helpers/llvm-arm-dis

#!/usr/bin/env perl

# This script uses llvm-mc to disassemble byte patterns into possible
# x86/x86_64/arm/aarch64 instructions, trying to find a way to represent
# that in any of the instruction sets.

use strict;
use warnings;
use Data::Dumper;

#vpush-vpop.s:@ CHECK-ARM: vpop  {s8, s9, s10, s11, s12} @ encoding: [0x05,0x4a,0xbd,0xec]
#vpush-vpop.s:@ CHECK-THUMB: vpush {d8, d9, d10, d11, d12} @ encoding: [0x2d,0xed,0x0a,0x8b]
my $syntax = "Syntax: $0 0xFFFF(FFFF)\n";
my @hexs = &fix_endian($ARGV[0]);
#die Dumper \@hexs;

foreach (@hexs) {
  my $code = $_->{'code'};
  my $arch = $_->{'arch'};
  print "$arch: $code: ";
  print `echo "$code" | llvm-mc -disassemble -triple $arch`;
}

sub fix_endian() {
  my ($hex) = @_;

  my ($a, $b, $c, $d) = ($hex =~ /^0?x?(\w{2})(\w{2})(\w{0,2})(\w{0,2})$/);
  print "$a $b $c $d\n";
  die $syntax unless $a and $b; # at least two bytes
  die $syntax if $c and not $d; # two or four bytes
  # ARM/T2/x86
  if ($c) {
    return (
      { 'code' => "0x$d 0x$c 0x$b 0x$a", 'arch' => "armv7" },
      { 'code' => "0x$b 0x$a 0x$d 0x$c", 'arch' => "thumbv7" },
      { 'code' => "0x$d 0x$c 0x$b 0x$a", 'arch' => "i686" },
      { 'code' => "0x$d 0x$c 0x$b 0x$a", 'arch' => "x86_64" }
    );
  } else {
      return ( { 'code' => "0x$b 0x$a", 'arch' => "thumb" } );
  }
}