c2ph

早期freebsd实现

#!/usr/contrib/bin/perl
#
#
#   c2ph (aka pstruct)
#   Tom Christiansen, <[e-mail]>
#   
#   As pstruct, dump C structures as generated from 'cc -g -S' stabs.
#   As c2ph, do this PLUS generate perl code for getting at the structures.
#
#   See the usage message for more.  If this isn't enough, read the code.
#

$RCSID = '$RCSfile: c2ph.SH,v $$Revision: 4.0.1.2 $$Date: 92/06/08 11:56:08 $';


######################################################################

# some handy data definitions.   many of these can be reset later.

$bitorder = 'b';  # ascending; set to B for descending bit fields

%intrinsics = 
%template = (
    'char', 			'c',
    'unsigned char', 		'C',
    'short',			's',
    'short int',		's',
    'unsigned short',		'S',
    'unsigned short int',	'S',
    'short unsigned int',	'S',
    'int',			'i',
    'unsigned int',		'I',
    'long',			'l',
    'long int',			'l',
    'unsigned long',		'L',
    'unsigned long',		'L',
    'long unsigned int',	'L',
    'unsigned long int',	'L',
    'long long',		'q',
    'long long int',		'q',
    'unsigned long long',	'Q',
    'unsigned long long int',	'Q',
    'float',			'f',
    'double',			'd',
    'pointer',			'p',
    'null',			'x',
    'neganull',			'X',
    'bit',			$bitorder,
); 

&buildscrunchlist;
delete $intrinsics{'neganull'};
delete $intrinsics{'bit'};
delete $intrinsics{'null'};

# use -s to recompute sizes
%sizeof = (
    'char', 			'1',
    'unsigned char', 		'1',
    'short',			'2',
    'short int',		'2',
    'unsigned short',		'2',
    'unsigned short int',	'2',
    'short unsigned int',	'2',
    'int',			'4',
    'unsigned int',		'4',
    'long',			'4',
    'long int',			'4',
    'unsigned long',		'4',
    'unsigned long int',	'4',
    'long unsigned int',	'4',
    'long long',		'8',
    'long long int',		'8',
    'unsigned long long',	'8',
    'unsigned long long int',	'8',
    'float',			'4',
    'double',			'8',
    'pointer',			'4',
);

($type_width, $member_width, $offset_width, $size_width) = (20, 20, 6, 5);

($offset_fmt, $size_fmt) = ('d', 'd');

$indent = 2;

$CC = 'cc';
$CFLAGS = '-g -S';
$DEFINES = '';

$perl++ if $0 =~ m#/?c2ph$#;

require 'getopts.pl';

eval '$'.$1.'$2;' while $ARGV[0] =~ /^([A-Za-z_]+=)(.*)/ && shift;

&Getopts('aixdpvtnws:') || &usage(0);

$opt_d && $debug++;
$opt_t && $trace++;
$opt_p && $perl++;
$opt_v && $verbose++;
$opt_n && ($perl = 0);

if ($opt_w) {
    ($type_width, $member_width, $offset_width) = (45, 35, 8);
} 
if ($opt_x) {
    ($offset_fmt, $offset_width, $size_fmt, $size_width) = ( 'x', '08', 'x', 04 );
}

eval '$'.$1.'$2;' while $ARGV[0] =~ /^([A-Za-z_]+=)(.*)/ && shift;

sub PLUMBER {
    select(STDERR);
    print "oops, apperent pager foulup\n";
    $isatty++;
    &usage(1);
} 

sub usage {
    local($oops) = @_;
    unless (-t STDOUT) {
	select(STDERR);
    } elsif (!$oops) {
	$isatty++;
	$| = 1;
	print "hit <RETURN> for further explanation: ";
	<STDIN>;
	open (PIPE, "|". ($ENV{PAGER} || 'more'));
	$SIG{PIPE} = PLUMBER;
	select(PIPE);
    } 

    print "usage: $0 [-dpnP] [var=val] [files ...]\n";

    exit unless $isatty;

    print <<EOF;

Options:

