QGIS » QGIS Application

N. L. wrote:

The problem seems to be related to the project file.

what does contain your project (datasources, layouts, etc.?)?

I rebuilt the project step by step and the bug was probably due to oversized csv attribute tables anchored in the project. If I leave them out now, Qgis seems to run stable.

N. L. wrote:

I rebuilt the project step by step and the bug was probably due to oversized csv attribute tables anchored in the project. If I leave them out now, Qgis seems to run stable.

tell us more: the csv was joined to what other datasource? how many columns has rhe CSV that causes the crash when joined? Does it crash if you add the CSV without joining it?

I tried to link the CSV to a shape file. The CSV file has more than 36000 lines and 226 MB.
The crash happens without making the layer with the linked CSV file visible. The loading of the project cannot be completed. In version 3.3 this error did not appear yet.

Seems related to #20412 - same backtrace. Probably there is a bug in the new code for handling of native events on Windows to catch USB drive insertion/removal.

I get a similar, but not entirely the same stack trace:

Crash ID: 6138756c706cb33aebe0925fbe4e95b5b12f5787

Stack Trace

QgsWinNativeEventFilter::nativeEventFilter qgswinnative.cpp:182
QAbstractEventDispatcher::filterNativeEvent :
CallWindowProcW :
CallWindowProcW :
GetLastInputInfo :
KiUserCallbackDispatcher :
NtUserPeekMessage :
PeekMessageW :
QEventDispatcherWin32::processEvents :
qt_plugin_query_metadata :
QEventLoop::exec :
QCoreApplication::exec :
main main.cpp:1505
WinMain mainwin.cpp:170
__scrt_common_main_seh exe_common.inl:253
BaseThreadInitThunk :
RtlUserThreadStart :

QGIS Info
QGIS Version: 3.4.1-Madeira
QGIS code revision: cdf878d4b5
Compiled against Qt: 5.11.2
Running against Qt: 5.11.2
Compiled against GDAL: 2.3.2
Running against GDAL: 2.3.2

System Info
CPU Type: x86_64
Kernel Type: winnt
Kernel Version: 10.0.14393

The line it points to is

182: unsigned long deviceType = reinterpret_cast<DEV_BROADCAST_HDR *>( lParam )->dbch_devicetype;
183: if ( deviceType == DBT_DEVTYP_VOLUME )

I had an outlandish theory that because I often get messages relating to Audio in DebugView around the crash,

such as

[10772] CxHDAudioDeviceControl::SetLEDStatusToDriver() + isPlaybackEndpoint = 1, ledStatus = 1

that it was something bizarre relating to an error somewhere caused by the difference between the concept of volume as a disk volume, as is meant in the QGIS message reading I think, and Sound Volume, but that seems somewhat far fetched.

The other thing that jumps out is
[9936] shell\lock\abovelock\shellcomponent\lockscreenapppositioner.cpp(256)\AboveLockAppHost.dll!00007FFDFC96323A: (caller: 00007FFE2D4784A3) ReturnHr(75) tid(232c) 8000FFFF Catastrophic failure

but it's hard to figure out exactly where things are going wrong from the traces.

Stephen - are you using a custom build? I'd love to put some debugging code in here, but unfortunately(?) I haven't been able to trigger this crash myself.

I think I can now reproduce it. I have a machine with a Thunderbolt connector and when that is plugged in or out to connect external monitors, mice, keyboard and network connections I get a crash. Happy to start a new bug thread in case these are not related, but I strongly suspect they are.

This is the DebugView output immediately preceeding the crash:

