Just a progress update. I believe I know why the Linux USB driver isn't working on the Nspire.

Hidden in a corner of the documentation for the USB controller lies this:

However, the queue heads (dQH) must be placed in on-chip RAM. A design limita- tion on burst size does not allow the queue heads to be placed in DRAM.

Basically, I think this means that queue heads must go in the SRAM. Linux isn't aware of this and is handing out DMA addresses that point to the SDRAM. Unfortunately there doesn't seem to be an obvious way to tell Linux to hand out SRAM DMA addresses to the USB driver.

In other words, this sucks because I know what's wrong but not actually sure how to solve this problem. :\

I assume you can install new stuff with those commands under Linux:
mkdir /mnt/initrd
mount -o loop initrd.tns /mnt/initrdThen you would throw additional ARM-compiled binaries in /mnt/initrd/usr/bin. That or you can always generate a new image with buildroot with the stuff you want on it, just take care the resulting image size will not exceed the Nspire memory.

Thanks, I think ill try what Juju suggested, it seems safer.

I have basically no ability to build a new image, since I have only basic skill with linux.

Hopefully, someone would be willing to post a pre-made image with the stuff that Lionel Debroux put in his config.

Anyway, what is the limit of the Nspire CX CAS's memory? I'd rather not accidentally kill my calculator.

Good, so linux is effectively in a sandbox, and can't mess with the regular TI-OS image/bootloader.

There is so much potential for this project. Sorry to keep bugging you guys for info

Ah well, one more newbie question, after I mount and put my stuff onto the image, do I just unmount it and my changes to it will be saved?

Clickpad & Touchpad models are not supported yet.

Told ya gcc would be pretty heavyweight. When you download the tar.gz it's something like 80MB.

Pretty small: http://bellard.org/tcc/