Re: [avrdude-dev] [patch #4020] Avrdude Safemode upgrade

[Brian Dean]

On Sat, Sep 17, 2005 at 08:40:28PM -0400, Brian Dean wrote:

> Update of patch #4020 (project avrdude):
> 
>                   Status:                    None => In Progress            
>              Assigned to:                    None => c_oflynn               
> 
>     _______________________________________________________
> 
> Follow-up Comment #1:
> 
> Colin, please check on this patch and verify whether it should go in
> or not.  I think this implements a reasonable set of conditions for
> enabling/disabling safemode, so if you are still OK with it, it's
> fine by me.

Another possbility, and I can't remember whether this was already
discussed or not, so if so please forgive me if it was - check the
memories being programmed and if any of them are "fuse" bytes, then
disable safemode.  That would make safemode transparent instead of
requiring one to specify -u on the command line in order to tell
avrdude that it is OK to update fuses.

Even better, would be to leave safemode enabled only for fuse bytes
that were not specified on the command line.  That way, the fuse bytes
that one intended to program can be while the others would remain
protected.  Same should hold for interactive mode - if a fuse byte is
modified there, subsequently disable safemode for that fuse byte.

This could be implemented by using a bit mask with a bit representing
the high, low, and extended fuse bytes.  As a fuse byte is programmed,
disable its bit in the mask.  When the program is ready to exit, see
which bits are still set and verify their contents against what was
read at the beginning.  Basically safemode would still operate as
intended, but would allow reprogramming of fuse bytes without having
to tell avrdude that it is OK first by specifying the -u flag.

-Brian
-- 
Brian Dean
ATmega128 based MAVRIC controllers
http://www.bdmicro.com/