Hydrogen

On Tue, October 21, 2008 2:37 pm, Sebastian Moors wrote:
> at the moment hydrogen refuses to start here... This is not the normal
> situation, the same executable works most of the time. Furthermore it
[snip]
> my computer or run "alsaconf", everything works fine. The problem
> appeared in the past on other systems too, often after jack crashed.
> Maybe someone is able to guess why the cast failed and what went wrong
> here..
[snip]
>
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0xb64166d0 (LWP 11096)]
> 0xb70af34b in __dynamic_cast () from /usr/lib/libstdc++.so.6
> (gdb) bt
> #0  0xb70af34b in __dynamic_cast () from /usr/lib/libstdc++.so.6
> #1  0x081e332a in H2Core::Sampler::makeTrackOutputQueues ()

Throwing out ideas...

If you try a dynamic_cast<> on an invalid pointer, it will segfault. 
Perhaps this code got executed before __audio_output was initialized.

Also, I recall once that I was having trouble running hydrogen through the
debugger.  It may have been this very segfault.  I think I resolved it by
upgrading gdb to 6.8.  (Sometimes, bugs can be caused by gdb -- and I
think I read something that made me suspect this was the case.)  However,
I also resolved it by _not_ using a debugger, but instead using core dumps
for segfault backtraces.

Also, what was H2's output before the segfault?

Thanks,
Gabriel

-- 
               G a b r i e l   M   B e d d i n g f i e l d

Hey Gabriel,

Gabriel M. Beddingfield wrote:
> Throwing out ideas...
> If you try a dynamic_cast<> on an invalid pointer, it will segfault. 
> Perhaps this code got executed before __audio_output was initialized.
>
> Also, I recall once that I was having trouble running hydrogen through the
> debugger.  It may have been this very segfault.  I think I resolved it by
> upgrading gdb to 6.8.  (Sometimes, bugs can be caused by gdb -- and I
> think I read something that made me suspect this was the case.)  However,
> I also resolved it by _not_ using a debugger, but instead using core dumps
> for segfault backtraces.
>
> Also, what was H2's output before the segfault?
>   
I was able to get some more details. The bug occurs outside gdb too,
so it can't be related to gdb. And more interesting:  It happens only if 
i start hydrogen
without commandline arguments or with hydrogen --driver null . It 
crashes *not* if i start with "--driver alsa"
or jack or whatever.. The output before the segfault can be found at the 
end of this mail..

- Sebastian





sebastian@...$ ./hydrogen --driver null

Hydrogen 0.9.4-svn583:584M [Oct 17 2008]  [http://www.hydrogen-music.org]
Copyright 2002-2008 Alessandro Cominu

Hydrogen comes with ABSOLUTELY NO WARRANTY
This is free software, and you are welcome to redistribute it
under certain conditions. See the file COPYING for details

VST_PATH not set, defaulting to 
/home/sebastian/vst:/usr/local/lib/vst:/usr/lib/vst
RemoteVSTClient: all cache files are up-to-date, not running scanner
(W) Preferences loadPreferences patternCategories node not found
(W)     static int H2Core::LocalFileMng::readXmlInt(TiXmlNode*, const 
QString&, int, bool, bool) 'jack_track_output_mode' node not found
(W)     static bool H2Core::LocalFileMng::readXmlBool(TiXmlNode*, const 
QString&, bool, bool) 'jack_ardour_transport_workaround' node not found
(W)     static int H2Core::LocalFileMng::readXmlInt(TiXmlNode*, const 
QString&, int, bool, bool) 'countoffset' node not found
(W)     static int H2Core::LocalFileMng::readXmlInt(TiXmlNode*, const 
QString&, int, bool, bool) 'playoffset' node not found
(W) SongReader  readSong Trying to load a song created with a different 
version of hydrogen.
(W) SongReader  readSong Song [/usr/src/htest/data/DefaultSong.h2song] 
saved with version 0.9.4-svn
(E) OssDriver   connect DSP ERROR_OPEN
(E)     void H2Core::audioEngine_startAudioDrivers() Error starting 
audio driver [audioDriver::connect()]
(E)     void H2Core::audioEngine_startAudioDrivers() Using the NULL 
output audio driver
(E)     void H2Core::audioEngine_startAudioDrivers() m_pMainBuffer_L == NULL
(E)     void H2Core::audioEngine_startAudioDrivers() m_pMainBuffer_R == NULL
(W)     static QString H2Core::LocalFileMng::readXmlString(TiXmlNode*, 
const QString&, const QString&, bool, bool) 'license' node not found
(W)     void H2Core::audioEngine_setSong(H2Core::Song*) Set song: 
Untitled Song
(E)     void H2Core::audioEngine_setupLadspaFX(unsigned int) nBufferSize=0
Segmentation fault

