RP2040 DEBUG=1 builds don't work

[dhalbert]

Version: fails with 7.0.0 main at this time and earlier

RP2040 board builds built with DEBUG=1 don't work: they crash. The crash can be seen with gdb:

(gdb) mon reset
Resetting target
(gdb) c
Continuing.

Program received signal SIGTRAP, Trace/breakpoint trap.
isr_hardfault () at sdk/src/rp2_common/pico_standard_link/crt0.S:99
99	decl_isr_bkpt isr_hardfault
(gdb) bt
#0  isr_hardfault () at sdk/src/rp2_common/pico_standard_link/crt0.S:99
#1  <signal handler called>
#2  0x1001d160 in port_init () at supervisor/port.c:97
#3  0x1001cfc2 in main () at ../../main.c:773
(gdb) 

If I set a breakpoint at main(), it does get past main(), and then soon hits the decl_isr_bkpt isr_hardfault.

MicroPython built as CMAKE_ARGS = -DMICROPY_BOARD=$(BOARD) -DCMAKE_BUILD_TYPE=Debug does not have this problem.

(I originally thought this was some kind of problem with the Segger J-Link and gdb, but that's because I never tried this with a non-debug build. In fact gdb works fine with a non-debug build, but it's not so useful.)

[tannewt]

This is weird because this is where the memory is being initialized: https://github.com/adafruit/circuitpython/blob/main/ports/raspberrypi/supervisor/port.c#L97

[DavePutz]

I ran some tests and found the if the -Og in the Makefile under the "ifeq ($(DEBUG), 1)" with -O3 resulted in a successful (and debugable) build. @tannewt ; since a normal build also uses -O3 would this be an acceptable change to PR?

[tannewt]

@DavePutz Yup, fine with me.

[microdev1]

Closing... fixed in #5210.