On 26.12.2010 19:41, David Brownell wrote:
via Google search, I found the following, which, in conjunction with the
Cortex-A9 reference manuals from ARM, may be of interest to folk looking
at OMAP4 and other recent TI chips for OpenOCD; it looked to answer some
of the questions I've heard on this list.
http://processors.wiki.ti.com/index.php/ICEPICK
(Also of course you'd use the ARMv7A manual too, but
large chunks of that already have debug support in OpenOCD,
( ... for Cortex-A8 cores as on OMAP3, (plus ARMv6/ARM11).
Linked from that page is a PDF document
http://processors.wiki.ti.com/images/f/f6/Router_Scan_Sequence-ICEpick-D.pdf
which may be more information than was public when ICEPICK-C support
was added to OpenOCD, enabling DaVinci and OMAP3 operations. (And FWIW,
the link-in-a-TAP sequences looked similar at first glance.)
The public OMAP4430 reference manual tells which cores are at which
address on the ICEpick-D ... e.g. the Cortex-A9 DAP is #9, the Cortex-M3
cores (Ducati) are at #4 and #5, and there are even two ARM968 cores at
addresses #2 and #3 (presumably for what older OMAPs used ARM7 cores).
(Plus of course the C6000x+ DSP.) Not clear which of those we should
be configuring into our scan_chain, maybe just the A9 to start. (That
will keep scans faster, and reduce power usage during debug.)
I'm thinking the current ICEpick support may need to be packaged as
specifically ICEpick-C support; maybe some can be shared with ICEPick-D,
I've not investigated yet.
Maybe this mail from the PandaBoard (OMAP4) mailing list might be
interesting here, too:
http://groups.google.com/group/pandaboard/msg/2a4eb67a994b9966
Best regards
Dirk
___
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development
