Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[Uros Bizjak]

Hello!

> Kai Tietz wrote:
>
>     I tested your suggested patch for i686-pc-mingw32 and for
>     x86_64-pc-mingw32 and it works. I would like to add this for mingw
>     targets, too..
>
>
>     Cheers,
>     Kai
>
>
> It is pre-approved if it passes testing.
> See http://gcc.gnu.org/ml/gcc-patches/2009-05/msg01458.html

Yes, please go ahead and enable __float128 for mingw targets.

Uros.

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[David Billinghurst]

Kai Tietz wrote:

> I tested your suggested patch for i686-pc-mingw32 and for
> x86_64-pc-mingw32 and it works. I would like to add this for mingw
> targets, too..
> 
> Cheers,
> Kai
> 

It is pre-approved if it passes testing.
See http://gcc.gnu.org/ml/gcc-patches/2009-05/msg01458.html

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[Kai Tietz]

Hello Uros,

2009/5/29 David Billinghurst <dbcygwin@gmail.com>:
> Uros Bizjak wrote:
>>
>> Hello!
>>
>>> This patch enables support for the __float128 (TFmode) IEEE quad type in
>>> libgcc on cygwin.  It uses the same approach as for IA-32 on darwin.
>>>
>>> With this patch the tests gcc.dg/torture/fp-int-convert-float128*.c pass
>>> on cygwin.  Previously the tests failed to link.  See:
>>>  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg00959.html
>>>  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg01028.html
>>
>> soft-fp is added for other x86 targets in "miscellaneous part" of
>> config.gcc. Assuming that soft-fp also works for mingw (it is trivial to add
>> soft-fp nowadays ;), we can simply add:
>>
>>   i[34567]86-*-pe | i[34567]86-*-cygwin* | i[34567]86-*-mingw* |
>> x86_64-*-mingw*)
>>       tmake_file="${tmake_file} i386/t-fprules-softfp soft-fp/t-softfp"
>>
>> to the "miscellaneous part" of config.gcc.
>> i386/32/t-fprules-softfp actually filters out TImode conversions that are
>> not valid on x86_32. Please note, that there is no i386/64/t-fprules-softp.
>>>
>>> +       ;;
>>> +i[34567]86-*-mingw*)
>>> +       extra_parts="crtbegin.o crtend.o crtfastmath.o"
>>>        tmake_file="i386/t-cygming i386/t-crtfm"
>>>        ;;
>>>  x86_64-*-mingw*)
>>>
>> Please add windows targets to existing target groups. It will just work
>> out-of-the-box.
>>
>> I have attached (untested) patch that implements suggested approaches. I'm
>> not able to test it on windows targets. If it bootstraps OK for you, then it
>> is OK for mainline SVN. Technically, testsuite failures are all regressions,
>> but this patch enables new functionality on cygwin targets, so you will have
>> to discuss backport to 4.4 with cygwin maintainers.
>
> I have committed this for cygwin only.  I can't test the other targets.
>

I tested your suggested patch for i686-pc-mingw32 and for
x86_64-pc-mingw32 and it works. I would like to add this for mingw
targets, too..

Cheers,
Kai

-- 
|  (\_/) This is Bunny. Copy and paste
| (='.'=) Bunny into your signature to help
| (")_(") him gain world domination

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[David Billinghurst]

Uros Bizjak wrote:
> Hello!
> 
>> This patch enables support for the __float128 (TFmode) IEEE quad type in
>> libgcc on cygwin.  It uses the same approach as for IA-32 on darwin.
>>
>> With this patch the tests gcc.dg/torture/fp-int-convert-float128*.c pass
>> on cygwin.  Previously the tests failed to link.  See:
>>  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg00959.html
>>  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg01028.html
>
> soft-fp is added for other x86 targets in "miscellaneous part" of 
> config.gcc. Assuming that soft-fp also works for mingw (it is trivial to 
> add soft-fp nowadays ;), we can simply add:
> 
>    i[34567]86-*-pe | i[34567]86-*-cygwin* | i[34567]86-*-mingw* | 
> x86_64-*-mingw*)
>        tmake_file="${tmake_file} i386/t-fprules-softfp soft-fp/t-softfp"
> 
> to the "miscellaneous part" of config.gcc.   
> 
> i386/32/t-fprules-softfp actually filters out TImode conversions that 
> are not valid on x86_32. Please note, that there is no 
> i386/64/t-fprules-softp.
>> +       ;;
>> +i[34567]86-*-mingw*)
>> +       extra_parts="crtbegin.o crtend.o crtfastmath.o"
>>         tmake_file="i386/t-cygming i386/t-crtfm"
>>         ;;
>>  x86_64-*-mingw*)
>>
> Please add windows targets to existing target groups. It will just work 
> out-of-the-box.
> 
> I have attached (untested) patch that implements suggested approaches. 
> I'm not able to test it on windows targets. If it bootstraps OK for you, 
> then it is OK for mainline SVN. Technically, testsuite failures are all 
> regressions, but this patch enables new functionality on cygwin targets, 
> so you will have to discuss backport to 4.4 with cygwin maintainers.

I have committed this for cygwin only.  I can't test the other targets.

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[David Billinghurst]

Uros Bizjak wrote:
> Hello!
> 
>> This patch enables support for the __float128 (TFmode) IEEE quad type in
>> libgcc on cygwin.  It uses the same approach as for IA-32 on darwin.
>>

> 
> I have attached (untested) patch that implements suggested approaches. 
> I'm not able to test it on windows targets. If it bootstraps OK for you, 
> then it is OK for mainline SVN. Technically, testsuite failures are all 
> regressions, but this patch enables new functionality on cygwin targets, 
> so you will have to discuss backport to 4.4 with cygwin maintainers.
> 
> Uros.

Thanks for your feedback and guidance.  I will test your patch.

	David

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[Dave Korn]

Joseph S. Myers wrote:
> On Fri, 22 May 2009, Uros Bizjak wrote:
> 
>> Assuming that soft-fp also works for mingw (it is trivial to add soft-fp
>> nowadays ;), we can simply add:
> 
> Note that for targets with symbol versioning a target-specific version 
> list is needed to reflect the versions at which these symbols were added 
> for that target - but I don't think this affects Windows targets.
> 

   Confirmed that it doesn't, but thanks for the heads-up.

    cheers,
      DaveK

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[Joseph S. Myers]

On Fri, 22 May 2009, Uros Bizjak wrote:

> Assuming that soft-fp also works for mingw (it is trivial to add soft-fp
> nowadays ;), we can simply add:

Note that for targets with symbol versioning a target-specific version 
list is needed to reflect the versions at which these symbols were added 
for that target - but I don't think this affects Windows targets.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: [cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[Uros Bizjak]

[-- Attachment #1: Type: text/plain, Size: 2981 bytes --]

Hello!

> This patch enables support for the __float128 (TFmode) IEEE quad type in
> libgcc on cygwin.  It uses the same approach as for IA-32 on darwin.
>
> With this patch the tests gcc.dg/torture/fp-int-convert-float128*.c pass
> on cygwin.  Previously the tests failed to link.  See:
>  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg00959.html
>  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg01028.html
>
> OK for trunk, then 4.4.
>
> 2009-05-21 David Billinghurst <billingd@gcc.gnu.org>
>
> * config.gcc: Enable support for the __float128 (TFmode) for 
> i?86-*-cygwin.
>
> * config.host: Enable support for the __float128 (TFmode) for 
> i?86-*-cygwin.
>
> Index: gcc/config.gcc
> ===================================================================
> --- gcc/config.gcc      (revision 147701)
> +++ gcc/config.gcc      (working copy)
> @@ -1267,7 +1267,7 @@
>         else
>                 tmake_eh_file="i386/t-sjlj-eh"
>         fi
> -       tmake_file="${tmake_file} ${tmake_eh_file} i386/t-cygming i386/t-cygwin"
> +       tmake_file="${tmake_file} ${tmake_eh_file} i386/t-cygming i386/t-cygwin i386/t-fprules-softfp soft-fp/t-softfp"
>         target_gtfiles="\$(srcdir)/config/i386/winnt.c"
>         extra_options="${extra_options} i386/cygming.opt"
>         extra_objs="winnt.o winnt-stubs.o"

soft-fp is added for other x86 targets in "miscellaneous part" of 
config.gcc. Assuming that soft-fp also works for mingw (it is trivial to 
add soft-fp nowadays ;), we can simply add:

    i[34567]86-*-pe | i[34567]86-*-cygwin* | i[34567]86-*-mingw* | 
x86_64-*-mingw*)
        tmake_file="${tmake_file} i386/t-fprules-softfp soft-fp/t-softfp"

to the "miscellaneous part" of config.gcc.

> Index: libgcc/config.host
> ===================================================================
> --- libgcc/config.host  (revision 147701)
> +++ libgcc/config.host  (working copy)
> @@ -323,8 +323,12 @@
>         ;;
>  i[34567]86-*-pe)
>         ;;
> -i[34567]86-*-cygwin* | i[34567]86-*-mingw*)
> +i[34567]86-*-cygwin*)
>         extra_parts="crtbegin.o crtend.o crtfastmath.o"
> +       tmake_file="i386/t-cygming i386/t-crtfm t-softfp i386/${host_address}/t-fprules-softfp"
>   

i386/32/t-fprules-softfp actually filters out TImode conversions that 
are not valid on x86_32. Please note, that there is no 
i386/64/t-fprules-softp.
> +       ;;
> +i[34567]86-*-mingw*)
> +       extra_parts="crtbegin.o crtend.o crtfastmath.o"
>         tmake_file="i386/t-cygming i386/t-crtfm"
>         ;;
>  x86_64-*-mingw*)
>
Please add windows targets to existing target groups. It will just work 
out-of-the-box.

