Bug#936207: biosig4c++: Python2 removal in sid/bullseye
On Tue, 5 Nov 2019 12:27:42 -0800 Steve Langasek wrote: > However, after applying that patch, the package fails to build because: > > - it tries to invoke python, which is not present. Fixed by setting >PYTHON=python3 in MAKEOPTS from debian/rules. > - the python3 pkgconfig handling is completely messed up in >python/setup.py; it tries to find a pkgconfig file in the system >directory (why, when it's part of the same source package we're just >building right now?), and when it doesn't find it, under python3 it >raises a different exception than ValueError, so the fallback code >doesn't work. And if I set PKG_CONFIG_PATH to point at the libbiosig.pc >in the parent directory, it just fails later at linking time because ../ >isn't on the linker path. > > I'm stopping my investigation there, it really looks like this needs some > upstream cleanup. > > -- > Steve Langasek Give me a lever long enough and a Free OS > Debian Developer to set it on, and I can move the world. > Ubuntu Developer https://www.debian.org/ > slanga...@ubuntu.com vor...@debian.org Dear Steve, I'm not sure what debian expects, and what kind of changes you would expect. If there is some documentation about this, please point me to that, and I'll have a look what I can do. If you say the "fallback code" does not work, beause of "rais[ing] a different exception than ValueError", perhaps removing that part (see attached patch_remove-python-setup-pkgconfig-exception.diff ) might solve this. In any case, the installation of biosig-python works if libbiosig is available on the system, in which case one needs to do only pip install numpy pkgconfig pip install Biosig-1.9.1.tar.gz where Biosig-1.9.1.tar.gz is just built in python/dist with this command: cd python && make release I provide the sources also here https://pub.ist.ac.at/~schloegl/biosig/prereleases/Biosig-1.9.1.tar.gz Kind regards, Alois diff --git a/biosig4c++/python/setup.py b/biosig4c++/python/setup.py index 8539a20a..be9e437b 100644 --- a/biosig4c++/python/setup.py +++ b/biosig4c++/python/setup.py @@ -12,17 +12,11 @@ except ImportError: import os import numpy.distutils.misc_util as mu -try: -import pkgconfig -PKG=pkgconfig.parse('libbiosig') -CPATH=PKG['include_dirs'] -LIBS=PKG['libraries'] -LIBDIR=PKG['library_dirs'] -except ValueError: -print('cannot load pkgconfig(libbiosig) - use default location') -CPATH='/usr/local/include' -LIBS='-lbiosig' -LIBDIR='/usr/local/lib' +import pkgconfig +PKG=pkgconfig.parse('libbiosig') +CPATH=PKG['include_dirs'] +LIBS=PKG['libraries'] +LIBDIR=PKG['library_dirs'] module_biosig = Extension('biosig', define_macros = [('MAJOR_VERSION', '1'), ('MINOR_VERSION', '9')],
Bug#925643: ignore error in missing symbol
The severity of this bug should be changed to minor for the following reason(s): The symbol being lost is not from the API of biosig, and no application using libbiosig could have used it in a meanigful way. The missing symbol _cxx1112basic_stringIcSt11char_traitsIcESaIcEE12_M_constructIPKcEEvT_S8_St20forward_iterator_tag is most likely a change in compiling tinyxml (which is c++), however libbiosig prodes a C interface. If c++ symbols are exported, it is unintentional, more overover, those are not part of the API of libbiosig (see biosig.h and biosig-dev.h). I admit that visibility of symbols should be managed in order to avoid such issues, and I plan to do so with the upcoming release (1.9.6). I could already provide a patch, however this would also change the symbols and would not help with the present version. No user application can break by removal of the symbol, because there is no application that could have made use of that symbol in any meaningful way. Therefore I suggest to ignore "dpkg-gensymbols: error: some symbols or patterns disappeared in the symbols file: see diff output below" and reduce the severity of this bug to minor. Cheers, Alois
Bug#910686: Pymol 2.2.0 sources available from github
Package: pymol Version: 2.2.0 Severity: wishlist The open source version of Pymol 2.2.0 is available from here now: https://github.com/schrodinger/pymol-open-source https://github.com/schrodinger/pymol-open-source/releases It seems, debian/watch file is not aware of this new release, attached is a revised watch file . May I also suggest to package pymol 2.2.0 ? Best regards, Alois # See uscan(1) for format # TODO: http://pymol.svn.sourceforge.net/viewvc/pymol/tags.tar.gz?view=tar # opts=uversionmangle=s/^/0~svn\./,downloadurlmangle=s/xmds\?view.*$/xmds\/trunk\/xmds-doc.tar.gz\?view=tar/,filenamemangle=s/^.*revision=(\d+)$/xmds-doc-0~svn.$1.tar.gz/ \ # # http://xmds.svn.sourceforge.net/viewvc/xmds/trunk/xmds-doc/ .*revision=(\d+)$ # #version=3 #opts=uversionmangle=s/^(?!1)0?([\d]+)/0\.$1/;s/(.*)rc(.*)/$1~$2/, \ # http://pymol.svn.sourceforge.net/viewvc/pymol/tags/ /viewvc/pymol/tags/(?:[rv])?([\d\.]+(?:r\d+)?)(?:rc\d+)?/ version=4 opts=filenamemangle=s/.+\/v?(\d\S+)\.tar\.gz/pymol-$1\.tar\.gz/ \ https://github.com/schrodinger/pymol-open-source/tags .*/v?(\d\S+)\.tar\.gz
Bug#783653: pymol: Segmentation fault immediately after loading pdb
I confirm that this bug occurs under some specific conditions. For testing I used this file http://www.pdb.org/pdb/files/1bl8.pdb which works fine on a local installation of pymol (Debian/Stretch). It works also when connecting to OSX (through xquartz 2.7.12 with "ssh -Y " to a Debian/Jessie server with Pymol installed. However, when I try to connect from my Debian/Stretch Laptop through "ssh -X " to the same Debian/Jessie server, and I try to open the same file on the server, pymol crashes with the following error message: > PyMOL(TM) Molecular Graphics System, Version 1.8.4.0. > Copyright (c) Schrodinger, LLC. > All Rights Reserved. > > Created by Warren L. DeLano, Ph.D. > > PyMOL is user-supported open-source software. Although some versions > are freely available, PyMOL is not in the public domain. > > If PyMOL is helpful in your work or study, then please volunteer > support for our ongoing efforts to create open and affordable scientific > software by purchasing a PyMOL Maintenance and/or Support subscription. > > More information can be found at "http://www.pymol.org;. > > Enter "help" for a list of commands. > Enter "help " for information on a specific command. > > Hit ESC anytime to toggle between text and graphics. > > Detected OpenGL version 2.0 or greater. Shaders available. > Detected GLSL version 1.30. > OpenGL graphics engine: > GL_VENDOR: VMware, Inc. > GL_RENDERER: Gallium 0.4 on llvmpipe (LLVM 3.8, 256 bits) > GL_VERSION: 3.0 Mesa 13.0.6 > Detected 32 CPU cores. Enabled multithreaded rendering. > HEADERMEMBRANE PROTEIN23-JUL-98 1BL8 > TITLE POTASSIUM CHANNEL (KCSA) FROM STREPTOMYCES LIVIDANS > COMPNDMOL_ID: 1; > COMPND 2 MOLECULE: PROTEIN (POTASSIUM CHANNEL PROTEIN); > COMPND 3 CHAIN: A, B, C, D; > COMPND 4 ENGINEERED: YES; > COMPND 5 MUTATION: YES > ObjectMolecule: Read secondary structure assignments. > ObjectMolecule: Read crystal symmetry information. > CmdLoad: "1bl8.pdb" loaded as "1bl8". > Segmentation fault I tried also with some other Debian machines, and it seems a contributing factor is whether the GL_RENDERER is 128 or 256 bits. > GL_RENDERER: Gallium 0.4 on llvmpipe (LLVM 3.8, 256 bits) It crashes in case of 256 bits, but does not crash when 128 bit renderer is reported. Maybe someone else can make use of this information ? Cheers, Alois
Bug#866115: [Pkg-monitoring-maintainers] Bug#866115: dist-upgrade to Stretch broke ganglia-web
