Deep Dive w/Scott: CircuitPython on a Raspberry Pi

Adafruit Industries

Join Scott as he works on bringing CircuitPython to the Raspberry Pi. Work on USB has stalled so he's taking a break by working on getting CircuitPython working over UART. Chat with Scott and lot of others on the Adafruit Discord at https://adafru.it/discord. Deep Dive happens every week. Normally Fridays at 2pm Pacific but occasionally shifted to Thursday at 2pm. Typically goes for two hours or more. Questions are welcome. Next week is on Friday. 0:00 Getting started 0:03:25 Hello everyone / Housekeeping 0:05:51 Today - working on Arm Cortex A as bare metal micro (raspberry pi 4) 0:06:37 last week: interrupts / tiny usb 0:07:29 Exception Levels. EL3, EL2 (hypervisor), EL1 (OS), EL0 (applications) 0:09:13 Last week - EL2 hypervisor bits needed to be set 0:10:47 Looked into ARM architecture reference manual HCR_EL2 (64 bit register) 0:12:39 TGE bit Trap General Exception 0:16:19 D1.13 asynchronous exception type, routing, masking and priorities 0:18:09 Non-secure mode, SCR, TDE “C - interrupt remains pending” TGE bit set to 1 0:19:44 the magic bit that needed to be set in broadcom peripherals repo 0:21:00 NVM3722 ARM and AUX_IRQhandler 0:23:25 Pointer boxing ( using ‘extra’ bits in 64 bit pointer ) 0:24:10 Open OCD, Gdb, and uart screens 0:25:06 CP repl on Rasbperry Pi 0:25:33 print(“hello world”) hangs / next week HDMI output, but first fix this bug 0:28:21 ARM exception decoding using cortex-a.py arm8a-exception 0:29:40 pointer greater than 32 bits - probably a problem 0:30:30 Memory mapping ( only the first ‘gig’ of memory mapped ) 0:37:19 question: can a zero 1.3 use with any lcd??? 0:40:55 aapcs arm call stack research / Procedure call standard for the Arm Architecture 0:43:51 decode failing - irq_entry and irq_exit call stack creation 0:45:20 Frame pointer 0:50:58 ARM ABI 0:53:40 Microsoft arm arch 64 ABI documentation - very similar 0:54:54 experiment with h=”hello world” 0:55:58 gdb) info frame 1:02:25 typical C function call 1:05:28 exceptions should save the ‘callers’ state 1:06:13 ‘101’ review of memory ( heap and stack ) 1:25:00 disassemble mp_obj_print_helper to look at use of frame pointer 1:28:00 back to github libunwind 1:34:35 Unwinding frames in python 1:35:10 CP uses PIO for neopixels on RP2040 1:37:52 aarch64-unwind.h 1:41:10 does it come down to the exception entry code needing to update frame pointer (x29) to be stored on the stack to a GDB "correct" value, rather than storing the previous frame pointer value when exception occurs? 1:51:20 Digikey has a small number of the LED glasses and nRF driver boards available 1:54:00 stack-unwind-samples 1:56:10 greyed out command completion from fish shell 1:59:08 code is compiled for 64 bit ARM 2:00:43 oh, my fish extension - otherwise it’s pretty standard .fishrc (omf and bobthefish) 2:02:44 looking at github rsta2/circle for examples 2:03:51 exception stub 2:04:32 edc/bass to use bash utilities in fish shell 2:05:12 circle irq handler - might like what we want “exceptionstub64.S” 2:16:23 trying out the experiment - it’s worse :-( 2:18:55 wrap up - shop at adafruit.com or digikey.com if not in stock 2:20:47 wake the cat - and catch the links at github tannewt rpi branch 2:21:55 bye spook Visit the Adafruit shop online - http://www.adafruit.com ----------------------------------------- LIVE CHAT IS HERE! http://adafru.it/discord Adafruit on Instagram: https://www.instagram.com/adafruit Subscribe to Adafruit on YouTube: http://adafru.it/subscribe New tutorials on the Adafruit Learning System: http://learn.adafruit.com/ -----------------------------------------