Re: [patch] Decompose gimplify_expr

[Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>]

Hello,

> > > > constant propagation is not performed anywhere in cfg cleanup.
> > >
> > > Constant propagation is performed in tree_merge_blocks when the PHI
> > > argument is a constant.  IIRC Richard Guenther had an example of this
> > > actually happening in one of his test cases.
> > 
> > OK, I forgot on this.  Nevertheless, the constant propagation itself of
> > course does not cause need to update SSA; removal of the unreachable
> > edges does.  So I still fail to see Daniel's point -- this trivial case
> > of constant propagation in cfg cleanup does not cause any "corner cases
> > in simple optimizations that nobody thinks about that bite us in the ass
> > later".
> 
> A constant propagation into an ARRAY_REF like
> 
>  # i = PHI <0>
>  # V_MAY_DEF <a.0_1>
>  # V_MAY_DEF <a.1_1>
>  a[i] = x;
> 
> causes
> 
>  # V_MUST_DEF <a.0>
>  a[0] = x;
> 
> where a.0 is not in SSA form and needs to be renamed.

umm, nice.  This cannot happen currently, just with your improvements to
alias analysis of arrays, I assume?  The consequences of our insane
system of representation of results of alias analysis still surprise me
sometimes :-)

Zdenek