On 2017-06-27 15:15, Daniel Pocock wrote: > > > On 27/06/17 14:58, Alois Schloegl wrote: >> Package: ganglia-web >> >> >> The following ganglia packages are installed: >> >> ganglia:~# dpkg -l|grep ganglia >> ii ganglia-modules-linux 1.3.6-2 >> amd64Ganglia extra modules for Linux (IO, filesystems, >> multicpu) >> ii ganglia-monitor 3.6.0-7+b1 >> amd64cluster monitoring toolkit - node daemon >> ii ganglia-monitor-python3.6.0-7 >> all cluster monitoring toolkit - python modules >> ii ganglia-webfrontend 3.6.1-3 >> all cluster monitoring toolkit - web front-end >> ii libganglia1 3.6.0-7+b1 >> amd64cluster monitoring toolkit - shared libraries >> >> >> After dist-upgrading from Debian 8 to Debian 9, the ganglia web >> interface is not working anymore, it just shows this error message. >> >> >> === START-OF-MESSAGE === >> >> Sorry, you do not have access to this resource. >> "); } try { $dwoo = new Dwoo($conf['dwoo_compiled_dir'], >> $conf['dwoo_cache_dir']); } catch (Exception $e) { print " >> There was an error initializing the Dwoo PHP Templating Engine: ". >> $e->getMessage() . " >> >> The compile directory should be owned and writable by the apache user. > > > On my system it is working, I see the following owner and permissions: > > $ ls -la /var/lib/ganglia-web/dwoo/ > total 16 > drwxr-xr-x 4 www-data root 4096 Jan 22 2015 . > drwxr-xr-x 4 root root 4096 Jan 22 2015 .. > drwxr-xr-x 2 www-data root 4096 Apr 4 2014 cache > drwxrwxrwx 3 www-data root 4096 Mar 16 2015 compiled > >> >> >> Do you have any suggests for debugging this issue ? >> > > Please send the output of this command: > > $ ls -la /var/lib/ganglia-web/dwoo/ > These permissions are the same at my machine. ganglia:~# ls -la /var/lib/ganglia-web/dwoo/ insgesamt 16 drwxr-xr-x 4 www-data root 4096 Jun 27 12:37 . drwxr-xr-x 4 root root 4096 Jun 27 12:37 .. drwxr-xr-x 2 www-data root 4096 Jul 7 2016 cache drwxrwxrwx 3 www-data root 4096 Jul 27 2015 compiled I tried further, purging and reinstalling ganglia-web and apache did not help. What solved the problem was enabling all modules with a2enmod Your choices are: access_compat actions alias allowmethods asis auth_basic auth_digest auth_form auth_kerb authn_anon authn_core authn_dbd authn_dbm authn_file authn_socache authnz_fcgi authnz_ldap authz_core authz_dbd authz_dbm authz_groupfile authz_host authz_owner authz_user autoindex buffer cache cache_disk cache_socache cern_meta cgi cgid charset_lite data dav dav_fs dav_lock dbd deflate dialup dir dump_io echo env expires ext_filter file_cache filter gnutls headers heartbeat heartmonitor http2 ident imagemap include info lbmethod_bybusyness lbmethod_byrequests lbmethod_bytraffic lbmethod_heartbeat ldap log_debug log_forensic lua macro mime mime_magic mpm_event mpm_itk mpm_prefork mpm_worker negotiation perl php7.0 proxy proxy_ajp proxy_balancer proxy_connect proxy_express proxy_fcgi proxy_fdpass proxy_ftp proxy_hcheck proxy_html proxy_http proxy_http2 proxy_scgi proxy_wstunnel ratelimit reflector remoteip reqtimeout request rewrite sed session session_cookie session_crypto session_dbd setenvif slotmem_plain slotmem_shm socache_dbm socache_memcache socache_shmcb speling ssl status substitute suexec unique_id userdir usertrack vhost_alias xml2enc and enabling all of them solved the problem. I'm not sure which of them I really need, but suspect that there is a missing dependency on either to php7 or gnutls module. Thanks for considering, Alois
Bug#866115: dist-upgrade to Stretch broke ganglia-web
Package: ganglia-web The following ganglia packages are installed: ganglia:~# dpkg -l|grep ganglia ii ganglia-modules-linux 1.3.6-2 amd64Ganglia extra modules for Linux (IO, filesystems, multicpu) ii ganglia-monitor 3.6.0-7+b1 amd64cluster monitoring toolkit - node daemon ii ganglia-monitor-python3.6.0-7 all cluster monitoring toolkit - python modules ii ganglia-webfrontend 3.6.1-3 all cluster monitoring toolkit - web front-end ii libganglia1 3.6.0-7+b1 amd64cluster monitoring toolkit - shared libraries After dist-upgrading from Debian 8 to Debian 9, the ganglia web interface is not working anymore, it just shows this error message. === START-OF-MESSAGE === Sorry, you do not have access to this resource. "); } try { $dwoo = new Dwoo($conf['dwoo_compiled_dir'], $conf['dwoo_cache_dir']); } catch (Exception $e) { print " There was an error initializing the Dwoo PHP Templating Engine: ". $e->getMessage() . " The compile directory should be owned and writable by the apache user. "; exit; } # Useful for addons. $GHOME = "."; if ($context == "meta" or $context == "control") { $title = "$self ${conf['meta_designator']} Report"; include_once "./header.php"; include_once "./meta_view.php"; } else if ($context == "tree") { $title = "$self ${conf['meta_designator']} Tree"; include_once "./header.php"; include_once "./grid_tree.php"; } else if ($context == "cluster" or $context == "cluster-summary") { if (preg_match('/cluster/i', $clustername)) $title = "$clustername Report"; else $title = "$clustername Cluster Report"; include_once "./header.php"; include_once "./cluster_view.php"; } else if ($context == "physical") { $title = "$clustername Physical View"; include_once "./header.php"; include_once "./physical_view.php"; } else if ($context == "node") { $title = "$hostname Node View"; include_once "./header.php"; include_once "./show_node.php"; } else if ($context == "host") { $title = "$hostname Host Report"; include_once "./header.php"; include_once "./host_view.php"; } else if ($context == "views") { $title = "$viewname view"; include_once "./header.php"; include_once "./views_view.php"; } else if ($context == "compare_hosts") { $title = "Compare Hosts"; include_once "./header.php"; include_once "./compare_hosts.php"; } else if ($context == "decompose_graph") { $title = "Decompose graph"; include_once "./header.php"; include_once "./decompose_graph.php"; } else { $title = "Unknown Context"; print "Unknown Context Error: Have you specified a host but not a cluster?."; } include_once "./footer.php"; ?> END === When searching the web, there where some hints towards a permission problem, but changing the permissions did not help. I also purged and reinstalled ganglia-web and the apache, but it did not help. Do you have any suggests for debugging this issue ?
Bug#857121: biosig4c++ FTBFS on architectures where char is unsigned
This bug was fixed upstream in commit 8e7c3d7e485e1d3af3f3094641282274424b7385 Attaches is the corresponding patch. Thanks for consideration, Alois commit 8e7c3d7e485e1d3af3f3094641282274424b7385 Author: Christoph Schmidt-HieberDate: Sat Dec 10 16:50:41 2016 +0100 Fix compilation with gcc6 on armhf. See Debian Bug #847526 diff --git a/biosig4c++/t210/codes.h b/biosig4c++/t210/codes.h index 4db65b2..53affbf 100644 --- a/biosig4c++/t210/codes.h +++ b/biosig4c++/t210/codes.h @@ -22,7 +22,7 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. #ifndef __CODES_H__ #define __CODES_H__ -static const char STR_END[]={-1,'\0'}; +static const char STR_END[]={(char)-1,'\0'}; static char STR_NULL[]=" unspecified/unknown "; static alfabetic _special[]={
Bug#847309: libthrust-dev: cub::DeviceReduce::Min causes segmentation fault
Package: libthrust-dev Version: 1.7.0 Severity: normal Dear Maintainer, I tried to compile and run relion 2 beta with cuda enable. nvidia-cuda-toolkit 7.5 from debian8/backports is installed on that machine. When testing the application, it crashed with a segmentation fault. The issue is also filed here, https://bitbucket.org/tcblab/relion2-beta/issues/164/seg-fault-on-relion2-with-cuda which includes also the data we used for testing, and long investigation of the issue. The problem went away when the output argument was initialized with +Infinity, and after upgrading to libthrust-dev 1.8.1. This indicates: 1) cub::DeviceReduce::Min from libthrust-dev 1.7.0 does not user a proper initialization, and depends on uninitialized data. 2) the problem seems to be solved in upstream in more recent version. w.r.t to debian, I'd like suggesting that nvidia-cuda-toolkit from backports should require libthrust-dev from backports. That could avoid similar problems for others. Best regards, Alois -- System Information: Debian Release: 8.6 APT prefers stable APT policy: (990, 'stable') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.16.0-4-amd64 (SMP w/12 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: systemd (via /run/systemd/system) libthrust-dev depends on no packages. libthrust-dev recommends no packages. Versions of packages libthrust-dev suggests: ii nvidia-cuda-toolkit 7.5.18-4~bpo8+1 -- no debconf information signature.asc Description: OpenPGP digital signature
Bug#831387: Files and directories with Umlaut do not get backed up through smb
Package: backuppc Version: 3.3.1-3 Severity: important I'm running backuppc 3.3.1-3 (from debian/testing) on Debian/Jessie. When trying to backup through smb files and directories with umlauts in its name, it will not backup those files, and the content of these directories is ignored. When trying to debug the problem, I found that the problem occurs only when starting through the web interface. When triggering the backup from the command line on the backup server the problem did not occur. root@backup2:~# dpkg -l|grep backuppc ii backuppc 3.3.1-3 amd64high-performance, enterprise-grade system for backing up PCs root@backup2:~# dpkg -l|grep smbclient ii libsmbclient:amd642:4.2.10+dfsg-0+deb8u3 amd64shared library for communication with SMB/CIFS servers ii smbclient 2:4.2.10+dfsg-0+deb8u3 amd64command-line SMB/CIFS clients for Unix I'm aware of this post: https://sourceforge.net/p/backuppc/mailman/message/35035066/ and have set according to Oliver Freyd's posting LANG=en_US.UTF-8 in /etc/init.d/backuppc I'm not sure whether this is a bug in backuppc or in the debian packaging, so this message is posting to debian bugs and the backuppc-users mailinglist. I did a number of tests with various configuration options (see below), but running the smb backup through the backuppc server fails to backup files and directories with umlaut in its name. Does anyone know how to fix this ? check LANG ## locale is set to en_US.UTF8 for root as well as user "backuppc" root@backup2:~# locale LANG=en_US.UTF-8 LANGUAGE= LC_CTYPE="en_US.UTF-8" LC_NUMERIC="en_US.UTF-8" LC_TIME="en_US.UTF-8" LC_COLLATE="en_US.UTF-8" LC_MONETARY="en_US.UTF-8" LC_MESSAGES="en_US.UTF-8" LC_PAPER="en_US.UTF-8" LC_NAME="en_US.UTF-8" LC_ADDRESS="en_US.UTF-8" LC_TELEPHONE="en_US.UTF-8" LC_MEASUREMENT="en_US.UTF-8" LC_IDENTIFICATION="en_US.UTF-8" LC_ALL= root@backup2:~# locale^C root@backup2:~# su -c locale backuppc LANG=en_US.UTF-8 LANGUAGE= LC_CTYPE="en_US.UTF-8" LC_NUMERIC="en_US.UTF-8" LC_TIME="en_US.UTF-8" LC_COLLATE="en_US.UTF-8" LC_MONETARY="en_US.UTF-8" LC_MESSAGES="en_US.UTF-8" LC_PAPER="en_US.UTF-8" LC_NAME="en_US.UTF-8" LC_ADDRESS="en_US.UTF-8" LC_TELEPHONE="en_US.UTF-8" LC_MEASUREMENT="en_US.UTF-8" LC_IDENTIFICATION="en_US.UTF-8" LC_ALL= Test 0: preparation ## For testing I've set up on the windows machines some test files and directories with Umlauts in it. So, when I do CLIENTHOST=mywindowsmachine.local su -c "smbclient //$CLIENTHOST/C$ -W IST -k -D /Users/schloegl/test.backuppc -c -E -d 5 -c tarmode full -TcX - \BOOTSECT.BAK " backuppc >t5.tar tar xvf t5.tar und run # find Users/ Users/ Users/schloegl Users/schloegl/test.backuppc Users/schloegl/test.backuppc/withUmlautÖtest Users/schloegl/test.backuppc/withUmlautÖtest/test..001.txt Users/schloegl/test.backuppc/withUmlautÖtest/testÖ.001.txt Users/schloegl/test.backuppc/noUmlaut Users/schloegl/test.backuppc/noUmlaut/test..001.txt Users/schloegl/test.backuppc/noUmlaut/testÖ.001.txt # cat Users/schloegl/test.backuppc/withUmlautÖtest/testÖ.001.txt test test which is fine, all other files test*.txt have the same content. Test 4: Triggering the backup from the web interface ## htop shows these processes: 0 0 0 5 4793 backuppc 20 0 65632 15980 3156 S 0.0 0.0 0:00.04 ├─ /usr/bin/perl /usr/share/backuppc/bin/BackupPC -d 60.4 0 60.4 13 5145 backuppc 20 0 82932 26176 4380 S 7.1 0.1 0:00.89 │ ├─ /usr/bin/perl /usr/share/backuppc/bin/BackupPC_dump -f mywindowsmachine.local 0 0 0 7 5149 backuppc 20 0 231M 12932 10708 R 20.9 0.0 0:02.81 │ │ ├─ /usr/bin/smbclient //w7test012.ist.local/C$ -W IST -k -N -E -d 5 -c tarmode full verbose -Tc - /Users/schloegl 35010 239 34771 8 5148 backuppc 20 0 64368 17460 4100 D 41.7 0.0 0:04.61 │ │ └─ /usr/bin/perl /usr/share/backuppc/bin/BackupPC_tarExtract mywindowsmachine.local C$ 0 0 0 0 11 4796 backuppc 30 10 46460 10048 3812 S 0.0 0.0 0:00.07 │ └─ /usr/bin/perl /usr/share/backuppc/bin/BackupPC_trashClean And the files on the backup server are: cat `ls -1td /var/lib/backuppc/pc/$CLIENTHOST/*/|head -1`fC\$/fUsers/fschloegl/ftest.backuppc/fwithUmlautÖtest/ftestÖ.001.txt cat: /var/lib/backuppc/pc/w7test012.ist.local/111/fC$/fUsers/fschloegl/ftest.backuppc/fwithUmlautÖtest/ftestÖ.001.txt: No such file or directory root@backup2:~# root@backup2:~# find `ls -1td /var/lib/backuppc/pc/$CLIENTHOST/*/|head -1`fC\$/fUsers/fschloegl/ftest.backuppc /var/lib/backuppc/pc/w7test012.ist.local/111/fC$/fUsers/fschloegl/ftest.backuppc /var/lib/backuppc/pc/w7test012.ist.local/111/fC$/fUsers/fschloegl/ftest.backuppc/fnoUmlaut ca /var/lib/backuppc/pc/w7test012.ist.local/111/fC$/fUsers/fschloegl/ftest.backuppc/fnoUmlaut/fPaxHeader
Bug#816283: NVRM: API mismatch on cuda/jessie-backports
On 2016-03-31 20:30, Luca Boccassi wrote: > On Thu, 2016-03-31 at 18:23 +0200, Alois Schloegl wrote: >> At the beginning, I used the .run installer, but thought to have cleaned >> this up. Obviously, this was not the case. The are a few more files >> installed on Sep 10 (mostly symbolic links). I've removed those and >> reinstalled nvidia-smi. That fixed the problem. >> >> >> Thanks a lot for the hint. > > Hello Alois, > > Are you happy with the result? Is it working fine now? If so, may I > close this bug now? > > Kind regards, > Luca Boccassi > Yes, thank you. The bug can be closed. Best, Alois
Bug#816283: NVRM: API mismatch on cuda/jessie-backports
On 03/31/2016 04:12 PM, Andreas Beckmann wrote: > On 2016-03-31 11:46, Alois Schloegl wrote: >> I'm trying to get Cuda working on debian/jessie. When running >> nvidia-smi, dmesg shows this message: >> >> [ 2355.945623] NVRM: API mismatch: the client has the version 352.41, but >> NVRM: this kernel module has the version 352.79. Please >> NVRM: make sure that this kernel module and all NVIDIA driver >> NVRM: components have the same version. >> [ 2355.970347] NVRM: nvidia_frontend_ioctl: minor 255, module->ioctl >> failed, error -22 >> >> Any ideas, where this client with version 352.41 comes from, and how to >> fix this ? > > We had 352.41-1 in experimental ... or did you install that driver from > the .run installer? > > This "client" is an old library ... try > > strace -e open nvidia-smi > > to see which library is is openening and then see where this is > pointing, it's probably /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.1 > being incorrect in some way. Yes, you are right. The are actually a bunch of other library files # ls -atlr /usr/lib/x86_64-linux-gnu/lib*352* -rwxr-xr-x 1 root root75400 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libGLESv2.so.352.41 -rwxr-xr-x 1 root root51080 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libGLESv1_CM.so.352.41 -rwxr-xr-x 1 root root 935112 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libEGL.so.352.41 -rwxr-xr-x 1 root root 166848 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-encode.so.352.41 -rwxr-xr-x 1 root root 14280360 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libcuda.so.352.41 -rwxr-xr-x 1 root root 15898960 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-opencl.so.352.41 -rwxr-xr-x 1 root root 159240 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-cfg.so.352.41 -rwxr-xr-x 1 root root 1652272 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvcuvid.so.352.41 -rwxr-xr-x 1 root root 197704 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-ifr.so.352.41 -rwxr-xr-x 1 root root86088 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-fbc.so.352.41 -rwxr-xr-x 1 root root 967360 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-ml.so.352.41 -rwxr-xr-x 1 root root 1790568 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-gtk2.so.352.41 -rwxr-xr-x 1 root root 1784456 Sep 10 2015 /usr/lib/x86_64-linux-gnu/libnvidia-gtk3.so.352.41 -rw-r--r-- 1 root root 42469400 Jan 14 00:53 /usr/lib/x86_64-linux-gnu/libnvidia-glcore.so.352.79 -rw-r--r-- 1 root root12600 Jan 14 00:58 /usr/lib/x86_64-linux-gnu/libnvidia-tls.so.352.79 -rw-r--r-- 1 root root 486048 Jan 14 02:05 /usr/lib/x86_64-linux-gnu/libnvidia-glsi.so.352.79 -rw-r--r-- 1 root root75400 Jan 14 02:19 /usr/lib/x86_64-linux-gnu/libGLESv2.so.352.79 -rw-r--r-- 1 root root51080 Jan 14 02:19 /usr/lib/x86_64-linux-gnu/libGLESv1_CM.so.352.79 -rw-r--r-- 1 root root 935112 Jan 14 02:19 /usr/lib/x86_64-linux-gnu/libEGL.so.352.79 -rw-r--r-- 1 root root 40648712 Jan 14 02:19 /usr/lib/x86_64-linux-gnu/libnvidia-eglcore.so.352.79 -rw-r--r-- 1 root root 48309464 Jan 14 02:20 /usr/lib/x86_64-linux-gnu/libnvidia-compiler.so.352.79 At the beginning, I used the .run installer, but thought to have cleaned this up. Obviously, this was not the case. The are a few more files installed on Sep 10 (mostly symbolic links). I've removed those and reinstalled nvidia-smi. That fixed the problem. Thanks a lot for the hint. signature.asc Description: OpenPGP digital signature
