Home > On-Demand Archives > Q&A Sessions >
Live Q&A - Instrumenting Embedded Applications
Dave Marples - Watch Now - Duration: 27:37
Hi Dave,
I found your presentation very informative, very exciting tools with a lot of potential! Making tracing accessible to newcomers/students will be a great investment for the industry.
Looking forward to getting hold of an Orbtrace mini :D (Checking Zyp's store regularly)
Hi,
Thanks for the kind words. We have a huge distance left to travel, but at least we've started the journey new and the tools we already have available are genuinely useful...hopefully folks will build on those to delivery some nicely differentiated capabilities.
Thanks for the presentation. Feels like a dream come true ! Proper unintrusive trace has always been one of the missing tool in my embedded software engineer toolkit. I guess I'll have to try out Orboculum now !
Can you shed a bit more light on which features require ORBTrace, ? and which can work with any debug probe (like J-Link or maybe ST-Link) ? I am currently using Segger RTT over 2 channels on a project : one channel for a CLI interface, another for logging messages. Segger's own tools in this case don't even support displaying properly both channels, I have to go through a custom Python script to exploit that. Maybe I could switch to using SWO and Orbuculum ;)
Hi Nathan,
The SWO link is unidirectional - from the target to the host, but that's generally the important direction of data transfer anyway. JLink and STLink will both terminate SWO and (at least for the JLink case) has been tested with Orbuculum, so you should be good to go.
If you want a proper bidirectional CLI then you can use a SWO channel for the target-to-host and write to a known memory location on the target (anywhere, as long as it's known) and trigger an otherwise unused interrupt by writing to the NVIC to collect the downlink data....obviously that side will need to be done from the debugger, so you can use the control interface of JLink or some gdb magic to do that... Sounds like the perfect job for a few lines of Python :-)
Good luck! Ping us on the discord if you get stuck.
09:29:08 From Stephane to Dave Marples(Direct Message) : https://embeddedonlineconference.com/speaker/Dave_Marples 09:35:44 From Keith J : Love what you guys are doing Dave. I do see a lot of expansion of the parallel output as unlikely as the cost of pins is kind of high. What you guys are doing with the SWO is awesome. 09:38:18 From BobF : Impressive. You highlight FPGA - AMD or INTEL ? PS: Note for amendment Slide 31 - it’s an FGPGA! 09:43:54 From Luke Moll : Apologies if this has been covered/answered already - where can the supporting files for the target be found (to send data to the ITM) as used in the examples? 09:44:23 From Quantum Leaps : Do the standard eval boards expose all the necessary pins? How about NUCLEO boards, for example? 09:44:23 From Jim Norton : Nice scope in the background! 09:44:57 From Keith J : So just scanning the site you mentioned, haven't had time to read yet, but did I see that you support Blackmagic probes? As I have one of those already. 09:50:08 From Simon Smith : Mike not working sorry. I bumped into you at the ARM Hitex conference and found this presentation much easier to follow. Our HW guys have only fitted a 6 pin connector on our boards (JTDI, NJTRST, JTDO-TSWO, SWCLK, SWDIOP, NRST, 3v3). Would Orbuculum work on it? How does it compare with tools such as SystemView or Tracealyzer with a J-Link? 09:50:22 From Luke Moll : If you were able to hook into the RTOS context switch and add ITM messages, would the bandwidth (SWO) be suitable to reconstruct the task execution on the PC? 09:50:22 From Jim Norton : Why do so many EE's not understand the usefulness of an UART? 09:50:27 From Keith J : Thanks Dave. Very interesting project. Would have never heard about it except for EoC! 09:50:27 From Jim Norton : :-) 09:51:14 From Luke Moll : Also - what's the overhead for logging compared to a UART or SEGGER RTT? 09:52:10 From BobF : IEEE Commag morphed into the IEEE IoT journal ? 09:52:15 From Simon Smith : Replying to "Mike not working s..." Thanks. Do you have a getting started tutorial? I looked at Discord, and it was too detailed to understand. 09:53:25 From Jim Norton : Hhahaha 09:55:58 From Luke Moll : Thanks for the talk + Q&A, will be looking for opportunities to use this at work :) 09:56:14 From Jim Norton : Do you believe RISC V will catch up with ARM? 09:57:13 From Jim Norton : Thank you very much! 09:58:50 From Charles Miller : Thank you Dave. Great project.
I really enjoyed this conference!! Hopefully we can have more from you. Thanks a lot