1 \input texinfo @c -*- texinfo -*-
3 @setfilename tcc-doc.info
4 @settitle Tiny C Compiler Reference Documentation
5 @dircategory Software development
7 * TCC: (tcc-doc). The Tiny C Compiler.
17 @center @titlefont{Tiny C Compiler Reference Documentation}
25 @node Top, Introduction, (dir), (dir)
26 @top Tiny C Compiler Reference Documentation
28 This manual documents version @value{VERSION} of the Tiny C Compiler.
31 * Introduction:: Introduction to tcc.
32 * Invoke:: Invocation of tcc (command line, options).
33 * Clang:: ANSI C and extensions.
34 * asm:: Assembler syntax.
35 * linker:: Output file generation and supported targets.
36 * Bounds:: Automatic bounds-checking of C code.
37 * Libtcc:: The libtcc library.
38 * devel:: Guide for Developers.
45 TinyCC (aka TCC) is a small but hyper fast C compiler. Unlike other C
46 compilers, it is meant to be self-relying: you do not need an
47 external assembler or linker because TCC does that for you.
49 TCC compiles so @emph{fast} that even for big projects @code{Makefile}s may
52 TCC not only supports ANSI C, but also most of the new ISO C99
53 standard and many GNUC extensions including inline assembly.
55 TCC can also be used to make @emph{C scripts}, i.e. pieces of C source
56 that you run as a Perl or Python script. Compilation is so fast that
57 your script will be as fast as if it was an executable.
59 TCC can also automatically generate memory and bound checks
60 (@pxref{Bounds}) while allowing all C pointers operations. TCC can do
61 these checks even if non patched libraries are used.
63 With @code{libtcc}, you can use TCC as a backend for dynamic code
64 generation (@pxref{Libtcc}).
66 TCC mainly supports the i386 target on Linux and Windows. There are alpha
67 ports for the ARM (@code{arm-tcc}) and the TMS320C67xx targets
68 (@code{c67-tcc}). More information about the ARM port is available at
69 @url{http://lists.gnu.org/archive/html/tinycc-devel/2003-10/msg00044.html}.
71 For usage on Windows, see also @url{tcc-win32.txt}.
74 @chapter Command line invocation
80 usage: tcc [options] [@var{infile1} @var{infile2}@dots{}] [@option{-run} @var{infile} @var{args}@dots{}]
85 @c man begin DESCRIPTION
86 TCC options are a very much like gcc options. The main difference is that TCC
87 can also execute directly the resulting program and give it runtime
90 Here are some examples to understand the logic:
93 @item @samp{tcc -run a.c}
94 Compile @file{a.c} and execute it directly
96 @item @samp{tcc -run a.c arg1}
97 Compile a.c and execute it directly. arg1 is given as first argument to
98 the @code{main()} of a.c.
100 @item @samp{tcc a.c -run b.c arg1}
101 Compile @file{a.c} and @file{b.c}, link them together and execute them. arg1 is given
102 as first argument to the @code{main()} of the resulting program.
104 Because multiple C files are specified, @option{--} are necessary to clearly
105 separate the program arguments from the TCC options.
108 @item @samp{tcc -o myprog a.c b.c}
109 Compile @file{a.c} and @file{b.c}, link them and generate the executable @file{myprog}.
111 @item @samp{tcc -o myprog a.o b.o}
112 link @file{a.o} and @file{b.o} together and generate the executable @file{myprog}.
114 @item @samp{tcc -c a.c}
115 Compile @file{a.c} and generate object file @file{a.o}.
117 @item @samp{tcc -c asmfile.S}
118 Preprocess with C preprocess and assemble @file{asmfile.S} and generate
119 object file @file{asmfile.o}.
121 @item @samp{tcc -c asmfile.s}
122 Assemble (but not preprocess) @file{asmfile.s} and generate object file
125 @item @samp{tcc -r -o ab.o a.c b.c}
126 Compile @file{a.c} and @file{b.c}, link them together and generate the object file @file{ab.o}.
132 TCC can be invoked from @emph{scripts}, just as shell scripts. You just
133 need to add @code{#!/usr/local/bin/tcc -run} at the start of your C source:
136 #!/usr/local/bin/tcc -run
141 printf("Hello World\n");
146 TCC can read C source code from @emph{standard input} when @option{-} is used in
147 place of @option{infile}. Example:
150 echo 'main()@{puts("hello");@}' | tcc -run -
154 @section Option summary
161 Generate an object file.
164 Put object file, executable, or dll into output file @file{outfile}.
166 @item -run source [args...]
167 Compile file @var{source} and run it with the command line arguments
168 @var{args}. In order to be able to give more than one argument to a
169 script, several TCC options can be given @emph{after} the
170 @option{-run} option, separated by spaces:
172 tcc "-run -L/usr/X11R6/lib -lX11" ex4.c
174 In a script, it gives the following header:
176 #!/usr/local/bin/tcc -run -L/usr/X11R6/lib -lX11
179 @item -mfloat-abi (ARM only)
180 Select the float ABI. Possible values: @code{softfp} and @code{hard}
183 Print only the compiler version and nothing else.
189 Show included files. As sole argument, print search dirs (as below).
192 Display compilation statistics.
194 @item -print-search-dirs
195 Print the configured installation directory and a list of library
196 and include directories tcc will search.
200 Preprocessor options:
204 Specify an additional include path. Include paths are searched in the
205 order they are specified.
207 System include paths are always searched after. The default system
208 include paths are: @file{/usr/local/include}, @file{/usr/include}
209 and @file{PREFIX/lib/tcc/include}. (@file{PREFIX} is usually
210 @file{/usr} or @file{/usr/local}).
213 Define preprocessor symbol @samp{sym} to
214 val. If val is not present, its value is @samp{1}. Function-like macros can
215 also be defined: @option{-DF(a)=a+1}
218 Undefine preprocessor symbol @samp{sym}.
223 Note: each of the following options has a negative form beginning with
227 @item -funsigned-char
228 Let the @code{char} type be unsigned.
231 Let the @code{char} type be signed.
234 Do not generate common symbols for uninitialized data.
236 @item -fleading-underscore
237 Add a leading underscore at the beginning of each C symbol.
239 @item -fms-extensions
240 Allow a MS C compiler extensions to the language. Curretly this
241 assume a nested named structure declaration without identifier behave
244 @item -fdollars-in-identifiers
245 Allow a dollars in identifiers.
247 @item -fnormalize-inc-dirs
248 Be more gcc compatible and remove non-existent or duplicate directories
249 from include paths. This helps to compile such packages as coreutils.
257 Disable all warnings.
261 Note: each of the following warning options has a negative form beginning with
265 @item -Wimplicit-function-declaration
266 Warn about implicit function declaration.
269 Warn about unsupported GCC features that are ignored by TCC.
271 @item -Wwrite-strings
272 Make string constants be of type @code{const char *} instead of @code{char
276 Abort compilation if warnings are issued.
279 Activate all warnings, except @option{-Werror}, @option{-Wunusupported} and
280 @option{-Wwrite-strings}.
288 Specify an additional static library path for the @option{-l} option. The
289 default library paths are @file{/usr/local/lib}, @file{/usr/lib} and @file{/lib}.
292 Link your program with dynamic library libxxx.so or static library
293 libxxx.a. The library is searched in the paths specified by the
294 @option{-L} option and @env{LIBRARY_PATH} variable.
297 Set the path where the tcc internal libraries (and include files) can be
298 found (default is @file{PREFIX/lib/tcc}).
301 Generate a shared library instead of an executable.
304 set name for shared library to be used at runtime
307 Generate a statically linked executable (default is a shared linked
311 Export global symbols to the dynamic linker. It is useful when a library
312 opened with @code{dlopen()} needs to access executable symbols.
315 Generate an object file combining all input files.
317 @item -Wl,-rpath=path
318 Put custom search path for dynamic libraries into executable.
320 @item -Wl,--whole-archive
321 Add a whole archive, not just the symbols of the archive that would
322 satisfy undefined symbols in the program.
324 @item -Wl,--no-whole-archive
325 Turn off the effect of the --whole-archive option for subsequent archive
328 @item -Wl,--oformat=fmt
329 Use @var{fmt} as output format. The supported output formats are:
332 ELF output format (default)
334 Binary image (only for executable output)
336 COFF output format (only for executable output for TMS320C67xx target)
339 @item -Wl,-subsystem=console/gui/wince/...
