No worries Tim - our situation was probably simpler than yours.
In our case the FPGA TAP simply needed to be sent a special command to "bypass" 
it and expose the M3 TAP.
There was no other initialization/setup to be done.
________________________________
From: Tim Newsome <[email protected]>
Sent: Tuesday 7 January 2020 18:55
To: Tommy Murphy <[email protected]>
Cc: OpenOCD ML <[email protected]>
Subject: Re: [OpenOCD-devel] handle targets that are inaccessible when 
connecting

Thanks, Tommy. I'd thought of doing it all in TCL before `init` but I feel it 
gets rather unwieldy, and I'd end up rewriting a bunch of the RISC-V target 
code in TCL just to make it happen.

Tim

On Mon, Jan 6, 2020 at 2:18 PM Tommy Murphy 
<[email protected]<mailto:[email protected]>> wrote:
Hi Tim

FWIW we (Microchip/Microsemi) have a similar situation with our 
SmartFusion/SmartFusion2 SoC FPGAs which have an FPGA top level TAP and also a 
hard Cortex-M3 with its own TAP in the MSS (Microcontroller Sub-system). The 
former has to be negotiated and a special command sent before the latter can be 
accessed. The attached target script shows how we manage this. Perhaps (some 
of) this might be relevant to your situation?

Hope this helps
Regards
Tommy

________________________________
From: Tim Newsome <[email protected]<mailto:[email protected]>>
Sent: Monday 6 January 2020 20:05
To: OpenOCD ML 
<[email protected]<mailto:[email protected]>>
Subject: [OpenOCD-devel] handle targets that are inaccessible when connecting

I have a RISC-V target with 2 TAP controllers chained together. The first 
controller connects to a core that's always accessible. The second controller 
connects to a core that's only accessible after the first core has done some 
setup, which (at least initially) requires the debugger.

If I just configure both of those and connect then OpenOCD errors out because 
examine() fails on the core behind the second controller because it's not 
powered up yet. How can I change the code to make this case work? Is there an 
example for some other target?

Tim
_______________________________________________
OpenOCD-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openocd-devel

Reply via email to