fdo#88006, Valgrind and ImplSVData (vcl part)

[julien2412]

Hello,

 By giving a try to fdo#88006, I retrieved a Valgrind trace (see
https://bugs.freedesktop.org/attachment.cgi?id=111712)
I noticed this (after having added "--track-origins=yes" to Valgrind line
106 of soffice script):
Uninitialised value was created by a heap allocation
...
ImplYield(bool, bool) (svapp.cxx:362)

Taking a look at svapp.cxx, I've got this:
    350     ImplSVData* pSVData = ImplGetSVData();
    351 
    352     // run timers that have timed out
    353     if ( !pSVData->mbNoCallTimer )
    354         while ( pSVData->mbNotAllTimerCalled )
    355             Timer::ImplTimerCallbackProc();
    356 
    357     pSVData->maAppData.mnDispatchLevel++;
... some comments
    362     pSVData->mpDefInst->Yield( i_bWait &&
!pSVData->maAppData.mbAppQuit && !pSVData->maAppData.mbNoYield, i_bAllEvents
);
See http://opengrok.libreoffice.org/xref/core/vcl/source/app/svapp.cxx#362

Here's the content of ImplGetSVData():
     79 ImplSVData::ImplSVData()
     80 {
     81     // init global instance data
     82     memset( this, 0, sizeof( ImplSVData ) );
     83     maHelpData.mbAutoHelpId = true;
     84     maNWFData.maMenuBarHighlightTextColor = Color( COL_TRANSPARENT
);
     85 }

I know it's been like this since 2007 at least but is it really ok to
initialize an object (which contains not only integral types but also
structure/classes) this way?

Julien



--
View this message in context: http://nabble.documentfoundation.org/fdo-88006-Valgrind-and-ImplSVData-vcl-part-tp4135029.html
Sent from the Dev mailing list archive at Nabble.com.