Lib for Arduino #189
Comments
|
I don't maintain it anymore. If you want to play around with it, it still exists at the 2.0.1 tag: |
|
Ok Thank you seems like I need visual studio, don't exist any version
already compiled or we can use only the files *.h ? Thank you
2017-01-09 13:51 GMT+00:00 Adam Crain <notifications@github.com>:
… Closed #189 <#189>.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#189 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AX2Vjy1mfyJAD-PcRPmOVwiks-Ko9kpXks5rQjt7gaJpZM4LeOLc>
.
|
|
The toolchain used is Atmel Studio: http://www.atmel.com/Microsite/atmel-studio/ Yes, it's based on the Visual Studio user interface. There is no precompiled version. |
|
Thanks for your answer and all of your work around dnp3, but now I
realise probably it will not help me. My processador its different .. its
arquitecture is risc from MSP430 Texas Instrument.
Thank you
2017-01-09 14:56 GMT+00:00 Adam Crain <notifications@github.com>:
… The toolchain used is Atmel Studio:
http://www.atmel.com/Microsite/atmel-studio/
Yes, it's based on the Visual Studio user interface.
There is no precompiled version.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#189 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AX2Vj2ChLDzsaKLn-AsKNf1nxbVKUab-ks5rQkqbgaJpZM4LeOLc>
.
|
|
I compiled OpenDNP3 for the Texas Instruments MSP432 a few times ago, and it was fairly straightforward. The tag pointed by Mr. Crain is a good starting point. If I remember correctly, a few classes have changed, but it's easy to find your way trough it. However, I would recommend you using the GCC based compiler provided by TI (available here: http://www.ti.com/tool/MSP430-GCC-OPENSOURCE ). I tried using TI's proprietary compiler, but I had a few issues and couldn't achieve it. Also, note that the resulting code takes a fair bit of space. On the MSP432 Launchpad, with 256KB, I had to use all -Os to have all my code fit into the memory (I had other libraries as well). Please note that my program wasn't doing much and barely interacted with OpenDNP3. It was more of a test of FreeRTOS then OpenDNP3 actually... But still, it is feasible. |
|
Hi Emgre, Thanks for your answer. I'am new at this kind of thing, The msp430 GCC only works with command line right? I'am not finding the correct way to work with this(http://www.ti.com/lit/ug/slau646b/slau646b.pdf checking this info). Do you can create a repo with that info ? Can be useful to more people :) Thanks for you help |
|
The GCC compiler can be used Inside TI Code Composer Studio (the Eclipse-based IDE). It's simply a matter of selecting the right compiler when you create the project. The code that I worked on is on my old laptop hard drive, since it was merely a test. I'll try to reinstall it on my laptop this weekend, but I can't promise you that I'll have time. |
|
Thanks for your precious help. I will wait. avrdude.exe: Device signature = 0x1e9801 (probably m2560) Any ideia ? thank you |
|
@emgre Very cool to know you got things going with the TI toolchain. @ricardolima91 I did get things running on the AtMega2650, but you may have to tweak the solution file b/c of hardcoded paths, etc. It's just not something I can support. The problem you're going to run into on the Mega is SRAM. It only has 8KB and I had to dial back all the buffer sizes to make it run... Even then, it barely fit and without profiling you run the risk of stack/heap collisions. I would't recommend even attempting a port onto anything with less than 16KB. Flash was not a problem. If I recall it used ~128 of 256kb leaving plenty of space for an application. Perhaps not so much if you also have to fit in an RTOS :(. In general, the 2.0.x has migrated away from supporting MCUs easily. Certain C++ types like unique_ptr / and std::function<void ()> have creep-ed into the code, although you could easily write your own simplified definitions for these. -Adam |
|
I found my original code, but it was based on an earlier version of OpenDNP3 and it was full of not related stuff, so I decided to create a new project from the beginning. I have something that compiles, but at runtime, it stops at OutstationContext.cpp:299, and by looking trough the debugger, it seems that StateIdle::instance (in OutstationStates.cpp) is not initialized properly (the _vptr are set to 0x00). My guess is that some optimisations discard this static variable (and its cousins). However, I can't slack the optimisations, otherwise the code won't fit on the Flash. I'm using GCC (arm-none-eabi) with By the way, the latest version is definitely harder for MCUs, especially with the unique_ptr/shared_ptr a bit everywhere that forces the creation on the heap, instead of the stack. The code is available here : https://github.com/emgre/opendnp3msp432 Émile Grégoire |
|
Thank you so much Émile Grégoire for sharing your code. I will explore that info :) i'am new at this but I will try. |
|
@emgre Interesting about the static initializers. I've seen some embedded toolchains (the esp8266 C++ compiler) that didn't call static initializers prior to invoking main(). Perhaps this is the cause? I'm probably going to make those flyweights initialize lazily in the getters since this requires less code. I wonder how the embedded toolchain treats each method of initialization.... |
|
@jadamcrain I fixed the issue with the static initializers and a whole bunch of other issues. I pretty simply had to call the _mainCRTstartup symbol generated by GCC to initialize the static variables. It does all the boilerplate (setting the SP, zeroing the BSS, initializing the static variables), but it doesn't copy the .data segment for an unknown reason. Anymay, I wrote the little loop myself and now it's working. There was also an issue from TI's default linker script where the stack pointer would point to the wrong end of the stack, making it overlapping the heap, then the BSS. That was a nice challenge to find. Also, the _init symbol would get optimized out unless we explicitely tell the linker script to not do so (but that's my fault... I think...) I have the MSP432 sending the first unsollicited null response succesfully. I just need to fix the Executor so that the MSP432 keep talking to the master. I haven't pushed anything yet on the repo, I want to have a working code. I hope to have time to finish it this week, but university is draining my time. By the way, doing this is a nice academic lesson on the ARM architecture and on code generation. I had great fun debugging that stuff. Émile |
|
That's awesome. Sounds a lot like my experience on the Atmel SAM3X8E. Did you write your own limited stubs for std::{unique_ptr, shared_ptr, function} or did your toolchain have something for these? |
|
I got it working! It took me just a few minutes the fix the rest of the code. The interaction with OpenDNP3 was clearly the easy part of this, getting the whole thing compiling as it should was the hard part. You can check the latest commit on my repo. Having the flyweights initialized lazily could help debugging the code. If a constructor fails, at least we will be able to locate it easily with the debugging information. Right now, if the constructor fails, we are still in initialization code (before the main), and you pretty much have to debug it instruction by instruction, walking in his shoes, just to find out what's going wrong. For std::unique_ptr et al., they are provided by the standard library, so no need to provide stubs. I'm linking with newlib-nano, and they seem pretty lightweight (the malloc seems very minimalistic, but it works). Here's a screenshot of the MSP432 talking like a chap via the serial port to a master built with OpenDNP3 and running on my computer. Émile |
|
Looks great. I'll be sure to point folks towards this as an example of porting the library to MCUs. -Adam |
|
It's really a shame that no one from the opendmp3 community tried to organize this for a normal Arduino library already. |
Where is the lib for Arduino ? In the past I know it exist, but now don't appear in any place in repo. Any reason ? Thank you
The text was updated successfully, but these errors were encountered: