Reading out the time zone on different architecures produces different results.

[Martin Gansser]

Hi,

the testsuite of the package cxxtools fails on armv7hl and i686.

In cxxtools there is a tool in "tools/cxxtz", which reads the tz database using cxxtools::Tz.

For some reason reading the offset for Europe/Berlin on 2018-11-7 13:00:06 leads to a offset of 0.89111h instead of exactly 1h.  This happens on the Fedora Build Server for armv7hl and i686 architecture [1].

I put in der check section the command "tools/cxxtz -z Europe/Berlin '2018-11-07 13:00:05'" so that it can be seen in the log file.

build.log results on armv8hl and i686
...
Executing(%check): /bin/sh -e /var/tmp/rpm-tmp.IfvkcP
+ umask 022
+ cd /builddir/build/BUILD
+ cd cxxtools-3.0
+ tools/cxxtz -z Europe/Berlin '2018-11-07 13:00:05'
2018-11-07 13:53:33 LMT 0
+ test/alltests
SmartPtr::RefCounted: OK
....

[1] https://koji.fedoraproject.org/koji/taskinfo?taskID=60983464
[2] https://martinkg.fedorapeople.org/ErrorReports/cxxtools-3.0-1.fc33.src.rpm

Regards
Martin

[Richard W.M. Jones]

On Tue, Feb 02, 2021 at 01:07:51PM -0000, Martin Gansser wrote:
> Hi,
>
> the testsuite of the package cxxtools fails on armv7hl and i686.
>
> In cxxtools there is a tool in "tools/cxxtz", which reads the tz
> database using cxxtools::Tz.
>
> For some reason reading the offset for Europe/Berlin on 2018-11-7
> 13:00:06 leads to a offset of 0.89111h instead of exactly 1h.  This
> happens on the Fedora Build Server for armv7hl and i686 architecture
> [1].

1h is the correct answer according to zdump (and common sense).

It seems more than just coincidence that it only fails on 32 bit
architectures.  I prepped the cxxtools 3.0 sources and ran this on a
x86-64 host (easier than finding an armv7 or i686 host):

$ ./configure CC="gcc" CFLAGS="-m32" CXXFLAGS="-m32" LDFLAGS="-m32"
$ make
$ tools/cxxtz -z Europe/Berlin '2018-11-07 13:00:05'
2018-11-07 13:53:33 LMT 0

so that reproduces the bug.

I would say this is an upstream bug and you should report it to them.
They should be easily able to reproduce it following the steps above
on any x86-64 host.

Rich.
> I put in der check section the command "tools/cxxtz -z Europe/Berlin '2018-11-07 13:00:05'" so that it can be seen in the log file.
>
> build.log results on armv8hl and i686
> ...
> Executing(%check): /bin/sh -e /var/tmp/rpm-tmp.IfvkcP
> umask 022
> cd /builddir/build/BUILD
> cd cxxtools-3.0
> tools/cxxtz -z Europe/Berlin '2018-11-07 13:00:05'
>
> 2018-11-07 13:53:33 LMT 0
> test/alltests
>
> SmartPtr::RefCounted: OK
> ....
>
> [1] https://koji.fedoraproject.org/koji/taskinfo?taskID=60983464
> [2] https://martinkg.fedorapeople.org/ErrorReports/cxxtools-3.0-1.fc33.src.rpm
>
> Regards
> Martin
> _______________________________________________
> devel mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/[e-mail]

[Martin Gansser]

Many Thanks for your feedback.

but when i run the following command on my x86-64 host,
i get the error message that the C compiler does't work.

[martin@fc33 cxxtools-3.0]$ ./configure CC="gcc" CFLAGS="-m32" CXXFLAGS="-m32" LDFLAGS="-m32"
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... /usr/bin/mkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether make supports nested variables... yes
checking whether make supports the include directive... yes (GNU style)
checking for gcc... gcc
checking whether the C compiler works... no
configure: error: in `/home/martin/rpmbuild/BUILD/cxxtools-3.0':
configure: error: C compiler cannot create executables
See `config.log' for more details

Martin

[Richard W.M. Jones]

On Tue, Feb 02, 2021 at 02:58:06PM -0000, Martin Gansser wrote:
> Many Thanks for your feedback.
>
> but when i run the following command on my x86-64 host,
> i get the error message that the C compiler does't work.
>
> [martin@fc33 cxxtools-3.0]$ ./configure CC="gcc" CFLAGS="-m32" CXXFLAGS="-m32" LDFLAGS="-m32"
> checking for a BSD-compatible install... /usr/bin/install -c
> checking whether build environment is sane... yes
> checking for a thread-safe mkdir -p... /usr/bin/mkdir -p
> checking for gawk... gawk
> checking whether make sets $(MAKE)... yes
> checking whether make supports nested variables... yes
> checking whether make supports the include directive... yes (GNU style)
> checking for gcc... gcc
> checking whether the C compiler works... no
> configure: error: in `/home/martin/rpmbuild/BUILD/cxxtools-3.0':
> configure: error: C compiler cannot create executables
> See `config.log' for more details

You'll need to look in config.log as the message says.  However the
most likely reason is you don't have the 32 bit toolchain installed,
in this case glibc-devel.i686, libstdc++-devel.i686 and
openssl-devel.i686.

Rich.

[Martin Gansser]

Thanks for your help.

Perhaps you can help me with another problem that exists with the testsuite on the ppc64le architecure [1]

the error message is:
serializationinfo::testRangeCheck: ASSERTION at serializationinfo-test.cpp:561
    exception of type std::range_error expected in siValue<double>(si)

[1] https://kojipkgs.fedoraproject.org//work/tasks/8769/61188769/build.log

Regards
Martin

[Richard W.M. Jones]

On Wed, Feb 03, 2021 at 09:20:37AM -0000, Martin Gansser wrote:
> Thanks for your help.
>
> Perhaps you can help me with another problem that exists with the testsuite on the ppc64le architecure [1]
>
> the error message is:
> serializationinfo::testRangeCheck: ASSERTION at serializationinfo-test.cpp:561
>    exception of type std::range_error expected in siValue<double>(si)

558   if (std::numeric_limits<long double>::max() > static_cast<long double>(std::numeric_limits<double>::max()))
559   {
560     si.setValue(static_cast<long double>(std::numeric_limits<double>::max()) * 1.01);
561     CXXTOOLS_UNIT_ASSERT_THROW(siValue<double>(si), std::range_error);

I'm not exactly sure what this code is doing, but I'm going to guess
it's confused that long double doesn't work like it does on x86-64.
On x86-64, long double is 80 bits, on ppc64le long double is, or is
going to be, 128 bits:
https://fedoraproject.org/wiki/Changes/PPC64LE_Float128_Transition

Take it to upstream is my advice.

Rich.
> [1] https://kojipkgs.fedoraproject.org//work/tasks/8769/61188769/build.log
>
> Regards
> Martin
> _______________________________________________
> devel mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/[e-mail]

[Martin Gansser]

Thanks, I'll discuss it with upstream.

Regards
Martin