *p++;
} else if (ch == '-') {
s = -1;
ch = *p++;
}
if (ch < '0' || ch > '9')
expect("exponent digits");
while (ch >= '0' && ch <= '9') {
exp_val = exp_val * 10 + ch - '0';
ch = *p++;
}
exp_val = exp_val * s;
2009-05-05 22:17:49 +04:00
/* now we can generate the number */
/* XXX: should patch directly float number */
d = (double)bn[1] * 4294967296.0 + (double)bn[0];
d = ldexp(d, exp_val - frac_bits);
t = toup(ch);
if (t == 'F') {
ch = *p++;
tok = TOK_CFLOAT;
/* float : should handle overflow */
tokc.f = (float)d;
} else if (t == 'L') {
ch = *p++;
2009-07-19 00:07:17 +04:00
#ifdef TCC_TARGET_PE
tok = TOK_CDOUBLE;
tokc.d = d;
#else
2009-05-05 22:17:49 +04:00
tok = TOK_CLDOUBLE;
/* XXX: not large enough */
tokc.ld = (long double)d;
2009-07-19 00:07:17 +04:00
#endif
2009-05-05 22:17:49 +04:00
} else {
tok = TOK_CDOUBLE;
tokc.d = d;
}
} else {
/* decimal floats */
if (ch == '.') {
if (q >= token_buf + STRING_MAX_SIZE)
goto num_too_long;
*q++ = ch;
ch = *p++;
float_frac_parse:
while (ch >= '0' && ch <= '9') {
if (q >= token_buf + STRING_MAX_SIZE)
goto num_too_long;
*q++ = ch;
ch = *p++;
}
}
if (ch == 'e' || ch == 'E') {
if (q >= token_buf + STRING_MAX_SIZE)
goto num_too_long;
*q++ = ch;
ch = *p++;
if (ch == '-' || ch == '+') {
if (q >= token_buf + STRING_MAX_SIZE)
goto num_too_long;
*q++ = ch;
ch = *p++;
}
if (ch < '0' || ch > '9')
expect("exponent digits");
while (ch >= '0' && ch <= '9') {
if (q >= token_buf + STRING_MAX_SIZE)
goto num_too_long;
*q++ = ch;
ch = *p++;
}
}
*q = '\0';
t = toup(ch);
errno = 0;
if (t == 'F') {
ch = *p++;
tok = TOK_CFLOAT;
tokc.f = strtof(token_buf, NULL);
} else if (t == 'L') {
ch = *p++;
2009-07-19 00:07:17 +04:00
#ifdef TCC_TARGET_PE
tok = TOK_CDOUBLE;
tokc.d = strtod(token_buf, NULL);
#else
2009-05-05 22:17:49 +04:00
tok = TOK_CLDOUBLE;
tokc.ld = strtold(token_buf, NULL);
2009-07-19 00:07:17 +04:00
#endif
2009-05-05 22:17:49 +04:00
} else {
tok = TOK_CDOUBLE;
tokc.d = strtod(token_buf, NULL);
}
}
} else {
unsigned long long n, n1;
int lcount, ucount, ov = 0;
const char *p1;
2009-05-05 22:17:49 +04:00
/* integer number */
*q = '\0';
q = token_buf;
if (b == 10 && *q == '0') {
b = 8;
q++;
}
n = 0;
while(1) {
t = *q++;
/* no need for checks except for base 10 / 8 errors */
if (t == '\0')
2009-05-05 22:17:49 +04:00
break;
else if (t >= 'a')
2009-05-05 22:17:49 +04:00
t = t - 'a' + 10;
else if (t >= 'A')
2009-05-05 22:17:49 +04:00
t = t - 'A' + 10;
else
2009-05-05 22:17:49 +04:00
t = t - '0';
if (t >= b)
tcc_error("invalid digit");
2009-05-05 22:17:49 +04:00
n1 = n;
n = n * b + t;
/* detect overflow */
if (n1 >= 0x1000000000000000ULL && n / b != n1)
ov = 1;
2009-05-05 22:17:49 +04:00
}
/* Determine the characteristics (unsigned and/or 64bit) the type of
the constant must have according to the constant suffix(es) */
lcount = ucount = 0;
p1 = p;
2009-05-05 22:17:49 +04:00
for(;;) {
t = toup(ch);
if (t == 'L') {
if (lcount >= 2)
tcc_error("three 'l's in integer constant");
if (lcount && *(p - 1) != ch)
tcc_error("incorrect integer suffix: %s", p1);
2009-05-05 22:17:49 +04:00
lcount++;
ch = *p++;
} else if (t == 'U') {
if (ucount >= 1)
tcc_error("two 'u's in integer constant");
2009-05-05 22:17:49 +04:00
ucount++;
ch = *p++;
} else {
break;
}
}
/* Determine if it needs 64 bits and/or unsigned in order to fit */
if (ucount == 0 && b == 10) {
if (lcount <= (LONG_SIZE == 4)) {
if (n >= 0x80000000U)
lcount = (LONG_SIZE == 4) + 1;
}
if (n >= 0x8000000000000000ULL)
ov = 1, ucount = 1;
} else {
if (lcount <= (LONG_SIZE == 4)) {
if (n >= 0x100000000ULL)
lcount = (LONG_SIZE == 4) + 1;
else if (n >= 0x80000000U)
ucount = 1;
}
if (n >= 0x8000000000000000ULL)
ucount = 1;
}
if (ov)
tcc_warning("integer constant overflow");
tok = TOK_CINT;
if (lcount) {
tok = TOK_CLONG;
if (lcount == 2)
tok = TOK_CLLONG;
}
if (ucount)
++tok; /* TOK_CU... */
2016-08-16 22:22:59 +03:00
tokc.i = n;
2009-05-05 22:17:49 +04:00
}
if (ch)
tcc_error("invalid number\n");
2009-05-05 22:17:49 +04:00
}
#define PARSE2(c1, tok1, c2, tok2) \
case c1: \
PEEKC(c, p); \
if (c == c2) { \
p++; \
tok = tok2; \
} else { \
tok = tok1; \
} \
break;
/* return next token without macro substitution */
static inline void next_nomacro1(void)
{
int t, c, is_long, len;
2009-05-05 22:17:49 +04:00
TokenSym *ts;
uint8_t *p, *p1;
unsigned int h;
p = file->buf_ptr;
redo_no_start:
c = *p;
switch(c) {
case ' ':
case '\t':
tok = c;
p++;
if (parse_flags & PARSE_FLAG_SPACES)
goto keep_tok_flags;
while (isidnum_table[*p - CH_EOF] & IS_SPC)
++p;
goto redo_no_start;
2009-05-05 22:17:49 +04:00
case '\f':
case '\v':
case '\r':
p++;
goto redo_no_start;
case '\\':
/* first look if it is in fact an end of buffer */
c = handle_stray1(p);
p = file->buf_ptr;
if (c == '\\')
goto parse_simple;
if (c != CH_EOF)
2009-05-05 22:17:49 +04:00
goto redo_no_start;
{
TCCState *s1 = tcc_state;
if ((parse_flags & PARSE_FLAG_LINEFEED)
&& !(tok_flags & TOK_FLAG_EOF)) {
tok_flags |= TOK_FLAG_EOF;
tok = TOK_LINEFEED;
goto keep_tok_flags;
2010-01-14 23:00:05 +03:00
} else if (!(parse_flags & PARSE_FLAG_PREPROCESS)) {
tok = TOK_EOF;
} else if (s1->ifdef_stack_ptr != file->ifdef_stack_ptr) {
tcc_error("missing #endif");
2010-01-14 23:00:05 +03:00
} else if (s1->include_stack_ptr == s1->include_stack) {
2009-05-05 22:17:49 +04:00
/* no include left : end of file. */
tok = TOK_EOF;
} else {
tok_flags &= ~TOK_FLAG_EOF;
/* pop include file */
2009-05-05 22:17:49 +04:00
/* test if previous '#endif' was after a #ifdef at
start of file */
if (tok_flags & TOK_FLAG_ENDIF) {
#ifdef INC_DEBUG
printf("#endif %s\n", get_tok_str(file->ifndef_macro_saved, NULL));
#endif
search_cached_include(s1, file->filename, 1)
->ifndef_macro = file->ifndef_macro_saved;
tok_flags &= ~TOK_FLAG_ENDIF;
