Firewall knock operator
fwknop -A <'proto/ports'> -R|-a|-s -D <'host'> [options]
fwknop implements an authorization scheme known as Single Packet Authorization (SPA) for strong service concealment. SPA requires only a single packet which is encrypted, non-replayable, and authenticated via an HMAC in order to communicate desired access to a service that is hidden behind a firewall in a default-drop filtering stance. The main application of SPA is to use a firewall to drop all attempts to connect to services such as SSH in order to make the exploitation of vulnerabilities (both 0-day and unpatched code) more difficult. Any service that is concealed by SPA naturally cannot be scanned for with Nmap. The fwknop project supports three different firewalls: iptables on Linux systems, pf on OpenBSD, and ipfw on FreeBSD and Mac OS X.
SPA is essentially next generation Port Knocking (PK), but solves many of the limitations exhibited by PK while retaining its core benefits. PK limitations include a general difficulty in protecting against replay attacks, asymmetric ciphers and HMAC schemes are not usually possible to reliably support, and it is trivially easy to mount a DoS attack against a PK server just by spoofing an additional packet into a PK sequence as it traverses the network (thereby convincing the PK server that the client doesn\(cqt know the proper sequence). All of these limitation are solved by SPA. At the same time, SPA hides services behind a default-drop firewall policy, acquires SPA data passively (usually via libpcap or other means), and implements standard cryptographic operations for SPA packet authentication and encryption/decryption.
This is the manual page for the fwknop client which is responsible for constructing SPA packets and sending them over the network. The server side is implemented by the fwknopd daemon which sniffs the network for SPA packets and interacts with the local firewall to allow SPA authenticated connections. It is recommended to read the fwknopd(8) manual page as well.
SPA packets generated by fwknop leverage HMAC for authenticated encryption in the encrypt-then-authenticate model. Although the usage of an HMAC is currently optional (enabled via the --use-hmac command line switch), it is highly recommended for three reasons: 1) without an HMAC, cryptographically strong authentication is not possible with fwknop unless GnuPG is used, but even then an HMAC should still be applied, 2) an HMAC applied after encryption protects against cryptanalytic CBC-mode padding oracle attacks such as the Vaudenay attack and related trickery (like the more recent "Lucky 13" attack against SSL), and 3) the code required by the fwknopd daemon to verify an HMAC is much more simplistic than the code required to decrypt an SPA packet, so an SPA packet without a proper HMAC isn\(cqt even sent through the decryption routines. Reason 3) is why an HMAC should still be used even when SPA packets are encrypted with GnuPG due to the fact that SPA data is not sent through libgpgme functions unless the HMAC checks out first. GnuPG and libgpgme are relatively complex bodies of code, and therefore limiting the ability of a potential attacker to interact with this code through an HMAC operation helps to maintain a stronger security stance. Generating an HMAC for SPA communications requires a dedicated key in addition to the normal encryption key, and both can be generated with the --key-gen option.
fwknop encrypts SPA packets either with the Rijndael block cipher or via GnuPG and associated asymmetric cipher. If the symmetric encryption method is chosen, then as usual the encryption key is shared between the client and server (see the fwknopd /etc/fwknop/access.conf file for details). The actual encryption key used for Rijndael encryption is generated via the standard PBKDF1 key derivation algorithm, and CBC mode is set. If the GnuPG method is chosen, then the encryption keys are derived from GnuPG key rings. SPA packets generated by fwknop running as a client adhere to the following format (before encryption and the HMAC is applied):
random data (16 digits) username timestamp software version mode (command mode (0) or access mode (1)) if command mode => command to execute else access mode => IP,proto,port message digest (SHA512 / SHA384 / SHA256 / SHA1 / MD5)
Each of the above fields are separated by a ":" character due to the variable length of several of the fields, and those that might contain ":" characters are base64 encoded. The message digest (SHA256 by default) is part of the data to be encrypted and is independent of the HMAC which is appended to the SPA packet data after encryption. The 16 digits of random data (about 53 bits) ensures that no two SPA packets are identical, and this is in addition to and independent of using PBKDF1 for key derivation for Rijndael in CBC mode (which uses an 8-byte random "salt" value). Because fwknopd tracks the SHA256 digest of all incoming valid SPA packets and throws out duplicates, replay attacks are not feasible against fwknop. Syslog alerts are generated if a replay is detected.
By default, the fwknop client sends authorization packets over UDP port 62201, but this can be altered with the --server-port argument (this requires fwknopd to be configured to acquire SPA data over the selected port). Also, fwknop can send the SPA packet over a random port via the --rand-port argument. See fwknopd(8) for further details. See the EXAMPLES section for example invocations of the fwknop client.
