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.
161 Preprocessor options:
165 Specify an additional include path. Include paths are searched in the
166 order they are specified.
168 System include paths are always searched after. The default system
169 include paths are: @file{/usr/local/include}, @file{/usr/include}
170 and @file{PREFIX/lib/tcc/include}. (@file{PREFIX} is usually
171 @file{/usr} or @file{/usr/local}).
174 Define preprocessor symbol @samp{sym} to
175 val. If val is not present, its value is @samp{1}. Function-like macros can
176 also be defined: @option{-DF(a)=a+1}
179 Undefine preprocessor symbol @samp{sym}.
184 Note: each warning option has a negative form beginning with @option{-Wno-}.
188 Warn about unsupported GCC features that are ignored by TCC.
190 @item -Wwrite-strings
191 Make string constants being of type @code{const char *} intead of @code{char
195 Abort compilation if warnings are issued.
198 Activate all warnings, except @option{-Werror}, @option{-Wunusupported} and
199 @option{-Wwrite-strings} (currently not useful).
207 Specify an additional static library path for the @option{-l} option. The
208 default library paths are @file{/usr/local/lib}, @file{/usr/lib} and @file{/lib}.
211 Link your program with dynamic library libxxx.so or static library
212 libxxx.a. The library is searched in the paths specified by the
216 Generate a shared library instead of an executable (@option{-o} option
220 Generate a statically linked executable (default is a shared linked
221 executable) (@option{-o} option must also be given).
224 Export global symbols to the dynamic linker. It is useful when a library
225 opened with @code{dlopen()} needs to access executable symbols.
228 Generate an object file combining all input files (@option{-o} option must
237 Generate run time debug information so that you get clear run time
238 error messages: @code{ test.c:68: in function 'test5()': dereferencing
239 invalid pointer} instead of the laconic @code{Segmentation
243 Generate additional support code to check
244 memory allocations and array/pointer bounds. @option{-g} is implied. Note
245 that the generated code is slower and bigger in this case.
248 Display N callers in stack traces. This is useful with @option{-g} or
253 Note: GCC options @option{-Ox}, @option{-fx} and @option{-mx} are
260 @settitle Tiny C Compiler
272 @chapter C language support
276 TCC implements all the ANSI C standard, including structure bit fields
277 and floating point numbers (@code{long double}, @code{double}, and
278 @code{float} fully supported).
280 @section ISOC99 extensions
282 TCC implements many features of the new C standard: ISO C99. Currently
283 missing items are: complex and imaginary numbers and variable length
286 Currently implemented ISOC99 features:
290 @item 64 bit @code{long long} types are fully supported.
292 @item The boolean type @code{_Bool} is supported.
294 @item @code{__func__} is a string variable containing the current
297 @item Variadic macros: @code{__VA_ARGS__} can be used for
298 function-like macros:
300 #define dprintf(level, __VA_ARGS__) printf(__VA_ARGS__)
304 @code{dprintf} can then be used with a variable number of parameters.
306 @item Declarations can appear anywhere in a block (as in C++).
308 @item Array and struct/union elements can be initialized in any order by
311 struct @{ int x, y; @} st[10] = @{ [0].x = 1, [0].y = 2 @};
313 int tab[10] = @{ 1, 2, [5] = 5, [9] = 9@};
316 @item Compound initializers are supported:
318 int *p = (int [])@{ 1, 2, 3 @};
320 to initialize a pointer pointing to an initialized array. The same
321 works for structures and strings.
323 @item Hexadecimal floating point constants are supported:
325 double d = 0x1234p10;
329 is the same as writing
331 double d = 4771840.0;
334 @item @code{inline} keyword is ignored.
336 @item @code{restrict} keyword is ignored.
339 @section GNU C extensions
341 TCC implements some GNU C extensions:
345 @item array designators can be used without '=':
347 int a[10] = @{ [0] 1, [5] 2, 3, 4 @};
350 @item Structure field designators can be a label:
352 struct @{ int x, y; @} st = @{ x: 1, y: 1@};
356 struct @{ int x, y; @} st = @{ .x = 1, .y = 1@};
359 @item @code{\e} is ASCII character 27.
361 @item case ranges : ranges can be used in @code{case}s:
365 printf("range 1 to 9\n");
368 printf("unexpected\n");
373 @item The keyword @code{__attribute__} is handled to specify variable or
374 function attributes. The following attributes are supported:
376 @item @code{aligned(n)}: align data to n bytes (must be a power of two).
378 @item @code{section(name)}: generate function or data in assembly
379 section name (name is a string containing the section name) instead
380 of the default section.
