| Summary: | KDE Partition Manager crashes while starting up | ||
|---|---|---|---|
| Product: | [Applications] partitionmanager | Reporter: | Clinton <clintonthegeek> |
| Component: | general | Assignee: | Andrius Štikonas <andrius> |
| Status: | RESOLVED NOT A BUG | ||
| Severity: | crash | CC: | benklop, emre, vishalrao |
| Priority: | NOR | Keywords: | drkonqi |
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Platform: | Ubuntu | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
| Attachments: |
output of lsblk command
fdisk -l output Additional backtrace output |
||
|
Description
Clinton
2016-10-10 00:32:06 UTC
I can't reproduce this on Neon User Edition running on VirtualBox. Can you provide any more info? What devices or partitions are there? Created attachment 101641 [details]
output of lsblk command
I faced the same issue. I have Neon 5.8 dev unstable edition, same backtrace as original reporter's. I have a SATA SSD, an NVMe m.2 SSD, an HDD, an external USB HDD, and a USB pen drive plugged in, was trying to format the pen drive when came across this issue. Attaching output of my lsblk command, let me know if anything further required. Correction to my previous comment "was trying to format the pen drive when came across this issue" - I didn't write correct - I meant to say I also faced this issue during startup of partition manager, when it scans the devices, it lists my devices and not sure if this is important but when it reaches my /dev/nvme0n1 is when the crash happens. Any system info I can provide? What commands to run and post output that would help? Output of running "sudo fdisk -l" minus the snipped sections of /dev/ramX items: https://paste.kde.org/p1fbzr2po (In reply to Vishal Rao from comment #4) > Correction to my previous comment "was trying to format the pen drive when > came across this issue" - I didn't write correct - I meant to say I also > faced this issue during startup of partition manager, when it scans the > devices, it lists my devices and not sure if this is important but when it > reaches my /dev/nvme0n1 is when the crash happens. > > Any system info I can provide? What commands to run and post output that > would help? Ok, knowing it might crash on nvme might be useful. I'll try to think how I can reproduce it, need to find a way to add nvme device to virtual machine since I don't have any nvme devices. But I won't be able to find time for that till weekend at the very least... Created attachment 101708 [details]
fdisk -l output
I've also encountered this crash, but I've got only 2 disks installed, and one is an SD card (on which i'm running Neon). I don't have any nvme devices - only an m.2 drive with Windows and associated default OS partitions on it (sda), and the SD card with a generic Neon install (sdb).
attached fdisk -l output
Created attachment 101709 [details]
Additional backtrace output
My backtrace is slightly different, if only because I have more debugging symbols installed.
Additionally, running at the command line suggests this may be due to missing SMART support: benklop@Thinkpad:~$ sudo partitionmanager QStandardPaths: XDG_RUNTIME_DIR not set, defaulting to '/tmp/runtime-root' Loaded backend plugin: "pmlibpartedbackendplugin" QLayout: Attempting to add QLayout "" to PartitionManagerWidget "m_PartitionManagerWidget", which already has a layout QLayout: Attempting to add QLayout "" to ListDevices "m_ListDevices", which already has a layout QLayout: Attempting to add QLayout "" to QWidget "m_ListOperations", which already has a layout QLayout: Attempting to add QLayout "" to QWidget "m_InfoPane", which already has a layout QLayout: Attempting to add QLayout "" to QWidget "m_TreeLog", which already has a layout "Using backend plugin: pmlibpartedbackendplugin (1)" "Scanning devices..." QDBusArgument: read from a write-only object QDBusArgument: read from a write-only object QDBusArgument: read from a write-only object "Device found: ATA TOSHIBA THNSFJ25" QFileInfo::absolutePath: Constructed with empty