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
183 Show included files. As sole argument, print search dirs. -vvv shows tries too.
186 Display compilation statistics.
190 Preprocessor options:
194 Specify an additional include path. Include paths are searched in the
195 order they are specified.
197 System include paths are always searched after. The default system
198 include paths are: @file{/usr/local/include}, @file{/usr/include}
199 and @file{PREFIX/lib/tcc/include}. (@file{PREFIX} is usually
200 @file{/usr} or @file{/usr/local}).
203 Define preprocessor symbol @samp{sym} to
204 val. If val is not present, its value is @samp{1}. Function-like macros can
205 also be defined: @option{-DF(a)=a+1}
208 Undefine preprocessor symbol @samp{sym}.
211 Preprocess only, to stdout or file (with -o).
217 Note: each of the following options has a negative form beginning with
221 @item -funsigned-char
222 Let the @code{char} type be unsigned.
225 Let the @code{char} type be signed.
228 Do not generate common symbols for uninitialized data.
230 @item -fleading-underscore
231 Add a leading underscore at the beginning of each C symbol.
233 @item -fms-extensions
234 Allow a MS C compiler extensions to the language. Currently this
235 assumes a nested named structure declaration without an identifier
236 behaves like an unnamed one.
238 @item -fdollars-in-identifiers
239 Allow dollar signs in identifiers
247 Disable all warnings.
251 Note: each of the following warning options has a negative form beginning with
255 @item -Wimplicit-function-declaration
256 Warn about implicit function declaration.
259 Warn about unsupported GCC features that are ignored by TCC.
261 @item -Wwrite-strings
262 Make string constants be of type @code{const char *} instead of @code{char
266 Abort compilation if warnings are issued.
269 Activate all warnings, except @option{-Werror}, @option{-Wunusupported} and
270 @option{-Wwrite-strings}.
278 Specify an additional static library path for the @option{-l} option. The
279 default library paths are @file{/usr/local/lib}, @file{/usr/lib} and @file{/lib}.
282 Link your program with dynamic library libxxx.so or static library
283 libxxx.a. The library is searched in the paths specified by the
284 @option{-L} option and @env{LIBRARY_PATH} variable.
287 Set the path where the tcc internal libraries (and include files) can be
288 found (default is @file{PREFIX/lib/tcc}).
291 Generate a shared library instead of an executable.
294 set name for shared library to be used at runtime
297 Generate a statically linked executable (default is a shared linked
301 Export global symbols to the dynamic linker. It is useful when a library
302 opened with @code{dlopen()} needs to access executable symbols.
305 Generate an object file combining all input files.
307 @item -Wl,-rpath=path
308 Put custom search path for dynamic libraries into executable.
310 @item -Wl,--enable-new-dtags
311 When putting a custom search path for dynamic libraries into the executable,
312 create the new ELF dynamic tag DT_RUNPATH instead of the old legacy DT_RPATH.
314 @item -Wl,--oformat=fmt
315 Use @var{fmt} as output format. The supported output formats are:
318 ELF output format (default)
320 Binary image (only for executable output)
322 COFF output format (only for executable output for TMS320C67xx target)
325 @item -Wl,--export-all-symbols
326 @item -Wl,--export-dynamic
327 Export global symbols to the dynamic linker. It is useful when a library
328 opened with @code{dlopen()} needs to access executable symbols.
330 @item -Wl,-subsystem=console/gui/wince/...
331 Set type for PE (Windows) executables.
333 @item -Wl,-[Ttext=# | section-alignment=# | file-alignment=# | image-base=# | stack=#]
334 Modify executable layout.
339 @item -Wl,-(no-)whole-archive
340 Turn on/off linking of all objects in archives.
348 Generate run time debug information so that you get clear run time
349 error messages: @code{ test.c:68: in function 'test5()': dereferencing
350 invalid pointer} instead of the laconic @code{Segmentation
354 Generate additional support code to check
355 memory allocations and array/pointer bounds. @option{-g} is implied. Note
356 that the generated code is slower and bigger in this case.
357 The bound checking code is not included in shared libaries. The main executable should always be compiled with the @option{-b}.
