jasper / jiv working under powerpc?

[Roland Stigge]

Hi,

as the Debian maintainer of jasper (JPEG 2000 library), I got a report
from a MacOS X user that jiv (the JPEG 2000 image viewer) segfaults
under MacOS X. Unfortunately, I don't have a powerpc (sponsors welcome
:), and the Debian Developer machines don't help here (bruckner is
currently unreachable, malo is restricted, and voltaire doesn't forward
X11).

Would someone please try jiv and tell me if the Debian version works? To
do this, install the package libjasper-runtime, create a JPEG 2000
image, e.g.:

  $ jasper -f test.jpg -F test.jp2 -T jp2 -O rate=0.01

then, run jiv:

  $ jiv test.jp2

Please CC me privately, I'm not subscribed.

Thanks!

bye,
  Roland

Re: jasper / jiv working under powerpc?

[Roland Stigge]

On Thu, 2004-07-22 at 10:39, Arne Caspari wrote:
> > Would someone please try jiv and tell me if the Debian version works?

> It does not segfault on my iBook G4. But I tried with a 1284x960 sized jpeg 
> file first and it did not gave a correct result: 
> The image was overlaid by strange vertical color bars. A 640x480 image 
> performed better but had still a blue vertical bar on the left side. 
> 
> Also jiv used up 100% processor time and flickered a lot. 

Thanks.

Imagemagick now uses jasper. Does /usr/bin/display yield the same
problem?

bye,
  Roland

Re: powerpcspe in Debian

[Roland Stigge]

On 30/10/12 21:20, Roger Leigh wrote:
> On Tue, Oct 30, 2012 at 06:44:24PM +0100, Sebastian Andrzej Siewior wrote:
>> On Tue, Oct 30, 2012 at 05:50:09PM +0100, Roland Stigge wrote:
>>> Yes, I'm interested.
>>>
>>> Do I need to setup a buildd? Or is there a buildd running that is
>>> supposed to build the current ~50 packages in "Needs-Build"? What would
>>> be the next steps?
>>
>> sbuild/buildd would be a good start. I could look for my old RFS if you want.
>> I used the DM layer with snapshots as a setup with around 10 or 20GiB space.
>> Anyway. The packages listed in Needs-Build are hardly interresting.
>> You need to look at BD-Uninstallable. Look at the list [0]. Most packages are
>> stuck due to debhelper. Look what is wrong, my guess is that you can't 
>> install
>> python.
>> For  manuall fixing you need a minimal build environment and I used always
>> schroot for that.
> 
> If you have any sbuild/schroot questions, please do give me a shout
> if you get stuck.
> 
> While schroot does support LVM snapshots, you might also want to
> look at using Btrfs snapshots (faster and more reliable, IME), or
> an overlay filesystem like aufs.  schroot supports both of these
> for snapshots as well.
> 
> Example configuration for Btrfs:
> 
> [sid-amd64-sbuild]
> aliases=unstable-amd64-sbuild
> type=btrfs-snapshot
> description=Debian sid/amd64 autobuilder
> btrfs-source-subvolume=/srv/chroots-btrfs/sid
> btrfs-snapshot-directory=/srv/chroots-btrfs/snapshot
> groups=root,sbuild
> root-groups=root,sbuild
> source-groups=root,sbuild
> source-root-groups=root,sbuild
> profile=sbuild
> command-prefix=eatmydata
> 
> /srv/chroots-btrfs is a Btrfs filesystem, with /srv/chroots-btrfs/sid
> being a Btrfs subvolume containing the build environment.

Good - thanks for your help!

Will see what I can do. Will be on VAC for the next 2 weeks. Will
probably continue Debian duties, but won't have access to a powerpcspe
during this time.

Will get back when I have packages ready.

Can you please try to not further automatically decrease the number of
available powerpcspe packages during this time? :-)

Thanks in advance,

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/50904dca.7080...@antcom.de

Re: powerpcspe in Debian

[Roland Stigge]

On 10/30/2012 11:25 PM, Kyle Moffett wrote:
> On Tue, Oct 30, 2012 at 2:59 PM, Roland Stigge  wrote:
>> Will see what I can do. Will be on VAC for the next 2 weeks. Will
>> probably continue Debian duties, but won't have access to a powerpcspe
>> during this time.
>>
>> Will get back when I have packages ready.
>>
>> Can you please try to not further automatically decrease the number of
>> available powerpcspe packages during this time? :-)
> 
> I think you misunderstand what causes packages to become "unavailable".
> 
> The official Debian ports are making forward progress and uploading
> new packages, but nobody is building powerpcspe packages to match.  As
> the old versions of the "architecture all" packages are aged out of
> the regular Debian repository, anything in the powerpcspe port that
> depends on that version of that package becomes uninstallable.

