Re: [fluid-dev] Crashing bug in fluidsynth (CVS)

[Josh Green]

On Thu, 2005-09-15 at 01:44 -0700, Mo DeJong wrote:
> On Thu, 15 Sep 2005 00:48:54 -0700
> Josh Green <address@hidden> wrote:
> 
> > There was a 64 bit related crash bug that was also fixed with this
> > section of code (I believe it was an AMD 64 bit system).  What
> > architecture is your system?  Can you tell if USE_LONGLONG is defined or
> > not?  This define determines the format of the fluid_phase_t structure
> > and the corresponding macros in fluid_phase.h.
> >     Josh
> 
> This is running in a 32 bit Win32 machine. As far as I can tell, this has 
> nothing
> to do with the decl type of fluid_phase_t, there is something very wrong with
> the pointer voice->sample->data, it seems that when the dsp code in
> fluid_voice.c is entered, the voice->sample->data has already gone bad
> and now points off to some invalid memory. It does not seem to be the
> same memory from the soundfont->sampledata that is seemed to point
> to originally. I have tried to track down the problem in gdb, but I am not
> having much luck.
> 
> cheers
> Mo DeJong
> 
> P.S.
> Did you notice that the whole ramsfont module is being ignored? It gets
> built but the code is not used.
> 

Does this problem occur with every SoundFont that you have tried?  I may
have to try and reproduce this problem in order to figure out whats
wrong.  Does your previous statement now no longer apply?

> The problem is that dsp_phase value is a really large
> number (1057488896) and the dsp_data[dsp_phase_index]
> array is not large enough for this index, so it segfaults.

Are the sample index and fraction incorrect?  Or it seems you are saying
now that there is something wrong with the pointer.  I'll setup mingw on
my VMWare windows virtual machine, since I've been meaning to do this
for some time.  Keep us posted :)  Cheers.
        Josh Green

signature.asc
Description: This is a digitally signed message part