359 There are five environment variables that can be used:
361 @item TCC_BOUNDS_WARN_POINTER_ADD
362 Print warning when pointer add creates an illegal pointer.
363 @item TCC_BOUNDS_PRINT_CALLS
364 Print bound checking calls. Can be used for debugging.
365 @item TCC_BOUNDS_PRINT_HEAP
366 Print heap objects that are not freed at exit of program.
367 @item TCC_BOUNDS_PRINT_STATISTIC
368 Print statistic information at exit of program.
369 @item TCC_BOUNDS_NEVER_FATAL
370 Try to continue in case of a bound checking error.
373 Note: @option{-b} is only available on i386 (linux and windows) and x86_64 (linux and windows) for the moment.
376 Display N callers in stack traces. This is useful with @option{-g} or
385 Generate makefile fragment with dependencies.
388 Use @file{depfile} as output for -MD.
390 @item -print-search-dirs
391 Print the configured installation directory and a list of library
392 and include directories tcc will search.
399 Target specific options:
403 Use an algorithm for bitfield alignment consistent with MSVC. Default is
406 @item -mfloat-abi (ARM only)
407 Select the float ABI. Possible values: @code{softfp} and @code{hard}
410 Do not use sse registers on x86_64
413 Pass command line to the i386/x86_64 cross compiler.
417 Note: GCC options @option{-Ox}, @option{-fx} and @option{-mx} are
421 @c man begin ENVIRONMENT
422 Environment variables that affect how tcc operates.
428 A colon-separated list of directories searched for include files,
429 directories given with @option{-I} are searched first.
432 A colon-separated list of directories searched for libraries for the
433 @option{-l} option, directories given with @option{-L} are searched first.
442 @settitle Tiny C Compiler
456 @chapter C language support
460 TCC implements all the ANSI C standard, including structure bit fields
461 and floating point numbers (@code{long double}, @code{double}, and
462 @code{float} fully supported).
464 @section ISOC99 extensions
466 TCC implements many features of the new C standard: ISO C99. Currently
467 missing items are: complex and imaginary numbers.
469 Currently implemented ISOC99 features:
473 @item variable length arrays.
475 @item 64 bit @code{long long} types are fully supported.
477 @item The boolean type @code{_Bool} is supported.
479 @item @code{__func__} is a string variable containing the current
482 @item Variadic macros: @code{__VA_ARGS__} can be used for
483 function-like macros:
485 #define dprintf(level, __VA_ARGS__) printf(__VA_ARGS__)
489 @code{dprintf} can then be used with a variable number of parameters.
491 @item Declarations can appear anywhere in a block (as in C++).
493 @item Array and struct/union elements can be initialized in any order by
496 struct @{ int x, y; @} st[10] = @{ [0].x = 1, [0].y = 2 @};
498 int tab[10] = @{ 1, 2, [5] = 5, [9] = 9@};
501 @item Compound initializers are supported:
503 int *p = (int [])@{ 1, 2, 3 @};
505 to initialize a pointer pointing to an initialized array. The same
506 works for structures and strings.
508 @item Hexadecimal floating point constants are supported:
510 double d = 0x1234p10;
514 is the same as writing
516 double d = 4771840.0;
519 @item @code{inline} keyword is ignored.
521 @item @code{restrict} keyword is ignored.
524 @section GNU C extensions
526 TCC implements some GNU C extensions:
530 @item array designators can be used without '=':
532 int a[10] = @{ [0] 1, [5] 2, 3, 4 @};
535 @item Structure field designators can be a label:
537 struct @{ int x, y; @} st = @{ x: 1, y: 1@};
541 struct @{ int x, y; @} st = @{ .x = 1, .y = 1@};
544 @item @code{\e} is ASCII character 27.
546 @item case ranges : ranges can be used in @code{case}s:
550 printf("range 1 to 9\n");
553 printf("unexpected\n");
558 @cindex aligned attribute
559 @cindex packed attribute
560 @cindex section attribute
561 @cindex unused attribute
562 @cindex cdecl attribute
563 @cindex stdcall attribute
564 @cindex regparm attribute
565 @cindex dllexport attribute
566 @cindex nodecorate attribute
568 @item The keyword @code{__attribute__} is handled to specify variable or
569 function attributes. The following attributes are supported:
572 @item @code{aligned(n)}: align a variable or a structure field to n bytes
573 (must be a power of two).
575 @item @code{packed}: force alignment of a variable or a structure field to
578 @item @code{section(name)}: generate function or data in assembly section
579 name (name is a string containing the section name) instead of the default
582 @item @code{unused}: specify that the variable or the function is unused.
584 @item @code{cdecl}: use standard C calling convention (default).