Of course. Thanks!

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/5090e44a.1040...@antcom.de

Re: powerpcspe in Debian

[Roland Stigge]

Hi Kyle, Roger and Sebastian,

On 30/10/12 21:20, Roger Leigh wrote:
>>> Do I need to setup a buildd? Or is there a buildd running that is
>>> supposed to build the current ~50 packages in "Needs-Build"? What would
>>> be the next steps?
>>
>> sbuild/buildd would be a good start. I could look for my old RFS if you want.
>> I used the DM layer with snapshots as a setup with around 10 or 20GiB space.
>> Anyway. The packages listed in Needs-Build are hardly interresting.
>> You need to look at BD-Uninstallable. Look at the list [0]. Most packages are
>> stuck due to debhelper. Look what is wrong, my guess is that you can't 
>> install
>> python.
>> For  manuall fixing you need a minimal build environment and I used always
>> schroot for that.
> 
> If you have any sbuild/schroot questions, please do give me a shout
> if you get stuck.
> 
> While schroot does support LVM snapshots, you might also want to
> look at using Btrfs snapshots (faster and more reliable, IME), or
> an overlay filesystem like aufs.  schroot supports both of these
> for snapshots as well.

I have a working rootfs and an sbuild chroot running now. :-)

Starting to submit patches to the BTS now. Until now, I just needed to
change linux (for linux-libc-dev), libffi and ltrace. The rest went fine
with forced installation.

I uploaded some important packages to

deb http://people.debian.org/~stigge/deb-powerpcspe/ ./

This enables dist-upgrading the rootfs image available at

http://wiki.debian.org/PowerPCSPEPort

to a working state and also to install build-essential for an sbuild
schroot.

How can we setup a more systematic process now, e.g. with buildd +
wanna-build access?

Thanks in advance,

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/50997a6f.3010...@antcom.de

Re: powerpcspe in Debian

[Roland Stigge]

Hi Sebastian,

On 09/11/12 13:43, Sebastian Andrzej Siewior wrote:
>> I uploaded some important packages to
>>
>> deb http://people.debian.org/~stigge/deb-powerpcspe/ ./
> 
> Okay.

As an update, I've set up new hardware now which includes fast disks and
btrfs snapshot schroots for sbuild.

Since the repository is growing, I moved from people.d.o to my server,
so you can use it as an augmented repo for debian-ports, e.g.:

deb http://ftp.de.debian.org/debian-ports/ sid main
deb http://www.antcom.de/powerpcspe/ unstable main

I'm updating this one constantly now with new built packages from the
BD-Uninstallable list. Including source where I needed to change it.

>> How can we setup a more systematic process now, e.g. with buildd +
>> wanna-build access?
> 
> Let me please look at those packages over the weekend first.
> 
> Can perl/python/gcc be built/installed?

Python is doing fine, I'm still working on perl and gcc (which are
installable, but still need builds). Also still owing the patch for
linux (for linux-libc-dev).

Will keep you updated.

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/509d1003.3060...@debian.org

Re: powerpcspe in Debian

[Roland Stigge]

Hi,

I now have current gcc, eglibc, perl and python building and used on
powerpcspe.

Some Detail issues:

At http://antcom.de/powerpcspe/patches/ there are my patches for
gcc-4.6, gcc-4.7 and gcc-defaults.

gcc-4.7 only builds if I adjust the build environment further, see
gcc-4.7.sh and http://bugs.debian.org/637232 etc.

I'd like to hear some comments about those patches before I file them
agaist gcc-* packages.

I'm especially wondering about our multiarch approach. Typically,
multiarch archs have >=2 archs (e.g. powerpc and powerpc64). However,
powerpcspe doesn't seem to have a sibling arch, therefore being "single"
multiarch, basically doing most stuff like the powerpc port for
simplicity. Is there anything that I should keep in mind here or that
I'm missing here?