382 @item @code{unused}: specify that the variable or the function is unused.
384 @item @code{cdecl}: use standard C calling convention.
386 @item @code{stdcall}: use Pascal-like calling convention.
390 Here are some examples:
392 int a __attribute__ ((aligned(8), section(".mysection")));
396 align variable @code{a} to 8 bytes and put it in section @code{.mysection}.
399 int my_add(int a, int b) __attribute__ ((section(".mycodesection")))
406 generate function @code{my_add} in section @code{.mycodesection}.
408 @item GNU style variadic macros:
410 #define dprintf(fmt, args@dots{}) printf(fmt, ## args)
413 dprintf("one arg %d\n", 1);
416 @item @code{__FUNCTION__} is interpreted as C99 @code{__func__}
417 (so it has not exactly the same semantics as string literal GNUC
418 where it is a string literal).
420 @item The @code{__alignof__} keyword can be used as @code{sizeof}
421 to get the alignment of a type or an expression.
423 @item The @code{typeof(x)} returns the type of @code{x}.
424 @code{x} is an expression or a type.
426 @item Computed gotos: @code{&&label} returns a pointer of type
427 @code{void *} on the goto label @code{label}. @code{goto *expr} can be
428 used to jump on the pointer resulting from @code{expr}.
430 @item Inline assembly with asm instruction:
431 @cindex inline assembly
432 @cindex assembly, inline
435 static inline void * my_memcpy(void * to, const void * from, size_t n)
438 __asm__ __volatile__(
443 "1:\ttestb $1,%b4\n\t"
447 : "=&c" (d0), "=&D" (d1), "=&S" (d2)
448 :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
456 TCC includes its own x86 inline assembler with a @code{gas}-like (GNU
457 assembler) syntax. No intermediate files are generated. GCC 3.x named
458 operands are supported.
460 @item @code{__builtin_types_compatible_p()} and @code{__builtin_constant_p()}
465 @section TinyCC extensions
469 @item @code{__TINYC__} is a predefined macro to @code{1} to
470 indicate that you use TCC.
472 @item @code{#!} at the start of a line is ignored to allow scripting.
474 @item Binary digits can be entered (@code{0b101} instead of
477 @item @code{__BOUNDS_CHECKING_ON} is defined if bound checking is activated.
481 @chapter TinyCC Assembler
483 Since version 0.9.16, TinyCC integrates its own assembler. TinyCC
484 assembler supports a gas-like syntax (GNU assembler). You can
485 desactivate assembler support if you want a smaller TinyCC executable
486 (the C compiler does not rely on the assembler).
488 TinyCC Assembler is used to handle files with @file{.S} (C
489 preprocessed assembler) and @file{.s} extensions. It is also used to
490 handle the GNU inline assembler with the @code{asm} keyword.
494 TinyCC Assembler supports most of the gas syntax. The tokens are the
499 @item C and C++ comments are supported.
501 @item Identifiers are the same as C, so you cannot use '.' or '$'.
503 @item Only 32 bit integer numbers are supported.
511 @item Integers in decimal, octal and hexa are supported.
513 @item Unary operators: +, -, ~.
515 @item Binary operators in decreasing priority order:
523 @item A value is either an absolute number or a label plus an offset.
524 All operators accept absolute values except '+' and '-'. '+' or '-' can be
525 used to add an offset to a label. '-' supports two labels only if they
526 are the same or if they are both defined and in the same section.
534 @item All labels are considered as local, except undefined ones.
536 @item Numeric labels can be used as local @code{gas}-like labels.
537 They can be defined several times in the same source. Use 'b'
538 (backward) or 'f' (forward) as suffix to reference them:
542 jmp 1b /* jump to '1' label before */
543 jmp 1f /* jump to '1' label after */
550 @cindex assembler directives
551 @cindex directives, assembler
567 All directives are preceeded by a '.'. The following directives are
571 @item .align n[,value]
572 @item .skip n[,value]
573 @item .space n[,value]
574 @item .byte value1[,value2...]
575 @item .word value1[,value2...]
576 @item .short value1[,value2...]
577 @item .int value1[,value2...]
578 @item .long value1[,value2...]
581 @item .section section
587 @section X86 Assembler
590 All X86 opcodes are supported. Only ATT syntax is supported (source
591 then destination operand order). If no size suffix is given, TinyCC
592 tries to guess it from the operand sizes.
