Full Disclosure mailing list archives

'Rowhammer' - Software-triggered DRAM corruption


From: Nick Boyce <nick.boyce () gmail com>
Date: Thu, 12 Mar 2015 16:00:20 +0000

(I'm just posting the news - haven't seen this here yet)

A team of Google security researchers recently reported on discoveries
they have made over the last few months which show it is possible to
alter contents of DRAM locations by simply *reading* the contents of
neighbouring locations.  Using this technique they were able to
develop an exploit which modified page tables to allow write access to
the whole of physical memory and thus take complete control of the
system. This is operating-system-agnostic.

    "As DRAM manufacturing scales down chip
    features to smaller physical dimensions, to fit
    more memory capacity onto a chip, it has
    become harder to prevent DRAM cells from
    interacting electrically with each other.  As a
    result, accessing one location in memory
    can disturb neighbouring locations, causing
    charge to leak into or out of neighbouring cells"

Ouch.

http://googleprojectzero.blogspot.co.uk/2015/03/exploiting-dram-rowhammer-bug-to-gain.html?m=1

The current saving grace appears to be that the attack only works
against DDR3 RAM - less dense DDR2 circuitry is not susceptible, and
DDR4 circuit designs contain mitigations against the attack (which
implies the manufacturers have known about this for some time).  The
authors state that ECC does not help, which is puzzling.

Also, this may only affect SODIMMs, not DIMMs, as Google was only able
to make the attack work on laptops - desktop machines so far remaining
unaffected.

[I *knew* it was a good idea to hang on to that old Athlon XP desktop :-)]

Fetch the popcorn ?

Nick
-- 
"Yes, that's all nice ... but can it check the quality of the music,
and improve it along the way, if the music sucks ?"
iptables -A INPUT -m state --state BIEBER -j DROP
http://tech.slashdot.org/comments.pl?sid=6946741&cid=49024417

_______________________________________________
Sent through the Full Disclosure mailing list
https://nmap.org/mailman/listinfo/fulldisclosure
Web Archives & RSS: http://seclists.org/fulldisclosure/


Current thread: