Posted by theuncommonmanu on 2015-04-13 17:28

I had the same problem: my cpp files were not compiled.
I found that the file was not compiled because while importing c++ files, the compiler is not automatically set as it is for C files.
To correct this, you have to select your cpp file in the project explorer, right click -> Properties -> C/C++ build -> Tool chain Editor -> Select tool
Here select “MCU G++ Compiler” instead of “NO TOOL”

Once done, on my side, the workbench now tries to compile the file but I have no compilation errors:
cc1plus: error: invalid option `thumb’
cc1plus: error: invalid option `float-abi=soft’
myfile.cpp:1: warning: -ffunction-sections may affect debugging on some targets

and I don’t know yet why?

Posted by tang on 2015-04-14 10:03

Hi everyone,

Converting C project to C++ project is not supported yet. However i found a trick to manually do it.
In the project directory, you should find a “.project” file. Add “org.eclipse.cdt.core.ccnature” under the “natures” parent element, then save the changes.
The project has to be refreshed (or close then re-open).

Under the project properties, you should find the G++ settings :

Hope it help you.

Kevin.

Posted by theuncommonmanu on 2015-04-14 14:17

Thanks. This workaround allows me to compile.

Nevertheless, I had to redefine the include paths (included the ones of the HAL driver) and the preprocessor options “USE_HAL_DRIVER” and “STM32F401xE” (in my case) via the following path(C/C++ General -> Paths and Symbols -> Symbols-> Gnu C++.

Posted by rupeshg on 2015-04-14 16:49

Kevin,
The workaround also worked for me too! Like uncommonmanu said, I had to add the path and symbols, also add the right linker file since the g++ linker was pointing to some default *.ld file.
I was able to create a dummy c++ test class and compile and run with the stmf439xx micro. Can’t wait to port our project from IAR to gcc. Goodbye 3K+ compiler license fees

-Rup