Skip to contentSkip to conversation

exec-shield-nx-2.6.9-A1

PATCH Albert Cahalan 2 participants 3 messages Oct 17, 2004

lore
Share & tools
lore
#0 lore raw

You have some bits in this patch that don't belong. They aren't even conditional on a config option or sysctl value.

First, you change the permission on the /proc/*/maps file. Normally a remote attacker is unable to read this anyway, and a local setuid attack has time to try until success. Changing the permission might be a good idea, mostly because it exposes filenames, but it should be a separate patch.

Second, you restrict wchan. Oddly, you don't allow for the target task's euid to play a role, and you chose the CAP_SYS_NICE bit instead of some other bit. Huh? One might guess from CAP_SYS_NICE that the feature has now become hopelessly slow. Same as with the maps file, this should be a separate patch.

#1 lore raw
Quoted 3 lines · Albert Cahalan

You have some bits in this patch that don't belong. They aren't even conditional on a config option or sysctl value.

maybe you misunderstood my mail. This was an announcement for exec-shield users. You can safely ignore it.

Quoted 2 lines

[...] One might guess from CAP_SYS_NICE that the feature has now become hopelessly slow. [...]

(thank you for the kind words, it is always heartening to read your mails! I too wish you good luck with your projects.)

Ingo

#2 lore raw
Quoted 5 lines · Ingo Molnar
Quoted 2 lines

[...] One might guess from CAP_SYS_NICE that the feature has now become hopelessly slow. [...]

(thank you for the kind words, it is always heartening to read your mails! I too wish you good luck with your projects.)

:-)

Well, CAP_SYS_NICE does kind of imply that there is an issue with CPU time, doesn't it? I didn't see any code that would change performance, so I suppose you figure that wchan is already too slow?

CAP_SYS_ADMIN is the misc. dumping ground normally.