1 \input texinfo @c -*- texinfo -*-
3 @setfilename tcc-doc.info
4 @settitle Tiny C Compiler Reference Documentation
13 @center @titlefont{Tiny C Compiler Reference Documentation}
24 @node Top, Introduction, (dir), (dir)
25 @top Tiny C Compiler Reference Documentation
27 This manual documents version @value{VERSION} of the Tiny C Compiler.
30 * Introduction:: Introduction to tcc.
31 * Invoke:: Invocation of tcc (command line, options).
32 * Bounds:: Automatic bounds-checking of C code.
33 * Libtcc:: The libtcc library.
40 TinyCC (aka TCC) is a small but hyper fast C compiler. Unlike other C
41 compilers, it is meant to be self-relying: you do not need an
42 external assembler or linker because TCC does that for you.
44 TCC compiles so @emph{fast} that even for big projects @code{Makefile}s may
47 TCC not only supports ANSI C, but also most of the new ISO C99
48 standard and many GNUC extensions including inline assembly.
50 TCC can also be used to make @emph{C scripts}, i.e. pieces of C source
51 that you run as a Perl or Python script. Compilation is so fast that
52 your script will be as fast as if it was an executable.
54 TCC can also automatically generate memory and bound checks
55 (@pxref{Bounds}) while allowing all C pointers operations. TCC can do
56 these checks even if non patched libraries are used.
58 With @code{libtcc}, you can use TCC as a backend for dynamic code
59 generation (@pxref{Libtcc}).
62 @chapter Command line invocation
64 [This manual documents version @value{VERSION} of the Tiny C Compiler]
70 usage: tcc [options] [@var{infile1} @var{infile2}@dots{}] [@option{-run} @var{infile} @var{args}@dots{}]
75 @c man begin DESCRIPTION
76 TCC options are a very much like gcc options. The main difference is that TCC
77 can also execute directly the resulting program and give it runtime
80 Here are some examples to understand the logic:
83 @item @samp{tcc -run a.c}
84 Compile @file{a.c} and execute it directly
86 @item @samp{tcc -run a.c arg1}
87 Compile a.c and execute it directly. arg1 is given as first argument to
88 the @code{main()} of a.c.
90 @item @samp{tcc a.c -run b.c arg1}
91 Compile @file{a.c} and @file{b.c}, link them together and execute them. arg1 is given
92 as first argument to the @code{main()} of the resulting program. Because
93 multiple C files are specified, @option{--} are necessary to clearly separate the
94 program arguments from the TCC options.
96 @item @samp{tcc -o myprog a.c b.c}
97 Compile @file{a.c} and @file{b.c}, link them and generate the executable @file{myprog}.
99 @item @samp{tcc -o myprog a.o b.o}
100 link @file{a.o} and @file{b.o} together and generate the executable @file{myprog}.
102 @item @samp{tcc -c a.c}
103 Compile @file{a.c} and generate object file @file{a.o}.
105 @item @samp{tcc -c asmfile.S}
106 Preprocess with C preprocess and assemble @file{asmfile.S} and generate
107 object file @file{asmfile.o}.
109 @item @samp{tcc -c asmfile.s}
110 Assemble (but not preprocess) @file{asmfile.s} and generate object file
113 @item @samp{tcc -r -o ab.o a.c b.c}
114 Compile @file{a.c} and @file{b.c}, link them together and generate the object file @file{ab.o}.
120 TCC can be invoked from @emph{scripts}, just as shell scripts. You just
121 need to add @code{#!/usr/local/bin/tcc -run} at the start of your C source:
124 #!/usr/local/bin/tcc -run
129 printf("Hello World\n");
135 @section Option summary
142 Display current TCC version.
145 Generate an object file (@option{-o} option must also be given).
148 Put object file, executable, or dll into output file @file{outfile}.
151 Set the path where the tcc internal libraries can be found (default is
152 @file{PREFIX/lib/tcc}).
155 Output compilation statistics.
157 @item -run source [args...]
159 Compile file @var{source} and run it with the command line arguments
160 @var{args}. In order to be able to give more than one argument to a
161 script, several TCC options can be given @emph{after} the
162 @option{-run} option, separated by spaces. Example:
165 tcc "-run -L/usr/X11R6/lib -lX11" ex4.c
168 In a script, it gives the following header:
171 #!/usr/local/bin/tcc -run -L/usr/X11R6/lib -lX11
173 int main(int argc, char **argv)
181 Preprocessor options:
185 Specify an additional include path. Include paths are searched in the
186 order they are specified.
188 System include paths are always searched after. The default system
189 include paths are: @file{/usr/local/include}, @file{/usr/include}
190 and @file{PREFIX/lib/tcc/include}. (@file{PREFIX} is usually
191 @file{/usr} or @file{/usr/local}).
194 Define preprocessor symbol @samp{sym} to
195 val. If val is not present, its value is @samp{1}. Function-like macros can
196 also be defined: @option{-DF(a)=a+1}
199 Undefine preprocessor symbol @samp{sym}.
204 Note: each of the following warning options has a negative form beginning with
208 @item -funsigned-char
209 Let the @code{char} type be unsigned.
212 Let the @code{char} type be signed.
220 Disable all warnings.
224 Note: each of the following warning options has a negative form beginning with
228 @item -Wimplicit-function-declaration
229 Warn about implicit function declaration.
232 Warn about unsupported GCC features that are ignored by TCC.
234 @item -Wwrite-strings
235 Make string constants be of type @code{const char *} instead of @code{char
239 Abort compilation if warnings are issued.
242 Activate all warnings, except @option{-Werror}, @option{-Wunusupported} and
243 @option{-Wwrite-strings}.
251 Specify an additional static library path for the @option{-l} option. The
252 default library paths are @file{/usr/local/lib}, @file{/usr/lib} and @file{/lib}.
255 Link your program with dynamic library libxxx.so or static library
256 libxxx.a. The library is searched in the paths specified by the
260 Generate a shared library instead of an executable (@option{-o} option
264 Generate a statically linked executable (default is a shared linked
265 executable) (@option{-o} option must also be given).
268 Export global symbols to the dynamic linker. It is useful when a library
269 opened with @code{dlopen()} needs to access executable symbols.
272 Generate an object file combining all input files (@option{-o} option must
281 Generate run time debug information so that you get clear run time
282 error messages: @code{ test.c:68: in function 'test5()': dereferencing
283 invalid pointer} instead of the laconic @code{Segmentation
287 Generate additional support code to check
288 memory allocations and array/pointer bounds. @option{-g} is implied. Note
289 that the generated code is slower and bigger in this case.
292 Display N callers in stack traces. This is useful with @option{-g} or
297 Note: GCC options @option{-Ox}, @option{-fx} and @option{-mx} are
304 @settitle Tiny C Compiler
316 @chapter C language support
320 TCC implements all the ANSI C standard, including structure bit fields
321 and floating point numbers (@code{long double}, @code{double}, and
322 @code{float} fully supported).
324 @section ISOC99 extensions
326 TCC implements many features of the new C standard: ISO C99. Currently
327 missing items are: complex and imaginary numbers and variable length
330 Currently implemented ISOC99 features:
334 @item 64 bit @code{long long} types are fully supported.
336 @item The boolean type @code{_Bool} is supported.
338 @item @code{__func__} is a string variable containing the current
341 @item Variadic macros: @code{__VA_ARGS__} can be used for
342 function-like macros:
344 #define dprintf(level, __VA_ARGS__) printf(__VA_ARGS__)
348 @code{dprintf} can then be used with a variable number of parameters.
350 @item Declarations can appear anywhere in a block (as in C++).
