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
241 @item -ftest-coverage
242 Create code coverage code. After running the resulting code an executable.tcov
243 or sofile.tcov file is generated with code coverage.
251 Disable all warnings.
255 Note: each of the following warning options has a negative form beginning with
259 @item -Wimplicit-function-declaration
260 Warn about implicit function declaration.
263 Warn about unsupported GCC features that are ignored by TCC.
265 @item -Wwrite-strings
266 Make string constants be of type @code{const char *} instead of @code{char
270 Abort compilation if warnings are issued.
273 Activate all warnings, except @option{-Werror}, @option{-Wunusupported} and
274 @option{-Wwrite-strings}.
282 Specify an additional static library path for the @option{-l} option. The
283 default library paths are @file{/usr/local/lib}, @file{/usr/lib} and @file{/lib}.
286 Link your program with dynamic library libxxx.so or static library
287 libxxx.a. The library is searched in the paths specified by the
288 @option{-L} option and @env{LIBRARY_PATH} variable.
291 Set the path where the tcc internal libraries (and include files) can be
292 found (default is @file{PREFIX/lib/tcc}).
295 Generate a shared library instead of an executable.
298 set name for shared library to be used at runtime
301 Generate a statically linked executable (default is a shared linked
305 Export global symbols to the dynamic linker. It is useful when a library
306 opened with @code{dlopen()} needs to access executable symbols.
309 Generate an object file combining all input files.
311 @item -Wl,-rpath=path
312 Put custom search path for dynamic libraries into executable.
314 @item -Wl,--enable-new-dtags
315 When putting a custom search path for dynamic libraries into the executable,
316 create the new ELF dynamic tag DT_RUNPATH instead of the old legacy DT_RPATH.
318 @item -Wl,--oformat=fmt
319 Use @var{fmt} as output format. The supported output formats are:
322 ELF output format (default)
324 Binary image (only for executable output)
326 COFF output format (only for executable output for TMS320C67xx target)
329 @item -Wl,--export-all-symbols
330 @item -Wl,--export-dynamic
331 Export global symbols to the dynamic linker. It is useful when a library
332 opened with @code{dlopen()} needs to access executable symbols.
334 @item -Wl,-subsystem=console/gui/wince/...
335 Set type for PE (Windows) executables.
337 @item -Wl,-[Ttext=# | section-alignment=# | file-alignment=# | image-base=# | stack=#]
338 Modify executable layout.
343 @item -Wl,-(no-)whole-archive
344 Turn on/off linking of all objects in archives.
352 Generate run time debug information so that you get clear run time
353 error messages: @code{ test.c:68: in function 'test5()': dereferencing
354 invalid pointer} instead of the laconic @code{Segmentation
358 Generate additional support code to check memory allocations and array/pointer
359 bounds (@pxref{Bounds}). @option{-g} is implied.
362 Display N callers in stack traces. This is useful with @option{-g} or @option{-b}.
363 With executables, additional support for stack traces is included.
365 A function @code{ int tcc_backtrace(const char *fmt, ...); } is provided
366 to trigger a stack trace with a message on demand.
375 Just output makefile fragment with dependencies
378 Like -M except mention only user header files, not system header files.
381 Generate makefile fragment with dependencies.
384 Like -MD except mention only user header files, not system header files.
387 Use @file{depfile} as output for -MD.
389 @item -print-search-dirs
390 Print the configured installation directory and a list of library
391 and include directories tcc will search.
398 Target specific options:
402 Use an algorithm for bitfield alignment consistent with MSVC. Default is
405 @item -mfloat-abi (ARM only)
406 Select the float ABI. Possible values: @code{softfp} and @code{hard}
409 Do not use sse registers on x86_64
412 Pass command line to the i386/x86_64 cross compiler.
416 Note: GCC options @option{-Ox}, @option{-fx} and @option{-mx} are
420 @c man begin ENVIRONMENT
421 Environment variables that affect how tcc operates.
427 A colon-separated list of directories searched for include files,
428 directories given with @option{-I} are searched first.
431 A colon-separated list of directories searched for libraries for the
432 @option{-l} option, directories given with @option{-L} are searched first.
441 @settitle Tiny C Compiler
455 @chapter C language support
459 TCC implements all the ANSI C standard, including structure bit fields
460 and floating point numbers (@code{long double}, @code{double}, and
461 @code{float} fully supported).
