"Matthias Pfaller" <leo@dachau.marco.de> writes:

  > Ian Dall wrote:
  >> I have done this on gcc 2.7.2.1 from FSF. I recall that someone did
  >> some work for gcc to make it handle 64 bit types better. Has that made
  >> it into the FSF distribution or should I get it from the netbsd source
  >> tree?

  > That was Jon Buller and me. I think Jon sent the code to Richard Kenner
  > (is this true Jon?). Should I mail the diffs to the list? I think Jon
  > (or I) did this already one or two years ago.

I looked back through my archives and found a copy of the message Jon
sent to Richard Kenner. These patches are not in gcc up to 2.7.2.2.
Maybe they will be in 2.8. I notice that they are *also* not in the
netbsd-current source tree! I also noticed that there are some other
much older differences between the ns32k.md in the netbsd tree and the
FSF distribution -- notably the ffs instruction definitions. I have merged
both these changes into my version.

I am making good progress with my changes, but still have a few weird
things to sort out. I have it so gcc recognizes it can combine
multiply and add into multiply add, but for some reason it recognizes
the double precision case but not the single precision case. Also it
missed some apparently obvious opportunities to combine. It appears to
generate correct inline memcpy code, but in at least one instance,
puts something in a stack slot when there are perfectly good registers
not being used!

Ian