Re: Re: Qualys Security Advisor -- The Stack Clash

[Daniel Micay <danielmicay () gmail com>]

On Mon, 2017-06-19 at 11:26 -0600, Jeff Law wrote:
> I would consider those two GCC BZs (68065, 66479) a separate an
> distinct
> issue.
>
> It is far more important to address design issues around the existing
> -fstack-check first.  I think we've got a pretty good handle on how to
> address those problems and discussions with the upstream GCC community
> have already started.
>
> In an ideal world we'll get to a place where the new -fstack-check
> does
> not change program semantics, never misses probes and is efficient
> enough to just turn on and forget everywhere.  The existing
> -fstack-check fails all three of those criteria.
>
> Jeff

AFAIK, the main efficiency issue (reserving a register) was fixed for
GCC 6. I might be missing something but it seems very cheap now, at
least for x86_64. It definitely doesn't really work though.

Is there an example of it changing program semantics? I haven't seen
anything since the generic arch stuff was fixed.