[PATCH] clean up support for testing DFP rounding

Ben Elliston bje@au1.ibm.com
Tue Nov 21 22:52:00 GMT 2006


Hi Janis

Sorry for the delay ..

On Wed, Nov 15, 2006 at 05:14:08PM -0800, Janis Johnson wrote:

> This patch cleans up the current testing support for DFP rounding, and
> uses a new macro that's similar to the one used in soft-fp support.
> 
> Tested on powerpc64-linux with --enable-decimal-float.  OK for mainline?
> 
> 2006-11-15  Janis Johnson  <janis187@us.ibm.com>
> 
> libdecnumber/
> 	* decRound.c: Move declarations to header file.
> 	* decRound.h: New.
> 
> gcc/
> 	* config/dfp-bit.h: Include decRound.h.
> 	(CONTEXT_ROUND): Delete.
> 	(DFP_INIT_ROUNDMODE): Define if not already defined.
> 	* config/dfp-bit.c (dfp_unary_op, dfp_binary_op, dfp_compare_op,
> 	DFP_TO_DFP, INT_TO_DFP, BFP_TO_DFP): Use DFP_INIT_ROUNDMODE.

It's probably necessary to update some Makefile dependencies for the
existence of decRound.h (perhaps in gcc/Makefile.in, but more likely
in libgcc.mk).  Can you check, please?

Otherwise, OK.

Cheers, Ben



More information about the Gcc-patches mailing list