Dirk Behme wrote:
> Additionally, short status update from me:
>
> As pointed out by Kees [1], the 'do always 7 clocks tap_get_tms_path()'
> might not be flexible enough for OMAP. So I played a little with
> Holger's jtag_move_to() [2] and used it to hack a new tap_get_tms_path()
> (see patch in attachment for recent status). Unfortunately, this doesn't
> work correctly yet:
>
> Adding some debugging output to original version of OpenOCD and the
> hacked one with 'new' tap_get_tms_path() I get the output below. Have to
> investigate why new version fails with 'Could not validate end ...'.
Had a look to it and now an additional 'please explain' question ;)
What is the idea behind jtag_validate_chain()? Mainly why does it
check the value returned from ir scan for 0x1 and 0x3:
if (val != 0x1)
if (val != 0x3)
?
Having a 6 bit IR, in this function ir_test is set to 8 bit length and
filled with all 1. Then a jtag_add_plain_ir_scan() is done and the
returned value is 0xC1, passing the two above tests. Why is returned
value 0xC1 ('1100 0001') and what does this mean? I.e. what is the
idea behind these checks?
I ask because with Holger's modified tap_get_tms_path() which doesn't
assume any more that always 7 clocks / TMS steps are done, I get back
0x41 and the checks in jtag_validate_chain() are failing (the 0x3 at
pos 6 check).
To fix this, I'd like to understand the idea behind
jtag_validate_chain() :)
Most probably the issue is with this TMS manipulation
BUFFER_ADD = tap_get_tms_path( tap_get_state(), tap_get_end_state() )
| (last_bit << 7);
in ft2232.c which most probably only works if always 7 clocks are done
for each state change (?). And this isn't the case any more with
improved ap_get_tms_path() from Holger.
Thanks for the help and best regards
Dirk
[1]
https://lists.berlios.de/pipermail/openocd-development/2009-February/004773.html
[2]
https://lists.berlios.de/pipermail/openocd-development/2009-February/004736.html
_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development