340 Set type for PE (Windows) executables.
342 @item -Wl,-[Ttext=# | section-alignment=# | file-alignment=# | image-base=# | stack=#]
343 Modify executable layout.
354 Generate run time debug information so that you get clear run time
355 error messages: @code{ test.c:68: in function 'test5()': dereferencing
356 invalid pointer} instead of the laconic @code{Segmentation
360 Generate additional support code to check
361 memory allocations and array/pointer bounds. @option{-g} is implied. Note
362 that the generated code is slower and bigger in this case.
364 Note: @option{-b} is only available on i386 when using libtcc for the moment.
367 Display N callers in stack traces. This is useful with @option{-g} or
376 Generate makefile fragment with dependencies.
379 Use @file{depfile} as output for -MD.
383 Note: GCC options @option{-Ox}, @option{-fx} and @option{-mx} are
387 @c man begin ENVIRONMENT
388 Environment variables that affect how tcc operates.
394 A colon-separated list of directories searched for include files,
395 directories given with @option{-I} are searched first.
398 A colon-separated list of directories searched for libraries for the
399 @option{-l} option, directories given with @option{-L} are searched first.
408 @settitle Tiny C Compiler
422 @chapter C language support
426 TCC implements all the ANSI C standard, including structure bit fields
427 and floating point numbers (@code{long double}, @code{double}, and
428 @code{float} fully supported).
430 @section ISOC99 extensions
432 TCC implements many features of the new C standard: ISO C99. Currently
433 missing items are: complex and imaginary numbers.
435 Currently implemented ISOC99 features:
439 @item variable length arrays.
441 @item 64 bit @code{long long} types are fully supported.
443 @item The boolean type @code{_Bool} is supported.
445 @item @code{__func__} is a string variable containing the current
448 @item Variadic macros: @code{__VA_ARGS__} can be used for
449 function-like macros:
451 #define dprintf(level, __VA_ARGS__) printf(__VA_ARGS__)
455 @code{dprintf} can then be used with a variable number of parameters.
457 @item Declarations can appear anywhere in a block (as in C++).
459 @item Array and struct/union elements can be initialized in any order by
462 struct @{ int x, y; @} st[10] = @{ [0].x = 1, [0].y = 2 @};
464 int tab[10] = @{ 1, 2, [5] = 5, [9] = 9@};
467 @item Compound initializers are supported:
469 int *p = (int [])@{ 1, 2, 3 @};
471 to initialize a pointer pointing to an initialized array. The same
472 works for structures and strings.
474 @item Hexadecimal floating point constants are supported:
476 double d = 0x1234p10;
480 is the same as writing
482 double d = 4771840.0;
485 @item @code{inline} keyword is ignored.
487 @item @code{restrict} keyword is ignored.
490 @section GNU C extensions
492 TCC implements some GNU C extensions:
496 @item array designators can be used without '=':
498 int a[10] = @{ [0] 1, [5] 2, 3, 4 @};
501 @item Structure field designators can be a label:
503 struct @{ int x, y; @} st = @{ x: 1, y: 1@};
507 struct @{ int x, y; @} st = @{ .x = 1, .y = 1@};
510 @item @code{\e} is ASCII character 27.
512 @item case ranges : ranges can be used in @code{case}s:
516 printf("range 1 to 9\n");
519 printf("unexpected\n");
524 @cindex aligned attribute
525 @cindex packed attribute
526 @cindex section attribute
527 @cindex unused attribute
528 @cindex cdecl attribute
529 @cindex stdcall attribute
530 @cindex regparm attribute
531 @cindex dllexport attribute
533 @item The keyword @code{__attribute__} is handled to specify variable or
534 function attributes. The following attributes are supported:
537 @item @code{aligned(n)}: align a variable or a structure field to n bytes
538 (must be a power of two).
540 @item @code{packed}: force alignment of a variable or a structure field to
543 @item @code{section(name)}: generate function or data in assembly section
544 name (name is a string containing the section name) instead of the default
547 @item @code{unused}: specify that the variable or the function is unused.
