Re: [avrdude-dev] [bug #40831] LUFA AVRISP-MKII fails with avrdude 6.0.1

[Joerg Wunsch]

As Colin O'Flynn wrote:

> The only difference is I think the STK600 has an interrupt
> endpoint (this is from memory so is frequently wrong), so need some
> additional detection code. Will look into the JTAGICE3 as well.

No, the STK600 is close to the AVRISPmkII, only that it adds JTAG
and HV programming abilities.

It's the JTAGICE3 that started to have interrupt endpoints: in
firmware version 2, there are two bulk EPs for the normal traffic,
and one interrupt EP to report events.

Alas, in firmware version 3 (as well as with the new Atmel-ICE), due
to switching to CMSIS-DAP, there are two interrupt EPs for everything,
and an additional protocol on top to distinguish between the various
packet purposes (command, response, event).  All in all, this protocol
is much crappier (among other things, even the request to get a
response, which itself is just a single byte, always transfers full
512 bytes!).  Welcome to the wonderful world of MS Windows
compatibility, where every device on the USB is a "human interface",
just since HID used to be one of the few things Windows always has a
generic driver for, so you don't need vendor drivers.  (In contrast,
on all Unix-like systems, there are generic drivers for every USB
device, so libusb never needs a custom driver.)

As a consequence, I would recommend to anyone who can afford it to
keep their JTAGICE3 firmware at version 2.x.  But it won't work
together with recent versions of Atmel Studio then.

The CMSIS-DAP compliant devices are mandated to have the string
"CMSIS-DAP" as part of their product string, so this could be used
to uniquely identify that situation when trying to detect the EPs.
-- 
cheers, Joerg               .-.-.   --... ...--   -.. .  DL8DTL

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