filename blkid: unknown file system type "" on "/dev/sda3" QFileInfo::absolutePath: Constructed with empty filename QFileInfo::absolutePath: Constructed with empty filename QFileInfo::absolutePath: Constructed with empty filename "Device found: Generic STORAGE DEVICE" getting smart status failed for "/dev/sdb" : Operation not supported QFileInfo::absolutePath: Constructed with empty filename KCrash: crashing... crashRecursionCounter = 2 KCrash: Application Name = partitionmanager path = /usr/bin pid = 6138 KCrash: Arguments: /usr/bin/partitionmanager KCrash: Attempting to start /usr/lib/x86_64-linux-gnu/libexec/drkonqi from kdeinit sock_file=/tmp/runtime-root/kdeinit5__0 [1]+ Stopped sudo partitionmanager benklop@Thinkpad:~$ QSocketNotifier: Invalid socket 7 and type 'Read', disabling... QSocketNotifier: Invalid socket 12 and type 'Read', disabling... QSocketNotifier: Invalid socket 8 and type 'Read', disabling... QSocketNotifier: Invalid socket 18 and type 'Read', disabling... Hi Ben, Isn't m.2 drive nvme? I thought that's what nvme is... Backtrace points to line QListWidgetItem(QIcon::fromTheme(d.iconName()).pixmap(IconSize(KIconLoader::Desktop)), d.prettyName()) I wonder what might be going wrong here... icon Name should default to hard drive icon if icon is missing...maybe it's pretty name... In the original reporter's (and mine) stacktrace it seems to indicate operator= for QString when it's trying to read/set Device::deviceNode() so something about scanning an m.2 device (mine is NVMe) while Ben's could be m.2 "SATA" based. Anything we (or I could do) to run/capture more debugging info? While trying to install source and building/debugging (not much success here, newbie) I noticed my stacktrace now has Device::setIconName() call as follows. doing a quick grep, it looks like only the dummy backend is calling setIconName in its scanDevice() method, unless, of course, I'm making some mistake: Thread 1 (Thread 0x7fb4f85eb940 (LWP 5843)): [KCrash Handler] #4 0x00007fb4f4d5e37c in QString::operator=(QString const&) () from /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #5 0x00007fb4f80fc337 in Device::setIconName (this=0x7fb4d4060950, name=...) at /home/vishal/SRC/kpmcore-2.2.1+p16.04+git20161011.0712/src/core/device.h:92 #6 0x000000000044e8a9 in ListDeviceWidgetItem::ListDeviceWidgetItem (d=..., this=0xab6cc0) at /workspace/build/src/gui/listdevices.cpp:32 #7 ListDevices::updateDevices (this=0x7eef20, devices=...) at /workspace/build/src/gui/listdevices.cpp:57 #8 0x000000000045294a in MainWindow::on_m_OperationStack_devicesChanged (this=0x7d5820) at /workspace/build/src/gui/mainwindow.cpp:623 #9 0x0000000000484885 in MainWindow::qt_metacall (this=0x7d5820, _c=QMetaObject::InvokeMetaMethod, _id=7, _a=0x7fb4d402c2c0) at /workspace/build/obj-x86_64-linux-gnu/src/moc_mainwindow.cpp:197 #10 0x00007fb4f4edba19 in QObject::event(QEvent*) () from /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 I should mention if I run my local build of partitionmanager/kpmcore it does not crash, not sure if it's because I made a debug build. But when I run the one that came installed in /usr/bin I get the previously posted stacktrace crash during startup. (In reply to Vishal Rao from comment #13) > I should mention if I run my local build of partitionmanager/kpmcore it does > not crash, not sure if it's because I made a debug build. But when I run the > one that came installed in /usr/bin I get the previously posted stacktrace > crash during startup. Thanks for slightly more detailed debug output. Let me look at the code and think what might be going wrong with the icon. But I am not sure if we will be able to test it if it doesn't crash with your local build. (In reply to Vishal Rao from comment #13) > I should mention if I run my local build of partitionmanager/kpmcore it does > not crash, not sure if it's because I made a debug build. But when I run the > one that came installed in /usr/bin I get the previously posted stacktrace > crash during startup. Could it be you