Hard fault crash when using HTTPServer:ESP32-C3

@c---

EDIT: closed in favor of #9749

Originally posted by @c--- in #9173 (comment)

Adafruit CircuitPython 9.2.0-beta.1 on 2024-10-11; Maker Go ESP32C3 Supermini with ESP32-C3

As far as I can tell, server.poll() might be the cause.

Currently using the following code.py that still crashes:

import wifi
import socketpool
from adafruit_httpserver import Server

pool = socketpool.SocketPool(wifi.radio)
server = Server(pool, debug=False)

while True:
   try:
      server.poll() 

   except Exception as e:
      print(str(e))

Is this a duplicate of #9460 ?

@mMerlin

Is this a duplicate of #9460 ?

It might be but it's a much simpler reproducer, which is great. We should try to fix this and then see ask @mMerlin if #9460 is fixed.

I consider this a showstopper for 9.2.0.

@c---

@c--- The test program above generates continuous ServerStopped exceptions. This is more easily seen if you do

while True:
   try:
      server.poll() 

   except Exception as e:
      print(type(e))

to print the type rather than the string form of the exception.

You do need a server.start(). However, in your original test program, you had:

server.start("0.0.0.0", 80)

If I add that, I do indeed get a hard crash. But instead of using 0.0.0., use this:

server.start(str(wifi.radio.ipv4_address))

In other words, for now, you need to specify the IP address specifically; "0.0.0.0" doesn't work. That's a bug we should fix.
You can also use server.serve_forever(...), which does the .start() and the .poll() loop for you.

What are the implications of this discovery for #9460 which was thought to be the same (or at least related) bug. That does not use the 0.0.0.0 address.

What are the implications of this discovery for #9460 which was thought to be the same (or at least related) bug. That does not use the 0.0.0.0 address.

I don't think these are the same after all, after looking into it further.

Reproduced, but only on ESP32-C3. Test program:

import wifi
import socketpool
from adafruit_httpserver import Server
import time

time.sleep(5)
print("IP addr", wifi.radio.ipv4_address)

pool = socketpool.SocketPool(wifi.radio)
server = Server(pool, debug=False)

# port 80 will get EADDRINUSE due to web workflow
# Either IP adddress will cause a crash eventually.
server.start("0.0.0.0", 8000)
#server.start(str(wifi.radio.ipv4_address), 8000)

while True:
   try:
      server.poll()
   except Exception as e:
      print(type(e))

Either servert.start() can cause a crash: the wildcard IP address or the actual IP address don't matter. It takes around 1-5 minutes to restart in safe mode, though I have seen it run longer. No interaction with the HTTP server is needed. I simply restart code.py and wait.

I have reproduced this only on a QT Py ESP32-C3. I cannot reproduce on a QT ESP32-S3, with no PSRAM. I was hoping it might have been a storage leak.

I'm closing this in favor of #9749. There are several red herrings above, which are irrelevant to the actual issue.

Navigation Menu

Repository navigation

Hard fault crash when using HTTPServer:ESP32-C3 #9735

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Hard fault crash when using HTTPServer:ESP32-C3 #9735

Description

Activity

jepler commented on Oct 20, 2024

dhalbert commented on Oct 20, 2024

dhalbert commented on Oct 20, 2024

mMerlin commented on Oct 21, 2024

dhalbert commented on Oct 21, 2024

dhalbert commented on Oct 22, 2024

dhalbert commented on Oct 23, 2024

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions