On Tue, 2005-02-22 at 16:10 -0500, Paul Davis wrote:
> looks very much like a race in H2 to me:
>
> >inline void audioEngine_process_clearAudioBuffers(uint32_t nFrames)
> >{
> >*---> memset( m_pMainBuffer_L, 0, nFrames * sizeof( float ) ); // cle
> >ar main out Left
> > memset( m_pMainBuffer_R, 0, nFrames * sizeof( float ) ); // cle
> >ar main out Right
>
>
> >(gdb) bt full
> >#0 0xb759949b in memset (dstpp=0x0, c=0, len=512)
>
> so, the port buffer pointer was null. going further down the stack:
>
> >#2 0x08074381 in audioEngine_process (nframes=128, arg=0x0) at
>
> i'm suprised to see that the "arg" is null.
>
> > sampleRate = 1.17287768e-35
>
> odd, though possibly OK.
>
Paul,
Thanks for your help. Basically the issue was that the process()
callback could get called before Hydrogen's audio engine was completely
initialized. For some reason the H2 developers could not reproduce the
crash.
The bugs (there were actually two) are now fixed, one by Comix, one by
myself. Everyone who was hitting this, please try H2 CVS.
Lee
