Split edge fails when performed on a doloop jump on x86 port

[Dale Johannesen]

On Feb 18, 2004, at 11:58 AM, Jim Wilson wrote:
> I believe the problem is that operand 2 is a MEM, but the pattern uses 
> register_operand.  There is an inconsistency in the pattern here, the 
> cosntraints allow operands that the predicates don't.

I've hit instances of this on ppc as well.  Would it possible to check 
for this
during bootstrap?  I bet there are more of these bugs lurking.