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
180 Print only the compiler version and nothing else.
186 Show included files. As sole argument, print search dirs (as below).
189 Display compilation statistics.
191 @item -print-search-dirs
192 Print the configured installation directory and a list of library
193 and include directories tcc will search.
197 Preprocessor options:
201 Specify an additional include path. Include paths are searched in the
202 order they are specified.
204 System include paths are always searched after. The default system
205 include paths are: @file{/usr/local/include}, @file{/usr/include}
206 and @file{PREFIX/lib/tcc/include}. (@file{PREFIX} is usually
207 @file{/usr} or @file{/usr/local}).
210 Define preprocessor symbol @samp{sym} to
211 val. If val is not present, its value is @samp{1}. Function-like macros can
212 also be defined: @option{-DF(a)=a+1}
215 Undefine preprocessor symbol @samp{sym}.
220 Note: each of the following warning options has a negative form beginning with
224 @item -funsigned-char
225 Let the @code{char} type be unsigned.
228 Let the @code{char} type be signed.
231 Do not generate common symbols for uninitialized data.
233 @item -fleading-underscore
234 Add a leading underscore at the beginning of each C symbol.
242 Disable all warnings.
246 Note: each of the following warning options has a negative form beginning with
250 @item -Wimplicit-function-declaration
251 Warn about implicit function declaration.
254 Warn about unsupported GCC features that are ignored by TCC.
256 @item -Wwrite-strings
257 Make string constants be of type @code{const char *} instead of @code{char
261 Abort compilation if warnings are issued.
264 Activate all warnings, except @option{-Werror}, @option{-Wunusupported} and
265 @option{-Wwrite-strings}.
273 Specify an additional static library path for the @option{-l} option. The
274 default library paths are @file{/usr/local/lib}, @file{/usr/lib} and @file{/lib}.
277 Link your program with dynamic library libxxx.so or static library
278 libxxx.a. The library is searched in the paths specified by the
282 Set the path where the tcc internal libraries (and include files) can be
283 found (default is @file{PREFIX/lib/tcc}).
286 Generate a shared library instead of an executable.
289 set name for shared library to be used at runtime
292 Generate a statically linked executable (default is a shared linked
296 Export global symbols to the dynamic linker. It is useful when a library
297 opened with @code{dlopen()} needs to access executable symbols.
300 Generate an object file combining all input files.
302 @item -Wl,-rpath=path
303 Put custom seatch path for dynamic libraries into executable.
305 @item -Wl,--oformat=fmt
306 Use @var{fmt} as output format. The supported output formats are:
309 ELF output format (default)
311 Binary image (only for executable output)
313 COFF output format (only for executable output for TMS320C67xx target)
316 @item -Wl,-subsystem=console/gui/wince/...
317 Set type for PE (Windows) executables.
319 @item -Wl,-[Ttext=# | section-alignment=# | file-alignment=# | image-base=# | stack=#]
320 Modify executable layout.
331 Generate run time debug information so that you get clear run time
332 error messages: @code{ test.c:68: in function 'test5()': dereferencing
333 invalid pointer} instead of the laconic @code{Segmentation
337 Generate additional support code to check
338 memory allocations and array/pointer bounds. @option{-g} is implied. Note
339 that the generated code is slower and bigger in this case.
341 Note: @option{-b} is only available on i386 for the moment.
344 Display N callers in stack traces. This is useful with @option{-g} or
353 Generate makefile fragment with dependencies.
356 Use @file{depfile} as output for -MD.
360 Note: GCC options @option{-Ox}, @option{-fx} and @option{-mx} are
367 @settitle Tiny C Compiler
380 @chapter C language support
384 TCC implements all the ANSI C standard, including structure bit fields
385 and floating point numbers (@code{long double}, @code{double}, and
386 @code{float} fully supported).
388 @section ISOC99 extensions
390 TCC implements many features of the new C standard: ISO C99. Currently
391 missing items are: complex and imaginary numbers.
