Hallo Steve, Am Sun, 30 Apr 2006 15:25:52 +0800 schrieb Steve Underwood: > > Are there any USB experts out there who would be willing to help get USB > FET tool support working with Linux?
I'm not a real expert, but someone who had to deal with an USB driver for booting und debugging self developed MSP430 boards through a TUSB3410 in 2004 (for 2.4.x kernels). An EZ430-F2013 should be on the way to me (as the TI's sales autoresponder told to me), and I will be in need to program the new F20xx processors out of my linux box within the next few months. > The standard Linux distribution includes a driver for the TUSB3410 USB > chip used in the TI USB FET tool. However,this driver is only designed > for the generic use of the TUSB3410 chip as a USB to serial interface. Yes, I took a short look on it some weeks ago, because my customer asked me for a 2.6-kernel port of the old driver... ;-( > In the FET tool the USB chip has been given a unique USB ID. This is no > problem, as additional IDs can be added from the command line of the > existing Linux driver. > > In the FET tool the USB chip has two endpoints, while as a generic USB > to serial interface it has only one. This is more of a problem, as I am > not a USB expert, and I don't know how to handle these things. The > current Linux TUSB3410 driver won't even load firmware into the device. > > It feels like it should not be a big problem to adapt the current Linux > driver, to make a FET tool driver. However, learning enough about USB to > make this work nicely will probably take me a while. Is there anyone > around prepared to help shortcut that? We do not need any standard IOCTL (that's the part that makes things really 'unhandy' on serial drivers) and could go with our own simple solution for the control channel. What's to do on my side is the port of the existing driver from 2.4 to 2.6 kernel code - a not so easy thing, because Greg Kroah-Hartman changed a lot within the usb code... ;-( You can download my existing code from <ftp://www.ibrro.de/ftp/bmls-kernel-driver.tgz> to have a look on it. I appologize for some german comments within the sources (I am native german speaker with not so much experience in the english language), but I hope you can go with the other parts that are commented with my special pidgin-english... ;-) The driver is triggered by the initial TUSB3410 vendor/product-ID. It tries to upload a firmware code to the TUSB and then to do a reconnect to the just uploaded firmware by using private IDs. This all is explained within the above given .tgz. You can find an example firmware at <ftp://www.ibrro.de/ftp/tusb3410-firmware.tgz>. I've tried to port the firmware loader part to the new 2.6-kernel some weeks ago, but I didn't succeed within a few hours because things in the kernel have changed so much. Should I try again? > Once a driver is available allowing communication between the > application and the FET tool, the rest should be easy. There appear no > significant hurdles to getting the existing Windows USB FET tool > software to work after communication is established. The firmware loader part could help a lot to develop the FET tool because this FET part of development could be done away from the kernel. The above given firmware example is written with the 'sdcc' and converted to a loadable firmware file with some (very) small tools (all included in the .tgz with their sources... :-). Do you think we should try to do the work together? I do not have much time within the next months, and even less time during the next few weeks. But I'm very interested to get a working FET tool on my linux box... ...and, again, I really have very little experience with the english language. Best regards, Bodo
