Bug#917959: octave: segfault on complex dot product
Hello Adam Knapp, you might run octave inside gdb to retrieve a backtrace of the crash. Run below command inside a terminal application and you should receive a file gdb-octave*.log that you might forward to this bug. gdb -q -batch -ex 'set pagination off' -ex 'set width 0' -ex run -ex 'bt full' --args octave 2>&1 | tee -a gdb-octave_$(date +%Y-%m-%d_%H-%M-%S).log Even better would be if debug symbol packages would be installed before like described in [1]. Kind regards, Bernhard [1] https://wiki.debian.org/HowToGetABacktrace
Bug#917959: octave: segfault on complex dot product
Dear Adam, Le mardi 01 janvier 2019 à 11:17 -0500, Adam Knapp a écrit : > Package: octave > Version: 4.4.1-2+b1 > Severity: important > Below, we have what I tried: > > octave:1> dot([i i], [i i]) > fatal: caught signal Segmentation fault -- stopping myself... > Segmentation fault > > > This should return 2, just like the real number version (and like Matlab). > > Other variations crash similarly: as column vectors, with a mixture of real > and complex numbers, using single components. > > The mathematically equivalent: > sum( [i i] .* conj([i i]) ) > works fine. This is likely a BLAS issue. Can you please report back which BLAS implementation you are using? (most likely OpenBLAS). Does the bug disappear when you switch to another implementation? (e.g. reference BLAS). You probably want to read this page on BLAS implementations and Debian: https://wiki.debian.org/DebianScience/LinearAlgebraLibraries Best, -- ⢀⣴⠾⠻⢶⣦⠀ Sébastien Villemot ⣾⠁⢠⠒⠀⣿⡁ Debian Developer ⢿⡄⠘⠷⠚⠋⠀ http://sebastien.villemot.name ⠈⠳⣄ http://www.debian.org signature.asc Description: This is a digitally signed message part
Bug#917959: octave: segfault on complex dot product
Control: tags -1 unreproducible moreinfo Control: severity -1 normal * Adam Knapp [2019-01-01 11:17]: Package: octave Version: 4.4.1-2+b1 Severity: important Below, we have what I tried: octave:1> dot([i i], [i i]) fatal: caught signal Segmentation fault -- stopping myself... Segmentation fault This should return 2, just like the real number version (and like Matlab). Other variations crash similarly: as column vectors, with a mixture of real and complex numbers, using single components. The mathematically equivalent: sum( [i i] .* conj([i i]) ) works fine. I cannot replicate this bug in my system (Octave 4.4.1, amd64): octave:1> dot([i i], [i i]) ans = 2 Unless you provide more information on the problem, I am afraid we will not be able to consider this bug report. I am hereby lowering its severity level to "normal". Thanks, Rafael
Bug#917959: octave: segfault on complex dot product
Package: octave Version: 4.4.1-2+b1 Severity: important Hello, Below, we have what I tried: octave:1> dot([i i], [i i]) fatal: caught signal Segmentation fault -- stopping myself... Segmentation fault This should return 2, just like the real number version (and like Matlab). Other variations crash similarly: as column vectors, with a mixture of real and complex numbers, using single components. The mathematically equivalent: sum( [i i] .* conj([i i]) ) works fine. Adam Knapp -- System Information: Debian Release: buster/sid APT prefers unstable APT policy: (500, 'unstable'), (500, 'testing'), (500, 'stable'), (500, 'oldstable'), (1, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 4.18.0-2-amd64 (SMP w/4 CPU cores) Locale: LANG=en_US.utf8, LC_CTYPE=en_US.utf8 (charmap=UTF-8), LANGUAGE=en_US.utf8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: systemd (via /run/systemd/system) LSM: AppArmor: enabled Versions of packages octave depends on: ii libamd21:5.3.0+dfsg-1 ii libarpack2 3.6.3-1 ii libasound2 1.1.7-2 ii libatlas3-base [liblapack.so.3]3.10.3-7+b1 ii libblas3 [libblas.so.3]3.8.0-2 ii libbz2-1.0 1.0.6-9 ii libc6 2.28-3 ii libcamd2 1:5.3.0+dfsg-1 ii libccolamd21:5.3.0+dfsg-1 ii libcholmod31:5.3.0+dfsg-1 ii libcolamd2 1:5.3.0+dfsg-1 ii libcxsparse3 1:5.3.0+dfsg-1 ii libfftw3-double3 3.3.8-2 ii libfftw3-single3 3.3.8-2 ii libfltk-gl1.3 1.3.4-7 ii libfltk1.3 1.3.4-7 ii libfreetype6 2.9.1-3 ii libgcc11:8.2.0-13 ii libglpk40 4.65-2 ii libgomp1 8.2.0-13 ii libklu11:5.3.0+dfsg-1 ii liblapack3 [liblapack.so.3]3.8.0-2 ii liboctave6 4.4.1-2+b1 ii libopenblas-base [liblapack.so.3] 0.3.4+ds-1 ii libportaudio2 19.6.0-1 ii libqhull7 2015.2-4 ii libqrupdate1 1.1.2-3 ii libqscintilla2-qt5-13 2.10.4+dfsg-1+b1 ii libqt5core5a 5.11.3+dfsg-1 ii libqt5gui5 5.11.3+dfsg-1 ii libqt5help55.11.3-1 ii libqt5network5 5.11.3+dfsg-1 ii libqt5opengl5 5.11.3+dfsg-1 ii libqt5printsupport55.11.3+dfsg-1 ii libqt5sql5 5.11.3+dfsg-1 ii libqt5widgets5 5.11.3+dfsg-1 ii libsndfile11.0.28-4 ii libstdc++6 8.2.0-13 ii libsuitesparseconfig5 1:5.3.0+dfsg-1 ii libumfpack51:5.3.0+dfsg-1 ii libx11-6 2:1.6.7-1 ii octave-common 4.4.1-2 ii texinfo6.5.0.dfsg.1-4+b1 ii zlib1g 1:1.2.11.dfsg-1 Versions of packages octave recommends: ii default-jre-headless 2:1.11-70 ii epstool3.09-1 ii gnuplot-x11 [gnuplot-nox] 5.2.5+dfsg1-1 ii libatlas3-base 3.10.3-7+b1 ii libopenblas-base 0.3.4+ds-1 ii octave-doc 4.4.1-2 ii pstoedit 3.73-1+b1 Versions of packages octave suggests: ii liboctave-dev 4.4.1-2+b1 -- no debconf information