Bug#816283: nvidia-cuda-toolkit: cuda broken after upgrade (x86_64, Jessie, GTX980)
In order to fix the installation of glx-alternative-nvidia, and fix some other cuda related issues, I tried to update cuda to jessie-backports. The update did not work out of the box, e.g After uninstalling nvidia-cuda and reinstallung with apt-get -t jessie-backports install nvidia-cuda-toolkit, there were still problems nvidia-smi fails with this message Failed to initialize NVML: GPU access blocked by the operating system dmesg shows "API mismatch: the client has the version 352.41". After the update, I noticed that the link pointed to /usr/lib/x86_64-linux-gnu/libEGL.so.1 -> libEGL.so.352.41 and dpkg -i libegl1-nvidia_352.79-1~bpo8+1_amd64.deb did not fix this, so manual extracted /usr/lib/x86_64-linux-gnu/libEGL.so.352.79 and set the above link to it. dmesg shows "API mismatch: the client has the version 352.41". The details are: # dmesg |grep -i "nvid\|nvrm" [ 16.782287] nvidia: module license 'NVIDIA' taints kernel. [ 16.813559] nvidia :03:00.0: enabling device ( -> 0003) [ 16.829365] nvidia :82:00.0: enabling device ( -> 0003) [ 16.845105] nvidia :83:00.0: enabling device ( -> 0003) [ 16.861001] [drm] Initialized nvidia-drm 0.0.0 20150116 for :02:00.0 on minor 0 [ 16.868905] [drm] Initialized nvidia-drm 0.0.0 20150116 for :03:00.0 on minor 1 [ 16.876789] [drm] Initialized nvidia-drm 0.0.0 20150116 for :82:00.0 on minor 2 [ 16.884678] [drm] Initialized nvidia-drm 0.0.0 20150116 for :83:00.0 on minor 3 [ 16.892467] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 352.79 Wed Jan 13 16:17:53 PST 2016 [ 18.702943] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci:80/:80:02.0/:82:00.1/sound/card2/input7 [ 18.711066] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci:00/:00:02.0/:02:00.1/sound/card0/input8 [ 18.711188] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci:00/:00:02.0/:02:00.1/sound/card0/input12 [ 18.711330] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci:00/:00:02.0/:02:00.1/sound/card0/input17 [ 18.711668] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci:00/:00:02.0/:02:00.1/sound/card0/input21 [ 18.754668] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci:80/:80:02.0/:82:00.1/sound/card2/input11 [ 18.766633] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci:80/:80:03.0/:83:00.1/sound/card3/input6 [ 18.767994] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci:00/:00:03.0/:03:00.1/sound/card1/input9 [ 18.770893] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci:00/:00:03.0/:03:00.1/sound/card1/input13 [ 18.771620] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci:00/:00:03.0/:03:00.1/sound/card1/input16 [ 18.771805] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci:00/:00:03.0/:03:00.1/sound/card1/input20 [ 18.820540] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci:80/:80:02.0/:82:00.1/sound/card2/input14 [ 18.832036] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci:80/:80:02.0/:82:00.1/sound/card2/input19 [ 18.843980] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci:80/:80:03.0/:83:00.1/sound/card3/input10 [ 18.855326] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci:80/:80:03.0/:83:00.1/sound/card3/input15 [ 18.867117] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci:80/:80:03.0/:83:00.1/sound/card3/input18 [ 731.363046] NVRM: API mismatch: the client has the version 352.41, but NVRM: this kernel module has the version 352.79. Please NVRM: make sure that this kernel module and all NVIDIA driver NVRM: components have the same version. [ 731.387767] NVRM: nvidia_frontend_ioctl: minor 255, module->ioctl failed, error -22 Here is the list of current packages: # dpkg -l|grep -i nvidia ii glx-alternative-nvidia 0.7.1~bpo8+1 amd64allows the selection of NVIDIA as GLX provider ii libcublas6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuBLAS Library ii libcuda1:amd64 352.79-1~bpo8+1 amd64NVIDIA CUDA Driver Library ii libcudart6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA CUDA Runtime Library ii libcufft6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuFFT Library ii libcufftw6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuFFTW Library ii libcuinj64-6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA CUINJ Library (64-bit) ii libcurand6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuRAND Library ii libcusparse6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuSPARSE Library ii libegl1-nvidia:amd64 352.79-1~bpo8+1 amd64NVIDIA binary EGL libraries ii libgl1-nvidia-glx:amd64352.79-1~bpo8+1 amd64
Bug#816283: nvidia-cuda-toolkit: cuda broken after upgrade (x86_64, Jessie, GTX980)
I was able to revert to a working state of cuda 6.5, by using dpkg -i and the packges in /var/cache/apt/archives. Below are the output of nvidia-smi and the gpuinfo test program. The package list is attached in "pkglist.nvidia.working.txt". Configuration of glx-alternative-nvidia failed with this message. -- begin -- Processing triggers for glx-alternative-nvidia (0.7.1) ... update-alternatives: error: alternative link /etc/modprobe.d/nvidia.conf is already managed by nvidia dpkg: error processing package glx-alternative-nvidia (--configure): subprocess installed post-installation script returned error exit status 2 Errors were encountered while processing: glx-alternative-nvidia E: Sub-process /usr/bin/dpkg returned an error code (1) -- end -- So, this issue is affected by #801952 [1], as well. When upgrading to testing or backports, nvidia-driver 352.79-3 and some other cuda dependancies are pulled in, which break cuda (package list ist attached). Now, when looking at the package list, it seems more a problem of nvidia-driver and not cuda. However, [2] says that 352.79 supports gtx980. Currently, I'm suspecting a problem with the kernel module nvidia-uvm [3]. [1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=801952 [2] http://www.nvidia.com/download/driverResults.aspx/97645/en-us [3] https://www.phoronix.com/scan.php?page=news_item=NVIDIA-New-Build-System - output of test programs - nvidia-smi shows this: +--+ | NVIDIA-SMI 352.41 Driver Version: 352.41 | |---+--+--+ | GPU NamePersistence-M| Bus-IdDisp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===+==+==| | 0 GeForce GTX 980 Off | :02:00.0 Off | N/A | | 26% 28CP046W / 180W | 15MiB / 4095MiB | 0% Default | +---+--+--+ | 1 GeForce GTX 980 Off | :03:00.0 Off | N/A | | 26% 29CP045W / 180W | 15MiB / 4095MiB | 0% Default | +---+--+--+ | 2 GeForce GTX 980 Off | :82:00.0 Off | N/A | | 26% 28CP042W / 180W | 15MiB / 4095MiB | 0% Default | +---+--+--+ | 3 GeForce GTX 980 Off | :83:00.0 Off | N/A | | 0% 30CP033W / 180W | 15MiB / 4095MiB | 0% Default | +---+--+--+ +-+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=| | No running processes found | +-+ and "../bin/gpuinfo" gives: Simple GPU info query. You have 4 nVidia GPGPU. DeviceID NameVersion Memory(Mb) #0 GeForce GTX 980 5.2 4095 #1 GeForce GTX 980 5.2 4095 #2 GeForce GTX 980 5.2 4095 #3 GeForce GTX 980 5.2 4095 Wrote by Xueming Li @ Yifan Cheng Lab, UCSF ii glx-alternative-nvidia 0.7.1 amd64allows the selection of NVIDIA as GLX provider ii libcublas6.5:amd64 6.5.19-3 amd64NVIDIA cuBLAS Library rc libcuda1:amd64 352.41-1 amd64NVIDIA CUDA Driver Library ii libcudart6.5:amd64 6.5.19-3 amd64NVIDIA CUDA Runtime Library ii libcufft6.5:amd64 6.5.19-3 amd64NVIDIA cuFFT Library ii libcufftw6.5:amd64 6.5.19-3 amd64NVIDIA cuFFTW Library ii libcurand6.5:amd64 6.5.19-3 amd64NVIDIA cuRAND Library ii libcusparse6.5:amd64 6.5.19-3 amd64NVIDIA cuSPARSE Library ii libegl1-nvidia:amd64 352.79-3 amd64NVIDIA binary EGL libraries ii libgl1-nvidia-glx:amd64352.79-3 amd64NVIDIA binary OpenGL libraries ii libgles1-nvidia:amd64 352.79-3 amd64NVIDIA binary OpenGL|ES 1.x libraries ii libgles2-nvidia:amd64 352.79-3