2009-05-05 22:17:49 +04:00
}
/* add end of include file debug info */
tcc_debug_eincl(tcc_state);
2009-05-05 22:17:49 +04:00
/* pop include stack */
tcc_close();
2009-05-05 22:17:49 +04:00
s1->include_stack_ptr--;
p = file->buf_ptr;
various stuff win32/Makefile ("for cygwin") removed - On cygwin, the normal ./configure && make can be used with either cygwin's "GCC for Win32 Toolchain" ./configure --cross-prefix=i686-w64-mingw32- or with an existing tcc: ./configure --cc=<old-tccdir>/tcc.exe tcctest.c: - exclude test_high_clobbers() on _WIN64 (does not work) tests2/95_bitfield.c: - use 'signed char' for ARM (where default 'char' is unsigned) tests: - remove -I "expr" diff option to allow tests with busybox-diff. libtcc.c, tcc.c: - removed -iwithprefix option. It is supposed to be combined with -iprefix which we don't have either. tccgen.c: - fix assignments and return of 'void', as in void f() { void *p, *q; *p = *q: return *p; } This appears to be allowed but should do nothing. tcc.h, libtcc.c, tccpp.c: - Revert "Introduce VIP sysinclude paths which are always searched first" This reverts commit 1d5e386b0a78393ac6b670c209a185849ec798a1. The patch was giving tcc's system includes priority over -I which is not how it should be. tccelf.c: - add DT_TEXTREL tag only if text relocations are actually used (which is likely not the case on x86_64) - prepare_dynamic_rel(): avoid relocation of unresolved (weak) symbols tccrun.c: - for HAVE_SELINUX, use two mappings to the same (real) file. (it was so once except the RX mapping wasn't used at all). tccpe.c: - fix relocation constant used for x86_64 (by Andrei E. Warentin) - #ifndef _WIN32 do "chmod 755 ..." to get runnable exes on cygwin. tccasm.c: - keep forward asm labels static, otherwise they will endup in dynsym eventually. configure, Makefile: - mingw32: respect ./configure options --bindir --docdir --libdir - allow overriding tcc when building libtcc1.a and libtcc.def with make XTCC=<tcc program to use> - use $(wildcard ...) for install to allow installing just a cross compiler for example make cross-arm make install - use name <target>-libtcc1.a build-tcc.bat: - add options: -clean, -b bindir
2017-10-11 19:13:43 +03:00
if (p == file->buffer)
tok_flags = TOK_FLAG_BOF|TOK_FLAG_BOL;
2009-05-05 22:17:49 +04:00
goto redo_no_start;
}
}
break;
case '\n':
file->line_num++;
tok_flags |= TOK_FLAG_BOL;
p++;
2009-07-06 23:12:45 +04:00
maybe_newline:
2009-05-05 22:17:49 +04:00
if (0 == (parse_flags & PARSE_FLAG_LINEFEED))
goto redo_no_start;
tok = TOK_LINEFEED;
goto keep_tok_flags;
case '#':
/* XXX: simplify */
PEEKC(c, p);
if ((tok_flags & TOK_FLAG_BOL) &&
2009-05-05 22:17:49 +04:00
(parse_flags & PARSE_FLAG_PREPROCESS)) {
file->buf_ptr = p;
preprocess(tok_flags & TOK_FLAG_BOF);
p = file->buf_ptr;
2009-07-06 23:12:45 +04:00
goto maybe_newline;
2009-05-05 22:17:49 +04:00
} else {
if (c == '#') {
p++;
tok = TOK_TWOSHARPS;
} else {
if (parse_flags & PARSE_FLAG_ASM_FILE) {
p = parse_line_comment(p - 1);
2009-05-05 22:17:49 +04:00
goto redo_no_start;
} else {
tok = '#';
}
}
}
break;
/* dollar is allowed to start identifiers when not parsing asm */
case '$':
if (!(isidnum_table[c - CH_EOF] & IS_ID)
|| (parse_flags & PARSE_FLAG_ASM_FILE))
goto parse_simple;
2009-05-05 22:17:49 +04:00
case 'a': case 'b': case 'c': case 'd':
case 'e': case 'f': case 'g': case 'h':
case 'i': case 'j': case 'k': case 'l':
case 'm': case 'n': case 'o': case 'p':
case 'q': case 'r': case 's': case 't':
case 'u': case 'v': case 'w': case 'x':
case 'y': case 'z':
2009-05-05 22:17:49 +04:00
case 'A': case 'B': case 'C': case 'D':
case 'E': case 'F': case 'G': case 'H':
case 'I': case 'J': case 'K':
2009-05-05 22:17:49 +04:00
case 'M': case 'N': case 'O': case 'P':
case 'Q': case 'R': case 'S': case 'T':
case 'U': case 'V': case 'W': case 'X':
case 'Y': case 'Z':
2009-05-05 22:17:49 +04:00
case '_':
parse_ident_fast:
p1 = p;
h = TOK_HASH_INIT;
h = TOK_HASH_FUNC(h, c);
while (c = *++p, isidnum_table[c - CH_EOF] & (IS_ID|IS_NUM))
2009-05-05 22:17:49 +04:00
h = TOK_HASH_FUNC(h, c);
len = p - p1;
2009-05-05 22:17:49 +04:00
if (c != '\\') {
TokenSym **pts;
/* fast case : no stray found, so we have the full token
and we have already hashed it */
h &= (TOK_HASH_SIZE - 1);
pts = &hash_ident[h];
for(;;) {
ts = *pts;
if (!ts)
break;
if (ts->len == len && !memcmp(ts->str, p1, len))
goto token_found;
pts = &(ts->hash_next);
}
2014-03-09 18:52:31 +04:00
ts = tok_alloc_new(pts, (char *) p1, len);
2009-05-05 22:17:49 +04:00
token_found: ;
} else {
/* slower case */
cstr_reset(&tokcstr);
cstr_cat(&tokcstr, (char *) p1, len);
2009-05-05 22:17:49 +04:00
p--;
PEEKC(c, p);
parse_ident_slow:
while (isidnum_table[c - CH_EOF] & (IS_ID|IS_NUM))
{
2009-05-05 22:17:49 +04:00
cstr_ccat(&tokcstr, c);
PEEKC(c, p);
}
ts = tok_alloc(tokcstr.data, tokcstr.size);
}
tok = ts->tok;
break;
case 'L':
t = p[1];
if (t != '\\' && t != '\'' && t != '\"') {
/* fast case */
goto parse_ident_fast;
} else {
PEEKC(c, p);
if (c == '\'' || c == '\"') {
is_long = 1;
goto str_const;
} else {
cstr_reset(&tokcstr);
cstr_ccat(&tokcstr, 'L');
goto parse_ident_slow;
}
}
break;
2016-05-04 22:27:39 +03:00
case '0': case '1': case '2': case '3':
2009-05-05 22:17:49 +04:00
case '4': case '5': case '6': case '7':
case '8': case '9':
2016-10-01 22:58:02 +03:00
t = c;
PEEKC(c, p);
2016-05-04 22:27:39 +03:00
/* after the first digit, accept digits, alpha, '.' or sign if
prefixed by 'eEpP' */
2009-05-05 22:17:49 +04:00
parse_num:
2016-10-01 22:58:02 +03:00
cstr_reset(&tokcstr);
2009-05-05 22:17:49 +04:00
for(;;) {
2016-10-01 22:58:02 +03:00
cstr_ccat(&tokcstr, t);
2016-05-04 22:27:39 +03:00
if (!((isidnum_table[c - CH_EOF] & (IS_ID|IS_NUM))
|| c == '.'
|| ((c == '+' || c == '-')
2016-10-01 22:58:02 +03:00
&& (((t == 'e' || t == 'E')
&& !(parse_flags & PARSE_FLAG_ASM_FILE
/* 0xe+1 is 3 tokens in asm */
&& ((char*)tokcstr.data)[0] == '0'
&& toup(((char*)tokcstr.data)[1]) == 'X'))
|| t == 'p' || t == 'P'))))
2016-05-04 22:27:39 +03:00
break;
2016-10-01 22:58:02 +03:00
t = c;
PEEKC(c, p);
2009-05-05 22:17:49 +04:00
}
/* We add a trailing '\0' to ease parsing */
cstr_ccat(&tokcstr, '\0');
tokc.str.size = tokcstr.size;
tokc.str.data = tokcstr.data;
2009-05-05 22:17:49 +04:00
tok = TOK_PPNUM;
break;
2009-05-05 22:17:49 +04:00
case '.':
/* special dot handling because it can also start a number */
PEEKC(c, p);
if (isnum(c)) {
2016-10-01 22:58:02 +03:00
t = '.';
2009-05-05 22:17:49 +04:00
goto parse_num;
} else if ((isidnum_table['.' - CH_EOF] & IS_ID)
&& (isidnum_table[c - CH_EOF] & (IS_ID|IS_NUM))) {
*--p = c = '.';
goto parse_ident_fast;
} else if (c == '.') {
PEEKC(c, p);
if (c == '.') {
p++;
tok = TOK_DOTS;
} else {
*--p = '.'; /* may underflow into file->unget[] */
tok = '.';
}
2009-05-05 22:17:49 +04:00
} else {
tok = '.';
}
break;
case '\'':
case '\"':
is_long = 0;
str_const:
cstr_reset(&tokcstr);
if (is_long)
cstr_ccat(&tokcstr, 'L');
cstr_ccat(&tokcstr, c);
p = parse_pp_string(p, c, &tokcstr);
cstr_ccat(&tokcstr, c);
cstr_ccat(&tokcstr, '\0');
tokc.str.size = tokcstr.size;
tokc.str.data = tokcstr.data;
tok = TOK_PPSTR;
2009-05-05 22:17:49 +04:00
break;
case '<':
PEEKC(c, p);
if (c == '=') {
p++;
tok = TOK_LE;
} else if (c == '<') {
PEEKC(c, p);
if (c == '=') {
p++;
tok = TOK_A_SHL;
} else {
tok = TOK_SHL;
}
} else {
tok = TOK_LT;
}
break;
case '>':
PEEKC(c, p);
if (c == '=') {
p++;
tok = TOK_GE;
} else if (c == '>') {
PEEKC(c, p);
if (c == '=') {
p++;
tok = TOK_A_SAR;
} else {
tok = TOK_SAR;
}
} else {
tok = TOK_GT;
}
break;
2009-05-05 22:17:49 +04:00
case '&':
PEEKC(c, p);
if (c == '&') {
p++;
tok = TOK_LAND;
} else if (c == '=') {
p++;
tok = TOK_A_AND;
} else {
tok = '&';
}
break;
2009-05-05 22:17:49 +04:00
case '|':
PEEKC(c, p);
if (c == '|') {
p++;
tok = TOK_LOR;
} else if (c == '=') {
p++;
tok = TOK_A_OR;
} else {
tok = '|';
}
break;
case '+':
PEEKC(c, p);
if (c == '+') {
p++;
tok = TOK_INC;
} else if (c == '=') {
p++;
tok = TOK_A_ADD;
} else {
tok = '+';
}
break;
2009-05-05 22:17:49 +04:00
case '-':
PEEKC(c, p);
if (c == '-') {
p++;
tok = TOK_DEC;
} else if (c == '=') {
p++;
tok = TOK_A_SUB;
} else if (c == '>') {
p++;
tok = TOK_ARROW;
} else {
tok = '-';
}
break;
PARSE2('!', '!', '=', TOK_NE)
PARSE2('=', '=', '=', TOK_EQ)
PARSE2('*', '*', '=', TOK_A_MUL)
PARSE2('%', '%', '=', TOK_A_MOD)
PARSE2('^', '^', '=', TOK_A_XOR)
2009-05-05 22:17:49 +04:00
/* comments or operator */
case '/':
PEEKC(c, p);
if (c == '*') {
p = parse_comment(p);
/* comments replaced by a blank */
tok = ' ';
goto keep_tok_flags;
2009-05-05 22:17:49 +04:00
} else if (c == '/') {
p = parse_line_comment(p);
tok = ' ';
goto keep_tok_flags;
2009-05-05 22:17:49 +04:00
} else if (c == '=') {
p++;
tok = TOK_A_DIV;
} else {
tok = '/';
}
break;
2009-05-05 22:17:49 +04:00
/* simple tokens */
case '(':
case ')':
case '[':
case ']':
case '{':
case '}':
case ',':
case ';':
case ':':
case '?':
case '~':
case '@': /* only used in assembler */
parse_simple:
2009-05-05 22:17:49 +04:00
tok = c;
p++;
break;
default:
if (c >= 0x80 && c <= 0xFF) /* utf8 identifiers */
goto parse_ident_fast;
if (parse_flags & PARSE_FLAG_ASM_FILE)
goto parse_simple;
tcc_error("unrecognized character \\x%02x", c);
2009-05-05 22:17:49 +04:00
break;
}
tok_flags = 0;
keep_tok_flags:
file->buf_ptr = p;
#if defined(PARSE_DEBUG)
various stuff win32/Makefile ("for cygwin") removed - On cygwin, the normal ./configure && make can be used with either cygwin's "GCC for Win32 Toolchain" ./configure --cross-prefix=i686-w64-mingw32- or with an existing tcc: ./configure --cc=<old-tccdir>/tcc.exe tcctest.c: - exclude test_high_clobbers() on _WIN64 (does not work) tests2/95_bitfield.c: - use 'signed char' for ARM (where default 'char' is unsigned) tests: - remove -I "expr" diff option to allow tests with busybox-diff. libtcc.c, tcc.c: - removed -iwithprefix option. It is supposed to be combined with -iprefix which we don't have either. tccgen.c: - fix assignments and return of 'void', as in void f() { void *p, *q; *p = *q: return *p; } This appears to be allowed but should do nothing. tcc.h, libtcc.c, tccpp.c: - Revert "Introduce VIP sysinclude paths which are always searched first" This reverts commit 1d5e386b0a78393ac6b670c209a185849ec798a1. The patch was giving tcc's system includes priority over -I which is not how it should be. tccelf.c: - add DT_TEXTREL tag only if text relocations are actually used (which is likely not the case on x86_64) - prepare_dynamic_rel(): avoid relocation of unresolved (weak) symbols tccrun.c: - for HAVE_SELINUX, use two mappings to the same (real) file. (it was so once except the RX mapping wasn't used at all). tccpe.c: - fix relocation constant used for x86_64 (by Andrei E. Warentin) - #ifndef _WIN32 do "chmod 755 ..." to get runnable exes on cygwin. tccasm.c: - keep forward asm labels static, otherwise they will endup in dynsym eventually. configure, Makefile: - mingw32: respect ./configure options --bindir --docdir --libdir - allow overriding tcc when building libtcc1.a and libtcc.def with make XTCC=<tcc program to use> - use $(wildcard ...) for install to allow installing just a cross compiler for example make cross-arm make install - use name <target>-libtcc1.a build-tcc.bat: - add options: -clean, -b bindir