The fwknop client is quite portable, and is known to run on various Linux distributions (all major distros and embedded ones such as OpenWRT as well), FreeBSD, OpenBSD, and Cygwin on Windows. There is also a library libfko that both fwknop and fwknopd use for SPA packet encryption/decryption and HMAC authentication operations. This library can be used to allow third party applications to use SPA subject to the terms of the GNU General Public License (GPL).
These required arguments can be specified via command-line or from within the ~/.fwknoprc file (see -n, --named-config option and the FWKNOPRC FILE section below).
-A, --access=<port list>
Provide a list of ports and protocols to access on a remote computer running fwknopd. The format of this list is “+<proto>/<port>...<proto>/<port>+”, e.g. “tcp/22,udp/53”. NOTE: The vast majority of usages for fwknop require the -A argument, but sending full commands with the --server-cmd argument via an SPA packet to be executed by fwknopd does not require this argument.
Direct the fwknop client to authenticate with the fwknopd daemon/service at the specified destination hostname or IP address. The connection mode is discovered by the fwknopd daemon/service when it decrypts and parses the authentication packet.
One of these options (see below) is required to tell the remote fwknopd daemon what IP should be allowed through the local firewall. It is recommend to use the -R or -a options instead of -s in order to harden SPA communications against possible Man-In-The-Middle (MITM) attacks, and on the server side set REQUIRE_SOURCE_ADDRESS variable in the /etc/fwknop/access.conf file. Note that the most secure option is -a so that fwknop does not have to issue an HTTP request to http://www.cipherdyne.org/cgi-bin/myip in order to resolve the externally routable IP address. Using -a requires that the user already knows what the external IP is for the network where fwknop is running.
Print a usage summary message and exit.
Load an encryption key/password from the specified file. The key file contains a line for each destination hostname or IP address, a colon (":"), optional space and the password, followed by a newline. Note that the last line has to have a terminating newline character. Also note: though this is a convenience, having a file on your system with clear text passwords is not a good idea and is not recommended. Having the fwknop client prompt you for the key is generally more secure. Note also that if a key is stored on disk, the fwknop rc file is a more powerful mechanism for specifying not only the key but other options as well.
Read the encryption key/password from stdin. This can be used to send the data via a pipe for example. This command is similar to --fd 0.
Specify the file descriptor number to read the key/password from. This command avoids the user being prompted for a password if none has been found in the user specific stanza, or none has been supplied on the command line. A file descriptor set to 0 is similar to the stdin command.
Load an HMAC key/password from the specified file. Similarly to the format for the --get-key option, the HMAC key file contains a line for each destination hostname or IP address, a colon (":"), optional space and the password, followed by a newline. Note that the last line has to have a terminating newline character. Also note: though this is a convenience, having a file on your system with clear text passwords is not a good idea and is not recommended. Having the fwknop client prompt you for the HMAC key is generally more secure. Note also that if a key is stored on disk, the fwknop rc file is a more powerful mechanism for specifying not only the HMAC key but other options as well.
Have fwknop generate both Rijndael and HMAC keys that can be used for SPA packet encryption. These keys are derived from /dev/urandom and then base64 encoded before being printed to stdout, and are meant to be included within the “$HOME/.fwknoprc” file (or the file referenced by --get-key). Such keys are generally more secure than passphrases that are typed in from the command line.
Specify the number of bytes for a generated Rijndael key. The maximum size is currently 128 bytes.
Specify the number of bytes for a generated HMAC key. The maximum size is currently 128 bytes.
Execute fwknop with the command-line arguments from the previous invocation (if any). The previous arguments are parsed out of the ~/.fwknop.run file.
-n, --named-config=<stanza name>
Specify the name of the configuration stanza in the “$HOME/.fwknoprc” file to pull configuration and command directives. These named stanzas alleviate the need for remembering the various command-line arguments for frequently used invocations of fwknop. See the section labeled, FWKNOPRC FILE below for a list of the valid configuration directives in the .fwknoprc file.
Specify the Rijndael key on the command line. Since the key may be visible to utilities such as ps under Unix, this form should only be used where security is not critical. Having the fwknop client either prompt you for the key or acquire via the “$HOME/.fwknoprc” file is generally more secure.
Specify the base64 encoded Rijndael key. Since the key may be visible to utilities such as ps under Unix, this form should only be used where security is not critical. Having the fwknop client either prompt you for the key or acquire via the “$HOME/.fwknoprc” file is generally more secure.
Specify the base64 encoded HMAC key. Since the key may be visible to utilities such as ps under Unix, this form should only be used where security is not critical. Having the fwknop client either prompt you for the key or acquire via the “$HOME/.fwknoprc” file is generally more secure.
Specify the raw HMAC key (not base64 encoded). Since the key may be visible to utilities such as ps under Unix, this form should only be used where security is not critical. Having the fwknop client either prompt you for the key or acquire via the “$HOME/.fwknoprc” file is generally more secure.
