[Openocd-development] OpenOCD Cortex-A8 / S5PC100 support...?

[Nick Pelling]

Hi everyone,

I've just plugged my trusty old Olimex ARM-USB-OCD (the Swiss Army
Knife of JTAG debuggers) into my shiny new S5PC100-based board and...
I'm struggling to see how to get it working even slightly.

Has anyone got anywhere with the s5pc100 or other CoreSight-based
Cortex A8s (apart from the OMAP3530 as per
http://elinux.org/BeagleBoardOpenOCD )? Firas Achkar mentioned trying
this on the urbetter board (which I think is actually the original
ODroid) a month ago, as did Matt Hsu in November 2009. But trying the
latest openocd's autoprobe (as David Brownell suggested then) didn't
seem to produce anything so revealing as an ID code.

Incidentally, I've read through the CoreSight chapter in the latest
Samsung datasheet but that had precisely nothing so useful as a TAP
ID code or a DAP ID code to get started. Stepping through every
occurrence of JTAG also revealed nothing useful.

Any suggestions? Has anyone tried the same thing for the S5PC110 (as
per the ODroid-T etc)? Might there be some hidden nest of Samsung SoC
OpenOCD developers in a far-flung corner of Korea?

Cheers, ....Nick Pelling....

[Austin, Alex]

I imagine the problem is voltage, actually. The OMAP3530 and other similar devices run their JTAG ports at 1.8V, and the ARM-USB-OCD doesn't like working below 3V.
________________________________________
From: openocd-development-***@lists.berlios.de [openocd-development-***@lists.berlios.de] On Behalf Of Nick Pelling [***@nanodome.com]
Sent: Friday, December 03, 2010 12:52 PM
To: openocd-***@lists.berlios.de
Subject: [Openocd-development] OpenOCD Cortex-A8 / S5PC100 support...?

Hi everyone,

I've just plugged my trusty old Olimex ARM-USB-OCD (the Swiss Army
Knife of JTAG debuggers) into my shiny new S5PC100-based board and...
I'm struggling to see how to get it working even slightly.

Has anyone got anywhere with the s5pc100 or other CoreSight-based
Cortex A8s (apart from the OMAP3530 as per
http://elinux.org/BeagleBoardOpenOCD )? Firas Achkar mentioned trying
this on the urbetter board (which I think is actually the original
ODroid) a month ago, as did Matt Hsu in November 2009. But trying the
latest openocd's autoprobe (as David Brownell suggested then) didn't
seem to produce anything so revealing as an ID code.

Incidentally, I've read through the CoreSight chapter in the latest
Samsung datasheet but that had precisely nothing so useful as a TAP
ID code or a DAP ID code to get started. Stepping through every
occurrence of JTAG also revealed nothing useful.

Any suggestions? Has anyone tried the same thing for the S5PC110 (as
per the ODroid-T etc)? Might there be some hidden nest of Samsung SoC
OpenOCD developers in a far-flung corner of Korea?

Cheers, ....Nick Pelling....

[Dirk Behme]

Post by Austin, Alex
I imagine the problem is voltage, actually. The OMAP3530 and other similar devices run their JTAG ports at 1.8V, and the ARM-USB-OCD doesn't like working below 3V.

Therefore the Flyswatter

http://www.tincantools.com/product.php?productid=16134&cat=0&page=1&featured

is popular for OMAP3 because it supports 1.8V.

Dirk

Post by Austin, Alex
________________________________________
Sent: Friday, December 03, 2010 12:52 PM
Subject: [Openocd-development] OpenOCD Cortex-A8 / S5PC100 support...?
Hi everyone,
I've just plugged my trusty old Olimex ARM-USB-OCD (the Swiss Army
Knife of JTAG debuggers) into my shiny new S5PC100-based board and...
I'm struggling to see how to get it working even slightly.
Has anyone got anywhere with the s5pc100 or other CoreSight-based
Cortex A8s (apart from the OMAP3530 as per
http://elinux.org/BeagleBoardOpenOCD )? Firas Achkar mentioned trying
this on the urbetter board (which I think is actually the original
ODroid) a month ago, as did Matt Hsu in November 2009. But trying the
latest openocd's autoprobe (as David Brownell suggested then) didn't
seem to produce anything so revealing as an ID code.
Incidentally, I've read through the CoreSight chapter in the latest
Samsung datasheet but that had precisely nothing so useful as a TAP
ID code or a DAP ID code to get started. Stepping through every
occurrence of JTAG also revealed nothing useful.
Any suggestions? Has anyone tried the same thing for the S5PC110 (as
per the ODroid-T etc)? Might there be some hidden nest of Samsung SoC
OpenOCD developers in a far-flung corner of Korea?
Cheers, ....Nick Pelling....
_______________________________________________
Openocd-development mailing list
https://lists.berlios.de/mailman/listinfo/openocd-development
_______________________________________________
Openocd-development mailing list
https://lists.berlios.de/mailman/listinfo/openocd-development

