fixes for m68k problems

[Jeffrey A Law]

  In message < 9812141822.AA23862@moene.indiv.nluug.nl >you write:
  >   > > This patch should fix the m68k bootstrapping problems.
  >   > > It was actually a latent bug.
  >   >
  >   > Well, let's say that it generated a lot of "latent heat".
  > 
  > >  Yea.  The odds of this bug being triggered without all
  > >  the recent work in reload is pretty slim.
  > 
  > Helaas, pindakaas:
  > 
  > stage1/xgcc -Bstage1/ -B/usr/snp/m68k-next-nextstep3/bin/  -DIN_GCC  
  >    -W -Wall -O2 -g -O2  -DHAVE_CONFIG_H     -I. -I../../egcs/gcc  
  > -I../../egcs/gcc/config -I../../egcs/gcc/../include \
  >    
[ ... ]
  > 
  > It might be that this is specific for NEXTSTEP3 (haven't seen any  
  > complaints from others) ...
  > 
  > BTW, I ran into the same crash with a snapshot CVS'd several days  
  > ago, so it is not the result of your changes, but your patch also  
  > doesn't fix it :-(
I just bootstrapped from the CVS tree without problems on m68k with a
bastardized version of 4.3/4.4 BSD.  I wonder if the compiler you're using
to build stage1 is mis-compiling stage1.  I'm using a gcc-2.5.8 +fixes variant
to build stage1.

jeff