This is a native Ruby implementation of the SMB Protocol Family. It currently supports:
The RubySMB library provides client-level and packet-level support for the protocol. A user can parse and manipulate raw SMB packets, or use the client to perform higher-level SMB operations.
See the Wiki for more information on this project's long-term goals, style guide, and developer tips.
RubySMB attempts to follow the Semantic Versioning 2.0.0 standard, however due to certain key architectural qualities, some reservations are made regarding what will trigger major and minor version bumps. Upstream consumers utilizing higher-level SMB operations can lock to minor (~= 3.0
) versions however consumers utilizing raw SMB packet manipulation are recommended to set restrictions for compatibility to allow update resolution for patch versions (~= 3.0.0
).
Add this line to your application's Gemfile:
gem 'ruby_smb'
And then execute:
$ bundle
Or install it yourself as:
$ gem install ruby_smb
All packets are implemented in a declarative style with BinData. Nested data structures are used where appropriate to give users an easy method of manipulating individual fields inside of a packet.
Note that these packets are defined using BinData and that their struction is subject to change across minor version bumps. See CONTRIBUTING.md for more information.
SMB1 Packets are made up of three basic components:
- The SMB Header - This is a standard SMB Header. All SMB1 packets use the same SMB header.
- The Parameter Block - This is where function parameters are passed across the wire in the packet. Parameter blocks will always have a 'Word Count' field that gives the size of the parameter block in words (2-bytes)
- The Data Block - This is the data section of the packet. The data block will always have a 'byte count' field that gives the size of the Data block in bytes.
The SMB Header can always just be declared as a field in the BinData DSL for the packet class, because its structure never changes. For the Parameter block and data blocks, we always define subclasses for this particular packet. They inherit the 'Word Count' and 'Byte Count' fields, along with the auto-calculation routines for those fields, from their ancestors. Any other fields are then defined in our subclass before we start the DSL declarations for the packet.
Example:
module RubySMB
module SMB1
module Packet
# This class represents an SMB1 TreeConnect Request Packet as defined in
# [2.2.4.7.1 Client Request Extensions](https://msdn.microsoft.com/en-us/library/cc246330.aspx)
class TreeConnectRequest < RubySMB::GenericPacket
# A SMB1 Parameter Block as defined by the {TreeConnectRequest}
class ParameterBlock < RubySMB::SMB1::ParameterBlock
and_x_block :andx_block
tree_connect_flags :flags
uint16 :password_length, label: 'Password Length', initial_value: 0x01
end
class DataBlock < RubySMB::SMB1::DataBlock
stringz :password, label: 'Password Field', initial_value: '', length: lambda { self.parent.parameter_block.password_length }
stringz :path, label: 'Resource Path'
stringz :service, label: 'Resource Type', initial_value: '?????'
end
smb_header :smb_header
parameter_block :parameter_block
data_block :data_block
def initialize_instance
super
smb_header.command = RubySMB::SMB1::Commands::SMB_COM_TREE_CONNECT
end
end
end
end
end
SMB2 Packets are far simpler than their older SMB1 counterparts. We still abstract out the SMB2 header since it is the same structure used for every packet. Beyond that, the SMB2 packet is relatively flat in comparison to SMB1.
Example:
module RubySMB
module SMB2
module Packet
# An SMB2 TreeConnectRequest Packet as defined in
# [2.2.9 SMB2 TREE_CONNECT Request](https://msdn.microsoft.com/en-us/library/cc246567.aspx)
class TreeConnectRequest < RubySMB::GenericPacket
endian :little
smb2_header :smb2_header
uint16 :structure_size, label: 'Structure Size', initial_value: 9
uint16 :flags, label: 'Flags', initial_value: 0x00
uint16 :path_offset, label: 'Path Offset', initial_value: 0x48
uint16 :path_length, label: 'Path Length', initial_value: lambda { self.path.length }
string :path, label: 'Path Buffer'
def initialize_instance
super
smb2_header.command = RubySMB::SMB2::Commands::TREE_CONNECT
end
def encode_path(path)
self.path = path.encode("utf-16le")
end
end
end
end
end
You can create an instance of any particular packet class, and then reach into the data structure to set or read explicit values in a fairly straightforward manner.
