Summary: | Plasmashell crashed after starting gparted | ||
---|---|---|---|
Product: | [Frameworks and Libraries] frameworks-solid | Reporter: | Simon <simon.vizzini83> |
Component: | general | Assignee: | Lukáš Tinkl <lukas> |
Status: | RESOLVED DUPLICATE | ||
Severity: | crash | CC: | admin, admin, ali.sherif10, anselmolsm, bhush94, burden.paul, bwucke+kdebugs, cdanilo25, com.tihon, cordlandwehr, daviduser, dinimito2, dritosweb, hlovdal, jejones3141, joe.m.ferner, jwiels, kde4, kde, kde, kde, kdelibs-bugs, mhallika, michael.messaggi, NearExternal, pedron.alexandre, petrov.rainkimry, plasma-bugs, public, putz.alexey, rainer+kdebugs, rajeeshknambiar, simonandric5, thomas.pasch |
Priority: | NOR | Keywords: | drkonqi |
Version: | unspecified | ||
Target Milestone: | --- | ||
Platform: | Ubuntu | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: | ||
Attachments: | dbus-monitor --system log |
Description
Simon
2015-04-23 20:11:46 UTC
*** Bug 346526 has been marked as a duplicate of this bug. *** Solid people, can you look at this. *** Bug 346818 has been marked as a duplicate of this bug. *** *** Bug 345145 has been marked as a duplicate of this bug. *** *** Bug 347641 has been marked as a duplicate of this bug. *** *** Bug 347754 has been marked as a duplicate of this bug. *** *** Bug 350301 has been marked as a duplicate of this bug. *** Solid::Backends::UDisks2::DeviceBackend::checkCache() does m_device->property(key.toUtf8()) at the crash site, where m_device is QDBusInterface and QObject::property(const char *) takes parameter char* instead of QByteArray - so, there seems to be a type mismatch. But, shouldn't this crash happen everytime in that case? Also - can we convert the property name to char* as the property in question is "MountPoint" which I imagine can contain Unicode characters? *** Bug 339964 has been marked as a duplicate of this bug. *** OK, so testing the interface qdbus --literal --system org.freedesktop.UDisks2 /org/freedesktop/UDisks2/block_devices/sda1 org.freedesktop.DBus.Properties.Get org.freedesktop.UDisks2.Filesystem MountPoints [Variant: [Argument: aay {{47, 98, 111, 111, 116, 0}}]] (that's "/boot" in ascii, null terminated) org.freedesktop.DBus.Properties.Get should always always return a variant (see) http://dbus.freedesktop.org/doc/dbus-specification.html which we're getting here, and that variant contains the an object of type "aay" . All good there. From the error we can tell that in qdbusmetaobject.cpp we're going into this path in QDBusMetaObjectGenerator::findType if (type == QVariant::Invalid || signature != QDBusMetaType::typeToSignature(type)) { // type is still unknown or doesn't match back to the signature that it // was expected to, so synthesize a fake type typeName = "QDBusRawType<0x" + signature.toHex() + ">*"; type = registerComplexDBusType(typeName); } Given udisks is already giving us a string in the most ridiculous way possible, it seems plausible that in some cases udisks is returning a property in the wrong format from the one they advertise in their interface. Tried looking at udisks code, but I can't read this glib stuff. Can someone reproduce this whilst running dbus-monitor --system ? Created attachment 93933 [details]
dbus-monitor --system log
As requested. It took me a few tries for it to occur, but eventually it happened, near the end of the log file. I had Firefox and Dolphin open at the time; I did access the volumes on the modem/pendrive before removing it ("unsafely"), then I reinserted it and Plasma crashed.
Thank you very much *** Bug 351762 has been marked as a duplicate of this bug. *** *** This bug has been marked as a duplicate of bug 346153 *** @David : can you add comments in duplicate reports? *** Bug 353156 has been marked as a duplicate of this bug. *** *** Bug 353981 has been marked as a duplicate of this bug. *** *** Bug 354554 has been marked as a duplicate of this bug. *** *** Bug 355033 has been marked as a duplicate of this bug. *** *** Bug 354998 has been marked as a duplicate of this bug. *** *** Bug 355476 has been marked as a duplicate of this bug. *** *** Bug 356158 has been marked as a duplicate of this bug. *** *** Bug 356148 has been marked as a duplicate of this bug. *** log above didn't have what I needed. Can someone give me a reliable way to 100% reproduce this. Ideally with docker or some such. (In reply to David Edmundson from comment #25) > log above didn't have what I needed. > > Can someone give me a reliable way to 100% reproduce this. Ideally with > docker or some such. https://bugs.kde.org/355476 has a bt generated when plasmashell crashed after starting docker. Hope it helps somehow. I'm not after a backtrace, I want guided steps on making my system crash. 355476 was generated after a systemctl start docker.service *** Bug 356410 has been marked as a duplicate of this bug. *** version bug#347641 - reproducible about 30% of the times. Steps to reproduce: insert modem, wait some 30s. If SD card reader and emulated DVD-ROM appear, the crash didn't occur, repeat. It rarely takes more than three tries, and using the modem on the move, that's more than enough to make it very cumbersome... especially that after upgrade to Xenial, plasmashell often refuses to come back up after the crash, necessitating reboot. It might be related to devices appearing, vanishing and reappearing in a very short sequence. A recent update broke the automatic usb_modeswitch call from udev that had been switching the modem part on in my modem. Now I'm calling usb_modeswitch manually from shell whenever I plug the modem in, and the crash has not occurred even once since then. Normally, udev calls the modeswitch as soon as the device makes itself known. usb_modeswitch doesn't just enable the modem - it makes whole device with all its parts (modem, DVD, reader) reappear with a new product id. Manually, I can't call usb_modeswitch so fast after the device is recognized by the system without the modem. *** Bug 359875 has been marked as a duplicate of this bug. *** *** Bug 361536 has been marked as a duplicate of this bug. *** *** Bug 361818 has been marked as a duplicate of this bug. *** *** Bug 365853 has been marked as a duplicate of this bug. *** *** Bug 368112 has been marked as a duplicate of this bug. *** *** Bug 368806 has been marked as a duplicate of this bug. *** *** Bug 370492 has been marked as a duplicate of this bug. *** *** Bug 371785 has been marked as a duplicate of this bug. *** *** Bug 372967 has been marked as a duplicate of this bug. *** *** Bug 373518 has been marked as a duplicate of this bug. *** *** Bug 374566 has been marked as a duplicate of this bug. *** *** Bug 374793 has been marked as a duplicate of this bug. *** *** Bug 375814 has been marked as a duplicate of this bug. *** *** Bug 377783 has been marked as a duplicate of this bug. *** |