Bug#816283: nvidia-cuda-toolkit: cuda broken after upgrade (x86_64, Jessie, GTX980)
On 2016-02-29 16:08, lumin wrote: > Hi, > > Thank you for reporting this bug. > > It seems like a GCC ABI issue, rather than that of CUDA > or nvidia-driver. > If the nvidia-driver package ships binary then they are > linked by GCC-5, which requires an different ABI to > the one of Jessie. Hence installing stuff from unstable > into Jessie may basically introduce ABI-broken ELFs. > If the truth lies there, this bug cannot be fixed. > And my advise is 1. try unstable 2. install nvidia stuff > > Does this bug appear in version 7.0.28-3 or some former version? > or is this the first time you trying to install these stuff from > unstable into stable? Before trying to upgrade, I was on nvidia-driver 352.41 nvidia-cuda-toolkit 6.5.19 from debian/testing. This was working. Now, I'm on nvidia-driver 352.79-4 nvidia-cuda-toolkit 7.0.28-4 from debian/unstable. This machine is part of a larger compute cluster, and I'd prefer to stick with Jessie, if possible. If your theory about incompatible ABI's is correct, it would not help installing "nvidia stuff", because the ABI would still be incompatible. Attached is the list of installed nvidia packages obtained with dpkg -l|grep -i nvidia Thanks, Alois > > Thanks :-) > ii glx-alternative-nvidia 0.7.1 amd64allows the selection of NVIDIA as GLX provider rc libcublas6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuBLAS Library ii libcublas7.0:amd64 7.0.28-4 amd64NVIDIA cuBLAS Library ii libcuda1:amd64 352.79-4 amd64NVIDIA CUDA Driver Library rc libcudart6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA CUDA Runtime Library ii libcudart7.0:amd64 7.0.28-4 amd64NVIDIA CUDA Runtime Library rc libcufft6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuFFT Library ii libcufft7.0:amd64 7.0.28-4 amd64NVIDIA cuFFT Library rc libcufftw6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuFFTW Library ii libcufftw7.0:amd64 7.0.28-4 amd64NVIDIA cuFFTW Library rc libcuinj64-6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA CUINJ Library (64-bit) ii libcuinj64-7.0:amd64 7.0.28-4 amd64NVIDIA CUINJ Library (64-bit) rc libcurand6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuRAND Library ii libcurand7.0:amd64 7.0.28-4 amd64NVIDIA cuRAND Library ii libcusolver7.0:amd64 7.0.28-4 amd64NVIDIA cuSOLVER Library rc libcusparse6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA cuSPARSE Library ii libcusparse7.0:amd64 7.0.28-4 amd64NVIDIA cuSPARSE Library ii libegl1-nvidia:amd64 352.79-4 amd64NVIDIA binary EGL libraries ii libgl1-nvidia-glx:amd64352.79-4 amd64NVIDIA binary OpenGL libraries ii libgles1-nvidia:amd64 352.79-4 amd64NVIDIA binary OpenGL|ES 1.x libraries ii libgles2-nvidia:amd64 352.79-4 amd64NVIDIA binary OpenGL|ES 2.x libraries rc libnppc6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA Performance Primitives core runtime library ii libnppc7.0:amd64 7.0.28-4 amd64NVIDIA Performance Primitives core runtime library rc libnppi6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA Performance Primitives for image processing runtime library ii libnppi7.0:amd64 7.0.28-4 amd64NVIDIA Performance Primitives for image processing runtime library rc libnpps6.5:amd64 6.5.19-3~bpo8+1 amd64NVIDIA Performance Primitives for signal processing runtime library ii libnpps7.0:amd64 7.0.28-4 amd64NVIDIA Performance Primitives for signal processing runtime library ii libnvidia-compiler:amd64 352.79-4 amd64NVIDIA runtime compiler library ii libnvidia-eglcore:amd64352.79-4 amd64NVIDIA binary EGL core libraries ii
Bug#816283: nvidia-cuda-toolkit: cuda broken after upgrade (x86_64, Jessie, GTX980)
Package: nvidia-cuda-toolkit Version: 7.0.28-4 Severity: important Dear Maintainer, I tried to upgrade the nvidia-drivers from 352.41-1 to the current version "352.79-3" and nvidia-cuda-toolkit to a matching version, cuda became unusable. Rebooting the machine did not help. I used the following two tests. First, nvidia-smi reports this error: # nvidia-smi Failed to initialize NVML: GPU access blocked by the operating system Second, I do have a short cuda program from here https://github.com/leschzinerlab/Motion-correction.git and can be installed with the following commands git clone https://github.com/leschzinerlab/Motion-correction.git cd Motion-correction/motioncorr_v2.1/src/ make -B ../bin/gpuinfo The compilation works without problems, but running the command ../bin/gpuinfo results in this message: Simple GPU info query. You have 32720 nVidia GPGPU. DeviceID NameVersion Memory(Mb) Failed to get GPU info of Device #0 Failed to get GPU info of Device #1 Failed to get GPU info of Device #2 Failed to get GPU info of Device #3 ... Failed to get GPU info of Device #32568 Failed to get GPU info of Device #32569 Failed to get GPU info of Device #32570 Failed to get GPU info of Device #32571 I tried a number of additional things, like using the packages in jessie-backports, and nvidia-cuda 7.0 from debian/unstable. None of these attempts solved the issue. Best, Alois -- System Information: Debian Release: 8.3 APT prefers stable APT policy: (990, 'stable'), (500, 'unstable'), (500, 'testing') Architecture: amd64 (x86_64) Kernel: Linux 3.16.0-4-amd64 (SMP w/12 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: systemd (via /run/systemd/system) Versions of packages nvidia-cuda-toolkit depends on: ii g++-4.8 4.8.4-1 ii g++-4.9 4.9.2-10 ii gcc-4.8 4.8.4-1 ii gcc-4.9 4.9.2-10 ii libc62.19-18+deb8u3 ii libgcc1 1:5.3.1-8 ii libnvvm3 7.0.28-4 ii libstdc++6 5.3.1-8 ii nvidia-cuda-dev 7.0.28-4 ii nvidia-profiler 7.0.28-4 ii ocl-icd-opencl-dev [opencl-dev] 2.2.3-1+deb8u1 Versions of packages nvidia-cuda-toolkit recommends: ii nvidia-cuda-doc 7.0.28-4 ii nvidia-cuda-gdb 7.0.28-4 pn nvidia-visual-profiler Versions of packages nvidia-cuda-toolkit suggests: pn libcupti-dev ii nvidia-driver 352.79-4 -- no debconf information
Bug#807016: octave-plplot: toggle_plplot_use is broken on Jessie
Package: octave-plplot Version: 5.10.0+dfsg-1 Severity: normal Tags: patch I've tryied to use plplot from Octave on Jessie. Octave and octave-plplot are installed. Starting octave and running "toggle_plplot_use cases the following problem: octave:1> use_plplot This function is deprecated. Use toggle_plplot_use instead. octave:2> toggle_plplot_use error: sq_string cannot be indexed with { error: evaluating argument list element number 1 error: called from: error: /usr/share/octave/site/m/PLplot/toggle_plplot_use.m at line 42, column 3 Attached are to possible patches, each of them works on Octave 3.8. patch_fix2_toggle_plplot_use.diff seems shorter and more elegant, though I did not check compatibility with older versions of Octave. Best regards, Alois -- System Information: Debian Release: 8.2 APT prefers stable APT policy: (990, 'stable'), (80, 'testing'), (50, 'unstable'), (40, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.16.0-4-amd64 (SMP w/4 CPU cores) Locale: LANG=de_AT.UTF-8, LC_CTYPE=de_AT.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: systemd (via /run/systemd/system) Versions of packages octave-plplot depends on: ii libc62.19-18+deb8u1 ii libgcc1 1:5.1.1-14 ii liboctave2 3.8.2-4 ii libplplot12 5.10.0+dfsg-1 ii libstdc++6 5.1.1-14 ii octave 3.8.2-4 octave-plplot recommends no packages. octave-plplot suggests no packages. -- no debconf information --- toggle_plplot_use.orig.m 2015-12-04 08:38:15.728228972 +0100 +++ toggle_plplot_use1.m 2015-12-04 08:39:37.018874374 +0100 @@ -39,7 +39,7 @@ use_plplot_path = plplot_octave_path; plplot_path_to_remove = char(strsplit(genpath(use_plplot_path),pathsep)); for use_plplot_i=1:size(plplot_path_to_remove)(1) - rmpath(plplot_path_to_remove{use_plplot_i}); + rmpath(strtrim(plplot_path_to_remove(use_plplot_i,:))); endfor if (strcmp (use_plplot_state, "on")) --- toggle_plplot_use.orig.m 2015-12-04 08:38:15.728228972 +0100 +++ toggle_plplot_use2.m 2015-12-04 08:37:18.766375290 +0100 @@ -37,7 +37,7 @@ endif use_plplot_path = plplot_octave_path; -plplot_path_to_remove = char(strsplit(genpath(use_plplot_path),pathsep)); +plplot_path_to_remove = strsplit(genpath(use_plplot_path),pathsep); for use_plplot_i=1:size(plplot_path_to_remove)(1) rmpath(plplot_path_to_remove{use_plplot_i}); endfor
