Bug#597962: Seems to have resolved itself
severity 597962 important kthxbye On Mon, Nov 15, 2010 at 20:10:43 +, Ian Campbell wrote: I suspect the ability to reproduce the issue depends somewhat on the pattern of other memory allocations going on etc since that will effect how quickly the erroneously freed memory gets reused (if at all), which might explain away the small number of people who are seeing this. I tried (reasonably hard) to construct a smaller test case which exhibited the problem but I was unable to do so. Seems to fit the bill for important: a bug which has a major effect on the usability of a package, without rendering it completely unusable to everyone. Cheers, Julien signature.asc Description: Digital signature
Processed: Re: Bug#597962: Seems to have resolved itself
Processing commands for cont...@bugs.debian.org: severity 597962 important Bug #597962 [quodlibet] quodlibet: Error Occurred dialog on playback Severity set to 'important' from 'grave' kthxbye Stopping processing here. Please contact me if you need assistance. -- 597962: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=597962 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#597962: Seems to have resolved itself
Problem exist, i used lates sid (all packages synced) and QL crushed with python-gobject 2.21.4+is.2.21.3-1, but work fine with 2.21.4 version. 2010/11/15 Ian Campbell i...@hellion.org.uk On Mon, 2010-11-15 at 14:13 -0500, Christine Spang wrote: On Sun, Nov 14, 2010 at 11:23:42AM -0500, Mike Paul wrote: I was going to take a shot at debugging this today, since it's on the RC bug list for squeeze, but it seems to have resolved itself at some point, at least in unstable. QL starts without errors, and it's playing music right now. The last time I tried it was weeks ago, so unfortunately I don't know which of the many package updates between then and now was responsible for the change. I also don't have a system running squeeze, so I can't test whether it works there. But it's working for me in sid. Thanks for taking a look at this, Mike! Were you able to reproduce this bug before? Ian, can you still reproduce this bug? I'm afraid so. I reverted my hacky patch and the issue reoccurred. The system was pretty out of date so I did an aptitude upgrade, which unfortunately didn't help (aptitude upgrade log appended anyway). I didn't think it was worth logging out and back in again or anything like that, but I can try if someone thinks it is worth a go. I'm hesitant to close this bug without hearing from Ian, but perhaps downgrading it would be appropriate since I have yet to encounter someone else who is able to reproduce, which suggests that it only appears (appeared?) in very limited circumstances. I think this bug probably more correctly belongs to the gst python bindings (or maybe the gobject bindings framework generally), since I can't see how this can be expected to work correctly unless they explicitly take action to integrate the gobject/gst reference counting of the C objects with python's reference counting when adding objects to queues (or whatever, I'm not that familiar with gst terminology). It can be worked around in the python user of the bindings by keeping extra python references to the objects in question, as my hacky patch does, but that doesn't seem right to me. I suspect the ability to reproduce the issue depends somewhat on the pattern of other memory allocations going on etc since that will effect how quickly the erroneously freed memory gets reused (if at all), which might explain away the small number of people who are seeing this. I tried (reasonably hard) to construct a smaller test case which exhibited the problem but I was unable to do so. Does python have a debugging facility which will explicitly poison memory on free? Since this issue is in the C bindings perhaps it wouldn't help much without explicit bindings support. Perhaps a similar facility exists in gobject and would be more use here? G_DEBUG=gc-friendly looks pretty close, does that help anyone else reproduce? (it seem to need a debug version of glib) Valgrind might also help, a bit of a long shot though. Ian. Aptitude 0.6.3: log report Mon, Nov 15 2010 19:39:16 + IMPORTANT: this log only lists intended actions; actions which fail due to dpkg problems may not be completed. Will install 81 packages, and remove 9 packages. 