Crash on Windows when importing dives via USB from OSTC2

[Anton Lundin]

On 19 April, 2017 - Stefan Fuchs wrote:

> Hi All,
> 
> unfortunately I see a new crash when importing via USB from my OSTC2.
> This happens under Windows 10 with current master and was not there
> about 2 weeks ago. It happens every time at the end of data transfer
> from the dive computer.
> 
> Stack trace attached. I already started to think about one of the latest
> commits being the root cause but from the stack trace I can't see a
> direct link to one of the changes I'm aware off.
> 

<snip/>

> 
> LEAK: 77 RenderObject
> LEAK: 1 Page
> LEAK: 1 Frame
> LEAK: 21 CachedResource
> LEAK: 169 WebCoreNode
> QWaitCondition: Destroyed while threads are still waiting
> Registers:
> eax=00000000 ebx=77a2f9a0 ecx=00000004 edx=00000000 esi=00000003 edi=00000000
> eip=7798e89c esp=03c78d44 ebp=03c78e1c iopl=0         nv up ei pl nz na po nc
> cs=0023  ss=002b  ds=002b  es=002b  fs=0053  gs=002b             efl=00000206
> 
> AddrPC   Params
> 7798E89C FFFFFFFF 00000003 00000004  ntdll.dll!_NtTerminateProcess at 8
> 7794A41E 00000003 77E8F3B0 FFFFFFFF  ntdll.dll!RtlExitUserProcess
> 7447ADC2 00000003 03C78E84 76D66CD7  KERNEL32.DLL!_ExitProcessImplementation at 4
> 76D66738 00000003 8BF28E59 00000004  msvcrt.dll!___crtExitProcess
> 76D66CD6 00000003 00000001 00000000  msvcrt.dll!doexit
> 76D666B1 00000003 00010001 00000000  msvcrt.dll!__exit
> 76D5BCC9 03C79560 00000004 03C79258  msvcrt.dll!_abort
> 03FF3AE2 00000003 03C7928C 03C79248  Qt5Cored.dll!qt_message_fatal  [/home/stefan/Entwicklung/Subsurface/mxe-2016-10-12/tmp-qtbase-i686-w64-mingw32.shared/qtbase-opensource-src-5.7.0/src/corelib/global/qlogging.cpp @ 1680]
> 03FF09B1 03C7928C 04330284 043E8E47  Qt5Cored.dll!fatal  [/home/stefan/Entwicklung/Subsurface/mxe-2016-10-12/tmp-qtbase-i686-w64-mingw32.shared/qtbase-opensource-src-5.7.0/src/corelib/global/qlogging.cpp @ 793]
> 03FEC8FA 043E8E47 043E8B60 00000A87  Qt5Cored.dll!qt_assert  [/home/stefan/Entwicklung/Subsurface/mxe-2016-10-12/tmp-qtbase-i686-w64-mingw32.shared/qtbase-opensource-src-5.7.0/src/corelib/global/qglobal.cpp @ 3063]
> 041434C1 03C79330 00000000 FFFFFFFF  Qt5Cored.dll!beginRemoveRows  [/home/stefan/Entwicklung/Subsurface/mxe-2016-10-12/tmp-qtbase-i686-w64-mingw32.shared/qtbase-opensource-src-5.7.0/src/corelib/itemmodels/qabstractitemmodel.cpp @ 2695]

> 004A215F 00000000 0000001F 00000000  subsurface.exe!setImportedDivesIndexes  [/home/stefan/Entwicklung/Subsurface/subsurface/desktop-widgets/downloadfromdivecomputer.cpp @ 736]

This looks to be curlprint.

> 004A06F5 03C7C164 03C794EC 03C79408  subsurface.exe!onDownloadThreadFinished  [/home/stefan/Entwicklung/Subsurface/subsurface/desktop-widgets/downloadfromdivecomputer.cpp @ 456]
> 00481A63 03C7C164 00000000 00000006  subsurface.exe!qt_static_metacall  [/home/stefan/Entwicklung/Subsurface/win32/subsurface/desktop-widgets/moc_downloadfromdivecomputer.cpp @ 298]
> 041A0F26 03C7C164 5E0508E0 00000004  Qt5Cored.dll!placeMetaCall  [/home/stefan/Entwicklung/Subsurface/mxe-2016-10-12/tmp-qtbase-i686-w64-mingw32.shared/qtbase-opensource-src-5.7.0/src/corelib/kernel/qobject.cpp @ 502]

Because we're in onDownloadThreadFinished, I'd guess its not
libdivecomputers fault. Maybe libdivecomputer generates some data which
we can't handle, but thats another thing.


There's also both a patch release and a major release of qt after qt 5.7.0.

Might be worth trying to to one of those.


//Anton


-- 
Anton Lundin	+46702-161604