gcc-2.95pre: Internal compiler error in `gen_add2_insn'

[David Edelsohn]

>>>>> Jeffrey A Law writes:

Jeff> Actually, I think we need to make the generic addsi/addi3 expander handle
Jeff> the counter register.

Jeff> I'm not sure if it's doc'd anywhere, but the add patterns are considered
Jeff> special much like the movxx patterns and need to handle some of these
Jeff> oddball cases.

	I cannot believe that this is the solution.  PowerPC cannot
operate on CTR directly.  This seems like exactly the type of case that
secondary reload is suppose to solve.  The addsi3 insn patterns would need
a scratch register just like reload is suppose to be generating when it
synthesizes the equivalent sequence.

David