On 26/10/2022 09:46, Xisco Fauli wrote:
Attached you can find the list of LibreOffice 7.4.2.3 crashes reported to https://crashreport.libreoffice.org/stats/version/7.4.2.3
In
Name Count First report Last Report ID Version Reason OS Stack Code Lines Last 4 UNO Commands
[...]
osl_releaseMutex 29399 16/06/28 22/10/25 12300b88-6e9e-4855-8b2e-45871b39b55c 7.4.2.3 Unhandled C++ Exception windows sal/osl/w32/mutex.cxx:83 cppu/source/typelib/typelib.cxx:2059 include/cppu/unotype.hxx:300 cppuhelper/source/shlib.cxx:310 cppuhelper/source/servicemanager.cxx:818 cppuhelper/source/servicemanager.cxx:1778 cppuhelper/source/servicemanager.cxx:1003 cppuhelper/source/servicemanager.cxx:970 desktop/source/app/app.cxx:1925 vcl/source/window/winproc.cxx:2799 vcl/inc/salframe.hxx:308 LeaveCriticalSection(pMutexImpl); { return cppu_detail_getUnoType(static_cast< T1 * >(NULL)); throw css::loader::CannotActivateFactoryException( cppuhelper::detail::loadSharedLibComponentFactory( loadImplementation(context, impl); std::shared_ptr< Data::Implementation > impl( return createInstanceWithContext(aServiceSpecifier, context_); xSMgr->createInstance('com.sun.star.bridge.OleApplicationRegistration'); ImplHandleUserEvent( const_cast<ImplSVEvent *>(static_cast<ImplSVEvent const *>(pEvent)) ); { return m_pProc && m_pProc( m_pWindow, nEvent, pEvent ); }
the "Count: 29399" is presumably the number of reports that, at the time the above document was generated, had the same "Signature: osl_releaseMutex" (<https://crashreport.libreoffice.org/stats/signature/osl_releaseMutex>) as <https://crashreport.libreoffice.org/stats/crash_details/12300b88-6e9e-4855-8b2e-45871b39b55c>?
That specific signature is apparently overly unspecific and groups together wildly different crash scenarios. When computing those crash signatures, is there a way to filter out such stack frames that have been identified as being too unspecific?