Specify path to the fwknop rc file (default is $HOME/.fwknoprc).
Save command line arguments to the $HOME/.fwknoprc stanza specified with the -n option. If the -n option is omitted, then the stanza name will default to the destination server value (hostname or IP) given with the -D argument.
Used with --save-rc-stanza to overwrite all of the variables for the specified stanza
Dump a list of the stanzas found in “$HOME/.fwknoprc”.
Display the last command-line arguments used by fwknop.
Save command line arguments to a specified file path. Without this option, and when --no-save-args is not also specified, then the default save args path is ~/.fwknop.run.
Do not save the command line arguments given when fwknop is executed.
Test mode. Generate the SPA packet data, but do not send it. Instead, print a break-down of the SPA data fields, then run the data through the decryption and decoding process and print the break-down again. This is primarily a debugging feature.
Instruct the fwknop client to write a newly created SPA packet out to the specified file so that it can be examined off-line.
Append the generated packet data to the file specified with the -B option.
Run the fwknop client in verbose mode. This causes fwknop to print some extra information about the current command and the resulting SPA data.
Display version information and exit.
Set HMAC mode for authenticated encryption of SPA communications. As of fwknop 2.5, this is an optional feature, but this will become the default in a future release.
Specify IP address that should be permitted through the destination fwknopd server firewall (this IP is encrypted within the SPA packet itself). This is useful to prevent a MITM attack where a SPA packet can be intercepted en-route and sent from a different IP than the original. Hence, if the fwknopd server trusts the source address on the SPA packet IP header then the attacker gains access. The -a option puts the source address within the encrypted SPA packet, and so thwarts this attack. The -a option is also useful to specify the IP that will be granted access when the SPA packet itself is spoofed with the --spoof-src option. Another related option is -R (see below) which instructs the fwknop client to automatically resolve the externally routable IP address the local system is connected to by querying a website that returns the actual IP address it sees from the calling system.
Use GPG encryption on the SPA packet (default if not specified is Rijndael). Note: Use of this option will also require a GPG recipient (see --gpg-recipient along with other GPG-related options below).
Set the HMAC digest algorithm for authenticated encryption of SPA packets. Choices are: MD5, SHA1, SHA256 (the default), SHA384, and SHA512.
The fwknopd server offers the ability to provide SPA access through an iptables firewall to an internal service by interfacing with the iptables NAT capabilities. So, if the fwknopd server is protecting an internal network on an RFC-1918 address space, an external fwknop client can request that the server port forward an external port to an internal IP, i.e. “+--NAT-access 192.168.10.2,55000+”. In this case, access will be granted to 192.168.10.2 via port 55000 to whatever service is requested via the --access argument (usually tcp/22). Hence, after sending such an SPA packet, one would then do “ssh -p 55000 [email protected]” and the connection would be forwarded on through to the internal 192.168.10.2 system automatically. Note that the port “55000” can be randomly generated via the --nat-rand-port argument (described later).
On the fwknopd server, a NAT operation can apply to the local system instead of being forwarded through the system. That is, for iptables firewalls, a connection to, say, port 55,000 can be translated to port 22 on the local system. By making use of the --nat-local argument, the fwknop client can be made to request such access. This means that any external attacker would only see a connection over port 55,000 instead of the expected port 22 after the SPA packet is sent.
Usually fwknop is used to request access to a specific port such as tcp/22 on a system running fwknopd. However, by using the --nat-port argument, it is possible to request access to a (again, such as tcp/22), but have this access granted via the specified port (so, the -p argument would then be used on the SSH client command line). See the --nat-local and --nat-access command line arguments to fwknop for additional details on gaining access to services via a NAT operation.
Usually fwknop is used to request access to a specific port such as tcp/22 on a system running fwknopd. However, by using the --nat-rand-port argument, it is possible to request access to a particular service (again, such as tcp/22), but have this access granted via a random translated port. That is, once the fwknop client has been executed in this mode and the random port selected by fwknop is displayed, the destination port used by the follow-on client must be changed to match this random port. For SSH, this is accomplished via the -p argument. See the --nat-local and --nat-access command line arguments to fwknop for additional details on gaining access to services via a NAT operation.
Specify the port number where fwknopd accepts packets via libpcap or ulogd pcap writer. By default fwknopd looks for authorization packets over UDP port 62201.
Set the protocol (udp, tcp, http, udpraw, tcpraw, or icmp) for the outgoing SPA packet. Note: The udpraw, tcpraw, and icmp modes use raw sockets and thus require root access to run. Also note: The tcp mode expects to establish a TCP connection to the server before sending the SPA packet. This is not normally done, but is useful for compatibility with the Tor for strong anonymity; see http://tor.eff.org/. In this case, the fwknopd server will need to be configured to listen on the target TCP port (which is 62201 by default).
