Just something I noticed with Paxtest

Discuss usability issues, general maintenance, and general support issues for a grsecurity-enabled system.

Just something I noticed with Paxtest

Postby GBit » Thu Jul 11, 2013 2:50 am

running paxtest blackhat I get considerably higher entropy across the board, except for

Main Executable Randomization (PIE) where I get 27 bits (whereas vanilla I get 28) and this appears to be consistent after rebooting.

Just wondering if this is an issue/ regression (27 bits seems adequate anyways, just curious/ thought I'd bring it up).

My config:
http://pastebin.com/X3qzCghY

Again, I don't think it's something major, or that 27 isn't enough. I just thought that this might help out in case a feature is not working as expected.

edit: It may also just be a difference between 3.8 and 3.10. The pax kernel is 3.10, the vanilla is 3.8.x.
GBit
 
Posts: 81
Joined: Mon Jun 04, 2012 3:31 pm

Re: Just something I noticed with Paxtest

Postby PaX Team » Sun Jul 14, 2013 6:52 pm

it's due to UDEREF and KERNEXEC which on amd64 rely on a special implementation detail (per-cpu pgd) that reduces the available userland address space size from 47 to 42 bits with a corresponding reduction in ASLR entropy as well.
PaX Team
 
Posts: 2310
Joined: Mon Mar 18, 2002 4:35 pm

Re: Just something I noticed with Paxtest

Postby GBit » Sun Jul 14, 2013 7:22 pm

Makes sense. Thank you.
GBit
 
Posts: 81
Joined: Mon Jun 04, 2012 3:31 pm


Return to grsecurity support