Fix a bug in hashtab.c and improve cse.c perfs

[Jeffrey A Law]

  In message < 19991030233307.A12193@schutzenberger.liafa.jussieu.fr >you write:
  > I've been evaluating hash tables. I do indeed get the 10% speed increase
  > in compiling gcc on a sparc, with a gcc 2.95.1.
  > 
  > Jeff, maybe it would be a good idea to copy those hash tables to the
  > 2.95 branch, if a 2.95.3 is on the table ?  This looks to me as a change
  > which should not break anything, is small and completely independent of
  > the rest of the compiler, and is much too good to be ignored...
Sorry.  No, this is not the kind of change we would want to make for gcc-2.95.3
if indeed such a release ever happens.  Those releases are for *critical* bugs,
the new hash table code doesn't have a prayer of meeting that criteria.  It
also likely fails the absolute safety criteria.




jeff