[Buildroot] Update of Buildroot toolchains on autobuild.b.o

[Buildroot] Update of Buildroot toolchains on autobuild.b.o

[Thomas Petazzoni]

Hello,

For information, I have updated the toolchains on autobuild.b.o. Here
are the changes:

 * All existing Buildroot pre-build toolchains have been rebuilt with
   Buildroot 2016.05.

 * Addition of a ARM Cortex-M4 noMMU toolchain.

 * Addition of a m68k 5208 noMMU toolchain.

 * Addition of a m68k 68040 toolchain.

Due to the addition of those new toolchains, for fairly specific
architecture, I'm expecting to see a pretty significant increase in
build failures.

Let me know if there are other toolchain combinations that you would
like to see tested on autobuild.b.o.

Best regards,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] Update of Buildroot toolchains on autobuild.b.o

[Bernd Kuhls]

[posted and mailed]

Thomas Petazzoni <thomas.petazzoni-
wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8@public.gmane.org> wrote in 
news:20160608220152.5b54ea7d at free-electrons.com:

> Let me know if there are other toolchain combinations that you would
> like to see tested on autobuild.b.o.

Hi Thomas,

a lot of multimedia packages (ffmpeg, mplayer & co.) contain support not only 
for mmx but also for sse*/avx*. Some of them fail in a mmx-only configuration 
so I would like to propose a toolchain based on this defconfig:

BR2_x86_pentium_mmx=y

We already have a p4 defconfig

http://autobuild.buildroot.net/toolchains/configs/br-i386-pentium4-
full.config

and defconfigs for x86_64 but none for older Intel cpu's.

Regards, Bernd

[Buildroot] Update of Buildroot toolchains on autobuild.b.o

[Bernd Kuhls]

[posted and mailed]

Thomas Petazzoni <thomas.petazzoni-
wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8@public.gmane.org> wrote in 
news:20160608220152.5b54ea7d at free-electrons.com:

>  * Addition of a ARM Cortex-M4 noMMU toolchain.

Hi Thomas,

afaics something is weird with this toolchain.

According to its defconfig it supports threads, but not NPTL however:

$ wget http://autobuild.buildroot.net/toolchains/configs/br-arm-cortex-m4-
full.conf
$ make defconfig BR2_DEFCONFIG=br-arm-cortex-m4-full.config

$ grep -i has_threads .config
BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS=y
# BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS_DEBUG is not set
# BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS_NPTL is not set
BR2_TOOLCHAIN_HAS_THREADS=y

But according to the output of gcc is does not support threads although it 
was compiled with --enable-threads

$ make toolchain
$ output/host/opt/ext-toolchain/bin/arm-buildroot-uclinux-uclibcgnueabi-gcc 
-v

Configured with: ./configure --prefix=/opt/br-arm-cortex-m4-full-2016.05-2-
g5dabb45/usr --sysconfdir=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/etc 
--enable-static --target=arm-buildroot-uclinux-uclibcgnueabi --with-
sysroot=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/usr/arm-buildroot-
uclinux-uclibcgnueabi/sysroot --disable-__cxa_atexit --with-gnu-ld --
disable-libssp --disable-multilib --with-gmp=/opt/br-arm-cortex-m4-full-
2016.05-2-g5dabb45/usr --with-mpfr=/opt/br-arm-cortex-m4-full-2016.05-2-
g5dabb45/usr --with-pkgversion='Buildroot 2016.05-00002-g5dabb45' --with-
bugurl=http://bugs.buildroot.net/ --disable-libsanitizer --disable-tls --
disable-libmudflap 

    	    	--enable-threads

 --with-mpc=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/usr --without-isl 
--without-cloog --with-float=soft --disable-decimal-float --with-abi=aapcs-
linux --with-cpu=cortex-m4 --with-float=soft --with-mode=thumb --enable-
languages=c,c++ --with-build-time-tools=/opt/br-arm-cortex-m4-full-2016.05-
2-g5dabb45/usr/arm-buildroot-uclinux-uclibcgnueabi/bin --disable-shared --
disable-libgomp

    	    	Thread model: single

gcc version 4.9.3 (Buildroot 2016.05-00002-g5dabb45) 

I noticed this while trying to fix some boost build errors.

