[p5sagit/p5-mst-13.2.git] / ext / B / B / Assembler.pm

#      Assembler.pm
#
#      Copyright (c) 1996 Malcolm Beattie
#
#      You may distribute under the terms of either the GNU General Public
#      License or the Artistic License, as specified in the README file.

package B::Assembler;
use Exporter;
use B qw(ppname);
use B::Asmdata qw(%insn_data @insn_name);
use Config qw(%Config);
require ByteLoader;		# we just need its $VERSIOM

@ISA = qw(Exporter);
@EXPORT_OK = qw(assemble_fh newasm endasm assemble);
$VERSION = 0.04;

use strict;
my %opnumber;
my ($i, $opname);
for ($i = 0; defined($opname = ppname($i)); $i++) {
    $opnumber{$opname} = $i;
}

my($linenum, $errors, $out); #	global state, set up by newasm

sub error {
    my $str = shift;
    warn "$linenum: $str\n";
    $errors++;
}

my $debug = 0;
sub debug { $debug = shift }

sub limcheck($$$$){
    my( $val, $lo, $hi, $loc ) = @_;
    if( $val < $lo || $hi < $val ){
        error "argument for $loc outside [$lo, $hi]: $val";
        $val = $hi;
    }
    return $val;
}

#
# First define all the data conversion subs to which Asmdata will refer
#

sub B::Asmdata::PUT_U8 {
    my $arg = shift;
    my $c = uncstring($arg);
    if (defined($c)) {
	if (length($c) != 1) {
	    error "argument for U8 is too long: $c";
	    $c = substr($c, 0, 1);
	}
    } else {
        $arg = limcheck( $arg, 0, 0xff, 'U8' );
	$c = chr($arg);
    }
    return $c;
}

sub B::Asmdata::PUT_U16 {
    my $arg = limcheck( $_[0], 0, 0xffff, 'U16' );
    pack("S", $arg);
}
sub B::Asmdata::PUT_U32 {
    my $arg = limcheck( $_[0], 0, 0xffffffff, 'U32' );
    pack("L", $arg);
}
sub B::Asmdata::PUT_I32 {
    my $arg = limcheck( $_[0], -0x80000000, 0x7fffffff, 'I32' );
    pack("L", $arg);
}
sub B::Asmdata::PUT_NV  { sprintf("%s\0", $_[0]) } # "%lf" looses precision and pack('d',...)
						   # may not even be portable between compilers
sub B::Asmdata::PUT_objindex { # could allow names here
    my $arg = limcheck( $_[0], 0, 0xffffffff, '*index' );
    pack("L", $arg);
} 
sub B::Asmdata::PUT_svindex { &B::Asmdata::PUT_objindex }
sub B::Asmdata::PUT_opindex { &B::Asmdata::PUT_objindex }
sub B::Asmdata::PUT_pvindex { &B::Asmdata::PUT_objindex }

sub B::Asmdata::PUT_strconst {
    my $arg = shift;
    my $str = uncstring($arg);
    if (!defined($str)) {
	error "bad string constant: $arg";
	$str = '';
    }
    if ($str =~ s/\0//g) {
	error "string constant argument contains NUL: $arg";
        $str = '';
    }
    return $str . "\0";
}

