tinycc/tcc-doc.texi

\input texinfo @c -*- texinfo -*-

@settitle Tiny C Compiler Reference Documentation
@titlepage
@sp 7
@center @titlefont{Tiny C Compiler Reference Documentation}
@sp 3
@end titlepage

@chapter Introduction

TinyCC (aka TCC) is a small but hyper fast C compiler. Unlike other C
compilers, it is meant to be self-suffisant: you do not need an
external assembler or linker because TCC does that for you.

TCC compiles so @emph{fast} that even for big projects @code{Makefile}s may
not be necessary. 

TCC not only supports ANSI C, but also most of the new ISO C99
standard and many GNUC extensions.

TCC can also be used to make @emph{C scripts}, i.e. pieces of C source
that you run as a Perl or Python script. Compilation is so fast that
your script will be as fast as if it was an executable.

TCC can also automatically generate memory and bound checks
(@xref{bounds}) while allowing all C pointers operations. TCC can do
these checks even if non patched libraries are used.

With @code{libtcc}, you can use TCC as a backend for dynamic code
generation (@xref{libtcc}).

@node invoke
@chapter Command line invocation

@section Quick start

@example
usage: tcc [-c] [-o outfile] [-Bdir] [-bench] [-Idir] [-Dsym[=val]] [-Usym]
           [-g] [-b] [-Ldir] [-llib] [-shared] [-static]
           [--] infile1 [infile2... --] [infile_args...]
@end example

TCC options are a very much like gcc. The main difference is that TCC
can also execute directly the resulting program and give it runtime
arguments.

Here are some examples to understand the logic:

@table @code
@item tcc a.c
Compile a.c and execute it directly

@item tcc a.c arg1
Compile a.c and execute it directly. arg1 is given as first argument to
the @code{main()} of a.c.

@item tcc -- a.c b.c -- arg1
Compile a.c and b.c, link them together and execute them. arg1 is given
as first argument to the @code{main()} of the resulting program. Because
multiple C files are specified, @code{--} are necessary to clearly separate the
program arguments from the TCC options.

@item tcc -o myprog a.c b.c
Compile a.c and b.c, link them and generate the executable myprog.

@item tcc -o myprog a.o b.o
link a.o and b.o together and generate the executable myprog.

@item tcc -c a.c
Compile a.c and generate object file a.o

@item tcc -r -o ab.o a.c b.c
Compile a.c and b.c, link them together and generate the object file ab.o.

@end table

Scripting:

TCC can be invoked from @emph{scripts}, just as shell scripts. You just
need to add @code{#!/usr/local/bin/tcc} at the start of your C source:

@example
#!/usr/local/bin/tcc
#include <stdio.h>

int main() 
{
    printf("Hello World\n");
    return 0;
}
@end example

@section Option summary

General Options:

@table @samp
@item -c
Generate an object file (@samp{-o} option must also be given).

@item -o outfile
Put object file, executable, or dll into output file @file{outfile}.

@item -Bdir
Set the path where the tcc internal libraries can be found (default is
@file{PREFIX/lib/tcc}).

@item -bench
Output compilation statistics.
@end table

Preprocessor options:

@table @samp
@item -Idir
Specify an additionnal include path. Include paths are searched in the
order they are specified.

System include paths are always searched after. The default system
include paths are: @file{/usr/local/include}, @file{/usr/include}
and @file{PREFIX/lib/tcc/include}. (@code{PREFIX} is usually
@file{/usr} or @file{/usr/local}).

@item -Dsym[=val]
Define preprocessor symbol 'sym' to
val. If val is not present, its value is '1'. Function-like macros can
also be defined: @code{'-DF(a)=a+1'}

@item -Usym
Undefine preprocessor symbol 'sym'.
@end table

C compiler options:

@table @samp
@item -g
Generate run time debug information so that you get clear run time
error messages: @code{ test.c:68: in function 'test5()': dereferencing
invalid pointer} instead of the laconic @code{Segmentation
fault}.

@item -b
Generate additionnal support code to check
memory allocations and array/pointer bounds. '-g' is implied. Note
that the generated code is slower and bigger in this case.
@end table

Linker options:

@table @samp
@item -Ldir
Specify an additionnal static library path for the @samp{-l} option. The
default library paths are @file{/usr/local/lib}, @file{/usr/lib} and @file{/lib}.

@item -lxxx
Link your program with dynamic library libxxx.so or static library
libxxx.a. The library is searched in the paths specified by the
@samp{-L} option.