Sebastian Moors wrote:
> Hey Gabriel,
>
> Gabriel M. Beddingfield wrote:
>   
>> Throwing out ideas...
>> If you try a dynamic_cast<> on an invalid pointer, it will segfault. 
>> Perhaps this code got executed before __audio_output was initialized.
>>
>> Also, I recall once that I was having trouble running hydrogen through the
>> debugger.  It may have been this very segfault.  I think I resolved it by
>> upgrading gdb to 6.8.  (Sometimes, bugs can be caused by gdb -- and I
>> think I read something that made me suspect this was the case.)  However,
>> I also resolved it by _not_ using a debugger, but instead using core dumps
>> for segfault backtraces.
>>
>> Also, what was H2's output before the segfault?
>>   
>>     
>  And more interesting:  It happens only if 
> i start hydrogen
> without commandline arguments or with hydrogen --driver null . It 
>
>   

After a fast look in main.cpp it turned out that there is no "--driver 
null" , so it is feasible why it is the same as "hydrogen" without 
arguments :)
- Sebastian

Sebastian Moors wrote:
> Hey Gabriel,
> so it can't be related to gdb. And more interesting:  It happens only if 
> i start hydrogen
> without commandline arguments or with hydrogen --driver null . It 
> crashes *not* if i start with "--driver alsa"
> or jack or whatever.. The output before the segfault can be found at the 
> end of this mail..

I can't reproduce this.  I've tried no arguments, with drivers 
jack/oss/alsa/null, with svn HEAD, with a local version of jackMidi, with jackd 
running, with jackd not running... I can't get it to segfault.

> VST_PATH not set, defaulting to 
> /home/sebastian/vst:/usr/local/lib/vst:/usr/lib/vst
> RemoteVSTClient: all cache files are up-to-date, not running scanner

I haven't seen this before.  How did you get VST support?  (Just curious)

> (E) OssDriver   connect DSP ERROR_OPEN

I tried to default to OSS.

> (E)     void H2Core::audioEngine_setupLadspaFX(unsigned int) nBufferSize=0
> Segmentation fault

Buffer size 0?

-Gabriel

Hi,


hydrogen segfaults here (debian sid, jack 0.116.1) when the audio device 
is not available (for instance, running rhythmbox / alsa) while starting 
up hydrogen
Here's the output:


<-- snip

JACK compiled with System V SHM support.
loading driver ..
Enhanced3DNow! detected
SSE2 detected
apparent rate = 0
creating alsa driver ... hw:0,0|hw:0,0|1024|2|0|0|0|nomon|swmeter|-|32bit
control device hw:0
the playback device "hw:0,0" is already in use. Please stop the 
application using it and run JACK again
cannot load driver module alsa
(E)     void H2Core::audioEngine_init() Error the audio engine is not in 
UNINITIALIZED state
(E) JackOutput    init Unknown status with JACK server.

-->


gdb:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fb7145fb710 (LWP 4438)]
0x00007fb712288e00 in jack_get_sample_rate () from /usr/lib/libjack.so.0
(gdb) bt
#0  0x00007fb712288e00 in jack_get_sample_rate () from /usr/lib/libjack.so.0
#1  0x0000000000575219 in H2Core::JackOutput::init ()
#2  0x00000000005625f6 in H2Core::createDriver ()
#3  0x00000000005634a0 in H2Core::audioEngine_startAudioDrivers ()
#4  0x00000000005641ac in H2Core::Hydrogen::Hydrogen ()
#5  0x0000000000564355 in H2Core::Hydrogen::get_instance ()
#6  0x00000000005763ad in H2Core::JackOutput::init ()
#7  0x00000000005625f6 in H2Core::createDriver ()
#8  0x00000000005634a0 in H2Core::audioEngine_startAudioDrivers ()
#9  0x00000000005641ac in H2Core::Hydrogen::Hydrogen ()
#10 0x0000000000564355 in H2Core::Hydrogen::get_instance ()
#11 0x00000000005a6327 in H2Core::SongReader::readSong ()
#12 0x00000000005a9d09 in H2Core::Song::load ()
#13 0x00000000005a9e02 in H2Core::Song::get_empty_song ()
#14 0x00000000004bdf85 in MainForm::MainForm ()
#15 0x00000000004eaf22 in main ()

Sebastian Moors wrote:
> hydrogen segfaults here (debian sid, jack 0.116.1) when the audio device 
> is not available (for instance, running rhythmbox / alsa) while starting 
> up hydrogen

Try the attached patch.

Thanks,
Gabriel

Hey,

thanks! The patch is working fine for me, i hope that fixes all the 
segfaults i had during application start in the past.
- Sebastian

Gabriel M. Beddingfield wrote:
> Sebastian Moors wrote:
>> hydrogen segfaults here (debian sid, jack 0.116.1) when the audio 
>> device is not available (for instance, running rhythmbox / alsa) 
>> while starting up hydrogen
>
> Try the attached patch.
>
> Thanks,
> Gabriel
>
>
> ------------------------------------------------------------------------
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by:
> SourcForge Community
> SourceForge wants to tell your story.
> http://p.sf.net/sfu/sf-spreadtheword
> ------------------------------------------------------------------------
>
> _______________________________________________
> Hydrogen-devel mailing list
> Hydrogen-devel@...
> https://lists.sourceforge.net/lists/listinfo/hydrogen-devel