sub B::Asmdata::PUT_pvcontents {
    my $arg = shift;
    error "extraneous argument: $arg" if defined $arg;
    return "";
}
sub B::Asmdata::PUT_PV {
    my $arg = shift;
    my $str = uncstring($arg);
    if( ! defined($str) ){
        error "bad string argument: $arg";
        $str = '';
    }
    return pack("L", length($str)) . $str;
}
sub B::Asmdata::PUT_comment_t {
    my $arg = shift;
    $arg = uncstring($arg);
    error "bad string argument: $arg" unless defined($arg);
    if ($arg =~ s/\n//g) {
	error "comment argument contains linefeed: $arg";
    }
    return $arg . "\n";
}
sub B::Asmdata::PUT_double { sprintf("%s\0", $_[0]) } # see PUT_NV above
sub B::Asmdata::PUT_none {
    my $arg = shift;
    error "extraneous argument: $arg" if defined $arg;
    return "";
}
sub B::Asmdata::PUT_op_tr_array {
    my $arg = shift;
    my @ary = split(/\s*,\s*/, $arg);
    if (@ary != 256) {
	error "wrong number of arguments to op_tr_array";
	@ary = (0) x 256;
    }
    return pack("S256", @ary);
}
# XXX Check this works
# Note: $arg >> 32 is a no-op on 32-bit systems
sub B::Asmdata::PUT_IV64 {
    my $arg = shift;
    return pack("LL", ($arg >> 16) >>16 , $arg & 0xffffffff);
}

sub B::Asmdata::PUT_IV {
    $Config{ivsize} == 4 ? &B::Asmdata::PUT_I32 : &B::Asmdata::PUT_IV64;
}

my %unesc = (n => "\n", r => "\r", t => "\t", a => "\a",
	     b => "\b", f => "\f", v => "\013");

sub uncstring {
    my $s = shift;
    $s =~ s/^"// and $s =~ s/"$// or return undef;
    $s =~ s/\\(\d\d\d|.)/length($1) == 3 ? chr(oct($1)) : ($unesc{$1}||$1)/eg;
    return $s;
}

sub strip_comments {
    my $stmt = shift;
    # Comments only allowed in instructions which don't take string arguments
    $stmt =~ s{
	(?sx)	# Snazzy extended regexp coming up. Also, treat
		# string as a single line so .* eats \n characters.
	^\s*	# Ignore leading whitespace
	(
	  [^"]*	# A double quote '"' indicates a string argument. If we
		# find a double quote, the match fails and we strip nothing.
	)
	\s*\#	# Any amount of whitespace plus the comment marker...
	.*$	# ...which carries on to end-of-string.
    }{$1};	# Keep only the instruction and optional argument.
    return $stmt;
}

# create the ByteCode header: magic, archname, ByteLoader $VERSION, ivsize,
# 	ptrsize, byteorder
# nvtype is irrelevant (floats are stored as strings)
# byteorder is strconst not U32 because of varying size issues

sub gen_header {
    my $header = "";

    $header .= B::Asmdata::PUT_U32(0x43424c50);	# 'PLBC'
    $header .= B::Asmdata::PUT_strconst('"' . $Config{archname}. '"');
    $header .= B::Asmdata::PUT_strconst(qq["$ByteLoader::VERSION"]);
    $header .= B::Asmdata::PUT_U32($Config{ivsize});
    $header .= B::Asmdata::PUT_U32($Config{ptrsize});
    $header .= B::Asmdata::PUT_strconst(sprintf(qq["0x%s"], $Config{byteorder}));

    $header;
}

sub parse_statement {
    my $stmt = shift;
    my ($insn, $arg) = $stmt =~ m{
	^\s*	# allow (but ignore) leading whitespace
	(.*?)	# Instruction continues up until...
	(?:	# ...an optional whitespace+argument group
	    \s+		# first whitespace.
	    (.*)	# The argument is all the rest (newlines included).
	)?$	# anchor at end-of-line
    }sx;
    if (defined($arg)) {
	if ($arg =~ s/^0x(?=[0-9a-fA-F]+$)//) {
	    $arg = hex($arg);
	} elsif ($arg =~ s/^0(?=[0-7]+$)//) {
	    $arg = oct($arg);
	} elsif ($arg =~ /^pp_/) {
	    $arg =~ s/\s*$//; # strip trailing whitespace
	    my $opnum = $opnumber{$arg};
	    if (defined($opnum)) {
		$arg = $opnum;
	    } else {
		error qq(No such op type "$arg");
		$arg = 0;
	    }
	}
    }
    return ($insn, $arg);
}