352 @item Array and struct/union elements can be initialized in any order by
355 struct @{ int x, y; @} st[10] = @{ [0].x = 1, [0].y = 2 @};
357 int tab[10] = @{ 1, 2, [5] = 5, [9] = 9@};
360 @item Compound initializers are supported:
362 int *p = (int [])@{ 1, 2, 3 @};
364 to initialize a pointer pointing to an initialized array. The same
365 works for structures and strings.
367 @item Hexadecimal floating point constants are supported:
369 double d = 0x1234p10;
373 is the same as writing
375 double d = 4771840.0;
378 @item @code{inline} keyword is ignored.
380 @item @code{restrict} keyword is ignored.
383 @section GNU C extensions
385 TCC implements some GNU C extensions:
389 @item array designators can be used without '=':
391 int a[10] = @{ [0] 1, [5] 2, 3, 4 @};
394 @item Structure field designators can be a label:
396 struct @{ int x, y; @} st = @{ x: 1, y: 1@};
400 struct @{ int x, y; @} st = @{ .x = 1, .y = 1@};
403 @item @code{\e} is ASCII character 27.
405 @item case ranges : ranges can be used in @code{case}s:
409 printf("range 1 to 9\n");
412 printf("unexpected\n");
417 @item The keyword @code{__attribute__} is handled to specify variable or
418 function attributes. The following attributes are supported:
420 @item @code{aligned(n)}: align data to n bytes (must be a power of two).
422 @item @code{section(name)}: generate function or data in assembly
423 section name (name is a string containing the section name) instead
424 of the default section.
426 @item @code{unused}: specify that the variable or the function is unused.
428 @item @code{cdecl}: use standard C calling convention.
430 @item @code{stdcall}: use Pascal-like calling convention.
434 Here are some examples:
436 int a __attribute__ ((aligned(8), section(".mysection")));
440 align variable @code{a} to 8 bytes and put it in section @code{.mysection}.
443 int my_add(int a, int b) __attribute__ ((section(".mycodesection")))
450 generate function @code{my_add} in section @code{.mycodesection}.
452 @item GNU style variadic macros:
454 #define dprintf(fmt, args@dots{}) printf(fmt, ## args)
457 dprintf("one arg %d\n", 1);
460 @item @code{__FUNCTION__} is interpreted as C99 @code{__func__}
461 (so it has not exactly the same semantics as string literal GNUC
462 where it is a string literal).
464 @item The @code{__alignof__} keyword can be used as @code{sizeof}
465 to get the alignment of a type or an expression.
467 @item The @code{typeof(x)} returns the type of @code{x}.
468 @code{x} is an expression or a type.
470 @item Computed gotos: @code{&&label} returns a pointer of type
471 @code{void *} on the goto label @code{label}. @code{goto *expr} can be
472 used to jump on the pointer resulting from @code{expr}.
474 @item Inline assembly with asm instruction:
475 @cindex inline assembly
476 @cindex assembly, inline
479 static inline void * my_memcpy(void * to, const void * from, size_t n)
482 __asm__ __volatile__(
487 "1:\ttestb $1,%b4\n\t"
491 : "=&c" (d0), "=&D" (d1), "=&S" (d2)
492 :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
500 TCC includes its own x86 inline assembler with a @code{gas}-like (GNU
501 assembler) syntax. No intermediate files are generated. GCC 3.x named
502 operands are supported.
504 @item @code{__builtin_types_compatible_p()} and @code{__builtin_constant_p()}
509 @section TinyCC extensions
513 @item @code{__TINYC__} is a predefined macro to @code{1} to
514 indicate that you use TCC.
516 @item @code{#!} at the start of a line is ignored to allow scripting.
518 @item Binary digits can be entered (@code{0b101} instead of
521 @item @code{__BOUNDS_CHECKING_ON} is defined if bound checking is activated.
525 @chapter TinyCC Assembler
527 Since version 0.9.16, TinyCC integrates its own assembler. TinyCC
528 assembler supports a gas-like syntax (GNU assembler). You can
529 desactivate assembler support if you want a smaller TinyCC executable
530 (the C compiler does not rely on the assembler).
