GSOC'18 project description

Gedare Bloom gedare at rtems.org
Mon Apr 30 16:09:30 UTC 2018


On Mon, Apr 30, 2018 at 6:05 AM, Sebastian Huber
<sebastian.huber at embedded-brains.de> wrote:
> Hello Vidushi,
>
> On 30/04/18 09:15, Vidushi Vashishth wrote:
>>
>> Hello!
>>
>> I am Vidushi Vashishth, and I have been selected for the Google Summer of
>> Code'18 program. I look forward to a productive and enriching summer with
>> the RTEMS community.
>
>
> I am looking forward to get some new insights into a network stack
> performance issue with 10Gbit/s Ethernet on a QorIQ T4240 with your tracing
> support.
>
>>
>> I will be working on enhancement of the tracing system of RTEMS, under the
>> guidance of mentors: Gedare Bloom, Chris Johns and Sebastian Huber. The main
>> objectives of my project are threefold:
>>
>> 1) Combine the Common Trace Format with the functionality of the Trace
>> Linker. I will be working with barectf and rtems-tld in this regard.
>> 2) Coming up with a synchronisation and transport mechanism to send trace
>> records to the host machine.
>> 3) Investigate the importance of live tracing functionality and addition
>> of kernel level tracing. I will design both possibilities and decide which
>> one to work on by the end of phase 1 of my evaluation.
>
>
> I work currently on a new low level infrastructure to support gathering of
> thread (create, delete, terminate, switch) and interrupt (entry, exit)
> events. Similar to the capture engine, but with less overhead. I would like
> to be able to gather at least 200000 events per second per processor (24 on
> the QorIQ T4240). This results in about 80MiB/s of data so that I can
> transfer it easily via a 1Gbit/s Ethernet interface.
>
> I don't know how to setup the CTF infrastructure to display this data. I
> hope that you can help me here.
>

Vidushi can add a task of creating a babeltrace formula for conversion
from your custom format to CTF to facilitate unified analysis. Just
need a good description and examples of the custom format I guess.

>>
>> The wiki page for my project is:
>> https://devel.rtems.org/wiki/GSoC/2018/EnhancementRuntimeTracing
>>
>> I will be documenting the progress of my project on my blog:
>> https://vidushivashishth.github.io/
>>
>> Both of these sources are empty right now but will be modified with
>> relevant information over the course of this summer.
>
>
> I would prefer to use the ticket system as much as possible for this
> project.
>

Yes, it would be good to have tickets at the start of each phase as a
way to track progress and issues. Vidushi, maybe you can begin by
pulling out of your proposal the few tasks in Phase 1 and create a new
ticket for each one?

> The main audience for the runtime tracing are the RTEMS users. A key issue
> is to provide adequate documentation so that it is easy to use. If you need
> this tracing then you probably have a nasty application issue and no time to
> struggle with setting up things with a project deadline looming. Where
> should this documentation go? A new chapter in the User Manual
>
> https://docs.rtems.org/branches/master/user/index.html
>
> ?
>

This is a good and important question. I think there should be a
section in this User Manual under Host Tools for the Babeltrace/Debug
tools that use the CTF output. There will also likely need to be
additions to the Classic API guide for "Configuring a System" and
improvements to description of the Capture Engine, which I'm currently
having trouble finding.

Gedare

> --
> Sebastian Huber, embedded brains GmbH
>
> Address : Dornierstr. 4, D-82178 Puchheim, Germany
> Phone   : +49 89 189 47 41-16
> Fax     : +49 89 189 47 41-09
> E-Mail  : sebastian.huber at embedded-brains.de
> PGP     : Public key available on request.
>
> Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
>
> _______________________________________________
> users mailing list
> users at rtems.org
> http://lists.rtems.org/mailman/listinfo/users



More information about the users mailing list