Mon 09 Sep 2013 03:33:12 PM UTC, comment #9:
Regarding my remark in comment #8, here is the respective
part of the SVN log:
r718 | joerg_wunsch | 2007-01-11 21:20:25 +0100 (Do, 11. Jan 2007) | 10 Zeilen
- butterfly.c (butterfly_initialize): when sending the 'T'
command (which is ignored by current AVR109 bootloaders),
send the first reply from the list of supported device
codes back rather than using avrdude.conf's idea about
an AVR910 device code. Apparently, this solves disagreements
between different versions of at least the ATmega8 AVR910
device code.
Closes bug #18727: Writing flash failed
Given the date of the change, the first AVRDUDE release
affected by it is 5.4. This would match your observation
that 5.2 is working fine.
|
Sun 08 Sep 2013 09:07:35 PM UTC, comment #8:
One discrepancy immediately catches attention:
The "OK" log says:
The "Failed" log says:
Given the large number of device codes that are supported by
the programming hardware, I really doubt this is actually a
bootloader. (Bootloaders usually support exactly one device,
the device they are operating on.)
Did you try with both, -c butterfly (aka. -c avr109), as well
as -c avr910? While both implementations share the same
roots, their behaviour is not exactly the same.
One of the consequences of this might be that in the working
log, we can see:
i.e. the device code 0x5e is chose, which is the correct one
for an ATtiny2313. In the "failed" log, we see:
i.e. it picks device code 0x13 which belongs to an AT90S1200.
This is done since the bootloader implementation just picks
the very first device code it can see, since many bootloaders
simply don't spend much attention on the device codes they
are claiming to support, so picking the first one is usually
the best choice.
However, for a standalone programmer, there are subtle
differences in the programming algorithm between AVR's
grandfather AT90S1200, and the ATtiny2313, which might explain
the failure.
|
Sun 08 Sep 2013 04:06:30 PM UTC, comment #7:
Sorry for being late. Just posting the requested log files.
kind regards
Christian
(file #29048, file #29049)
|
Tue 03 Sep 2013 12:20:33 PM UTC, comment #6:
See comment #2: please get us a full -vvvv log of both
versions.
Yes, the avrdude.conf format has changed in the current version
(and that's the major reason why the next version will be
called 6.0). Remember that you can always specify an
avrdude.conf file that matches your binary using -C, e.g.
Note that bug #38965 has been closed upon request by its
originator, since he couldn't see the bug anymore.
|
Sun 30 Dec 2012 01:14:05 PM UTC, comment #5:
I noticed that the failing attempts with versions 5.11.xyz seem to flash much slower than the succeeded attempt with version 5.2 so my guess: Is this issue related to https://savannah.nongnu.org/bugs/?37965?
I don't know how to use stty to configure serial port at the moment so can't evaluate.
|
Sun 30 Dec 2012 12:49:37 PM UTC, comment #4:
Here's my command I used to make the flashing:
avrdude -p t2313 -P /dev/ttyUSB0 -c avr109 -e -U flash:w:ornamatrix.hex -b 115200
Using the last trunk from SVN repository does not even start anything: "avrdude: error at /usr/local/etc/avrdude.conf:297: programmer type must be written as "id_type"". So it seems it does not come around with the "old style" config file. Using the avrdude.cong from latest trunk it worked to me axactly the same way as avrdude version 5.11.1 -- it fails with the same output (as it seems to me).
|
Sat 29 Dec 2012 08:56:11 PM UTC, comment #3:
At first sorry for my german bug report.
I connected theflasher again and flashed using both versions of avrdude, the outputs of each are attached.
And no, I didn't try the latest svn version but I'll do and report the result here soon.
Thanks so far!
Christian
(file #27163, file #27164)
|
Sat 29 Dec 2012 08:28:07 PM UTC, comment #2:
Did you try the current svn version? Maybe the bug is already fixed.
|
Sat 29 Dec 2012 05:39:44 PM UTC, comment #1:
Communication language here is commonly English, as this is
an internation effort.
Please submit more details:
. what kind of programmer is it (the -c option)
. the complete commandline
. the log of both versions run with -vvvv (four times option
"-v"); note that all trace data go to stderr, so you have
to redirect FD 2: "avrdude ... 2> logfile.txt"; please
provide these logfiles as an attachment here
|
Sat 29 Dec 2012 04:24:43 PM UTC, original submission:
Moin moin allerseits,
ich möchte auf einen Bug in der (in den aktuellen -- 29.12.2012 -- Ubuntu-Repos vorhandenen) Version 5.11.1 von avrdude hinweisen:
Auf dem 29C3 des Chaos Computer Clubs im CCH in Hamburg habe ich von blinkenarea.org eine kleine Schaltung "OrnaMatrix" (http://wiki.blinkenarea.org/index.php/OrnaMatrix) und einen kleinen ISP vom gleichen Entwickler erstanden. Nach anfänglichen Schwierigkeiten konnten wir sicher stellen, dass keine Hardwarefehler vorhanden sind.
Die avrdude Version 5.11.1 auf meinem LinuxMint13 war nicht in der Lage, den ATTiny2313 mit einem neuen HEX-File zu versorgen. Erst nachdem ich aus dem Downloadbereich die Version 5.2 des avrdude geladen, configured, make, make installed habe, konnte ich den Flashvorgang erfolgreich durchführen.
Ich bin mir nicht sicher, welche weiteren Informationen zur Fehlersuche benötigt werden, stehe also gerne für Fragen zur Verfügung.
Gruß
Christian Pohl
|