sub assemble_insn {
    my ($insn, $arg) = @_;
    my $data = $insn_data{$insn};
    if (defined($data)) {
	my ($bytecode, $putsub) = @{$data}[0, 1];
	my $argcode = &$putsub($arg);
	return chr($bytecode).$argcode;
    } else {
	error qq(no such instruction "$insn");
	return "";
    }
}

sub assemble_fh {
    my ($fh, $out) = @_;
    my $line;
    my $asm = newasm($out);
    while ($line = <$fh>) {
	assemble($line);
    }
    endasm();
}

sub newasm {
    my($outsub) = @_;

    die "Invalid printing routine for B::Assembler\n" unless ref $outsub eq 'CODE';
    die <<EOD if ref $out;
Can't have multiple byteassembly sessions at once!
	(perhaps you forgot an endasm()?)
EOD

    $linenum = $errors = 0;
    $out = $outsub;

    $out->(gen_header());
}

sub endasm {
    if ($errors) {
	die "There were $errors assembly errors\n";
    }
    $linenum = $errors = $out = 0;
}

sub assemble {
    my($line) = @_;
    my ($insn, $arg);
    $linenum++;
    chomp $line;
    if ($debug) {
	my $quotedline = $line;
	$quotedline =~ s/\\/\\\\/g;
	$quotedline =~ s/"/\\"/g;
	$out->(assemble_insn("comment", qq("$quotedline")));
    }
    if( $line = strip_comments($line) ){
        ($insn, $arg) = parse_statement($line);
        $out->(assemble_insn($insn, $arg));
        if ($debug) {
	    $out->(assemble_insn("nop", undef));
        }
    }
}

1;

__END__

=head1 NAME

B::Assembler - Assemble Perl bytecode

=head1 SYNOPSIS

	use B::Assembler qw(newasm endasm assemble);
	newasm(\&printsub);	# sets up for assembly
	assemble($buf); 	# assembles one line
	endasm();		# closes down

	use B::Assembler qw(assemble_fh);
	assemble_fh($fh, \&printsub);	# assemble everything in $fh

=head1 DESCRIPTION

See F<ext/B/B/Assembler.pm>.

=head1 AUTHORS

Malcolm Beattie, C<mbeattie@sable.ox.ac.uk>
Per-statement interface by Benjamin Stuhl, C<sho_pi@hotmail.com>

