On Sat, Dec 25, 2010 at 11:22:48PM +0100, wylda@volny.cz wrote:
Hi guys,
there are bug reports showing random crashes with audio on, during winecfg etc. I was bit by those too... There is probably something wrong how regressions are handled here, which goes on and thus there are unneeded bugreports which cost everyone's time to read, evaluate, search, mark, retest... Why don't you use revert more offen like kernel guys?
Back to the sound issue - my zero programming skill standing just on sound despair, i guess in dlls/winealsa.drv/mixer.c ALSA_MixerInit() should be
int card, err, capcontrols=0
otherwise things like:
mixdev[mixnum].chans += capcontrols;
mixdev[mixnum].lines = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(line) * mixdev[mixnum].chans);
mixdev[mixnum].controls = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(control) * CONTROLSPERLINE*mixdev[mixnum].chans);
goes crazy like in http://bugs.winehq.org/show_bug.cgi?id=25231#c2. Could someone review this piece of code, please?
well spotted, the patch incorrectly removed the = 0.
I submitted a patch readding it.
This specific regression I did not hear about it yet, if someone sees such a regression on bisect, please get the right people in the bugzilla (like the author of the patch) and/or post to the list pointing out the regression.
The point is ... speak up if you see such a clear regression.
Ciao, Marcus