The growing powerpcspe repo is still available at my server, augmenting
debian-ports:

deb http://ftp.de.debian.org/debian-ports/ sid main
deb http://www.antcom.de/powerpcspe/ unstable main

One question about the boost library: In debian-ports, there is already
version 1.49.1.1, but in common unstable there is only 1.49.0.1. Is
there an important reason for this? Otherwise, I would just stay with
the version 1.49.0.1 built from unstable.

Thanks in advance,

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/50a38bc4.1050...@antcom.de

Re: powerpcspe in Debian

[Roland Stigge]

Hi,

On 19/11/12 21:58, Sebastian Andrzej Siewior wrote:
>> At http://antcom.de/powerpcspe/patches/ there are my patches for
>> gcc-4.6, gcc-4.7 and gcc-defaults.
> 
> Please remove *-multilib support from your patches, it should build without
> it.

Already done, as you can see in the patches in the powerpcspe specific
bugs and discussed with Matthias.

>> gcc-4.7 only builds if I adjust the build environment further, see
>> gcc-4.7.sh and http://bugs.debian.org/637232 etc.
> 
> gcc-4.7.sh should go. You should build a compiler with multiarch support,
> recompile libc and friends in order to place the libraries there. That was
> imho the point where I took a break :) So that could end up in work…

Yes, that should work now also. Stuff is rebuilding currently anyway.
About 4000 binary packages in my repo currently. Will get back to you
when questions occur.

>> The growing powerpcspe repo is still available at my server, augmenting
>> debian-ports:
>>
>> deb http://ftp.de.debian.org/debian-ports/ sid main
>> deb http://www.antcom.de/powerpcspe/ unstable main
>>
>> One question about the boost library: In debian-ports, there is already
>> version 1.49.1.1, but in common unstable there is only 1.49.0.1. Is
>> there an important reason for this? Otherwise, I would just stay with
>> the version 1.49.0.1 built from unstable.
> 
> Could you be more specific here please? I see 1.49.0-3.1 at
>  http://packages.qa.debian.org/b/boost1.49.html
>  http://buildd.debian-ports.org/status/package.php?p=boost1.49
> which looks fine. The boost-defaults package is 1.49.0.1. I don't see 1.49.1.1
> anywhere.

Well, boost1.49 is actually not there, but boost-defaults 1.49.1.1 :

http://ftp.de.debian.org/debian-ports/pool-powerpcspe/main/b/boost-defaults/

while only 1.49.0.1 is in unstable/experimental source. That's what I
meant. Sorry for the confusion.

Thanks,

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/50aab9db.1020...@antcom.de

Generic solution for overflowing GOT table

[Roland Stigge]

Hi,

on powerpc (and powerpcspe) in Debian, there are several packages FTBFS
like this (package quickfix just as an example):

make[4]: Entering directory 
`/build/buildd-quickfix_1.13.3+dfsg-4-powerpc-X8NgWA/quickfix-1.13.3+dfsg/src/python'
/bin/bash ../../libtool --tag=CXX   --mode=compile /usr/bin/g++ -DHAVE_CONFIG_H 
-I. -I../.. -I. -I.. -I../..-O3 -Wall -ansi -Wpointer-arith -Wwrite-strings 
-I/usr/include/boost  -I/usr/include/mysql -I/usr/mysql/include -I/usr/include 
-I/usr/include/postgresql -I/usr/include/libxml2  -I/usr/include/python2.7 
-I/usr/lib/ruby/1.8/powerpc-linux -MT QuickfixPython.lo -MD -MP -MF 
.deps/QuickfixPython.Tpo -c -o QuickfixPython.lo QuickfixPython.cpp
libtool: compile:  /usr/bin/g++ -DHAVE_CONFIG_H -I. -I../.. -I. -I.. -I../.. 
-O3 -Wall -ansi -Wpointer-arith -Wwrite-strings -I/usr/include/boost 
-I/usr/include/mysql -I/usr/mysql/include -I/usr/include 
-I/usr/include/postgresql -I/usr/include/libxml2 -I/usr/include/python2.7 
-I/usr/lib/ruby/1.8/powerpc-linux -MT QuickfixPython.lo -MD -MP -MF 
.deps/QuickfixPython.Tpo -c QuickfixPython.cpp  -fPIC -DPIC -o 
.libs/QuickfixPython.o
QuickfixPython.cpp: In function 'PyObject* 
_wrap_DataDictionary_getGroup(PyObject*, PyObject*)':
QuickfixPython.cpp:306005:9: warning: unused variable 'res' [-Wunused-variable]
QuickfixPython.cpp: In function 'PyObject* 
_wrap_DataDictionary_getFieldTag(PyObject*, PyObject*)':
QuickfixPython.cpp:304789:96: warning: 'temp3' may be used uninitialized in 
this function [-Wuninitialized]
QuickfixPython.cpp: In function 'PyObject* 
_wrap_DataDictionary_getGroup(PyObject*, PyObject*)':
QuickfixPython.cpp:305978:18: warning: 'temp5' may be used uninitialized in 
this function [-Wuninitialized]
QuickfixPython.cpp:305999:96: warning: 'temp4' may be used uninitialized in 
this function [-Wuninitialized]
/tmp/ccj2yRZf.s: Assembler messages:
/tmp/ccj2yRZf.s:1756016: Error: operand out of range (0x8000 is not 
between 0x8000 and 0x7fff)
/tmp/ccj2yRZf.s:1756588: Error: operand out of range (0x8018 is not 
between 0x8000 and 0x7fff)
/tmp/ccj2yRZf.s:1756627: Error: operand out of range (0x8004 is not 
between 0x8000 and 0x7fff)
[...]

