Loop optimizer issues

[Zdenek Dvorak]

Hello,

> >> > just for sure -- nobody will protest if I drop handling of libcalls
> >> > in new loop optimizer? This definitely should not be neccesary to handle
> >> > on rtl level.
> >> 
> >> How do you plan to handle ports where multiplication or addition
> >> requires a libcall?
> 
> Zdenek> in no special way.  The optimizations done on ast level would then have
> Zdenek> higher priority for optimizing them when possible.
> 
> 	Does "new loop optimizer" refer to a revision to the RTL loop
> optimizer or a re-write of the loop optimizer in the Tree-SSA
> infrastructure?

rtl one.

> LIBCALLs should not be present in GIMPLE, should they?

I hope so; this is exactly why I would prefer to handle things there
rather than figting it later in rtl.

Zdenek