[Xenomai-core] LTTng for Xenomai: Supporters wanted

[Xenomai-core] LTTng for Xenomai: Supporters wanted

[Jan Kiszka]

[-- Attachment #1.1: Type: text/plain, Size: 2132 bytes --]

Hi,

looks like we have some new vacancies in this project. On the other
hand, there is now the chance to improve the tool situation of Xenomai
significantly:

I've managed to hack (almost) recent LTTng on top of 2.6.20-ipipe-1.8-01
and Xenomai trunk. Thanks to Jean-Olivier's work for his diploma thesis
[1], we already had a patch to LTTV, the LTTng viewer. But as LTTng is
still in stormy waters (kernel markers etc...), the backend interface
for Xenomai/I-pipe needed a new concept to meet the current, hopefully
now stable design. You can find the result attached in form of a
screenshot of LTTV visualising parts of a simple latency run.

The patches I have here currently need some unwinding from my personal
queue, therefore nothing to publish yet. I will look at this again soon,
how soon also depends on the feedback and the interest to join the party
(actually, I would personally prefer to leave it soon again :) ). To
give you an impression what needs to be done to make this tool usable:

 o Stabilise the "xenoltt" plugin of LTTV (it currently quickly crashes
   under my fingers)
 o Split the Xenomai probe blob into several modules so that loading
   those can be used control which subsystem gets traced
 o Integrate the full set of events into the plugin, the XML facility
   description, and of course the Xenomai code
 o Enhance the visualisation plugin to make use of all events

Specifically the work on the plugin, which is kind of "off-topic" to the
Xenomai core project, needs someone who is a bit familiar with GTK
development (or would like to become this) and who is ideally also
interested in using the result. Maintaining the trace marks in Xenomai
also needs to be done, but this is no significant work once they are in
shape again and are actually used. Also, don't worry about the lowest
bits in I-pipe, that patch is trivial (but not void...).

Looking forward to feedback and to hear from volunteers. I'm convinced
that such a tracing tool will open thrilling perspective for Xenomai
system designers!

Jan

[1] http://ltt.polymtl.ca/svn/lttng-xenomai/

[-- Attachment #1.2: LTTng-Xenomai.png --]
[-- Type: image/png, Size: 41509 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 249 bytes --]

Re: [Xenomai-core] LTTng for Xenomai: Supporters wanted

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 1049 bytes --]

Jan Kiszka wrote:
> Hi,
> 
> looks like we have some new vacancies in this project. On the other
> hand, there is now the chance to improve the tool situation of Xenomai
> significantly:
> 
> I've managed to hack (almost) recent LTTng on top of 2.6.20-ipipe-1.8-01
> and Xenomai trunk. Thanks to Jean-Olivier's work for his diploma thesis
> [1], we already had a patch to LTTV, the LTTng viewer. But as LTTng is
> still in stormy waters (kernel markers etc...), the backend interface
> for Xenomai/I-pipe needed a new concept to meet the current, hopefully
> now stable design. You can find the result attached in form of a
> screenshot of LTTV visualising parts of a simple latency run.
> 
> The patches I have here currently need some unwinding from my personal
> queue, therefore nothing to publish yet. I will look at this again soon,

Patches are now available at

	http://www.rts.uni-hannover.de/rtaddon/patches

Still only as a series, but maintaining multiple queues would be a bit
too much effort for this.

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]