According to others struggling publicly on the net, there are several
workarounds available for similar issues.

However, things like -fsection-anchors, -mon-toc, etc. don't worked
for me with quickfix (-fPIC necessary for lib) so I'm wondering if
there is a generic solution for this powerpc specific issue available,
instead of fixing all those packages individually?

(Please CC me.)

Thanks in advance,

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/513dcc6e.7070...@antcom.de

Re: Generic solution for overflowing GOT table

[Roland Stigge]

On 03/11/2013 01:22 PM, Roland Stigge wrote:
> make[4]: Entering directory 
> `/build/buildd-quickfix_1.13.3+dfsg-4-powerpc-X8NgWA/quickfix-1.13.3+dfsg/src/python'
> /bin/bash ../../libtool --tag=CXX   --mode=compile /usr/bin/g++ 
> -DHAVE_CONFIG_H -I. -I../.. -I. -I.. -I../..-O3 -Wall -ansi 
> -Wpointer-arith -Wwrite-strings -I/usr/include/boost  -I/usr/include/mysql 
> -I/usr/mysql/include -I/usr/include -I/usr/include/postgresql 
> -I/usr/include/libxml2  -I/usr/include/python2.7 
> -I/usr/lib/ruby/1.8/powerpc-linux -MT QuickfixPython.lo -MD -MP -MF 
> .deps/QuickfixPython.Tpo -c -o QuickfixPython.lo QuickfixPython.cpp
> libtool: compile:  /usr/bin/g++ -DHAVE_CONFIG_H -I. -I../.. -I. -I.. -I../.. 
> -O3 -Wall -ansi -Wpointer-arith -Wwrite-strings -I/usr/include/boost 
> -I/usr/include/mysql -I/usr/mysql/include -I/usr/include 
> -I/usr/include/postgresql -I/usr/include/libxml2 -I/usr/include/python2.7 
> -I/usr/lib/ruby/1.8/powerpc-linux -MT QuickfixPython.lo -MD -MP -MF 
> .deps/QuickfixPython.Tpo -c QuickfixPython.cpp  -fPIC -DPIC -o 
> .libs/QuickfixPython.o
> QuickfixPython.cpp: In function 'PyObject* 
> _wrap_DataDictionary_getGroup(PyObject*, PyObject*)':
> QuickfixPython.cpp:306005:9: warning: unused variable 'res' 
> [-Wunused-variable]
> QuickfixPython.cpp: In function 'PyObject* 
> _wrap_DataDictionary_getFieldTag(PyObject*, PyObject*)':
> QuickfixPython.cpp:304789:96: warning: 'temp3' may be used uninitialized in 
> this function [-Wuninitialized]
> QuickfixPython.cpp: In function 'PyObject* 
> _wrap_DataDictionary_getGroup(PyObject*, PyObject*)':
> QuickfixPython.cpp:305978:18: warning: 'temp5' may be used uninitialized in 
> this function [-Wuninitialized]
> QuickfixPython.cpp:305999:96: warning: 'temp4' may be used uninitialized in 
> this function [-Wuninitialized]
> /tmp/ccj2yRZf.s: Assembler messages:
> /tmp/ccj2yRZf.s:1756016: Error: operand out of range (0x8000 is 
> not between 0x8000 and 0x7fff)
> /tmp/ccj2yRZf.s:1756588: Error: operand out of range (0x8018 is 
> not between 0x8000 and 0x7fff)
> /tmp/ccj2yRZf.s:1756627: Error: operand out of range (0x8004 is 
> not between 0x8000 and 0x7fff)
> [...]

