08:32 <zapb_> I have a device with two cores, access to the second core is disabled after reset. I need to write to a specific register to enable / power-up the second core. What is the best (reset) event to automatically write to this register? Does anybody know a similar and supported device?

08:33 <Haohmaru> is this supposed to be done thru the debugger or is this entirely a firmware question/

08:34 <zapb_> No, it's supposed to be done by the debugger

08:35 <Haohmaru> shouldn't it be done somewhere around after openocd "connects and does the typical things" ?

08:36 <Haohmaru> i mean at the same point where you might "begin to flash a new firmware" or "begin debugging"

08:39 <zapb_> Yes, somewhere in the 'reset event' system, but OpenOCD goes crazy after the reset because it cannot examine the second core because it's not accessible anymore

08:40 <Haohmaru> oh

08:41 <Haohmaru> so it has to be somewhere before this "examine" thing

08:53 <zapb_> Yep, already deferred the examine process which works the first time but not work after reset

12:21 <zapb_> borneoa___, do you know a target with a similar behaviour that we already support / have a config file for?

12:29 <karlp> if I had ot guess, I'd look at one of the renesas rcar stuff, not sure it's all merged yet?

12:35 <karlp> perhaps the event tap-enable?

13:13 <borneoa___> zapb_: yes

13:15 <borneoa___> zapb_: stm32mp15x, 21x, 23x and 25x. While booting on cortex-A the cortex-M is not accessible

13:18 <borneoa___> zapb_: in the script upstream I have added several events to handle it. There is a clock to be enabled and a bit to set in DBGMCU . The fix is done at reset and examine events to care about OpenOCD start and at reset

13:19 <borneoa___> zapb_: on which device are you working, if you can share the info...

14:01 <zapb_> borneoa___, thanks! Cannot say at the moment but will bring everything upstream asap.

14:01 <zapb_> borneoa___, noticed that we have a $halt variable in the the reset events, maybe that's the trick. will check later

14:02 <zapb_> borneoa___, thanks a lot!

14:12 <zapb_> borneoa___, maybe one additional question: I would assume that once examined OpenOCD tries to poll your CM core. and at reset the core is not accessible. how do you prevent that?

14:13 <PaulFertser> borneoa___: hey! BTW, saw your commit about Tcl/Tk events but I'm a bit confused. I understand that comment in the sense that OpenOCD can take inspiration from Tk and support some kind of templating for the event handler body to make it easier to DRY and write more generic event handlers.