594 Currently, MMX opcodes are supported but not SSE ones.
596 @chapter TinyCC Linker
599 @section ELF file generation
602 TCC can directly output relocatable ELF files (object files),
603 executable ELF files and dynamic ELF libraries without relying on an
606 Dynamic ELF libraries can be output but the C compiler does not generate
607 position independent code (PIC). It means that the dynamic librairy
608 code generated by TCC cannot be factorized among processes yet.
610 TCC linker eliminates unreferenced object code in libraries. A single pass is
611 done on the object and library list, so the order in which object files and
612 libraries are specified is important (same constraint as GNU ld). No grouping
613 options (@option{--start-group} and @option{--end-group}) are supported.
615 @section ELF file loader
617 TCC can load ELF object files, archives (.a files) and dynamic
620 @section GNU Linker Scripts
621 @cindex scripts, linker
622 @cindex linker scripts
623 @cindex GROUP, linker command
624 @cindex FILE, linker command
625 @cindex OUTPUT_FORMAT, linker command
626 @cindex TARGET, linker command
628 Because on many Linux systems some dynamic libraries (such as
629 @file{/usr/lib/libc.so}) are in fact GNU ld link scripts (horrible!),
630 the TCC linker also supports a subset of GNU ld scripts.
632 The @code{GROUP} and @code{FILE} commands are supported. @code{OUTPUT_FORMAT}
633 and @code{TARGET} are ignored.
635 Example from @file{/usr/lib/libc.so}:
638 Use the shared library, but some functions are only in
639 the static library, so try that secondarily. */
640 GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a )
644 @chapter TinyCC Memory and Bound checks
646 @cindex memory checks
648 This feature is activated with the @option{-b} (@pxref{Invoke}).
650 Note that pointer size is @emph{unchanged} and that code generated
651 with bound checks is @emph{fully compatible} with unchecked
652 code. When a pointer comes from unchecked code, it is assumed to be
653 valid. Even very obscure C code with casts should work correctly.
655 For more information about the ideas behind this method, see
656 @url{http://www.doc.ic.ac.uk/~phjk/BoundsChecking.html}.
658 Here are some examples of caught errors:
662 @item Invalid range with standard string function:
670 @item Out of bounds-error in global or local arrays:
680 @item Out of bounds-error in malloc'ed data:
684 tab = malloc(20 * sizeof(int));
692 @item Access of freed memory:
696 tab = malloc(20 * sizeof(int));
708 tab = malloc(20 * sizeof(int));
717 @chapter The @code{libtcc} library
719 The @code{libtcc} library enables you to use TCC as a backend for
720 dynamic code generation.
722 Read the @file{libtcc.h} to have an overview of the API. Read
723 @file{libtcc_test.c} to have a very simple example.
725 The idea consists in giving a C string containing the program you want
726 to compile directly to @code{libtcc}. Then you can access to any global
727 symbol (function or variable) defined.
729 @chapter Developer's guide
731 This chapter gives some hints to understand how TCC works. You can skip
732 it if you do not intend to modify the TCC code.
734 @section File reading
736 The @code{BufferedFile} structure contains the context needed to read a
737 file, including the current line number. @code{tcc_open()} opens a new
738 file and @code{tcc_close()} closes it. @code{inp()} returns the next
743 @code{next()} reads the next token in the current
744 file. @code{next_nomacro()} reads the next token without macro
747 @code{tok} contains the current token (see @code{TOK_xxx})
748 constants. Identifiers and keywords are also keywords. @code{tokc}
749 contains additional infos about the token (for example a constant value
750 if number or string token).
754 The parser is hardcoded (yacc is not necessary). It does only one pass,
759 @item For initialized arrays with unknown size, a first pass
760 is done to count the number of elements.
762 @item For architectures where arguments are evaluated in
763 reverse order, a first pass is done to reverse the argument order.
769 The types are stored in a single 'int' variable. It was choosen in the
770 first stages of development when tcc was much simpler. Now, it may not
771 be the best solution.
774 #define VT_INT 0 /* integer type */
775 #define VT_BYTE 1 /* signed byte type */
776 #define VT_SHORT 2 /* short type */
777 #define VT_VOID 3 /* void type */
778 #define VT_PTR 4 /* pointer */
779 #define VT_ENUM 5 /* enum definition */
780 #define VT_FUNC 6 /* function type */
781 #define VT_STRUCT 7 /* struct/union definition */
782 #define VT_FLOAT 8 /* IEEE float */
783 #define VT_DOUBLE 9 /* IEEE double */
784 #define VT_LDOUBLE 10 /* IEEE long double */
785 #define VT_BOOL 11 /* ISOC99 boolean type */
786 #define VT_LLONG 12 /* 64 bit integer */
787 #define VT_LONG 13 /* long integer (NEVER USED as type, only
789 #define VT_BTYPE 0x000f /* mask for basic type */
790 #define VT_UNSIGNED 0x0010 /* unsigned type */
791 #define VT_ARRAY 0x0020 /* array type (also has VT_PTR) */
792 #define VT_BITFIELD 0x0040 /* bitfield modifier */
794 #define VT_STRUCT_SHIFT 16 /* structure/enum name shift (16 bits left) */
797 When a reference to another type is needed (for pointers, functions and
798 structures), the @code{32 - VT_STRUCT_SHIFT} high order bits are used to
799 store an identifier reference.
801 The @code{VT_UNSIGNED} flag can be set for chars, shorts, ints and long
804 Arrays are considered as pointers @code{VT_PTR} with the flag
807 The @code{VT_BITFIELD} flag can be set for chars, shorts, ints and long
808 longs. If it is set, then the bitfield position is stored from bits
809 VT_STRUCT_SHIFT to VT_STRUCT_SHIFT + 5 and the bit field size is stored
810 from bits VT_STRUCT_SHIFT + 6 to VT_STRUCT_SHIFT + 11.
812 @code{VT_LONG} is never used except during parsing.
814 During parsing, the storage of an object is also stored in the type
818 #define VT_EXTERN 0x00000080 /* extern definition */
819 #define VT_STATIC 0x00000100 /* static variable */
820 #define VT_TYPEDEF 0x00000200 /* typedef definition */
825 All symbols are stored in hashed symbol stacks. Each symbol stack
826 contains @code{Sym} structures.
828 @code{Sym.v} contains the symbol name (remember
829 an idenfier is also a token, so a string is never necessary to store
830 it). @code{Sym.t} gives the type of the symbol. @code{Sym.r} is usually
831 the register in which the corresponding variable is stored. @code{Sym.c} is
832 usually a constant associated to the symbol.
834 Four main symbol stacks are defined:
839 for the macros (@code{#define}s).
842 for the global variables, functions and types.
845 for the local variables, functions and types.
847 @item global_label_stack
848 for the local labels (for @code{goto}).
851 for GCC block local labels (see the @code{__label__} keyword).
855 @code{sym_push()} is used to add a new symbol in the local symbol
856 stack. If no local symbol stack is active, it is added in the global
859 @code{sym_pop(st,b)} pops symbols from the symbol stack @var{st} until
860 the symbol @var{b} is on the top of stack. If @var{b} is NULL, the stack
863 @code{sym_find(v)} return the symbol associated to the identifier
864 @var{v}. The local stack is searched first from top to bottom, then the
869 The generated code and datas are written in sections. The structure
870 @code{Section} contains all the necessary information for a given
871 section. @code{new_section()} creates a new section. ELF file semantics
872 is assumed for each section.
874 The following sections are predefined:
879 is the section containing the generated code. @var{ind} contains the
880 current position in the code section.
883 contains initialized data
886 contains uninitialized data
889 @itemx lbounds_section
890 are used when bound checking is activated
893 @itemx stabstr_section
894 are used when debugging is actived to store debug information
897 @itemx strtab_section
898 contain the exported symbols (currently only used for debugging).
902 @section Code generation
903 @cindex code generation
905 @subsection Introduction
907 The TCC code generator directly generates linked binary code in one
908 pass. It is rather unusual these days (see gcc for example which
909 generates text assembly), but it can be very fast and surprisingly
912 The TCC code generator is register based. Optimization is only done at
913 the expression level. No intermediate representation of expression is
914 kept except the current values stored in the @emph{value stack}.
916 On x86, three temporary registers are used. When more registers are
917 needed, one register is spilled into a new temporary variable on the stack.
919 @subsection The value stack
920 @cindex value stack, introduction
922 When an expression is parsed, its value is pushed on the value stack
923 (@var{vstack}). The top of the value stack is @var{vtop}. Each value
924 stack entry is the structure @code{SValue}.
