[-- Attachment #1: Type: text/plain, Size: 961 bytes --] Dear Team, While trying to configure musl package on power arch(ppc64le) we are getting the below configure issue Reflink:- musl - musl - an implementation of the standard library for Linux-based systems (musl-libc.org)<https://git.musl-libc.org/cgit/musl/> Musl version 1.2.1:- Download link:- https://git.musl-libc.org/cgit/musl/snapshot/musl-1.2.1.tar.gz Detailed issue. Architure used:- uname -a Linux f1a836851f19 4.18.0-348.12.2.el8_5.ppc64le #1 SMP Mon Jan 17 06:49:39 EST 2022 ppc64le ppc64le ppc64le GNU/Linux Issue faced:- ./configure ........... checking preprocessor condition _CALL_ELF == 2... true checking preprocessor condition __LITTLE_ENDIAN__... true checking preprocessor condition _SOFT_FLOAT... false configured for powerpc64 variant: powerpc64le checking whether compiler's long double definition matches float.h... no ./configure: error: unsupported long double type Regards, Ankit Paraskar [-- Attachment #2: Type: text/html, Size: 3348 bytes --]
Are you using a toolchain that now supports IEEE128 long double? Musl
was not going to support the previous IBM 128 double-double long
double format. Now that PPC64 Linux supports IEEE128, there may be a
mismatch.
Thanks, David
On Fri, Jun 10, 2022 at 7:15 AM Ankit Paraskar <Ankit.Paraskar@ibm.com> wrote:
>
> Dear Team,
>
>
>
> While trying to configure musl package on power arch(ppc64le) we are getting the below configure issue
>
>
>
> Reflink:-
>
> musl - musl - an implementation of the standard library for Linux-based systems (musl-libc.org)
>
>
>
> Musl version 1.2.1:-
>
> Download link:- https://git.musl-libc.org/cgit/musl/snapshot/musl-1.2.1.tar.gz
>
>
>
>
>
> Detailed issue.
>
>
>
> Architure used:-
>
> uname -a
>
> Linux f1a836851f19 4.18.0-348.12.2.el8_5.ppc64le #1 SMP Mon Jan 17 06:49:39 EST 2022 ppc64le ppc64le ppc64le GNU/Linux
>
>
>
>
>
> Issue faced:-
>
> ./configure
>
> ...........
>
> checking preprocessor condition _CALL_ELF == 2... true
>
> checking preprocessor condition __LITTLE_ENDIAN__... true
>
> checking preprocessor condition _SOFT_FLOAT... false
>
> configured for powerpc64 variant: powerpc64le
>
> checking whether compiler's long double definition matches float.h... no
>
> ./configure: error: unsupported long double type
>
>
>
>
>
>
>
> Regards,
>
> Ankit Paraskar
>
>
On Fri, Jun 10, 2022 at 11:14:03AM +0000, Ankit Paraskar wrote:
> Dear Team,
>
> While trying to configure musl package on power arch(ppc64le) we are getting the below configure issue
>
> Reflink:-
> musl - musl - an implementation of the standard library for Linux-based systems (musl-libc.org)<https://git.musl-libc.org/cgit/musl/>
>
> Musl version 1.2.1:-
> Download link:- https://git.musl-libc.org/cgit/musl/snapshot/musl-1.2.1.tar..gz
>
>
> Detailed issue.
>
> Architure used:-
> uname -a
> Linux f1a836851f19 4.18.0-348.12.2.el8_5.ppc64le #1 SMP Mon Jan 17 06:49:39 EST 2022 ppc64le ppc64le ppc64le GNU/Linux
>
>
> Issue faced:-
> ../configure
> ............
> checking preprocessor condition _CALL_ELF == 2... true
> checking preprocessor condition __LITTLE_ENDIAN__... true
> checking preprocessor condition _SOFT_FLOAT... false
> configured for powerpc64 variant: powerpc64le
> checking whether compiler's long double definition matches float.h... no
> ../configure: error: unsupported long double type
You need a toolchain built for powerpc64le-linux-musl target where
long double has the same size/range/representation as double. The musl
ABI for ppc64 uses this because the legacy IBM double-double type for
long double does not match our policy of only using IEEE-conforming
types for long double (which various code in musl depends on), and
support for IEEE quad on ppc64 is limited (even moreso at the time the
port was added) and would have tied us/users to particular compilers
that support it.
If you're trying to use the musl-gcc wrapper on a glibc host, you
~may~ be able to make it work by setting the CC variable to something
to override, like CC="gcc -mno-long-double-128" or CC="gcc
-mlong-double-64" (not sure which if either of those works; the GCC
docs seem to be omitting it but referring to it elsewhere).
Rich
[-- Attachment #1: Type: text/plain, Size: 2542 bytes --] Thank you for the suggestion , this really clarified the error. Would also appreciate to share information on code contribution for musl repo, aspect of CLA Contributor License Agreement - Wikipedia<https://en.wikipedia.org/wiki/Contributor_License_Agreement> requirements. Thanks, Ankit P Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Rich Felker<mailto:dalias@libc.org> Sent: 10 June 2022 18:15 To: Ankit Paraskar<mailto:Ankit.Paraskar@ibm.com> Cc: musl@lists.openwall.com<mailto:musl@lists.openwall.com> Subject: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for must repo On Fri, Jun 10, 2022 at 11:14:03AM +0000, Ankit Paraskar wrote: > Dear Team, > > While trying to configure musl package on power arch(ppc64le) we are getting the below configure issue > > Reflink:- > musl - musl - an implementation of the standard library for Linux-based systems (musl-libc.org)<https://git.musl-libc.org/cgit/musl/ > > > Musl version 1.2.1:- > Download link:- https://git.musl-libc.org/cgit/musl/snapshot/musl-1.2.1.tar..gz > > > Detailed issue. > > Architure used:- > uname -a > Linux f1a836851f19 4.18.0-348.12.2.el8_5.ppc64le #1 SMP Mon Jan 17 06:49:39 EST 2022 ppc64le ppc64le ppc64le GNU/Linux > > > Issue faced:- > ../configure > ............ > checking preprocessor condition _CALL_ELF == 2... true > checking preprocessor condition __LITTLE_ENDIAN__... true > checking preprocessor condition _SOFT_FLOAT... false > configured for powerpc64 variant: powerpc64le > checking whether compiler's long double definition matches float.h... no > ../configure: error: unsupported long double type You need a toolchain built for powerpc64le-linux-musl target where long double has the same size/range/representation as double. The musl ABI for ppc64 uses this because the legacy IBM double-double type for long double does not match our policy of only using IEEE-conforming types for long double (which various code in musl depends on), and support for IEEE quad on ppc64 is limited (even moreso at the time the port was added) and would have tied us/users to particular compilers that support it. If you're trying to use the musl-gcc wrapper on a glibc host, you ~may~ be able to make it work by setting the CC variable to something to override, like CC="gcc -mno-long-double-128" or CC="gcc -mlong-double-64" (not sure which if either of those works; the GCC docs seem to be omitting it but referring to it elsewhere). Rich [-- Attachment #2: Type: text/html, Size: 4751 bytes --]
[-- Attachment #1: Type: text/plain, Size: 2903 bytes --] ++ Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Ankit Paraskar<mailto:Ankit.Paraskar@ibm.com> Sent: 29 June 2022 17:50 To: Rich Felker<mailto:dalias@libc.org> Cc: musl@lists.openwall.com<mailto:musl@lists.openwall.com> Subject: RE: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for musl repo Thank you for the suggestion , this really clarified the error. Would also appreciate to share information on code contribution for musl repo, aspect of CLA Contributor License Agreement - Wikipedia<https://en.wikipedia.org/wiki/Contributor_License_Agreement> requirements. Thanks, Ankit P Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Rich Felker<mailto:dalias@libc.org> Sent: 10 June 2022 18:15 To: Ankit Paraskar<mailto:Ankit.Paraskar@ibm.com> Cc: musl@lists.openwall.com<mailto:musl@lists.openwall.com> Subject: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for must repo On Fri, Jun 10, 2022 at 11:14:03AM +0000, Ankit Paraskar wrote: > Dear Team, > > While trying to configure musl package on power arch(ppc64le) we are getting the below configure issue > > Reflink:- > musl - musl - an implementation of the standard library for Linux-based systems (musl-libc.org)<https://git.musl-libc.org/cgit/musl/ > > > Musl version 1.2.1:- > Download link:- https://git.musl-libc.org/cgit/musl/snapshot/musl-1.2.1.tar..gz > > > Detailed issue. > > Architure used:- > uname -a > Linux f1a836851f19 4.18.0-348.12.2.el8_5.ppc64le #1 SMP Mon Jan 17 06:49:39 EST 2022 ppc64le ppc64le ppc64le GNU/Linux > > > Issue faced:- > ../configure > ............ > checking preprocessor condition _CALL_ELF == 2... true > checking preprocessor condition __LITTLE_ENDIAN__... true > checking preprocessor condition _SOFT_FLOAT... false > configured for powerpc64 variant: powerpc64le > checking whether compiler's long double definition matches float.h... no > ../configure: error: unsupported long double type You need a toolchain built for powerpc64le-linux-musl target where long double has the same size/range/representation as double. The musl ABI for ppc64 uses this because the legacy IBM double-double type for long double does not match our policy of only using IEEE-conforming types for long double (which various code in musl depends on), and support for IEEE quad on ppc64 is limited (even moreso at the time the port was added) and would have tied us/users to particular compilers that support it. If you're trying to use the musl-gcc wrapper on a glibc host, you ~may~ be able to make it work by setting the CC variable to something to override, like CC="gcc -mno-long-double-128" or CC="gcc -mlong-double-64" (not sure which if either of those works; the GCC docs seem to be omitting it but referring to it elsewhere). Rich [-- Attachment #2: Type: text/html, Size: 5620 bytes --]
[-- Attachment #1: Type: text/plain, Size: 3422 bytes --] Dear Team/Rich, Any updates ? Regards, Ankit Paraskar Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Ankit Paraskar<mailto:Ankit.Paraskar@ibm.com> Sent: 05 July 2022 19:24 To: Rich Felker<mailto:dalias@libc.org> Cc: musl@lists.openwall.com<mailto:musl@lists.openwall.com>; Gerrit Huizenga [Notes]<mailto:gerrit@us.ibm.com>; Prashant Khoje<mailto:Prashant.Khoje@ibm.com> Subject: RE: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for musl repo ++ Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Ankit Paraskar<mailto:Ankit.Paraskar@ibm.com> Sent: 29 June 2022 17:50 To: Rich Felker<mailto:dalias@libc.org> Cc: musl@lists.openwall.com<mailto:musl@lists.openwall.com> Subject: RE: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for musl repo Thank you for the suggestion , this really clarified the error. Would also appreciate to share information on code contribution for musl repo, aspect of CLA Contributor License Agreement - Wikipedia<https://en.wikipedia.org/wiki/Contributor_License_Agreement> requirements. Thanks, Ankit P Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows From: Rich Felker<mailto:dalias@libc.org> Sent: 10 June 2022 18:15 To: Ankit Paraskar<mailto:Ankit.Paraskar@ibm.com> Cc: musl@lists.openwall.com<mailto:musl@lists.openwall.com> Subject: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for must repo On Fri, Jun 10, 2022 at 11:14:03AM +0000, Ankit Paraskar wrote: > Dear Team, > > While trying to configure musl package on power arch(ppc64le) we are getting the below configure issue > > Reflink:- > musl - musl - an implementation of the standard library for Linux-based systems (musl-libc.org)<https://git.musl-libc.org/cgit/musl/ > > > Musl version 1.2.1:- > Download link:- https://git.musl-libc.org/cgit/musl/snapshot/musl-1.2.1.tar..gz > > > Detailed issue. > > Architure used:- > uname -a > Linux f1a836851f19 4.18.0-348.12.2.el8_5.ppc64le #1 SMP Mon Jan 17 06:49:39 EST 2022 ppc64le ppc64le ppc64le GNU/Linux > > > Issue faced:- > ../configure > ............ > checking preprocessor condition _CALL_ELF == 2... true > checking preprocessor condition __LITTLE_ENDIAN__... true > checking preprocessor condition _SOFT_FLOAT... false > configured for powerpc64 variant: powerpc64le > checking whether compiler's long double definition matches float.h... no > ../configure: error: unsupported long double type You need a toolchain built for powerpc64le-linux-musl target where long double has the same size/range/representation as double. The musl ABI for ppc64 uses this because the legacy IBM double-double type for long double does not match our policy of only using IEEE-conforming types for long double (which various code in musl depends on), and support for IEEE quad on ppc64 is limited (even moreso at the time the port was added) and would have tied us/users to particular compilers that support it. If you're trying to use the musl-gcc wrapper on a glibc host, you ~may~ be able to make it work by setting the CC variable to something to override, like CC="gcc -mno-long-double-128" or CC="gcc -mlong-double-64" (not sure which if either of those works; the GCC docs seem to be omitting it but referring to it elsewhere). Rich [-- Attachment #2: Type: text/html, Size: 6863 bytes --]
On Thu Jul 21, 2022 at 7:46 AM CEST, Ankit Paraskar wrote:
> Dear Team/Rich,
>
> Any updates ?
>
> Regards,
> Ankit Paraskar
>
there is no CLA needed to submit patches to musl.
Ankit,
The problem is with the toolchain that you are using, not with Musl
Libc. The Musl community is not going to fix it for you. Please stop
waiting for them or implying that they should fix the problem with
your build environment.
Thanks, David
On Thu, Jul 21, 2022 at 8:48 AM Ankit Paraskar <Ankit.Paraskar@ibm.com> wrote:
>
> Dear Team/Rich,
>
> Any updates ?
>
>
>
> Regards,
>
> Ankit Paraskar
>
>
>
>
>
> Sent from Mail for Windows
>
>
>
> From: Ankit Paraskar
> Sent: 05 July 2022 19:24
> To: Rich Felker
> Cc: musl@lists.openwall.com; Gerrit Huizenga [Notes]; Prashant Khoje
> Subject: RE: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for musl repo
>
>
>
> ++
>
>
>
> Sent from Mail for Windows
>
>
>
> From: Ankit Paraskar
> Sent: 29 June 2022 17:50
> To: Rich Felker
> Cc: musl@lists.openwall.com
> Subject: RE: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for musl repo
>
>
>
> Thank you for the suggestion , this really clarified the error.
>
> Would also appreciate to share information on code contribution for musl repo, aspect of CLA Contributor License Agreement - Wikipedia requirements.
>
>
>
> Thanks,
>
> Ankit P
>
>
>
>
>
> Sent from Mail for Windows
>
>
>
> From: Rich Felker
> Sent: 10 June 2022 18:15
> To: Ankit Paraskar
> Cc: musl@lists.openwall.com
> Subject: [EXTERNAL] Re: [musl] Configuration gives error on ppc64le arch for must repo
>
>
>
> On Fri, Jun 10, 2022 at 11:14:03AM +0000, Ankit Paraskar wrote:
> > Dear Team,
> >
> > While trying to configure musl package on power arch(ppc64le) we are getting the below configure issue
> >
> > Reflink:-
> > musl - musl - an implementation of the standard library for Linux-based systems (musl-libc.org)<https://git.musl-libc.org/cgit/musl/ >
> >
> > Musl version 1.2.1:-
> > Download link:- https://git.musl-libc.org/cgit/musl/snapshot/musl-1.2.1.tar..gz
> >
> >
> > Detailed issue.
> >
> > Architure used:-
> > uname -a
> > Linux f1a836851f19 4.18.0-348.12.2.el8_5.ppc64le #1 SMP Mon Jan 17 06:49:39 EST 2022 ppc64le ppc64le ppc64le GNU/Linux
> >
> >
> > Issue faced:-
> > ../configure
> > ............
> > checking preprocessor condition _CALL_ELF == 2... true
> > checking preprocessor condition __LITTLE_ENDIAN__... true
> > checking preprocessor condition _SOFT_FLOAT... false
> > configured for powerpc64 variant: powerpc64le
> > checking whether compiler's long double definition matches float.h... no
> > ../configure: error: unsupported long double type
>
> You need a toolchain built for powerpc64le-linux-musl target where
> long double has the same size/range/representation as double. The musl
> ABI for ppc64 uses this because the legacy IBM double-double type for
> long double does not match our policy of only using IEEE-conforming
> types for long double (which various code in musl depends on), and
> support for IEEE quad on ppc64 is limited (even moreso at the time the
> port was added) and would have tied us/users to particular compilers
> that support it.
>
> If you're trying to use the musl-gcc wrapper on a glibc host, you
> ~may~ be able to make it work by setting the CC variable to something
> to override, like CC="gcc -mno-long-double-128" or CC="gcc
> -mlong-double-64" (not sure which if either of those works; the GCC
> docs seem to be omitting it but referring to it elsewhere).
>
> Rich
>
>
>
>
>
>