Regards, Bernd

[Buildroot] Update of Buildroot toolchains on autobuild.b.o

[Thomas Petazzoni]

Hello,

On Sun, 12 Jun 2016 17:28:19 +0200, Bernd Kuhls wrote:

> According to its defconfig it supports threads, but not NPTL however:

The fact that it doesn't support NPTL is expected I believe (we're
talking about ARMv7-M).

> $ wget http://autobuild.buildroot.net/toolchains/configs/br-arm-cortex-m4-
> full.conf
> $ make defconfig BR2_DEFCONFIG=br-arm-cortex-m4-full.config
> 
> $ grep -i has_threads .config
> BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS=y
> # BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS_DEBUG is not set
> # BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS_NPTL is not set
> BR2_TOOLCHAIN_HAS_THREADS=y
> 
> But according to the output of gcc is does not support threads although it 
> was compiled with --enable-threads
> 
> $ make toolchain
> $ output/host/opt/ext-toolchain/bin/arm-buildroot-uclinux-uclibcgnueabi-gcc 
> -v
> 
> Configured with: ./configure --prefix=/opt/br-arm-cortex-m4-full-2016.05-2-
> g5dabb45/usr --sysconfdir=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/etc 
> --enable-static --target=arm-buildroot-uclinux-uclibcgnueabi --with-
> sysroot=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/usr/arm-buildroot-
> uclinux-uclibcgnueabi/sysroot --disable-__cxa_atexit --with-gnu-ld --
> disable-libssp --disable-multilib --with-gmp=/opt/br-arm-cortex-m4-full-
> 2016.05-2-g5dabb45/usr --with-mpfr=/opt/br-arm-cortex-m4-full-2016.05-2-
> g5dabb45/usr --with-pkgversion='Buildroot 2016.05-00002-g5dabb45' --with-
> bugurl=http://bugs.buildroot.net/ --disable-libsanitizer --disable-tls --
> disable-libmudflap 
> 
>     	    	--enable-threads
> 
>  --with-mpc=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/usr --without-isl 
> --without-cloog --with-float=soft --disable-decimal-float --with-abi=aapcs-
> linux --with-cpu=cortex-m4 --with-float=soft --with-mode=thumb --enable-
> languages=c,c++ --with-build-time-tools=/opt/br-arm-cortex-m4-full-2016.05-
> 2-g5dabb45/usr/arm-buildroot-uclinux-uclibcgnueabi/bin --disable-shared --
> disable-libgomp
> 
>     	    	Thread model: single
> 
> gcc version 4.9.3 (Buildroot 2016.05-00002-g5dabb45) 
> 
> I noticed this while trying to fix some boost build errors.

Indeed weird. Waldemar, do you have an idea? uClibc_config.h contains:

#undef __HAS_NO_THREADS__
#define __LINUXTHREADS_OLD__ 1
#undef __LINUXTHREADS_NEW__
#define __UCLIBC_HAS_THREADS__ 1

Thanks,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] Update of Buildroot toolchains on autobuild.b.o

[Waldemar Brodkorb]