532 TinyCC Assembler is used to handle files with @file{.S} (C
533 preprocessed assembler) and @file{.s} extensions. It is also used to
534 handle the GNU inline assembler with the @code{asm} keyword.
538 TinyCC Assembler supports most of the gas syntax. The tokens are the
543 @item C and C++ comments are supported.
545 @item Identifiers are the same as C, so you cannot use '.' or '$'.
547 @item Only 32 bit integer numbers are supported.
555 @item Integers in decimal, octal and hexa are supported.
557 @item Unary operators: +, -, ~.
559 @item Binary operators in decreasing priority order:
567 @item A value is either an absolute number or a label plus an offset.
568 All operators accept absolute values except '+' and '-'. '+' or '-' can be
569 used to add an offset to a label. '-' supports two labels only if they
570 are the same or if they are both defined and in the same section.
578 @item All labels are considered as local, except undefined ones.
580 @item Numeric labels can be used as local @code{gas}-like labels.
581 They can be defined several times in the same source. Use 'b'
582 (backward) or 'f' (forward) as suffix to reference them:
586 jmp 1b /* jump to '1' label before */
587 jmp 1f /* jump to '1' label after */
594 @cindex assembler directives
595 @cindex directives, assembler
612 All directives are preceeded by a '.'. The following directives are
616 @item .align n[,value]
617 @item .skip n[,value]
618 @item .space n[,value]
619 @item .byte value1[,value2...]
620 @item .word value1[,value2...]
621 @item .short value1[,value2...]
622 @item .int value1[,value2...]
623 @item .long value1[,value2...]
627 @item .section section
633 @section X86 Assembler
636 All X86 opcodes are supported. Only ATT syntax is supported (source
637 then destination operand order). If no size suffix is given, TinyCC
638 tries to guess it from the operand sizes.
640 Currently, MMX opcodes are supported but not SSE ones.
642 @chapter TinyCC Linker
645 @section ELF file generation
648 TCC can directly output relocatable ELF files (object files),
649 executable ELF files and dynamic ELF libraries without relying on an
652 Dynamic ELF libraries can be output but the C compiler does not generate
653 position independent code (PIC). It means that the dynamic library
654 code generated by TCC cannot be factorized among processes yet.
656 TCC linker eliminates unreferenced object code in libraries. A single pass is
657 done on the object and library list, so the order in which object files and
658 libraries are specified is important (same constraint as GNU ld). No grouping
659 options (@option{--start-group} and @option{--end-group}) are supported.
661 @section ELF file loader
663 TCC can load ELF object files, archives (.a files) and dynamic
666 @section GNU Linker Scripts
667 @cindex scripts, linker
668 @cindex linker scripts
669 @cindex GROUP, linker command
670 @cindex FILE, linker command
671 @cindex OUTPUT_FORMAT, linker command
672 @cindex TARGET, linker command
674 Because on many Linux systems some dynamic libraries (such as
675 @file{/usr/lib/libc.so}) are in fact GNU ld link scripts (horrible!),
676 the TCC linker also supports a subset of GNU ld scripts.
678 The @code{GROUP} and @code{FILE} commands are supported. @code{OUTPUT_FORMAT}
679 and @code{TARGET} are ignored.
681 Example from @file{/usr/lib/libc.so}:
684 Use the shared library, but some functions are only in
685 the static library, so try that secondarily. */
686 GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a )
690 @chapter TinyCC Memory and Bound checks
692 @cindex memory checks
694 This feature is activated with the @option{-b} (@pxref{Invoke}).
696 Note that pointer size is @emph{unchanged} and that code generated
697 with bound checks is @emph{fully compatible} with unchecked
698 code. When a pointer comes from unchecked code, it is assumed to be
699 valid. Even very obscure C code with casts should work correctly.
701 For more information about the ideas behind this method, see
702 @url{http://www.doc.ic.ac.uk/~phjk/BoundsChecking.html}.
704 Here are some examples of caught errors:
708 @item Invalid range with standard string function:
716 @item Out of bounds-error in global or local arrays:
726 @item Out of bounds-error in malloc'ed data:
730 tab = malloc(20 * sizeof(int));
738 @item Access of freed memory:
742 tab = malloc(20 * sizeof(int));
754 tab = malloc(20 * sizeof(int));
763 @chapter The @code{libtcc} library
765 The @code{libtcc} library enables you to use TCC as a backend for
766 dynamic code generation.
768 Read the @file{libtcc.h} to have an overview of the API. Read
769 @file{libtcc_test.c} to have a very simple example.
771 The idea consists in giving a C string containing the program you want
772 to compile directly to @code{libtcc}. Then you can access to any global
773 symbol (function or variable) defined.
775 @chapter Developer's guide
777 This chapter gives some hints to understand how TCC works. You can skip
778 it if you do not intend to modify the TCC code.
780 @section File reading
782 The @code{BufferedFile} structure contains the context needed to read a
783 file, including the current line number. @code{tcc_open()} opens a new
784 file and @code{tcc_close()} closes it. @code{inp()} returns the next
789 @code{next()} reads the next token in the current
790 file. @code{next_nomacro()} reads the next token without macro
793 @code{tok} contains the current token (see @code{TOK_xxx})
794 constants. Identifiers and keywords are also keywords. @code{tokc}
795 contains additional infos about the token (for example a constant value
796 if number or string token).
800 The parser is hardcoded (yacc is not necessary). It does only one pass,
805 @item For initialized arrays with unknown size, a first pass
806 is done to count the number of elements.
808 @item For architectures where arguments are evaluated in
809 reverse order, a first pass is done to reverse the argument order.
815 The types are stored in a single 'int' variable. It was choosen in the
816 first stages of development when tcc was much simpler. Now, it may not
817 be the best solution.
820 #define VT_INT 0 /* integer type */
821 #define VT_BYTE 1 /* signed byte type */
822 #define VT_SHORT 2 /* short type */
823 #define VT_VOID 3 /* void type */
824 #define VT_PTR 4 /* pointer */
825 #define VT_ENUM 5 /* enum definition */
826 #define VT_FUNC 6 /* function type */
827 #define VT_STRUCT 7 /* struct/union definition */
828 #define VT_FLOAT 8 /* IEEE float */
829 #define VT_DOUBLE 9 /* IEEE double */
830 #define VT_LDOUBLE 10 /* IEEE long double */
831 #define VT_BOOL 11 /* ISOC99 boolean type */
832 #define VT_LLONG 12 /* 64 bit integer */
833 #define VT_LONG 13 /* long integer (NEVER USED as type, only
835 #define VT_BTYPE 0x000f /* mask for basic type */
836 #define VT_UNSIGNED 0x0010 /* unsigned type */
837 #define VT_ARRAY 0x0020 /* array type (also has VT_PTR) */
838 #define VT_BITFIELD 0x0040 /* bitfield modifier */
840 #define VT_STRUCT_SHIFT 16 /* structure/enum name shift (16 bits left) */
843 When a reference to another type is needed (for pointers, functions and
844 structures), the @code{32 - VT_STRUCT_SHIFT} high order bits are used to
845 store an identifier reference.
847 The @code{VT_UNSIGNED} flag can be set for chars, shorts, ints and long
850 Arrays are considered as pointers @code{VT_PTR} with the flag
853 The @code{VT_BITFIELD} flag can be set for chars, shorts, ints and long
854 longs. If it is set, then the bitfield position is stored from bits
855 VT_STRUCT_SHIFT to VT_STRUCT_SHIFT + 5 and the bit field size is stored
856 from bits VT_STRUCT_SHIFT + 6 to VT_STRUCT_SHIFT + 11.
858 @code{VT_LONG} is never used except during parsing.
860 During parsing, the storage of an object is also stored in the type
864 #define VT_EXTERN 0x00000080 /* extern definition */
865 #define VT_STATIC 0x00000100 /* static variable */
866 #define VT_TYPEDEF 0x00000200 /* typedef definition */
871 All symbols are stored in hashed symbol stacks. Each symbol stack
872 contains @code{Sym} structures.
874 @code{Sym.v} contains the symbol name (remember
875 an idenfier is also a token, so a string is never necessary to store
876 it). @code{Sym.t} gives the type of the symbol. @code{Sym.r} is usually
877 the register in which the corresponding variable is stored. @code{Sym.c} is
878 usually a constant associated to the symbol.
880 Four main symbol stacks are defined:
885 for the macros (@code{#define}s).
888 for the global variables, functions and types.
891 for the local variables, functions and types.
893 @item global_label_stack
894 for the local labels (for @code{goto}).
897 for GCC block local labels (see the @code{__label__} keyword).
901 @code{sym_push()} is used to add a new symbol in the local symbol
902 stack. If no local symbol stack is active, it is added in the global
905 @code{sym_pop(st,b)} pops symbols from the symbol stack @var{st} until
906 the symbol @var{b} is on the top of stack. If @var{b} is NULL, the stack
909 @code{sym_find(v)} return the symbol associated to the identifier
910 @var{v}. The local stack is searched first from top to bottom, then the
915 The generated code and datas are written in sections. The structure
916 @code{Section} contains all the necessary information for a given
917 section. @code{new_section()} creates a new section. ELF file semantics
918 is assumed for each section.
920 The following sections are predefined:
925 is the section containing the generated code. @var{ind} contains the
926 current position in the code section.
929 contains initialized data
932 contains uninitialized data
935 @itemx lbounds_section
936 are used when bound checking is activated
939 @itemx stabstr_section
940 are used when debugging is actived to store debug information
943 @itemx strtab_section
944 contain the exported symbols (currently only used for debugging).
948 @section Code generation
949 @cindex code generation
951 @subsection Introduction
953 The TCC code generator directly generates linked binary code in one
954 pass. It is rather unusual these days (see gcc for example which
955 generates text assembly), but it can be very fast and surprisingly
958 The TCC code generator is register based. Optimization is only done at
959 the expression level. No intermediate representation of expression is
960 kept except the current values stored in the @emph{value stack}.
962 On x86, three temporary registers are used. When more registers are
963 needed, one register is spilled into a new temporary variable on the stack.
965 @subsection The value stack
966 @cindex value stack, introduction
968 When an expression is parsed, its value is pushed on the value stack
969 (@var{vstack}). The top of the value stack is @var{vtop}. Each value
970 stack entry is the structure @code{SValue}.
972 @code{SValue.t} is the type. @code{SValue.r} indicates how the value is
973 currently stored in the generated code. It is usually a CPU register
974 index (@code{REG_xxx} constants), but additional values and flags are
978 #define VT_CONST 0x00f0
979 #define VT_LLOCAL 0x00f1
980 #define VT_LOCAL 0x00f2
981 #define VT_CMP 0x00f3
982 #define VT_JMP 0x00f4
983 #define VT_JMPI 0x00f5
984 #define VT_LVAL 0x0100
985 #define VT_SYM 0x0200
986 #define VT_MUSTCAST 0x0400
987 #define VT_MUSTBOUND 0x0800
988 #define VT_BOUNDED 0x8000
989 #define VT_LVAL_BYTE 0x1000
990 #define VT_LVAL_SHORT 0x2000
991 #define VT_LVAL_UNSIGNED 0x4000
992 #define VT_LVAL_TYPE (VT_LVAL_BYTE | VT_LVAL_SHORT | VT_LVAL_UNSIGNED)
998 indicates that the value is a constant. It is stored in the union
999 @code{SValue.c}, depending on its type.
1002 indicates a local variable pointer at offset @code{SValue.c.i} in the
1006 indicates that the value is actually stored in the CPU flags (i.e. the
1007 value is the consequence of a test). The value is either 0 or 1. The
1008 actual CPU flags used is indicated in @code{SValue.c.i}.