393 Currently implemented ISOC99 features:
397 @item variable length arrays.
399 @item 64 bit @code{long long} types are fully supported.
401 @item The boolean type @code{_Bool} is supported.
403 @item @code{__func__} is a string variable containing the current
406 @item Variadic macros: @code{__VA_ARGS__} can be used for
407 function-like macros:
409 #define dprintf(level, __VA_ARGS__) printf(__VA_ARGS__)
413 @code{dprintf} can then be used with a variable number of parameters.
415 @item Declarations can appear anywhere in a block (as in C++).
417 @item Array and struct/union elements can be initialized in any order by
420 struct @{ int x, y; @} st[10] = @{ [0].x = 1, [0].y = 2 @};
422 int tab[10] = @{ 1, 2, [5] = 5, [9] = 9@};
425 @item Compound initializers are supported:
427 int *p = (int [])@{ 1, 2, 3 @};
429 to initialize a pointer pointing to an initialized array. The same
430 works for structures and strings.
432 @item Hexadecimal floating point constants are supported:
434 double d = 0x1234p10;
438 is the same as writing
440 double d = 4771840.0;
443 @item @code{inline} keyword is ignored.
445 @item @code{restrict} keyword is ignored.
448 @section GNU C extensions
450 TCC implements some GNU C extensions:
454 @item array designators can be used without '=':
456 int a[10] = @{ [0] 1, [5] 2, 3, 4 @};
459 @item Structure field designators can be a label:
461 struct @{ int x, y; @} st = @{ x: 1, y: 1@};
465 struct @{ int x, y; @} st = @{ .x = 1, .y = 1@};
468 @item @code{\e} is ASCII character 27.
470 @item case ranges : ranges can be used in @code{case}s:
474 printf("range 1 to 9\n");
477 printf("unexpected\n");
482 @cindex aligned attribute
483 @cindex packed attribute
484 @cindex section attribute
485 @cindex unused attribute
486 @cindex cdecl attribute
487 @cindex stdcall attribute
488 @cindex regparm attribute
489 @cindex dllexport attribute
491 @item The keyword @code{__attribute__} is handled to specify variable or
492 function attributes. The following attributes are supported:
495 @item @code{aligned(n)}: align a variable or a structure field to n bytes
496 (must be a power of two).
498 @item @code{packed}: force alignment of a variable or a structure field to
501 @item @code{section(name)}: generate function or data in assembly section
502 name (name is a string containing the section name) instead of the default
505 @item @code{unused}: specify that the variable or the function is unused.
507 @item @code{cdecl}: use standard C calling convention (default).
509 @item @code{stdcall}: use Pascal-like calling convention.
511 @item @code{regparm(n)}: use fast i386 calling convention. @var{n} must be
512 between 1 and 3. The first @var{n} function parameters are respectively put in
513 registers @code{%eax}, @code{%edx} and @code{%ecx}.
515 @item @code{dllexport}: export function from dll/executable (win32 only)
519 Here are some examples:
521 int a __attribute__ ((aligned(8), section(".mysection")));
525 align variable @code{a} to 8 bytes and put it in section @code{.mysection}.
528 int my_add(int a, int b) __attribute__ ((section(".mycodesection")))
535 generate function @code{my_add} in section @code{.mycodesection}.
537 @item GNU style variadic macros:
539 #define dprintf(fmt, args@dots{}) printf(fmt, ## args)
542 dprintf("one arg %d\n", 1);
545 @item @code{__FUNCTION__} is interpreted as C99 @code{__func__}
546 (so it has not exactly the same semantics as string literal GNUC
547 where it is a string literal).
549 @item The @code{__alignof__} keyword can be used as @code{sizeof}
550 to get the alignment of a type or an expression.
552 @item The @code{typeof(x)} returns the type of @code{x}.
553 @code{x} is an expression or a type.
