On Sat, May 03, 2014 at 09:32:45AM -0700, Duane Ellis wrote:
> On May 3, 2014, at 7:48 AM, Paul Fertser <[email protected]> wrote:
> 
> >>> 
> >> Multi level ROM tables (not currently supported), and ROM tables
> >> with bugs (no HW guys have have bugs, its only us SW guys right?)
> >> they may never get fixed.
> > 
> > Displaying of nested ROM tables is already available in 0.8.0, dbgbase
> > autodetection is being worked on here: http://openocd.zylin.com/1920
> > . But what exactly are you talking about when you say about
> > autodetecting AP ports specifically,
> 
> Hmm - I have the git-head and it does not seem to work well, let me
> dig into it for a while

dap_find_ap works by simply returning the first AP of the matching
type, the ROM tables are not involved here. But I can see how it can
get wrong with multiple cores on the same DAP, so yes, we need a way
to be able to manually specify that, I agree.

> The cortex_M has assumptions about the ports that are wrong.  Some
> of this has to do with the idea of “private memory space” verses
> shared memory space. Example: Cortex_M TRM - document id:
> DDI-0337-E, Section 4.1 - page 4-2 (aka: page 88) These are not
> accessible from the SYSTEM MEMORY AP port.
> 
> My example here might be a bit mis-leading, the point is that some
> CPUs do not have the same memory map as others. Another example is
> accessing memory as PHYSICAL (through the AXI ap) vrs VIRTUAL (i.e.:
> via the cpu)

Hm, so there should be a way to specify access method for individual
regions manually, right?

> > Yep, I see, indeed. Too bad ARM folks didn't think about proper
> > topology auto-discovery when designing CoreSight :/
> 
> There are other aspects of this:
> 
> The platform may have 8 cpus.  I do not want to connect and attach,
> and probe every single CPU.  Perhaps I only want to connect and
> attach to the (NTH) cpu.
> 
> With auto discovery, how do I tell the system to *SKIP* a specific
> CPU, there is no means to do that.

I think they should have provided some robust DAP-level solution to
auto-enumeration, topology discovery etc.

Since that's not happened, well, we should provide some means for
manual configuration indeed.

-- 
Be free, use free (http://www.gnu.org/philosophy/free-sw.html) software!
mailto:[email protected]

------------------------------------------------------------------------------
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.  Get 
unparalleled scalability from the best Selenium testing platform available.
Simple to use. Nothing to install. Get started now for free."
http://p.sf.net/sfu/SauceLabs
_______________________________________________
OpenOCD-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openocd-devel

Reply via email to