2017-10-11 19:13:43 +03:00
printf("token = %d %s\n", tok, get_tok_str(tok, &tokc));
2009-05-05 22:17:49 +04:00
#endif
}
/* return next token without macro substitution. Can read input from
macro_ptr buffer */
static void next_nomacro_spc(void)
{
if (macro_ptr) {
redo:
tok = *macro_ptr;
if (tok) {
TOK_GET(&tok, &macro_ptr, &tokc);
2009-05-05 22:17:49 +04:00
if (tok == TOK_LINENUM) {
file->line_num = tokc.i;
goto redo;
}
}
} else {
next_nomacro1();
}
//printf("token = %s\n", get_tok_str(tok, &tokc));
2009-05-05 22:17:49 +04:00
}
ST_FUNC void next_nomacro(void)
2009-05-05 22:17:49 +04:00
{
do {
next_nomacro_spc();
} while (tok < 256 && (isidnum_table[tok - CH_EOF] & IS_SPC));
2009-05-05 22:17:49 +04:00
}
static void macro_subst(
TokenString *tok_str,
Sym **nested_list,
const int *macro_str
);
2014-04-12 08:00:13 +04:00
/* substitute arguments in replacement lists in macro_str by the values in
args (field d) and return allocated string */
static int *macro_arg_subst(Sym **nested_list, const int *macro_str, Sym *args)
2009-05-05 22:17:49 +04:00
{
int t, t0, t1, spc;
const int *st;
2009-05-05 22:17:49 +04:00
Sym *s;
CValue cval;
2009-05-05 22:17:49 +04:00
TokenString str;
CString cstr;
tok_str_new(&str);
t0 = t1 = 0;
2009-05-05 22:17:49 +04:00
while(1) {
TOK_GET(&t, &macro_str, &cval);
2009-05-05 22:17:49 +04:00
if (!t)
break;
if (t == '#') {
/* stringize */
TOK_GET(&t, &macro_str, &cval);
2009-05-05 22:17:49 +04:00
if (!t)
goto bad_stringy;
2009-05-05 22:17:49 +04:00
s = sym_find2(args, t);
if (s) {
cstr_new(&cstr);
cstr_ccat(&cstr, '\"');
st = s->d;
2009-05-05 22:17:49 +04:00
spc = 0;
while (*st >= 0) {
TOK_GET(&t, &st, &cval);
if (t != TOK_PLCHLDR
&& t != TOK_NOSUBST
&& 0 == check_space(t, &spc)) {
const char *s = get_tok_str(t, &cval);
while (*s) {
if (t == TOK_PPSTR && *s != '\'')
add_char(&cstr, *s);
else
cstr_ccat(&cstr, *s);
++s;
}
}
2009-05-05 22:17:49 +04:00
}
cstr.size -= spc;
cstr_ccat(&cstr, '\"');
2009-05-05 22:17:49 +04:00
cstr_ccat(&cstr, '\0');
#ifdef PP_DEBUG
printf("\nstringize: <%s>\n", (char *)cstr.data);
2009-05-05 22:17:49 +04:00
#endif
/* add string */
cval.str.size = cstr.size;
cval.str.data = cstr.data;
tok_str_add2(&str, TOK_PPSTR, &cval);
cstr_free(&cstr);
2009-05-05 22:17:49 +04:00
} else {
bad_stringy:
expect("macro parameter after '#'");
2009-05-05 22:17:49 +04:00
}
} else if (t >= TOK_IDENT) {
s = sym_find2(args, t);
if (s) {
int l0 = str.len;
st = s->d;
2009-05-05 22:17:49 +04:00
/* if '##' is present before or after, no arg substitution */
if (*macro_str == TOK_PPJOIN || t1 == TOK_PPJOIN) {
/* special case for var arg macros : ## eats the ','
if empty VA_ARGS variable. */
if (t1 == TOK_PPJOIN && t0 == ',' && gnu_ext && s->type.t) {
if (*st <= 0) {
/* suppress ',' '##' */
str.len -= 2;
} else {
/* suppress '##' and add variable */
str.len--;
goto add_var;
}
}
} else {
add_var:
if (!s->next) {
/* Expand arguments tokens and store them. In most
cases we could also re-expand each argument if
used multiple times, but not if the argument
contains the __COUNTER__ macro. */
TokenString str2;
sym_push2(&s->next, s->v, s->type.t, 0);
tok_str_new(&str2);
macro_subst(&str2, nested_list, st);
tok_str_add(&str2, 0);
s->next->d = str2.str;
}
st = s->next->d;
}
for(;;) {
int t2;
TOK_GET(&t2, &st, &cval);
if (t2 <= 0)
break;
tok_str_add2(&str, t2, &cval);
2009-05-05 22:17:49 +04:00
}
2017-05-08 07:38:09 +03:00
if (str.len == l0) /* expanded to empty string */
tok_str_add(&str, TOK_PLCHLDR);
2009-05-05 22:17:49 +04:00
} else {
tok_str_add(&str, t);
}
} else {
tok_str_add2(&str, t, &cval);
}
t0 = t1, t1 = t;
2009-05-05 22:17:49 +04:00
}
tok_str_add(&str, 0);
return str.str;
}
static char const ab_month_name[12][4] =
{
"Jan", "Feb", "Mar", "Apr", "May", "Jun",
"Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
};
static int paste_tokens(int t1, CValue *v1, int t2, CValue *v2)
{
CString cstr;
int n, ret = 1;
cstr_new(&cstr);
if (t1 != TOK_PLCHLDR)
cstr_cat(&cstr, get_tok_str(t1, v1), -1);
n = cstr.size;
if (t2 != TOK_PLCHLDR)
cstr_cat(&cstr, get_tok_str(t2, v2), -1);
cstr_ccat(&cstr, '\0');
tcc_open_bf(tcc_state, ":paste:", cstr.size);
