PaX vs precompiled headers (Was: gcc 4.8.1 and precompiled header support for NetBSD hosts)

[Valery Ushakov <uwe%stderr.spb.ru@localhost>]

On Wed, Sep 11, 2013 at 15:20:09 +0200, Martin Husemann wrote:

> However, one bit is still open, and I'd like to get feedback on the hack
> before pushing it upstream: for precompiled header files gcc assumes it
> can pick a steady (over multiple compiler runs) address and mmap a part of
> the precompiled header file at that address. No relocation supported.
> 
> I told them what I think of this design (http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58379)
> but we have to live with it for now.

This is still a problem and while the latest (~an year ago) comment
from Andrew Pinski hints that this might get fixed, we are stuck with
the current precompiled headers implementation in gcc at least for a
few more years.

I wonder if we should explicitly disable ASLR for the cc1* binaries.
This attack vector is too far out there and c++ projects in the wild
do use precompiled headers.

-uwe