GCC?

[juancferrer]

Any info on getting GCC to compile and program for this thing?

[senatorpenguin]

The instructions on building the Codesourcery ARM toolchain worked for me from claymore.engineer.gvsu.edu/egr326/LM3S6965
The section Tips & Tricks on modifying the FTDI drivers for Mac are also important.
Once I clean up my code, I'll post on how to use the bootloader and Stellaris libraries to develop using autotools, and also how to use the peripherals on the Evalbot.

[jethroevb]

During the years, I have worked on projects that use many different processor architectures from many different vendors. I've learned not to taint my main computing environment with different compiler versions and toolchains.

Instead, when presented with a new architecture, what I do first is, I download and configure a fresh pre-made VMWare virtual appliance of a Linux distribution. CLI only, no GUI is needed. I prefer Debian but other distributions work as well.

www.vmware.com/appliances/


Secondly, if it's an ARM architecture like the EvalBot, I download CodeSourcery GCC lite edition and install it in the virtual machine:

www.codesourcery.com/sgpp/lite/arm


Then, I set up the virtual machine to have access to project directories on my host computer. I then use native IDE software (XCode in my case) running on the host machine to edit files and keep track of the revision history.

When it is time to compile something, I log into the virtual machine using SSH and change to the project directory that is actually residing on the host side. SSH in the virtual machine is configured for passwordless login (key file). Program upload to target device (and debugging) can be handed by native software (on the host) or virtual software (in the virtual machine) as long as it's a USB peripheral.

The benefit is, that I can keep several virtual machines with tools and libraries that I know to be working. Still, I also get to use a native IDE and version handling tool that are common for all projects. When developing for wildy different architectures, vendor-provided IDEs are more trouble than they're worth.

[imlost]

I am attempting to test full Mingw GCC port on a Windows host:

opensource.zylin.com/gccbinary.html

The only issue they state is a floating point compile issue for the Cortex M3.

They also offer MIPS/ARM/PPC tool-chains for MacOS PPC/x86, Linux, and Windows.

[artvandelay]

I'm mainly following the instructions on:

www.synthetos.com/blog/

(Scroll down to entry from 12/5/2010)

This mainly walks you through setup of the CodeSourcery GCC lite edition that jethroevb mentioned. Setting it up in a VM sounds like a great idea; in fact, I thought about doing that. Hopefully I won't regret dumping it in my main OS X system :-)

I have the cross-compiler up and running, and at the moment I'm building the StellarisWare examples from TI (download link for that code is on the blog I mentioned above).

[artvandelay]

Alright, finally got my toolchain working (can compile StellarisWare examples) and OpenOCD running and communicating with the board. I was able to dump the Flash image successfully.

Time for bed; hopefully tomorrow I can get some code flashed onto the board.

[chemuduguntar]

Jan 22, 2011 1:52:56 GMT -5 artvandelay said:

Alright, finally got my toolchain working (can compile StellarisWare examples) and OpenOCD running and communicating with the board. I was able to dump the Flash image successfully.

Time for bed; hopefully tomorrow I can get some code flashed onto the board.

x2

it's 7am here, time for bed

[imlost]

#even after checking the build setup the Win gcc Binaries borked building the libs
arm-elf-gcc -print-search-dirs

=(

[ckanode]

Hmm... maybe someone can help. I'll still do research and reading to try to find the answer myself. I am a newbie. I've done some C/C++ programming, but that's been with IDEs where the environment is set up for me, so I've had little experience actually setting up my own environment.

I'm on Windows XP (I do hope to set up VMs like jethroevb has suggested) and using Cygwin.
C:/StellarisWare/
C:/Evalbot/ - project folder

Building the examples in the StellarisWare folder was successful using make. However, when I try to compile in my project folder, I am having file path issues. I used martinhubacek's MakeFile as a model for constructing mine. It fails on trying to find the inc folder in StellarisWare.

Next thing that I am going to try is to move my project folder into the StellarisWare folder. I don't want to, but if I can get it to work... I just want to start programming the bot.