Kernel Bug Lay Undiscovered for Eight Years
A Linux kernel bug in network socket initialization could allow an attacker to acquire root privileges to inflict damages. The security hole applies to all kernels of the 2.4 and 2.6 series and has been around possibly since 2001.
Source of the bug is a NULL pointer dereference caused by an incorrectly implemented function during socket initialization. Pointers should have redirected the function to a predefined stub routine, but some protocols leave these pointers uninitialized. The list of affected protocols is pretty long: Appletalk, IPX, X.25, IRDA, Bluetooth, ISDN, AX25, SCTP via IPv6 and IUCV, although there may be more.
The kernel validates the pointers before executing most of the relevant functions. However, it was found that the sock_sendpage() routine bybasses the validation and accepts the NULL pointer. An attacker can thereby inflict some damage with code on the first page executed with root privileges. A precondition, however, is that the attacker already has code modification privileges as user, which precludes any remote attack possibilities.
A few exploits were already prepared for the vulnerability, but a relevant kernel patch now exists in Git. Most distros should implement their own security patches for the affected protocols for their delivered kernel versions.
great post
http://www.cocoschanel.com
http://www.gucciguccis.com
http://www.urboots.com
http://www.handbags2012.com
http://www.louisvuittonslv.com
http://www.uggmalls.com
Naivete
If on the other hand it had been known about and exploits available, and it had never been fixed,then there would be something to get upset about."
The comment above may be a bit naive? Just because it was not found by kernel developers until recently does not mean that the bug may not have been exploited.
Why
If on the other hand it had been known about and exploits available, and it had never been fixed,then there would be something to get upset about.
kernel typo - (I hope)
Uh, that applies to all kernels of the 2.4 - 2.6 series...
quote: "2.6, from 2.6.0 up to and including 2.6.30.4"
I hope this was just a typo, because leaving out that it affects the 2.6 series would be a large blunder.