Fix PR tree-optimization/20940

Diego Novillo
Wed Mar 16 17:56:00 GMT 2005

On Wed, Mar 16, 2005 at 12:34:00PM -0500, Richard Kenner wrote:
>     Oh?  This could be a problem.  I thought that fold was going to
>     evolve into 2 separate modes, one for front end use and another
>     for middle-end use.
> Which fold would then be used for a tree-level combiner?
The regular fold().  A tree combiner must be able to regimplify
and reconstruct the SSA web.


More information about the Gcc-patches mailing list