-w	wide; short for: type_width=45 member_width=35 offset_width=8
-x	hex; short for:  offset_fmt=x offset_width=08 size_fmt=x size_width=04

-n  	do not generate perl code  (default when invoked as pstruct)
-p  	generate perl code         (default when invoked as c2ph)
-v	generate perl code, with C decls as comments

-i	do NOT recompute sizes for intrinsic datatypes
-a	dump information on intrinsics also

-t 	trace execution
-d	spew reams of debugging output

-slist  give comma-separated list a structures to dump


Var Name        Default Value    Meaning

EOF

    &defvar('CC', 'which_compiler to call');
    &defvar('CFLAGS', 'how to generate *.s files with stabs');
    &defvar('DEFINES','any extra cflags or cpp defines, like -I, -D, -U');

    print "\n";

    &defvar('type_width', 'width of type field   (column 1)');
    &defvar('member_width', 'width of member field (column 2)');
    &defvar('offset_width', 'width of offset field (column 3)');
    &defvar('size_width', 'width of size field   (column 4)');

    print "\n";

    &defvar('offset_fmt', 'sprintf format type for offset');
    &defvar('size_fmt', 'sprintf format type for size');

    print "\n";

    &defvar('indent', 'how far to indent each nesting level');

   print <<'EOF';

    If any *.[ch] files are given, these will be catted together into
    a temporary *.c file and sent through:
	    $CC $CFLAGS $DEFINES 
    and the resulting *.s groped for stab information.  If no files are
    supplied, then stdin is read directly with the assumption that it
    contains stab information.  All other liens will be ignored.  At
    most one *.s file should be supplied.

EOF
    close PIPE;
    exit 1;
} 

sub defvar {
    local($var, $msg) = @_;
    printf "%-16s%-15s  %s\n", $var, eval "\$$var", $msg;
} 

$recurse = 1;

if (@ARGV) {
    if (grep(!/\.[csh]$/,@ARGV)) {
	warn "Only *.[csh] files expected!\n";
	&usage;
    } 
    elsif (grep(/\.s$/,@ARGV)) {
	if (@ARGV > 1) { 
	    warn "Only one *.s file allowed!\n";
	    &usage;
	}
    } 
    elsif (@ARGV == 1 && $ARGV[0] =~ /\.c$/) {
	local($dir, $file) = $ARGV[0] =~ m#(.*/)?(.*)$#;
	$chdir = "cd $dir; " if $dir;
	&system("$chdir$CC $CFLAGS $DEFINES $file") && exit 1;
	$ARGV[0] =~ s/\.c$/.s/;
    } 
    else {
	$TMP = "/tmp/c2ph.$$.c";
	&system("cat @ARGV > $TMP") && exit 1;
	&system("cd /tmp; $CC $CFLAGS $DEFINES $TMP") && exit 1;
	unlink $TMP;
	$TMP =~ s/\.c$/.s/;
	@ARGV = ($TMP);
    } 
}

if ($opt_s) {
    for (split(/[\s,]+/, $opt_s)) {
	$interested{$_}++;
    } 
} 


$| = 1 if $debug;

main: {

    if ($trace) {
	if (-t && !@ARGV) { 
	    print STDERR "reading from your keyboard: ";
	} else {
	    print STDERR "reading from " . (@ARGV ? "@ARGV" : "<STDIN>").": ";
	}
    }

STAB: while (<>) {
	if ($trace && !($. % 10)) {
	    $lineno = $..'';
	    print STDERR $lineno, "\b" x length($lineno);
	} 
	next unless /^\s*\.stabs\s+/;
	$line = $_;
	s/^\s*\.stabs\s+//; 
	&stab; 
    }
    print STDERR "$.\n" if $trace;
    unlink $TMP if $TMP;

    &compute_intrinsics if $perl && !$opt_i;

    print STDERR "resolving types\n" if $trace;

    &resolve_types;
    &adjust_start_addrs;

