[tree-ssa] bootstrap miscomparison failure on PPC

[Diego Novillo]

On Fri, 2003-07-04 at 18:06, law@redhat.com wrote:
> In message <1057331693.3147.58.camel@frodo.toronto.redhat.com>, Diego Novillo w
> rites:
>  >Jeff,
>  >
>  >One of these two changes introduced a bootstrap regression on PPC.  We
>  >miscompare gcc.o.  I'm still not sure which one yet.
> Why don't you restart -- I just checked in a bugfix for the folder
> change.   Let me know how it goes.
> 
Oh, our messages crossed paths.  Undoing the tree-ssa.c patch allowed me
to bootstrap.  When doing constant replacements, we should fold the
statement, but in this case it's not clear that this will help since we
are replacing the whole RHS of the assignment with a constant.  I'm not
sure what may be wrong with it.


Diego.