[PyKDE] high event volumes causing seg faults

[Patrick Stinson]

I'm using alsa's rawmidi interface to read midi messages from hardware, then 
pumping them to my qt application code. The problem thus far has been dealing 
with the higher (~50-100/sec) volumes of messages received in a dedicated 
thread, and integrating them safely into the main qt event thread. 

My midi class has a blocking read method that produces messages. a python 
thread is created to call it, and messages are put into a Queue.Queue object. 
Another python thread reads from the queue and uses QApplication.postEvent to 
get the message into the qt thread. All is fine until I call a method like 
'QSlider.setValue' that changes the way a widget looks from my overridden 
QWidget.event() method. I've tried setting a single-millisecond timer to 
generate events in place of the midi object, and of course, everything works 
fine. Why is it that the slider dies when the 'events', however proxied, 
originate in the midi thread?

Sometimes I'm getting negative ref counts; what could be left over or deleted 
so fast that python/qt couldn't keep up?


here is the simple result from an attempt at a 
[midi thread: Queue.put()] -> 
[dispatch thread: Queue.get(), QApplication.postEvent()] ->
[qt thread: event(),QEvent.data()]
:

bash-2.05b$ pksampler
WAD Enabled
pkaudio: found 4 devices, but only using 1.
pkaudio: ->> Found device: hw:PCI,0
pkaudio:          outputs: 2
pkaudio:           inputs: 2
pkaudio:   [[loaded device 1]]
pkaudio: ->> Found device: hw:Extigy,0
pkaudio:          outputs: 6
pkaudio:           inputs: 2
pkaudio: sample rates: 48000
pkaudio: ->> Found device: hw:Extigy,2
pkaudio:          outputs: 2
pkaudio:           inputs: 0
pkaudio: sample rates: 48000
PKSampler: driver buffer size: 2048
PKSampler: driver sample rate: 44100
PKSampler: All daemons up and running
WAD: Collecting debugging information...
WAD: Segmentation fault.
#12  0x40238247 in __clone()
#11  0x40030d40 in pthread_start_thread()
#10  0x080ee9bd in t_bootstrap()
#9   0x080c2da9 in PyEval_CallObjectWithKeywords()
#8   0x0805c3fe in PyObject_Call()
#7   0x080642d3 in instancemethod_call()
#6   0x0805c3fe in PyObject_Call()
#5   0x0810d125 in function_call()
#4   0x080c162f in PyEval_EvalCodeEx()
#3   0x080bf82f in eval_frame()
#2   0x080c3119 in call_function()
#1   0x0810d8d6 in PyCFunction_Call()
#0   0x417f58b1 in _Z26sipDo_QCustomEvent_setDataP7_objectS0_()


--------------------------------------


I've had several horrendous stack traces, but I think the following one best 
suits the extent of my problem...

PKSampler: All daemons up and running
WAD: Collecting debugging information...
SegFault: [ C stack trace ]

#50  0x080b9ef4 in PyEval_EvalCode()
#49  0x080c162f in PyEval_EvalCodeEx()
#48  0x080bf82f in eval_frame()
#47  0x080c3119 in call_function()
#46  0x0810d8d6 in PyCFunction_Call()
#45  0x418de03b in _Z28sipDo_QApplication_exec_loopP7_objectS0_()
#44  0x40563b11 in _ZN12QApplication4execEv()
#43  0x40576928 in _ZN10QEventLoop4execEv()
#42  0x40576a86 in _ZN10QEventLoop9enterLoopEv()
#41  0x4050fea4 in _ZN10QEventLoop13processEventsEj()
#40  0x4055301d in _ZN10QEventLoop14activateTimersEv()
#39  0x418db9fb in _ZN15sipQApplication6notifyEP7QObjectP6QEvent()
#38  0x40562cb5 in _ZN12QApplication6notifyEP7QObjectP6QEvent()
#37  0x405638c5 in _ZN12QApplication14internalNotifyEP7QObjectP6QEvent()
#36  0x41598b6f in _ZN9sipQTimer5eventEP6QEvent()
#35  0x405df932 in _ZN6QTimer5eventEP6QEvent()
#34  0x408da8eb in _ZN6QTimer7timeoutEv()
#33  0x405be454 in _ZN7QObject15activate_signalEi()
#32  0x405be620 in _ZN7QObject15activate_signalEP15QConnectionListP8QUObject()
#31  0x41907334 in _ZN10sipProxyqt9qt_invokeEiP8QUObject()
#30  0x41523175 in _ZN10sipProxyqt9proxySlotEv()
#29  0x40019fdf in _Z13sipEmitToSlotP7sipSlotP7_object()
#28  0x080c2da9 in PyEval_CallObjectWithKeywords()
#27  0x0805c3fe in PyObject_Call()
#26  0x080642d3 in instancemethod_call()
#25  0x0805c3fe in PyObject_Call()
#24  0x0810d125 in function_call()
#23  0x080c162f in PyEval_EvalCodeEx()
#22  0x080bf82f in eval_frame()
#21  0x080c325b in call_function()
#20  0x080c3463 in fast_function()
#19  0x080bf82f in eval_frame()
#18  0x080c325b in call_function()
#17  0x080c3463 in fast_function()
#16  0x080bf82f in eval_frame()
#15  0x080c3119 in call_function()
#14  0x0810d8d6 in PyCFunction_Call()
#13  0x41b4223d in _Z23sipDo_PK_Effect_ControlP7_objectS0_()
#12  0x40020157 in sipMapCppToSelf()
#11  0x4001b713 in sipNewCppToSelf()
#10  0x0805ea92 in PyInstance_New()
#9   0x080c2da9 in PyEval_CallObjectWithKeywords()
#8   0x0805c3fe in PyObject_Call()
#7   0x080642d3 in instancemethod_call()
#6   0x0805c3fe in PyObject_Call()
#5   0x0810d125 in function_call()
#4   0x080c16a8 in PyEval_EvalCodeEx()
#3   0x08082a95 in _Py_Dealloc()
#2   0x0810a7c6 in frame_dealloc()
#1   0x08082a8a in _Py_Dealloc()
#0   0x08082a1e in _Py_ForgetReference()

pkaudio: buffer thread exiting.
pkaudio: engine thread exiting.
PKSampler: All daemons stopped cleanly.
[40724 refs]
Fatal Python error: Objects/frameobject.c:403 object at 0x81ff8a4 has negative 
ref count -1
WAD: Collecting debugging information...
WAD: Abort.
#14  0x08054ca1 in ?()
#13  0x4017390c in __libc_start_main()
#12  0x08054d5a in main()
#11  0x08055638 in Py_Main()
#10  0x080e5b61 in Py_Finalize()
#9   0x080e45ca in PyInterpreterState_Clear()
#8   0x080e4a9c in PyThreadState_Clear()
#7   0x08082a95 in _Py_Dealloc()
#6   0x0810a7b6 in frame_dealloc()
#5   0x0807ec70 in _Py_NegativeRefcount()
#4   0x080e838b in Py_AtExit()
#3   0x401885fb in __GI_abort()
#2   0x40187094 in __GI_raise()
#1   0x4003386b in __pthread_raise()
#0   0x401872e1 in __kill()