On 04/04/2011 9.43, Øyvind Harboe wrote:
Thanks for testing! That's a pretty good list to kick things off!
Features still to be implemented on PandaBoard (and all others Cortex-A*)
are:
- software breakpoints
Presumably the issue here is how to deal with caching?
I didn't investigate too much on this issue, but yes, maybe caching can
be a problem.
- tracing
- support for ETM (Embedded Trace Macrocell)
- a reliable way to use two or more cores on cortex-A9
What's the issue here?
Well, maybe I've missed something here. I've never had the need to debug
the 2nd core (often it is enabled by the OS, so bootloaders don't use
it). Anyway I wonder which is the suggested way to debug 2 cores at the
same time. Maybe start 2 openocd servers on different ports (4444 +
another port) and open two telnet sessions? Do we need two JTAG dongles?
>> - maybe many others ;-)
NOTE: I want to point out that I'm working a lot with bootloaders so I
didn't test it very well with a real OS running or with GDB.
Bootloaders and RTOS's are perhaps the most important application
of JTAG debuggers?
I think you're right because OS have themselves a lot of good methods to
debug them. On Linux you can use KDB,KGDB,ktrace,kprobes, not to mention
good old printk ;-)
_______________________________________________
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development
