Reload bug

[Jan Hubicka]

> > No, it is valid for high parts too as long as the register is possible
> > (imagine the xmm being replaced by eax).
> 
> Did I forget to say that this example is bogus? The logic doesn't of course 
> trigger for %eax so we don't pessimize anything in that case.

What would prevent it from match on (subreg:DI (reg eax) 4)? That was
the original problem that made me to change the behaviour (we can't
reload instructions needing two such subregged operands in -fPIC as we
simply run out of registers)

Honza
> 
> -- 
> Eric Botcazou