optab.pl

It s a Linux disassemble, can set break point, disassemble ELF file.

#!/usr/bin/perl
#
# optab.pl
# Generate source/include files from input file optab-x86.dat
#
# Note: this script sets up the assembler/disassembler files
# in similar manner to NASM. However, because of NASM's
# restricted licensing, I was forced to "reinvent the wheel"
# so to speak. As a result, the overall idea of how this
# works is similar to NASM, but the actual implementation
# is different.
#
# $Id: optab.pl,v 1.1.1.1 2004/04/26 00:40:21 pa33 Exp $

$InputFile = "optab-x86.dat";
$SourceOutputFileD = "dtab-x86.c";
$SourceOutputFileA = "atab-x86.c";
$IncludeOutputFileN = "inames-x86.h";
$IncludeOutputFile = "optab-x86.h";

open (INFILE, $InputFile) || die "Unable to open $InputFile";

# Current line number
$linenum = 0;

# Offset in our opcode array
$offset = 0;

while (<INFILE>)
{
  $linenum++;

  # skip comments
  next if /^\s*#/;

  # Remove newlines
  chomp;

  # Split line into substrings
  split;

  # skip blank lines
  next if $#_ == -1;

  # Make sure there are at least three fields in the line
  if ($#_ < 2)
  {
    warn "Line $linenum contains less than three fields\n";
    next;
  }

  # Format the line - convert intel syntax to our bitmask names
  # etc
  ($FormattedLine) = &format(@_);

  if ($FormattedLine)
  {
    ## print "format = $FormattedLine\n";
    ## print "0 = $_[0], 1 = $_[1], 2 = $_[2] 3 = $_[3]\n";
    $CodeName = "Code_$_[0]";
    push @$CodeName, $FormattedLine;

    # Our opcodes are stored as two digits separated by
    # commas (XX,YY,ZZ), so extract the first byte of the
    # opcode, and store it in our Op array
    $ii = substr($_[2], 0, 2);
    $CodeName = "Op_$ii";
    push @$CodeName, $offset;
  }

  $oparray{$_[0]}++;
  $offset++;
}

close INFILE;

@opcodes = sort keys(%oparray);

# Generate assembly table file

print STDERR "Generating ../source/$SourceOutputFileA\n";
open (OUTFILE, ">../source/$SourceOutputFileA") ||
  die "Unable to open ../source/$SourceOutputFileA";

print OUTFILE "/*\n";
print OUTFILE " * This file was generated by optab.pl - do not edit.\n";
print OUTFILE " */\n\n";

print OUTFILE "#include \"common-x86.h\"\n";
print OUTFILE "#include \"operands-x86.h\"\n";
print OUTFILE "#include \"optab-x86.h\"\n";
print OUTFILE "#include \"regs-x86.h\"\n";

print OUTFILE "\n";

foreach $ii (@opcodes)
{
  print OUTFILE "static struct x86OpCode Instruction_${ii}[] = {\n";
  $CodeName = "Code_${ii}";
  foreach $jj (@$CodeName)
  {
    print OUTFILE "  $jj\n";
  }
  print OUTFILE "  { -1, -1, { -1, -1, -1 }, 0, -1, -1, { -1, -1, -1 } }\n};\n\n";
}

print OUTFILE "struct x86OpCode *x86Instructions[] = {";
$first = 1;
foreach $ii (@opcodes)
{
  if (!$first) { print OUTFILE ","; } else { $first = 0; }
  print OUTFILE "\n  Instruction_${ii}";
}
print OUTFILE "\n};\n";

close OUTFILE;

