Bug#943396: FTBFS on armhf: testsuite segfault

2020-10-27 Thread Bernhard Übelacker 
Dear Maintainer,
a short addition.
The backtrace that I received building with CC=gcc-10 seems
to be the same as in #972665, therefore they might be related?

Kind regards,
Bernhard

Bug#943396: FTBFS on armhf: testsuite segfault

2019-10-25 Thread Graham Inggs 
On Thu, 24 Oct 2019 at 21:30, Andreas Beckmann  wrote:
> Is this reproducible in buster, too, or limited to sid/bullseye?

Reproducible builds [1] says sid and bullseye only.


[1] https://tests.reproducible-builds.org/debian/history/armhf/fftw3.html

Bug#943396: FTBFS on armhf: testsuite segfault

2019-10-24 Thread Andreas Beckmann 
On Thu, 24 Oct 2019 13:13:43 +0100 Iain Lane  wrote:
> Package: src:fftw3
> Version: 3.3.8-2

> We found this in an Ubuntu test rebuild, but I just confirmed on
> harris.d.o:

Is this reproducible in buster, too, or limited to sid/bullseye?

Andreas

Bug#943396: FTBFS on armhf: testsuite segfault

2019-10-24 Thread Sebastien Bacher 
The bug also has been reported upstream on
https://github.com/FFTW/fftw3/issues/182 which includes a valgrind log.
Getting a backtrace is a bit tricky since the test hangs under gdb...

Cheers,

Bug#943396: FTBFS on armhf: testsuite segfault

2019-10-24 Thread Iain Lane 
Package: src:fftw3
Version: 3.3.8-2
Severity: serious
Tags: patch

Hiya,

We found this in an Ubuntu test rebuild, but I just confirmed on
harris.d.o:

> perl -w ./check.pl -r --estimate -c=5 -v `pwd`/bench
> Executing "/home/laney/fftw3-3.3.8/tests/bench -o estimate --verbose=1   
> --verify '//obrd7560' --verify '//ibrd7560' --verify '//ofrd7560' --verify 
> '//ifrd7560' --verify 'obrd7560' --verify 'ibrd7560' --verify 'ofrd7560' 
> --verify 'ifrd7560' --verify '//obcd7560' --verify '//ibcd7560' --verify 
> '//ofcd7560' --verify '//ifcd7560' --verify 'obcd7560' --verify 'ibcd7560' 
> --verify 'ofcd7560' --verify 'ifcd7560' --verify 'okd5e00*18' --verify 
> 'ikd5e00*18' --verify 'obrd8x3x7v7' --verify 'ibrd8x3x7v7' --verify 
> 'ofrd8x3x7v7' --verify 'ifrd8x3x7v7' --verify '//obcd8x3x7v7' --verify 
> '//ibcd8x3x7v7' --verify '//ofcd8x3x7v7' --verify '//ifcd8x3x7v7' --verify 
> 'obcd8x3x7v7' --verify 'ibcd8x3x7v7' --verify 'ofcd8x3x7v7' --verify 
> 'ifcd8x3x7v7' --verify 'okd17472o00' --verify 'ikd17472o00' --verify 
> 'obr5x8x10x14v6' --verify 'ibr5x8x10x14v6' --verify 'ofr5x8x10x14v6' --verify 
> 'ifr5x8x10x14v6' --verify '//obc5x8x10x14v6' --verify '//ibc5x8x10x14v6' 
> --verify '//ofc5x8x10x14v6' --verify '//ifc5x8x10x14v6' --verify 
> 'obc5x8x10x14v6' --verify 'ibc5x8x10x14v6' --verify 'ofc5x8x10x14v6' --verify 
> 'ifc5x8x10x14v6'"
> //obrd7560 2.36623e-07 7.01972e-07 2.57654e-07
> //ibrd7560 2.70078e-07 7.01972e-07 2.48174e-07
> //ofrd7560 2.32727e-07 7.01972e-07 2.3547e-07
> Segmentation fault

I can try to get a backtrace if you want, or feel free to grab the build
from harris:~laney/fftw3-3.3.8 and poke at it. Daniel Van Vugt from the
Ubuntu team investigated this a little bit and found that disabling Neon
works around this problem. See the MR:

  https://salsa.debian.org/science-team/fftw3/merge_requests/1

Would appreciate your thoughts.

Cheers,

-- 
Iain Lane  [ i...@orangesquash.org.uk ]
Debian Developer   [ la...@debian.org ]
Ubuntu Developer   [ la...@ubuntu.com ]