After setting the environment variables ARMGCC_DIR and PATH I was able to build the sources successfully. And this bin works as the prebuilt one when loaded from A5 core through Rpmsg_Demo example.
Since not all the users are experts, I suggest to improve the documentation here in the following way:
in paragraph Build M4 binary on Windows 7 machine, replace “Download GNU ARM Embedded Toolchain and put the contents in C:\MinGW\bin” with “Download the zip package of GNU ARM Embedded Toolchain and extract its content inside C:\MinGW\bin. You should see the folders arm-none-eabi, bin, lib and share below C:\MinGW\bin”
in the same paragraph, add the new step 6 “set the environment variables(My Computer → Properties → Advanced System settings → Environment variables). ARMGCC_DIR=C:\MinGW\bin PATH=C:\MinGW\bin”
in the same paragraph, rename point 6 to 7 and change build.bat to build_all.bat. There is no build.bat file in FreeRTOS source code.
Now this thread has become a little bit “dirty”, so I prefer to open a new one, to find the reason why the sources can’t be compiled when gcc is invoked by Keil DS-5/DS-MDK
The first commit are changes done by NXP, I cannot specifically comment whether those are necessary. The second commit was just a warning which appears when using newer GCC versions, I don’t think that change has actual impact.
Hi @stefan.tx
Do you mean that NXP patched the file for imx7 and didn’t patched it for vf61?
Could you provide a link to the NXP repository with NXP implementation of rpmsg?
Thanks
NXP does not support VF61, the VF61 branch is entirely a Toradex thing. And yes, we did not port the NXP 1.0.1 changes to the VF61 branch, so far we also did not had a need to since rpmsg worked with Linux… That said, it might make sense to actually backport these changes, we just did not came around to do it…
@stefan.tx thank you very much for your support.
In the meanwhile I’m working with Keil engineers to investigate some strange behavior from their side.
