========
duane> # Example: Declare a chip, for example some Xilinx FPGA
duane> jtag newtap xilinx -chip xilinx -ir-length 6 -ir-capture
???? -ir-mask ????
rick> [I don't like chip]
duane> I prefer "chip" for one reason. Physically that is what is
present on the board.
rick> That's a pretty narrow focus for the designator. If someone had 2
boards that shared a JTAG chain,
I think multi-board _today_ would be a very rarefied event in the world
that OpenOCD lives in.
Earlier in the life of OpenOCD - it was *RARE* to see multi-tap chips
(+2 years ago) Today - it is more common and in fact to watch the
questions on the list - many ARM926 chips are multi-tap. It is quite
confusing to users. And in some cases - I have seen a rare mention of
multi-chip chains. That however is understandable - you have a schematic
in front of you. I have *NEVER* seen an actual hardware that has
multi-board level jtag chains. I suspect - if they exist - it is for
system level board interconnect testing. I suspect
"board-level-interconnect' is rare out side of a "factory or very very
high end" configuration.
OpenOCD just does not live in that arena.
For that reason, I think today - "chip" is sufficient.
========
duane> [about the TAP syntax] What if we changed the syntax like this:
duane> jtag create CHIPNAME TAPNAME .... options ...
[snip]
rick> What namespace are the chips and taps put into?
Per your suggest, chip & tap remain a parameter to the "jtag configure"
command.
rick> What does target creation look like?
Target creation - today uses the "--chain-position NUMBER",
rick> Do they need to specify the chip and tap as two separate arguments?
Good question. What about this: Put a "dot" between the two - it is
sort of like TK path names.
It would become: "--chain-position CHIP.TAP"
====================
duane> [about name spaces]
rick> [about name spaces]
Agreed - Namespace is becoming a bigger issue.
Thus - per your suggestion, all "tap" based things require either (1)
the "jtag" command prefix or (2) a "path-like" name.
====================
BTW - this back and forth is very helpful - Thank you.
Sometimes my head gets wound up in places it should not be!
-Duane.
_______________________________________________
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development