memcpy(file->buffer, cstr.data, cstr.size);
tok_flags = 0;
for (;;) {
next_nomacro1();
if (0 == *file->buf_ptr)
break;
if (is_space(tok))
continue;
tcc_warning("pasting \"%.*s\" and \"%s\" does not give a valid"
" preprocessing token", n, cstr.data, (char*)cstr.data + n);
ret = 0;
break;
}
tcc_close();
//printf("paste <%s>\n", (char*)cstr.data);
cstr_free(&cstr);
return ret;
}
/* handle the '##' operator. Return NULL if no '##' seen. Otherwise
return the resulting string (which must be freed). */
static inline int *macro_twosharps(const int *ptr0)
{
int t;
CValue cval;
TokenString macro_str1;
int start_of_nosubsts = -1;
const int *ptr;
/* we search the first '##' */
for (ptr = ptr0;;) {
TOK_GET(&t, &ptr, &cval);
if (t == TOK_PPJOIN)
break;
if (t == 0)
return NULL;
}
tok_str_new(&macro_str1);
//tok_print(" $$$", ptr0);
for (ptr = ptr0;;) {
TOK_GET(&t, &ptr, &cval);
if (t == 0)
break;
if (t == TOK_PPJOIN)
continue;
while (*ptr == TOK_PPJOIN) {
int t1; CValue cv1;
/* given 'a##b', remove nosubsts preceding 'a' */
if (start_of_nosubsts >= 0)
macro_str1.len = start_of_nosubsts;
/* given 'a##b', remove nosubsts preceding 'b' */
while ((t1 = *++ptr) == TOK_NOSUBST)
;
if (t1 && t1 != TOK_PPJOIN) {
TOK_GET(&t1, &ptr, &cv1);
if (t != TOK_PLCHLDR || t1 != TOK_PLCHLDR) {
if (paste_tokens(t, &cval, t1, &cv1)) {
t = tok, cval = tokc;
} else {
tok_str_add2(&macro_str1, t, &cval);
t = t1, cval = cv1;
}
}
}
}
if (t == TOK_NOSUBST) {
if (start_of_nosubsts < 0)
start_of_nosubsts = macro_str1.len;
} else {
start_of_nosubsts = -1;
}
tok_str_add2(&macro_str1, t, &cval);
}
tok_str_add(&macro_str1, 0);
//tok_print(" ###", macro_str1.str);
return macro_str1.str;
}
/* peek or read [ws_str == NULL] next token from function macro call,
walking up macro levels up to the file if necessary */
static int next_argstream(Sym **nested_list, TokenString *ws_str)
{
int t;
const int *p;
Sym *sa;
for (;;) {
if (macro_ptr) {
p = macro_ptr, t = *p;
if (ws_str) {
while (is_space(t) || TOK_LINEFEED == t || TOK_PLCHLDR == t)
tok_str_add(ws_str, t), t = *++p;
}
if (t == 0) {
end_macro();
/* also, end of scope for nested defined symbol */
sa = *nested_list;
while (sa && sa->v == 0)
sa = sa->prev;
if (sa)
sa->v = 0;
continue;
}
} else {
ch = handle_eob();
if (ws_str) {
while (is_space(ch) || ch == '\n' || ch == '/') {
if (ch == '/') {
int c;
uint8_t *p = file->buf_ptr;
PEEKC(c, p);
if (c == '*') {
p = parse_comment(p);
file->buf_ptr = p - 1;
} else if (c == '/') {
p = parse_line_comment(p);
file->buf_ptr = p - 1;
} else
break;
ch = ' ';
}
if (ch == '\n')
file->line_num++;
if (!(ch == '\f' || ch == '\v' || ch == '\r'))
tok_str_add(ws_str, ch);
cinp();
}
}
t = ch;
}
if (ws_str)
return t;
next_nomacro_spc();
return tok;
}
}
2009-05-05 22:17:49 +04:00
/* do macro substitution of current token with macro 's' and add
result to (tok_str,tok_len). 'nested_list' is the list of all
macros we got inside to avoid recursing. Return non zero if no
substitution needs to be done */
static int macro_subst_tok(
TokenString *tok_str,
Sym **nested_list,
Sym *s)
2009-05-05 22:17:49 +04:00
{
Sym *args, *sa, *sa1;
int parlevel, t, t1, spc;
2009-05-05 22:17:49 +04:00
TokenString str;
char *cstrval;
CValue cval;
2009-05-05 22:17:49 +04:00
CString cstr;
char buf[32];
2009-05-05 22:17:49 +04:00
/* if symbol is a macro, prepare substitution */
/* special macros */
if (tok == TOK___LINE__ || tok == TOK___COUNTER__) {
t = tok == TOK___LINE__ ? file->line_num : pp_counter++;
snprintf(buf, sizeof(buf), "%d", t);
2009-05-05 22:17:49 +04:00
cstrval = buf;
t1 = TOK_PPNUM;
goto add_cstr1;
} else if (tok == TOK___FILE__) {
cstrval = file->filename;
goto add_cstr;
} else if (tok == TOK___DATE__ || tok == TOK___TIME__) {
time_t ti;
struct tm *tm;
time(&ti);
tm = localtime(&ti);
if (tok == TOK___DATE__) {
snprintf(buf, sizeof(buf), "%s %2d %d",
2009-05-05 22:17:49 +04:00
ab_month_name[tm->tm_mon], tm->tm_mday, tm->tm_year + 1900);
} else {
snprintf(buf, sizeof(buf), "%02d:%02d:%02d",
2009-05-05 22:17:49 +04:00
tm->tm_hour, tm->tm_min, tm->tm_sec);
}
cstrval = buf;
add_cstr:
t1 = TOK_STR;
add_cstr1:
cstr_new(&cstr);
cstr_cat(&cstr, cstrval, 0);
cval.str.size = cstr.size;
cval.str.data = cstr.data;
2009-05-05 22:17:49 +04:00
tok_str_add2(tok_str, t1, &cval);
cstr_free(&cstr);
} else if (s->d) {
int saved_parse_flags = parse_flags;
int *joined_str = NULL;
int *mstr = s->d;
2009-05-05 22:17:49 +04:00
if (s->type.t == MACRO_FUNC) {
/* whitespace between macro name and argument list */
TokenString ws_str;
tok_str_new(&ws_str);
spc = 0;
parse_flags |= PARSE_FLAG_SPACES | PARSE_FLAG_LINEFEED
| PARSE_FLAG_ACCEPT_STRAYS;
/* get next token from argument stream */
t = next_argstream(nested_list, &ws_str);
if (t != '(') {
/* not a macro substitution after all, restore the
* macro token plus all whitespace we've read.