586 @item @code{stdcall}: use Pascal-like calling convention.
588 @item @code{regparm(n)}: use fast i386 calling convention. @var{n} must be
589 between 1 and 3. The first @var{n} function parameters are respectively put in
590 registers @code{%eax}, @code{%edx} and @code{%ecx}.
592 @item @code{dllexport}: export function from dll/executable (win32 only)
594 @item @code{nodecorate}: do not apply any decorations that would otherwise be applied when exporting function from dll/executable (win32 only)
598 Here are some examples:
600 int a __attribute__ ((aligned(8), section(".mysection")));
604 align variable @code{a} to 8 bytes and put it in section @code{.mysection}.
607 int my_add(int a, int b) __attribute__ ((section(".mycodesection")))
614 generate function @code{my_add} in section @code{.mycodesection}.
616 @item GNU style variadic macros:
618 #define dprintf(fmt, args@dots{}) printf(fmt, ## args)
621 dprintf("one arg %d\n", 1);
624 @item @code{__FUNCTION__} is interpreted as C99 @code{__func__}
625 (so it has not exactly the same semantics as string literal GNUC
626 where it is a string literal).
628 @item The @code{__alignof__} keyword can be used as @code{sizeof}
629 to get the alignment of a type or an expression.
631 @item The @code{typeof(x)} returns the type of @code{x}.
632 @code{x} is an expression or a type.
634 @item Computed gotos: @code{&&label} returns a pointer of type
635 @code{void *} on the goto label @code{label}. @code{goto *expr} can be
636 used to jump on the pointer resulting from @code{expr}.
638 @item Inline assembly with asm instruction:
639 @cindex inline assembly
640 @cindex assembly, inline
643 static inline void * my_memcpy(void * to, const void * from, size_t n)
646 __asm__ __volatile__(
651 "1:\ttestb $1,%b4\n\t"
655 : "=&c" (d0), "=&D" (d1), "=&S" (d2)
656 :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
664 TCC includes its own x86 inline assembler with a @code{gas}-like (GNU
665 assembler) syntax. No intermediate files are generated. GCC 3.x named
666 operands are supported.
668 @item @code{__builtin_types_compatible_p()} and @code{__builtin_constant_p()}
671 @item @code{#pragma pack} is supported for win32 compatibility.
675 @section TinyCC extensions
679 @item @code{__TINYC__} is a predefined macro to indicate that you use TCC.
681 @item @code{#!} at the start of a line is ignored to allow scripting.
683 @item Binary digits can be entered (@code{0b101} instead of
686 @item @code{__BOUNDS_CHECKING_ON} is defined if bound checking is activated.
691 @chapter TinyCC Assembler
693 Since version 0.9.16, TinyCC integrates its own assembler. TinyCC
694 assembler supports a gas-like syntax (GNU assembler). You can
695 deactivate assembler support if you want a smaller TinyCC executable
696 (the C compiler does not rely on the assembler).
698 TinyCC Assembler is used to handle files with @file{.S} (C
699 preprocessed assembler) and @file{.s} extensions. It is also used to
700 handle the GNU inline assembler with the @code{asm} keyword.
704 TinyCC Assembler supports most of the gas syntax. The tokens are the
709 @item C and C++ comments are supported.
711 @item Identifiers are the same as C, so you cannot use '.' or '$'.
713 @item Only 32 bit integer numbers are supported.
721 @item Integers in decimal, octal and hexa are supported.
723 @item Unary operators: +, -, ~.
725 @item Binary operators in decreasing priority order:
733 @item A value is either an absolute number or a label plus an offset.
734 All operators accept absolute values except '+' and '-'. '+' or '-' can be
735 used to add an offset to a label. '-' supports two labels only if they
736 are the same or if they are both defined and in the same section.
744 @item All labels are considered as local, except undefined ones.
746 @item Numeric labels can be used as local @code{gas}-like labels.
747 They can be defined several times in the same source. Use 'b'
748 (backward) or 'f' (forward) as suffix to reference them:
752 jmp 1b /* jump to '1' label before */
753 jmp 1f /* jump to '1' label after */
760 @cindex assembler directives
761 @cindex directives, assembler
762 @cindex align directive
763 @cindex skip directive
764 @cindex space directive
765 @cindex byte directive
766 @cindex word directive
767 @cindex short directive
768 @cindex int directive
769 @cindex long directive
770 @cindex quad directive
771 @cindex globl directive
772 @cindex global directive
773 @cindex section directive
774 @cindex text directive
775 @cindex data directive
776 @cindex bss directive
777 @cindex fill directive
778 @cindex org directive
779 @cindex previous directive
780 @cindex string directive
781 @cindex asciz directive
782 @cindex ascii directive
784 All directives are preceded by a '.'. The following directives are
788 @item .align n[,value]
789 @item .skip n[,value]
790 @item .space n[,value]
791 @item .byte value1[,...]
