Re: Dataflow branch review #5

[Paolo Bonzini <paolo dot bonzini at lu dot unisi dot ch>]

> >      case RTX_EXTRA:
> >        ...
> > -      if (!CONSTANT_P (op))
> > -        return 0;
> > -      else
> > -	/* As for RTX_CONST_OBJ.  */
> > -	return -3;
> > +      return 0;
> >
> > Why this change?  I don't see why, and I couldn't find a ChangeLog
> > entry for it.
>
> svn blame points to you [Paolo].  why did you do this?

This was probably committed by mistake in the first fwprop commit to the 
branch.  It should actually have been in the mainline too, though I 
cannot find the relevant gcc-patches.  The reason is simply that

#define CONSTANT_P(X)   \
  (GET_RTX_CLASS (GET_CODE (X)) == RTX_CONST_OBJ)

so there is no CONSTANT_P object in class RTX_EXTRA anymore.  These were
CONST_VECTOR and CONSTANT_P_RTX; the former was moved to RTX_CONST_OBJ 
and the latter was removed with the tree-ssa merge.

Paolo