Example:
2.3.3 :001 > packet = RubySMB::SMB1::Packet::TreeConnectRequest.new
=> {:smb_header=>{:protocol=>4283649346, :command=>117, :nt_status=>0, :flags=>{:reply=>0, :opbatch=>0, :oplock=>0, :canonicalized_paths=>1, :case_insensitive=>1, :reserved=>0, :buf_avail=>0, :lock_and_read_ok=>0}, :flags2=>{:reserved1=>0, :is_long_name=>0, :reserved2=>0, :signature_required=>0, :compressed=>0, :security_signature=>0, :eas=>0, :long_names=>1, :unicode=>0, :nt_status=>1, :paging_io=>1, :dfs=>0, :extended_security=>0, :reparse_path=>0}, :pid_high=>0, :security_features=>"\x00\x00\x00\x00\x00\x00\x00\x00", :reserved=>0, :tid=>0, :pid_low=>0, :uid=>0, :mid=>0}, :parameter_block=>{:word_count=>4, :andx_block=>{:andx_command=>255, :andx_reserved=>0, :andx_offset=>0}, :flags=>{:reserved=>0, :extended_response=>1, :extended_signature=>0, :reserved2=>0, :disconnect=>0, :reserved3=>0, :reserved4=>0}, :password_length=>1}, :data_block=>{:byte_count=>8, :password=>"", :path=>"", :service=>"?????"}}
2.3.3 :002 > packet.parameter_block
=> {:word_count=>4, :andx_block=>{:andx_command=>255, :andx_reserved=>0, :andx_offset=>0}, :flags=>{:reserved=>0, :extended_response=>1, :extended_signature=>0, :reserved2=>0, :disconnect=>0, :reserved3=>0, :reserved4=>0}, :password_length=>1}
2.3.3 :003 > packet.parameter_block.flags
=> {:reserved=>0, :extended_response=>1, :extended_signature=>0, :reserved2=>0, :disconnect=>0, :reserved3=>0, :reserved4=>0}
2.3.3 :004 > packet.parameter_block.flags.extended_signature = 1
=> 1
2.3.3 :005 > packet.parameter_block.flags
=> {:reserved=>0, :extended_response=>1, :extended_signature=>1, :reserved2=>0, :disconnect=>0, :reserved3=>0, :reserved4=>0}
2.3.3 :006 >
2.3.3 :006 > packet.data_block.password = 'guest'
=> "guest"
2.3.3 :007 > packet.data_block.password
=> "guest"
2.3.3 :008 > packet.data_block
=> {:byte_count=>13, :password=>"guest", :path=>"", :service=>"?????"}
2.3.3 :009 >
You can also pass field/value pairs into the packet constructor as arguments, defaulting individual fields as desired.
Example:
2.3.3 :017 > packet = RubySMB::SMB2::Packet::TreeConnectRequest.new(path:'test')
=> {:smb2_header=>{:protocol=>4266872130, :structure_size=>64, :credit_charge=>0, :nt_status=>0, :command=>0, :credits=>0, :flags=>{:reserved3=>0, :signed=>0, :related_operations=>0, :async_command=>0, :reply=>0, :reserved2=>0, :reserved1=>0, :replay_operation=>0, :dfs_operation=>0}, :next_command=>0, :message_id=>0, :process_id=>65279, :tree_id=>0, :session_id=>0, :signature=>"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"}, :structure_size=>9, :flags=>0, :path_offset=>72, :path_length=>4, :path=>"test"}
2.3.3 :018 > packet.path
=> "test"
Sometimes you need to read a binary data and apply one of the packet structures to it. For example, when you are reading a response packet, you will need to read the raw response string into an actual packet class. This is done using the #read class method.
2.3.3 :014 > blob = "\xFFSMB+\x00\x00\x00\x00\x98\x01`\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01\x00\x00\x00\x00"
=> "\xFFSMB+\u0000\u0000\u0000\u0000\x98\u0001`\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0001\u0000\u0000\u0000\u0000"
2.3.3 :015 > packet = RubySMB::SMB1::Packet::EchoResponse.read(blob)
=> {:smb_header=>{:protocol=>4283649346, :command=>43, :nt_status=>0, :flags=>{:reply=>1, :opbatch=>0, :oplock=>0, :canonicalized_paths=>1, :case_insensitive=>1, :reserved=>0, :buf_avail=>0, :lock_and_read_ok=>0}, :flags2=>{:reserved1=>0, :is_long_name=>0, :reserved2=>0, :signature_required=>0, :compressed=>0, :security_signature=>0, :eas=>0, :long_names=>1, :unicode=>0, :nt_status=>1, :paging_io=>1, :dfs=>0, :extended_security=>0, :reparse_path=>0}, :pid_high=>0, :security_features=>"\x00\x00\x00\x00\x00\x00\x00\x00", :reserved=>0, :tid=>0, :pid_low=>0, :uid=>0, :mid=>0}, :parameter_block=>{:word_count=>1, :sequence_number=>0}, :data_block=>{:byte_count=>0, :data=>""}}
2.3.3 :016 >
Any structure or packet in RubySMB can also be converted back into a binary blob using BinData's #to_binary_s method.