Spoof the source address from which the fwknop client sends SPA packets. This requires root on the client side access since a raw socket is required to accomplish this. Note that the --spoof-user argument can be given in this mode in order to pass any REQUIRE_USERNAME keyword that might be specified in /etc/fwknop/access.conf.
Instruct the fwknop client to send an SPA packet over a random destination port between 10,000 and 65535. The fwknopd server must use a PCAP_FILTER variable that is configured to accept such packets. For example, the PCAP_FILTER variable could be set to: “+udp dst portrange 10000-65535+”.
This is an important option, and instructs the fwknop client and the fwknopd daemon/service to query a web server that returns the caller\(cqs IP address (as seen by the web server). In some cases, this is needed to determine the IP address that should be allowed through the iptables policy at the remote fwknopd server side. This is useful if the fwknop client is being used on a system that is behind an obscure NAT address. Presently, fwknop uses the URL: http://www.cipherdyne.org/cgi-bin/myip to resolve the caller IP. Note that it is generally more secure to use the -a option if the externally routable IP address for the client is already known to the user since this elminates the need for fwknop to issue an HTTP request.
Override the default URL used for resolving the source IP address. For best results, the URL specified here should point to a web service that provides just an IP address in the body of the HTTP response.
Instruct the fwknop client to form an SPA packet that contains the special-case IP address “+0.0.0.0+” which will inform the destination fwknopd SPA server to use the source IP address from which the SPA packet originates as the IP that will be allowed through upon modification of the firewall ruleset. This option is useful if the fwknop client is deployed on a machine that is behind a NAT device and the external IP is not known. However, usage of this option is not recommended, and either the -a or -R options should be used instead. The permit-address options -s, -R and -a are mutually exclusive.
Set the source port for outgoing SPA packet.
Specify the length of time (seconds) that the remote firewall rule that grants access to a service is to remain active. The default maintained by fwknopd is 30 seconds, but any established connection can be kept open after the initial accept rule is deleted through the use of a connection tracking mechanism that may be offered by the firewall.
-C, --server-cmd=<command to execute>
Instead of requesting access to a service with an SPA packet, the --server-cmd argument specifies a command that will be executed by the fwknopd server. The command is encrypted within the SPA packet and sniffed off the wire (as usual) by the fwknopd server.
Specify an HTTP proxy that the fwknop client will use to send the SPA packet through. Using this option will automatically set the SPA packet transmission mode (usually set via the --server-proto argument) to "http". You can also specify the proxy port by adding ":<port>" to the proxy host name or ip.
Specify the message digest algorithm to use in the SPA data. Choices are: MD5, SHA1, SHA256 (the default), SHA384, and SHA512.
Specify the encryption mode when AES is used for encrypting SPA packets. The default is CBC mode, but others can be chosen such as CFB or OFB as long as this is also specified in the /etc/fwknop/access.conf file on the server side via the ENCRYPTION_MODE variable. In general, it is recommended to not include this argument and let the default (CBC) apply. Note that the string “legacy” can be specified in order to generate SPA packets with the old initialization vector strategy used by versions of fwknop prior to 2.5. With the 2.5 release, fwknop generates initialization vectors in a manner that is compatible with OpenSSL via the PBKDF1 algorithm.
By default, the fwknopd daemon on the server side enforces time synchronization between the clocks running on client and server systems. The fwknop client places the local time within each SPA packet as a time stamp to be validated by the fwknopd server after decryption. However, in some circumstances, if the clocks are out of sync and the user on the client system does not have the required access to change the local clock setting, it can be difficult to construct and SPA packet with a time stamp the server will accept. In this situation, the --time-offset-plus option can allow the user to specify an offset (e.g. “60sec” “60min” “2days” etc.) that is added to the local time.
This is similar to the --time-offset-plus option (see above), but subtracts the specified time offset instead of adding it to the local time stamp.
Set the HTTP User-Agent for resolving the external IP via -R, or for sending SPA packets over HTTP.
Specify the username that is included within SPA packet. This allows the fwknop client to satisfy any non-root REQUIRE_USERNAME keyword on the fwknopd server (--spoof-src mode requires that the fwknop client is executed as root).
In -P icmp mode, specify the ICMP type value that will be set in the SPA packet ICMP header. The default is echo reply.
In -P icmp mode, specify the ICMP code value that will be set in the SPA packet ICMP header. The default is zero.
Note that the usage of GPG for SPA encryption/decryption can and should involve GPG keys that are signed by each side (client and server). The basic procedure for this involves the following steps after the client key has been transferred to the server and vice-versa:
[spaserver]# gpg --import client.asc [spaserver]# gpg --edit-key 1234ABCD Command> sign [spaclient]$ gpg --import server.asc [spaclient]$ gpg --edit-key ABCD1234 Command> sign
More comprehensive information on this can be found here: http://www.cipherdyne.org/fwknop/docs/gpghowto.html.