463 @section ISOC99 extensions
465 TCC implements many features of the new C standard: ISO C99. Currently
466 missing items are: complex and imaginary numbers.
468 Currently implemented ISOC99 features:
472 @item variable length arrays.
474 @item 64 bit @code{long long} types are fully supported.
476 @item The boolean type @code{_Bool} is supported.
478 @item @code{__func__} is a string variable containing the current
481 @item Variadic macros: @code{__VA_ARGS__} can be used for
482 function-like macros:
484 #define dprintf(level, __VA_ARGS__) printf(__VA_ARGS__)
488 @code{dprintf} can then be used with a variable number of parameters.
490 @item Declarations can appear anywhere in a block (as in C++).
492 @item Array and struct/union elements can be initialized in any order by
495 struct @{ int x, y; @} st[10] = @{ [0].x = 1, [0].y = 2 @};
497 int tab[10] = @{ 1, 2, [5] = 5, [9] = 9@};
500 @item Compound initializers are supported:
502 int *p = (int [])@{ 1, 2, 3 @};
504 to initialize a pointer pointing to an initialized array. The same
505 works for structures and strings.
507 @item Hexadecimal floating point constants are supported:
509 double d = 0x1234p10;
513 is the same as writing
515 double d = 4771840.0;
518 @item @code{inline} keyword is ignored.
520 @item @code{restrict} keyword is ignored.
523 @section GNU C extensions
525 TCC implements some GNU C extensions:
529 @item array designators can be used without '=':
531 int a[10] = @{ [0] 1, [5] 2, 3, 4 @};
534 @item Structure field designators can be a label:
536 struct @{ int x, y; @} st = @{ x: 1, y: 1@};
540 struct @{ int x, y; @} st = @{ .x = 1, .y = 1@};
543 @item @code{\e} is ASCII character 27.
545 @item case ranges : ranges can be used in @code{case}s:
549 printf("range 1 to 9\n");
552 printf("unexpected\n");
557 @cindex aligned attribute
558 @cindex packed attribute
559 @cindex section attribute
560 @cindex unused attribute
561 @cindex cdecl attribute
562 @cindex stdcall attribute
563 @cindex regparm attribute
564 @cindex dllexport attribute
565 @cindex nodecorate attribute
567 @item The keyword @code{__attribute__} is handled to specify variable or
568 function attributes. The following attributes are supported:
571 @item @code{aligned(n)}: align a variable or a structure field to n bytes
572 (must be a power of two).
574 @item @code{packed}: force alignment of a variable or a structure field to
577 @item @code{section(name)}: generate function or data in assembly section
578 name (name is a string containing the section name) instead of the default
581 @item @code{unused}: specify that the variable or the function is unused.
583 @item @code{cdecl}: use standard C calling convention (default).
585 @item @code{stdcall}: use Pascal-like calling convention.
587 @item @code{regparm(n)}: use fast i386 calling convention. @var{n} must be
588 between 1 and 3. The first @var{n} function parameters are respectively put in
589 registers @code{%eax}, @code{%edx} and @code{%ecx}.
591 @item @code{dllexport}: export function from dll/executable (win32 only)
593 @item @code{nodecorate}: do not apply any decorations that would otherwise be applied when exporting function from dll/executable (win32 only)
597 Here are some examples:
599 int a __attribute__ ((aligned(8), section(".mysection")));
603 align variable @code{a} to 8 bytes and put it in section @code{.mysection}.
606 int my_add(int a, int b) __attribute__ ((section(".mycodesection")))
613 generate function @code{my_add} in section @code{.mycodesection}.
615 @item GNU style variadic macros:
617 #define dprintf(fmt, args@dots{}) printf(fmt, ## args)
620 dprintf("one arg %d\n", 1);
623 @item @code{__FUNCTION__} is interpreted as C99 @code{__func__}
624 (so it has not exactly the same semantics as string literal GNUC
625 where it is a string literal).
627 @item The @code{__alignof__} keyword can be used as @code{sizeof}
628 to get the alignment of a type or an expression.
630 @item The @code{typeof(x)} returns the type of @code{x}.
631 @code{x} is an expression or a type.
633 @item Computed gotos: @code{&&label} returns a pointer of type
634 @code{void *} on the goto label @code{label}. @code{goto *expr} can be
635 used to jump on the pointer resulting from @code{expr}.
