News for hackers. Stuff that matters.
Re:Debugging mechanism in Intel CPUs allows seizing control ... by payne92 on Monday January 09, 2017 @11:10PM (#13362499)
I am continually perplexed at how hardware designers end up with these kinds of "features" on external ports.

Remember direct DMA attacks on FireWire, etc.?

SMH.

Re:Debugging mechanism in Intel CPUs allows seizing control ... by Animats on Tuesday January 10, 2017 @12:11AM (#13362770)
I tried to get that disabled in Linux. There's a register in the FireWire interface which sets the range of addresses into which external memory access requests can be made. By default, it was at 4GB, even on 64 bit systems. Setting it to zero would break some external kernel debugger, though, so it was left open.

(Firewire isn't DMA. PCI-E is DMA. Firewire is a network cable, over which software sends packets. LAN over FireWire was sometimes done. But there's an optional feature which recognizes special packets for doing word-size loads and stores. This is usually used to talk to dumb slave devices, where you write "registers" to make things happen. It's not needed on a computer, and it's far too slow for bulk transfers. FireWire isn't inherently more vulnerable than a LAN port.)

Re:Debugging mechanism in Intel CPUs allows seizing control ... by kalleboo on Tuesday January 10, 2017 @04:35AM (#13363724)
IIRC Apple's "fix" was to just disable recognition of new FireWire devices when the computer was locked. So a nasty device could still compromise you, but it would be one that you plugged in yourself.
Re:Debugging mechanism in Intel CPUs allows seizing control ... by mikeash on Tuesday January 10, 2017 @10:06AM (#13365291)
It looks like your recollection is indeed correct. See the entry for CVE-2011-3215 here: https://support.apple.com/en-us/HT202348
Re:Debugging mechanism in Intel CPUs allows seizing control ... by psybin on Tuesday January 10, 2017 @02:06AM (#13363235)
Firewire in most setups allows DMA access to the host, but only to the first 32 bits of RAM. There's tools for unlocking, reading out keys from memory while asleep, all sorts of nasties.
Re:Debugging mechanism in Intel CPUs allows seizing control ... by Animats on Tuesday January 10, 2017 @03:04AM (#13363441)
In Linux, that's a constant in the Linux driver.
Re:Debugging mechanism in Intel CPUs allows seizing control ... by andreyf on Monday January 09, 2017 @11:21PM (#13362544)
I'd be curious to see a list of devices which ship with this enabled. I imagine it's shorter than the article makes it sound.

I imagine DMA was a trade-off for speed. Security is about such trade-offs.

Re:Debugging mechanism in Intel CPUs allows seizing control ... by psybin on Monday January 09, 2017 @11:30PM (#13362576)
It's a common feature for very high speed writes in hardware, DMA in firewire and thunderbolt is no mistake.
Re:Debugging mechanism in Intel CPUs allows seizing control ... by yuhong on Monday January 09, 2017 @11:12PM (#13362513)
The fun thing is that it is probably a very useful feature for those who know how to use it. You can debug your BIOS this way for example.
Re:Debugging mechanism in Intel CPUs allows seizing control ... by viraptor on Monday January 09, 2017 @11:44PM (#13362649)
Sure, but you should disable it before shipping the final version. It's the same with programming microcontrollers - you have full read write and debug available. But you should really flip the right fuses to disallow changes in the shipped product.