[fluid-dev] A MIDI file that causes fluidsynth to segfault

[Ken Restivo]

Attached is a MIDI file that causes fluidsynth to segfault on my machine (Core 
Duo 1.66Ghz 2GM RAM, linux 2.8.19-rt15), jackd 102.20-1, fluidsynth 
1.0.8.2006.12.24-2-SVN)

This is with no softsynths running except one fluidsynth instance, using the 
Steinway Grand soundfont (the 80MB one with all the crossfades). No other JACk 
apps running except qjackctl, and the only other ALSA app being arecordmidi-- 
that's it.

bFreeBoB ERR: Xrun on connection 1
cannot read event response from client [piano] (Connection reset by peer)
subgraph starting at qjackctl-3515 timed out (subgraph_wait_fd=12, status = 0, 
state = Finished)
subgraph starting at qjackctl-3515 timed out (subgraph_wait_fd=12, status = 0, 
state = Finished)
LibFreeBoB ERR: SLAVE XMT : Buffer underrun! 128 (0 / 224) (0 / 0 )
LibFreeBoB ERR: Xrun on connection 1
subgraph starting at qjackctl-3515 timed out (subgraph_wait_fd=12, status = 0, 
state = Finished)
cannot read event response from client [piano] (Connection reset by peer)

Yeah, I can understand an Xrun maybe, at the point where this happens the CPU 
usage is getting pretty high, like around 70% or so. But the entire fluidsynth 
losing control of its bladder and quitting? Seems extreme to me. The docs for 
the soundfont (from 2001) state that the font might give some crackles "on a 
slow CPU", but my current CPU would is faster than anything that was 
commercially available in 2001.

The noodling trailing off at the end of the file is me going, hello?, wait a 
minute, why is there no sound coming out? 

Makes me a sad panda.

-ken

2007-02-21.3753.mid
Description: Binary data

pgpXQZ9ozVO87.pgp
Description: PGP signature