Re: [PATCH]: Rewrite reassociation pass => Ada bootstrap failure on x86-linux

[Jeffrey A Law <law at redhat dot com>]

> Thanks for looking into this!
Just part of the "job".  The GCAC bootstrap is still running and
probably will still be running when I hop on the plane tomorrow.  :-)

My current theory is the GNAT sources are triggering an optimization
that is rarely or never triggered by the rest of the GCC sources and
the code to implement that optimization is somehow GC unsafe.  That's
a theory I can test quickly when I get back to SLC.



> When this feature was added, there were some discussions
> on this list, may be the following still works:
> 
> http://gcc.gnu.org/ml/gcc/2003-12/msg00748.html
> <<
> [root@dberlin bin]# grep exec-shield /etc/sysctl.conf
> kernel.exec-shield = 0
> kernel.exec-shield-randomize = 0
> 
> (throwing it in sysctl.conf makes it automatically do it for you on
> bootup, you'll still have to sysctl -w these manually if you don't want
> to reboot)
The exec-shield stuff has changed a little through time -- I've been
disabling randomization via one of the rc.foo scripts and peeking at
the values in /proc shows that it ought to be disabled.  But reality
is I'm still seeing randomization.

I'm sure there's a good explanation for that, but given that I'll be
back in SLC shortly I'm not terribly inclined to spend a lot of
time figuring out why the anti-randomization stuff isn't working
on my laptop.

jeff