792 @item .word value1[,...]
793 @item .short value1[,...]
794 @item .int value1[,...]
795 @item .long value1[,...]
796 @item .quad immediate_value1[,...]
799 @item .section section
803 @item .fill repeat[,size[,value]]
806 @item .string string[,...]
807 @item .asciz string[,...]
808 @item .ascii string[,...]
811 @section X86 Assembler
814 All X86 opcodes are supported. Only ATT syntax is supported (source
815 then destination operand order). If no size suffix is given, TinyCC
816 tries to guess it from the operand sizes.
818 Currently, MMX opcodes are supported but not SSE ones.
821 @chapter TinyCC Linker
824 @section ELF file generation
827 TCC can directly output relocatable ELF files (object files),
828 executable ELF files and dynamic ELF libraries without relying on an
831 Dynamic ELF libraries can be output but the C compiler does not generate
832 position independent code (PIC). It means that the dynamic library
833 code generated by TCC cannot be factorized among processes yet.
835 TCC linker eliminates unreferenced object code in libraries. A single pass is
836 done on the object and library list, so the order in which object files and
837 libraries are specified is important (same constraint as GNU ld). No grouping
838 options (@option{--start-group} and @option{--end-group}) are supported.
840 @section ELF file loader
842 TCC can load ELF object files, archives (.a files) and dynamic
845 @section PE-i386 file generation
848 TCC for Windows supports the native Win32 executable file format (PE-i386). It
849 generates EXE files (console and gui) and DLL files.
851 For usage on Windows, see also tcc-win32.txt.
853 @section GNU Linker Scripts
854 @cindex scripts, linker
855 @cindex linker scripts
856 @cindex GROUP, linker command
857 @cindex FILE, linker command
858 @cindex OUTPUT_FORMAT, linker command
859 @cindex TARGET, linker command
861 Because on many Linux systems some dynamic libraries (such as
862 @file{/usr/lib/libc.so}) are in fact GNU ld link scripts (horrible!),
863 the TCC linker also supports a subset of GNU ld scripts.
865 The @code{GROUP} and @code{FILE} commands are supported. @code{OUTPUT_FORMAT}
866 and @code{TARGET} are ignored.
868 Example from @file{/usr/lib/libc.so}:
871 Use the shared library, but some functions are only in
872 the static library, so try that secondarily. */
873 GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a )
877 @chapter TinyCC Memory and Bound checks
879 @cindex memory checks
881 This feature is activated with the @option{-b} (@pxref{Invoke}).
883 Note that pointer size is @emph{unchanged} and that code generated
884 with bound checks is @emph{fully compatible} with unchecked
885 code. When a pointer comes from unchecked code, it is assumed to be
886 valid. Even very obscure C code with casts should work correctly.
888 For more information about the ideas behind this method, see
889 @url{http://www.doc.ic.ac.uk/~phjk/BoundsChecking.html}.
891 Here are some examples of caught errors:
895 @item Invalid range with standard string function:
903 @item Out of bounds-error in global or local arrays:
913 @item Out of bounds-error in malloc'ed data:
917 tab = malloc(20 * sizeof(int));
925 @item Access of freed memory:
929 tab = malloc(20 * sizeof(int));
941 tab = malloc(20 * sizeof(int));
949 Signal handlers are not compatible with bounds checking. The code
950 below can be used to protect signal handlers.
951 The call to __bound_checking(1) will disable bounds checking in the
954 The BOUNDS_CHECKING_OFF and BOUNDS_CHECKING_ON can also be used to
955 disable bounds checking for some code. This is not recommended.