Bug#742503: [Pkg-octave-devel] Bug#742503: octave 3.8.1 requires java
On 2015-07-16 07:00, Mike Miller wrote: On Wed, Jul 15, 2015 at 08:59:32 -0400, Mike Miller wrote: On Wed, Jul 15, 2015 at 10:00:33 +0200, Alois Schloegl wrote: Octave should not have a strong dependency on openjdk. Any of the following solutions would do: 1) the dependency should be changed to recommended, and/or 2) it should allow for some alternative java engine (e.g. gcj or whatever). (see also [4] ), 3) keeping the java dependency in a octave package octave-java as it was before 3.8.x Before #1 or #3 are feasible, some upstream work will need to be done to allow Octave to detect the (presence of / version of / path to) JRE at runtime. Please see upstream bug #40111. Minor clarification: currently, when Octave is compiled with Java support but the JRE is not actually present, the error is handled and reported: octave:1 javaMethod (getRuntime, java.lang.Runtime) error: javaMethod: /usr/lib/jvm/java-7-openjdk-amd64/jre/lib/amd64/server/libjvm.so: failed to load: /usr/lib/jvm/java-7-openjdk-amd64/jre/lib/amd64/server/libjvm.so: cannot open shared object file: No such file or directory octave:2 So Octave does handle this to some degree. The error reporting is not ideal and could be improved as part of #40111. But it does look feasible to be able to run Octave without a JRE. Users who opt out of installing default-jre-headless will see errors like the above if they try to use java functions, or functions that rely on java such as those in the octave-io package. Thoughts? I changed the control file in the debian package (see attachments). default-jre-headless is now not in Depends but in Build-Depends and Recommends. The resulting octave-debian packages does the job. In case somebody wants to use the java-interface, the above error message seems good enough. Please consider repackaging Octave with the modified control file. Alois --- octave-3.8.4.deb/DEBIAN/control.orig 2015-07-16 09:11:00.860280895 +0200 +++ octave-3.8.4/DEBIAN/control 2015-07-16 09:38:50.716915697 +0200 @@ -1,10 +1,11 @@ Package: octave Version: 3.8.2-4 -Architecture: arm64 +Architecture: amd64 Maintainer: Debian Octave Group pkg-octave-de...@lists.alioth.debian.org -Installed-Size: 3327 -Depends: libamd2.3.1, libarpack2 (= 2.1), libblas3 | libblas.so.3, libc6 (= 2.17), libcamd2.3.1, libccolamd2.8.0, libcholmod2.1.2, libcolamd2.8.0, libcxsparse3.1.2, libfftw3-double3, libfftw3-single3, libfltk-gl1.3 (= 1.3.0), libfltk1.3 (= 1.3.1), libfontconfig1 (= 2.11), libfreetype6 (= 2.2.1), libgcc1 (= 1:4.7), libgl1-mesa-glx | libgl1, libglpk36 (= 4.51), libglu1-mesa | libglu1, libgomp1 (= 4.2.1), libgraphicsmagick++3, libgraphicsmagick3 (= 1.3.5), liblapack3 | liblapack.so.3, liboctave2 (= 3.8.2-4), libqhull6 (= 2012.1), libqrupdate1 (= 1.0), libqscintilla2-11, libqt4-network (= 4:4.5.3), libqtcore4 (= 4:4.7.0~beta1), libqtgui4 (= 4:4.8.0), libstdc++6 (= 4.9), libumfpack5.6.2, libx11-6, texinfo, octave-common (= 3.8.2-4), default-jre-headless -Recommends: gnuplot-x11 | gnuplot-qt, libopenblas-base | libatlas3-base, pstoedit +Installed-Size: 3497 +Build-Depends: default-jre-headless +Depends: libamd2.3.1, libarpack2 (= 2.1), libblas3 | libblas.so.3, libc6 (= 2.14), libcamd2.3.1, libccolamd2.8.0, libcholmod2.1.2, libcolamd2.8.0, libcxsparse3.1.2, libfftw3-double3, libfftw3-single3, libfltk-gl1.3 (= 1.3.0), libfltk1.3 (= 1.3.1), libfontconfig1 (= 2.11), libfreetype6 (= 2.2.1), libgcc1 (= 1:4.1.1), libgl1-mesa-glx | libgl1, libglpk36 (= 4.51), libglu1-mesa | libglu1, libgomp1 (= 4.2.1), libgraphicsmagick++3, libgraphicsmagick3 (= 1.3.5), liblapack3 | liblapack.so.3, liboctave2 (= 3.8.2-4), libqhull6 (= 2012.1), libqrupdate1 (= 1.0), libqscintilla2-11, libqt4-network (= 4:4.5.3), libqtcore4 (= 4:4.7.0~beta1), libqtgui4 (= 4:4.8.0), libstdc++6 (= 4.9), libumfpack5.6.2, libx11-6, texinfo, octave-common (= 3.8.2-4) +Recommends: gnuplot-x11 | gnuplot-qt, libopenblas-base | libatlas3-base, pstoedit, default-jre-headless Suggests: octave-info, octave-doc, octave-htmldoc Conflicts: octave-java, octave3.2 Breaks: dynare (= 4.4.1-1), libsbml5-octave (= 5.8.0-2), octave-audio (= 1.1.4-4), octave-biosig (= 1.3.0-2+b1), octave-communications (= 1.2.0-1), octave-control (= 2.6.2-1), octave-econometrics (= 1:1.1.1-2), octave-fixed (= 0.7.10-5+b1), octave-gdf (= 0.1.2-2+b1), octave-general (= 1.3.2-2), octave-geometry (= 1.7.0-1), octave-gmt (= 4.5.11-1), octave-gsl (= 1.0.8-5), octave-image (= 2.0.0-3), octave-io ( 1.3), octave-java (= 1.2.9-2), octave-lhapdf (= 5.9.1-3), octave-linear-algebra (= 2.2.0-1), octave-miscellaneous (= 1.2.0-2), octave-mpi (= 1.1.1-1), octave-nan (= 2.5.9-1), octave-nlopt (= 2.4.1+dfsg-1), octave-nurbs (= 1.3.7-1), octave-ocs (= 0.1.3-1), octave-octcdf (= 1.1.6-1), octave-octgpr (= 1.2.0-3), octave-odepkg (= 0.8.4-1), octave-openmpi-ext (= 1.1.0-2), octave-optim (= 1.2.2-2), octave-optiminterp (= 0.3.4-1), octave-parallel (= 2.2.0-1), octave-pfstools (= 1.8.5-1
Bug#742503: [Pkg-octave-devel] Bug#742503: octave 3.8.1 requires java
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Recently, I tried to remove openjdk-[6-8] from a number of machines because of the large number of open security issues with openjdk (currently there are 23, 21 and 20 open security issues on openjdk-6, openjdk-7, and openjdk-8, resp.). I learned that due to some package dependencies, octave gets removed too. That's not desired. Octave should not have a strong dependency on openjdk. Any of the following solutions would do: 1) the dependency should be changed to recommended, and/or 2) it should allow for some alternative java engine (e.g. gcj or whatever). (see also [4] ), 3) keeping the java dependency in a octave package octave-java as it was before 3.8.x Alois [1] https://security-tracker.debian.org/tracker/source-package/openjdk-6 [2] https://security-tracker.debian.org/tracker/source-package/openjdk-7 [3] https://security-tracker.debian.org/tracker/source-package/openjdk-8 [4] https://lists.alioth.debian.org/pipermail/pkg-octave-devel/2013-December /010469.html -BEGIN PGP SIGNATURE- Version: GnuPG v2 iQEcBAEBCAAGBQJVphMhAAoJEBdbD2J3nb+MIMsH/2oevdIY3lJsIrLpna49mKcj zQo1Mc3zunsEZWLG8R+FyezLUrhnS0zeAIsjVSFbk3UznqaBuAgic+PjOE5uJDoU DvkC8pi81VDIHvycIsNiHhaKXdxtmTZIVf24b3lIWB+N/18STqP9SzjByu4klLXR pwG2Uq32cbbyWPm/xwSnQMew/5dWd8tEdsDaGGn1yD2pneiv2g4DbVSiVtGgEoBA b7b66vLVwFH9jHXbzxRhaLU3Gbntk8KX2/F2YHjtGGjaM/GUQ7aEC1R0UTSfXfnM /fpNkWhFmZW+CPFN2gmo7sU0xJqSTkU0tDLTE9FlO2KEuIAQjarFITVb0Fl7+mQ= =ApCl -END PGP SIGNATURE- -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#698286: incorrect value for max vmem is reported : possible fix
The attached patch should address this issue. The patch is essentially this commit af73e397da01a4f70c01387f284e87262e42e7a3 Author: Dave Love d.l...@liverpool.ac.uk Date: Thu Apr 21 12:44:36 2011 + Introduce lUlong64 to fix wrong vmem reporting on 64-bit GNU/Linux (from Univa) from http://arc.liv.ac.uk/repos/git/sge combined with a minor change for backporting to 6.2u5-7.1 --- a/gridengine_add64bit_support.patch 2013-06-05 14:24:17.415377163 +0200 +++ b/gridengine_add64bit_support.patch 2013-06-05 15:15:59.137610718 +0200 @@ -366,7 +366,7 @@ + if (pos 0) { + /* someone has called lGetPosUlong64() */ + /* makro with an invalid nm*/ -+ CRITICAL((SGE_EVENT, SFNMAX, MSG_CULL_GETPOSULONG64_GOTINVALIDPOSITION )); ++ CRITICAL((SGE_EVENT, MSG_CULL_GETPOSULONG64_GOTINVALIDPOSITION )); + DEXIT; + abort(); + } I've tested the patch using this procedure: On wheezy machine, I downloaded the source package of gridengine 6.2u5-7.1 apt-get -b source gridengine Then applied the attached patch to the extracted source, and rebuild the package in the following way: dpkg-buildpackage -rfakeroot -uc -b and it builds cleanly and can be installed. diff --git a/source/common/basis_types.h b/source/common/basis_types.h index aa4fb28..12d21f3 100644 --- a/source/common/basis_types.h +++ b/source/common/basis_types.h @@ -30,6 +30,7 @@ * All Rights Reserved. * / +/* Portions of this code are Copyright (c) 2011 Univa Corporation. */ /*___INFO__MARK_END__*/ #include sys/types.h @@ -122,17 +123,29 @@ extern C { # define u_long32 u_long #endif +#if defined(TARGET_64BIT) +# define u_long64 u_long +#elif defined(WIN32NATIVE) +# define u_long64 unsigned long long +#elif defined(FREEBSD) || defined(NETBSD) +# define u_long64 uint64_t +#else +# define u_long64 unsigned long long +#endif + #define U_LONG32_MAX 4294967295UL #define LONG32_MAX 2147483647 /* set sge_u32 and sge_x32 for 64 or 32 bit machines */ /* sge_uu32 for strictly unsigned, not nice, but did I use %d for an unsigned? */ #if defined(TARGET_64BIT) || defined(FREEBSD) || defined(NETBSD) +# define sge_u64%ld # define sge_u32%d # define sge_uu32 %u # define sge_x32%x # define sge_fu32 d #else +# define sge_u64%lld # define sge_u32%ld # define sge_uu32 %lu # define sge_x32%lx diff --git a/source/daemons/common/procfs.c b/source/daemons/common/procfs.c index f78513a..6882ca5 100644 --- a/source/daemons/common/procfs.c +++ b/source/daemons/common/procfs.c @@ -28,6 +28,7 @@ * All Rights Reserved. * / +/* Portions of this code are Copyright (c) 2011 Univa Corporation. */ /*___INFO__MARK_END__*/ #if !defined(COMPILE_DC) @@ -554,7 +555,7 @@ time_t last_time lSetPosUlong(pr, pos_utime, utime); lSetPosUlong(pr, pos_stime, stime); - lSetPosUlong(pr, pos_vsize, vsize); + lSetPosUlong64(pr, pos_vsize, vsize); close(fd); } @@ -770,7 +771,7 @@ time_t last_time proc_elem-proc.pd_stime = ((double)lGetPosUlong(pr, pos_stime)) / sysconf(_SC_CLK_TCK); /* could retrieve uid/gid using stat() on stat file */ - proc_elem-vmem = lGetPosUlong(pr, pos_vsize); + proc_elem-vmem = lGetPosUlong64(pr, pos_vsize); /* * I/O accounting diff --git a/source/libs/cull/cull_db.c b/source/libs/cull/cull_db.c index f848902..4bbd6b2 100644 --- a/source/libs/cull/cull_db.c +++ b/source/libs/cull/cull_db.c @@ -28,6 +28,7 @@ * All Rights Reserved. * / +/* Portions of this code are Copyright (c) 2011 Univa Corporation. */ /*___INFO__MARK_END__*/ #include stdio.h #include stdlib.h @@ -383,6 +384,9 @@ lList *lJoin(const char *name, int nm0, const lList *lp0, case lUlongT: needed = (ep0-cont[lp0_pos].ul == ep1-cont[lp1_pos].ul); break; +case lUlong64T: + needed = (ep0-cont[lp0_pos].ul64 == ep1-cont[lp1_pos].ul64); + break; case lStringT: needed = !strcmp(ep0-cont[lp0_pos].str, ep1-cont[lp1_pos].str); break; diff --git a/source/libs/cull/cull_dump_scan.c b/source/libs/cull/cull_dump_scan.c index 95472bb..2df8af7 100644 --- a/source/libs/cull/cull_dump_scan.c +++ b/source/libs/cull/cull_dump_scan.c @@ -28,6 +28,7 @@ * All Rights Reserved. * / +/* Portions of this code are Copyright (c) 2011 Univa Corporation. */ /*___INFO__MARK_END__*/ #include stdio.h #include stdlib.h @@ -63,6 +64,7 @@ static int fGetKet(FILE *fp); static int fGetDescr(FILE *fp, lDescr *dp); static int fGetInt(FILE *fp,
Bug#701246: biosig4c++: ftbfs with GCC-4.8
The same bug has been already reported against liboctave-dev http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=705485 This bug report can be merged. Alois -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#701246: debian bugreport in biosig4c++: ftbfs with GCC-4.8
In file included from /usr/include/octave-3.6.2/octave/mexproto.h:52:0, from /usr/include/octave-3.6.2/octave/mex.h:77, from mexSLOAD.cpp:15: /usr/include/c++/4.8/cstdlib: In function 'long long int std::abs(long long int)': /usr/include/c++/4.8/cstdlib:174:20: error: declaration of C function 'long long int std::abs(long long int)' conflicts with abs(long long __x) { return __builtin_llabs (__x); } The files /usr/include/octave-3.6.2/octave/mexproto.h and /usr/include/octave-3.6.2/octave/mex.h are not part of biosig4c++ but from octave. Therefore, this bug report should be assigned to octave (not biosig4c++). Alois -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#703591: pdftk - support for AES encrypted pdf files missing
Package: pdftk Version: 1.44-7 Severity: Wishlist I got an pdf file that could not be opened with pdftk. pdftk /tmp/q.pdf data_dump Error: Failed to open PDF file: /tmp/q.pdf Done. Input errors, so no output created. Unfortunately, the file contains personal information, so I can not share it. The same pdf can be opened with evince and okular (without needing a password). Investigating the issue, I found this site: http://www.thbimage.com/pdf/pdftk_can_t_open_pdf_file_gnr2sr_5lv_1_news_motzarella_org_.html It describes the same problem, and my file contains also this sequence .../StrF/StdCF/CF/StdCF/CFM/AESV2/AuthEvent/DocOpen/Length 16/StmF/StdCF/U indicating an AES encryption. The error message is different than in http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=454140 so it's a different bug. The workaround is, opening the file with okular and printing it to a new pdf file. However, it would be great if pdftk could also handle AES encrypted pdf files in seamless way. Alois -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#698286: incorrect value for max vmem is reported
Package: gridengine* Version: 6.2u5-1 When a job has finished, the max vmem value as it was actually used is reported to the user. Unfortunately, the reported value is never larger than 4G also the application has clearly used more than 4G of Memory. I've been testing this with Octave as well as mathematica. Based on some preliminary tests, it seems that the reported value is the (real value module 2^32). Thus, if more then 4G of vmem are used, the reported maxvmem value is meaningless. attached are: - an Octave sample script, allocating 600M of random double precision numbers resulting in 4.8GB memory need. - the script for submission to the grid engine qsub memtest03.m The e.mail report contains this: Job 4709696 (MEMTEST02) Complete User = schloegl Queue= mai...@nick13.ista.local Host = nick13.ista.local Start Time = 01/16/2013 11:19:11 End Time = 01/16/2013 11:19:36 User Time= 00:00:23 System Time = 00:00:02 Wallclock Time = 00:00:25 CPU = 00:00:25 Max vmem = 710.246M Exit Status = 0 The correct value of Max vmem should be 4.47GB or more. But its close to mod(8*6e8,2^32)=505032704 (this is plausible when we assume 200MB overhead by Octave). maxvmem is an important variable of optimizing the use of a cluster. Users can control how much memory is is required, which is then used by the grid engine for scheduleing. For this reason, its important that a meaningful value for maxvmem is reported. Therefore I suggest a severity level important. Severity: important Alois % Octave script clear x=randn(6e8,1); size(x) whos exit memtest03.sh Description: Bourne shell script
Bug#631831: [Pkg-octave-devel] Bug#631831: Bug#631831: build octave with --enable-openmp
On 06/30/11 04:11, Jordi Gutiérrez Hermoso wrote: On 29 June 2011 18:54,schlo...@ist.ac.at wrote: On Tue, Jun 28, 2011 at 09:03:32AM +0200, Alois Schloegl wrote: adding -lgomp to mkoctfile call does not do it. And I've not found any another option to enable openmp at this stage. Did you pass -fopenmp to the compiler? Thomas I tried, but mkoctfile does not accept -fopenmp $ mkoctfile --mex -fopenmp covm_mex.cpp -lz -lgomp mkoctfile: unrecognized argument -fopenmp Pass it as an environment variable: CXXFLAGS=-O2 -fopenmp mkoctifle --mex covm_mex.cpp -lz -lgomp Does that work? Jordi, thanks for the hint. I tried this too, as well as CFLAGS=-g -O2 -pthread -fopenmp mkoctfile-3.2.4 --mex covm_mex.cpp -lz -lgomp CXXFLAGS=-g -O2 -pthread -fopenmp mkoctfile-3.2.4 --mex covm_mex.cpp -lz -lgomp In none of these case, more than one CPU core was used. The ratio between cputime() and wall time toc() was always close to 1, and also top showed only 100 % cpu usage at most; with openmp enabled, I saw up to 1200% at the 12 core machine. Alois -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#631831: [Pkg-octave-devel] Bug#631831: build octave with --enable-openmp
On 06/27/11 20:23, Thomas Weber wrote: Hi Alois, On Mon, Jun 27, 2011 at 05:30:18PM +0200, Alois Schloegl wrote: Package: octave3.2 Severity: wishlist Octave should be configured with --enable-openmp. This would enable the use of multiple cpus. https://mailman.cae.wisc.edu/pipermail/octave-maintainers/2011-May/023838.html Is it really needed to build Octave with OpenMP support or just octave-nan? Thomas adding -lgomp to mkoctfile call does not do it. And I've not found any another option to enable openmp at this stage. If Octave is not built with --enable-openmp, the mex-files in octave-nan are not built with openmp enabled. Alois -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#611070: Package: installation-reports
