I did the recommended fixes with the exception of adding reference links to the fingerprints, because I was unsure what was meant by this.

@droberson I'm sure you didn't create the fingerprints on your own, you must have got them from somewhere, right? By reference links to the fingerprints , I meant the list of websites from which you were able to get these fingerprints.

@rewanth1997 I referenced the RFC and original tool which does not have a website anymore and hasn't been maintained since 2004. I gathered these fingerprints myself by connecting to multiple devices that I was able to positively identify; I own the devices personally, they are devices that I administer at work or stuff at friends' houses.

When you connect to a telnet service, they typically send a series of options requests/other commands. These come in a 3-byte format:

0xff command value

This script extracts those commands and outputs them as the decimal value of their ASCII values so they can be copied and pasted easily.

Connecting to several dozen Cisco IOS devices of different models and versions will yield the same "fingerprint" regardless of if someone has changed or removed the banner because they share the same implementation of a telnet server. The same applies to Linux telnetd, regardless of distro, the architecture of the device, or if the administrator has set a non-stock banner.

In practice, most telnet implementations will have a unique "fingerprint" based on the data sent by the server upon connection. This has been a very reliable method of identifying at least the manufacturer/vendor of a device running telnet.

@cldrn @dmiller-nmap This looks good to me and is working good. Final confirmation please.

I appreciate your work on this script, but I don't see why we can't make these into service matches in nmap-service-probes. If, as I understand it, the method is to collect all the Telnet directives (IACs) until the first non-IAC byte, then you should be able to put these into that file just after # Specific vendor telnet options that should be matched more accurately by prompt, etc:

softmatch telnet m|^\xff\xfb\x01(?!\xff)| p|APC PDU/UPS devices or Windows CE telnetd|
softmatch telnet m|^\xff\xfb\x01\xff\xfd\x03\xff\xfb\x03\xff\xfd\x1f(?!xff)| p/Aruba telnetd/
softmatch telnet m|^\xff\xfd\x03(?!xff)| p/Cisco telnetd/
softmatch telnet m|^\xff\xfb\x01\xff\xfb\x03\xff\xfd\x18\xff\xfd\x1f(?!xff)| p/Cisco IOS telnetd/
softmatch telnet m|^\xff\xfd\x1f(?!xff)| p/Cowrie Honeypot telnetd/
softmatch telnet m|^\xff\xfd\x03\xff\xfb\x03\xff\xfb\x01(?!xff)| p/Enterasys telnetd/
softmatch telnet m|^\xff\xfb\x01\xff\xfb\x03(?!xff)| p/HP LaserJet telnetd/ d/printer/
softmatch telnet m|^\xff\xfb\x03\xff\xfb\x01(?!xff)| p/HP Integrated Lights Out telnetd/ d/remote management/
softmatch telnet m|^\xff\xfc\x01(?!xff)| p/HP JetDirect telnetd/ d/printer/
softmatch telnet m|^\xff\xfb\x01\xff\xfb\x01\xff\xfb\x01\xff\xfb\x03\xff\xfd\x18\xff\xfd\x1f(?!xff)| p/Huawei telnetd/
softmatch telnet m|^\xff\xfd\x18\xff\xfd\x20\xff\xfd\x23\xff\xfd\x27(?!xff)| p/Linux telnetd/ o/Linux/ cpe:/o:linux:linux_kernel/a
softmatch telnet m|^\xff\xfd\x25\xff\xfb\x01\xff\xfb\x03\xff\xfd\x27\xff\xfd\x1f\xff\xfd\x00\xff\xfb\x00(?!xff)| p/Microsoft Telnet Service telnetd/
softmatch telnet m|^\xff\xfd\x25\xff\xfb\x01\xff\xfd\x03\xff\xfd\x1f\xff\xfd\x00\xff\xfb\x00(?!xff)| p/Windows NT 4.0 telnetd/ o/Windows/ cpe:/o:microsoft:windows_nt:4.0/a
softmatch telnet m|^\xff\xfb\x01\xff\xfb\x03\xff\xfb\x00\xff\xfd\x01\xff\xfd\x00(?!xff)| p/Moxa Serial to Ethernet telnetd/

You will notice that we already have one exactly like this for Huawei devices.

Can you try this and let us know if it produces results you would expect? The service matching engine is much faster than NSE when it comes to this sort of thing, and more people use it.

Shucks. This is a way better solution. I should have looked into this more beforehand.

@droberson If you can confirm these softmatch lines look/work OK, we will add them and still credit you in the CHANGELOG.

@dmiller-nmap Many of these signatures were present already in nmap-service-probes

Order mattered for some of these, too. For instance HP laserjet and Moxa devices:
softmatch telnet m|^\xff\xfb\x01\xff\xfb\x03(?!xff)| p/HP LaserJet telnetd/ d/printer/
softmatch telnet m|^\xff\xfb\x01\xff\xfb\x03\xff\xfb\x00\xff\xfd\x01\xff\xfd\x00(?!xff)| p/Moxa Serial to Ethernet telnetd/

This was incorrectly reporting Moxa devices as LaserJet printers.

The following have been verified and did not have entries in this file to my knowledge:

softmatch telnet m|^\xff\xfd\x18(?!\xff)| p/GE Multilin/
softmatch telnet m|^\xff\xfb\x01\xff\xfb\x03\xff\xfb\x00\xff\xfd\x01\xff\xfd\x00(?!xff)| p/Moxa Serial to Ethernet telnetd/
softmatch telnet m|^\xff\xfd\x1f(?!xff)| p/Cowrie Honeypot/
softmatch telnet m|^\xff\xfb\x01\xff\xfd\x03\xff\xfb\x03\xff\xfd\x1f(?!xff)| p/Aruba telnetd/

Ah, I see I was missing a backslash in the negative lookahead. I'll fix that and commit; that should fix the order problem. Thanks!