Instruct fwknop to acquire GnuPG key password from a running gpg-agent instance (if available).
Specify the path to the GnuPG directory; normally this path is derived from the home directory of the user that is running the fwknop client (so the default is ~/.gnupg). This is useful when a “root” user wishes to log into a remote machine whose sshd daemon/service does not permit root login.
--gpg-recipient=<key ID or Name>
Specify the GnuPG key ID, e.g. “+1234ABCD+” (see the output of "gpg–list-keys") or the key name (associated email address) of the recipient of the Single Packet Authorization message. This key is imported by the fwknopd server and the associated private key is used to decrypt the SPA packet. The recipient\(cqs key must first be imported into the client GnuPG key ring.
--gpg-signer-key=<key ID or Name>
Specify the GnuPG key ID, e.g. “+ABCD1234+” (see the output of "gpg --list-keys") or the key name to use when signing the SPA message. The user is prompted for the associated GnuPG password to create the signature. This adds a cryptographically strong mechanism to allow the fwknopd daemon on the remote server to authenticate who created the SPA message.
Instruct fwknop to not acquire a passphrase for usage of GnuPG signing key. This option is provided to make SPA packet construction easier for client-side operations in automated environments where the passphrase for the signing key has been removed from the GnuPG key ring. However, it is usually better to leverage gpg-agent instead of using this option.
The .fwknoprc file is used to set various parameters to override default program parameters at runtime. It also allows for additional named configuration stanzas for setting program parameters for a particular invocation.
The fwknop client will create this file if it does not exist in the user\(cqs home directory. This initial version has some sample directives that are commented out. It is up to the user to edit this file to meet their needs.
The .fwknoprc file contains a default configuration area or stanza which holds global configuration directives that override the program defaults. You can edit this file and create additional named stanzas that can be specified with the -n or --named-config option. Parameters defined in the named stanzas will override any matching default stanza directives. Note that command-line options will still override any corresponding .fwknoprc directives.
There are directives to match most of the command-line parameters fwknop supports. Here is the current list of each directive along with a brief description and its matching command-line option(s):
Specify the hostname or IP of the destination (fwknopd) server (-D, --destination).
Specify the address to allow within the SPA data. Note: This parameter covers the -a, -s, and -R command-line options. You can specify a hostname or IP address (the -a option), specify the word "source" to tell the fwknopd server to accept the source IP of the packet as the IP to allow (the -s option), or use the word "resolve" to have fwknop resolve the external network IP via HTTP request (the -R option).
ACCESS <port list>
Set the one or more protocol/ports to open on the firewall (-A, --access). The format of this list is “+<proto>/<port>...<proto>/<port>+”, e.g. “tcp/22,udp/53”.
Set the server port to use for sending the SPA packet (-p, --server-port).
Set the protocol to use for sending the SPA packet (-P, --server-proto).
This is the passphrase that is used for SPA packet encryption and applies to both Rijndael or GPG encryption modes. The actual encryption key that is used for Rijndael is derived from the PBKDF1 algorithm, and the GPG key is derived from the specified GPG key ring.
KEY_BASE64 <base64 encoded passphrase>
Specify the encryption passphrase as a base64 encoded string. This allows non-ascii characters to be included in the base64-decoded key.
Set HMAC mode for authenticated encryption of SPA packets. This will have fwknop prompt the user for a dedicated HMAC key that is independent of the encryption key. Alternatively, the HMAC key can be specified with the HMAC_KEY or HMAC_KEY_BASE64 directives (see below).
Specify the HMAC key for authenticated encryption of SPA packets. This supports both Rijndael and GPG encryption modes, and is applied according to the encrypt-then-authenticate model.
HMAC_KEY_BASE64 <base64 encoded key>
Specify the HMAC key as a base64 encoded string. This allows non-ascii characters to be included in the base64-decoded key.
HMAC_DIGEST_TYPE <digest algorithm>
Set the HMAC digest algorithm used for authenticated encryption of SPA packets. Choices are: MD5, SHA1, SHA256 (the default), SHA384, and SHA512.
Set the source port to use for sending the SPA packet (-S, --source-port).
Set the firewall rule timeout value (-f, --fw-timeout).
Set to a URL that will be used for resolving the source IP address (--resolve-url).
Set a value to apply to the timestamp in the SPA packet. This can be either a positive or negative value (--time-offset-plus/minus).
Specify the encryption mode when AES is used. This variable is a synonym for the -M, --encryption-mode command line argument. In general, it is recommended to not include this argument and let the default (CBC) apply. Note that the string “legacy” can be specified in order to generate SPA packets with the old initialization vector strategy used by versions of fwknop prior to 2.5.