555 @item Computed gotos: @code{&&label} returns a pointer of type
556 @code{void *} on the goto label @code{label}. @code{goto *expr} can be
557 used to jump on the pointer resulting from @code{expr}.
559 @item Inline assembly with asm instruction:
560 @cindex inline assembly
561 @cindex assembly, inline
564 static inline void * my_memcpy(void * to, const void * from, size_t n)
567 __asm__ __volatile__(
572 "1:\ttestb $1,%b4\n\t"
576 : "=&c" (d0), "=&D" (d1), "=&S" (d2)
577 :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
585 TCC includes its own x86 inline assembler with a @code{gas}-like (GNU
586 assembler) syntax. No intermediate files are generated. GCC 3.x named
587 operands are supported.
589 @item @code{__builtin_types_compatible_p()} and @code{__builtin_constant_p()}
592 @item @code{#pragma pack} is supported for win32 compatibility.
596 @section TinyCC extensions
600 @item @code{__TINYC__} is a predefined macro to indicate that you use TCC.
602 @item @code{#!} at the start of a line is ignored to allow scripting.
604 @item Binary digits can be entered (@code{0b101} instead of
607 @item @code{__BOUNDS_CHECKING_ON} is defined if bound checking is activated.
612 @chapter TinyCC Assembler
614 Since version 0.9.16, TinyCC integrates its own assembler. TinyCC
615 assembler supports a gas-like syntax (GNU assembler). You can
616 desactivate assembler support if you want a smaller TinyCC executable
617 (the C compiler does not rely on the assembler).
619 TinyCC Assembler is used to handle files with @file{.S} (C
620 preprocessed assembler) and @file{.s} extensions. It is also used to
621 handle the GNU inline assembler with the @code{asm} keyword.
625 TinyCC Assembler supports most of the gas syntax. The tokens are the
630 @item C and C++ comments are supported.
632 @item Identifiers are the same as C, so you cannot use '.' or '$'.
634 @item Only 32 bit integer numbers are supported.
642 @item Integers in decimal, octal and hexa are supported.
644 @item Unary operators: +, -, ~.
646 @item Binary operators in decreasing priority order:
654 @item A value is either an absolute number or a label plus an offset.
655 All operators accept absolute values except '+' and '-'. '+' or '-' can be
656 used to add an offset to a label. '-' supports two labels only if they
657 are the same or if they are both defined and in the same section.
665 @item All labels are considered as local, except undefined ones.
667 @item Numeric labels can be used as local @code{gas}-like labels.
668 They can be defined several times in the same source. Use 'b'
669 (backward) or 'f' (forward) as suffix to reference them:
673 jmp 1b /* jump to '1' label before */
674 jmp 1f /* jump to '1' label after */
681 @cindex assembler directives
682 @cindex directives, assembler
683 @cindex align directive
684 @cindex skip directive
685 @cindex space directive
686 @cindex byte directive
687 @cindex word directive
688 @cindex short directive
689 @cindex int directive
690 @cindex long directive
691 @cindex quad directive
692 @cindex globl directive
693 @cindex global directive
694 @cindex section directive
695 @cindex text directive
696 @cindex data directive
697 @cindex bss directive
698 @cindex fill directive
699 @cindex org directive
700 @cindex previous directive
701 @cindex string directive
702 @cindex asciz directive
703 @cindex ascii directive
705 All directives are preceeded by a '.'. The following directives are
709 @item .align n[,value]
710 @item .skip n[,value]
711 @item .space n[,value]
712 @item .byte value1[,...]
713 @item .word value1[,...]
714 @item .short value1[,...]
715 @item .int value1[,...]
716 @item .long value1[,...]
717 @item .quad immediate_value1[,...]
720 @item .section section
724 @item .fill repeat[,size[,value]]
727 @item .string string[,...]
728 @item .asciz string[,...]
729 @item .ascii string[,...]