637 @item Inline assembly with asm instruction:
638 @cindex inline assembly
639 @cindex assembly, inline
642 static inline void * my_memcpy(void * to, const void * from, size_t n)
645 __asm__ __volatile__(
650 "1:\ttestb $1,%b4\n\t"
654 : "=&c" (d0), "=&D" (d1), "=&S" (d2)
655 :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
663 TCC includes its own x86 inline assembler with a @code{gas}-like (GNU
664 assembler) syntax. No intermediate files are generated. GCC 3.x named
665 operands are supported.
667 @item @code{__builtin_types_compatible_p()} and @code{__builtin_constant_p()}
670 @item @code{#pragma pack} is supported for win32 compatibility.
674 @section TinyCC extensions
678 @item @code{__TINYC__} is a predefined macro to indicate that you use TCC.
680 @item @code{#!} at the start of a line is ignored to allow scripting.
682 @item Binary digits can be entered (@code{0b101} instead of
688 @chapter TinyCC Assembler
690 Since version 0.9.16, TinyCC integrates its own assembler. TinyCC
691 assembler supports a gas-like syntax (GNU assembler). You can
692 deactivate assembler support if you want a smaller TinyCC executable
693 (the C compiler does not rely on the assembler).
695 TinyCC Assembler is used to handle files with @file{.S} (C
696 preprocessed assembler) and @file{.s} extensions. It is also used to
697 handle the GNU inline assembler with the @code{asm} keyword.
701 TinyCC Assembler supports most of the gas syntax. The tokens are the
706 @item C and C++ comments are supported.
708 @item Identifiers are the same as C, so you cannot use '.' or '$'.
710 @item Only 32 bit integer numbers are supported.
718 @item Integers in decimal, octal and hexa are supported.
720 @item Unary operators: +, -, ~.
722 @item Binary operators in decreasing priority order:
730 @item A value is either an absolute number or a label plus an offset.
731 All operators accept absolute values except '+' and '-'. '+' or '-' can be
732 used to add an offset to a label. '-' supports two labels only if they
733 are the same or if they are both defined and in the same section.
741 @item All labels are considered as local, except undefined ones.
743 @item Numeric labels can be used as local @code{gas}-like labels.
744 They can be defined several times in the same source. Use 'b'
745 (backward) or 'f' (forward) as suffix to reference them:
749 jmp 1b /* jump to '1' label before */
750 jmp 1f /* jump to '1' label after */
757 @cindex assembler directives
758 @cindex directives, assembler
759 @cindex align directive
760 @cindex skip directive
761 @cindex space directive
762 @cindex byte directive
763 @cindex word directive
764 @cindex short directive
765 @cindex int directive
766 @cindex long directive
767 @cindex quad directive
768 @cindex globl directive
769 @cindex global directive
770 @cindex section directive
771 @cindex text directive
772 @cindex data directive
773 @cindex bss directive
774 @cindex fill directive
775 @cindex org directive
776 @cindex previous directive
777 @cindex string directive
778 @cindex asciz directive
779 @cindex ascii directive
781 All directives are preceded by a '.'. The following directives are
785 @item .align n[,value]
786 @item .skip n[,value]
787 @item .space n[,value]
788 @item .byte value1[,...]
789 @item .word value1[,...]
790 @item .short value1[,...]
791 @item .int value1[,...]
792 @item .long value1[,...]
793 @item .quad immediate_value1[,...]
796 @item .section section
800 @item .fill repeat[,size[,value]]
803 @item .string string[,...]
804 @item .asciz string[,...]
805 @item .ascii string[,...]
808 @section X86 Assembler
811 All X86 opcodes are supported. Only ATT syntax is supported (source
812 then destination operand order). If no size suffix is given, TinyCC
813 tries to guess it from the operand sizes.
815 Currently, MMX opcodes are supported but not SSE ones.
818 @chapter TinyCC Linker
821 @section ELF file generation
824 TCC can directly output relocatable ELF files (object files),
825 executable ELF files and dynamic ELF libraries without relying on an
828 Dynamic ELF libraries can be output but the C compiler does not generate
829 position independent code (PIC). It means that the dynamic library
830 code generated by TCC cannot be factorized among processes yet.
