[Mep-dev] March Mail Resend

[Michelle]

Our mail server had some issues from a host transfer. Here's what I've been "going on about" for the past week or so. 

We found out a lot about the status of Ekiga, and I think we should help fix it for Beagleboard. 

-=-=-=-=--=-=-=-=-=-

28 Feb 2010

I have revision B7 Beagleboards (TI OMAP processor, which has an ARM Cortex A8 CPU).

I'm running Angstrom, and Ekiga was obtained from the software repository. 

Ekiga crashes, though, and I'm trying to figure out why. I'm hoping someone on the list might be able to point out where I'm going wrong. 

I ran dbg following the instructions at http://wiki.ekiga.org/index.php/Debugging_Ekiga, and got the following.

Program received signal SIGSEGV, Segmentation fault.
0x4000ac6c in ?? () from /lib/ld-linux.so.3

The full file created by gdb is here:
http://microwaveengineeringproject.blogspot.com/2010/02/gdb-results-for-ekiga-on-beagleboard.html

I know that ld-linux.so.* libraries find and load the shared libraries needed by a program, get the program ready to run, and then run it. I saw that Ekiga needed ptlib and libopal. These libraries are installed, along with their -dbg versions for gdb, and -dev version

I'm confident Ekiga should work on the Beagleboard because I found this photograph by Koen on Flickr.

http://www.flickr.com/photos/koenkooi/3208840095/

Ekiga relies upon the Open Phone Abstraction Library, and implements H.323, which is a teleconferencing protocol. I am hoping this will be useful to start with and/or learn from for MEP. 

However, there may be some work that needs to be done to get Ekiga up and running on Beagleboard. It's been ported to ARM, and is in the Angstrom software repository, so probability is high that it's in a useful or very near to useful state already. 

Looking forward to any advice you all might have!

-Michelle W5NYV


-=-=-=-=--=-=-=-=-=-

28 February 2010

James W8ISS had some advice to step back one version and see if that would work. It turns out that the version I was attemping to use (3.2.0) had a known bug where it crashed with an empty account list. I was hopeful that this was the issue, so I moved on to the "unstable" Angstrom repository, where a more recent version of Ekiga was found (3.2.6). So, I tried that. That did not work.

I haven't been able to find any evidence of Ekiga "working" on beagleboard, except for the photograph from Koen on Flickr and some traffic on lists about Ekiga crashing. It's supposed to be ported to ARM, so I am optimistic that most of the work has already been done. 

Ekiga seems like a very good candidate for demonstrating MEP, and I believe it should be easier to get Ekiga working than to write another application from scratch. 

This brings us to Help Wanted - who would like to join the project of getting Ekiga working on Beagleboard for MEP? Do you know of anyone (or can you find evidence of anyone) working on Ekiga for beagleboard that we can collaborate with?

-Michelle W5NYV

-=-=-=-=--=-=-=-=-=-

28 February 2010

Here's a bit of conversation about Ekiga on Beagleboard that happened today.


(10:59:26 AM) Michelle: Hello!
(10:59:50 AM) Michelle: Anyone have Ekiga working on a Beagleboard? 
(11:02:41 AM) johanbr [~j at JBrannlund2.MathStat.Dal.Ca] entered the room.
(11:17:44 AM) Snark: Abraxas3d, someone proposed patches recently for arm support, if I remember well
(11:19:33 AM) Snark: https://bugzilla.gnome.org/show_bug.cgi?id=611308
(11:19:46 AM) Snark: ^^^ that was that bug
(11:22:29 AM) Michelle: That looks promising - I'm still in the process of figuring out what is most likely causing Ekiga to crash. Is there any way I can help with getting Ekiga working on ARM? 
(11:24:32 AM) Michelle: Do I take these attachments and add them to my source for ptlib and ekiga, to try them out? 
(11:25:19 AM) Michelle: they look like scripts, to me. 
(11:25:51 AM) Snark: they are patches
(11:26:03 AM) Snark: notice that they only add support for video
(11:26:15 AM) Snark: which would imply the rest is already working properly
(11:26:45 AM) ced117 [~ced117 at AStrasbourg-157-1-11-114.w90-40.abo.wanadoo.fr] entered the room.
(11:35:17 AM) Michelle: Segmentation faults are all I seem to be getting so far. The beagleboard "ideas" website described Ekiga as "already compiled for the arm-7 on Angstrom (Angstrom being the OS). So The project would be to optimize it for the Beagle (and probably some debugging as Ekiga on beagle apparently is not that stable)." When I read this entry on the table of Beagleboard Ideas (from 2009), I took it to mean that a large part of or almost all of the work was probably done. The stability problems appear on beagleboard various forums - people that were able to get it running experience crashes and other setbacks. 
(11:36:10 AM) Michelle: I'm optimistic it's well within striking range, if not already solved. I've just not had a lot of luck finding the magic recipe or thing I'm missing. 
(11:36:19 AM) Michelle: the link to the bug report is very helpful!
(11:36:42 AM) Michelle: at the very least, encouraging. 
(11:43:38 AM) Snark: if you can get traces, that would help find bugs
(11:44:10 AM) Snark: http://wiki.ekiga.org/index.php/Debugging_Ekiga
(11:44:16 AM) Snark: ^ explanations how to debug
(11:45:39 AM) Michelle: I do indeed have some traces, and have that link marked for reference. I don't want to waste anyone's time with traces that are due to trivial problems on my end, so I haven't submitted any quite yet. 
(11:50:34 AM) Snark: ok
(11:50:48 AM) Snark: does the fact that parts of ekiga come as plugins help?
(11:53:13 AM) [L] [~iruan at dyn128-54-240-70.ucsd.edu] entered the room.
(11:53:35 AM) Michelle: I confess I don't know enough about plugins (other than the basic programming theory) in Ekiga to know if it helps or hurts. When installing, there is a directory of plugins, and they are populated. If there is a missing plugin, I would probably not be able to tell without being better oriented to using and installing plugins.
(11:54:45 AM) Michelle: I'll do my best to check, though.
(11:54:57 AM) Snark: which plugins are enabled?
(11:55:07 AM) Snark: they should all be optional things
(11:55:18 AM) Snark: so removing them to remove their problems could be an idea
(11:59:47 AM) Michelle: none currently installed in the release considered stable from Angstrom (3.2.0), and I'm still working through the rebuild-from-source for 3.2.6. I do remember seeing plugins on the last attempt, and will try your idea when it rebuilds. I appreciate the advice very much.
(12:02:11 PM) Snark: 3.2.0 had some crashes here and there, which have been fixed since