@item -shared
Generate a shared library instead of an executable (@samp{-o} option
must also be given).

@item -static
Generate a statically linked executable (default is a shared linked
executable) (@samp{-o} option must also be given).

@item -r
Generate an object file combining all input files (@samp{-o} option must
also be given).

@end table

@chapter C language support

@section ANSI C

TCC implements all the ANSI C standard, including structure bit fields
and floating point numbers (@code{long double}, @code{double}, and
@code{float} fully supported). The following limitations are known:

@itemize
 @item The preprocessor tokens are the same as C. It means that in some
  rare cases, preprocessed numbers are not handled exactly as in ANSI
  C. This approach has the advantage of being simpler and FAST!
@end itemize

@section ISOC99 extensions

TCC implements many features of the new C standard: ISO C99. Currently
missing items are: complex and imaginary numbers and variable length
arrays.

Currently implemented ISOC99 features:

@itemize

@item 64 bit @code{'long long'} types are fully supported.

@item The boolean type @code{'_Bool'} is supported.

@item @code{'__func__'} is a string variable containing the current
function name.

@item Variadic macros: @code{__VA_ARGS__} can be used for
   function-like macros:
@example
    #define dprintf(level, __VA_ARGS__) printf(__VA_ARGS__)
@end example
@code{dprintf} can then be used with a variable number of parameters.

@item Declarations can appear anywhere in a block (as in C++).

@item Array and struct/union elements can be initialized in any order by
  using designators:
@example
    struct { int x, y; } st[10] = { [0].x = 1, [0].y = 2 };

    int tab[10] = { 1, 2, [5] = 5, [9] = 9};
@end example
    
@item Compound initializers are supported:
@example
    int *p = (int []){ 1, 2, 3 };
@end example
to initialize a pointer pointing to an initialized array. The same
works for structures and strings.

@item Hexadecimal floating point constants are supported:
@example
          double d = 0x1234p10;
@end example
is the same as writing 
@example
          double d = 4771840.0;
@end example

@item @code{'inline'} keyword is ignored.

@item @code{'restrict'} keyword is ignored.
@end itemize

@section GNU C extensions

TCC implements some GNU C extensions:

@itemize

@item array designators can be used without '=': 
@example
    int a[10] = { [0] 1, [5] 2, 3, 4 };
@end example

@item Structure field designators can be a label: 
@example
    struct { int x, y; } st = { x: 1, y: 1};
@end example
instead of
@example
    struct { int x, y; } st = { .x = 1, .y = 1};
@end example

@item @code{'\e'} is ASCII character 27.

@item case ranges : ranges can be used in @code{case}s:
@example
    switch(a) {
    case 1 ... 9:
          printf("range 1 to 9\n");
          break;
    default:
          printf("unexpected\n");
          break;
    }
@end example

@item The keyword @code{__attribute__} is handled to specify variable or
function attributes. The following attributes are supported:
  @itemize
  @item @code{aligned(n)}: align data to n bytes (must be a power of two).

  @item @code{section(name)}: generate function or data in assembly
  section name (name is a string containing the section name) instead
  of the default section.

  @item @code{unused}: specify that the variable or the function is unused.

  @item @code{cdecl}: use standard C calling convention.

  @item @code{stdcall}: use Pascal-like calling convention.

  @end itemize

Here are some examples:
@example
    int a __attribute__ ((aligned(8), section(".mysection")));
@end example

align variable @code{'a'} to 8 bytes and put it in section @code{.mysection}.

@example
    int my_add(int a, int b) __attribute__ ((section(".mycodesection"))) 
    {
        return a + b;
    }
@end example

generate function @code{'my_add'} in section @code{.mycodesection}.

@item GNU style variadic macros:
@example
    #define dprintf(fmt, args...) printf(fmt, ## args)

    dprintf("no arg\n");
    dprintf("one arg %d\n", 1);
@end example

@item @code{__FUNCTION__} is interpreted as C99 @code{__func__} 
(so it has not exactly the same semantics as string literal GNUC
where it is a string literal).

@end itemize

@section TinyCC extensions

@itemize

@item @code{__TINYC__} is a predefined macro to @code{'1'} to
indicate that you use TCC.

@item @code{'#!'} at the start of a line is ignored to allow scripting.

@item Binary digits can be entered (@code{'0b101'} instead of
@code{'5'}).

@item @code{__BOUNDS_CHECKING_ON} is defined if bound checking is activated.

@end itemize

@chapter TinyCC Linker

@section ELF file generation

TCC can directly output relocatable ELF files (object files),
executable ELF files and dynamic ELF libraries without relying on an
external linker.

