Exception causing insns in delay slots

[law@redhat.com]

 In message <20020427.232129.123986259.davem@redhat.com>, "David S. Miller" 
writes:
 > I keep forgetting the make the Sparc backend do likewise when
 > generating code for UltraSPARC or later, for the same reason.
Yea.  I don't remember the exact numbers, but it was a percentage point or
two for integer codes.  Enough to be measurable :-)

What I keep meaning to do is switch the PA to default to PA8000 scheduling as
they're probably the most commonly found machines these days.

jeff