549 @item @code{cdecl}: use standard C calling convention (default).
551 @item @code{stdcall}: use Pascal-like calling convention.
553 @item @code{regparm(n)}: use fast i386 calling convention. @var{n} must be
554 between 1 and 3. The first @var{n} function parameters are respectively put in
555 registers @code{%eax}, @code{%edx} and @code{%ecx}.
557 @item @code{dllexport}: export function from dll/executable (win32 only)
561 Here are some examples:
563 int a __attribute__ ((aligned(8), section(".mysection")));
567 align variable @code{a} to 8 bytes and put it in section @code{.mysection}.
570 int my_add(int a, int b) __attribute__ ((section(".mycodesection")))
577 generate function @code{my_add} in section @code{.mycodesection}.
579 @item GNU style variadic macros:
581 #define dprintf(fmt, args@dots{}) printf(fmt, ## args)
584 dprintf("one arg %d\n", 1);
587 @item @code{__FUNCTION__} is interpreted as C99 @code{__func__}
588 (so it has not exactly the same semantics as string literal GNUC
589 where it is a string literal).
591 @item The @code{__alignof__} keyword can be used as @code{sizeof}
592 to get the alignment of a type or an expression.
594 @item The @code{typeof(x)} returns the type of @code{x}.
595 @code{x} is an expression or a type.
597 @item Computed gotos: @code{&&label} returns a pointer of type
598 @code{void *} on the goto label @code{label}. @code{goto *expr} can be
599 used to jump on the pointer resulting from @code{expr}.
601 @item Inline assembly with asm instruction:
602 @cindex inline assembly
603 @cindex assembly, inline
606 static inline void * my_memcpy(void * to, const void * from, size_t n)
609 __asm__ __volatile__(
614 "1:\ttestb $1,%b4\n\t"
618 : "=&c" (d0), "=&D" (d1), "=&S" (d2)
619 :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
627 TCC includes its own x86 inline assembler with a @code{gas}-like (GNU
628 assembler) syntax. No intermediate files are generated. GCC 3.x named
629 operands are supported.
631 @item @code{__builtin_types_compatible_p()} and @code{__builtin_constant_p()}
634 @item @code{#pragma pack} is supported for win32 compatibility.
638 @section TinyCC extensions
642 @item @code{__TINYC__} is a predefined macro to indicate that you use TCC.
644 @item @code{#!} at the start of a line is ignored to allow scripting.
646 @item Binary digits can be entered (@code{0b101} instead of
649 @item @code{__BOUNDS_CHECKING_ON} is defined if bound checking is activated.
654 @chapter TinyCC Assembler
656 Since version 0.9.16, TinyCC integrates its own assembler. TinyCC
657 assembler supports a gas-like syntax (GNU assembler). You can
658 desactivate assembler support if you want a smaller TinyCC executable
659 (the C compiler does not rely on the assembler).
661 TinyCC Assembler is used to handle files with @file{.S} (C
662 preprocessed assembler) and @file{.s} extensions. It is also used to
663 handle the GNU inline assembler with the @code{asm} keyword.
667 TinyCC Assembler supports most of the gas syntax. The tokens are the
672 @item C and C++ comments are supported.
674 @item Identifiers are the same as C, so you cannot use '.' or '$'.
676 @item Only 32 bit integer numbers are supported.
684 @item Integers in decimal, octal and hexa are supported.
686 @item Unary operators: +, -, ~.
688 @item Binary operators in decreasing priority order:
696 @item A value is either an absolute number or a label plus an offset.
697 All operators accept absolute values except '+' and '-'. '+' or '-' can be
698 used to add an offset to a label. '-' supports two labels only if they
699 are the same or if they are both defined and in the same section.
707 @item All labels are considered as local, except undefined ones.
709 @item Numeric labels can be used as local @code{gas}-like labels.
