Re: [avr-chat] Avrdude and JTAG / Dragon...

[David VanHorn]

> Well, the debugging models of AVR Studio (which the JTAG ICE and
> Dragon are tuned for) and GDB are vastly different, and AVaRICE isn't
> really the best optimized tool ever, so it ends up causing heavy JTAG
> traffic if you're doing something unlucky (which already a single step
> at the C level can account for).  It's usually not much of a problem
> though if you ensure to keep the AVR going for a longer period of
> time.  I tend to set many temporary breakpoints, rather than
> single-stepping.

Something to watch for..
In the data sheet, they mention that the jtag breakpointing is done by
reprogramming a register.
I read this a while back, probably in the M128 data sheet, that the
registers used for this are similar to program memory, in that they
have a short, finite lifetime.  I've had some wierd issues in the past
with debugging that may have been caused by wearout of this register.

I had a look, and I can't put my finger on this ref.. Maybe someone
else remembers?