Hi,
Thomas Petazzoni wrote,

> Hello,
> 
> On Sun, 12 Jun 2016 17:28:19 +0200, Bernd Kuhls wrote:
> 
> > According to its defconfig it supports threads, but not NPTL however:
> 
> The fact that it doesn't support NPTL is expected I believe (we're
> talking about ARMv7-M).
> 
> > $ wget http://autobuild.buildroot.net/toolchains/configs/br-arm-cortex-m4-
> > full.conf
> > $ make defconfig BR2_DEFCONFIG=br-arm-cortex-m4-full.config
> > 
> > $ grep -i has_threads .config
> > BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS=y
> > # BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS_DEBUG is not set
> > # BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS_NPTL is not set
> > BR2_TOOLCHAIN_HAS_THREADS=y
> > 
> > But according to the output of gcc is does not support threads although it 
> > was compiled with --enable-threads
> > 
> > $ make toolchain
> > $ output/host/opt/ext-toolchain/bin/arm-buildroot-uclinux-uclibcgnueabi-gcc 
> > -v
> > 
> > Configured with: ./configure --prefix=/opt/br-arm-cortex-m4-full-2016.05-2-
> > g5dabb45/usr --sysconfdir=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/etc 
> > --enable-static --target=arm-buildroot-uclinux-uclibcgnueabi --with-
> > sysroot=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/usr/arm-buildroot-
> > uclinux-uclibcgnueabi/sysroot --disable-__cxa_atexit --with-gnu-ld --
> > disable-libssp --disable-multilib --with-gmp=/opt/br-arm-cortex-m4-full-
> > 2016.05-2-g5dabb45/usr --with-mpfr=/opt/br-arm-cortex-m4-full-2016.05-2-
> > g5dabb45/usr --with-pkgversion='Buildroot 2016.05-00002-g5dabb45' --with-
> > bugurl=http://bugs.buildroot.net/ --disable-libsanitizer --disable-tls --
> > disable-libmudflap 
> > 
> >     	    	--enable-threads
> > 
> >  --with-mpc=/opt/br-arm-cortex-m4-full-2016.05-2-g5dabb45/usr --without-isl 
> > --without-cloog --with-float=soft --disable-decimal-float --with-abi=aapcs-
> > linux --with-cpu=cortex-m4 --with-float=soft --with-mode=thumb --enable-
> > languages=c,c++ --with-build-time-tools=/opt/br-arm-cortex-m4-full-2016.05-
> > 2-g5dabb45/usr/arm-buildroot-uclinux-uclibcgnueabi/bin --disable-shared --
> > disable-libgomp
> > 
> >     	    	Thread model: single
> > 
> > gcc version 4.9.3 (Buildroot 2016.05-00002-g5dabb45) 
> > 
> > I noticed this while trying to fix some boost build errors.
> 
> Indeed weird. Waldemar, do you have an idea? uClibc_config.h contains:
> 
> #undef __HAS_NO_THREADS__
> #define __LINUXTHREADS_OLD__ 1
> #undef __LINUXTHREADS_NEW__
> #define __UCLIBC_HAS_THREADS__ 1

When looking at the output of some precompiled toolchain for ARM
noMMU used by Emcraft:
./bin/arm-uclinuxeabi-gcc -v                                                                                                        
Using built-in specs.
Target: arm-uclinuxeabi
Configured with:
/scratch/julian/2010q1-release-uclinux-lite/src/gcc-4.4-2010q1/configure
--build=i686-pc-linux-gnu --host=i686-pc-linux-gnu
--target=arm-uclinuxeabi --enable-threads --disable-libmudflap
--disable-libssp --disable-libstdcxx-pch
--enable-extra-sgxxlite-multilibs --with-gnu-as --with-gnu-ld
--with-specs='%{O2:%{!fno-remove-local-statics:
-fremove-local-statics}}
%{O*:%{O|O0|O1|O2|Os:;:%{!fno-remove-local-statics:
%-fremove-local-statics}}}' --enable-languages=c,c++
%--disable-shared --disable-lto --with-pkgversion='Sourcery G++ Lite
%2010q1-189'
%--with-bugurl=https://support.codesourcery.com/GNUToolchain/
%--disable-nls --prefix=/opt/codesourcery
%--with-sysroot=/opt/codesourcery/arm-uclinuxeabi/libc
%--with-build-sysroot=/scratch/julian/2010q1-release-uclinux-lite/install/arm-uclinuxeabi/libc
%--with-gmp=/scratch/julian/2010q1-release-uclinux-lite/obj/host-libs-2010q1-189-arm-uclinuxeabi-i686-pc-linux-gnu/usr
%--with-mpfr=/scratch/julian/2010q1-release-uclinux-lite/obj/host-libs-2010q1-189-arm-uclinuxeabi-i686-pc-linux-gnu/usr
%--with-ppl=/scratch/julian/2010q1-release-uclinux-lite/obj/host-libs-2010q1-189-arm-uclinuxeabi-i686-pc-linux-gnu/usr
%--with-host-libstdcxx='-static-libgcc
%-Wl,-Bstatic,-lstdc++,-Bdynamic -lm'
%--with-cloog=/scratch/julian/2010q1-release-uclinux-lite/obj/host-libs-2010q1-189-arm-uclinuxeabi-i686-pc-linux-gnu/usr
%--disable-libgomp --enable-poison-system-directories
%--with-build-time-tools=/scratch/julian/2010q1-release-uclinux-lite/install/arm-uclinuxeabi/bin
%--with-build-time-tools=/scratch/julian/2010q1-release-uclinux-lite/install/arm-uclinuxeabi/bin
Thread model: single
gcc version 4.4.1 (Sourcery G++ Lite 2010q1-189) 

