jtag_avr_spy.doc

AVR JTAG Software Debug Tools

Documentation

The jtag avr spy dumps the JTAG ICE traffic to RS232 in
an array.

The JTAG is nothing else then a synchronous protocol
like a shift register (e.g. 74HCT164) : the TAP controller 
in this case is a bit more intelligent. 
This means that in order to read data you have to clock in
(= write) data with the JTAG clk. So you have to create a 2
way communication.

The TAP controller uses a state machine to guide the data into
right register set. The instruction register can be seen
as an address decoder to address the right register set.
The data register shifts the data into and out of this
selected register set.

I used some letters to clarify that the TAP controller
is writing the instruction/data register and which data
is to be read or written.

I made an assumption that the pause state - mentioned in the
datasheet - willnot be reached. In that case I used a
question mark.


How is it organised ?

Spy tool

Every time a new sequence has been started up then the jtag
spy outputs in the first column a letter : I for instruction
register and D for data register.

In the second column there will be an I for data to be shifted
in and O for data to be shifted out of the selected register. This
happens at the same time.
The jtag spy immediately outputs the hexadecimal byte without any
spaces.

Data into an instruction register cannot be higher than 4 bits 
(seems to be a requirement of the JTAG).
This is a jtag reference point that there is just one hexadecimal
byte. So you can see that there is just one command to get or set
a value/register/data/memory.

This tool resides into the AVR ATmega16 but with some minor changes
you can run it on any avr.


Host tool

Theodore Roth wrote a complete set to initiate the JTAG ICE box : the
jtagice-0.0.2.tar.gz. This excellent tool complements the jtag spy
to what it is. You can read more details in the software.


The result

Normally spoken you have to use 2 serial ports but for some reason it
doesnot work very well on my machine. So I had to use 2 GNU/Linux systems :
one on which I captured the output of the spy tool and one to run the
tool of Ted.
I captured some of the results with redirection of the /dev/ttySx to text
(as ofcourse the unix way, thanks Karl) so I can show you some results
but not all of it.
The way it si done

I did some analysis on the results : you can find this in the result
directory.

Armand