Re: stable GCC 3.4 configurations

[Joerg Sonnenberger <joerg@xxxxxxxxxxxxxxxxx>]

On Tue, Mar 22, 2005 at 11:25:57PM +0100, Joerg Sonnenberger wrote:
> On Tue, Mar 22, 2005 at 11:15:43PM +0100, Andreas Bartelt wrote:
> > Just curious - are these problems, when gcc optimizations are enabled, 
> > caused by gcc or by DFly bugs? How can critical parts in the code be 
> > identified?
> 
> It's more a combination of different assumptions the optimisations are
> based upon.

Follow up to myself, the culprit is the (lost patch to remove ?) use
of -fstrict-aliasing with -O2+ and -Os, which is definitely not supported
by some parts of our code base.

You can try to compile with -fno-strict-aliasing, but we make no guaranties
about the result.

I'm thinking about patching it out :)

Joerg