I would say it is normal for any *-uclinux-* toolchain using
linuxthreads instead of NPTL.

best regards
 Waldemar

[Buildroot] Thread support in Cortex m4 toolchains, was: Re: Update of Buildroot toolchains on autobuild.b.o

[Bernd Kuhls]

Am Mon, 13 Jun 2016 23:48:25 +0200 schrieb Waldemar Brodkorb:

> Thread model: single
> gcc version 4.4.1 (Sourcery G++ Lite 2010q1-189) 
> 
> I would say it is normal for any *-uclinux-* toolchain using
> linuxthreads instead of NPTL.

Hi Waldemar,

how to distinguish a toolchain using linuxthreads from a
toolchain lacking thread support?

The toolchain from
http://autobuild.buildroot.net/toolchains/configs/br-arm-full-
nothread.config
shows a similar output to the m4 toolchain

$ output/host/opt/ext-toolchain/bin/arm-buildroot-linux-uclibcgnueabihf-
gcc -v
Using built-in specs.
COLLECT_GCC=/home/bernd/buildroot/br2/output/host/opt/ext-toolchain/bin/
arm-buildroot-linux-uclibcgnueabihf-gcc.br_real
COLLECT_LTO_WRAPPER=/home/bernd/buildroot/br2/output/host/opt/ext-
toolchain/bin/../libexec/gcc/arm-buildroot-linux-uclibcgnueabihf/4.9.3/
lto-wrapper
Target: arm-buildroot-linux-uclibcgnueabihf
Configured with: ./configure --prefix=/opt/br-arm11-full-
nothread-2016.05-2-g5dabb45/usr --sysconfdir=/opt/br-arm11-full-
nothread-2016.05-2-g5dabb45/etc --enable-static --target=arm-buildroot-
linux-uclibcgnueabihf --with-sysroot=/opt/br-arm11-full-
nothread-2016.05-2-g5dabb45/usr/arm-buildroot-linux-uclibcgnueabihf/
sysroot --disable-__cxa_atexit --with-gnu-ld --disable-libssp --disable-
multilib --with-gmp=/opt/br-arm11-full-nothread-2016.05-2-g5dabb45/usr --
with-mpfr=/opt/br-arm11-full-nothread-2016.05-2-g5dabb45/usr --with-
pkgversion='Buildroot 2016.05-00002-g5dabb45' --with-bugurl=http://
bugs.buildroot.net/ --disable-libsanitizer --disable-tls --disable-
libmudflap --disable-threads --disable-libitm --disable-libatomic --with-
mpc=/opt/br-arm11-full-nothread-2016.05-2-g5dabb45/usr --without-isl --
without-cloog --disable-decimal-float --with-abi=aapcs-linux --with-
cpu=arm1176jzf-s --with-fpu=vfp --with-float=hard --with-mode=arm --
enable-languages=c,c++ --with-build-time-tools=/opt/br-arm11-full-
nothread-2016.05-2-g5dabb45/usr/arm-buildroot-linux-uclibcgnueabihf/bin --
enable-shared --disable-libgomp
Thread model: single
gcc version 4.9.3 (Buildroot 2016.05-00002-g5dabb45) 

