Buggy peephole prevents bootstrap (i386)

[Jan Hubicka]

> On Fri, 13 Oct 2000, Zack Weinberg wrote:
> 
> > This transformation was done to save space - the add instruction is
> > three bytes shorter than the corresponding compare.  However, add
> > doesn't set the condition codes the same way cmp and sub do.  It
> > appears (from staring at the confusing Intel reference manual) that
> > add's condition codes are usable only with the signed conditional
> 
Hi
I am sorry for causing the bootstrap problems - it still works for me for
some purpose.
The problem with add condition code changes seems to be quite tricky -
I suggest to revert the patch for now and I will try to get around that.
It seems to the patterns are just wrong by having neg on bad side - I ve
messed up the symetricity of add instruction...

Honza