Just as a small follow-up: In the above case, -fpic instead of -fPIC
fixes the issue. Is this acceptable in Debian for shared libs?

Thanks in advance,

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/513e7026.5060...@antcom.de

Re: Re: Generic solution for overflowing GOT table

[Roland Stigge]

Hi,

you wrote:
> On Tue, Mar 12, 2013 at 01:00:38AM +0100, Roland Stigge wrote:
>> Just as a small follow-up: In the above case, -fpic instead of -fPIC
>> fixes the issue. Is this acceptable in Debian for shared libs?
>
> This reduces the size of the file and therefor works around the too
> large offsets. However it is not acceptable.

Thanks for your note. In the pursuit for the right solution for powerpc
here, can you please tell me, or point me to some ressource about, why?
Maybe there is some other general strategy?

For reference, I also described the issue with some more detail about
this issue here:

http://lists.debian.org/debian-devel/2013/03/msg00223.html

Thanks in advance,

Roland


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/51410331.7060...@antcom.de

Re: powerpcspe status? toolchain issues

[Roland Stigge]

Hi Helmut,

sorry for not fixing earlier. Currently, I don't have the time, but will
need to fix within the next months, because I need powerpcspe in
projects for the next years.

Please decide yourself if you test-rebuild it for now. It will be broken
for the next few days at least.

Thanks for testing!

Roland


On 06/11/2014 05:07 PM, Helmut Grohne wrote:
> Hi,
> 
> I would like to know what the status is for powerpcspe. The build log
> statistics don't look too promising:
> http://buildd.debian-ports.org/stats/graph-big.png
> 
> With gcc-4.8 cross compiling eglibc 2.19 started to fail:
> https://jenkins.debian.net/view/rebootstrap/job/rebootstrap_powerpcspe_gcc48_nobiarch/5/console
> 
> gcc-4.9 doesn't even build a stage1, but runs into ICEs instead.
> https://jenkins.debian.net/view/rebootstrap/job/rebootstrap_powerpcspe_gcc49_nobiarch/5/console
> 
> Are there intentions to fix this or should I stop testing
> cross-buildability of powerpcspe with rebootstrap[1]?
> 
> Please CC me in replies.
> 
> Helmut
> 
> [1] https://wiki.debian.org/HelmutGrohne/rebootstrap
> 


-- 
To UNSUBSCRIBE, email to debian-powerpc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/53987721.6090...@debian.org

Glibc on powerpcspe

[Roland Stigge]

Hi!

I only recently started the transition from eglibc to glibc on
powerpcspe. See the attached patch.

Unfortunately, this still contains some rough architecture specific
changes: Besides conditionals around powerpc (vs. powerpcspe) specific
asm which should be ok, I just deleted some powerpc specific files (e.g.
s_fabs.S in sysdeps/powerpc/fpu/ and sysdeps/powerpc/powerpc32/fpu/) to
force generic implementations of those functions instead of powerpc
specific.

This is of course not a solution to patch into the standard Debian package.

Now my question to more experienced glibc developers: What would be the
best way here to do those changes powerpcspe/e500 compatibly and at the
same time keep the default powerpc asm files etc.?

