RTEMS Trace Tool
isaac.gutekunst at vecna.com
Wed Feb 10 16:20:01 UTC 2016
I've been poking around the trace infrastructure myself lately. I don't
actually know what you're trying to accomplish, and have not actually
used the all the tools.
When you install rtems, it puts rtems-tld in the install path. That
might be helpful to you.
Some very helpful resources for understanding the capture engine are the
testsuites/smptests/smpcapture01 and 02
I haven't actually run them, and at the moment am using my own capture
backend for a different project, but it's very helpful.
From seeing the usage, you can read the source in
cpukit/libmisc/capture as Joel mentioned and make more sense of it.
One limitation I have discovered, that may be solvable, is that you
cannot both capture data, and read it back at the same time, which is
important if you want to stream some kind of trace live.
I've implemented a barebones lttng-live protocol TCP server, and am in
the final stages of debugging the prototype. I feel it could potentially
be useful here. Let me know if you're interested in hearing more.
On 02/10/2016 10:32 AM, Joel Sherrill wrote:
> On Wed, Feb 10, 2016 at 4:11 AM, Arwa Khannoussi
> <khannoussi.arwa at gmail.com <mailto:khannoussi.arwa at gmail.com>> wrote:
> I just started using the RTEMS trace tool, after having a look at
> https://devel.rtems.org/wiki/Developer/Projects/Open/TraceTool and
> It's mentioned to contact Chris Johns for more informations, so here
> I am.
> Chris is spotty right now so it might be a bit before he replies.
> He mentored this GSoC project and it was one step in a multi-step process to
> get everything we want for tracing and timeline visualization.
> My questions are:
> - Is this tool still functional, because there is no updates since 2009?
> The code should be as functional as it ever was except for what I will
> describe below.
> - I'm using RTEMS 4.11 not 4.10 does that make any problem?
> At some level yes, but the code you are looking at is a snapshot of a
> work in progress. I would NOT use an older RTEMS just to be in sync with
> it. The trace code should be updated to track RTEMS. The trace code is
> the code at a more immature stage of development.
> - In the sources given at
> https://code.google.com/archive/p/rtems-tracing/ the file wrapper.h
> used in rtems_trace_generator.py is missing so I tried to fixe the
> bugs caused by the absence of this file but the problem is that
> there is not the sources of the two principals functions
> "rtems_capture_trace_record" and "rtems_capture_trace_log".
> My question is where could I find those two functions?
> The capture engine had a significant rework between 4.10 and 4.11 to be
> SMP safe. Methods were renamed. The code is in cpukit/libmisc/capture. I
> don't know if there is user level documentation on the APIs or not off
> the top of my head. Check the code/.h files in that directory.
> The trace wrapper generator should now be in the rtems-tools git
> repository on git.rtems.org <http://git.rtems.org>.
> starting around slide 14 is some info on it. There is a long term plan
> but everything has been done by volunteers. The ultimate goal is to
> automatically generate trace wrappers (now part of rtems-tools), capture
> the events (capture engine), have a static or dynamic connection to a
> host, provide the log in Common Trace Format, and visualize the timeline
> (Linux Trace Toolkit).
> So parts of the chain exist and just need polishing and tying together.
> Certainly this is an area ripe for a GSoC project. :)
> Thank you,
> KHANNOUSSI Arwa.
> KHANNOUSSI Arwa
> Ingénieur Informatique
> Master 2 Logiciels pour les Systèmes Embarqués -Université de
> Bretagne Occidentale.
> Tèl: 0612675563
> users mailing list
> users at rtems.org <mailto:users at rtems.org>
> users mailing list
> users at rtems.org
More information about the users