PowerPC ASM_OUTPUT_{FLOAT,DOUBLE} and ASM_OUTPUT_REG_{PUSH,POP}

[Jeffrey A Law]

  In message < 9903070529.AA29138@marc.watson.ibm.com >you write:
  > 	I am planning to install the following patch, but I want to make
  > sure that no one knows of any reason these changes should not be made.
  > 
  > 	The PowerPC port currently emits float and double constants (other
  > than unrepresentable numbers) as floating-point values as opposed to IEEE
  > 754 bit-patterns.  I think this is a hold-over from a time when the port
  > did not use GCC's REAL_ARITHMETIC support.  This format triggers a bug in
  > the GNU assembler for the smallest denormalized number.  This change makes
  > the port consistent with most other ports.
Using .longs sounds like the right thing to do -- that's how most ports work
now.

jeff