=cut
Commit	Line	Data
a798dbf2	1	# Assembler.pm
	2	#
	3	# Copyright (c) 1996 Malcolm Beattie
	4	#
	5	# You may distribute under the terms of either the GNU General Public
	6	# License or the Artistic License, as specified in the README file.
059a8bb7	7
a798dbf2	8	package B::Assembler;
	9	use Exporter;
	10	use B qw(ppname);
	11	use B::Asmdata qw(%insn_data @insn_name);
059a8bb7	12	use Config qw(%Config);
059a8bb7	13	require ByteLoader; # we just need its $VERSIOM
a798dbf2	14
a798dbf2	15	@ISA = qw(Exporter);
059a8bb7	16	@EXPORT_OK = qw(assemble_fh newasm endasm assemble);
f4abc3e7	17	$VERSION = 0.04;
a798dbf2	18
	19	use strict;
	20	my %opnumber;
	21	my ($i, $opname);
	22	for ($i = 0; defined($opname = ppname($i)); $i++) {
	23	$opnumber{$opname} = $i;
	24	}
	25
059a8bb7	26	my($linenum, $errors, $out); # global state, set up by newasm
a798dbf2	27
	28	sub error {
	29	my $str = shift;
	30	warn "$linenum: $str\n";
	31	$errors++;
	32	}
	33
	34	my $debug = 0;
	35	sub debug { $debug = shift }
	36
f4abc3e7	37	sub limcheck($$$$){
	38	my( $val, $lo, $hi, $loc ) = @_;
	39	if( $val < $lo \|\| $hi < $val ){
	40	error "argument for $loc outside [$lo, $hi]: $val";
	41	$val = $hi;
	42	}
	43	return $val;
	44	}
	45
a798dbf2	46	#
	47	# First define all the data conversion subs to which Asmdata will refer
	48	#
	49
	50	sub B::Asmdata::PUT_U8 {
	51	my $arg = shift;
	52	my $c = uncstring($arg);
	53	if (defined($c)) {
	54	if (length($c) != 1) {
	55	error "argument for U8 is too long: $c";
	56	$c = substr($c, 0, 1);
	57	}
	58	} else {
f4abc3e7	59	$arg = limcheck( $arg, 0, 0xff, 'U8' );
a798dbf2	60	$c = chr($arg);
	61	}
	62	return $c;
	63	}
	64
f4abc3e7	65	sub B::Asmdata::PUT_U16 {
	66	my $arg = limcheck( $_[0], 0, 0xffff, 'U16' );
	67	pack("S", $arg);
	68	}
	69	sub B::Asmdata::PUT_U32 {
	70	my $arg = limcheck( $_[0], 0, 0xffffffff, 'U32' );
	71	pack("L", $arg);
	72	}
	73	sub B::Asmdata::PUT_I32 {
	74	my $arg = limcheck( $_[0], -0x80000000, 0x7fffffff, 'I32' );
	75	pack("L", $arg);
	76	}
	77	sub B::Asmdata::PUT_NV { sprintf("%s\0", $_[0]) } # "%lf" looses precision and pack('d',...)
059a8bb7	78	# may not even be portable between compilers
f4abc3e7	79	sub B::Asmdata::PUT_objindex { # could allow names here
	80	my $arg = limcheck( $_[0], 0, 0xffffffff, '*index' );
	81	pack("L", $arg);
	82	}
cf743617	83	sub B::Asmdata::PUT_svindex { &B::Asmdata::PUT_objindex }
cf743617	84	sub B::Asmdata::PUT_opindex { &B::Asmdata::PUT_objindex }
059a8bb7	85	sub B::Asmdata::PUT_pvindex { &B::Asmdata::PUT_objindex }
a798dbf2	86
	87	sub B::Asmdata::PUT_strconst {
	88	my $arg = shift;
f4abc3e7	89	my $str = uncstring($arg);
f4abc3e7	90	if (!defined($str)) {
a798dbf2	91	error "bad string constant: $arg";
f4abc3e7	92	$str = '';
a798dbf2	93	}
f4abc3e7	94	if ($str =~ s/\0//g) {
a798dbf2	95	error "string constant argument contains NUL: $arg";
f4abc3e7	96	$str = '';
a798dbf2	97	}
f4abc3e7	98	return $str . "\0";
a798dbf2	99	}
	100
	101	sub B::Asmdata::PUT_pvcontents {
	102	my $arg = shift;
	103	error "extraneous argument: $arg" if defined $arg;
	104	return "";
	105	}
	106	sub B::Asmdata::PUT_PV {