732 @section X86 Assembler
735 All X86 opcodes are supported. Only ATT syntax is supported (source
736 then destination operand order). If no size suffix is given, TinyCC
737 tries to guess it from the operand sizes.
739 Currently, MMX opcodes are supported but not SSE ones.
742 @chapter TinyCC Linker
745 @section ELF file generation
748 TCC can directly output relocatable ELF files (object files),
749 executable ELF files and dynamic ELF libraries without relying on an
752 Dynamic ELF libraries can be output but the C compiler does not generate
753 position independent code (PIC). It means that the dynamic library
754 code generated by TCC cannot be factorized among processes yet.
756 TCC linker eliminates unreferenced object code in libraries. A single pass is
757 done on the object and library list, so the order in which object files and
758 libraries are specified is important (same constraint as GNU ld). No grouping
759 options (@option{--start-group} and @option{--end-group}) are supported.
761 @section ELF file loader
763 TCC can load ELF object files, archives (.a files) and dynamic
766 @section PE-i386 file generation
769 TCC for Windows supports the native Win32 executable file format (PE-i386). It
770 generates EXE files (console and gui) and DLL files.
772 For usage on Windows, see also tcc-win32.txt.
774 @section GNU Linker Scripts
775 @cindex scripts, linker
776 @cindex linker scripts
777 @cindex GROUP, linker command
778 @cindex FILE, linker command
779 @cindex OUTPUT_FORMAT, linker command
780 @cindex TARGET, linker command
782 Because on many Linux systems some dynamic libraries (such as
783 @file{/usr/lib/libc.so}) are in fact GNU ld link scripts (horrible!),
784 the TCC linker also supports a subset of GNU ld scripts.
786 The @code{GROUP} and @code{FILE} commands are supported. @code{OUTPUT_FORMAT}
787 and @code{TARGET} are ignored.
789 Example from @file{/usr/lib/libc.so}:
792 Use the shared library, but some functions are only in
793 the static library, so try that secondarily. */
794 GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a )
798 @chapter TinyCC Memory and Bound checks
800 @cindex memory checks
802 This feature is activated with the @option{-b} (@pxref{Invoke}).
804 Note that pointer size is @emph{unchanged} and that code generated
805 with bound checks is @emph{fully compatible} with unchecked
806 code. When a pointer comes from unchecked code, it is assumed to be
807 valid. Even very obscure C code with casts should work correctly.
809 For more information about the ideas behind this method, see
810 @url{http://www.doc.ic.ac.uk/~phjk/BoundsChecking.html}.
812 Here are some examples of caught errors:
816 @item Invalid range with standard string function:
824 @item Out of bounds-error in global or local arrays:
834 @item Out of bounds-error in malloc'ed data:
838 tab = malloc(20 * sizeof(int));
846 @item Access of freed memory:
850 tab = malloc(20 * sizeof(int));
862 tab = malloc(20 * sizeof(int));
871 @chapter The @code{libtcc} library
873 The @code{libtcc} library enables you to use TCC as a backend for
874 dynamic code generation.
876 Read the @file{libtcc.h} to have an overview of the API. Read
877 @file{libtcc_test.c} to have a very simple example.
879 The idea consists in giving a C string containing the program you want
880 to compile directly to @code{libtcc}. Then you can access to any global
881 symbol (function or variable) defined.
884 @chapter Developer's guide
886 This chapter gives some hints to understand how TCC works. You can skip
887 it if you do not intend to modify the TCC code.
889 @section File reading
891 The @code{BufferedFile} structure contains the context needed to read a
892 file, including the current line number. @code{tcc_open()} opens a new
893 file and @code{tcc_close()} closes it. @code{inp()} returns the next
898 @code{next()} reads the next token in the current
899 file. @code{next_nomacro()} reads the next token without macro
902 @code{tok} contains the current token (see @code{TOK_xxx})
903 constants. Identifiers and keywords are also keywords. @code{tokc}
904 contains additional infos about the token (for example a constant value
905 if number or string token).