710 They can be defined several times in the same source. Use 'b'
711 (backward) or 'f' (forward) as suffix to reference them:
715 jmp 1b /* jump to '1' label before */
716 jmp 1f /* jump to '1' label after */
723 @cindex assembler directives
724 @cindex directives, assembler
725 @cindex align directive
726 @cindex skip directive
727 @cindex space directive
728 @cindex byte directive
729 @cindex word directive
730 @cindex short directive
731 @cindex int directive
732 @cindex long directive
733 @cindex quad directive
734 @cindex globl directive
735 @cindex global directive
736 @cindex section directive
737 @cindex text directive
738 @cindex data directive
739 @cindex bss directive
740 @cindex fill directive
741 @cindex org directive
742 @cindex previous directive
743 @cindex string directive
744 @cindex asciz directive
745 @cindex ascii directive
747 All directives are preceded by a '.'. The following directives are
751 @item .align n[,value]
752 @item .skip n[,value]
753 @item .space n[,value]
754 @item .byte value1[,...]
755 @item .word value1[,...]
756 @item .short value1[,...]
757 @item .int value1[,...]
758 @item .long value1[,...]
759 @item .quad immediate_value1[,...]
762 @item .section section
766 @item .fill repeat[,size[,value]]
769 @item .string string[,...]
770 @item .asciz string[,...]
771 @item .ascii string[,...]
774 @section X86 Assembler
777 All X86 opcodes are supported. Only ATT syntax is supported (source
778 then destination operand order). If no size suffix is given, TinyCC
779 tries to guess it from the operand sizes.
781 Currently, MMX opcodes are supported but not SSE ones.
784 @chapter TinyCC Linker
787 @section ELF file generation
790 TCC can directly output relocatable ELF files (object files),
791 executable ELF files and dynamic ELF libraries without relying on an
794 Dynamic ELF libraries can be output but the C compiler does not generate
795 position independent code (PIC). It means that the dynamic library
796 code generated by TCC cannot be factorized among processes yet.
798 TCC linker eliminates unreferenced object code in libraries. A single pass is
799 done on the object and library list, so the order in which object files and
800 libraries are specified is important (same constraint as GNU ld). No grouping
801 options (@option{--start-group} and @option{--end-group}) are supported.
803 @section ELF file loader
805 TCC can load ELF object files, archives (.a files) and dynamic
808 @section PE-i386 file generation
811 TCC for Windows supports the native Win32 executable file format (PE-i386). It
812 generates EXE files (console and gui) and DLL files.
814 For usage on Windows, see also tcc-win32.txt.
816 @section GNU Linker Scripts
817 @cindex scripts, linker
818 @cindex linker scripts
819 @cindex GROUP, linker command
820 @cindex FILE, linker command
821 @cindex OUTPUT_FORMAT, linker command
822 @cindex TARGET, linker command
824 Because on many Linux systems some dynamic libraries (such as
825 @file{/usr/lib/libc.so}) are in fact GNU ld link scripts (horrible!),
826 the TCC linker also supports a subset of GNU ld scripts.
828 The @code{GROUP} and @code{FILE} commands are supported. @code{OUTPUT_FORMAT}
829 and @code{TARGET} are ignored.
831 Example from @file{/usr/lib/libc.so}:
834 Use the shared library, but some functions are only in
835 the static library, so try that secondarily. */
836 GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a )
840 @chapter TinyCC Memory and Bound checks
842 @cindex memory checks
844 This feature is activated with the @option{-b} (@pxref{Invoke}).
846 Note that pointer size is @emph{unchanged} and that code generated
847 with bound checks is @emph{fully compatible} with unchecked
848 code. When a pointer comes from unchecked code, it is assumed to be
849 valid. Even very obscure C code with casts should work correctly.
851 For more information about the ideas behind this method, see
852 @url{http://www.doc.ic.ac.uk/~phjk/BoundsChecking.html}.
854 Here are some examples of caught errors:
858 @item Invalid range with standard string function:
866 @item Out of bounds-error in global or local arrays:
876 @item Out of bounds-error in malloc'ed data:
880 tab = malloc(20 * sizeof(int));
888 @item Access of freed memory:
892 tab = malloc(20 * sizeof(int));
904 tab = malloc(20 * sizeof(int));
913 @chapter The @code{libtcc} library
915 The @code{libtcc} library enables you to use TCC as a backend for
916 dynamic code generation.
918 Read the @file{libtcc.h} to have an overview of the API. Read
919 @file{libtcc_test.c} to have a very simple example.
