[ILUG] Hack attempt??

[Fergal Daly]

At 12:26 06/03/00, Justin Mason wrote:
>On that note, has anyone tried out some of non-executable-stack patches
>for Linux?  And *why* aren't they an option in the standard kernel?!  I
>know it still leaves the possibility of overflows in heap buffers, but
>since most of the bugs are stack-based it would reduce the opportunities
>greatly.

I was reading the kernel mailing list summaries a while ago and there was a 
big argument about this going on. I think the gist was that the stack 
doesn't actually need to be executable, you can just overflow an incorrect 
return address onto it which points into somewhere else in the actual 
program and let that piece of code do the damage (I think it may be 
possible to pass args as well, so you can still do an arbitrary system 
command). If they made the stack unexecutable, the hackers would start 
using the slightly more difficult way of doing things, but people would now 
think that stack exploits are no longer a problem and wouldn't be so 
bothered about checking buffers.

Sounded slightly silly to me, anything which makes life a little more 
difficult for hackers and breaks all the root kits that are out there 
sounds worthwhile. Coders should be checking for overflow for other reasons 
than security.

Anyway, I think Linus was on the side of leaving things as they are, so 
that finished that,

Fergal