DIGEST_TYPE <digest algorithm>
Set the SPA message digest type (-m, --digest-type). Choices are: MD5, SHA1, SHA256 (the default), SHA384, and SHA512.
Set to Y to specify the use of GPG for encryption (--gpg-encryption).
Set to Y to have fwknop interface with a GPG agent instance for the GPG key password (--gpg-agent). Agent information itself is specified with the GPG_AGENT_INFO environmental variable.
This is the passphrase that is used for signing SPA packet data in GPG encryption mode, and is a synonym for the KEY variable (i.e. the signing passphrase can be specified with the KEY variable instead). The SPA packet is encrypted with the remote server key and signed with the local client key.
GPG_SIGNING_PW_BASE64 <base64 encoded passphrase>
Specify the GPG signing passphrase as a base64 encoded string. This allows non-ascii characters to be included in the base64-decoded key.
GPG_SIGNER <key ID or Name>
Specify the GPG key name or ID for signing the GPG-encrypted SPA data (--gpg-signer-key).
GPG_RECIPIENT <key ID or Name>
Specify the GPG key name or ID for the recipient of the GPG-encrypted SPA data (--gpg-recipient-key).
Specify the GPG home directory (--gpg-home-dir). Defaults to ~/.gnupg.
Set the username in the SPA data to the specified value (-U, --spoof-user).
Set the source IP of the outgoing SPA packet to the specified value (-Q, --spoof-source).
Send the SPA packet over a randomly assigned port (-r, --rand-port).
Load an encryption key/password from a file (-G, --get-key).
HTTP_USER_AGENT <agent string>
Set the HTTP User-Agent for resolving the external IP via -R, or for sending SPA packets over HTTP (-u, --user-agent).
Gain NAT access to an internal service protected by the fwknop server (-N, --nat-access).
Access a local service via a forwarded port on the fwknopd server system (--nat-local).
Specify the port to forward to access a service via NAT (--nat-port).
Have the fwknop client assign a random port for NAT access (--nat-rand-port).
SPOOF_USER, GPG_AGENT_INFO (only used in --gpg-agent mode).
Because fwknop places the IP to be allowed through the firewall within the encrypted SPA payload (unless -s is used which is not recommended and can be prohibited in the fwknopd server configuration), SPA packets can easily be spoofed, and this is a good thing in this context. That is, the source IP of an SPA packet is ignored by the fwknopd daemon (when the REQUIRE_SOURCE_ADDRESS variable is set in the /etc/fwknop/access.conf file) and only the IP that is contained within an authenticated and properly decrypted SPA packet is granted access through the firewall. This makes it possible to make it appear as though, say, www.yahoo.com is trying to authenticate to a target system but in reality the actual connection will come from a seemingly unrelated IP.
The following examples illustrate the command line arguments that could be supplied to the fwknop client in a few situations:
The most effective and easiest way to use fwknop is to have the client generate both an encryption key and an HMAC key, and then save them to the “$HOME/.fwknoprc” file along with access request specifics. The keys will also need to be transferred to the system where fwknopd is running. The also client supports a separate set of encryption and HMAC keys for each SPA destination if multiple fwknopd servers are running on different systems.
So, assuming that the IP 220.127.116.11 is the system where fwknopd is deployed and SSH is protected by the firewall on that system in a default-drop stance, run the client like so to generate encryption and HMAC keys:
$ fwknop -A tcp/22 --use-hmac -R -D 18.104.22.168 --key-gen --save-rc-stanza --verbose [+] Wrote Rijndael and HMAC keys to rc file: /home/user/.fwknoprc
With the access request arguments and encryption and HMAC keys generated and saved in “$HOME/.fwknoprc”, the keys themselves need to be transferred to the 22.214.171.124 system where fwknopd is running. As always, this should be done via some secure means such as SSH before SPA is enabled and SSHD is blocked by the firewall. Here is what the new 126.96.36.199 stanza looks like in the ~/.fwknoprc file:
$ tail -n 8 /home/user/.fwknoprc [188.8.131.52] ACCESS tcp/22 SPA_SERVER 184.108.40.206 KEY_BASE64 HvUtIOramehLGKimD4ECXOzinaH4h3U8H1WXum7b54Q= HMAC_KEY_BASE64 DLeLf93a3yBT2vhEpM+dWlirGta5GU+jdyG5uXp4461HgOtbqMem4gX0Bp2PJGzYZlbbcavcOM00UPm+0GqkXA== USE_HMAC Y VERBOSE Y RESOLVE_IP_HTTP Y
The keys are base64 encoded blobs of random data, and both the KEY_BASE64 and HMAC_KEY_BASE64 lines should be copied verbatim and placed within the /etc/fwknop/access.conf file on 220.127.116.11. Once this is done, fwknopd can be started on that system, a default-drop policy against SSH connections can be put in place, and then access to SSH is managed via fwknop. To access SSH, just use the -n argument to reference the 18.104.22.168 stanza out of the .fwknoprc file (some --verbose output is included for illustration):
$ fwknop -n 22.214.171.124 FKO Field Values: ================= Random Value: 8950423288486978 Username: mbr Timestamp: 1370194770 FKO Version: 2.5 Message Type: 1 (Access msg) Message String: 126.96.36.199,tcp/22 Nat Access: <NULL> Server Auth: <NULL> Client Timeout: 0 (seconds) Digest Type: 3 (SHA256) HMAC Type: 3 (SHA256) Encryption Type: 1 (Rijndael) Encryption Mode: 2 (CBC) ... $ ssh -l user 188.8.131.52 [email protected]'s password:
The most common usage of fwknop is to gain access to SSH running on a remote system that has the fwknopd daemon deployed along with a default-drop firewall policy. The following command illustrates this where IP 184.108.40.206 is the IP to be allowed through the firewall running on 220.127.116.11 (note that the /etc/fwknop/access.conf file consumed by fwknopd will need to have matching encryption and HMAC keys, and configuration specifics can be found in the fwknopd(8) manual page). Also, note the examples below prompt the user to supply the encryption and HMAC keys via stdin instead of writing them to disk as in the case of using the “$HOME/.fwknoprc” file in the example above. However, all of the following examples can be converted to using the ~/.fwknoprc file just by adding the --save-rc-stanza argument:
$ fwknop -A tcp/22 --use-hmac -a 18.104.22.168 -D 22.214.171.124 Enter encryption key: Enter HMAC key: $ ssh -l user 126.96.36.199 [email protected]'s password:
If the --verbose flag is added to the command line, then some SPA packet specifics are printed to stdout (not all output is shown for brevity):
$ fwknop -A tcp/22 --use-hmac -a 188.8.131.52 -D 184.108.40.206 --verbose Enter encryption key: Enter HMAC key: Random Value: 1916307060193417 Username: mbr Timestamp: 1368498909 FKO Version: 2.5 Message Type: 1 (Access msg) Message String: 220.127.116.11,tcp/22 Nat Access: <NULL> Server Auth: <NULL> Client Timeout: 0 (seconds) Digest Type: 3 (SHA256) HMAC Type: 3 (SHA256) Encryption Type: 1 (Rijndael) Encryption Mode: 2 (CBC)
Simultaneous access to multiple services is also supported, and here is an example of requesting access to both SSH and OpenVPN on 18.104.22.168:
$ fwknop -A "tcp/22,tcp/1194" --use-hmac -a 22.214.171.124 -D 126.96.36.199
There are many cases where an fwknop client is deployed on a network behind a NAT device and the externally routable IP is not known to the user. In this case, use the IP resolution service available at http://www.cipherdyne.org/cgi-bin/myip via the -R command line switch in order to derive the external client IP address. This is a safer method of acquiring the client IP address than using the -s option mentioned earlier in this manual page because the source IP is put within the encrypted packet instead of having the fwknopd daemon grant the requested access from whatever IP address the SPA packet originates (i.e. using -s opens the possibility of a MITM attack):
$ fwknop -A tcp/22 --use-hmac -R -D 188.8.131.52
Use the Single Packet Authorization mode to gain access to SSH and this time use GnuPG keys to encrypt and decrypt:
$ fwknop -A tcp/22 --use-hmac --gpg-sign ABCD1234 --gpg--recipient 1234ABCD -R -D 184.108.40.206
Instruct the fwknop server running at 220.127.116.11 to allow 18.104.22.168 to connect to SSH, but spoof the authorization packet from an IP associated with www.yahoo.com (requires root on the fwknop client OS):
# fwknop --spoof-src "www.yahoo.com" -A tcp/22 --use-hmac -a 22.214.171.124 -D 126.96.36.199
When fwknopd is running on an iptables firewall with systems deployed behind it, it is possible to take advantage of the NAT capabilities offered by iptables in order to transparently reach systems behind the firewall via SPA. Here is an example where the fwknop client is used to gain access to SSH running on the non-routable IP 192.168.10.23 that is deployed on the network behind 188.8.131.52. In this case, the SSH connection made to 184.108.40.206 is translated via NAT to the 192.168.10.2 system automatically:
$ fwknop -A tcp/22 -N 192.168.10.2:22 -R -D 220.127.116.11
With the 2.5 release, fwknop underwent significant changes in its usage of cryptography including the addition of support for HMAC authenticated encryption for both Rijndael and GnuPG modes, ensuring the proper usage of PBKDF1 for key derivation when SPA packets are encrypted with Rijndael, and several bugs were fixed from previous versions of fwknop. In general, this implies that when Rijndael is used, SPA packets produced by the 2.5 release are incompatible with previous versions of fwknop. The GnuPG encryption mode is unaffected by these updates. However, even with Rijndael is used, backwards compatibility is supported through setting the legacy encryption mode with -M on the fwknop client command line and/or the ENCRYPTION_MODE variable in the /etc/fwknop/access.conf file. This way, a pre-2.5 server can decrypt SPA packets produced by a 2.5 and later client (set -M legacy), and a 2.5 and later server can decrypt SPA packets produced by pre-2.5 clients (set ENCRYPTION_MODE legacy in the access.conf file). Note that HMAC is only supported as of 2.5 and is an optional feature, so backwards compatibility is only for configurations that don\(cqt use an HMAC on either side. It is strongly recommended to upgrade all fwknop clients and servers to 2.5 and use the new HMAC mode for properly authenticated SPA communications. The backwards compatibility support is used to make it easier to upgrade clients and servers with a phased approach.