* whitespace is intentionally not merged to preserve
* newlines. */
parse_flags = saved_parse_flags;
tok_str_add(tok_str, tok);
if (parse_flags & PARSE_FLAG_SPACES) {
int i;
for (i = 0; i < ws_str.len; i++)
tok_str_add(tok_str, ws_str.str[i]);
}
tok_str_free_str(ws_str.str);
return 0;
} else {
tok_str_free_str(ws_str.str);
}
do {
next_nomacro(); /* eat '(' */
} while (tok == TOK_PLCHLDR);
2009-05-05 22:17:49 +04:00
/* argument macro */
args = NULL;
sa = s->next;
/* NOTE: empty args are allowed, except if no args */
for(;;) {
do {
next_argstream(nested_list, NULL);
} while (is_space(tok) || TOK_LINEFEED == tok);
empty_arg:
2009-05-05 22:17:49 +04:00
/* handle '()' case */
if (!args && !sa && tok == ')')
break;
if (!sa)
tcc_error("macro '%s' used with too many args",
2009-05-05 22:17:49 +04:00
get_tok_str(s->v, 0));
tok_str_new(&str);
parlevel = spc = 0;
/* NOTE: non zero sa->t indicates VA_ARGS */
while ((parlevel > 0 ||
(tok != ')' &&
(tok != ',' || sa->type.t)))) {
if (tok == TOK_EOF || tok == 0)
break;
2009-05-05 22:17:49 +04:00
if (tok == '(')
parlevel++;
else if (tok == ')')
parlevel--;
if (tok == TOK_LINEFEED)
tok = ' ';
if (!check_space(tok, &spc))
tok_str_add2(&str, tok, &tokc);
next_argstream(nested_list, NULL);
2009-05-05 22:17:49 +04:00
}
if (parlevel)
expect(")");
2009-05-05 22:17:49 +04:00
str.len -= spc;
tok_str_add(&str, -1);
2009-05-05 22:17:49 +04:00
tok_str_add(&str, 0);
sa1 = sym_push2(&args, sa->v & ~SYM_FIELD, sa->type.t, 0);
sa1->d = str.str;
2009-05-05 22:17:49 +04:00
sa = sa->next;
if (tok == ')') {
/* special case for gcc var args: add an empty
var arg argument if it is omitted */
if (sa && sa->type.t && gnu_ext)
goto empty_arg;
break;
2009-05-05 22:17:49 +04:00
}
if (tok != ',')
expect(",");
2009-05-05 22:17:49 +04:00
}
if (sa) {
tcc_error("macro '%s' used with too few args",
2009-05-05 22:17:49 +04:00
get_tok_str(s->v, 0));
}
parse_flags = saved_parse_flags;
2009-05-05 22:17:49 +04:00
/* now subst each arg */
mstr = macro_arg_subst(nested_list, mstr, args);
/* free memory */
sa = args;
while (sa) {
sa1 = sa->prev;
tok_str_free_str(sa->d);
if (sa->next) {
tok_str_free_str(sa->next->d);
sym_free(sa->next);
}
2009-05-05 22:17:49 +04:00
sym_free(sa);
sa = sa1;
}
}
2009-05-05 22:17:49 +04:00
sym_push2(nested_list, s->v, 0, 0);
parse_flags = saved_parse_flags;
joined_str = macro_twosharps(mstr);
macro_subst(tok_str, nested_list, joined_str ? joined_str : mstr);
2009-05-05 22:17:49 +04:00
/* pop nested defined symbol */
sa1 = *nested_list;
*nested_list = sa1->prev;
sym_free(sa1);
if (joined_str)
tok_str_free_str(joined_str);
if (mstr != s->d)
tok_str_free_str(mstr);
2009-05-05 22:17:49 +04:00
}
return 0;
}
/* do macro substitution of macro_str and add result to
(tok_str,tok_len). 'nested_list' is the list of all macros we got
inside to avoid recursing. */
static void macro_subst(
TokenString *tok_str,
Sym **nested_list,
const int *macro_str
)
2009-05-05 22:17:49 +04:00
{
Sym *s;
int t, spc, nosubst;
CValue cval;
spc = nosubst = 0;
2009-05-05 22:17:49 +04:00
while (1) {
TOK_GET(&t, &macro_str, &cval);
if (t <= 0)
2009-05-05 22:17:49 +04:00
break;
if (t >= TOK_IDENT && 0 == nosubst) {
s = define_find(t);
if (s == NULL)
goto no_subst;
2009-05-05 22:17:49 +04:00
/* if nested substitution, do nothing */
if (sym_find2(*nested_list, t)) {
/* and mark it as TOK_NOSUBST, so it doesn't get subst'd again */
tok_str_add2(tok_str, TOK_NOSUBST, NULL);
2009-05-05 22:17:49 +04:00
goto no_subst;
}
{
TokenString *str = tok_str_alloc();
str->str = (int*)macro_str;
begin_macro(str, 2);
tok = t;
macro_subst_tok(tok_str, nested_list, s);
if (macro_stack != str) {
/* already finished by reading function macro arguments */
break;
}
macro_str = macro_ptr;
end_macro ();
}
if (tok_str->len)
spc = is_space(t = tok_str->str[tok_str->lastlen]);
2009-05-05 22:17:49 +04:00
} else {
if (t == '\\' && !(parse_flags & PARSE_FLAG_ACCEPT_STRAYS))
tcc_error("stray '\\' in program");
no_subst:
if (!check_space(t, &spc))
2009-05-05 22:17:49 +04:00
tok_str_add2(tok_str, t, &cval);
if (nosubst) {
if (nosubst > 1 && (spc || (++nosubst == 3 && t == '(')))
continue;
nosubst = 0;
}
if (t == TOK_NOSUBST)
nosubst = 1;
2009-05-05 22:17:49 +04:00
}
/* GCC supports 'defined' as result of a macro substitution */
if (t == TOK_DEFINED && pp_expr)
nosubst = 2;
2009-05-05 22:17:49 +04:00
}
}
/* return next token with macro substitution */
ST_FUNC void next(void)
2009-05-05 22:17:49 +04:00
{
redo:
if (parse_flags & PARSE_FLAG_SPACES)
next_nomacro_spc();
else
next_nomacro();
if (macro_ptr) {
if (tok == TOK_NOSUBST || tok == TOK_PLCHLDR) {
/* discard preprocessor markers */
2009-05-05 22:17:49 +04:00
goto redo;
} else if (tok == 0) {
/* end of macro or unget token string */
end_macro();
goto redo;
}
} else if (tok >= TOK_IDENT && (parse_flags & PARSE_FLAG_PREPROCESS)) {
Sym *s;
/* if reading from file, try to substitute macros */
s = define_find(tok);
if (s) {
2016-05-04 22:14:39 +03:00
Sym *nested_list = NULL;
tokstr_buf.len = 0;
macro_subst_tok(&tokstr_buf, &nested_list, s);
tok_str_add(&tokstr_buf, 0);
begin_macro(&tokstr_buf, 0);
goto redo;
2009-05-05 22:17:49 +04:00
}
}
/* convert preprocessor tokens into C tokens */
if (tok == TOK_PPNUM) {
if (parse_flags & PARSE_FLAG_TOK_NUM)
parse_number((char *)tokc.str.data);
} else if (tok == TOK_PPSTR) {
if (parse_flags & PARSE_FLAG_TOK_STR)
parse_string((char *)tokc.str.data, tokc.str.size - 1);
2009-05-05 22:17:49 +04:00
}
}
/* push back current token and set current token to 'last_tok'. Only
identifier case handled for labels. */
ST_INLN void unget_tok(int last_tok)
2009-05-05 22:17:49 +04:00
{
TokenString *str = tok_str_alloc();
tok_str_add2(str, tok, &tokc);
