[PATCH, x86_64]: Use soft-fp for TFmode (128bit) FP.

[Uros Bizjak]

Hello!

This patch implements TFmode (128bit) floating point by using soft-fp 
for x86_64 targets. To support TImode FP-int conversions, a bunch of 
support files is added to soft-fp directory (+ a couple of small 
bugfixes, as can be seen in attached soft-fp diff).

The bulk of sfp-machine.h is taken from libc and is further extended to 
support TImode integers. Perhaps somebody will be interested in the 
fact, that sfp-machine.h from libc can't be compiled due to the 
definition of _FP_NANFRAC_Q and 64bit _FP_W_TYPE_SIZE. The attached 
sfp-machine.h compiles OK.

The patch was bootstrapped on x86_64 and regression tested for all 
default languages.

Following ChangeLog is without config/soft-fp changes:

2007-04-12  Uros Bizjak  <ubizjak@gmail.com>

        * config.gcc (x86_64-*-linux*, x86_64-*-kfreebsd*-gnu,
        x86_64-*-knetbsd*-gnu): Add i386/t-fprules-softfp64
        and soft-fp/t-softfp to tmake_file.
        * config/i386/t-fprules-softfp64: New file.
        * config/i386/sfp-machine.h: New file.
        * config/i386/t-linux64 (softfp_wrap_start): New.
        (softfp_wrap_end): New.
        * config/i386/i386.c (ix86_scalar_mode_supported): TFmode is
        supported for TARGET_64BIT.

Results, obtained with patched gcc are:

XPASS: gcc.dg/torture/fp-int-convert-float128-timode.c  -O0  (test for 
excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128-timode.c  -O1  (test for 
excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128-timode.c  -O2  (test for 
excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128-timode.c  -O3 
-fomit-frame-pointer  (test for excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128-timode.c  -O3 -g  (test 
for excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128-timode.c  -Os  (test for 
excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128.c  -O0  (test for excess 
errors)
XPASS: gcc.dg/torture/fp-int-convert-float128.c  -O1  (test for excess 
errors)
XPASS: gcc.dg/torture/fp-int-convert-float128.c  -O2  (test for excess 
errors)
XPASS: gcc.dg/torture/fp-int-convert-float128.c  -O3 
-fomit-frame-pointer  (test for excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128.c  -O3 -g  (test for 
excess errors)
XPASS: gcc.dg/torture/fp-int-convert-float128.c  -Os  (test for excess 
errors)

BTW: It is possible to support TFmode on SSE 32bit targets, but what 
does ABI says for argument/result passing of __float128?

Uros.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: i386-fp128.diff
Type: text/x-patch
Size: 4463 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20070412/9140196c/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: soft-fp.diff
Type: text/x-patch
Size: 24207 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20070412/9140196c/attachment-0001.bin>