921 The idea consists in giving a C string containing the program you want
922 to compile directly to @code{libtcc}. Then you can access to any global
923 symbol (function or variable) defined.
926 @chapter Developer's guide
928 This chapter gives some hints to understand how TCC works. You can skip
929 it if you do not intend to modify the TCC code.
931 @section File reading
933 The @code{BufferedFile} structure contains the context needed to read a
934 file, including the current line number. @code{tcc_open()} opens a new
935 file and @code{tcc_close()} closes it. @code{inp()} returns the next
940 @code{next()} reads the next token in the current
941 file. @code{next_nomacro()} reads the next token without macro
944 @code{tok} contains the current token (see @code{TOK_xxx})
945 constants. Identifiers and keywords are also keywords. @code{tokc}
946 contains additional infos about the token (for example a constant value
947 if number or string token).
951 The parser is hardcoded (yacc is not necessary). It does only one pass,
956 @item For initialized arrays with unknown size, a first pass
957 is done to count the number of elements.
959 @item For architectures where arguments are evaluated in
960 reverse order, a first pass is done to reverse the argument order.
966 The types are stored in a single 'int' variable. It was chosen in the
967 first stages of development when tcc was much simpler. Now, it may not
968 be the best solution.
971 #define VT_INT 0 /* integer type */
972 #define VT_BYTE 1 /* signed byte type */
973 #define VT_SHORT 2 /* short type */
974 #define VT_VOID 3 /* void type */
975 #define VT_PTR 4 /* pointer */
976 #define VT_ENUM 5 /* enum definition */
977 #define VT_FUNC 6 /* function type */
978 #define VT_STRUCT 7 /* struct/union definition */
979 #define VT_FLOAT 8 /* IEEE float */
980 #define VT_DOUBLE 9 /* IEEE double */
981 #define VT_LDOUBLE 10 /* IEEE long double */
982 #define VT_BOOL 11 /* ISOC99 boolean type */
983 #define VT_LLONG 12 /* 64 bit integer */
984 #define VT_LONG 13 /* long integer (NEVER USED as type, only
986 #define VT_BTYPE 0x000f /* mask for basic type */
987 #define VT_UNSIGNED 0x0010 /* unsigned type */
988 #define VT_ARRAY 0x0020 /* array type (also has VT_PTR) */
989 #define VT_VLA 0x20000 /* VLA type (also has VT_PTR and VT_ARRAY) */
990 #define VT_BITFIELD 0x0040 /* bitfield modifier */
991 #define VT_CONSTANT 0x0800 /* const modifier */
992 #define VT_VOLATILE 0x1000 /* volatile modifier */
993 #define VT_DEFSIGN 0x2000 /* signed type */
995 #define VT_STRUCT_SHIFT 18 /* structure/enum name shift (14 bits left) */
998 When a reference to another type is needed (for pointers, functions and
999 structures), the @code{32 - VT_STRUCT_SHIFT} high order bits are used to
1000 store an identifier reference.
1002 The @code{VT_UNSIGNED} flag can be set for chars, shorts, ints and long
1005 Arrays are considered as pointers @code{VT_PTR} with the flag
1006 @code{VT_ARRAY} set. Variable length arrays are considered as special
1007 arrays and have flag @code{VT_VLA} set instead of @code{VT_ARRAY}.
1009 The @code{VT_BITFIELD} flag can be set for chars, shorts, ints and long
1010 longs. If it is set, then the bitfield position is stored from bits
1011 VT_STRUCT_SHIFT to VT_STRUCT_SHIFT + 5 and the bit field size is stored
1012 from bits VT_STRUCT_SHIFT + 6 to VT_STRUCT_SHIFT + 11.
1014 @code{VT_LONG} is never used except during parsing.
1016 During parsing, the storage of an object is also stored in the type
1020 #define VT_EXTERN 0x00000080 /* extern definition */
1021 #define VT_STATIC 0x00000100 /* static variable */
1022 #define VT_TYPEDEF 0x00000200 /* typedef definition */
1023 #define VT_INLINE 0x00000400 /* inline definition */
1024 #define VT_IMPORT 0x00004000 /* win32: extern data imported from dll */
1025 #define VT_EXPORT 0x00008000 /* win32: data exported from dll */
1026 #define VT_WEAK 0x00010000 /* win32: data exported from dll */
1031 All symbols are stored in hashed symbol stacks. Each symbol stack
1032 contains @code{Sym} structures.
