Slashdot Mirror
New 'Phlashing' Attack Sabotages Hardware
yahoi writes "A new type of denial-of-service attack, called permanent denial-of-service (PDOS), damages a system so badly that it requires replacement or reinstallation of hardware. A researcher has discovered how to abuse firmware update mechanisms with what he calls 'phlashing' — a type of remote PDOS attack."
I'm sick of this naming phad.
FINALLY! *This* is bricking
Is it possible to exploit firmware from the outside, unless the person has enabled remote management and is using the default password?
Those two rarely go hand in hand.
However, I think we'll see a lot of trojans with firmware payloads. How many people use the WRT54G? And how many access points are unsecured with the name "linksys"? Those people probably didn't change their admin password.
Simple solution: Hardware button. You have to press it to flash the router, and you have a minute after you press it to upload the firmware. Should be an easy thing to do and provide a great amount of protection.
Phlashing? And he calls his demo code PhlashDance? Good way to make this seem completely silly. "Damn it, we've been phlashdanced!" That'll really get management to up your security budget, if they ever stop laughing.
It figures that when "bricking" might be remotely appropriate, they pick something worse.
It could have been remote bricking, BOIP(brick over IP), brick-and-run, packet bricking, warbricking.
Even brick-o-gram(landshark).
Sigh...
nah - his tool's called PhlashDance, which made me go all warm and fuzzy at the thought of Jennifer Beals stamping on my fimware in her heels :P
One swallow does not a fellatrix make
Dear Sir, I am the former son of the Nigerian dictator Sonni Abacha. I would like to give you several million dollars. To receive this, please add a static IP to your D-Link router and reboot it.
If you haven't made a developer cry, you've wasted a day.
He used to be able to turn any working piece of kit into a piece of metal art in about 20 seconds, EVERYTHING was always a BIOS issue and he would NEVER check with anyone before replacing the BIOS.
Lets be clear about how dumb this person was, he had a BIOS that worked on his test servers and would then apply that to all the other servers INDEPENDENT OF HARDWARE OR OS. He would then start the machines (which of course wouldn't start) declare them "broken" and say the issue was with the software.
We did some low level hardware stuff in our software and it did break the boxes sometimes so it took 2 months of painful testing and debugging which found nothing, it only came about because one of the team had a heavy night and decided to "rest" in the server room and saw the moron apply the BIOS to a server that had been running and then scurry out to blame the team again.
Basic rule after then was BIOS set to read-only and locked down with a secure password, to this day my BIOS has a password thanks to the sheer physical shock of realising how dumb some people can be.
An Eye for an Eye will make the whole world blind - Gandhi
This isn't exactly a new problem...in the early days, you could fry a monitor by setting the video card to absurd refresh rates, and you could destroy hard disks by issuing bogus stepping commands to the heads and slamming them into the stops.
I'm sorry, but every device out there should have two factory reset switches:
1 to reset user data, akin to a standard BIOS "reset to factory settings"
1 to re-flash the BIOS to the factory-installed version of the BIOS, to de-brick devices.
Furthermore, if there is anything a user can do that is designed to update the machine in a way that's irreversible without a password setting a BIOS or boot password, a hardware switch should be pressed as the information is saved. While this won't prevent social engineering, it will prevent pure software exploits from making the hardware unusable.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
I experimented with a technique (that worked) on the Commodore 64. You could address the floppy drive directly to move the drive head to the innermost position, which was on the opposite side of the "track 0" microswitch. Then you deliberately crash the CPU on the drive. When it POSTs it moves the head inward to track 0 to initialize. Since the head is on the wrong side of the switch it never gets there, makes a terrible noise, and gives up.