nasmstab

开源的nasm编译器源码,研究编译器原理很有帮且

#!/usr/bin/perl
 
sub StabLine  ($ $ $ $ $ $) {
        local ($comment,$n_strx,$type,$other,$desc,$value) = @_;
        print $comment;
        print "","dd",$n_strx;
        print "","db",$type;
        print "","db",$other;
        print "","dw",$desc;
        print "","dd","0" . $value . "h";
}

sub RStabLine  ($ $ $ $ $) {
        local ($comment,$offset,$info,$type,$symbol) = @_;
        print $comment;
        print "","dd",$offset;
        print "","db",$type;
        print "","db",$symbol;
        print "","dw",$info;
}

#this sub exists because i've no idea how to print non-ascii numbers in perl

sub OutBin ( $ $ ) {
        local ($offset, $shnum) = @_;
        seek(FINAL,$offset,0);
        if ( $shnum == 2 ) { printf FINAL "\x02" } ;
        if ( $shnum == 3 ) { printf FINAL "\x03" } ;
        if ( $shnum == 4 ) { printf FINAL "\x04" } ;
        if ( $shnum == 5 ) { printf FINAL "\x05" } ;
        if ( $shnum == 6 ) { printf FINAL "\x06" } ;
        if ( $shnum == 7 ) { printf FINAL "\x07" } ;
        if ( $shnum == 8 ) { printf FINAL "\x08" } ;
        if ( $shnum == 9 ) { printf FINAL "\x09" } ;
        if ( $shnum == 10 ) { printf FINAL "\x0a" } ;
        if ( $shnum == 11 ) { printf FINAL "\x0b" } ;
        if ( $shnum == 12 ) { printf FINAL "\x0c" } ;
        if ( $shnum == 13 ) { printf FINAL "\x0d" } ;
        if ( $shnum == 14 ) { printf FINAL "\x0e" } ;
        if ( $shnum == 15 ) { printf FINAL "\x0f" } ;
}

sub DispHelp () {
        $\="\n";
        print "Usage:";
        print "\t-f,--input-file";
        print "\t\tThe input file name (only required option)";
        print "\t-o,--output-file";
        print "\t\tThe output file name (if not specified, *.asm becomes *.o";
        print "\t\tand anything else becomes a.out)";
        print "\t-l,--list-file";
        print "\t\tThe listing file's name (default: trailing .asm is
removed";
        print "\t\tif there and .lst is appended)";
        print "\t-s,--second-asm-file";
        print "\t\tThe second asm file's name (default: trailing .asm is";
        print "\t\tremoved if there and .nasm is appended)";
        print "\n";
        exit ;
}

if ( $ARGV[0] eq "" ) { $ARGV[0] = "-h" };

$i = 0;
$filename = "";
$outname  = "";

while ( $ARGV[$i] ne "" ) {
        $_ = $ARGV[$i];
        if ( m/^-/ ) {
                if ( m/^-f$/ ) { $filename = $ARGV[++$i] };
                if ( m/^-o$/ ) { $outname  = $ARGV[++$i] };
                if ( m/^-l$/ ) { $listname = $ARGV[++$i] };
                if ( m/^-s$/ ) { $asmname  = $ARGV[++$i] };
                if ( m/^-h$/ ) { DispHelp };
        } elsif ( m/^--\w+/ ) {
                if ( m/^--input-file$/ )      { $filename = $ARGV[++$i] };
                if ( m/^--output-file$/ )     { $outname  = $ARGV[++$i] };
                if ( m/^--list-file$/ )       { $listname = $ARGV[++$i] };
                if ( m/^--second-asm-file$/ ) { $asmname  = $ARGV[++$i] };
                if ( m/^--help/ ) { DispHelp };
        } elsif ( m/^--$/ ) {
                while ( $ARGV[++$i] ) {
                        $NasmOptions .= " ";
                        $NasmOptions .= $_;
                };
        } else {
                DispHelp()
        };
        $i++;
};

if ( $filename eq "" ) { DispHelp() };

if ( $outname eq "" ) {
        $outname = $filename;
        $outname =~ s/\.asm/.o/;
        if ( $outname eq $filename ) { $outname = "a.out" };
};

if ( $listname eq "" ) {
        $listname = $filename;
        $listname =~ s/\.asm//;
        $listname .= ".lst";
};

if ( $asmname eq "" ) {
        $asmname = $filename;
        $asmname =~ s/\.asm//;
        $asmname .= ".nasm";
};

$err = `nasm -f elf ${filename} -l ${listname} -o ${outname} `;

if ( $err ) { die "\n$err\n"};

open(LISTFILE,"${listname}") or die "\n $0: Could not reopen list file!\n";
open(ASMFILE,">${asmname}")  or die "\n $0: Could not open asm file!\n";

select ASMFILE;

open(OLDASM,$filename) or die "\n$0: Cannot open file $filename\n";

while ( $x = <OLDASM> ) {
        print $x;
}

@stab = ("n_desc", "value");
@rel_stab = ("offset");
$i = 0;
$current_section = "";
$has_text = 'FALSE';
$midst_of_macro = 'FALSE';
$line_dec = 0 ;