909 The parser is hardcoded (yacc is not necessary). It does only one pass,
914 @item For initialized arrays with unknown size, a first pass
915 is done to count the number of elements.
917 @item For architectures where arguments are evaluated in
918 reverse order, a first pass is done to reverse the argument order.
924 The types are stored in a single 'int' variable. It was choosen in the
925 first stages of development when tcc was much simpler. Now, it may not
926 be the best solution.
929 #define VT_INT 0 /* integer type */
930 #define VT_BYTE 1 /* signed byte type */
931 #define VT_SHORT 2 /* short type */
932 #define VT_VOID 3 /* void type */
933 #define VT_PTR 4 /* pointer */
934 #define VT_ENUM 5 /* enum definition */
935 #define VT_FUNC 6 /* function type */
936 #define VT_STRUCT 7 /* struct/union definition */
937 #define VT_FLOAT 8 /* IEEE float */
938 #define VT_DOUBLE 9 /* IEEE double */
939 #define VT_LDOUBLE 10 /* IEEE long double */
940 #define VT_BOOL 11 /* ISOC99 boolean type */
941 #define VT_LLONG 12 /* 64 bit integer */
942 #define VT_LONG 13 /* long integer (NEVER USED as type, only
944 #define VT_BTYPE 0x000f /* mask for basic type */
945 #define VT_UNSIGNED 0x0010 /* unsigned type */
946 #define VT_ARRAY 0x0020 /* array type (also has VT_PTR) */
947 #define VT_VLA 0x20000 /* VLA type (also has VT_PTR and VT_ARRAY) */
948 #define VT_BITFIELD 0x0040 /* bitfield modifier */
949 #define VT_CONSTANT 0x0800 /* const modifier */
950 #define VT_VOLATILE 0x1000 /* volatile modifier */
951 #define VT_SIGNED 0x2000 /* signed type */
953 #define VT_STRUCT_SHIFT 18 /* structure/enum name shift (14 bits left) */
956 When a reference to another type is needed (for pointers, functions and
957 structures), the @code{32 - VT_STRUCT_SHIFT} high order bits are used to
958 store an identifier reference.
960 The @code{VT_UNSIGNED} flag can be set for chars, shorts, ints and long
963 Arrays are considered as pointers @code{VT_PTR} with the flag
964 @code{VT_ARRAY} set. Variable length arrays are considered as special
965 arrays and have flag @code{VT_VLA} set instead of @code{VT_ARRAY}.
967 The @code{VT_BITFIELD} flag can be set for chars, shorts, ints and long
968 longs. If it is set, then the bitfield position is stored from bits
969 VT_STRUCT_SHIFT to VT_STRUCT_SHIFT + 5 and the bit field size is stored
970 from bits VT_STRUCT_SHIFT + 6 to VT_STRUCT_SHIFT + 11.
972 @code{VT_LONG} is never used except during parsing.
974 During parsing, the storage of an object is also stored in the type
978 #define VT_EXTERN 0x00000080 /* extern definition */
979 #define VT_STATIC 0x00000100 /* static variable */
980 #define VT_TYPEDEF 0x00000200 /* typedef definition */
981 #define VT_INLINE 0x00000400 /* inline definition */
982 #define VT_IMPORT 0x00004000 /* win32: extern data imported from dll */
983 #define VT_EXPORT 0x00008000 /* win32: data exported from dll */
984 #define VT_WEAK 0x00010000 /* win32: data exported from dll */
989 All symbols are stored in hashed symbol stacks. Each symbol stack
990 contains @code{Sym} structures.
992 @code{Sym.v} contains the symbol name (remember
993 an idenfier is also a token, so a string is never necessary to store
994 it). @code{Sym.t} gives the type of the symbol. @code{Sym.r} is usually
995 the register in which the corresponding variable is stored. @code{Sym.c} is
996 usually a constant associated to the symbol like its address for normal
997 symbols, and the number of entries for symbols representing arrays.