Example:
2.3.3 :012 > packet = RubySMB::SMB1::Packet::EchoResponse.new
=> {:smb_header=>{:protocol=>4283649346, :command=>43, :nt_status=>0, :flags=>{:reply=>1, :opbatch=>0, :oplock=>0, :canonicalized_paths=>1, :case_insensitive=>1, :reserved=>0, :buf_avail=>0, :lock_and_read_ok=>0}, :flags2=>{:reserved1=>0, :is_long_name=>0, :reserved2=>0, :signature_required=>0, :compressed=>0, :security_signature=>0, :eas=>0, :long_names=>1, :unicode=>0, :nt_status=>1, :paging_io=>1, :dfs=>0, :extended_security=>0, :reparse_path=>0}, :pid_high=>0, :security_features=>"\x00\x00\x00\x00\x00\x00\x00\x00", :reserved=>0, :tid=>0, :pid_low=>0, :uid=>0, :mid=>0}, :parameter_block=>{:word_count=>1, :sequence_number=>0}, :data_block=>{:byte_count=>0, :data=>""}}
2.3.3 :013 > packet.to_binary_s
=> "\xFFSMB+\x00\x00\x00\x00\x98\x01`\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01\x00\x00\x00\x00"
Sitting on top of the packet layer in RubySMB is the RubySMB::Client class. This is the abstraction that most users of RubySMB will interact with. It provides simple convenience methods for performing SMB actions. It handles the creation, sending and receiving of packets for the user, providing reasonable defaults in many cases.
The RubySMB client is capable of multi-protocol negotiation. The user simply specifies whether SMB1 and/or SMB2 should be supported, and the client negotiates the protocol and dialect behind the scenes.
In the following example, we tell the client that both SMB1 and SMB2 should be supported. The client will then negotiate with the server which version should be used.
Negotiation Example:
sock = TCPSocket.new address, 445
dispatcher = RubySMB::Dispatcher::Socket.new(sock)
client = RubySMB::Client.new(dispatcher, username: 'msfadmin', password: 'msfadmin')
client.negotiate
RubySMB uses the Ruby NTLM gem for authentication. While the client will not currently attempt older basic authentication on its own, it will attempt an anonymous login if no user credentials are supplied.
Authenticated Example:
sock = TCPSocket.new address, 445
dispatcher = RubySMB::Dispatcher::Socket.new(sock)
client = RubySMB::Client.new(dispatcher, username: 'msfadmin', password: 'msfadmin')
client.negotiate
client.authenticate
Anonymous Example:
sock = TCPSocket.new address, 445
dispatcher = RubySMB::Dispatcher::Socket.new(sock)
client = RubySMB::Client.new(dispatcher, username: '', password: '')
client.negotiate
client.authenticate
While there is one RubySMB::Client object that supports both SMB1 and SMB2, once the library connects to an SMB tree, it returns a protocol-specific RubySMB::Tree object. This Tree object executes all subsequent file operations on the tree.
In the below example we see a simple script to connect to a remote tree, and list all files in a given sub-directory.
Example:
sock = TCPSocket.new address, 445
dispatcher = RubySMB::Dispatcher::Socket.new(sock)
client = RubySMB::Client.new(dispatcher, username: 'msfadmin', password: 'msfadmin')
client.negotiate
client.authenticate
begin
tree = client.tree_connect('TEST_SHARE')
puts "Connected to #{path} successfully!"
rescue StandardError => e
puts "Failed to connect to #{path}: #{e.message}"
end
files = tree.list(directory: 'subdir1')
files.each do |file|
create_time = file.create_time.to_datetime.to_s
access_time = file.last_access.to_datetime.to_s
change_time = file.last_change.to_datetime.to_s
file_name = file.file_name.encode("UTF-8")
puts "FILE: #{file_name}\n\tSIZE(BYTES):#{file.end_of_file}\n\tSIZE_ON_DISK(BYTES):#{file.allocation_size}\n\tCREATED:#{create_time}\n\tACCESSED:#{access_time}\n\tCHANGED:#{change_time}\n\n"
end
It is useful to have Wireshark and a reference SMB client, such as Impacket's installed to help debug and compare output:
Configure Wireshark in Debian-based systems to be able to capture traffic without root user privileges:
sudo apt-get install wireshark
sudo dpkg-reconfigure wireshark-common
sudo addgroup wireshark
sudo usermod -a -G wireshark <USERNAME>
sudo apt-get install python-setuptools
sudo easy_install pyasn1 pycrypto
- Download from GitHub (https://github.com/coresecurity/impacket)
sudo python setup.py install
cd examples && python smbclient.py <USER>:<PASS>@<WINDOWS HOST IP>
In situations where WireShark reports some requests/responses as malformed (not parsed correctly),
Microsoft Network Monitor can be used instead.
For example, the LookupSids
response is not parsed correctly by WireShark, whereas it is by this tool.
This software can be installed on a Windows machine:
- Download & install the software
- Open it
- Click the
Start
button (or press F5) present at the top bar
The SMB requests will be present under All Traffic -> My Traffic -> System (4)
.
ruby_smb
is released under a 3-clause BSD license. See LICENSE.txt for full text.
- Fork it ( https://github.com/rapid7/ruby_smb/fork )
- Create your feature branch (
git checkout -b my-new-feature
) - Commit your changes (
git commit -am 'Add some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create a new Pull Request