832 TCC linker eliminates unreferenced object code in libraries. A single pass is
833 done on the object and library list, so the order in which object files and
834 libraries are specified is important (same constraint as GNU ld). No grouping
835 options (@option{--start-group} and @option{--end-group}) are supported.
837 @section ELF file loader
839 TCC can load ELF object files, archives (.a files) and dynamic
842 @section PE-i386 file generation
845 TCC for Windows supports the native Win32 executable file format (PE-i386). It
846 generates EXE files (console and gui) and DLL files.
848 For usage on Windows, see also tcc-win32.txt.
850 @section GNU Linker Scripts
851 @cindex scripts, linker
852 @cindex linker scripts
853 @cindex GROUP, linker command
854 @cindex FILE, linker command
855 @cindex OUTPUT_FORMAT, linker command
856 @cindex TARGET, linker command
858 Because on many Linux systems some dynamic libraries (such as
859 @file{/usr/lib/libc.so}) are in fact GNU ld link scripts (horrible!),
860 the TCC linker also supports a subset of GNU ld scripts.
862 The @code{GROUP} and @code{FILE} commands are supported. @code{OUTPUT_FORMAT}
863 and @code{TARGET} are ignored.
865 Example from @file{/usr/lib/libc.so}:
868 Use the shared library, but some functions are only in
869 the static library, so try that secondarily. */
870 GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a )
874 @chapter TinyCC Memory and Bound checks
876 @cindex memory checks
878 This feature is activated with the @option{-b} option (@pxref{Invoke}).
879 Here are some examples of caught errors:
883 @item Invalid range with standard string function:
891 @item Out of bounds-error in global or local arrays:
901 @item Out of bounds-error in malloc'ed data:
905 tab = malloc(20 * sizeof(int));
913 @item Access of freed memory:
917 tab = malloc(20 * sizeof(int));
929 tab = malloc(20 * sizeof(int));
936 TCC defines @code{__BOUNDS_CHECKING_ON} if activated.
938 There are five environment variables that can be used to control the behavior:
940 @item TCC_BOUNDS_WARN_POINTER_ADD
941 - Print warning when pointer add creates an illegal pointer.
942 @item TCC_BOUNDS_PRINT_CALLS
943 - Print bound checking calls. Can be used for debugging.
944 @item TCC_BOUNDS_PRINT_HEAP
945 - Print heap objects that are not freed at exit of program.
946 @item TCC_BOUNDS_PRINT_STATISTIC
947 - Print statistic information at exit of program.
948 @item TCC_BOUNDS_NEVER_FATAL
949 - Try to continue in case of a bound checking error.
952 Also, a function @code{__bounds_checking(x)} can be used to turn off/on bounds
953 checking from usercode (see below).
957 @item Only available on i386 (linux and windows), x86_64 (linux and windows),
958 arm, arm64 and riscv64 for the moment.
959 @item The generated code is slower and bigger.
960 @item The bound checking code is not included in shared libraries. The main
961 executable should always be compiled with the @option{-b}.
962 @item Pointer size is @emph{unchanged} and code generated with bound checks is
963 @emph{fully compatible} with unchecked code. When a pointer comes from
964 unchecked code, it is assumed to be valid. Even very obscure C code with
965 casts should work correctly.
966 @item Signal handlers are not compatible with bounds checking. The
967 bounds checking code disables checking in signal/sigaction handlers.
968 The fork() function call in a multi threaded application is also a problem.
969 The bound checking code fixes this for the child process.
970 @item The reason that signals and fork have problems is that we use locking
971 inside the bounds checking code.
972 Inside a signal handler we can not use locks. Also in a multi threaded
973 application after a fork the child process can have the lock set
975 @item The BOUNDS_CHECKING_OFF and BOUNDS_CHECKING_ON can also be used to
976 disable bounds checking for some code.
977 @item The __bounds_checking call adds a value to a thread local value.
978 The value starts at 0. If the value is not 0 the code is not checked
979 for bounds checking errors.
983 #if defined(__TINYC__) && __BOUNDS_CHECKING_ON
984 extern void __bounds_checking (int x);
985 # define BOUNDS_CHECKING_OFF __bounds_checking(1)
986 # define BOUNDS_CHECKING_ON __bounds_checking(-1)
988 # define BOUNDS_CHECKING_OFF
989 # define BOUNDS_CHECKING_ON
993 For more information about the ideas behind this method, see
994 @url{http://www.doc.ic.ac.uk/~phjk/BoundsChecking.html}.
