Recent AVRDUDE unhappy with efuse values in atmega328_isp targets

[GoogleCodeExporter]

The EFUSE byte on m328 and some other chips only has three bits that are 
actually used, and there seems to have been some disagreement on how to handle 
the unused bits so that "verify" operations on the fuse work correctly.
The original makefiles claim that avrdude wanted these bits specified as ones, 
so that an EFUSE setting of xxxxx101 became FD.  The current avrdude wants them 
to be zero (05 in the makefile.)
Interestingly, other chips have different settings (apparently in the 
avrdude.conf file.)  DF still works on 1284, for example.

Also check atmega328_pro8 settings.

Original issue reported on code.google.com by wes...@gmail.com on 30 Jun 2014 at 1:55

[GoogleCodeExporter]

The issue is caused by the avrdude.conf config file.
It doesn't treat the unused/reserved bits consistently
across all the parts.
I found some avrdude bug reports on this that go back 6 years so it is not
a newly discovered issue.
Obviously it isn't being perceived as a priority to fix.
https://savannah.nongnu.org/bugs/?22248
https://savannah.nongnu.org/bugs/?21954

It is such a simple fix,
I'm actually quite surprised it hasn't been fixed in all this time.
Maybe we should ping Joerg and see if we can elevate fixing this?

Problem is, it has existed for so long, that depending on it is fixed,
certain things may break since anybody using avrdude is currently doing work 
arounds to make it work using the existing config file.

Original comment by bperry...@gmail.com on 30 Jun 2014 at 3:40