Thanks in advance,

Roland
Index: glibc-2.19/sysdeps/powerpc/powerpc32/fpu/s_copysign.S
===
--- glibc-2.19.orig/sysdeps/powerpc/powerpc32/fpu/s_copysign.S	2014-11-10 14:30:05.192139074 +0100
+++ /dev/null	1970-01-01 00:00:00.0 +
@@ -1,59 +0,0 @@
-/* Copy a sign bit between floating-point values.
-   Copyright (C) 1997-2014 Free Software Foundation, Inc.
-   This file is part of the GNU C Library.
-
-   The GNU C Library is free software; you can redistribute it and/or
-   modify it under the terms of the GNU Lesser General Public
-   License as published by the Free Software Foundation; either
-   version 2.1 of the License, or (at your option) any later version.
-
-   The GNU C Library is distributed in the hope that it will be useful,
-   but WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-   Lesser General Public License for more details.
-
-   You should have received a copy of the GNU Lesser General Public
-   License along with the GNU C Library; if not, see
-   .  */
-
-/* This has been coded in assembler because GCC makes such a mess of it
-   when it's coded in C.  */
-
-#include 
-#include 
-
-ENTRY(__copysign)
-/* double [f1] copysign (double [f1] x, double [f2] y);
-   copysign(x,y) returns a value with the magnitude of x and
-   with the sign bit of y.  */
-	stwu	r1,-16(r1)
-	cfi_adjust_cfa_offset (16)
-	stfd	fp2,8(r1)
-	lwz	r3,8+HIWORD(r1)
-	cmpwi   r3,0
-	addir1,r1,16
-	cfi_adjust_cfa_offset (-16)
-	blt L(0)
-	fabsfp1,fp1
-	blr
-L(0):	fnabs   fp1,fp1
-	blr
-	END (__copysign)
-
-weak_alias (__copysign,copysign)
-
-/* It turns out that it's safe to use this code even for single-precision.  */
-weak_alias (__copysign,copysignf)
-strong_alias(__copysign,__copysignf)
-
-#ifdef NO_LONG_DOUBLE
-weak_alias (__copysign,copysignl)
-strong_alias(__copysign,__copysignl)
-#endif
-#ifdef IS_IN_libm
-# if LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)
-compat_symbol (libm, __copysign, copysignl, GLIBC_2_0)
-# endif
-#elif LONG_DOUBLE_COMPAT(libc, GLIBC_2_0)
-compat_symbol (libc, __copysign, copysignl, GLIBC_2_0)
-#endif
Index: glibc-2.19/sysdeps/powerpc/powerpc32/fpu/s_copysignl.S
===
--- glibc-2.19.orig/sysdeps/powerpc/powerpc32/fpu/s_copysignl.S	2014-11-10 14:30:05.192139074 +0100
+++ /dev/null	1970-01-01 00:00:00.0 +
@@ -1,49 +0,0 @@
-/* Copy a sign bit between floating-point values.
-   IBM extended format long double version.
-   Copyright (C) 2006-2014 Free Software Foundation, Inc.
-   This file is part of the GNU C Library.
-
-   The GNU C Library is free software; you can redistribute it and/or
-   modify it under the terms of the GNU Lesser General Public
-   License as published by the Free Software Foundation; either
-   version 2.1 of the License, or (at your option) any later version.
-
-   The GNU C Library is distributed in the hope that it will be useful,
-   but WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-   Lesser General Public License for more details.
-
-   You should have received a copy of the GNU Lesser General Public
-   License along with the GNU C Library; if not, see
-   .  */
-
-#include 
-#include 
-
-ENTRY(__copysignl)
-/* long double [f1,f2] copysign (long double [f1,f2] x, long double [f3,f4] y);
-   copysign(x,y) returns a value with the magnitude of x and
-   with the sign bit of y.  */
-	stwu	r1,-16(r1)
-	cfi_adjust_cfa_offset (16)
-	stfd	fp3,8(r1)
-	fmr	fp0,fp1
-	fabs	fp1,fp1
-	fcmpu	cr7,fp0,fp1
-	lwz	r3,8+HIWORD(r1)
-	cmpwi	cr6,r3,0
-	addi	r1,r1,16
-	cfi_adjust_cfa_offset (-16)
-	beq	cr7,L(0)
-	fneg	fp2,fp2
-L(0):	bgelr	cr6
-	fneg	fp1,fp1
-	fneg	fp2,fp2
-	blr
-END (__copysignl)
-
-#ifdef IS_IN_libm
-long_double_symbol (libm, __copysignl, copysignl)
-#else
-long_double_symbol (libc, __copysignl, copysignl)
-#endif
Index: glibc-2.19/sysdeps/powerpc/fpu/math_ldbl.h
===
--- glibc-2.19.orig/sysdeps/powerpc/fpu