997 @chapter The @code{libtcc} library
999 The @code{libtcc} library enables you to use TCC as a backend for
1000 dynamic code generation.
1002 Read the @file{libtcc.h} to have an overview of the API. Read
1003 @file{libtcc_test.c} to have a very simple example.
1005 The idea consists in giving a C string containing the program you want
1006 to compile directly to @code{libtcc}. Then you can access to any global
1007 symbol (function or variable) defined.
1010 @chapter Developer's guide
1012 This chapter gives some hints to understand how TCC works. You can skip
1013 it if you do not intend to modify the TCC code.
1015 @section File reading
1017 The @code{BufferedFile} structure contains the context needed to read a
1018 file, including the current line number. @code{tcc_open()} opens a new
1019 file and @code{tcc_close()} closes it. @code{inp()} returns the next
1024 @code{next()} reads the next token in the current
1025 file. @code{next_nomacro()} reads the next token without macro
1028 @code{tok} contains the current token (see @code{TOK_xxx})
1029 constants. Identifiers and keywords are also keywords. @code{tokc}
1030 contains additional infos about the token (for example a constant value
1031 if number or string token).
1035 The parser is hardcoded (yacc is not necessary). It does only one pass,
1040 @item For initialized arrays with unknown size, a first pass
1041 is done to count the number of elements.
1043 @item For architectures where arguments are evaluated in
1044 reverse order, a first pass is done to reverse the argument order.
1050 The types are stored in a single 'int' variable. It was chosen in the
1051 first stages of development when tcc was much simpler. Now, it may not
1052 be the best solution.
1055 #define VT_INT 0 /* integer type */
1056 #define VT_BYTE 1 /* signed byte type */
1057 #define VT_SHORT 2 /* short type */
1058 #define VT_VOID 3 /* void type */
1059 #define VT_PTR 4 /* pointer */
1060 #define VT_ENUM 5 /* enum definition */
1061 #define VT_FUNC 6 /* function type */
1062 #define VT_STRUCT 7 /* struct/union definition */
1063 #define VT_FLOAT 8 /* IEEE float */
1064 #define VT_DOUBLE 9 /* IEEE double */
1065 #define VT_LDOUBLE 10 /* IEEE long double */
1066 #define VT_BOOL 11 /* ISOC99 boolean type */
1067 #define VT_LLONG 12 /* 64 bit integer */
1068 #define VT_LONG 13 /* long integer (NEVER USED as type, only
1070 #define VT_BTYPE 0x000f /* mask for basic type */
1071 #define VT_UNSIGNED 0x0010 /* unsigned type */
1072 #define VT_ARRAY 0x0020 /* array type (also has VT_PTR) */
1073 #define VT_VLA 0x20000 /* VLA type (also has VT_PTR and VT_ARRAY) */
1074 #define VT_BITFIELD 0x0040 /* bitfield modifier */
1075 #define VT_CONSTANT 0x0800 /* const modifier */
1076 #define VT_VOLATILE 0x1000 /* volatile modifier */
1077 #define VT_DEFSIGN 0x2000 /* signed type */
1079 #define VT_STRUCT_SHIFT 18 /* structure/enum name shift (14 bits left) */
1082 When a reference to another type is needed (for pointers, functions and
1083 structures), the @code{32 - VT_STRUCT_SHIFT} high order bits are used to
1084 store an identifier reference.
1086 The @code{VT_UNSIGNED} flag can be set for chars, shorts, ints and long
1089 Arrays are considered as pointers @code{VT_PTR} with the flag
1090 @code{VT_ARRAY} set. Variable length arrays are considered as special
1091 arrays and have flag @code{VT_VLA} set instead of @code{VT_ARRAY}.
1093 The @code{VT_BITFIELD} flag can be set for chars, shorts, ints and long
1094 longs. If it is set, then the bitfield position is stored from bits
1095 VT_STRUCT_SHIFT to VT_STRUCT_SHIFT + 5 and the bit field size is stored
1096 from bits VT_STRUCT_SHIFT + 6 to VT_STRUCT_SHIFT + 11.
1098 @code{VT_LONG} is never used except during parsing.