This toolchain, as expected, lacks the definition of 
_GLIBCXX_HAS_GTHREADS in
output/host/opt/ext-toolchain/arm-buildroot-linux-uclibcgnueabihf/include/
c++/4.9.3/arm-buildroot-linux-uclibcgnueabihf/bits/c++config.h

The same is valid for the Cortex m4 toolchain however despite the fact 
that it claims to have thread support in Kconfig. This leads to build 
errors in the boost package:
http://autobuild.buildroot.net/
results/422/4220e275dcc90e64545ad2d5ed3e0b51044f6426/

./boost/config/requires_threads.hpp:29:4: error: #error "Threading 
support unavaliable: it has been explicitly disabled with 
BOOST_DISABLE_THREADS"
 #  error "Threading support unavaliable: it has been explicitly disabled 
with BOOST_DISABLE_THREADS"

Boost checks for _GLIBCXX_HAS_GTHREADS to determine thread support:
https://github.com/boostorg/config/blob/develop/include/boost/config/
stdlib/libstdcpp3.hpp#L37

Afaics we have two choices: Fix the thread support in gcc or remove the 
setting that the m4 toolchain supports threads.

Regards, Bernd

[Buildroot] Thread support in Cortex m4 toolchains, was: Re: Update of Buildroot toolchains on autobuild.b.o

[Waldemar Brodkorb]

Hi Bernd,
Bernd Kuhls wrote,

> Am Mon, 13 Jun 2016 23:48:25 +0200 schrieb Waldemar Brodkorb:
> 
> > Thread model: single
> > gcc version 4.4.1 (Sourcery G++ Lite 2010q1-189) 
> > 
> > I would say it is normal for any *-uclinux-* toolchain using
> > linuxthreads instead of NPTL.
> 
> Hi Waldemar,
> 
> how to distinguish a toolchain using linuxthreads from a
> toolchain lacking thread support?
> 
> The toolchain from
> http://autobuild.buildroot.net/toolchains/configs/br-arm-full-
> nothread.config
> shows a similar output to the m4 toolchain
> 
> $ output/host/opt/ext-toolchain/bin/arm-buildroot-linux-uclibcgnueabihf-
> gcc -v
> Using built-in specs.
> COLLECT_GCC=/home/bernd/buildroot/br2/output/host/opt/ext-toolchain/bin/
> arm-buildroot-linux-uclibcgnueabihf-gcc.br_real
> COLLECT_LTO_WRAPPER=/home/bernd/buildroot/br2/output/host/opt/ext-
> toolchain/bin/../libexec/gcc/arm-buildroot-linux-uclibcgnueabihf/4.9.3/
> lto-wrapper
> Target: arm-buildroot-linux-uclibcgnueabihf
> Configured with: ./configure --prefix=/opt/br-arm11-full-
> nothread-2016.05-2-g5dabb45/usr --sysconfdir=/opt/br-arm11-full-
> nothread-2016.05-2-g5dabb45/etc --enable-static --target=arm-buildroot-
> linux-uclibcgnueabihf --with-sysroot=/opt/br-arm11-full-
> nothread-2016.05-2-g5dabb45/usr/arm-buildroot-linux-uclibcgnueabihf/
> sysroot --disable-__cxa_atexit --with-gnu-ld --disable-libssp --disable-
> multilib --with-gmp=/opt/br-arm11-full-nothread-2016.05-2-g5dabb45/usr --
> with-mpfr=/opt/br-arm11-full-nothread-2016.05-2-g5dabb45/usr --with-
> pkgversion='Buildroot 2016.05-00002-g5dabb45' --with-bugurl=http://
> bugs.buildroot.net/ --disable-libsanitizer --disable-tls --disable-
> libmudflap --disable-threads --disable-libitm --disable-libatomic --with-
> mpc=/opt/br-arm11-full-nothread-2016.05-2-g5dabb45/usr --without-isl --
> without-cloog --disable-decimal-float --with-abi=aapcs-linux --with-
> cpu=arm1176jzf-s --with-fpu=vfp --with-float=hard --with-mode=arm --
> enable-languages=c,c++ --with-build-time-tools=/opt/br-arm11-full-
> nothread-2016.05-2-g5dabb45/usr/arm-buildroot-linux-uclibcgnueabihf/bin --
> enable-shared --disable-libgomp
> Thread model: single
> gcc version 4.9.3 (Buildroot 2016.05-00002-g5dabb45) 
> 
> This toolchain, as expected, lacks the definition of 
> _GLIBCXX_HAS_GTHREADS in
> output/host/opt/ext-toolchain/arm-buildroot-linux-uclibcgnueabihf/include/
> c++/4.9.3/arm-buildroot-linux-uclibcgnueabihf/bits/c++config.h
> 
> The same is valid for the Cortex m4 toolchain however despite the fact 
> that it claims to have thread support in Kconfig. This leads to build 
> errors in the boost package:
> http://autobuild.buildroot.net/
> results/422/4220e275dcc90e64545ad2d5ed3e0b51044f6426/
> 
> ./boost/config/requires_threads.hpp:29:4: error: #error "Threading 
> support unavaliable: it has been explicitly disabled with 
> BOOST_DISABLE_THREADS"
>  #  error "Threading support unavaliable: it has been explicitly disabled 
> with BOOST_DISABLE_THREADS"
> 
> Boost checks for _GLIBCXX_HAS_GTHREADS to determine thread support:
> https://github.com/boostorg/config/blob/develop/include/boost/config/
> stdlib/libstdcpp3.hpp#L37
> 
> Afaics we have two choices: Fix the thread support in gcc or remove the 
> setting that the m4 toolchain supports threads.