[David Brownell]

Post by Austin, Alex
I imagine the problem is voltage,
actually. The OMAP3530 and other similar devices run their
JTAG ports at 1.8V,

True, and important. Thought I'm not sure
quite what you mean by "Similar". Don't
the Sitara chips run at 3V3 not 1V8? And
I thought I heard about a DaVinci-family
Cortex-A8, at that voltage too. That may
have been a rumor; I've not tracked such.

and the ARM-USB-OCD doesn't like working

Post by Austin, Alex
below 3V.

Not without a level shifter, like the one
which I used to debug beagle with Olimex-Tiny.

More convenient, doubtless, to use a JTAG that
has a built in level shifter. That's why folk
talk FlySwatter so often with respect to Beagle.

... and why I'd like to see XDS100 support become
possible in OpenOCD. (It includes both 1V8
support, and the TI-14 JTAG connector that's used
instead of standard ARM-20 connectors.

[Nick Pelling]

Hi everyone,

Ah, I should have added that voltage was the very first thing I
checked. On the S5PC100 the JTAG runs at VDD_EXT which has a valid
operating range from below 1.8V to above 3.3V, and the board I'm
trying to bring up has VDD_EXT set to 3.3V (as designed, measured to
~3.25V), so this really shouldn't be the issue here.

Cheers, ....Nick Pelling....

[Nick Pelling]

Hi everyone,

Using the Bus Pirate's JTAG macros, my S5PC100 returns a valid 3.3V
JTAG chain with 250 devices (it's a highly integrated SoC, so 250
sounds like a plausible number), each with IDCODE = 0: while a
Cortex-M3 microcontroller returns a single-device chain with the
correct id value.

Hence I'm pretty convinced that this S5PC100 board is connected and
functioning properly, but am struggling for ideas as to how best to
start constructing a suitable target cfg file for it. I've asked
Samsung for documentation listing all its JTAG codes, but that may
take some time to arrive. Any other suggestions?

Thanks, ....Nick Pelling....

[Tomek CEDRO]

(...)constructing a suitable target cfg file for it. I've asked Samsung for
documentation listing all its JTAG codes, but that may take some time to

Hello Nick, the other problem is that Samsung may want you to sign the
NDA and keep quiet about the device, as they did when I asked for some
documentation for one device that I wanted to use in open source
project - they did not agree :-(

If that helps I have some OMAP 35xx devel boards at my site (its
Cortex-A8 I think) at the moment, so I can verify/support your
research on my hardware here, but I have some other priority tasks to
finish so m y time is limited, so I can help you with the problem at
the moment sorry :-)

Best regars,
Tomek

[Peter Stuge]

Post by Tomek CEDRO

I've asked Samsung for documentation listing all its JTAG codes,
but that may take some time to

Hello Nick, the other problem is that Samsung may want you to sign the
NDA and keep quiet about the device, as they did when I asked for some
documentation for one device that I wanted to use in open source
project - they did not agree

If you run into this please let me know what particular device it is.


//Peter

[Nick Pelling]

Hi everyone,

Tomek CEDRO wrote: (replying to Nick Pelling)

Post by Tomek CEDRO

I've asked Samsung for documentation listing all its JTAG codes,
but that may take some time to

Hello Nick, the other problem is that Samsung may want you to sign the
NDA and keep quiet about the device, as they did when I asked for some
documentation for one device that I wanted to use in open source
project - they did not agree

If you run into this please let me know what particular device it is.

I now have the BSDL file for the S5PC100 (through Samsung support
channels) which is great, but my next step is to try to find out
whether or not it is covered by the NDA I've already signed - many
manufacturers post their BSDLs openly (for instance, Luminary Micro),
but this one wasn't on Samsung Semi's support site.

There's nothing particularly surprising in there, apart from some
JTAG boundary constraints which I suspect really ought to have been
included in the User Manual.

Cheers, ....Nick Pelling....

[Tomek CEDRO]

(...) constructing a suitable target cfg file for it. I've asked Samsung for
documentation listing all its JTAG codes,

Btw. have you tried UrJTAG and the "discovery" command - it helps
finding such registers on unknown devices :-)