Re: patch/RFC: PR optimization/8423

[Zack Weinberg <zack at codesourcery dot com>]

On Wed, Nov 06, 2002 at 11:23:37AM -0800, Richard Henderson wrote:
> On Tue, Nov 05, 2002 at 01:55:00PM -0800, Zack Weinberg wrote:
> > Staring at toplev.c for awhile leads me to believe that the right fix
> > for this is to run GCSE _before_ CSE, and make GCSE aware of
> > CONSTANT_P_RTX.  The appended patch does just that.  Since this is a
> > drastic change, I will also check what overall effect this has on code
> > quality by examining bootstrap times and the size of the generated
> > compiler.
> 
> I wonder if this means we can get rid of one of the local cse
> passes?  It wouldn't surprise me if this is indeed now redundant,
> since gcse has acquired some local propagation skills.
> 
> Failing improvements in both compile quality and compile time,
> I think we should simply xfail this test.

What actually happens is it blows up compiling libgcc, due to some
construct or other that simplify-rtx.c doesn't know how to handle but
cse's simplifiers do.  I haven't yet dug into this.

I think xfailing this for the time being is the right move.  I may
continue to poke at this on basic-improvements - we know local CSE is
dog slow, it would certainly be nice to get rid of one of the passes.

zw