956 It is better to fix the code.
960 #ifdef __BOUNDS_CHECKING_ON
961 extern void __bound_checking (int no_check);
962 #define BOUNDS_CHECKING_OFF __bound_checking(1)
963 #define BOUNDS_CHECKING_ON __bound_checking(-1)
965 #define BOUNDS_CHECKING_OFF
966 #define BOUNDS_CHECKING_ON
969 void real_signal_handler(int sig, siginfo_t *info, void *ucontext)
974 void signal_handler(int sig, void *info, void *ucontext)
977 real_signal_handler(sig, info, data);
984 @chapter The @code{libtcc} library
986 The @code{libtcc} library enables you to use TCC as a backend for
987 dynamic code generation.
989 Read the @file{libtcc.h} to have an overview of the API. Read
990 @file{libtcc_test.c} to have a very simple example.
992 The idea consists in giving a C string containing the program you want
993 to compile directly to @code{libtcc}. Then you can access to any global
994 symbol (function or variable) defined.
997 @chapter Developer's guide
999 This chapter gives some hints to understand how TCC works. You can skip
1000 it if you do not intend to modify the TCC code.
1002 @section File reading
1004 The @code{BufferedFile} structure contains the context needed to read a
1005 file, including the current line number. @code{tcc_open()} opens a new
1006 file and @code{tcc_close()} closes it. @code{inp()} returns the next
1011 @code{next()} reads the next token in the current
1012 file. @code{next_nomacro()} reads the next token without macro
1015 @code{tok} contains the current token (see @code{TOK_xxx})
1016 constants. Identifiers and keywords are also keywords. @code{tokc}
1017 contains additional infos about the token (for example a constant value
1018 if number or string token).
1022 The parser is hardcoded (yacc is not necessary). It does only one pass,
1027 @item For initialized arrays with unknown size, a first pass
1028 is done to count the number of elements.
1030 @item For architectures where arguments are evaluated in
1031 reverse order, a first pass is done to reverse the argument order.
1037 The types are stored in a single 'int' variable. It was chosen in the
1038 first stages of development when tcc was much simpler. Now, it may not
1039 be the best solution.
1042 #define VT_INT 0 /* integer type */
1043 #define VT_BYTE 1 /* signed byte type */
1044 #define VT_SHORT 2 /* short type */
1045 #define VT_VOID 3 /* void type */
1046 #define VT_PTR 4 /* pointer */
1047 #define VT_ENUM 5 /* enum definition */
1048 #define VT_FUNC 6 /* function type */
1049 #define VT_STRUCT 7 /* struct/union definition */
1050 #define VT_FLOAT 8 /* IEEE float */
1051 #define VT_DOUBLE 9 /* IEEE double */
1052 #define VT_LDOUBLE 10 /* IEEE long double */
1053 #define VT_BOOL 11 /* ISOC99 boolean type */
1054 #define VT_LLONG 12 /* 64 bit integer */
1055 #define VT_LONG 13 /* long integer (NEVER USED as type, only
1057 #define VT_BTYPE 0x000f /* mask for basic type */
1058 #define VT_UNSIGNED 0x0010 /* unsigned type */
1059 #define VT_ARRAY 0x0020 /* array type (also has VT_PTR) */
1060 #define VT_VLA 0x20000 /* VLA type (also has VT_PTR and VT_ARRAY) */
1061 #define VT_BITFIELD 0x0040 /* bitfield modifier */
1062 #define VT_CONSTANT 0x0800 /* const modifier */
1063 #define VT_VOLATILE 0x1000 /* volatile modifier */
1064 #define VT_DEFSIGN 0x2000 /* signed type */
1066 #define VT_STRUCT_SHIFT 18 /* structure/enum name shift (14 bits left) */
1069 When a reference to another type is needed (for pointers, functions and
1070 structures), the @code{32 - VT_STRUCT_SHIFT} high order bits are used to
1071 store an identifier reference.
1073 The @code{VT_UNSIGNED} flag can be set for chars, shorts, ints and long
1076 Arrays are considered as pointers @code{VT_PTR} with the flag
1077 @code{VT_ARRAY} set. Variable length arrays are considered as special
1078 arrays and have flag @code{VT_VLA} set instead of @code{VT_ARRAY}.
1080 The @code{VT_BITFIELD} flag can be set for chars, shorts, ints and long
1081 longs. If it is set, then the bitfield position is stored from bits
1082 VT_STRUCT_SHIFT to VT_STRUCT_SHIFT + 5 and the bit field size is stored
1083 from bits VT_STRUCT_SHIFT + 6 to VT_STRUCT_SHIFT + 11.
1085 @code{VT_LONG} is never used except during parsing.