I have attached (untested) patch that implements suggested approaches. 
I'm not able to test it on windows targets. If it bootstraps OK for you, 
then it is OK for mainline SVN. Technically, testsuite failures are all 
regressions, but this patch enables new functionality on cygwin targets, 
so you will have to discuss backport to 4.4 with cygwin maintainers.

Uros.

[-- Attachment #2: p.diff.txt --]
[-- Type: text/plain, Size: 1225 bytes --]

Index: gcc/config.gcc
===================================================================
--- gcc/config.gcc	(revision 147792)
+++ gcc/config.gcc	(working copy)
@@ -3127,6 +3127,9 @@
 	i[34567]86-*-linux* | x86_64-*-linux* | i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu)
 		tmake_file="${tmake_file} i386/t-fprules-softfp soft-fp/t-softfp i386/t-linux"
 		;;
+	i[34567]86-*-pe | i[34567]86-*-cygwin* | i[34567]86-*-mingw* | x86_64-*-mingw*)
+		tmake_file="${tmake_file} i386/t-fprules-softfp soft-fp/t-softfp"
+		;;
 	ia64*-*-linux*)
 		tmake_file="${tmake_file} ia64/t-fprules-softfp soft-fp/t-softfp"
 		;;
Index: libgcc/config.host
===================================================================
--- libgcc/config.host	(revision 147792)
+++ libgcc/config.host	(working copy)
@@ -562,7 +562,9 @@
 case ${host} in
 i[34567]86-*-darwin* | x86_64-*-darwin* | \
   i[34567]86-*-kfreebsd*-gnu | x86_64-*-kfreebsd*-gnu | \
-  i[34567]86-*-linux* | x86_64-*-linux*)
+  i[34567]86-*-linux* | x86_64-*-linux* |\
+  i[34567]86-*-pe | i[34567]86-*-cygwin* | \
+  i[34567]86-*-mingw* | x86_64-*-mingw*)
 	if test "${host_address}" = 32; then
 		tmake_file="${tmake_file} t-softfp i386/${host_address}/t-fprules-softfp"
 	fi

[cygwin patch] Enable Support for __float128 (TFmode) IEEE quad type

[David Billinghurst]

This patch enables support for the __float128 (TFmode) IEEE quad type in
libgcc on cygwin.  It uses the same approach as for IA-32 on darwin.

With this patch the tests gcc.dg/torture/fp-int-convert-float128*.c pass
on cygwin.  Previously the tests failed to link.  See:
  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg00959.html
  - http://gcc.gnu.org/ml/gcc-patches/2009-05/msg01028.html

OK for trunk, then 4.4.

2009-05-21  David Billinghurst <billingd@gcc.gnu.org>

	* config.gcc: Enable support for the __float128 (TFmode) for i?86-*-cygwin.

	* config.host: Enable support for the __float128 (TFmode) for i?86-*-cygwin.


Index: gcc/config.gcc
===================================================================
--- gcc/config.gcc      (revision 147701)
+++ gcc/config.gcc      (working copy)
@@ -1267,7 +1267,7 @@
         else
                 tmake_eh_file="i386/t-sjlj-eh"
         fi
-       tmake_file="${tmake_file} ${tmake_eh_file} i386/t-cygming i386/t-cygwin"
+       tmake_file="${tmake_file} ${tmake_eh_file} i386/t-cygming i386/t-cygwin i386/t-fprules-softfp soft-fp/t-softfp"
         target_gtfiles="\$(srcdir)/config/i386/winnt.c"
         extra_options="${extra_options} i386/cygming.opt"
         extra_objs="winnt.o winnt-stubs.o"

Index: libgcc/config.host
===================================================================
--- libgcc/config.host  (revision 147701)
+++ libgcc/config.host  (working copy)
@@ -323,8 +323,12 @@
         ;;
  i[34567]86-*-pe)
         ;;
-i[34567]86-*-cygwin* | i[34567]86-*-mingw*)
+i[34567]86-*-cygwin*)
         extra_parts="crtbegin.o crtend.o crtfastmath.o"
+       tmake_file="i386/t-cygming i386/t-crtfm t-softfp i386/${host_address}/t-fprules-softfp"
+       ;;
+i[34567]86-*-mingw*)
+       extra_parts="crtbegin.o crtend.o crtfastmath.o"
         tmake_file="i386/t-cygming i386/t-crtfm"
         ;;
  x86_64-*-mingw*)