For emphasis, if the fwknopd server is upgraded to 2.5 (or later), but older clients cannot be upgraded at the same time, then for each SOURCE stanza in the /etc/fwknop/access.conf file, add the following line:
In addition, if the KEY variable has an encryption key longer than 16 bytes, it will need to be truncated to 16 characters in the access.conf file in order for pre-2.5 clients to work properly. This limitation is fixed in 2.5, and provides additional motiviation for upgrading all clients and servers to 2.5 or later.
Now, flipping the scenario around, if the fwknop clients are upgraded but the fwknopd server is still at a pre-2.5 version, then add the -M legacy argument to the fwknop command line:
$ fwknop -A tcp/22 -M legacy -R -D 18.104.22.168
The fwknop client requires libfko which is normally included with both source and binary distributions, and is a dedicated library developed by the fwknop project. Whenever the fwknopd server is used, libpcap is a required dependency. However, the upcoming 2.6 release will offer a UDP listener mode along with privilege separation support and will not require libpcap in this mode. In UDP listener mode, even though fwknopd binds to a UDP port, SPA packets are never acknowledged so from an attacker\(cqs perspective there is no difference between fwknopd sniffing the wire passively vs. listening on a UDP socket in terms of what can be scanned for.
For GPG functionality, GnuPG must also be correctly installed and configured along with the libgpgme library.
To take advantage of all of the authentication and access management features of the fwknopd daemon/service a functioning iptables, ipfw, or pf firewall is required on the underlying operating system.
The most comprehensive way to gain diagnostic information on fwknop is to run the test suite test-fwknop.pl script located in the test/ directory in the fwknop sources. The test suite sends fwknop through a large number of run time tests, has valgrind support, validates both SPA encryption and HMAC results against OpenSSL, and even has its own built in fuzzer for SPA communications. For more basic diagnostic information, fwknop can be executed with the -T (or --test) command line option. This will have fwknop simply create and print the SPA packet information, then run it through a decrypt/decode cycle and print it again. In addition, the --verbose command line switch is useful to see various SPA packet specifics printed to stdout.
More information on Single Packet Authorization can be found in the paper “Single Packet Authorization with fwknop” available at http://www.cipherdyne.org/fwknop/docs/SPA.html. A comprehensive tutorial on fwknop operations and theory can be found at http://www.cipherdyne.org/fwknop/docs/fwknop-tutorial.html. This tutorial also includes information about the design of fwknop that may be worth reading for those interested in why fwknop is different from other SPA implementations.
fwknop uses the git versioning system as its source code repository along with Github for tracking of issues and milestones:
$ git clone https://github.com/mrash/fwknop.git fwknop.git
Additional commentary on Single Packet Authorization can be found via Michael Rash\(cqs Twitter feed: http://twitter.com/michaelrash, @michaelrash
This “C” version of fwknop was derived from the original Perl-based version on which many people who are active in the open source community have contributed. See the CREDITS file in the fwknop sources, or visit http://www.cipherdyne.org/fwknop/docs/contributors.html to view the online list of contributors. A few contributors deserve to be singled out including: Franck Joncourt, Max Kastanas, Vlad Glagolev, Sean Greven, Hank Leininger, Fernando Arnaboldi, and Erik Gomez.
The phrase “Single Packet Authorization” was coined by MadHat and Simple Nomad at the BlackHat Briefings of 2005.
Send bug reports to [email protected] or [email protected], or open a new issue on Github (see https://github.com/mrash/fwknop.git). Suggestions and/or comments are always welcome as well. Additional information may be found in the fwknop mailing list archives (see: https://lists.sourceforge.net/lists/listinfo/fwknop-discuss).
fwknop is distributed under the GNU General Public License (GPL) version 2, and the latest version may be downloaded from http://www.cipherdyne.org.