tok_str_add(str, 0);
begin_macro(str, 1);
2009-05-05 22:17:49 +04:00
tok = last_tok;
}
ST_FUNC void preprocess_start(TCCState *s1, int is_asm)
{
various stuff win32/Makefile ("for cygwin") removed - On cygwin, the normal ./configure && make can be used with either cygwin's "GCC for Win32 Toolchain" ./configure --cross-prefix=i686-w64-mingw32- or with an existing tcc: ./configure --cc=<old-tccdir>/tcc.exe tcctest.c: - exclude test_high_clobbers() on _WIN64 (does not work) tests2/95_bitfield.c: - use 'signed char' for ARM (where default 'char' is unsigned) tests: - remove -I "expr" diff option to allow tests with busybox-diff. libtcc.c, tcc.c: - removed -iwithprefix option. It is supposed to be combined with -iprefix which we don't have either. tccgen.c: - fix assignments and return of 'void', as in void f() { void *p, *q; *p = *q: return *p; } This appears to be allowed but should do nothing. tcc.h, libtcc.c, tccpp.c: - Revert "Introduce VIP sysinclude paths which are always searched first" This reverts commit 1d5e386b0a78393ac6b670c209a185849ec798a1. The patch was giving tcc's system includes priority over -I which is not how it should be. tccelf.c: - add DT_TEXTREL tag only if text relocations are actually used (which is likely not the case on x86_64) - prepare_dynamic_rel(): avoid relocation of unresolved (weak) symbols tccrun.c: - for HAVE_SELINUX, use two mappings to the same (real) file. (it was so once except the RX mapping wasn't used at all). tccpe.c: - fix relocation constant used for x86_64 (by Andrei E. Warentin) - #ifndef _WIN32 do "chmod 755 ..." to get runnable exes on cygwin. tccasm.c: - keep forward asm labels static, otherwise they will endup in dynsym eventually. configure, Makefile: - mingw32: respect ./configure options --bindir --docdir --libdir - allow overriding tcc when building libtcc1.a and libtcc.def with make XTCC=<tcc program to use> - use $(wildcard ...) for install to allow installing just a cross compiler for example make cross-arm make install - use name <target>-libtcc1.a build-tcc.bat: - add options: -clean, -b bindir
2017-10-11 19:13:43 +03:00
CString cstr;
tccpp_new(s1);
s1->include_stack_ptr = s1->include_stack;
s1->ifdef_stack_ptr = s1->ifdef_stack;
file->ifdef_stack_ptr = s1->ifdef_stack_ptr;
pp_expr = 0;
pp_counter = 0;
pp_debug_tok = pp_debug_symv = 0;
pp_once++;
s1->pack_stack[0] = 0;
s1->pack_stack_ptr = s1->pack_stack;
set_idnum('$', s1->dollars_in_identifiers ? IS_ID : 0);
set_idnum('.', is_asm ? IS_ID : 0);
various stuff win32/Makefile ("for cygwin") removed - On cygwin, the normal ./configure && make can be used with either cygwin's "GCC for Win32 Toolchain" ./configure --cross-prefix=i686-w64-mingw32- or with an existing tcc: ./configure --cc=<old-tccdir>/tcc.exe tcctest.c: - exclude test_high_clobbers() on _WIN64 (does not work) tests2/95_bitfield.c: - use 'signed char' for ARM (where default 'char' is unsigned) tests: - remove -I "expr" diff option to allow tests with busybox-diff. libtcc.c, tcc.c: - removed -iwithprefix option. It is supposed to be combined with -iprefix which we don't have either. tccgen.c: - fix assignments and return of 'void', as in void f() { void *p, *q; *p = *q: return *p; } This appears to be allowed but should do nothing. tcc.h, libtcc.c, tccpp.c: - Revert "Introduce VIP sysinclude paths which are always searched first" This reverts commit 1d5e386b0a78393ac6b670c209a185849ec798a1. The patch was giving tcc's system includes priority over -I which is not how it should be. tccelf.c: - add DT_TEXTREL tag only if text relocations are actually used (which is likely not the case on x86_64) - prepare_dynamic_rel(): avoid relocation of unresolved (weak) symbols tccrun.c: - for HAVE_SELINUX, use two mappings to the same (real) file. (it was so once except the RX mapping wasn't used at all). tccpe.c: - fix relocation constant used for x86_64 (by Andrei E. Warentin) - #ifndef _WIN32 do "chmod 755 ..." to get runnable exes on cygwin. tccasm.c: - keep forward asm labels static, otherwise they will endup in dynsym eventually. configure, Makefile: - mingw32: respect ./configure options --bindir --docdir --libdir - allow overriding tcc when building libtcc1.a and libtcc.def with make XTCC=<tcc program to use> - use $(wildcard ...) for install to allow installing just a cross compiler for example make cross-arm make install - use name <target>-libtcc1.a build-tcc.bat: - add options: -clean, -b bindir
2017-10-11 19:13:43 +03:00
cstr_new(&cstr);
if (s1->cmdline_defs.size)
cstr_cat(&cstr, s1->cmdline_defs.data, s1->cmdline_defs.size);
cstr_printf(&cstr, "#define __BASE_FILE__ \"%s\"\n", file->filename);
if (is_asm)
cstr_printf(&cstr, "#define __ASSEMBLER__ 1\n");
2020-01-18 00:58:39 +03:00
if (s1->output_type == TCC_OUTPUT_MEMORY)
cstr_printf(&cstr, "#define __TCC_RUN__ 1\n");
if (!is_asm && s1->output_type != TCC_OUTPUT_PREPROCESS)
cstr_cat(&cstr, "#include \"tcc_predefs.h\"\n", -1);
if (s1->cmdline_incl.size)
cstr_cat(&cstr, s1->cmdline_incl.data, s1->cmdline_incl.size);
//printf("%s\n", (char*)cstr.data);
*s1->include_stack_ptr++ = file;
tcc_open_bf(s1, "<command line>", cstr.size);
memcpy(file->buffer, cstr.data, cstr.size);
cstr_free(&cstr);
parse_flags = is_asm ? PARSE_FLAG_ASM_FILE : 0;
tok_flags = TOK_FLAG_BOL | TOK_FLAG_BOF;
}
/* cleanup from error/setjmp */
ST_FUNC void preprocess_end(TCCState *s1)
{
while (macro_stack)
end_macro();
macro_ptr = NULL;
while (file)
tcc_close();
tccpp_delete(s1);
}
2016-10-18 00:24:01 +03:00
ST_FUNC void tccpp_new(TCCState *s)
{
int i, c;
const char *p, *r;
/* init isid table */
for(i = CH_EOF; i<128; i++)
set_idnum(i,
is_space(i) ? IS_SPC
