25-09-2007, Grant Edwards: > On 2007-09-25, David Pettit <[email protected]> wrote: >> >>> I also CCed patches here, hoping that there will be some testing and feed >>> back, but... So, if TI EZ430 USB tool is used by more people, i can start >>> effort again. >> >> I think your efforts would be appreciated. I now have three neatly stacked >> in my desk drawer at work (vendor reps love to give gifts). If I can use >> them with Linux, I will take them home with me.
> The stock driver in the 2.6.22 kernel works fine along with the > udev rule that sets the configurationValue to 2. To use this thing was possible right after driver was written (~2.6.11) of course. Question is "how to use". Because they accepted just one very tiny patch to add device ID, now udev rules are a bit smaller. But i want to use the device easily and in Debian. I want to have easily changeable firmware, because windows drivers have newer version of it (and you better don't see ugliness of the sources of the vendor's firmware :). And there are people out there, who wrote their own firmware, because usb chip itself have many additional features. The main problem is linux usb stack, as it is today, after 3 rewrites. Design prohibits driver to change usb configuration of the device. This is because usb driver in linux is not driver of the usb hardware, but usb-interface (as in usb spec.) driver. Changing configuration of the device, cuts trunk of the "driver", and everything falls down. Current usb gate-keeper played dumb with me, when i was trying to discuss this fatal design flaw. I was flamed, but whatever. <http://article.gmane.org/gmane.linux.usb.devel/50656> Those programmers, who don't like to be in touch with hardware don't deserve my angriness. The USB chips by TI are very "interesting" and have many "features". At first view, if compare FTDI USB and TI USB, latter have much more additional external components. I thought, this is rather complicated. But complication from programming point of view led to all the f*cks with linux usb stack. External firmware is abother very interesting topic in it. Original hardware design have an serial eeprom interface to load (by bootloader) application firmware off-chip. *But* Timings of the loading are too big to pass usb spec on how quickly device must set itself up. Thus, device have additional eeprom chip near it, but it is rather useless, just to store *few* bytes of customer's IDs. Patented bootstraping with ID for bootloader, switching to firmware with new ID is "another funky thing". Aaarrhh. Software and softhardware sucks, didn't you know ?:) -- -o--=O`C #oo'L O <___=E M