-=-=-=-=--=-=-=-=-=-

2 March 2010

Hi Michelle,

The old version of ekiga was able to do the loopback test, but more recent versions keep crashing. The latest version (3.3.1+git) finally gives something usable in gdb:

Program received signal SIGSEGV, Segmentation fault.
0x409f4c60 in __gnu_cxx::__detail::_Bitmap_counter<__gnu_cxx::bitmap_allocator<PContainerReference>::_Alloc_block*>::_M_reset(long) () from /usr/lib/ekiga/3.3.1/libekiga.so
(gdb) bt
#0 0x409f4c60 in __gnu_cxx::__detail::_Bitmap_counter<__gnu_cxx::bitmap_allocator<PContainerReference>::_Alloc_block*>::_M_reset(long) ()
from /usr/lib/ekiga/3.3.1/libekiga.so
#1 0x409f615c in __gnu_cxx::bitmap_allocator<PContainerReference>::allocate(unsigned int) () from /usr/lib/ekiga/3.3.1/libekiga.so
#2 0x409e9668 in PContainer::PContainer(int) () from /usr/lib/ekiga/3.3.1/libekiga.so
#3 0x409e96f8 in PCollection::PCollection(int) () from /usr/lib/ekiga/3.3.1/libekiga.so
#4 0x409ea6cc in PArrayObjects::PArrayObjects(int) () from /usr/lib/ekiga/3.3.1/libekiga.so
#5 0x409ae6b4 in PPluginManager::GetPluginManager() () from /usr/lib/ekiga/3.3.1/libekiga.so
#6 0x417c80c0 in ?? () from /usr/lib/libpt.so.2.8.0
#7 0x417c80c0 in ?? () from /usr/lib/libpt.so.2.8.0
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

This looks like a problem in opal or ptlib. I lack the time or skills to look into this further :(

regards,

Koen
-=-=-=-=--=-=-=-=-=-

4 March 2010

A couple of us will be at the San Bernardino Microwave Society meeting tonight at 7:00pm in Corona, CA. Please come and check out the meeting if you're able - tonight's presentation is described below:

"At the 4 March 2010 SBMS meeting the "Tech Talk" will be Tony, KC6QHP talking about a wide range of
activities to support the construction of a 47 GHz radio. Including; EDM machining, making waveguide switches,
embedded microcontroller for TR switching, building a suitable tripod head, acquisition of hard to get parts (caps,
wire bonding wire, substrates, chips, epoxies, etc.) putting together circuits with MMICs, interfacing to and from
bare die MMICs, etc. The SBMS meets at the American Legion Hall 1024 Main Street (south of the 91 freeway) in
Corona, CA at 1900 hours local time on the first Thursday of each month. Check out the SBMS web site at
http://www.ham-radio.com/sbms/."

Link to this month's SBMS Newsletter here:

http://www.ham-radio.com/sbms/newsletters/2010nwsltrs/03sbms2010.pdf

-Michelle W5NYV


Potestatem obscuri lateris nescis.



-=-=-=-=--=-=-=-=-=-

4 March 2010

Michelle 提到:
> Hi Paul,
> I was wondering if you were able to get Ekiga working on a Beagleboard or another ARM processor?
> 
> We're hoping to use Ekiga on Beagleboard for an amateur radio project (www.delmarnorth.com/microwave) and you were recommended as someone to ask about whether or not Ekiga "works" on ARM7. 
> 
> 
> Very truly yours,
> -Michelle W5NYV
> 
> 
> Potestatem obscuri lateris nescis.
> 
> 

Hi Michelle,

I didn't work on beagleboard. But if you can run Debian on beagleboard with Xorg, ekiga should be
run well.

I'm actually working on babbage board (SoC: iMX.51).

Recently I made a patch to patch ptlib and ekiga for bpp 16 support.
https://bugzilla.gnome.org/show_bug.cgi?id=611308

I recompile ptlib and ekiga by applying these patches. And add "--disable-xv" when ./configure for
ekiga. On my babbage board, ekiga crashes X because there are some problems with the Xv driver.
The X driver is the evil close source software so I can do nothing there. So I use XImage instead of
Xv and thus make those patches.

Regards,
Paul Liu