sh-elf: PR target/15886

[Joern Rennecke]

> This looks odd for TARGET_SH1.  pr is saved at the address pointed
> by frame_pointer_rtx + frame size for TARGET_SH1, but it seems that
> the final frame size can't be known here because allocate_initial_values
> is called in regalloc phase.

Yes, indeed.  It appears it should use gen_rtx_MEM (Pmode, return_address_pointer_rtx) both for TARGET_SH5 and ! TARGET_SH5.

> Is it wrong to just return NULL_RTX for that case?

Yes, it is wrong, because that would invalidate the RETURN_ADDRESS_POINTER
case of initial_elimination_offset.