[9936] avcore\npsm\localprovider\baseprovider\lib\baseprovider.cpp(604)\NPSMDesktopProvider.dll!00007FFDDD972140: (caller: 00007FFDDD9726E5) ReturnHr(2607) tid(232c) 80070490 Element not found.
[33204] ..\..\..\src\gui\qgsmapcanvas.cpp(1344) : (QgsMapCanvas::keyPressEvent) [14770ms] Ignoring key: 16777251
[9936] avcore\npsm\localprovider\baseprovider\lib\baseprovider.cpp(604)\NPSMDesktopProvider.dll!00007FFDDD972140: (caller: 00007FFDDD9726E5) ReturnHr(2608) tid(232c) 80070490 Element not found.
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4458) tid(3624) 80070057 The parameter is incorrect.
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4459) tid(5418) 80070057 The parameter is incorrect.
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4460) tid(54dc) 80070057 The parameter is incorrect.
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4461) tid(7ff8) 80070057 The parameter is incorrect.
[10772] *******************************************************************************
[10772] Types supported for 22
[10772] *******************************************************************************
[10772] Types supported for 25
[10772] DeviceIoControl failed with error 87: The parameter is incorrect.
[10772] DeviceIoControl failed with error 87: The parameter is incorrect.
[9936] onecoreuap\base\diagnosis\platform\notifications\platform\database\databasehelpers.cpp(143)\wpncore.dll!00007FFE100D1459: (caller: 00007FFE100D21AF) ReturnHr(274) tid(6448) 80070490 Element not found.
[10772] CCxHDAudioHDMI::get_MasterVolume
[10772] CCxHDAudioHDMI::get_MasterVolume: Created ()
[10772] CCxHDAudioHDMI::get_MasterVolume(): Before SetEndPointInfo()
[10772] CCxHDAudioHDMI::get_MasterVolume(): After SetEndPointInfo()
[10772] CCxHDAudioHDMI::get_MasterVolume(): Before QI<ICxHDMasterVolumeControl>
[10772] CCxHDAudioHDMI::get_MasterVolume(): After QI<ICxHDMasterVolumeControl>
[10772] CCHDAudioSPDIF::Final QI<ICxHDMasterVolumeControl>
[10772] CCxHDAudioHDMI::get_MasterVolume
[10772] CCxHDAudioHDMI::get_MasterVolume: Created ()
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4462) tid(5418) 80070057 The parameter is incorrect.
[10772] CCxHDAudioHDMI::get_MasterVolume(): Before SetEndPointInfo()
[10772] CCxHDAudioHDMI::get_MasterVolume(): After SetEndPointInfo()
[10772] CCxHDAudioHDMI::get_MasterVolume(): Before QI<ICxHDMasterVolumeControl>
[10772] CCxHDAudioHDMI::get_MasterVolume(): After QI<ICxHDMasterVolumeControl>
[10772] CCHDAudioSPDIF::Final QI<ICxHDMasterVolumeControl>
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4463) tid(7ff8) 80070057 The parameter is incorrect.
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4464) tid(54dc) 80070057 The parameter is incorrect.
[9936] shell\lib\bindctx.cpp(128)\explorerframe.dll!00007FFDFC4A4280: (caller: 00007FFDFC4CA68B) ReturnHr(4465) tid(3624) 80070057 The parameter is incorrect.
[9936] avcore\npsm\localprovider\baseprovider\lib\baseprovider.cpp(604)\NPSMDesktopProvider.dll!00007FFDDD972140: (caller: 00007FFDDD9726E5) ReturnHr(2609) tid(232c) 80070490 Element not found.
[33204] ..\..\..\src\app\qgscrashhandler.cpp(34) : (QgsCrashHandler::handle) [6696ms] CRASH!!!

P.S., I am using qgis-rel-dev-g7.4.2.bat from OSGeo4W (i.e with debugging symbols) as I don't have a build system on Windows.

My apologies, it is not a Lightning connector, it is a Thunderbolt connector.

I have a Thunderbolt USB type-C port on my laptop.
I just tried an ejection of a USB drive connected to this port and QGIS crashed right away.
(same crash ID: Crash ID: 46bdfc53c9e65ba01bf6e65f60a8fdcad5a2dd77).
Same behaviour when I connect a USB drive.

Interestingly, crash may happen when nothing particular is done.
(QGIS crashed with this ID while I was away from computer..)
I assume that that there is some Windows 10 check in background.. (?)

Nyall Dawson wrote:

Please test with the next round of QGIS 3.6 nightlies (not 3.4 - it hasn't been backported yet). I've added some extra safety checks which (fingers crossed!) might prevent this.

Did you mean 3.5? I tried installing using OSGeo4W, but it only lists 3.5 as far as I can see.

Did you mean 3.5?

yes

I installed the dev version (it says QGIS eab40d0e4c in the title bar), still crashes in idle phase for me. I'm still able to force a crash by connecting to network shares. Like Stephen I'm also using a docking station with thunderbolt, I don't get any crashes however when I disconnect/connect the docking station (running QGIS 3.4.x).

That build is too old -- it's missing the latest related fix

Yeah, think there is a build problem with the nightlies ATM. So will have to wait to fully test.

I installed the dev version again using OSGeo4W, I now get revision eab40d0e4c (don't know if this is the latest one?). Unfortunately still crashes for me.

Markus - wrote:

I installed the dev version again using OSGeo4W, I now get revision eab40d0e4c (don't know if this is the latest one?). Unfortunately still crashes for me.

is still the same as #20386-26

Giovanni Manghi wrote:

Markus - wrote:

I installed the dev version again using OSGeo4W, I now get revision eab40d0e4c (don't know if this is the latest one?). Unfortunately still crashes for me.

is still the same as #20386-26

Ah yes, sorry about that.

Looks promising so far with d4cf6a - no crash when removing the thunderbolt cable, and no crash in idle phase yet

I'm running revision 7cce9b3eb9 and have not been able to force a crash by adding/removing network shares. No crash in idle yet either, though I've just been running it for an hour, but looks promising. Will this be backported to 3.4.2, 3.4.3 or 3.4.4? Thanks for the fast handling of the issue, great work.

Markus - wrote:

I'm running revision 7cce9b3eb9 and have not been able to force a crash by adding/removing network shares. No crash in idle yet either, though I've just been running it for an hour, but looks promising. Will this be backported to 3.4.2, 3.4.3 or 3.4.4? Thanks for the fast handling of the issue, great work.

It is backported to the release-3_4 branch and will appear in the 3.4.2 point release.