998 Variable length array types use @code{Sym.c} as a location on the stack
999 which holds the runtime sizeof for the type.
1001 Four main symbol stacks are defined:
1006 for the macros (@code{#define}s).
1009 for the global variables, functions and types.
1012 for the local variables, functions and types.
1014 @item global_label_stack
1015 for the local labels (for @code{goto}).
1018 for GCC block local labels (see the @code{__label__} keyword).
1022 @code{sym_push()} is used to add a new symbol in the local symbol
1023 stack. If no local symbol stack is active, it is added in the global
1026 @code{sym_pop(st,b)} pops symbols from the symbol stack @var{st} until
1027 the symbol @var{b} is on the top of stack. If @var{b} is NULL, the stack
1030 @code{sym_find(v)} return the symbol associated to the identifier
1031 @var{v}. The local stack is searched first from top to bottom, then the
1036 The generated code and datas are written in sections. The structure
1037 @code{Section} contains all the necessary information for a given
1038 section. @code{new_section()} creates a new section. ELF file semantics
1039 is assumed for each section.
1041 The following sections are predefined:
1046 is the section containing the generated code. @var{ind} contains the
1047 current position in the code section.
1050 contains initialized data
1053 contains uninitialized data
1055 @item bounds_section
1056 @itemx lbounds_section
1057 are used when bound checking is activated
1060 @itemx stabstr_section
1061 are used when debugging is actived to store debug information
1063 @item symtab_section
1064 @itemx strtab_section
1065 contain the exported symbols (currently only used for debugging).
1069 @section Code generation
1070 @cindex code generation
1072 @subsection Introduction
1074 The TCC code generator directly generates linked binary code in one
1075 pass. It is rather unusual these days (see gcc for example which
1076 generates text assembly), but it can be very fast and surprisingly
1079 The TCC code generator is register based. Optimization is only done at
1080 the expression level. No intermediate representation of expression is
1081 kept except the current values stored in the @emph{value stack}.
1083 On x86, three temporary registers are used. When more registers are
1084 needed, one register is spilled into a new temporary variable on the stack.
1086 @subsection The value stack
1087 @cindex value stack, introduction
1089 When an expression is parsed, its value is pushed on the value stack
1090 (@var{vstack}). The top of the value stack is @var{vtop}. Each value
1091 stack entry is the structure @code{SValue}.
1093 @code{SValue.t} is the type. @code{SValue.r} indicates how the value is
1094 currently stored in the generated code. It is usually a CPU register
1095 index (@code{REG_xxx} constants), but additional values and flags are
1099 #define VT_CONST 0x00f0
1100 #define VT_LLOCAL 0x00f1
1101 #define VT_LOCAL 0x00f2
1102 #define VT_CMP 0x00f3
1103 #define VT_JMP 0x00f4
1104 #define VT_JMPI 0x00f5
1105 #define VT_LVAL 0x0100
1106 #define VT_SYM 0x0200
1107 #define VT_MUSTCAST 0x0400
1108 #define VT_MUSTBOUND 0x0800
1109 #define VT_BOUNDED 0x8000
1110 #define VT_LVAL_BYTE 0x1000
1111 #define VT_LVAL_SHORT 0x2000
1112 #define VT_LVAL_UNSIGNED 0x4000
1113 #define VT_LVAL_TYPE (VT_LVAL_BYTE | VT_LVAL_SHORT | VT_LVAL_UNSIGNED)
1119 indicates that the value is a constant. It is stored in the union
1120 @code{SValue.c}, depending on its type.
1123 indicates a local variable pointer at offset @code{SValue.c.i} in the
1127 indicates that the value is actually stored in the CPU flags (i.e. the
1128 value is the consequence of a test). The value is either 0 or 1. The
1129 actual CPU flags used is indicated in @code{SValue.c.i}.
