Mantis Bugtracker

Viewing Issue Simple Details Jump to Notes ] View Advanced ] Issue History ] Print ]
ID Category Severity Reproducibility Date Submitted Last Update
0000945 [Slicer3] Base Code crash always 2010-08-10 15:53 2010-08-14 18:10
Reporter inorton View Status public  
Assigned To pieper
Priority normal Resolution open  
Status assigned  
Summary 0000945: Nightlies hang on linux / ubuntu 10.04 and others
Description I'm trying to use the 3.7 linux-x86 nightlies but Slicer hangs on startup. Have tried several 3.7 nightlies from 7/24-8/5 and the issue occurs in all of them. This is not present in the 3.6.1 nightly (8/5) or the 3.6 stable (6/10) - those start just fine.

./Slicer3 and "Slicer3 -t" usually hangs on "Discovered IA_FEMesh Loadable Module (adding to cache)"
--no_modules and --no_module_cache hangs on "Initializing Models Module".
--test_mode just hangs

I should clarify that the splash window still redraws, so by "hang" I mean that startup doesn't go any further but appears not to be totally crashed. Slicer3-real cpu utilization is reported by top as Note: 0000090%, and I have let it sit for several minutes with no further progress in each case. I initially thought this might be an issue only with two recently (<1mo) upgraded computers - old library conflict or something. However, the same happens on a clean install (this morning).
Additional Information
Tags No tags attached.
Attached Files txt file icon bt1.txt [^] (47,445 bytes) 2010-08-10 15:53 [Show Content]
txt file icon bt2.txt [^] (23,696 bytes) 2010-08-10 15:59 [Show Content]

- Relationships

-  Notes
(0001928)
inorton (developer)
2010-08-10 15:54

Well, the clean build runs fine so that is a start.

I moved all modules out, but Slicer immediately exits (message below)
unless several modules are present:
CommandLine, ScriptedModule, SlicerDaemon, Volumes, SlicerVolumes, and
all of the tractography .so modules in lib/Slicer3/Modules.

After replacing only those modules above, Slicer runs to splash but the
hang behavior is the same (stuck on Volumes module).

-Isaiah

error:
"""
/home/isaiah/sw/Slicer3-3.7-alpha-2010-08-04-linux-x86/bin/Slicer3-real:
error while loading shared libraries:
libSlicerTractographyFiducialSeeding.so: cannot open shared object file:
No such file or directory

Launch
/home/isaiah/sw/Slicer3-3.7-alpha-2010-08-04-linux-x86/bin/Slicer3-real
returned with error: child process exited abnormally
"""


On 08/05/2010 04:43 PM, Steve Pieper wrote:
> > Hi Isaiah -
> >
> > Interesting - the stack tract helps. It looks like MapFrame is
> > getting called recursively, triggered by one of the shared libraries
> > that is being loaded.
> >
> > I'd say the way to debug this would be to move all the shared
> > libraries out of lib/Slicer3/Modules and see if the problem goes
> > away. Then re-introduce them one by one and see if we can find the
> > one that triggers the problem.
> >
> > Thanks for your help with this,
> > Steve
(0001929)
inorton (developer)
2010-08-10 15:55

Shoot - well, I'm not sure how to debug this one remotely. It does not happen on my linux machines. From the stack trace it looks like there is a render being done before the render window is fully initialized. I thought that meant that RequestRender is being called (perhaps indirectly) from a module's constructor and then getting 'update'd from the splash screen.

Can you do me a favor and send the gdb stack trace from the hang with --no-spash is specified?

Thanks,
Steve

On Aug/6/10 1:49 PM, Isaiah Norton wrote:
> Yeah, I retried no splash, test mode, -t, no modules, no module cache,
> and maybe 1-2 other ones that looked relevant... No dice.
>
> Thanks,
>
> -Isaiah
>
> On 08/06/2010 01:43 PM, Steve Pieper wrote:
>> Isaiah - can you try adding the --no-splash argument and see if that
>> changes the behavior?
>>
>> Thanks,
>> Steve
(0001930)
inorton (developer)
2010-08-10 15:57

From: Bill Lorensen <bill.lorensen@gmail.com>
To: swallace@nmr.mgh.harvard.edu
Cc: slicer-devel@bwh.harvard.edu
Date: Thu, 5 Aug 2010 18:40:00 -0400


Sounds like the problem I had a month or so ago on Fedora.

On Thu, Aug 5, 2010 at 6:23 PM, <swallace@nmr.mgh.harvard.edu> wrote:
> Had a potentially similar issue with Ubuntu 10.04 and building; it
> wouldn't hang on building Affine Registration. Don't know if this is the
> same issue, but thought I'd share, in case it's useful.
>
>> I'm having the same trouble on Ubuntu 8.10. I'm trying a new build now
>> and will report back.
>>
>> Katie
(0001931)
inorton (developer)
2010-08-10 15:59

Here's the trace with --no_splash, "bt2". Bug report here: http://www.na-mic.org/Bug/view.php?id=945. [^] For what it's worth, I've tried three distros now and have gotten the same result: Ubuntu 10.04, Fedora 11, and Knoppix 6.2 (running in a virtualbox). A student here has seen same thing in a 32-bit ubuntu 10.04 install on virtualbox on a 64-bit mac.
(0001933)
pieper (administrator)
2010-08-14 18:10

I plan to review this with Isaiah on his laptop in the next few days.

- Issue History
Date Modified Username Field Change
2010-08-10 15:53 inorton New Issue
2010-08-10 15:53 inorton File Added: bt1.txt
2010-08-10 15:54 inorton Note Added: 0001928
2010-08-10 15:55 inorton Note Added: 0001929
2010-08-10 15:57 inorton Note Added: 0001930
2010-08-10 15:59 inorton Note Added: 0001931
2010-08-10 15:59 inorton File Added: bt2.txt
2010-08-14 18:10 pieper Note Added: 0001933
2010-08-14 18:10 pieper Assigned To => pieper
2010-08-14 18:10 pieper Status new => assigned


Mantis 1.1.4[^]
Copyright © 2000 - 2008 Mantis Group
Powered by Mantis Bugtracker