Bug#679826: zsnes: segfaults on start in testing i386

[Ron]

On Sun, Jul 01, 2012 at 11:13:23PM +0200, Etienne Millon wrote:
> * Ron <ron at debian.org> [120701 23:07]:
> > Strace probably won't help a lot here, the problem isn't really tied
> > to interactions with system calls - my first guess would probably be
> > that something in zsnes isn't checking an error return that it
> > should, and continuing to call libao after some earlier operation
> > failed.
> > 
> > It could be something entirely different to that too, but tracing
> > the entry to where this fell over should trip over whatever it
> > actually is.
> 
> It has been filed as #679826, (unrelated to multiarch) -- I'll try
> again once it has been solved.

FWIW severity 'grave' for #679826 looks just a tad hysterical.

Since zsnes is known to work if you actually provide it with I/O devices,
then at most it's probably important, or possibly even just normal, since
it shouldn't segfault, but it would be perfectly legitimate for it to just
report "You have no devices to use this with.  Bad luck for you." and exit
in a more orderly fashion than crashing.

It's not clear from that report that it even got so far as touching libao
before it bailed out in that particular test - but it does point the finger
at zsnes needing to do some better sanity checking of error conditions.

 Cheers,
 Ron