1100 During parsing, the storage of an object is also stored in the type
1104 #define VT_EXTERN 0x00000080 /* extern definition */
1105 #define VT_STATIC 0x00000100 /* static variable */
1106 #define VT_TYPEDEF 0x00000200 /* typedef definition */
1107 #define VT_INLINE 0x00000400 /* inline definition */
1108 #define VT_IMPORT 0x00004000 /* win32: extern data imported from dll */
1109 #define VT_EXPORT 0x00008000 /* win32: data exported from dll */
1110 #define VT_WEAK 0x00010000 /* win32: data exported from dll */
1115 All symbols are stored in hashed symbol stacks. Each symbol stack
1116 contains @code{Sym} structures.
1118 @code{Sym.v} contains the symbol name (remember
1119 an identifier is also a token, so a string is never necessary to store
1120 it). @code{Sym.t} gives the type of the symbol. @code{Sym.r} is usually
1121 the register in which the corresponding variable is stored. @code{Sym.c} is
1122 usually a constant associated to the symbol like its address for normal
1123 symbols, and the number of entries for symbols representing arrays.
1124 Variable length array types use @code{Sym.c} as a location on the stack
1125 which holds the runtime sizeof for the type.
1127 Four main symbol stacks are defined:
1132 for the macros (@code{#define}s).
1135 for the global variables, functions and types.
1138 for the local variables, functions and types.
1140 @item global_label_stack
1141 for the local labels (for @code{goto}).
1144 for GCC block local labels (see the @code{__label__} keyword).
1148 @code{sym_push()} is used to add a new symbol in the local symbol
1149 stack. If no local symbol stack is active, it is added in the global
1152 @code{sym_pop(st,b)} pops symbols from the symbol stack @var{st} until
1153 the symbol @var{b} is on the top of stack. If @var{b} is NULL, the stack
1156 @code{sym_find(v)} return the symbol associated to the identifier
1157 @var{v}. The local stack is searched first from top to bottom, then the
1162 The generated code and data are written in sections. The structure
1163 @code{Section} contains all the necessary information for a given
1164 section. @code{new_section()} creates a new section. ELF file semantics
1165 is assumed for each section.
1167 The following sections are predefined:
1172 is the section containing the generated code. @var{ind} contains the
1173 current position in the code section.
1176 contains initialized data
1179 contains uninitialized data
1181 @item bounds_section
1182 @itemx lbounds_section
1183 are used when bound checking is activated
1186 @itemx stabstr_section
1187 are used when debugging is active to store debug information
1189 @item symtab_section
1190 @itemx strtab_section
1191 contain the exported symbols (currently only used for debugging).
1195 @section Code generation
1196 @cindex code generation
1198 @subsection Introduction
1200 The TCC code generator directly generates linked binary code in one
1201 pass. It is rather unusual these days (see gcc for example which
1202 generates text assembly), but it can be very fast and surprisingly
1205 The TCC code generator is register based. Optimization is only done at
1206 the expression level. No intermediate representation of expression is
1207 kept except the current values stored in the @emph{value stack}.
1209 On x86, three temporary registers are used. When more registers are
1210 needed, one register is spilled into a new temporary variable on the stack.
1212 @subsection The value stack
1213 @cindex value stack, introduction
1215 When an expression is parsed, its value is pushed on the value stack
1216 (@var{vstack}). The top of the value stack is @var{vtop}. Each value
1217 stack entry is the structure @code{SValue}.
1219 @code{SValue.t} is the type. @code{SValue.r} indicates how the value is
1220 currently stored in the generated code. It is usually a CPU register
1221 index (@code{REG_xxx} constants), but additional values and flags are
1225 #define VT_CONST 0x00f0
1226 #define VT_LLOCAL 0x00f1
1227 #define VT_LOCAL 0x00f2
1228 #define VT_CMP 0x00f3
1229 #define VT_JMP 0x00f4
1230 #define VT_JMPI 0x00f5
1231 #define VT_LVAL 0x0100
1232 #define VT_SYM 0x0200
1233 #define VT_MUSTCAST 0x0400
1234 #define VT_MUSTBOUND 0x0800
1235 #define VT_BOUNDED 0x8000
1236 #define VT_LVAL_BYTE 0x1000
1237 #define VT_LVAL_SHORT 0x2000
1238 #define VT_LVAL_UNSIGNED 0x4000
1239 #define VT_LVAL_TYPE (VT_LVAL_BYTE | VT_LVAL_SHORT | VT_LVAL_UNSIGNED)
1245 indicates that the value is a constant. It is stored in the union
1246 @code{SValue.c}, depending on its type.
