Texas Instruments L137 project

Here are two source files to start with main.c 6kBytes and L137_gpio.c 9kBytes. Compare the Spectrum kit’s setting or reading a bit—painful to follow via a debugger of deeply nested calls when a pointer will do. We take the minimum #defines and test a single file, then split them up later into more manageable include files and source files. We do not like to include all in a “land grab” of what you might encounter.

Spectrum Digital L137 board

The Texas Instruments OMAP-L137/ TMS320C6747 processor has a 300 MHz DSP and 300 MHz ARM9 core. The board was designed by Spectrum Digital. We wanted to develop software for an ARM processor with built-in Ethernet and a JTAG port on-board. The Spectrum board was reasonably priced and the compiler was not code size limited. There was not that much supplied with the board either; what about the Micrel switch? And how does the DSP fit in with the promised MontaVista Linux kernel?


We started with the examples and then merged several files to incrementally run all the examples in the test directory. Most of the peripherals were tested, however, the code was running on the DSP, not the ARM. On the L137 processor, the DSP comes up first. On the OMAP3, the ARM comes up first. We also never managed to get the serial port running, but did not spend a lot of effort on it. The habit of both silicon vendors and board support folk placing copyright notices on the header files for names of registers is a tad silly. Must we all go and retype the stuff in from the databook? We left their notices in and like the Linux community, went and stomped all over the code. We wanted shorter names and put in the full pointer access as it was an editing “copy and paste” exercise. Use as you see fit, but we are not claiming any copyright protection. The other peripherals used the Spectrum Digital test subdirectories and added some extra source, however, we will not be sanitizing these.

The strange references in the comments are TeX commands for BibTeX, and Doxygen. We started trying to cross-reference documentation in 2009 but were side tracked by other more interesting stuff—like a BeagleBoard. This evaluation was done in the evenings from September to December in 2009 while waiting for work during the Global Financial Circus at an automation company.