08:48 <gamiee> btw, just as info, I guess I managed to do the best reset mechanism as it is possible to do for the Bouffalo chips.

08:48 <gamiee> I just wanted to say that doing this was magnificient experience for me, both writing the reset mechanism and scripts in TCL, but also doing flash bank driver.

08:48 <gamiee> Now, I just hope it is all good and will be suitable for upstream :P

08:50 <Haohmaru> "QC PASSED"

10:43 <Guest99> I use openocd like this: openocd -f interface/jlink.cfg -c "adapter serial 11223344"....

10:43 <Guest99> However, I get "Error: invalid subcommand "serial 11223344"

10:43 <Guest99> anyidea ?

10:44 <Haohmaru> was that intended to set the adapter speed?

10:44 <Guest99> I have several JLink connected to this host

10:45 <Guest99> I want to select one using its serial number

10:45 <Haohmaru> oh, okay

10:46 <Guest99> that's how it's done here: https://github.com/openocd-org/openocd/blob/master/tcl/interface/jlink.cfg

10:46 <Haohmaru> what version of openocd are you using?

11:12 <PaulFertser> Yes, you need to read the user manual that was installed along with the binary you're running.

11:32 <Guest99> ok thanks, it's 0.11.0, you're right I didn't check the version and if the command was present

11:33 <Guest99> now I see that's "jlink serial xxxxx", thanks !

11:36 <karlp> or update to 0.12... :)

12:02 <Haohmaru> ^ +1

12:19 <Guest99> yes, I'll update

13:10 <Guest99> I usually use jlinksdk to connect and interact with my device. When I run the JLink commander, this is what I get:

13:10 <Guest99> # Connecting to target via SWD

13:10 <Guest99> # Found SW-DP with ID 0x4C013477

13:10 <Guest99> # DPIDR: 0x4C013477

13:10 <Guest99> # CoreSight SoC-600 or later (DPv3 detected)

13:10 <Guest99> # Detecting available APs

13:10 <Guest99> # APSpace base (BASEPTR0): 0x00000000

13:10 <Guest99> # APSpace size (DPIDR1.ASIZE): 32-bit (4 GB)

13:10 <Guest99> # Top-level ROM table, that describes AP map, found

13:10 <Guest99> # Scanning top-level ROM table and nested ones to find APs

13:10 <Guest99> # AP[0] (APAddr 0x00010000): APB-AP (IDR: 0x24770006)

13:10 <Guest99> # AP[1] (APAddr 0x00020000): AXI-AP (IDR: 0x24770014)

13:10 <Guest99> # AP[2] (APAddr 0x00030000): APB-AP (IDR: 0x24770006)

13:10 <Guest99> # AP[3] (APAddr 0x00040000): AHB-AP (IDR: 0x24770005)

13:10 <Guest99> I'm trying to get the same thing with openocd.

13:10 <Guest99> This is what I tried:

13:10 <Guest99> swd newdap mychip swd -expected-id 0x4C013477

13:10 <Guest99> dap create mychip.dap -chain-position mychip.swd -adiv6

13:10 <Guest99> ...

13:10 <Guest99> Any idea ?

13:11 <Guest99> I might have the config wrong, I'm working on it

13:12 <PaulFertser> Guest99: please always use some pastebin if you need to share more than two lines.

13:13 <Guest99> ok, sorry, I thought it was short enough

13:13 <PaulFertser> Guest99: adiv6 requires current code, even 0.12 is too old for that.

13:14 <PaulFertser> And use some exiting config for reference, e.g. tcl/target/ampere_qs_mq.cfg

13:15 <Guest99> I built openocd from github

13:15 <Guest99> I'll check if it's up-to-date

13:16 <PaulFertser> Well, probably 0.12 is actually new enough but I still recommend current HEAD.

13:16 <PaulFertser> So use existing config from the same version for reference and use pastebin to show full -d3 output please.

13:17 <Guest99> I've build current HEAD. Ok thanks, I'll do that.

13:23 <Guest99> https://pastebin.com/iyxH3FBH

13:25 <PaulFertser> Guest99: hm, "VTarget = 1.193 V" why is that? It's the jlink measuring VTref input and using that for JTAG/SWD signalling.

13:25 <Guest99> target is 1.2V

13:26 <Guest99> a log of openocd running on this ampere_qs_mq.cfg target would be helpful

13:27 <PaulFertser> OK, if it's 1.2 V and your jlink supports that it's correct.

13:27 <PaulFertser> Guest99: but "SWD ack not OK: 2 WAIT" looks wrong, it's as if the target is in wrong state. What is it doing at the moment you're trying to connect?

13:27 <Guest99> yes, the jlink work with Jlink commander

13:28 <Guest99> it's doing nothing, the mcu isn't running

13:28 <Guest99> I don't even need to start the cortex_m, I just want to access the mem-ap

13:29 <PaulFertser> Guest99: you define the CPU target on default AP0, that is, you omit -ap-num , is that right?

13:29 <PaulFertser> Guest99: for that do not define the arm target.

13:29 <PaulFertser> Define a mem_ap target on the appropriate AP number instead.

13:31 <Guest99> mem_ap are on their ap-num target.

13:31 <Guest99> "you define the CPU target on default AP0" -> you mean targets mychip.mcu

13:32 <PaulFertser> Hm, I see now you're defining all of them. And they all fail.

13:32 <Guest99> yes

13:35 <Guest99> I guess I would need to analyze the swd traffic and see what's going on. I don't have enough time for this right now :(

13:35 <PaulFertser> Guest99: for adiv6 AP num looks like 0x00010000

13:35 <PaulFertser> The lowest 3 digits should be zero

13:36 <PaulFertser> Guest99: nope, the problem is above SWD layer.

13:36 <PaulFertser> Use numbers from APAddr there, not just 0, 1, 2.

13:37 <PaulFertser> Guest99: that is mostly likely going to advance your efforts

13:37 <Guest99> that's what I'm trying, my connection to the target just failed, I'll be right back

13:38 <PaulFertser> (I figured that from "Invalid AP#" and checking the source code of the validation function, then it became obvious it's the immediate problem in this setup)

13:38 <Guest99> ok thanks!

13:55 <PaulFertser> Guest99: please share new output with correct ap numbers.

13:58 <Guest99> It's better: https://pastebin.com/f2ZgQWBy

14:02 <Guest99> PaulFertser ^

14:08 <Guest99> something strange, If I first run the JLink (and a DORMAT-TO-SWD that I need to run manually) and _then_ openocd, it works perfectly

14:09 <Guest99> I'll try to do only the DORMAT-TO-SWD on my side, then run openocd and I'll let you know

14:13 <Guest99> ok, the power init of my test script was not good. It now works. Thanks PaulFertser, using ap-num with addresses was the key.

14:41 <PaulFertser> Guest99: yay!

14:43 <PaulFertser> borneoa___: hm, probably -ap-num parameter and the error "Invalid AP#" is confusing in ADIv6 case where an AP _address_ is expected.

17:26 <borneoa___> PaulFertser: agree. There is another feature of ADIv6 that we do not support in OpenOCD yet, that is AP behind an AP. We will probably need to specify the whole path, e.g. dap_name:@addr_ap1/@addr_ap2/@addr_ap3... Maybe the parameter should be simply -ap and the messages adapted accordingly.

17:33 <PaulFertser> borneoa___: I do not have any specific suggestion, just an observation that it was the confusing part here for Guest99. Wasn't obvious to me either till I read the code in question.

17:34 <PaulFertser> Probably just more specific error diagnostics for different invalid cases printed by is_ap_num_valid() would be enough?