home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
EnigmA Amiga Run 1999 March
/
EnigmA AMIGA RUN 35 (1999)(G.R. Edizioni)(IT)[!][issue 1999-03].iso
/
earcd
/
devel
/
vbcc-68k
/
doc
/
vbccalpha.doc
< prev
next >
Wrap
Text File
|
1999-01-01
|
4KB
|
134 lines
vbcc - C compiler (c) in 1995-99 by Volker Barthelmann
INTRODUCTION
vbcc is a free portable and retargetable ANSI C compiler.
It is clearly split into a target independant and a target dependant
part and supports emulating datatypes of the target machine on any
other machine so that it is possible to e.g. make a crosscompiler for
a 64bit machine on a 32bit machine.
This document only deals with the target dependant parts of the
DEC Alpha version.
This is a pre-alpha version!
LEGAL
vbcc is (c) in 1995-99 by Volker Barthelmann. All code is written by me
and may be freely redistributed as long as no modifications are made
and nothing is charged for it.
Non-commercial usage of vbcc is allowed without any restrictions.
Commercial usage needs my written consent.
Sending me money, gifts, postcards etc. would of course be very nice
and may encourage further development of vbcc, but is not legally or
morally necessary to use vbcc.
ADDITIONAL OPTIONS FOR THIS VERSION
-merge-constants
Place identical floating point constants at the same
memory location. This can reduce program size and increase
compilation time.
-const-in-data
By default constant data will be placed in the code
section (and therefore is accessable with faster pc-relative
addressing modes). Using this option it will be placed in the
data section.
Note that on operating systems with memory protection this
option will disable write-protection of constant data.
-no-builtins
Do not replace certain builtin functions by inline code.
This may be useful if you use this code generator with
another frontend than vbcc.
stdarg.h will not work with -no-builtins.
SOME INTERNALS
The current version generates assembly output for use with the GNU
assembler. The generated code should work on systems with 21064,
21066, 21164 and higher Alpha CPUs.
The register names are:
$0 through $31 for the general purpose registers and
$f0 through $f31 for the floating point registers
The registers $0-$8, $16-$28, $f0, $f1 and $f10-$f30
are used as scratch registers (i.e. they can be destroyed in function
calls), all other registers are preserved. Of course $31 and $f31
cannot be used.
The first 6 function arguments which have arithmetic or pointer types
are passed in registers $16/$f16 through $21/$f21.
Integers and pointers are returned in $0, floats and doubles in $f0.
All other types are returned by passing the function the address
of the result as a hidden argument - so when you call such a function
without a proper declaration in scope you can expect a crash.
The elementary data types are represented like:
type size in bits alignment in bytes
char 8 1
short 16 2
int 32 4
long 64 8
all pointers 64 8
float 32 4
double 64 8
STDARG
A possible <stdarg.h> could look like this:
typedef struct {
char *regbase;
char *membase;
int arg;
} va_list;
char *__va_start(void);
int __va_fixargs(void);
#define va_start(vl,dummy) \
(vl.arg=__va_fixargs(),vl.regbase=__va_start(),vl.membase=vl.regbase+(6-vl.arg)*16)
#define va_end(vl) (vl.regbase=vl.membase=0)
#define __va_size(type) ((sizeof(type)+7)/8*8)
#define va_arg(vl,type) \
( \
((__typeof(type)&15)<=8&&++vl.arg<=6) ? \
( \
((__typeof(type)&15)==5||(__typeof(type)&15)==6) ? \
(vl.regbase+=16,*(type *)(vl.regbase-8)) \
: \
(vl.regbase+=16,*(type *)(vl.regbase-16)) \
) \
: \
(vl.membase+=__va_size(type),*(type *)(vl.membase-__va_size(type))) \
)
KNOWN PROBLEMS
- generated code is very poor
- nested function calls which pass parameters on the stack will not work
- several other problems
Volker Barthelmann volker@vb.franken.de