    $sum = 2 + $type_width + $member_width;
    $pmask1 = "%-${type_width}s %-${member_width}s"; 
    $pmask2 = "%-${sum}s %${offset_width}${offset_fmt}%s %${size_width}${size_fmt}%s";

    if ($perl) {
	# resolve template -- should be in stab define order, but even this isn't enough.
	print STDERR "\nbuilding type templates: " if $trace;
	for $i (reverse 0..$#type) {
	    next unless defined($name = $type[$i]);
	    next unless defined $struct{$name};
	    $build_recursed = 0;
	    &build_template($name) unless defined $template{&psou($name)} ||
					$opt_s && !$interested{$name};
	} 
	print STDERR "\n\n" if $trace;
    }

    print STDERR "dumping structs: " if $trace;


    foreach $name (sort keys %struct) {
	next if $opt_s && !$interested{$name};
	print STDERR "$name " if $trace;

	undef @sizeof;
	undef @typedef;
	undef @offsetof;
	undef @indices;
	undef @typeof;

	$mname = &munge($name);

	$fname = &psou($name);

	print "# " if $perl && $verbose;
	$pcode = '';
	print "$fname {\n" if !$perl || $verbose; 
	$template{$fname} = &scrunch($template{$fname}) if $perl;
	&pstruct($name,$name,0); 
	print "# " if $perl && $verbose;
	print "}\n" if !$perl || $verbose; 
	print "\n" if $perl && $verbose;

	if ($perl) {
	    print "$pcode";

	    printf("\nsub %-32s { %4d; }\n\n", "${mname}'struct", $countof{$name});

	    print <<EOF;
sub ${mname}'typedef { 
    local(\$${mname}'index) = shift;
    defined \$${mname}'index 
	? \$${mname}'typedef[\$${mname}'index] 
	: \$${mname}'typedef;
}
EOF

	    print <<EOF;
sub ${mname}'sizeof { 
    local(\$${mname}'index) = shift;
    defined \$${mname}'index 
	? \$${mname}'sizeof[\$${mname}'index] 
	: \$${mname}'sizeof;
}
EOF

	    print <<EOF;
sub ${mname}'offsetof { 
    local(\$${mname}'index) = shift;
    defined \$${mname}index 
	? \$${mname}'offsetof[\$${mname}'index] 
	: \$${mname}'sizeof;
}
EOF

	    print <<EOF;
sub ${mname}'typeof { 
    local(\$${mname}'index) = shift;
    defined \$${mname}index 
	? \$${mname}'typeof[\$${mname}'index] 
	: '$name';
}
EOF
    

	    print "\$${mname}'typedef = '" . &scrunch($template{$fname}) 
		. "';\n";

	    print "\$${mname}'sizeof = $sizeof{$name};\n\n";


	    print "\@${mname}'indices = (", &squishseq(@indices), ");\n";

	    print "\n";

	    print "\@${mname}'typedef[\@${mname}'indices] = (",
			join("\n\t", '', @typedef), "\n    );\n\n";
	    print "\@${mname}'sizeof[\@${mname}'indices] = (",
			join("\n\t", '', @sizeof), "\n    );\n\n";
	    print "\@${mname}'offsetof[\@${mname}'indices] = (",
			join("\n\t", '', @offsetof), "\n    );\n\n";
	    print "\@${mname}'typeof[\@${mname}'indices] = (",
			join("\n\t", '', @typeof), "\n    );\n\n";

	    $template_printed{$fname}++;
	    $size_printed{$fname}++;
	} 
	print "\n";
    }

    print STDERR "\n" if $trace;

    unless ($perl && $opt_a) { 
	print "\n1;\n";
	exit;
    }



    foreach $name (sort bysizevalue keys %intrinsics) {
	next if $size_printed{$name};
	print '$',&munge($name),"'sizeof = ", $sizeof{$name}, ";\n";
    }

    print "\n";

    sub bysizevalue { $sizeof{$a} <=> $sizeof{$b}; }


    foreach $name (sort keys %intrinsics) {
	print '$',&munge($name),"'typedef = '", $template{$name}, "';\n";
    }

    print "\n1;\n";
	
    exit;
}

########################################################################################


sub stab {
    next unless /:[\$\w]+(\(\d+,\d+\))?=[\*\$\w]+/;  # (\d+,\d+) is for sun
    s/"// 						|| next;
    s/",([x\d]+),([x\d]+),([x\d]+),.*// 		|| next;

    next if /^\s*$/;

