07:28 <zapb_> mawk, ah okay, interesting. I thought the the "newer" devices all have an option byte loading command

07:28 <mawk> you mean to update the effective option bytes from the stored value?

07:29 <mawk> well NRST is enough to lose access to the debugger but apparently from the MCU side it's not effective yet for some reason

07:29 <mawk> but I don't know exactly which part of the code fails because it's in a closed source blob

07:29 <mawk> and I obviously can't debug it

07:29 <zapb_> yes, because the option bytes are loaded from flash memory only during the startup phase of the MCU

07:30 <zapb_> newer devices have the option the reload the option bytes which can be triggered by software

07:30 <mawk> I see

07:30 <zapb_> for example, the L4 series has the "option_load" command in OpenOCD

07:30 <mawk> online I saw a blurb from documentation that says POR is necessary if the debugger is still attached when RDP is enabled

07:30 <mawk> so maybe I can get around that by loading a program in SRAM and disconnect the debugger?

07:31 <mawk> and the program waits for a second and then does a sysresetreq

07:31 <mawk> I have to find where I found the blurb

07:32 <zapb_> You can try, I have not much experience with the F7 in that regard but probably this won't work

07:33 <mawk> :(

07:33 <mawk> the problem seems related to debuggers, but I'm not entirely sure

07:33 <mawk> ARM doc says whether SYSRESETREQ behaves like a POR or just like a local reset is implementation defined, but apparently on stm32 it's just a local reset