1010 If any code is generated which destroys the CPU flags, this value MUST be
1011 put in a normal register.
1015 indicates that the value is the consequence of a conditional jump. For VT_JMP,
1016 it is 1 if the jump is taken, 0 otherwise. For VT_JMPI it is inverted.
1018 These values are used to compile the @code{||} and @code{&&} logical
1021 If any code is generated, this value MUST be put in a normal
1022 register. Otherwise, the generated code won't be executed if the jump is
1026 is a flag indicating that the value is actually an lvalue (left value of
1027 an assignment). It means that the value stored is actually a pointer to
1030 Understanding the use @code{VT_LVAL} is very important if you want to
1031 understand how TCC works.
1034 @itemx VT_LVAL_SHORT
1035 @itemx VT_LVAL_UNSIGNED
1036 if the lvalue has an integer type, then these flags give its real
1037 type. The type alone is not enough in case of cast optimisations.
1040 is a saved lvalue on the stack. @code{VT_LLOCAL} should be eliminated
1041 ASAP because its semantics are rather complicated.
1044 indicates that a cast to the value type must be performed if the value
1045 is used (lazy casting).
1048 indicates that the symbol @code{SValue.sym} must be added to the constant.
1052 are only used for optional bound checking.
1056 @subsection Manipulating the value stack
1059 @code{vsetc()} and @code{vset()} pushes a new value on the value
1060 stack. If the previous @var{vtop} was stored in a very unsafe place(for
1061 example in the CPU flags), then some code is generated to put the
1062 previous @var{vtop} in a safe storage.
1064 @code{vpop()} pops @var{vtop}. In some cases, it also generates cleanup
1065 code (for example if stacked floating point registers are used as on
1068 The @code{gv(rc)} function generates code to evaluate @var{vtop} (the
1069 top value of the stack) into registers. @var{rc} selects in which
1070 register class the value should be put. @code{gv()} is the @emph{most
1071 important function} of the code generator.
1073 @code{gv2()} is the same as @code{gv()} but for the top two stack
1076 @subsection CPU dependent code generation
1077 @cindex CPU dependent
1078 See the @file{i386-gen.c} file to have an example.
1083 must generate the code needed to load a stack value into a register.
1086 must generate the code needed to store a register into a stack value
1090 @itemx gfunc_param()
1092 should generate a function call
1094 @item gfunc_prolog()
1095 @itemx gfunc_epilog()
1096 should generate a function prolog/epilog.
1099 must generate the binary integer operation @var{op} on the two top
1100 entries of the stack which are guaranted to contain integer types.
1102 The result value should be put on the stack.
1105 same as @code{gen_opi()} for floating point operations. The two top
1106 entries of the stack are guaranted to contain floating point values of
1109 @item gen_cvt_itof()
1110 integer to floating point conversion.
1112 @item gen_cvt_ftoi()
1113 floating point to integer conversion.
1115 @item gen_cvt_ftof()
1116 floating point to floating point of different size conversion.
1118 @item gen_bounded_ptr_add()
1119 @item gen_bounded_ptr_deref()
1120 are only used for bounds checking.
1124 @section Optimizations done
1125 @cindex optimizations
1126 @cindex constant propagation
1127 @cindex strength reduction
1128 @cindex comparison operators
1129 @cindex caching processor flags
1130 @cindex flags, caching
1131 @cindex jump optimization
1132 Constant propagation is done for all operations. Multiplications and
1133 divisions are optimized to shifts when appropriate. Comparison
1134 operators are optimized by maintaining a special cache for the
1135 processor flags. &&, || and ! are optimized by maintaining a special
1136 'jump target' value. No other jump optimization is currently performed
1137 because it would require to store the code in a more abstract fashion.
1139 @unnumbered Concept Index
1146 @c texinfo-column-for-description: 32