1087 During parsing, the storage of an object is also stored in the type
1091 #define VT_EXTERN 0x00000080 /* extern definition */
1092 #define VT_STATIC 0x00000100 /* static variable */
1093 #define VT_TYPEDEF 0x00000200 /* typedef definition */
1094 #define VT_INLINE 0x00000400 /* inline definition */
1095 #define VT_IMPORT 0x00004000 /* win32: extern data imported from dll */
1096 #define VT_EXPORT 0x00008000 /* win32: data exported from dll */
1097 #define VT_WEAK 0x00010000 /* win32: data exported from dll */
1102 All symbols are stored in hashed symbol stacks. Each symbol stack
1103 contains @code{Sym} structures.
1105 @code{Sym.v} contains the symbol name (remember
1106 an identifier is also a token, so a string is never necessary to store
1107 it). @code{Sym.t} gives the type of the symbol. @code{Sym.r} is usually
1108 the register in which the corresponding variable is stored. @code{Sym.c} is
1109 usually a constant associated to the symbol like its address for normal
1110 symbols, and the number of entries for symbols representing arrays.
1111 Variable length array types use @code{Sym.c} as a location on the stack
1112 which holds the runtime sizeof for the type.
1114 Four main symbol stacks are defined:
1119 for the macros (@code{#define}s).
1122 for the global variables, functions and types.
1125 for the local variables, functions and types.
1127 @item global_label_stack
1128 for the local labels (for @code{goto}).
1131 for GCC block local labels (see the @code{__label__} keyword).
1135 @code{sym_push()} is used to add a new symbol in the local symbol
1136 stack. If no local symbol stack is active, it is added in the global
1139 @code{sym_pop(st,b)} pops symbols from the symbol stack @var{st} until
1140 the symbol @var{b} is on the top of stack. If @var{b} is NULL, the stack
1143 @code{sym_find(v)} return the symbol associated to the identifier
1144 @var{v}. The local stack is searched first from top to bottom, then the
1149 The generated code and data are written in sections. The structure
1150 @code{Section} contains all the necessary information for a given
1151 section. @code{new_section()} creates a new section. ELF file semantics
1152 is assumed for each section.
1154 The following sections are predefined:
1159 is the section containing the generated code. @var{ind} contains the
1160 current position in the code section.
1163 contains initialized data
1166 contains uninitialized data
1168 @item bounds_section
1169 @itemx lbounds_section
1170 are used when bound checking is activated
1173 @itemx stabstr_section
1174 are used when debugging is active to store debug information
1176 @item symtab_section
1177 @itemx strtab_section
1178 contain the exported symbols (currently only used for debugging).
1182 @section Code generation
1183 @cindex code generation
1185 @subsection Introduction
1187 The TCC code generator directly generates linked binary code in one
1188 pass. It is rather unusual these days (see gcc for example which
1189 generates text assembly), but it can be very fast and surprisingly
1192 The TCC code generator is register based. Optimization is only done at
1193 the expression level. No intermediate representation of expression is
1194 kept except the current values stored in the @emph{value stack}.
1196 On x86, three temporary registers are used. When more registers are
1197 needed, one register is spilled into a new temporary variable on the stack.
1199 @subsection The value stack
1200 @cindex value stack, introduction
1202 When an expression is parsed, its value is pushed on the value stack
1203 (@var{vstack}). The top of the value stack is @var{vtop}. Each value
1204 stack entry is the structure @code{SValue}.
1206 @code{SValue.t} is the type. @code{SValue.r} indicates how the value is
1207 currently stored in the generated code. It is usually a CPU register
1208 index (@code{REG_xxx} constants), but additional values and flags are
1212 #define VT_CONST 0x00f0
1213 #define VT_LLOCAL 0x00f1
1214 #define VT_LOCAL 0x00f2
1215 #define VT_CMP 0x00f3
1216 #define VT_JMP 0x00f4
1217 #define VT_JMPI 0x00f5
1218 #define VT_LVAL 0x0100
1219 #define VT_SYM 0x0200
1220 #define VT_MUSTCAST 0x0400
1221 #define VT_MUSTBOUND 0x0800
1222 #define VT_BOUNDED 0x8000
1223 #define VT_LVAL_BYTE 0x1000
1224 #define VT_LVAL_SHORT 0x2000
1225 #define VT_LVAL_UNSIGNED 0x4000
1226 #define VT_LVAL_TYPE (VT_LVAL_BYTE | VT_LVAL_SHORT | VT_LVAL_UNSIGNED)
1232 indicates that the value is a constant. It is stored in the union
1233 @code{SValue.c}, depending on its type.
