Tue 18 Dec 2012 04:20:41 PM UTC, comment #9:
That seems to have fixed it. r1126 works for me with Dragon firmware version 7.24.
By the way, your JTAGICE3 patches work for me as well, with that same xmega128a3(u).
|
Tue 18 Dec 2012 09:21:48 AM UTC, comment #8:
I think I found a suitable workaround: for firmware >= 7.x,
and Xmega devices, don't trigger a RESET during initialization.
Please test the SVN version again, and thanks for reporting
it in the first place!
|
Tue 18 Dec 2012 08:39:33 AM UTC, comment #7:
Yes, it appears to be a genuine firmware bug (because these
response codes are completely out of context). After
upgrading my Dragon to 7.24, I see the same.
Since earlier AVRDUDE versions could work around the bug,
I'll see what I can do about it.
|
Mon 17 Dec 2012 11:44:57 PM UTC, comment #6:
I downgraded my AVR Dragon firmware using avr32studio 2.5. It's now at FW version 6.11:
JTAG ICE mkII sign-on message:
Communications protocol version: 1
M_MCU:
boot-loader FW version: 255
firmware version: 6.11
hardware version: 1
S_MCU:
boot-loader FW version: 255
firmware version: 6.11
hardware version: 6
Serial number: 00:a2:00:00:40:28
Device ID: AVRDRAGON
I can now program the part with both the release and svn versions of AVRDUDE.
Pretty strange. I'd love to keep helping solve the problem, if you know what the next steps should be.
|
Mon 17 Dec 2012 10:51:19 PM UTC, comment #5:
> I'm not sure how to, but I'll spend some time tonight figuring it out.
I'm afraid you have to install an older version of AVR Studio
for this. I think firmware 7.14 (7.e in Atmel's notation)
correlates to some Studio 5.x version.
Thanks for your effort, the error picture is really puzzling.
|
Mon 17 Dec 2012 10:16:44 PM UTC, comment #4:
I'm not sure how to, but I'll spend some time tonight figuring it out. I'm certainly willing to.
This Dragon, with this firmware revision, also works fine with Atmel's avr32program command in Linux, and with Atmel Studio 6 in Windows.
I'm happy to help debug this. Let me know any other information you need me to gather.
|
Mon 17 Dec 2012 09:37:53 PM UTC, comment #3:
That firmware's behaviour is really strange, but I noticed you've also
got a earlier hardware revision than me (6 vs. 7), no idea whether
that makes a notable difference. The result codes are really
unexplainable. 0x84 in response to "program enable" is the result
code for "read PC", but it rather starts that your Dragon is not
generating the expected "break" event after the "program enable".
Likewise, the response 0x80 to the "read memory" command (for the
signature) is quite odd. 0x80 is just "RSP_OK", but of course, a
"read memory" command expects "RSP_MEMORY" (0x82) in return.
Curious, do you have the chance to downgrade the Dragon's firmware to
an earlier version?
|
Mon 17 Dec 2012 09:03:32 PM UTC, comment #2:
Thanks for the log file. Looks really strange.
Alas, I cannot reproduce it with my Dragon (but my firmware is 7.14,
yours is 7.24). Can you please also mail me the -vvvv log from the
old (working) version, for a crosscheck?
|
Mon 17 Dec 2012 07:33:46 PM UTC, comment #1:
Please repeat the failing command, and add -vvvv (four times
option "v") to the commandline, so we can see what exactly
is failing.
|
Mon 17 Dec 2012 07:29:40 PM UTC, original submission:
Programming with the AVR Dragon in JTAG mode works for me with AVRDUDE 5.11.1, but doesn't work with the latest version in SVN (r1124).
I've attached a log showing the error messages. In the log, '/usr/bin/avrdude' is the released version 5.11.1 that comes with Ubuntu 12.10, and 'avrdude' in the PATH is SVN r1124.
|