1034 @code{Sym.v} contains the symbol name (remember
1035 an idenfier is also a token, so a string is never necessary to store
1036 it). @code{Sym.t} gives the type of the symbol. @code{Sym.r} is usually
1037 the register in which the corresponding variable is stored. @code{Sym.c} is
1038 usually a constant associated to the symbol like its address for normal
1039 symbols, and the number of entries for symbols representing arrays.
1040 Variable length array types use @code{Sym.c} as a location on the stack
1041 which holds the runtime sizeof for the type.
1043 Four main symbol stacks are defined:
1048 for the macros (@code{#define}s).
1051 for the global variables, functions and types.
1054 for the local variables, functions and types.
1056 @item global_label_stack
1057 for the local labels (for @code{goto}).
1060 for GCC block local labels (see the @code{__label__} keyword).
1064 @code{sym_push()} is used to add a new symbol in the local symbol
1065 stack. If no local symbol stack is active, it is added in the global
1068 @code{sym_pop(st,b)} pops symbols from the symbol stack @var{st} until
1069 the symbol @var{b} is on the top of stack. If @var{b} is NULL, the stack
1072 @code{sym_find(v)} return the symbol associated to the identifier
1073 @var{v}. The local stack is searched first from top to bottom, then the
1078 The generated code and datas are written in sections. The structure
1079 @code{Section} contains all the necessary information for a given
1080 section. @code{new_section()} creates a new section. ELF file semantics
1081 is assumed for each section.
1083 The following sections are predefined:
1088 is the section containing the generated code. @var{ind} contains the
1089 current position in the code section.
1092 contains initialized data
1095 contains uninitialized data
1097 @item bounds_section
1098 @itemx lbounds_section
1099 are used when bound checking is activated
1102 @itemx stabstr_section
1103 are used when debugging is active to store debug information
1105 @item symtab_section
1106 @itemx strtab_section
1107 contain the exported symbols (currently only used for debugging).
1111 @section Code generation
1112 @cindex code generation
1114 @subsection Introduction
1116 The TCC code generator directly generates linked binary code in one
1117 pass. It is rather unusual these days (see gcc for example which
1118 generates text assembly), but it can be very fast and surprisingly
1121 The TCC code generator is register based. Optimization is only done at
1122 the expression level. No intermediate representation of expression is
1123 kept except the current values stored in the @emph{value stack}.
1125 On x86, three temporary registers are used. When more registers are
1126 needed, one register is spilled into a new temporary variable on the stack.
1128 @subsection The value stack
1129 @cindex value stack, introduction
1131 When an expression is parsed, its value is pushed on the value stack
1132 (@var{vstack}). The top of the value stack is @var{vtop}. Each value
1133 stack entry is the structure @code{SValue}.
1135 @code{SValue.t} is the type. @code{SValue.r} indicates how the value is
1136 currently stored in the generated code. It is usually a CPU register
1137 index (@code{REG_xxx} constants), but additional values and flags are
1141 #define VT_CONST 0x00f0
1142 #define VT_LLOCAL 0x00f1
1143 #define VT_LOCAL 0x00f2
1144 #define VT_CMP 0x00f3
1145 #define VT_JMP 0x00f4
1146 #define VT_JMPI 0x00f5
1147 #define VT_LVAL 0x0100
1148 #define VT_SYM 0x0200
1149 #define VT_MUSTCAST 0x0400
1150 #define VT_MUSTBOUND 0x0800
1151 #define VT_BOUNDED 0x8000
1152 #define VT_LVAL_BYTE 0x1000
1153 #define VT_LVAL_SHORT 0x2000
1154 #define VT_LVAL_UNSIGNED 0x4000
1155 #define VT_LVAL_TYPE (VT_LVAL_BYTE | VT_LVAL_SHORT | VT_LVAL_UNSIGNED)
1161 indicates that the value is a constant. It is stored in the union
1162 @code{SValue.c}, depending on its type.