28.6 MB of disk space will be freed === [REMOVE, NOT USED] libqt4-assistant [REMOVE, NOT USED] libqt4-help [REMOVE, NOT USED] libqt4-scripttools [REMOVE, NOT USED] libqt4-test [REMOVE, NOT USED] python-qt4 [REMOVE, NOT USED] python-qt4-dbus [REMOVE, NOT USED] python-reportlab [REMOVE, NOT USED] python-reportlab-accel [REMOVE, NOT USED] python-sip [HOLD] gthumb [HOLD] gthumb-data [HOLD] lvm2 [UPGRADE] acpid 1:2.0.6-1 - 1:2.0.6-2 [UPGRADE] base-files 5.9 - 5.10 [UPGRADE] bcc 0.16.17-3 - 0.16.17-3.1 [UPGRADE] bin86 0.16.17-3 - 0.16.17-3.1 [UPGRADE] bind9-doc 1:9.7.2.dfsg.P2-1 - 1:9.7.2.dfsg.P2-2 [UPGRADE] bind9-host 1:9.7.2.dfsg.P2-1 - 1:9.7.2.dfsg.P2-2 [UPGRADE] bsdmainutils 8.0.15 - 8.0.16 [UPGRADE] cdebconf 0.150 - 0.151 [UPGRADE] cpp-4.4 4.4.5-6 - 4.4.5-7 [UPGRADE] cups 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-bsd 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-client 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-common 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-ppdc 1.4.4-6 - 1.4.4-7 [UPGRADE] debian-cd 3.1.3 - 3.1.4 [UPGRADE] debianutils 3.4.1 - 3.4.2 [UPGRADE] debootstrap 1.0.25 - 1.0.26 [UPGRADE] dnsutils 1:9.7.2.dfsg.P2-1 - 1:9.7.2.dfsg.P2-2 [UPGRADE] eatmydata 15-3 - 15-4 [UPGRADE] elks-libc 0.16.17-3 - 0.16.17-3.1 [UPGRADE] evolution 2.30.3-3 - 2.30.3-4 [UPGRADE] evolution-common 2.30.3-3 - 2.30.3-4 [UPGRADE] evolution-plugins 2.30.3-3 - 2.30.3-4 [UPGRADE] foomatic-filters 4.0.5-5 - 4.0.5-6 [UPGRADE] g++-4.4 4.4.5-6 - 4.4.5-7 [UPGRADE] gcc-4.4 4.4.5-6 - 4.4.5-7 [UPGRADE] gcc-4.4-base 4.4.5-6 - 4.4.5-7 [UPGRADE] gdm 2.20.11-2 - 2.20.11-4 [UPGRADE] gnome-about 2.30.2-1 - 2.30.2-2 [UPGRADE] gnome-desktop-data 2.30.2-1 -
Bug#597962: Seems to have resolved itself
On Mon, 2010-11-15 at 14:13 -0500, Christine Spang wrote: Thanks for taking a look at this, Mike! Were you able to reproduce this bug before? Yes, I encountered it originally when it was known as bug #592587, and it was resolved by upgrading to python-gobject version 2.21.4-1, then broken again by version 2.21.4+is.2.21.3-1. -- Mike Paul m...@wyzardry.net -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#597962: Seems to have resolved itself
On Sun, Nov 14, 2010 at 11:23:42AM -0500, Mike Paul wrote: I was going to take a shot at debugging this today, since it's on the RC bug list for squeeze, but it seems to have resolved itself at some point, at least in unstable. QL starts without errors, and it's playing music right now. The last time I tried it was weeks ago, so unfortunately I don't know which of the many package updates between then and now was responsible for the change. I also don't have a system running squeeze, so I can't test whether it works there. But it's working for me in sid. Thanks for taking a look at this, Mike! Were you able to reproduce this bug before? Ian, can you still reproduce this bug? I'm hesitant to close this bug without hearing from Ian, but perhaps downgrading it would be appropriate since I have yet to encounter someone else who is able to reproduce, which suggests that it only appears (appeared?) in very limited circumstances. Christine -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#597962: Seems to have resolved itself
On Mon, 2010-11-15 at 14:13 -0500, Christine Spang wrote: On Sun, Nov 14, 2010 at 11:23:42AM -0500, Mike Paul wrote: I was going to take a shot at debugging this today, since it's on the RC bug list for squeeze, but it seems to have resolved itself at some point, at least in unstable. QL starts without errors, and it's playing music right now. The last time I tried it was weeks ago, so unfortunately I don't know which of the many package updates between then and now was responsible for the change. I also don't have a system running squeeze, so I can't test whether it works there. But it's working for me in sid. Thanks for taking a look at this, Mike! Were you able to reproduce this bug before? Ian, can you still reproduce this bug? I'm afraid so. I reverted my hacky patch and the issue reoccurred. The system was pretty out of date so I did an aptitude upgrade, which unfortunately didn't help (aptitude upgrade log appended anyway). I didn't think it was worth logging out and back in again or anything like that, but I can try if someone thinks it is worth a go. I'm hesitant to close this bug without hearing from Ian, but perhaps downgrading it would be