Package: installation-reports Boot method: CD Image version: http://cdimage.debian.org/cdimage/daily-builds/daily/arch-latest/i386/iso-cd/ Date: Jan 25 2011 10:00-11:00 CET GMT+1 Machine: Asus EeePC 1008PGO Processor: Atom N450 Memory: 1 GB Partitions: df -Tl will do; the raw partition table is preferred [default settings, single partition] # df -Tl FilesystemType 1K-blocks Used Available Use% Mounted on /dev/sda1 ext3 238395008 3926224 222359012 2% / tmpfstmpfs 512408 0512408 0% /lib/init/rw udev tmpfs 508060 232507828 1% /dev tmpfstmpfs 512408 0512408 0% /dev/shm Output of lspci -knn (or lspci -nn): # lspci -knn 00:00.0 Host bridge [0600]: Intel Corporation N10 Family DMI Bridge [8086:a010] Subsystem: ASUSTeK Computer Inc. Device [1043:83ac] Kernel driver in use: agpgart-intel 00:02.0 VGA compatible controller [0300]: Intel Corporation N10 Family Integrated Graphics Controller [8086:a011] Subsystem: ASUSTeK Computer Inc. Device [1043:83ac] Kernel driver in use: i915 00:02.1 Display controller [0380]: Intel Corporation N10 Family Integrated Graphics Controller [8086:a012] Subsystem: ASUSTeK Computer Inc. Device [1043:83ac] 00:1b.0 Audio device [0403]: Intel Corporation N10/ICH 7 Family High Definition Audio Controller [8086:27d8] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:8398] Kernel driver in use: HDA Intel 00:1c.0 PCI bridge [0604]: Intel Corporation N10/ICH 7 Family PCI Express Port 1 [8086:27d0] (rev 02) Kernel driver in use: pcieport 00:1c.2 PCI bridge [0604]: Intel Corporation N10/ICH 7 Family PCI Express Port 3 [8086:27d4] (rev 02) Kernel driver in use: pcieport 00:1c.3 PCI bridge [0604]: Intel Corporation N10/ICH 7 Family PCI Express Port 4 [8086:27d6] (rev 02) Kernel driver in use: pcieport 00:1d.0 USB Controller [0c03]: Intel Corporation N10/ICH 7 Family USB UHCI Controller #1 [8086:27c8] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:83ad] Kernel driver in use: uhci_hcd 00:1d.1 USB Controller [0c03]: Intel Corporation N10/ICH 7 Family USB UHCI Controller #2 [8086:27c9] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:83ad] Kernel driver in use: uhci_hcd 00:1d.2 USB Controller [0c03]: Intel Corporation N10/ICH 7 Family USB UHCI Controller #3 [8086:27ca] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:83ad] Kernel driver in use: uhci_hcd 00:1d.3 USB Controller [0c03]: Intel Corporation N10/ICH 7 Family USB UHCI Controller #4 [8086:27cb] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:83ad] Kernel driver in use: uhci_hcd 00:1d.7 USB Controller [0c03]: Intel Corporation N10/ICH 7 Family USB2 EHCI Controller [8086:27cc] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:83ad] Kernel driver in use: ehci_hcd 00:1e.0 PCI bridge [0604]: Intel Corporation 82801 Mobile PCI Bridge [8086:2448] (rev e2) 00:1f.0 ISA bridge [0601]: Intel Corporation NM10 Family LPC Controller [8086:27bc] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:83ad] 00:1f.2 SATA controller [0106]: Intel Corporation N10/ICH7 Family SATA AHCI Controller [8086:27c1] (rev 02) Subsystem: ASUSTeK Computer Inc. Device [1043:83ad] Kernel driver in use: ahci 01:00.0 Ethernet controller [0200]: Atheros Communications AR8132 Fast Ethernet [1969:1062] (rev c0) Subsystem: ASUSTeK Computer Inc. Device [1043:838a] Kernel driver in use: atl1c 02:00.0 Network controller [0280]: Atheros Communications Inc. AR9285 Wireless Network Adapter (PCI-Express) [168c:002b] (rev 01) Subsystem: AzureWave Device [1a3b:1106] Kernel driver in use: ath9k Base System Installation Checklist: [O] = OK, [E] = Error (please elaborate below), [ ] = didn't try it Initial boot: [O] Detect network card:[O] Configure network: [O] Detect CD: [O] Load installer modules: [O] Detect hard drives: [O] Partition hard drives: [ ] Install base system:[O] Clock/timezone setup: [O] User/password setup:[O] Install tasks: [O] Install boot loader:[O] Overall install:[O] Comments/Problems: No Problems Description of the install, in prose, and any thoughts, comments and ideas you had during the initial install. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#419556: [Pkg-octave-devel] Bug#419556: ambigous results in matrix multipliication
Original-Nachricht Datum: Mon, 16 Apr 2007 21:38:12 +0200 Von: Thomas Weber [EMAIL PROTECTED] An: Alois Schloegl [EMAIL PROTECTED], [EMAIL PROTECTED] Betreff: Re: [Pkg-octave-devel] Bug#419556: ambigous results in matrix multipliication Hi, Am Montag, 16. April 2007 17:32 schrieb Alois Schloegl: One expects that a matrix multiplication (A*B)' gives the same result than for (B'*A'). This is not the case for the debian octave package (tested with 2.9.10-3 and 2.1.73-13) Here is the test script: repmat(0,2)*repmat(NaN,2) repmat(NaN,2)*repmat(0,2) repmat(0,2)*repmat(inf,2) repmat(inf,2)*repmat(0,2) version octave_config_info('BLAS_LIBS') I take it for granted that you didn't have ATLAS installed. Is this correct? Actually, it was the same machine that I used for compiling Octave with Atlas. So atlas-base-dev (which requires also atlas-base) was installed when I tested Octave 2.9.10-3. Unless I'm mistaken, the reference implementation of dgemm is supposed to do exactly this, see http://velveeta.che.wisc.edu/octave/lists/bug-octave/2002/310 IMHO, this is a bug (and not a feature) of dgemm. Especially, because (A*B)' is not the same as (B'*A'). octave:16 [nan nan]*[0;0] ans = 0 octave:17 [0,0]*[nan nan]' ans = NaN octave:20 [0,0]*[inf inf]' ans = NaN octave:21 [inf inf]*[0;0] ans = 0 The problem seems to be in the lapack or blas library. Because this problem disappears (the result is always NaN) when I compile Octave from the sources using Atlas. I guess the best fix for this problem is installing one of the atlas3-* packages. Do you mean that installing atlas3-* besides the debian octave package solves the problem? I do not think so, because the machine I used had Atlas installed. If you mean, that atlas3-* should be installed when the Octave package is compiled for debian, yes I agree to that. Perhaps, the dependency between Octave and atlas-* need to be changed, such that atlas is mandatory instead of one of several alternatives. Regards Thomas Regards, Alois -- Feel free - 10 GB Mailbox, 100 FreeSMS/Monat ... Jetzt GMX TopMail testen: http://www.gmx.net/de/go/topmail -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#419556: ambigous results in matrix multipliication
Package: Octave2.9 Status: install ok installed Priority: optional Section: math Installed-Size: 26220 Maintainer: Debian Octave Group [EMAIL PROTECTED] Architecture: i386 Version: 2.9.10-3 Severity: Important One expects that a matrix multiplication (A*B)' gives the same result than for (B'*A'). This is not the case for the debian octave package (tested with 2.9.10-3 and 2.1.73-13) Here is the test script: repmat(0,2)*repmat(NaN,2) repmat(NaN,2)*repmat(0,2) repmat(0,2)*repmat(inf,2) repmat(inf,2)*repmat(0,2) version octave_config_info('BLAS_LIBS') octave2.9:27 repmat(0,2)*repmat(NaN,2) ans = NaN NaN NaN NaN octave2.9:28 repmat(NaN,2)*repmat(0,2) ans = 0 0 0 0 octave2.9:29 repmat(0,2)*repmat(inf,2) ans = NaN NaN NaN NaN octave2.9:30 repmat(inf,2)*repmat(0,2) ans = 0 0 0 0 octave2.9:31 version ans = 2.9.10 octave2.9:32 octave_config_info('BLAS_LIBS') ans = -llapack-3 -lblas-3 The problem seems to be in the lapack or blas library. Because this problem disappears (the result is always NaN) when I compile Octave from the sources using Atlas. The octae_config_info says octave:64 version ans = 2.9.10+ octave:65 octave_config_info('BLAS_LIBS') ans = -llapack -lcblas -lf77blas -latlas A related problem has been also discussed here and here http://velveeta.che.wisc.edu/octave/lists/bug-octave/2001/104 http://velveeta.che.wisc.edu/octave/lists/bug-octave/2002/293 Obviously, it is not fixed for the debian package. Alois -- Feel free - 10 GB Mailbox, 100 FreeSMS/Monat ... Jetzt GMX TopMail testen: http://www.gmx.net/de/go/topmail -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]