TI EvalBot and Windows, Eclipse style
[Work in progress]
This guide is based on Brendan Robert's guide for linux which can be found here.
I finally got my Evalbot from Texas Instruments, only took them over six months to ship it. Anyway, after I got it, I started researching for examples, guides, tutorials and most importantly: the best IDE that I could use.
Documentation regarding the board is pretty scarce and the most IDE's out there have all sorts of limitations and bloatware.
This is when I decided that I would go throug the minimal steps of setting up Eclipse as your primary IDE on Windows, so you could easily install and configure a proper development environment and get started by writing your first HelloWorld project.
Eclipse CDT - Your first task is to download Eclipse CDT. Unzip it somewhere safe.
StellarisWare - While this step is optional, I highly recommend downloading the example sources from StellarisWare, as they contain libraries (and sample code) that'll prove very useful.
CodeSourcery G++ Lite - This is the toolchain, so you'll definitely be needing this one. Just grab the Windows installer and install it.
GNU ARM Eclipse-plugin - Eclipse wouldn't know what to do with you projects without this. Install the plugin according to the instructions, just make sure you do it the recommended way!
LMFlash - We need to be able to download the application to the board.
After all this is done, it's time to create a new C project in Eclipse. Remember to choose ARM Windows GCC (Sourcery G++ Lite) when creating the project.
We'll be using Martin Hubacek's Nunchuk library for this HelloWorld project, so download that here. After unzipping it, copy the following files to the root of your HelloWorld project directory:
I2Clib.c
I2Clib.h
oled.c
oled.h
startup_gcc.c
main.ld
You're almost there! We will now configure all the project settings so the project knows where everything is and will compile without a fuss.
Open up your project properties and browse to C/C++ Build > Settings, change the following (or make sure they're correct):
In ARM Sourcery Windows GCC Assembler -> Miscellaneous
Specify this for Other flags option: -c -fmessage-length=0
In ARM Sourcery Windows GCC C Compiler -> Directories, the following include paths (-I) should be listed
StellarisWare
StellarisWare/driverlib
StellarisWare/utils
In ARM Sourcery Windows GCC C Linker -> General
For script file (-T), specify the main.ld that you copied from Martin's project
Check Remove unused sections (-Xlinker --gc-sections)
Uncheck everything else
In ARM Sourcery Windows GCC C Linker -> Libraries
Add "driver" under the top libraries (-l) section
Add StellarisWare and StellarisWare/driverlib/gcc to library search path (-L)
In ARM Sourcery Windows GCC C Linker -> Miscellaneous
Set other flags option to this: --entry ResetISR
Add a new C file to your project, call it something like HelloWorld.c and paste the following in it:
#include "inc/lm3s9b92.h" #include "inc/hw_i2c.h" #include "oled.h" void SysTickHandler() {} int main(void) { init_display(); dispString("Hello World!", 0, 0); while (1) { } }
If it compiles without errors, that's great news!
The final and most important step is the flashing part. We'll be using LMFlash to download the application to the Evalbot. Start off by creating a new text file called manual_build.bat and paste the following in it:
@echo off echo Ready to begin flashing. echo. pause echo. cd Debug arm-none-eabi-objcopy -O binary HelloBot.elf HelloBot.bin "C:\Program Files\Texas Instruments\Stellaris\LM Flash Programmer\LMFlash.exe" --quick-set=ek-lm3s9b92 --verify --reset "%CD%\HelloBot.bin" echo. echo. echo Flashing done! echo. pause
Now, of course, you do need to change the paths to fit your environment, including HelloBot.* and the path to LMFlash.exe.
This batch file is mainly useful for problematic flashing, hence the manual-part.
Create a similar batch file to be able to flash it after compiling, automatically. Call it something like build.bat:
@echo off cd Debug arm-none-eabi-objcopy -O binary HelloBot.elf HelloBot.bin "C:\Program Files\Texas Instruments\Stellaris\LM Flash Programmer\LMFlash.exe" --quick-set=ek-lm3s9b92 --verify --reset "%CD%\HelloBot.bin"
OPTIONAL: If you want to, you can just add the full path to the build.bat file to Eclipse projects post-build, so after building it'll automatically start flashing, which is what I like to do.
You can now plug the Evalbot in with the USB cable provided (use the ICSI port on the board), turn the power on (it might require battery power even though it's plugged in) and install the drivers. You can find the drivers in the StellarisWare directory.
Since you've propably already built the HelloWorld project, there should be a Debug directory containing the ELF file. If so, go ahead and run manual_build.bat and see the magic happen.
If all goes well you've got your first compiled application running on your Evalbot.
Congratulations!
EDIT: Still looking at debugging options, might be able to use arm-none-eabi-gdb --eval-command "target remote COM1" to connect to the board and debug, but not sure yet, will keep trying.