# Generate disassembly table file

print STDERR "Generating ../source/$SourceOutputFileD\n";
open (OUTFILE, ">../source/$SourceOutputFileD") ||
  die "Unable to open ../source/$SourceOutputFileD";

print OUTFILE "/*\n";
print OUTFILE " * This file was generated by optab.pl - do not edit.\n";
print OUTFILE " */\n\n";

print OUTFILE "#include \"common-x86.h\"\n";
print OUTFILE "#include \"operands-x86.h\"\n";
print OUTFILE "#include \"$IncludeOutputFile\"\n";
print OUTFILE "#include \"regs-x86.h\"\n";

print OUTFILE "\nstatic struct x86OpCode Instructions[] = {\n";

foreach $ii (@opcodes)
{
  $CodeName = "Code_$ii";
  foreach $jj (@$CodeName)
  {
    print OUTFILE "  $jj\n";
  }
}

print OUTFILE "  { 0, 0, { 0, 0, 0 }, 0, 0, 0, { 0, 0, 0 } }\n";
print OUTFILE "};\n\n";

for ($c = 0; $c < 256; $c++)
{
  $h = sprintf "%02X", $c;
  print OUTFILE "static struct x86OpCode *OpCode_${h}[] = {\n";
  $CodeName = "Op_$h";
  foreach $jj (@$CodeName)
  {
    print OUTFILE "  Instructions + $jj,\n";
  }
  print OUTFILE "  0\n};\n\n";
}

print OUTFILE "struct x86OpCode **x86OpCodes[] = {\n";
for ($c = 0; $c < 256; $c++)
{
  printf OUTFILE "  OpCode_%02X,\n", $c;
}
print OUTFILE "};\n";

close OUTFILE;

print STDERR "Generating ../include/$IncludeOutputFileN\n";
open (OUTFILE, ">../include/$IncludeOutputFileN") ||
  die "Unable to open ../include/$IncludeOutputFileN";

print OUTFILE "/*\n";
print OUTFILE " * This file was generated by optab.pl - do not edit.\n";
print OUTFILE " */\n\n";

print OUTFILE "#ifndef INCLUDED_inames_x86_h\n";
print OUTFILE "#define INCLUDED_inames_x86_h\n\n";

print OUTFILE "static const char *x86InstructionNames[] = {";
$first = 1;
foreach $ii (@opcodes)
{
  if (!$first) { print OUTFILE ","; } else { $first = 0; }
  $lower = $ii;
  $lower =~ tr/A-Z/a-z/;
  #print OUTFILE "  \"$lower\",\n";
  print OUTFILE "\n  \"$lower\"";
}
print OUTFILE "\n};\n";

print OUTFILE "\n#endif /* INCLUDED_inames_x86_h */\n";

close OUTFILE;

print STDERR "Generating ../include/$IncludeOutputFile\n";
open (OUTFILE, ">../include/$IncludeOutputFile") ||
  die "Unable to open ../include/$IncludeOutputFile";

print OUTFILE "/*\n";
print OUTFILE " * This file was generated by optab.pl - do not edit.\n";
print OUTFILE " */\n\n";

print OUTFILE "#ifndef INCLUDED_optab_x86_h\n";
print OUTFILE "#define INCLUDED_optab_x86_h\n\n";

print OUTFILE "enum\n{\n";

print OUTFILE join(",\n", map { "  I_$_" } @opcodes), "\n";

print OUTFILE "};\n";

print OUTFILE "\n";
print OUTFILE "extern struct x86OpCode **x86OpCodes[];\n";
print OUTFILE "extern struct x86OpCode *x86Instructions[];\n";
#print OUTFILE "extern const char *x86InstructionNames[];\n";

print OUTFILE "\n#endif /* INCLUDED_optab_x86_h */\n";

close OUTFILE;

# This function takes an array of strings as input, formats
# them, and returns an array entry for our Instructions[]
# array.
#
# Inputs: arg1 - instruction name
#         arg2 - list of operands separated by commas
#         arg3 - list of opcode bytes separated by commas
#         arg4 - (optional) additional info such as modrm values
sub format
{
  local ($inst, $operands, $codes, $modrm) = @_;
  local $opcnt = 0;
  local $oparg1 = "-1";
  local $oparg2 = "-1";
  local $oparg3 = "-1";

  # format the operands

  $operands =~ s/fpureg/reg_fpu/g;
  $operands =~ s/imm(\d+)/imm|bits$1/g;
  $operands =~ s/imm/immediate/g;
  $operands =~ s/mem(\d+)/mem|bits$1/g;
  $operands =~ s/mem/memory/g;
  $operands =~ s/mmxreg/reg_mmx/g;
  $operands =~ s/moffs/memoffs/g;
  $operands =~ s/rel(\d+)/relative|bits$1/g;
  $operands =~ s/rm(\d+)/regmem|bits$1/g;
  $operands =~ s/segreg/reg_sr/g;
  $operands =~ s/xmmreg/reg_xmm/g;
  $operands =~ s/conreg/reg_control/g;
  $operands =~ s/debreg/reg_debug/g;
  $operands =~ s/ptr(\d+):(\d+)/seg$1|off$2/g;
  if ($operands !~ /_rm/) { $operands =~ s/rm/regmem/g; }

  # Add a BITS32 to OFF32 opcodes so they will get priority over
  # their 16 bit versions, if any
  $operands =~ s/off32/off32|bits32/g;

  # Handle rDD(xxx) situations
  ($op1, $op2, $op3) = split(/,/, $operands, 3);

  if ($op1 =~ s/r(\d+)\((\w+)\)/register|bits$1/g) { $oparg1 = "r_$2"; }
  if ($op1 =~ s/r\((\w+)\)/register/g) { $oparg1 = "r_$1"; }
  if ($op1 =~ s/const(\d+)/constant/g) { $oparg1 = $1; }
  if ($op1 =~ s/reg(\d+)_rm/reg$1/g) { $oparg1 = "MODFIELD_RM"; }

  if ($op2 =~ s/r(\d+)\((\w+)\)/register|bits$1/g) { $oparg2 = "r_$2"; }
  if ($op2 =~ s/r\((\w+)\)/register/g) { $oparg2 = "r_$1"; }
  if ($op2 =~ s/const(\d+)/constant/g) { $oparg2 = $1; }
  if ($op2 =~ s/reg(\d+)_rm/reg$1/g) { $oparg2 = "MODFIELD_RM"; }

  if ($op3 =~ s/r(\d+)\((\w+)\)/register|bits$1/g) { $oparg3 = "r_$2"; }
  if ($op3 =~ s/r\((\w+)\)/register/g) { $oparg3 = "r_$1"; }
  if ($op3 =~ s/const(\d+)/constant/g) { $oparg3 = $1; }
  if ($op3 =~ s/reg(\d+)_rm/reg$1/g) { $oparg3 = "MODFIELD_RM"; }

  if (length($op3) != 0)
  {
    $opcnt = 3;
    $finalop = "$op1, $op2, $op3";
  }
  elsif (length($op2) != 0)
  {
    $opcnt = 2;
    $finalop = "$op1, $op2, 0";
  }
  elsif (length($op1) != 0)
  {
    $opcnt = 1;
    $finalop = "$op1, 0, 0";
  }
  else
  {
    $opcnt = 0;
    $finalop = "0, 0, 0";
  }

  $finalop = '0, 0, 0', $opcnt = 0 if $operands eq 'void';
  $finalop = 'BITS32, 0, 0', $opcnt = 0 if $operands eq 'void32';
  $finalop =~ tr/a-z/A-Z/;

  # We must convert opargs to uppercase in case they hold
  # a register index (ie: r_al -> R_AL)
  $oparg1 =~ tr/a-z/A-Z/;
  $oparg2 =~ tr/a-z/A-Z/;
  $oparg3 =~ tr/a-z/A-Z/;

  # codes will come in looking something like: F1,F2,F3
  # Replace the commas with a \x and stick a \x on the front
  # to get the first opcode (\xF1\xF2\xF3)

  $oplen = $codes =~ tr/\,/\,/;
  $oplen++;

  $codes =~ s/\,/\\x/g;
  $codes = "\\x" . $codes;

  # If there is no digit associated with the opcode, make the field
  # -1. If the digit is /r, make the field "REGRM", otherwise it's
  # numerical value. A value of "c" indicates a register code
  # that is added to the opcode.
  $modrm = -1 if $modrm eq '';
  $modrm = 'REGRM' if $modrm eq 'r';
  $modrm = 'REGCODE' if $modrm eq 'c';
  $modrm = 'FPUCODE' if $modrm eq 'f';

  ("{ I_$inst, $opcnt, { $finalop }, \"$codes\", $oplen, $modrm, { $oparg1, $oparg2, $oparg3 } },");
}