1236 indicates a local variable pointer at offset @code{SValue.c.i} in the
1240 indicates that the value is actually stored in the CPU flags (i.e. the
1241 value is the consequence of a test). The value is either 0 or 1. The
1242 actual CPU flags used is indicated in @code{SValue.c.i}.
1244 If any code is generated which destroys the CPU flags, this value MUST be
1245 put in a normal register.
1249 indicates that the value is the consequence of a conditional jump. For VT_JMP,
1250 it is 1 if the jump is taken, 0 otherwise. For VT_JMPI it is inverted.
1252 These values are used to compile the @code{||} and @code{&&} logical
1255 If any code is generated, this value MUST be put in a normal
1256 register. Otherwise, the generated code won't be executed if the jump is
1260 is a flag indicating that the value is actually an lvalue (left value of
1261 an assignment). It means that the value stored is actually a pointer to
1264 Understanding the use @code{VT_LVAL} is very important if you want to
1265 understand how TCC works.
1268 @itemx VT_LVAL_SHORT
1269 @itemx VT_LVAL_UNSIGNED
1270 if the lvalue has an integer type, then these flags give its real
1271 type. The type alone is not enough in case of cast optimisations.
1274 is a saved lvalue on the stack. @code{VT_LVAL} must also be set with
1275 @code{VT_LLOCAL}. @code{VT_LLOCAL} can arise when a @code{VT_LVAL} in
1276 a register has to be saved to the stack, or it can come from an
1277 architecture-specific calling convention.
1280 indicates that a cast to the value type must be performed if the value
1281 is used (lazy casting).
1284 indicates that the symbol @code{SValue.sym} must be added to the constant.
1288 are only used for optional bound checking.
1292 @subsection Manipulating the value stack
1295 @code{vsetc()} and @code{vset()} pushes a new value on the value
1296 stack. If the previous @var{vtop} was stored in a very unsafe place(for
1297 example in the CPU flags), then some code is generated to put the
1298 previous @var{vtop} in a safe storage.
1300 @code{vpop()} pops @var{vtop}. In some cases, it also generates cleanup
1301 code (for example if stacked floating point registers are used as on
1304 The @code{gv(rc)} function generates code to evaluate @var{vtop} (the
1305 top value of the stack) into registers. @var{rc} selects in which
1306 register class the value should be put. @code{gv()} is the @emph{most
1307 important function} of the code generator.
1309 @code{gv2()} is the same as @code{gv()} but for the top two stack
1312 @subsection CPU dependent code generation
1313 @cindex CPU dependent
1314 See the @file{i386-gen.c} file to have an example.
1319 must generate the code needed to load a stack value into a register.
1322 must generate the code needed to store a register into a stack value
1326 @itemx gfunc_param()
1328 should generate a function call
1330 @item gfunc_prolog()
1331 @itemx gfunc_epilog()
1332 should generate a function prolog/epilog.
1335 must generate the binary integer operation @var{op} on the two top
1336 entries of the stack which are guaranteed to contain integer types.
1338 The result value should be put on the stack.
1341 same as @code{gen_opi()} for floating point operations. The two top
1342 entries of the stack are guaranteed to contain floating point values of
1345 @item gen_cvt_itof()
1346 integer to floating point conversion.
1348 @item gen_cvt_ftoi()
1349 floating point to integer conversion.
1351 @item gen_cvt_ftof()
1352 floating point to floating point of different size conversion.
1354 @item gen_bounded_ptr_add()
1355 @item gen_bounded_ptr_deref()
1356 are only used for bounds checking.
1360 @section Optimizations done
1361 @cindex optimizations
1362 @cindex constant propagation
1363 @cindex strength reduction
1364 @cindex comparison operators
1365 @cindex caching processor flags
1366 @cindex flags, caching
1367 @cindex jump optimization
1368 Constant propagation is done for all operations. Multiplications and
1369 divisions are optimized to shifts when appropriate. Comparison
1370 operators are optimized by maintaining a special cache for the
1371 processor flags. &&, || and ! are optimized by maintaining a special
1372 'jump target' value. No other jump optimization is currently performed
1373 because it would require to store the code in a more abstract fashion.
1375 @unnumbered Concept Index
1382 @c texinfo-column-for-description: 32