: isid(i) ? IS_ID
: isnum(i) ? IS_NUM
: 0);
for(i = 128; i<256; i++)
set_idnum(i, IS_ID);
/* init allocators */
tal_new(&toksym_alloc, TOKSYM_TAL_LIMIT, TOKSYM_TAL_SIZE);
tal_new(&tokstr_alloc, TOKSTR_TAL_LIMIT, TOKSTR_TAL_SIZE);
memset(hash_ident, 0, TOK_HASH_SIZE * sizeof(TokenSym *));
memset(s->cached_includes_hash, 0, sizeof s->cached_includes_hash);
cstr_new(&cstr_buf);
cstr_realloc(&cstr_buf, STRING_MAX_SIZE);
tok_str_new(&tokstr_buf);
tok_str_realloc(&tokstr_buf, TOKSTR_MAX_SIZE);
2020-01-18 00:58:39 +03:00
tok_ident = TOK_IDENT;
p = tcc_keywords;
while (*p) {
r = p;
for(;;) {
c = *r++;
if (c == '\0')
break;
}
tok_alloc(p, r - p - 1);
p = r;
}
/* we add dummy defines for some special macros to speed up tests
and to have working defined() */
define_push(TOK___LINE__, MACRO_OBJ, NULL, NULL);
define_push(TOK___FILE__, MACRO_OBJ, NULL, NULL);
define_push(TOK___DATE__, MACRO_OBJ, NULL, NULL);
define_push(TOK___TIME__, MACRO_OBJ, NULL, NULL);
define_push(TOK___COUNTER__, MACRO_OBJ, NULL, NULL);
}
2016-10-18 00:24:01 +03:00
ST_FUNC void tccpp_delete(TCCState *s)
{
int i, n;
dynarray_reset(&s->cached_includes, &s->nb_cached_includes);
/* free tokens */
n = tok_ident - TOK_IDENT;
if (n > total_idents)
total_idents = n;
for(i = 0; i < n; i++)
tal_free(toksym_alloc, table_ident[i]);
tcc_free(table_ident);
table_ident = NULL;
/* free static buffers */
cstr_free(&tokcstr);
cstr_free(&cstr_buf);
cstr_free(&macro_equal_buf);
tok_str_free_str(tokstr_buf.str);
/* free allocators */
tal_delete(toksym_alloc);
toksym_alloc = NULL;
tal_delete(tokstr_alloc);
tokstr_alloc = NULL;
}
/* ------------------------------------------------------------------------- */
/* tcc -E [-P[1]] [-dD} support */
static void tok_print(const char *msg, const int *str)
{
FILE *fp;
int t, s = 0;
CValue cval;
fp = tcc_state->ppfp;
fprintf(fp, "%s", msg);
while (str) {
TOK_GET(&t, &str, &cval);
if (!t)
break;
fprintf(fp, " %s" + s, get_tok_str(t, &cval)), s = 1;
}
fprintf(fp, "\n");
}
static void pp_line(TCCState *s1, BufferedFile *f, int level)
{
int d = f->line_num - f->line_ref;
if (s1->dflag & 4)
return;
if (s1->Pflag == LINE_MACRO_OUTPUT_FORMAT_NONE) {
;
} else if (level == 0 && f->line_ref && d < 8) {
while (d > 0)
fputs("\n", s1->ppfp), --d;
} else if (s1->Pflag == LINE_MACRO_OUTPUT_FORMAT_STD) {
fprintf(s1->ppfp, "#line %d \"%s\"\n", f->line_num, f->filename);
} else {
fprintf(s1->ppfp, "# %d \"%s\"%s\n", f->line_num, f->filename,
level > 0 ? " 1" : level < 0 ? " 2" : "");
}
f->line_ref = f->line_num;
}
static void define_print(TCCState *s1, int v)
{
FILE *fp;
Sym *s;
s = define_find(v);
if (NULL == s || NULL == s->d)
return;
fp = s1->ppfp;
fprintf(fp, "#define %s", get_tok_str(v, NULL));
if (s->type.t == MACRO_FUNC) {
Sym *a = s->next;
fprintf(fp,"(");
if (a)
for (;;) {
fprintf(fp,"%s", get_tok_str(a->v & ~SYM_FIELD, NULL));
if (!(a = a->next))
break;
fprintf(fp,",");
}
fprintf(fp,")");
}
tok_print("", s->d);
}
static void pp_debug_defines(TCCState *s1)
{
int v, t;
const char *vs;
FILE *fp;
t = pp_debug_tok;
if (t == 0)
return;
file->line_num--;
pp_line(s1, file, 0);
file->line_ref = ++file->line_num;
fp = s1->ppfp;
v = pp_debug_symv;
vs = get_tok_str(v, NULL);
if (t == TOK_DEFINE) {
define_print(s1, v);
} else if (t == TOK_UNDEF) {
fprintf(fp, "#undef %s\n", vs);
} else if (t == TOK_push_macro) {
fprintf(fp, "#pragma push_macro(\"%s\")\n", vs);
} else if (t == TOK_pop_macro) {
fprintf(fp, "#pragma pop_macro(\"%s\")\n", vs);
}
pp_debug_tok = 0;
}
static void pp_debug_builtins(TCCState *s1)
{
int v;
for (v = TOK_IDENT; v < tok_ident; ++v)
define_print(s1, v);
}
/* Add a space between tokens a and b to avoid unwanted textual pasting */
static int pp_need_space(int a, int b)
{
return 'E' == a ? '+' == b || '-' == b
: '+' == a ? TOK_INC == b || '+' == b
: '-' == a ? TOK_DEC == b || '-' == b
: a >= TOK_IDENT ? b >= TOK_IDENT
: a == TOK_PPNUM ? b >= TOK_IDENT
: 0;
}
/* maybe hex like 0x1e */
static int pp_check_he0xE(int t, const char *p)
{
if (t == TOK_PPNUM && toup(strchr(p, 0)[-1]) == 'E')
return 'E';
return t;
}
/* Preprocess the current file */
ST_FUNC int tcc_preprocess(TCCState *s1)
{
BufferedFile **iptr;
int token_seen, spcs, level;
const char *p;
char white[400];
parse_flags = PARSE_FLAG_PREPROCESS
| (parse_flags & PARSE_FLAG_ASM_FILE)
| PARSE_FLAG_LINEFEED
| PARSE_FLAG_SPACES
| PARSE_FLAG_ACCEPT_STRAYS
;
/* Credits to Fabrice Bellard's initial revision to demonstrate its
capability to compile and run itself, provided all numbers are
given as decimals. tcc -E -P10 will do. */
if (s1->Pflag == LINE_MACRO_OUTPUT_FORMAT_P10)
parse_flags |= PARSE_FLAG_TOK_NUM, s1->Pflag = 1;
#ifdef PP_BENCH
/* for PP benchmarks */
do next(); while (tok != TOK_EOF);
return 0;
#endif
if (s1->dflag & 1) {
pp_debug_builtins(s1);
s1->dflag &= ~1;
}
token_seen = TOK_LINEFEED, spcs = 0;
pp_line(s1, file, 0);
for (;;) {
iptr = s1->include_stack_ptr;
next();
if (tok == TOK_EOF)
break;
level = s1->include_stack_ptr - iptr;
if (level) {
if (level > 0)
pp_line(s1, *iptr, 0);
pp_line(s1, file, level);
}
if (s1->dflag & 7) {
pp_debug_defines(s1);
if (s1->dflag & 4)
continue;
}
if (is_space(tok)) {
if (spcs < sizeof white - 1)
white[spcs++] = tok;
continue;
} else if (tok == TOK_LINEFEED) {
spcs = 0;
if (token_seen == TOK_LINEFEED)
continue;
++file->line_ref;
} else if (token_seen == TOK_LINEFEED) {
pp_line(s1, file, 0);
} else if (spcs == 0 && pp_need_space(token_seen, tok)) {
white[spcs++] = ' ';
}
white[spcs] = 0, fputs(white, s1->ppfp), spcs = 0;
fputs(p = get_tok_str(tok, &tokc), s1->ppfp);
token_seen = pp_check_he0xE(tok, p);
}
return 0;
}
/* ------------------------------------------------------------------------- */