didn't build dummy backend in your local build? I have built dummy backend also as part of kpmcore and noted both the default partitionmanager binary and my built one both try to load the backends from /usr/lib/x86.../qt5/plugins folder, so I tried replacing the original with my built ones and only the default partitionmanager binary from /usr/bin crashed and not my built one. One other difference I see is that when I run the default /usr/bin/partitionmanager it seems to trigger KDE-SU UI while my binary triggers/picks up gksudo... see different console output below during early run: vishal@skybird:/usr/bin$ ./partitionmanager org.kde.kdesu: Daemon not safe (not sgid), not using it. -------- vishal@skybird:~/SRC/partitionmanager-2.2.1+p16.04+git20160920.0957/build/src$ ./partitionmanager (gksudo:2327): Gtk-WARNING **: Unable to locate theme engine in module_path: "adwaita", (gksudo:2327): Gtk-WARNING **: Unable to locate theme engine in module_path: "adwaita", Just an update that while the issue occurs on my updated Neon dev unstable edition, it does not occur on a freshly installed neon user edition 5.8.2 at this time. I can reproduce this easily on KDE neon dev. ed. -stable and can get logs if needed. forvelin@stormgarden:~$ sudo partitionmanager QStandardPaths: XDG_RUNTIME_DIR not set, defaulting to '/tmp/runtime-root' Loaded backend plugin: "pmlibpartedbackendplugin" QLayout: Attempting to add QLayout "" to PartitionManagerWidget "m_PartitionManagerWidget", which already has a layout QLayout: Attempting to add QLayout "" to ListDevices "m_ListDevices", which already has a layout QLayout: Attempting to add QLayout "" to QWidget "m_ListOperations", which already has a layout QLayout: Attempting to add QLayout "" to QWidget "m_InfoPane", which already has a layout QLayout: Attempting to add QLayout "" to QWidget "m_TreeLog", which already has a layout "Using backend plugin: pmlibpartedbackendplugin (1)" "Scanning devices..." QDBusArgument: read from a write-only object QDBusArgument: read from a write-only object QDBusArgument: read from a write-only object "Device found: ATA INTEL SSDSC2BW18" getting temp failed for "/dev/sda" : No such file or directory blkid: unknown file system type "" on "/dev/sda2" QFileInfo::absolutePath: Constructed with empty filename "Partition ‘/dev/sda2’ is not properly aligned (first sector: 1001470, modulo: 2046)." KCrash: crashing... crashRecursionCounter = 2 KCrash: Application Name = partitionmanager path = /usr/bin pid = 4349 KCrash: Arguments: /usr/bin/partitionmanager KCrash: Attempting to start /usr/lib/x86_64-linux-gnu/libexec/drkonqi from kdeinit sock_file=/tmp/runtime-root/kdeinit5__0 (In reply to Emre Unlu from comment #18) > I can reproduce this easily on KDE neon dev. ed. -stable and can get logs if > needed. dev stable edition has the same package as dev unstable (where kpmcore is newer than partitionmanager). But I think today there was a new build of partitionmanager, so it would be interesting to see if anything changes once it gets into neon repos. Unfortunately console output is not really useful... There isn't anything that could help to find this crash. Ok, I can see crash in neon myself now. Indeed as Vishal wrote earlier, crash is gone after rebuilding exactly the same source code. So I'm almost sure this bug is just due to ABI break in kpmcore and partition manager was not recompiled in neon dev edition... If that's not the case please reopen this bug once newer package reaches neon (I'm not sure when it will be though). Just a small update: Looks like partitionmanager packaging repository disappeared, that's there were no newer snapshots of partitionmanager for a month... The reason for disappearance seems to be that debian removed their partitionmanager repo. You can see in https://packages.qa.debian.org/p/partitionmanager.html that partitionmanager was removed from testing... Just FYI, got this version: 2.2.1+p16.04+git20161026.1000-0 in Neon dev unstable update just now and it seems to be working okay. |