Post by @rqm

rqm
@rqm
Public en edited

New local kernel exploit for #OpenBSD called #bigfish gets attackers root with KARL/SMAP/SMEP enabled, with building blocks in the chain going back ~37 years to the original BSD sources:

https://calif.io/research/big-fish

It also has really cool ANSI art. 😉

Replying to @rqm
Tomáš
@prahou @merveilles.town

@rqm brb, running syspatch

@prahou@merveilles.town Just come over to -current

Replying to @rqm
Tomáš
@prahou @merveilles.town

@rqm never!

@prahou@merveilles.town Stables are for horses, and we know horses sink

Replying to @rqm
Tomáš
@prahou @merveilles.town

@rqm walked into that one

wow.. well done

@prahou@merveilles.town a horse, an artist, and a priest walk into a swamp...

Replying to @rqm
Tomáš
@prahou @merveilles.town

@rqm go on

Replying to @rqm
Haelwenn /элвэн/ :triskell:
@lanodan @queer.hacktivis.me
@rqm @prahou <Shrek> This is MY swamp!
Replying to @rqm
Tomáš
@prahou @merveilles.town

@rqm also is it -CURRENT because FISH and WATER? #openbsd

Replying to @rqm
Dr. Bruce Simpson
@bms @mastodon.bsd.cafe
Public en edited

@rqm The first mitigation that comes to mind is that of Capsicum where the capability to invoke shmat() is never granted. Only the mapping of POSIX anonymous shm segments is permitted. Granted, that is FreeBSD specific, but it points towards the incompleteness of the pledge model in OpenBSD. In-kernel mitigation is most likely only possible with PAC/MTE on arm64 or PKS on x86-64 which is neither supported nor adopted.

Replying to @rqm
Slow Servers
@slowservers @bsd.cafe

@rqm Pretty impressive!

I am looking forward to getting this patched on all of my servers, but for my hosts I am hoping for 8.0 as it resolves some issues for me.

I do have direct SSH access on hosts, but it's under a very restricted shell and I'm not sure you can spin up enough processes to pull it off.

```
vmuser:\
:umask=022:\
:datasize=16M:\
:coredumpsize=1024:\
:openfiles=20:\
:maxproc=10:\
:tc=auth-defaults:
```