Re: Pentium DEATH in user-mode

linux kernel account (linker@nightshade.z.ml.org)
Sun, 9 Nov 1997 20:20:03 -0500 (EST)


IT DOES NOT BREAK TRAMPOLINES! (OR SIGNALS) The most recent of the patches
can DETECT THEM and allow them to work unimpeaded (albet a little slower).

As I reall (It's been a while), It even has options to log their use (so
you can 'fix' them)

The patch (when allowing trampolienes) still leaves some room for buffer
exploits, but it's much more difficult.. As I recall I posted to that list
a while back, offering $50 (I think) to anyone who could take a legit
buffer bug and come up with an exploit that worked with trampolines
enabled.. (I think I also asked that the winner also submit an app that
was either commercial or could be found on a linux archive that the patch
broke)

I did not get any submissions.. :) (Good.. Cause I need my $$)

If there is a reason not to include this patch, it isn't because it breaks
anything... Yes, it can give a false sence of security.. But any admin who
is lulled by that prob wont have any security at all.. As a kernel option
I see no reason not to include it.

Please, dont talk negitavly about a patch you have neither used nor
understand!

On Sun, 9 Nov 1997, Aaron Tiensivu wrote:

> > Note that this bug means ANY buffer overflow bug, even on non-setuid apps,
> > is now an entry point for an attacker to crash your machine. Got users on
> > your system who wrote their own CGI apps in C? Ouch. Let's make that
> > non-executable stack patch part of the mainstream kernel.
>
> It would prolly already be in the kernel if it didn't break trampolines.
>