Dynamic ELF libraries can be output but the C compiler does not generate
position independant code (PIC) code. It means that the dynamic librairy
code generated by TCC cannot be factorized among processes yet.

TCC linker cannot currently suppress unused object code. But TCC
will soon integrate a novel feature not found in GNU tools: unused code
will be suppressed at the function or variable level, provided you only
use TCC to compile your files.

@section ELF file loader

TCC can load ELF object files, archives (.a files) and dynamic
libraries (.so).

@section GNU Linker Scripts

Because on many Linux systems some dynamic libraries (such as
@file{/usr/lib/libc.so}) are in fact GNU ld link scripts (horrible!),
TCC linker also support a subset of GNU ld scripts.

The @code{GROUP} and @code{FILE} commands are supported.

Example from @file{/usr/lib/libc.so}:
@example
/* GNU ld script
   Use the shared library, but some functions are only in
   the static library, so try that secondarily.  */
GROUP ( /lib/libc.so.6 /usr/lib/libc_nonshared.a )
@end example

@node bounds
@chapter TinyCC Memory and Bound checks

This feature is activated with the @code{'-b'} (@xref{invoke}).

Note that pointer size is @emph{unchanged} and that code generated
with bound checks is @emph{fully compatible} with unchecked
code. When a pointer comes from unchecked code, it is assumed to be
valid. Even very obscure C code with casts should work correctly.

To have more information about the ideas behind this method, check at 
@url{http://www.doc.ic.ac.uk/~phjk/BoundsChecking.html}.

Here are some examples of catched errors:

@table @asis

@item Invalid range with standard string function:
@example
{
    char tab[10];
    memset(tab, 0, 11);
}
@end example

@item Bound error in global or local arrays:
@example
{
    int tab[10];
    for(i=0;i<11;i++) {
        sum += tab[i];
    }
}
@end example

@item Bound error in allocated data:
@example
{
    int *tab;
    tab = malloc(20 * sizeof(int));
    for(i=0;i<21;i++) {
        sum += tab4[i];
    }
    free(tab);
}
@end example

@item Access to a freed region:
@example
{
    int *tab;
    tab = malloc(20 * sizeof(int));
    free(tab);
    for(i=0;i<20;i++) {
        sum += tab4[i];
    }
}
@end example

@item Freeing an already freed region:
@example
{
    int *tab;
    tab = malloc(20 * sizeof(int));
    free(tab);
    free(tab);
}
@end example

@end table

@node libtcc
@chapter The @code{libtcc} library

The @code{libtcc} library enables you to use TCC as a backend for
dynamic code generation. 

Read the @file{libtcc.h} to have an overview of the API. Read
@file{libtcc_test.c} to have a very simple example.

The idea consists in giving a C string containing the program you want
to compile directly to @code{libtcc}. Then the @code{main()} function of
the compiled string can be launched.

@chapter Developper's guide

This chapter gives some hints to understand how TCC works. You can skip
it if you do not intend to modify the TCC code.

@section File reading

The @code{BufferedFile} structure contains the context needed to read a
file, including the current line number. @code{tcc_open()} opens a new
file and @code{tcc_close()} closes it. @code{inp()} returns the next
character.

@section Lexer

@code{next()} reads the next token in the current
file. @code{next_nomacro()} reads the next token without macro
expansion.

@code{tok} contains the current token (see @code{TOK_xxx})
constants. Identifiers and keywords are also keywords. @code{tokc}
contains additionnal infos about the token (for example a constant value
if number or string token).

@section Parser

The parser is hardcoded (yacc is not necessary). It does only one pass,
except:

@itemize

@item For initialized arrays with unknown size, a first pass 
is done to count the number of elements.

@item For architectures where arguments are evaluated in 
reverse order, a first pass is done to reverse the argument order.

@end itemize

@section Types

The types are stored in a single 'int' variable. It was choosen in the
first stages of development when tcc was much simpler. Now, it may not
be the best solution.