while ( $x = <LISTFILE> ) {
        if ( $x =~ m/[^;]*%include/ ) {
                $x = <LISTFILE>;
                while ( $x =~ m/\s+\d+\s+\<\d+\>\s+/ ) {
                        $x = <LISTFILE>;
                        $line_dec++;
                }
        }
        if ( $current_section eq ".text" ) {
                if ( $x =~ m/^\s+(\S+)\s+(\S+)\s+(\S+)\s+[^<]+$/ ) {
                        $stab[$i++] = $1-$line_dec;     #linenum
                        $stab[$i++] = $2;               #offset
                        $count++;
                        if ( $3 =~ m/-/ ) {
                                $x = <LISTFILE>;
                                $line_dec++;
                        }
                        $midst_of_macro = 'FALSE';
                } elsif ( $x =~ m/^\s+(\S+)\s+(\S+)\s+(\S+)\s+<\d+>/ ) {
                        if ( $midst_of_macro eq 'TRUE' ) {
                                $stab[$i] = $stab[$i-2]; #same linenum
                                $line_dec++;
                        } else {
                                $stab[$i] = $1 - ++$line_dec;
                                $midst_of_macro = 'TRUE';
                        }
                        $count++;
                        $i++;
                        $stab[$i++] = $2;
                        if ( $3 =~ m/-/ ) {
                                $x = <LISTFILE>;
                                $line_dec++;
                        }

                }
                $has_text = 'TRUE';
        } elsif ( $x =~ m/\s+\S+\s+\S+\s+\S+\s+<\d+>/ ) { # is it a macro?
                $line_dec++;
        }
        if ( $x =~ s/(section|segment)\s+([^\s]+)/$2/ ) {
                        $current_section = $2;
        }
};

close LISTFILE;

unless ( $has_text eq "TRUE" ) {
        $err = `nasm -f elf ${asmname} -o ${outname}`;
        print STDERR $err;
        exit;
}

#Write Stab section
$, = "\t";      #output field separator
$\ = "\n";      #output record separator

print "section .stab noalloc";
StabLine(";header",1,0,0,$count+1,length($filename)*2+3);
StabLine(";so",length($asmname)+2,"064h",0,0,0);

$offset = 12;
$i = 0;
$j = 0;
$rel_stab[$j++] = $offset + 8;

while ( $stab[$i] ) {
        StabLine(";N_SLINE" . " " . ( ($i+2) / 2 ), 0, "044h", 0,
                $stab[$i++], $stab[$i++]);
        $offset += 12;
        $rel_stab[$j++] = $offset + 8;
}

#Write .rel.stab section
print "\n\nsection .rel.stab noalloc";

open (READELF,"readelf -s ${outname} |") or die "\n$0: Could not run readelf\n";

while ( $x = <READELF> ) {
   if ( $x =~ m/\s+(\d+):\s+00000000\s+\d+\s+SECTION\s+\w+\s+\w+\s+1\s+/){ $textsymnum = $1;
   };
};
close READELF;

$i = 0;

while ( $rel_stab[$i] ne "" ) {
        RStabLine(";relocation for N_SLINE " . ($i), $rel_stab[$i], 0, 1, $textsymnum);
        $i++;
} ;

#Write .stabstr section

print "\n\nsection .stabstr noalloc";

print "","db","0";
print "","db",'"' . $asmname . '"';
print "","db","0";
print "","db",'"' . $asmname . '"' ;
print "","db","0";

close ASMFILE;

$err = `nasm -f elf ${asmname} -o ${outname}`;

if ( $err ) { die "\n$err\n" } ;

open (READELF,"readelf -h -S ${outname} |") or die "\n$0: Could not run readelf\n";


while ( $x = <READELF> ) {
        if ( $x =~ m/Start\s+of\s+section\s+headers:\s+(\d+)\s+/ ) {
                $shoff = $1;
        }
        if ( $x =~ m/Size\s+of\s+section\s+headers:\s+(\d+)\s+/ ) {
                $shentsize = $1;
        }
        if ( $x =~ m/\[\s*(\d+)\]\s+.rel.stab\s+/ ) {
                $relnum = $1;
        }
        if ( $x =~ m/\[\s*(\d+)\]\s+.stab\s+/ ) {
                $stabnum = $1;
        }
        if ( $x =~ m/\[\s*(\d+)\]\s+.stabstr\s+/ ) {
                $stabstrnum = $1;
        }
        if ( $x =~ m/\[\s*(\d+)\]\s+.symtab\s+/ ) {
                $symtabnum = $1;
        }
}
close READELF;

sysopen (FINAL,"${outname}",2,0) or die "\n$0: Could not open ${outname}";
$, = "";        #output field separator
$\ = "";        #output record separator

#set .rel.stab->type to rel
OutBin($shoff + ($shentsize * $relnum) + 4,9);

#set .rel.stab->link to .symtab
OutBin($shoff + ($shentsize * $relnum) + 24,$symtabnum);

#set .rel.stab->info to .stab
OutBin($shoff + ($shentsize * $relnum) + 28,$stabnum);

#set .rel.stab->entsize to 8
OutBin($shoff + ($shentsize * $relnum) + 36,8);

#set .stab->link to .stabstr
OutBin($shoff + ($shentsize * $stabnum) + 24,$stabstrnum);

#set .stab->entsize to 12
OutBin($shoff + ($shentsize * $stabnum) + 36,12);

#set .stabstr->type to strtab
OutBin($shoff + ($shentsize * $stabstrnum) + 4,3);

close FINAL;

#Date: 17 Mar 2002 15:51:20 -0800
#From: [e-mail] (kired)
#Newsgroups: alt.lang.asm