Linux/Moose Worm Targets Routers, Modems, and Embedded Systems
An anonymous reader writes: Security firm ESET has published a report on new malware that targets Linux-based communication devices (modems, routers, and other internet-connected systems) to create a giant proxy network for manipulating social media. It's also capable of hijacking DNS settings. The people controlling the system use it for selling "follows," "likes," and so forth on social media sites like Twitter, Instagram, Vine, Facebook, and Google+. Affected router manufacturers include: Actiontec, Hik Vision, Netgear, Synology, TP-Link, ZyXEL, and Zhone. The researchers found that even some medical devices were vulnerable to the worm, though it wasn't designed specifically to work with them.
Remote management login+password. Telnet connection.
Neither of which is enabled on our TP-Link router.
This is not a story, and not really a Linux problem. The worm relies on weak passwords to execute code. This is about as newsworthy as telling me that car thieves found a way to exploit Fords that have the keys left in them.
It's news not because of OS(I don't know if they bothered; but exploits at the 'just use the default password against the external telnet interface' level would work against basically any OS, and the only real obstacle to executing a payload with the functions described would be that some of the really nasty VXworks-based devices are so RAM-starved that they can barely do their job, much less run malware at the same time); but because the security of nearly all 'consumer', and a disturbing number of more expensive, embedded devices is still utter shit.
It is bad enough that such plastic-box devices typically are shipping software well behind the curve(2.6X kernels, http servers with vulnerabilities that were closed upstream months before the device in question was released, that sort of thing); but 'default configuration leaves telnet listening on the WAN port, with weak credentials for root login' goes well beyond 'bug' and right into 'We Just Don't Care' territory. Even better, the same damn story has been true for at least the past decade, probably longer(though its importance has increased as the cost has fallen and number of little embedded boxes lurking around has skyrocketed).
At least on the desktop and server, some of the worst insecure-by-default atrocities have been ironed out, so attackers are now moderately likely to need to use vaguely clever vulnerabilities(even if they can often get away with ones that were patched months ago) or social engineering; but embedded crap hasn't even reached that level of security.
The fact that telnet is even there(outside of 'recovery' scenarios, where the emergency nature of the situation and availability of only the most limited resources make super-simple protocols like telnet and TFTP valuable) when OpenSSH has been available for the last 15 years, and less liberally licensed versions a bit longer, is disgusting in itself. Having it on the WAN, much less by default, is just depraved.