@example
#define VT_INT        0  /* integer type */
#define VT_BYTE       1  /* signed byte type */
#define VT_SHORT      2  /* short type */
#define VT_VOID       3  /* void type */
#define VT_PTR        4  /* pointer */
#define VT_ENUM       5  /* enum definition */
#define VT_FUNC       6  /* function type */
#define VT_STRUCT     7  /* struct/union definition */
#define VT_FLOAT      8  /* IEEE float */
#define VT_DOUBLE     9  /* IEEE double */
#define VT_LDOUBLE   10  /* IEEE long double */
#define VT_BOOL      11  /* ISOC99 boolean type */
#define VT_LLONG     12  /* 64 bit integer */
#define VT_LONG      13  /* long integer (NEVER USED as type, only
                            during parsing) */
#define VT_BTYPE      0x000f /* mask for basic type */
#define VT_UNSIGNED   0x0010  /* unsigned type */
#define VT_ARRAY      0x0020  /* array type (also has VT_PTR) */
#define VT_BITFIELD   0x0040  /* bitfield modifier */

#define VT_STRUCT_SHIFT 16   /* structure/enum name shift (16 bits left) */
@end example

When a reference to another type is needed (for pointers, functions and
structures), the @code{32 - VT_STRUCT_SHIFT} high order bits are used to
store an identifier reference.

The @code{VT_UNSIGNED} flag can be set for chars, shorts, ints and long
longs.

Arrays are considered as pointers @code{VT_PTR} with the flag
@code{VT_ARRAY} set.

The @code{VT_BITFIELD} flag can be set for chars, shorts, ints and long
longs. If it is set, then the bitfield position is stored from bits
VT_STRUCT_SHIFT to VT_STRUCT_SHIFT + 5 and the bit field size is stored
from bits VT_STRUCT_SHIFT + 6 to VT_STRUCT_SHIFT + 11.

@code{VT_LONG} is never used except during parsing.

During parsing, the storage of an object is also stored in the type
integer:

@example
#define VT_EXTERN  0x00000080  /* extern definition */
#define VT_STATIC  0x00000100  /* static variable */
#define VT_TYPEDEF 0x00000200  /* typedef definition */
@end example

@section Symbols

All symbols are stored in hashed symbol stacks. Each symbol stack
contains @code{Sym} structures.

@code{Sym.v} contains the symbol name (remember
an idenfier is also a token, so a string is never necessary to store
it). @code{Sym.t} gives the type of the symbol. @code{Sym.r} is usually
the register in which the corresponding variable is stored. @code{Sym.c} is
usually a constant associated to the symbol.

Four main symbol stacks are defined:

@table @code

@item define_stack
for the macros (@code{#define}s).

@item global_stack
for the global variables, functions and types.

@item local_stack
for the local variables, functions and types.

@item label_stack
for the local labels (for @code{goto}).

@end table

@code{sym_push()} is used to add a new symbol in the local symbol
stack. If no local symbol stack is active, it is added in the global
symbol stack.

@code{sym_pop(st,b)} pops symbols from the symbol stack @var{st} until
the symbol @var{b} is on the top of stack. If @var{b} is NULL, the stack
is emptied.

@code{sym_find(v)} return the symbol associated to the identifier
@var{v}. The local stack is searched first from top to bottom, then the
global stack.

@section Sections

The generated code and datas are written in sections. The structure
@code{Section} contains all the necessary information for a given
section. @code{new_section()} creates a new section. ELF file semantics
is assumed for each section.

The following sections are predefined:

@table @code

@item text_section
is the section containing the generated code. @var{ind} contains the
current position in the code section.

@item data_section
contains initialized data

@item bss_section
contains uninitialized data

@item bounds_section
@itemx lbounds_section
are used when bound checking is activated

@item stab_section
@itemx stabstr_section
are used when debugging is actived to store debug information

@item symtab_section
@itemx strtab_section
contain the exported symbols (currently only used for debugging).

@end table

@section Code generation

@subsection Introduction

The TCC code generator directly generates linked binary code in one
pass. It is rather unusual these days (see gcc for example which
generates text assembly), but it allows to be very fast and surprisingly
not so complicated.

The TCC code generator is register based. Optimization is only done at
the expression level. No intermediate representation of expression is
kept except the current values stored in the @emph{value stack}.

On x86, three temporary registers are used. When more registers are
needed, one register is flushed in a new local variable.

@subsection The value stack

When an expression is parsed, its value is pushed on the value stack
(@var{vstack}). The top of the value stack is @var{vtop}. Each value
stack entry is the structure @code{SValue}.