1131 If any code is generated which destroys the CPU flags, this value MUST be
1132 put in a normal register.
1136 indicates that the value is the consequence of a conditional jump. For VT_JMP,
1137 it is 1 if the jump is taken, 0 otherwise. For VT_JMPI it is inverted.
1139 These values are used to compile the @code{||} and @code{&&} logical
1142 If any code is generated, this value MUST be put in a normal
1143 register. Otherwise, the generated code won't be executed if the jump is
1147 is a flag indicating that the value is actually an lvalue (left value of
1148 an assignment). It means that the value stored is actually a pointer to
1151 Understanding the use @code{VT_LVAL} is very important if you want to
1152 understand how TCC works.
1155 @itemx VT_LVAL_SHORT
1156 @itemx VT_LVAL_UNSIGNED
1157 if the lvalue has an integer type, then these flags give its real
1158 type. The type alone is not enough in case of cast optimisations.
1161 is a saved lvalue on the stack. @code{VT_LLOCAL} should be eliminated
1162 ASAP because its semantics are rather complicated.
1165 indicates that a cast to the value type must be performed if the value
1166 is used (lazy casting).
1169 indicates that the symbol @code{SValue.sym} must be added to the constant.
1173 are only used for optional bound checking.
1177 @subsection Manipulating the value stack
1180 @code{vsetc()} and @code{vset()} pushes a new value on the value
1181 stack. If the previous @var{vtop} was stored in a very unsafe place(for
1182 example in the CPU flags), then some code is generated to put the
1183 previous @var{vtop} in a safe storage.
1185 @code{vpop()} pops @var{vtop}. In some cases, it also generates cleanup
1186 code (for example if stacked floating point registers are used as on
1189 The @code{gv(rc)} function generates code to evaluate @var{vtop} (the
1190 top value of the stack) into registers. @var{rc} selects in which
1191 register class the value should be put. @code{gv()} is the @emph{most
1192 important function} of the code generator.
1194 @code{gv2()} is the same as @code{gv()} but for the top two stack
1197 @subsection CPU dependent code generation
1198 @cindex CPU dependent
1199 See the @file{i386-gen.c} file to have an example.
1204 must generate the code needed to load a stack value into a register.
1207 must generate the code needed to store a register into a stack value
1211 @itemx gfunc_param()
1213 should generate a function call
1215 @item gfunc_prolog()
1216 @itemx gfunc_epilog()
1217 should generate a function prolog/epilog.
1220 must generate the binary integer operation @var{op} on the two top
1221 entries of the stack which are guaranted to contain integer types.
1223 The result value should be put on the stack.
1226 same as @code{gen_opi()} for floating point operations. The two top
1227 entries of the stack are guaranted to contain floating point values of
1230 @item gen_cvt_itof()
1231 integer to floating point conversion.
1233 @item gen_cvt_ftoi()
1234 floating point to integer conversion.
1236 @item gen_cvt_ftof()
1237 floating point to floating point of different size conversion.
1239 @item gen_bounded_ptr_add()
1240 @item gen_bounded_ptr_deref()
1241 are only used for bounds checking.
1245 @section Optimizations done
1246 @cindex optimizations
1247 @cindex constant propagation
1248 @cindex strength reduction
1249 @cindex comparison operators
1250 @cindex caching processor flags
1251 @cindex flags, caching
1252 @cindex jump optimization
1253 Constant propagation is done for all operations. Multiplications and
1254 divisions are optimized to shifts when appropriate. Comparison
1255 operators are optimized by maintaining a special cache for the
1256 processor flags. &&, || and ! are optimized by maintaining a special
1257 'jump target' value. No other jump optimization is currently performed
1258 because it would require to store the code in a more abstract fashion.
1260 @unnumbered Concept Index
1267 @c texinfo-column-for-description: 32