[avrdude-dev] EDBG is up for testing

[Joerg Wunsch]

Hello everyone,

implementation of the Atmel EDBG protocol is mostly done now.  All is
in SVN, so anyone who wants (and can) is invited to test it.

Basically, the protocol is a wrapper around the old existing JTAGICE3
protocol, so all the internals have not been touched. (*) The fact the
debugger is using EDBG protocol is detected by the occurence of the
substring "CMSIS-DAP" in the USB product string, as this is mandated
by the CMSIS-DAP document.

(*) There's one difference, the SPI_MULTI command which AVRDUDE is so
fond of is no longer supported in ISP mode.  I have yet to verify
whether we would still want to use it somehow.

In order to allow for the JTAGICE3 with both, the old V2.x (non-EDBG)
as well as the 3.x (EDBG) firmware to be detected, I extended the
"usbpid" key in programmer definitions in the config file to allow a
list of USB PIDs.  However, even though it's syntactically allowed,
most programmer implementations ignore the list and only take the
first USB PID that has been given in the file; only the "jtag3"
programmer implementation has been extended to really handle the list.
Each given USB PID is then probed successively, until a match was
found (or none at all).  That way, one can always specify "-c jtag3"
on the commandline, regardless of the firmware version (and thus
different USB PID, 0x2110 vs. 0x2140).

Since the protocol is just the same, the new debug dongle called
"Atmel-ICE" (**) is also supported that way.  (This refers to the AVR
part, obviously.  Regarding the ARM part, I'd hope that a CMSIS-DAP
aware OpenOCD would be able to talk to that.)

(**) http://store.atmel.com/PartDetail.aspx?q=p:10500375;c:100112#tc:description

Also supported is the onboard EDBG chip that can be found on the new
"Xplained Pro" evaluation boards.

If all goes well, I intend to roll a release anytime soon.

Not yet done: I didn't do any attempt to implement a Win32 variant
that uses the Win32 API HID library.  By now, only the libusb backend
is done.  If anyone wants to step forward and attempting an
implementation based on the Win32 HID API, they are most welcome.  The
advantage for Windows users there is that they don't need a separate
driver.  (Windows doesn't offer a real generic USB driver as the
various Unices do.)
-- 
cheers, Joerg               .-.-.   --... ...--   -.. .  DL8DTL

http://www.sax.de/~joerg/
Never trust an operating system you don't have sources for. ;-)