Re: Qualys Security Advisory - The Stack Clash

[Daniel Micay <danielmicay () gmail com>]

On Mon, 2017-06-19 at 09:40 -0600, kseifried () redhat com wrote:
> On 06/19/2017 09:28 AM, Qualys Security Advisory wrote:
> > Qualys Security Advisory
> >
> > The Stack Clash
>
> I just want to publicly thank Qualys for working with the Open Source
> community so we (Linux and *BSD) could all get this fixed properly.
> There was a lot of work from everyone involved and it all went pretty
> smoothly.

Fixing it properly would really also include fixing these:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68065
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66479

and actually implementing -fstack-check as not just a no-op in Clang.

Windows has working stack probes, even in Windows XP and perhaps even
earlier. LLVM has working stack probes there (not sure if GCC deals with
it properly) yet doesn't make them available elsewhere.

Rust is 'memory safe' but has this same stack exhaustion issue. It
didn't used to have the issue, since it kept around the LLVM segmented
stack code generation after it dropped segmented stacks to check for
stack overflow in function preludes. That got dropped for a 1-3%
performance win from using stack probes instead... which was a good
idea, but without implementing stack probes... making it a terrible
idea. It was deferred to some later date. That was in July 2015, and 2
years later it's not done.