    $size = $3 if $3;


    $line = $_;

    if (($name, $pdecl) = /^([\$ \w]+):[tT]((\d+)(=[rufs*](\d+))+)$/) {
	print "$name is a typedef for some funky pointers: $pdecl\n" if $debug;
	&pdecl($pdecl);
	next;
    }



    if (/(([ \w]+):t(\d+|\(\d+,\d+\)))=r?(\d+|\(\d+,\d+\))(;\d+;\d+;)?/) {  
	local($ident) = $2;
	push(@intrinsics, $ident);
	$typeno = &typeno($3);
	$type[$typeno] = $ident;
	print STDERR "intrinsic $ident in new type $typeno\n" if $debug; 
	next;
    }

    if (($name, $typeordef, $typeno, $extra, $struct, $_) 
	= /^([\$ \w]+):([ustT])(\d+|\(\d+,\d+\))(=[rufs*](\d+))?(.*)$/) 
    {
	$typeno = &typeno($typeno);  # sun foolery
    } 
    elsif (/^[\$\w]+:/) {
	next; # variable
    }
    else { 
	warn "can't grok stab: <$_> in: $line " if $_;
	next;
    } 

    #warn "got size $size for $name\n";
    $sizeof{$name} = $size if $size;

    s/;[-\d]*;[-\d]*;$//;  # we don't care about ranges

    $typenos{$name} = $typeno;

    unless (defined $type[$typeno]) {
	&panic("type 0??") unless $typeno;
	$type[$typeno] = $name unless defined $type[$typeno];
	printf "new type $typeno is $name" if $debug;
	if ($extra =~ /\*/ && defined $type[$struct]) {
	    print ", a typedef for a pointer to " , $type[$struct] if $debug;
	}
    } else {
	printf "%s is type %d", $name, $typeno if $debug;
	print ", a typedef for " , $type[$typeno] if $debug;
    } 
    print "\n" if $debug;
    #next unless $extra =~ /[su*]/;

    #$type[$struct] = $name;

    if ($extra =~ /[us*]/) {
	&sou($name, $extra);
	$_ = &sdecl($name, $_, 0);
    }
    elsif (/^=ar/) {
	print "it's a bare array typedef -- that's pretty sick\n" if $debug;
	$_ = "$typeno$_";
	$scripts = '';
	$_ = &adecl($_,1);

    }
    elsif (s/((\w+):t(\d+|\(\d+,\d+\)))?=r?(;\d+;\d+;)?//) {  # the ?'s are for gcc
	push(@intrinsics, $2);
	$typeno = &typeno($3);
	$type[$typeno] = $2;
	print STDERR "intrinsic $2 in new type $typeno\n" if $debug; 
    }
    elsif (s/^=e//) { # blessed by thy compiler; mine won't do this
	&edecl;
    } 
    else {
	warn "Funny remainder for $name on line $_ left in $line " if $_;
    } 
}

sub typeno {  # sun thinks types are (0,27) instead of just 27
    local($_) = @_;
    s/\(\d+,(\d+)\)/$1/;
    $_;
} 

sub pstruct {
    local($what,$prefix,$base) = @_; 
    local($field, $fieldname, $typeno, $count, $offset, $entry); 
    local($fieldtype);
    local($type, $tname); 
    local($mytype, $mycount, $entry2);
    local($struct_count) = 0;
    local($pad, $revpad, $length, $prepad, $lastoffset, $lastlength, $fmt);
    local($bits,$bytes);
    local($template);


    local($mname) = &munge($name);

    sub munge { 
	local($_) = @_;
	s/[\s\$\.]/_/g;
	$_;
    }

    local($sname) = &psou($what);

    $nesting++;

    for $field (split(/;/, $struct{$what})) {
	$pad = $prepad = 0;
	$entry = ''; 
	($fieldname, $typeno, $count, $offset, $length) = split(/,/, $field);