926 @code{SValue.t} is the type. @code{SValue.r} indicates how the value is
927 currently stored in the generated code. It is usually a CPU register
928 index (@code{REG_xxx} constants), but additional values and flags are
932 #define VT_CONST 0x00f0
933 #define VT_LLOCAL 0x00f1
934 #define VT_LOCAL 0x00f2
935 #define VT_CMP 0x00f3
936 #define VT_JMP 0x00f4
937 #define VT_JMPI 0x00f5
938 #define VT_LVAL 0x0100
939 #define VT_SYM 0x0200
940 #define VT_MUSTCAST 0x0400
941 #define VT_MUSTBOUND 0x0800
942 #define VT_BOUNDED 0x8000
943 #define VT_LVAL_BYTE 0x1000
944 #define VT_LVAL_SHORT 0x2000
945 #define VT_LVAL_UNSIGNED 0x4000
946 #define VT_LVAL_TYPE (VT_LVAL_BYTE | VT_LVAL_SHORT | VT_LVAL_UNSIGNED)
952 indicates that the value is a constant. It is stored in the union
953 @code{SValue.c}, depending on its type.
956 indicates a local variable pointer at offset @code{SValue.c.i} in the
960 indicates that the value is actually stored in the CPU flags (i.e. the
961 value is the consequence of a test). The value is either 0 or 1. The
962 actual CPU flags used is indicated in @code{SValue.c.i}.
964 If any code is generated which destroys the CPU flags, this value MUST be
965 put in a normal register.
969 indicates that the value is the consequence of a conditional jump. For VT_JMP,
970 it is 1 if the jump is taken, 0 otherwise. For VT_JMPI it is inverted.
972 These values are used to compile the @code{||} and @code{&&} logical
975 If any code is generated, this value MUST be put in a normal
976 register. Otherwise, the generated code won't be executed if the jump is
980 is a flag indicating that the value is actually an lvalue (left value of
981 an assignment). It means that the value stored is actually a pointer to
984 Understanding the use @code{VT_LVAL} is very important if you want to
985 understand how TCC works.
989 @itemx VT_LVAL_UNSIGNED
990 if the lvalue has an integer type, then these flags give its real
991 type. The type alone is not enough in case of cast optimisations.
994 is a saved lvalue on the stack. @code{VT_LLOCAL} should be eliminated
995 ASAP because its semantics are rather complicated.
998 indicates that a cast to the value type must be performed if the value
999 is used (lazy casting).
1002 indicates that the symbol @code{SValue.sym} must be added to the constant.
1006 are only used for optional bound checking.
1010 @subsection Manipulating the value stack
1013 @code{vsetc()} and @code{vset()} pushes a new value on the value
1014 stack. If the previous @var{vtop} was stored in a very unsafe place(for
1015 example in the CPU flags), then some code is generated to put the
1016 previous @var{vtop} in a safe storage.
1018 @code{vpop()} pops @var{vtop}. In some cases, it also generates cleanup
1019 code (for example if stacked floating point registers are used as on
1022 The @code{gv(rc)} function generates code to evaluate @var{vtop} (the
1023 top value of the stack) into registers. @var{rc} selects in which
1024 register class the value should be put. @code{gv()} is the @emph{most
1025 important function} of the code generator.
1027 @code{gv2()} is the same as @code{gv()} but for the top two stack
1030 @subsection CPU dependent code generation
1031 @cindex CPU dependent
1032 See the @file{i386-gen.c} file to have an example.
1037 must generate the code needed to load a stack value into a register.
1040 must generate the code needed to store a register into a stack value
1044 @itemx gfunc_param()
1046 should generate a function call
1048 @item gfunc_prolog()
1049 @itemx gfunc_epilog()
1050 should generate a function prolog/epilog.
1053 must generate the binary integer operation @var{op} on the two top
1054 entries of the stack which are guaranted to contain integer types.
1056 The result value should be put on the stack.
1059 same as @code{gen_opi()} for floating point operations. The two top
1060 entries of the stack are guaranted to contain floating point values of
1063 @item gen_cvt_itof()
1064 integer to floating point conversion.
1066 @item gen_cvt_ftoi()
1067 floating point to integer conversion.
1069 @item gen_cvt_ftof()
1070 floating point to floating point of different size conversion.
1072 @item gen_bounded_ptr_add()
1073 @item gen_bounded_ptr_deref()
1074 are only used for bounds checking.
1078 @section Optimizations done
1079 @cindex optimizations
1080 @cindex constant propagation
1081 @cindex strength reduction
1082 @cindex comparison operators
1083 @cindex caching processor flags
1084 @cindex flags, caching
1085 @cindex jump optimization
1086 Constant propagation is done for all operations. Multiplications and
1087 divisions are optimized to shifts when appropriate. Comparison
1088 operators are optimized by maintaining a special cache for the
1089 processor flags. &&, || and ! are optimized by maintaining a special
1090 'jump target' value. No other jump optimization is currently performed
1091 because it would require to store the code in a more abstract fashion.
1093 @unnumbered Concept Index
1100 @c texinfo-column-for-description: 32