appropriate since I have yet to encounter someone else who is able to reproduce, which suggests that it only appears (appeared?) in very limited circumstances. I think this bug probably more correctly belongs to the gst python bindings (or maybe the gobject bindings framework generally), since I can't see how this can be expected to work correctly unless they explicitly take action to integrate the gobject/gst reference counting of the C objects with python's reference counting when adding objects to queues (or whatever, I'm not that familiar with gst terminology). It can be worked around in the python user of the bindings by keeping extra python references to the objects in question, as my hacky patch does, but that doesn't seem right to me. I suspect the ability to reproduce the issue depends somewhat on the pattern of other memory allocations going on etc since that will effect how quickly the erroneously freed memory gets reused (if at all), which might explain away the small number of people who are seeing this. I tried (reasonably hard) to construct a smaller test case which exhibited the problem but I was unable to do so. Does python have a debugging facility which will explicitly poison memory on free? Since this issue is in the C bindings perhaps it wouldn't help much without explicit bindings support. Perhaps a similar facility exists in gobject and would be more use here? G_DEBUG=gc-friendly looks pretty close, does that help anyone else reproduce? (it seem to need a debug version of glib) Valgrind might also help, a bit of a long shot though. Ian. Aptitude 0.6.3: log report Mon, Nov 15 2010 19:39:16 + IMPORTANT: this log only lists intended actions; actions which fail due to dpkg problems may not be completed. Will install 81 packages, and remove 9 packages. 28.6 MB of disk space will be freed === [REMOVE, NOT USED] libqt4-assistant [REMOVE, NOT USED] libqt4-help [REMOVE, NOT USED] libqt4-scripttools [REMOVE, NOT USED] libqt4-test [REMOVE, NOT USED] python-qt4 [REMOVE, NOT USED] python-qt4-dbus [REMOVE, NOT USED] python-reportlab [REMOVE, NOT USED] python-reportlab-accel [REMOVE, NOT USED] python-sip [HOLD] gthumb [HOLD] gthumb-data [HOLD] lvm2 [UPGRADE] acpid 1:2.0.6-1 - 1:2.0.6-2 [UPGRADE] base-files 5.9 - 5.10 [UPGRADE] bcc 0.16.17-3 - 0.16.17-3.1 [UPGRADE] bin86 0.16.17-3 - 0.16.17-3.1 [UPGRADE] bind9-doc 1:9.7.2.dfsg.P2-1 - 1:9.7.2.dfsg.P2-2 [UPGRADE] bind9-host 1:9.7.2.dfsg.P2-1 - 1:9.7.2.dfsg.P2-2 [UPGRADE] bsdmainutils 8.0.15 - 8.0.16 [UPGRADE] cdebconf 0.150 - 0.151 [UPGRADE] cpp-4.4 4.4.5-6 - 4.4.5-7 [UPGRADE] cups 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-bsd 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-client 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-common 1.4.4-6 - 1.4.4-7 [UPGRADE] cups-ppdc 1.4.4-6 - 1.4.4-7 [UPGRADE] debian-cd 3.1.3 - 3.1.4 [UPGRADE] debianutils 3.4.1 - 3.4.2 [UPGRADE] debootstrap 1.0.25 - 1.0.26 [UPGRADE] dnsutils 1:9.7.2.dfsg.P2-1 - 1:9.7.2.dfsg.P2-2 [UPGRADE] eatmydata 15-3 - 15-4 [UPGRADE] elks-libc 0.16.17-3 - 0.16.17-3.1 [UPGRADE] evolution 2.30.3-3 - 2.30.3-4 [UPGRADE] evolution-common 2.30.3-3 - 2.30.3-4 [UPGRADE] evolution-plugins 2.30.3-3 - 2.30.3-4 [UPGRADE] foomatic-filters 4.0.5-5 - 4.0.5-6 [UPGRADE] g++-4.4 4.4.5-6 - 4.4.5-7 [UPGRADE] gcc-4.4 4.4.5-6 - 4.4.5-7 [UPGRADE] gcc-4.4-base 4.4.5-6 - 4.4.5-7 [UPGRADE] gdm 2.20.11-2 - 2.20.11-4 [UPGRADE] gnome-about 2.30.2-1 - 2.30.2-2 [UPGRADE] gnome-desktop-data 2.30.2-1 - 2.30.2-2 [UPGRADE] gnome-nettool 2.30.0-2 - 2.30.0-3 [UPGRADE] kernel-wedge 2.70 - 2.71 [UPGRADE] lib32gcc1 1:4.4.5-6 - 1:4.4.5-7 [UPGRADE] lib32stdc++6 4.4.5-6 - 4.4.5-7 [UPGRADE] libaudio-dev 1.9.2-3 - 1.9.2-4 [UPGRADE] libaudio2 1.9.2-3 - 1.9.2-4 [UPGRADE] libbind9-60 1:9.7.2.dfsg.P2-1 -
Bug#597962: Seems to have resolved itself
I was going to take a shot at debugging this today, since it's on the RC bug list for squeeze, but it seems to have resolved itself at some point, at least in unstable. QL starts without errors, and it's playing music right now. The last time I tried it was weeks ago, so unfortunately I don't know which of the many package updates between then and now was responsible for the change. I also don't have a system running squeeze, so I can't test whether it works there. But it's working for me in sid. -- Mike Paul m...@wyzardry.net -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