@code{SValue.t} is the type. @code{SValue.r} indicates how the value is
currently stored in the generated code. It is usually a CPU register
index (@code{REG_xxx} constants), but additionnal values and flags are
defined:

@example
#define VT_CONST     0x00f0
#define VT_LLOCAL    0x00f1
#define VT_LOCAL     0x00f2
#define VT_CMP       0x00f3
#define VT_JMP       0x00f4
#define VT_JMPI      0x00f5
#define VT_LVAL      0x0100
#define VT_SYM       0x0200
#define VT_MUSTCAST  0x0400
#define VT_MUSTBOUND 0x0800
#define VT_BOUNDED   0x8000
#define VT_LVAL_BYTE     0x1000
#define VT_LVAL_SHORT    0x2000
#define VT_LVAL_UNSIGNED 0x4000
#define VT_LVAL_TYPE     (VT_LVAL_BYTE | VT_LVAL_SHORT | VT_LVAL_UNSIGNED)
@end example

@table @code

@item VT_CONST
indicates that the value is a constant. It is stored in the union
@code{SValue.c}, depending on its type.

@item VT_LOCAL
indicates a local variable pointer at offset @code{SValue.c.i} in the
stack.

@item VT_CMP
indicates that the value is actually stored in the CPU flags (i.e. the
value is the consequence of a test). The value is either 0 or 1. The
actual CPU flags used is indicated in @code{SValue.c.i}. 

If any code is generated which destroys the CPU flags, this value MUST be
put in a normal register.

@item VT_JMP
@itemx VT_JMPI
indicates that the value is the consequence of a jmp. For VT_JMP, it is
1 if the jump is taken, 0 otherwise. For VT_JMPI it is inverted.

These values are used to compile the @code{||} and @code{&&} logical
operators.

If any code is generated, this value MUST be put in a normal
register. Otherwise, the generated code won't be executed if the jump is
taken.

@item VT_LVAL
is a flag indicating that the value is actually an lvalue (left value of
an assignment). It means that the value stored is actually a pointer to
the wanted value. 

Understanding the use @code{VT_LVAL} is very important if you want to
understand how TCC works.

@item VT_LVAL_BYTE
@itemx VT_LVAL_SHORT
@itemx VT_LVAL_UNSIGNED
if the lvalue has an integer type, then these flags give its real
type. The type alone is not suffisant in case of cast optimisations.

@item VT_LLOCAL
is a saved lvalue on the stack. @code{VT_LLOCAL} should be suppressed
ASAP because its semantics are rather complicated.

@item VT_MUSTCAST
indicates that a cast to the value type must be performed if the value
is used (lazy casting).

@item VT_SYM
indicates that the symbol @code{SValue.sym} must be added to the constant.

@item VT_MUSTBOUND
@itemx VT_BOUNDED
are only used for optional bound checking.

@end table

@subsection Manipulating the value stack

@code{vsetc()} and @code{vset()} pushes a new value on the value
stack. If the previous @code{vtop} was stored in a very unsafe place(for
example in the CPU flags), then some code is generated to put the
previous @code{vtop} in a safe storage.

@code{vpop()} pops @code{vtop}. In some cases, it also generates cleanup
code (for example if stacked floating point registers are used as on
x86).

The @code{gv(rc)} function generates code to evaluate @code{vtop} (the
top value of the stack) into registers. @var{rc} selects in which
register class the value should be put. @code{gv()} is the @emph{most
important function} of the code generator.

@code{gv2()} is the same as @code{gv()} but for the top two stack
entries.

@subsection CPU dependent code generation

See the @file{i386-gen.c} file to have an example.

@table @code

@item load()
must generate the code needed to load a stack value into a register.

@item store()
must generate the code needed to store a register into a stack value
lvalue.

@item gfunc_start()
@itemx gfunc_param()
@itemx gfunc_call()
should generate a function call

@item gfunc_prolog()
@itemx gfunc_epilog()
should generate a function prolog/epilog.

@item gen_opi(op)
must generate the binary integer operation @var{op} on the two top
entries of the stack which are guaranted to contain integer types.

The result value should be put on the stack.

@item gen_opf(op)
same as @code{gen_opi()} for floating point operations. The two top
entries of the stack are guaranted to contain floating point values of
same types.

@item gen_cvt_itof()
integer to floating point conversion.

@item gen_cvt_ftoi()
floating point to integer conversion.

@item gen_cvt_ftof()
floating point to floating point of different size conversion.

@item gen_bounded_ptr_add()
@item gen_bounded_ptr_deref()
are only used for bound checking.

@end table

@section Optimizations done

Constant propagation is done for all operations. Multiplications and
divisions are optimized to shifts when appropriate. Comparison
operators are optimized by maintaining a special cache for the
processor flags. &&, || and ! are optimized by maintaining a special
'jump target' value. No other jump optimization is currently performed
because it would require to store the code in a more abstract fashion.