1249 indicates a local variable pointer at offset @code{SValue.c.i} in the
1253 indicates that the value is actually stored in the CPU flags (i.e. the
1254 value is the consequence of a test). The value is either 0 or 1. The
1255 actual CPU flags used is indicated in @code{SValue.c.i}.
1257 If any code is generated which destroys the CPU flags, this value MUST be
1258 put in a normal register.
1262 indicates that the value is the consequence of a conditional jump. For VT_JMP,
1263 it is 1 if the jump is taken, 0 otherwise. For VT_JMPI it is inverted.
1265 These values are used to compile the @code{||} and @code{&&} logical
1268 If any code is generated, this value MUST be put in a normal
1269 register. Otherwise, the generated code won't be executed if the jump is
1273 is a flag indicating that the value is actually an lvalue (left value of
1274 an assignment). It means that the value stored is actually a pointer to
1277 Understanding the use @code{VT_LVAL} is very important if you want to
1278 understand how TCC works.
1281 @itemx VT_LVAL_SHORT
1282 @itemx VT_LVAL_UNSIGNED
1283 if the lvalue has an integer type, then these flags give its real
1284 type. The type alone is not enough in case of cast optimisations.
1287 is a saved lvalue on the stack. @code{VT_LVAL} must also be set with
1288 @code{VT_LLOCAL}. @code{VT_LLOCAL} can arise when a @code{VT_LVAL} in
1289 a register has to be saved to the stack, or it can come from an
1290 architecture-specific calling convention.
1293 indicates that a cast to the value type must be performed if the value
1294 is used (lazy casting).
1297 indicates that the symbol @code{SValue.sym} must be added to the constant.
1301 are only used for optional bound checking.
1305 @subsection Manipulating the value stack
1308 @code{vsetc()} and @code{vset()} pushes a new value on the value
1309 stack. If the previous @var{vtop} was stored in a very unsafe place(for
1310 example in the CPU flags), then some code is generated to put the
1311 previous @var{vtop} in a safe storage.
1313 @code{vpop()} pops @var{vtop}. In some cases, it also generates cleanup
1314 code (for example if stacked floating point registers are used as on
1317 The @code{gv(rc)} function generates code to evaluate @var{vtop} (the
1318 top value of the stack) into registers. @var{rc} selects in which
1319 register class the value should be put. @code{gv()} is the @emph{most
1320 important function} of the code generator.
1322 @code{gv2()} is the same as @code{gv()} but for the top two stack
1325 @subsection CPU dependent code generation
1326 @cindex CPU dependent
1327 See the @file{i386-gen.c} file to have an example.
1332 must generate the code needed to load a stack value into a register.
1335 must generate the code needed to store a register into a stack value
1339 @itemx gfunc_param()
1341 should generate a function call
1343 @item gfunc_prolog()
1344 @itemx gfunc_epilog()
1345 should generate a function prolog/epilog.
1348 must generate the binary integer operation @var{op} on the two top
1349 entries of the stack which are guaranteed to contain integer types.
1351 The result value should be put on the stack.
1354 same as @code{gen_opi()} for floating point operations. The two top
1355 entries of the stack are guaranteed to contain floating point values of
1358 @item gen_cvt_itof()
1359 integer to floating point conversion.
1361 @item gen_cvt_ftoi()
1362 floating point to integer conversion.
1364 @item gen_cvt_ftof()
1365 floating point to floating point of different size conversion.
1369 @section Optimizations done
1370 @cindex optimizations
1371 @cindex constant propagation
1372 @cindex strength reduction
1373 @cindex comparison operators
1374 @cindex caching processor flags
1375 @cindex flags, caching
1376 @cindex jump optimization
1377 Constant propagation is done for all operations. Multiplications and
1378 divisions are optimized to shifts when appropriate. Comparison
1379 operators are optimized by maintaining a special cache for the
1380 processor flags. &&, || and ! are optimized by maintaining a special
1381 'jump target' value. No other jump optimization is currently performed
1382 because it would require to store the code in a more abstract fashion.
1384 @unnumbered Concept Index
1391 @c texinfo-column-for-description: 32