[PATCH] PowerPC extendsfdf2 MEM alternative

[Richard Henderson]

On Sat, Mar 12, 2005 at 01:46:59PM -0500, David Edelsohn wrote:
> 	Is there a reason that output_constant_def() does not have the
> compress_float_constant() machinery?

Because the code above has to be modified to account for the difference.

What's your test case for this?  I'm confused about how the path
wouldn't be expand_assignment -> store_expr -> emit_move_insn.


r~