1165 indicates a local variable pointer at offset @code{SValue.c.i} in the
1169 indicates that the value is actually stored in the CPU flags (i.e. the
1170 value is the consequence of a test). The value is either 0 or 1. The
1171 actual CPU flags used is indicated in @code{SValue.c.i}.
1173 If any code is generated which destroys the CPU flags, this value MUST be
1174 put in a normal register.
1178 indicates that the value is the consequence of a conditional jump. For VT_JMP,
1179 it is 1 if the jump is taken, 0 otherwise. For VT_JMPI it is inverted.
1181 These values are used to compile the @code{||} and @code{&&} logical
1184 If any code is generated, this value MUST be put in a normal
1185 register. Otherwise, the generated code won't be executed if the jump is
1189 is a flag indicating that the value is actually an lvalue (left value of
1190 an assignment). It means that the value stored is actually a pointer to
1193 Understanding the use @code{VT_LVAL} is very important if you want to
1194 understand how TCC works.
1197 @itemx VT_LVAL_SHORT
1198 @itemx VT_LVAL_UNSIGNED
1199 if the lvalue has an integer type, then these flags give its real
1200 type. The type alone is not enough in case of cast optimisations.
1203 is a saved lvalue on the stack. @code{VT_LVAL} must also be set with
1204 @code{VT_LLOCAL}. @code{VT_LLOCAL} can arise when a @code{VT_LVAL} in
1205 a register has to be saved to the stack, or it can come from an
1206 architecture-specific calling convention.
1209 indicates that a cast to the value type must be performed if the value
1210 is used (lazy casting).
1213 indicates that the symbol @code{SValue.sym} must be added to the constant.
1217 are only used for optional bound checking.
1221 @subsection Manipulating the value stack
1224 @code{vsetc()} and @code{vset()} pushes a new value on the value
1225 stack. If the previous @var{vtop} was stored in a very unsafe place(for
1226 example in the CPU flags), then some code is generated to put the
1227 previous @var{vtop} in a safe storage.
1229 @code{vpop()} pops @var{vtop}. In some cases, it also generates cleanup
1230 code (for example if stacked floating point registers are used as on
1233 The @code{gv(rc)} function generates code to evaluate @var{vtop} (the
1234 top value of the stack) into registers. @var{rc} selects in which
1235 register class the value should be put. @code{gv()} is the @emph{most
1236 important function} of the code generator.
1238 @code{gv2()} is the same as @code{gv()} but for the top two stack
1241 @subsection CPU dependent code generation
1242 @cindex CPU dependent
1243 See the @file{i386-gen.c} file to have an example.
1248 must generate the code needed to load a stack value into a register.
1251 must generate the code needed to store a register into a stack value
1255 @itemx gfunc_param()
1257 should generate a function call
1259 @item gfunc_prolog()
1260 @itemx gfunc_epilog()
1261 should generate a function prolog/epilog.
1264 must generate the binary integer operation @var{op} on the two top
1265 entries of the stack which are guaranted to contain integer types.
1267 The result value should be put on the stack.
1270 same as @code{gen_opi()} for floating point operations. The two top
1271 entries of the stack are guaranted to contain floating point values of
1274 @item gen_cvt_itof()
1275 integer to floating point conversion.
1277 @item gen_cvt_ftoi()
1278 floating point to integer conversion.
1280 @item gen_cvt_ftof()
1281 floating point to floating point of different size conversion.
1283 @item gen_bounded_ptr_add()
1284 @item gen_bounded_ptr_deref()
1285 are only used for bounds checking.
1289 @section Optimizations done
1290 @cindex optimizations
1291 @cindex constant propagation
1292 @cindex strength reduction
1293 @cindex comparison operators
1294 @cindex caching processor flags
1295 @cindex flags, caching
1296 @cindex jump optimization
1297 Constant propagation is done for all operations. Multiplications and
1298 divisions are optimized to shifts when appropriate. Comparison
1299 operators are optimized by maintaining a special cache for the
1300 processor flags. &&, || and ! are optimized by maintaining a special
1301 'jump target' value. No other jump optimization is currently performed
1302 because it would require to store the code in a more abstract fashion.
1304 @unnumbered Concept Index
1311 @c texinfo-column-for-description: 32