Switching to a broken skin can crash Mixxx

[ronso0]

Bug Description

I just tried to debug the VU meter issue with Traktmixxx skin reported here

This makes Mixxx crash:

• copy Traktmixxx to ressource dir (here /usr/share/mixxx/skins/), actually not recommended
• start Mixxx with a new config dir
• boots with LateNight
• open preferences, select Traktmixxx
• skin seems to load, but crashes immediatley after

---

This way, however, I can start with Traktmixxx and it seems to work (at least not crash after some basic testing, track load, play etc):

• copy Traktmixxx to ressource dir (here /usr/share/mixxx/skins/), actually not recommended
• start Mixxx with a new config dir
• boots with LateNight
• switch skin or color theme so that ResizableSkin [skin] is written to mixxx.cfg
• close
• open mixxx.cfg, set ResizableSkin Traktmixxx
• start Mixxx
• works

BUT

• start with Traktmixxx as described
• switch to LateNight
• switch back to Traktmixxx
• CRASH

---

Backtrace:
backtrace-LateNight-to-Traktmixxx.txt
btw, since when are gdb backtraces this huge? 13869 lines???

Version

2.3.4

OS

Ubuntu Studio 20.04 / X11 / xfce 4.14 / Qt 5.12.8

[ronso0]

maybe the cause is the same that led to this segfault?
#10893 (comment)

[ronso0]

starting with --disableVuMeterGL doesn't avoid the crash btw.

[ronso0]

I updated the description:
starting with Traktmixxx by writing it to mixxx.cfg manually works, but any following switch (to another skin and) back to Traktmixxx crashes Mixxx.

@m0dB Do you have any idea? Can you read anything from that backtrace?

[daschuer]

I can see the same segfault, original reported here:
#10893 (comment)
And analyzed here:
#10893 (comment)
and workaround here:
m0dB@c9af1e0

#0  0x0000000000000000 in  ()
#1  0x00007fffeb04f7e2 in QXcbWindow::handleNativeEvent(xcb_generic_event_t*) (this=<optimized out>, event=0x7fffe40e43c0) at qxcbwindow.cpp:1696
#2  0x00007fffeb03d404 in QXcbConnection::handleXcbEvent(xcb_generic_event_t*) (this=this@entry=0x5555563ce3c0, event=event@entry=0x7fffe40e43c0) at qxcbconnection.cpp:562
#3  0x00007fffeb03d78a in QXcbConnection::processXcbEvents(QFlags<QEventLoop::ProcessEventsFlag>) (this=0x5555563ce3c0, flags=flags@entry=...) at qxcbconnection.cpp:1027
#4  0x00007fffeb068327 in xcbSourceDispatch(GSource*, GSourceFunc, gpointer) (source=<optimized out>) at qxcbeventdispatcher.cpp:103
#5  0x00007ffff28dd17d in g_main_context_dispatch () at /lib/x86_64-linux-gnu/libglib-2.0.so.0
#6  0x00007ffff28dd400 in  () at /lib/x86_64-linux-gnu/libglib-2.0.so.0
#7  0x00007ffff28dd4a3 in g_main_context_iteration () at /lib/x86_64-linux-gnu/libglib-2.0.so.0
#8  0x00007ffff3ceb435 in QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) (this=0x555556480b20, flags=...) at kernel/qeventdispatcher_glib.cpp:422
#9  0x00007ffff3c923ab in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) (this=this@entry=0x7fffffffdce0, flags=..., flags@entry=...) at ../../include/QtCore/../../src/corelib/global/qflags.h:140
#10 0x00007ffff3c9a116 in QCoreApplication::exec() () at ../../include/QtCore/../../src/corelib/global/qflags.h:120
#11 0x000055555570ac42 in (anonymous namespace)::runMixxx (args=..., app=0x7fffffffdd70) at ../../src/main.cpp:37
#12 main(int, char**) (argc=<optimized out>, argv=<optimized out>) at ../../src/main.cpp:112

[ronso0]

I removed all VU meters and all spinnies from Traktmixxx and still Mixxx crashes when switching to it.
Maybe worth noting: there are 6 Waveform widgets created per deck.

The end of the backtrace looks exactly like @daschuer's above.
Full backtrace
bt Traktmixxx without vumeters and spinnies.txt

[daschuer]

Unfortunately I can no longer reproduce the issue.
Is this really a critical bug for the 2.3.5 milestone or can we proceed without a fix?

[ronso0]

I labeled it critical because user data must not crash Mixxx, but it's not clear if this is a 2.3.4 regression, and a fix is not in sight. So let's proceed!

[ronso0]

Update: Traktmixxx doesn't crash Mixxx 2.4

I'll close this for now.