Did not have time to test but the code changes look good to me. Great job tracking this down, interrupts during critical sections are never fun to debug.

Nice work! Thank you!

Testing using three pairs of QT Py boards, one of each pair running 10.0.0-alpha.4, the other running artifacts from this PR:

• (2) ESP32-S2 running HTTP server
• (2) ESP32-S2 running HTTPS server
• (2) ESP32-S3 running HTTPS server

Looks good so far. Each of the 10.0.0-alpha.4 boards has already encountered a safemode. None of the PR#10328 boards have had any issues. Will let it run for a few days and report if anything unexpected happens.

import time
import os
import microcontroller
import supervisor
import storage
import gc
import board
import digitalio
import traceback
import wifi
import socketpool
import ssl
import adafruit_connection_manager
from adafruit_httpserver import Server, Request, Response, REQUEST_HANDLED_RESPONSE_SENT

time.sleep(3)  # wait for serial
print(f"{gc.mem_free()=}")

if microcontroller.nvm[0] == 0xAA:
    print(f'{"="*25}\nreset from safemode.py')
    microcontroller.nvm[0] = 0x55
else:
    print(f'{"="*25}\nreset clean')

wifi.radio.connect(os.getenv("WIFI_SSID"), os.getenv("WIFI_PASSWORD"))
pool = adafruit_connection_manager.get_radio_socketpool(wifi.radio)
server = Server(
    pool,
    root_path="/static",
    https=True,
    certfile="cert.pem",
    keyfile="key.pem",
    debug=True,
)

@server.route("/")
def base(request: Request):
    resp = f"{storage.getmount("/").label} t={time.monotonic_ns()} mem={gc.mem_free()}"
    print(resp)
    return Response(request, resp)

server.start(str(wifi.radio.ipv4_address))
while True:
    try:
        server.poll()
    except Exception as ex:
        traceback.print_exception(ex, ex, ex.__traceback__)

HTTP Server is the same, but with the server configured without HTTPS.

No clients are attempting to connect.

Also, there is a safemode.py file that will automatically reset the board, and set nvm to indicate safemode.

Result:

With over 60 hours of completely automatic operation, pre-PR S2 & S3 devices experienced numerous intermitent safemodes and resets (as expected), and PR S2 & S3 devices ran continuously without any issues at all (as expected).

Just for fun, started up a client to repeatedly make requests on each device to confirm the servers were still operational: all servers are serving successfully. HTTP response times typically under 50ms, HTTPS response times typically just over 1/3 of a second. Measurements are from the client perspective. The first request to a device usually takes longer, sometimes much longer, probably due to TCP and / or TLS setup.

Resounding success, thanks @eightycc !