	107	my $arg = shift;
f4abc3e7	108	my $str = uncstring($arg);
	109	if( ! defined($str) ){
	110	error "bad string argument: $arg";
	111	$str = '';
	112	}
	113	return pack("L", length($str)) . $str;
a798dbf2	114	}
d53d2b93	115	sub B::Asmdata::PUT_comment_t {
a798dbf2	116	my $arg = shift;
	117	$arg = uncstring($arg);
	118	error "bad string argument: $arg" unless defined($arg);
	119	if ($arg =~ s/\n//g) {
	120	error "comment argument contains linefeed: $arg";
	121	}
	122	return $arg . "\n";
	123	}
059a8bb7	124	sub B::Asmdata::PUT_double { sprintf("%s\0", $_[0]) } # see PUT_NV above
a798dbf2	125	sub B::Asmdata::PUT_none {
	126	my $arg = shift;
	127	error "extraneous argument: $arg" if defined $arg;
	128	return "";
	129	}
	130	sub B::Asmdata::PUT_op_tr_array {
	131	my $arg = shift;
	132	my @ary = split(/\s,\s/, $arg);
	133	if (@ary != 256) {
	134	error "wrong number of arguments to op_tr_array";
	135	@ary = (0) x 256;
	136	}
059a8bb7	137	return pack("S256", @ary);
a798dbf2	138	}
a798dbf2	139	# XXX Check this works
f4abc3e7	140	# Note: $arg >> 32 is a no-op on 32-bit systems
a798dbf2	141	sub B::Asmdata::PUT_IV64 {
a798dbf2	142	my $arg = shift;
f4abc3e7	143	return pack("LL", ($arg >> 16) >>16 , $arg & 0xffffffff);
	144	}
	145
	146	sub B::Asmdata::PUT_IV {
	147	$Config{ivsize} == 4 ? &B::Asmdata::PUT_I32 : &B::Asmdata::PUT_IV64;
a798dbf2	148	}
	149
	150	my %unesc = (n => "\n", r => "\r", t => "\t", a => "\a",
	151	b => "\b", f => "\f", v => "\013");
	152
	153	sub uncstring {
	154	my $s = shift;
	155	$s =~ s/^"// and $s =~ s/"$// or return undef;
	156	$s =~ s/\\(\d\d\d\|.)/length($1) == 3 ? chr(oct($1)) : ($unesc{$1}\|\|$1)/eg;
	157	return $s;
	158	}
	159
	160	sub strip_comments {
	161	my $stmt = shift;
	162	# Comments only allowed in instructions which don't take string arguments
	163	$stmt =~ s{
	164	(?sx) # Snazzy extended regexp coming up. Also, treat
	165	# string as a single line so .* eats \n characters.
	166	^\s* # Ignore leading whitespace
	167	(
	168	[^"]* # A double quote '"' indicates a string argument. If we
	169	# find a double quote, the match fails and we strip nothing.
	170	)
	171	\s*\# # Any amount of whitespace plus the comment marker...
	172	.*$ # ...which carries on to end-of-string.
	173	}{$1}; # Keep only the instruction and optional argument.
	174	return $stmt;
	175	}
	176
059a8bb7	177	# create the ByteCode header: magic, archname, ByteLoader $VERSION, ivsize,
	178	# ptrsize, byteorder
	179	# nvtype is irrelevant (floats are stored as strings)
	180	# byteorder is strconst not U32 because of varying size issues
	181
	182	sub gen_header {
	183	my $header = "";
	184
	185	$header .= B::Asmdata::PUT_U32(0x43424c50); # 'PLBC'
	186	$header .= B::Asmdata::PUT_strconst('"' . $Config{archname}. '"');
	187	$header .= B::Asmdata::PUT_strconst(qq["$ByteLoader::VERSION"]);
	188	$header .= B::Asmdata::PUT_U32($Config{ivsize});
	189	$header .= B::Asmdata::PUT_U32($Config{ptrsize});
	190	$header .= B::Asmdata::PUT_strconst(sprintf(qq["0x%s"], $Config{byteorder}));
	191
	192	$header;
	193	}
	194
a798dbf2	195	sub parse_statement {
	196	my $stmt = shift;
	197	my ($insn, $arg) = $stmt =~ m{
a798dbf2	198	^\s* # allow (but ignore) leading whitespace
	199	(.*?) # Instruction continues up until...
	200	(?: # ...an optional whitespace+argument group
	201	\s+ # first whitespace.
	202	(.*) # The argument is all the rest (newlines included).
	203	)?$ # anchor at end-of-line
f4abc3e7	204	}sx;
a798dbf2	205	if (defined($arg)) {
	206	if ($arg =~ s/^0x(?=[0-9a-fA-F]+$)//) {
	207	$arg = hex($arg);
	208	} elsif ($arg =~ s/^0(?=[0-7]+$)//) {
	209	$arg = oct($arg);
	210	} elsif ($arg =~ /^pp_/) {
	211	$arg =~ s/\s*$//; # strip trailing whitespace
	212	my $opnum = $opnumber{$arg};
	213	if (defined($opnum)) {
	214	$arg = $opnum;
	215	} else {
	216	error qq(No such op type "$arg");
	217	$arg = 0;
	218	}
	219	}
	220	}
	221	return ($insn, $arg);
	222	}
	223
	224	sub assemble_insn {
	225	my ($insn, $arg) = @_;
	226	my $data = $insn_data{$insn};
	227	if (defined($data)) {
	228	my ($bytecode, $putsub) = @{$data}[0, 1];
	229	my $argcode = &$putsub($arg);
	230	return chr($bytecode).$argcode;
	231	} else {
	232	error qq(no such instruction "$insn");
	233	return "";
	234	}
	235	}
	236
	237	sub assemble_fh {
	238	my ($fh, $out) = @_;
059a8bb7	239	my $line;
059a8bb7	240	my $asm = newasm($out);
a798dbf2	241	while ($line = <$fh>) {
059a8bb7	242	assemble($line);
a798dbf2	243	}
059a8bb7	244	endasm();
	245	}
	246
	247	sub newasm {
	248	my($outsub) = @_;
	249
	250	die "Invalid printing routine for B::Assembler\n" unless ref $outsub eq 'CODE';
	251	die <<EOD if ref $out;
	252	Can't have multiple byteassembly sessions at once!
	253	(perhaps you forgot an endasm()?)
	254	EOD
	255
	256	$linenum = $errors = 0;
	257	$out = $outsub;
	258
	259	$out->(gen_header());
	260	}
	261
	262	sub endasm {
a798dbf2	263	if ($errors) {
059a8bb7	264	die "There were $errors assembly errors\n";
	265	}
	266	$linenum = $errors = $out = 0;
	267	}
	268
	269	sub assemble {
	270	my($line) = @_;
	271	my ($insn, $arg);
	272	$linenum++;
	273	chomp $line;
	274	if ($debug) {
	275	my $quotedline = $line;
	276	$quotedline =~ s/\\/\\\\/g;
	277	$quotedline =~ s/"/\\"/g;
	278	$out->(assemble_insn("comment", qq("$quotedline")));
	279	}
f4abc3e7	280	if( $line = strip_comments($line) ){
	281	($insn, $arg) = parse_statement($line);
	282	$out->(assemble_insn($insn, $arg));
	283	if ($debug) {
	284	$out->(assemble_insn("nop", undef));
	285	}
a798dbf2	286	}
	287	}
	288
	289	1;
7f20e9dd	290
	291	__END__
	292
	293	=head1 NAME
	294
	295	B::Assembler - Assemble Perl bytecode
	296
	297	=head1 SYNOPSIS
	298
059a8bb7	299	use B::Assembler qw(newasm endasm assemble);
	300	newasm(\&printsub); # sets up for assembly
	301	assemble($buf); # assembles one line
	302	endasm(); # closes down
	303
	304	use B::Assembler qw(assemble_fh);
	305	assemble_fh($fh, \&printsub); # assemble everything in $fh
7f20e9dd	306
	307	=head1 DESCRIPTION
	308
	309	See F<ext/B/B/Assembler.pm>.
	310
059a8bb7	311	=head1 AUTHORS
7f20e9dd	312
7f20e9dd	313	Malcolm Beattie, C<mbeattie@sable.ox.ac.uk>
059a8bb7	314	Per-statement interface by Benjamin Stuhl, C<sho_pi@hotmail.com>
7f20e9dd	315
7f20e9dd	316	=cut