I would prefer first choice.
Can you try something like the attached patch for your gcc build?

I managed to compile ARM noMMU boost with that.
I have done no runtime testing and there is an open issue regarding
noMMU ARM and threads support discussed on the uClinux mailinglist.

best regards
 Waldemar
-------------- next part --------------
A non-text attachment was scrubbed...
Name: uclinux-threads.patch
Type: text/x-diff
Size: 471 bytes
Desc: not available
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20160616/169a59d1/attachment.bin>

[Buildroot] Thread support in Cortex m4 toolchains, was: Re: Update of Buildroot toolchains on autobuild.b.o

[Bernd Kuhls]

Am Thu, 16 Jun 2016 23:17:24 +0200 schrieb Waldemar Brodkorb:

> Hi Bernd,
> Bernd Kuhls wrote,
> 
>> Afaics we have two choices: Fix the thread support in gcc or remove the
>> setting that the m4 toolchain supports threads.
> 
> I would prefer first choice.
> Can you try something like the attached patch for your gcc build?

Hi,

thanks for the patch, I tested it with gcc-4.9.3 using this defconfig:

BR2_arm=y
BR2_cortex_m4=y
BR2_KERNEL_MIRROR="https://www.kernel.org/pub"
BR2_GNU_MIRROR="http://ftp.gnu.org/pub/gnu"
BR2_JLEVEL=16
# BR2_COMPILER_PARANOID_UNSAFE_PATH is not set
BR2_KERNEL_HEADERS_4_5=y
BR2_TOOLCHAIN_BUILDROOT_INET_RPC=y
BR2_TOOLCHAIN_BUILDROOT_LOCALE=y
BR2_TOOLCHAIN_BUILDROOT_CXX=y
BR2_INIT_NONE=y
# BR2_PACKAGE_BUSYBOX is not set
# BR2_TARGET_ROOTFS_TAR is not set

and could compile boost as well.

Tested-by: Bernd Kuhls <bernd.kuhls@t-online.de>
(compile-tested boost)

Regards, Bernd