lukeswitz / atomgps_wigler Goto Github PK

Getting a ton of errors in serial output

StartM5Atom initializing...OK
E (63) gpio: gpio_set_level(227): GPIO output gpio_num error

They pause when we wait for the GPS fix, then resume in bursts of dozens per second before and after a new filename is created.

10:46:35.052 -> E (223) gpio: gpio_set_level(227): GPIO output gpio_num error
10:46:35.052 -> E (228) gpio: gpio_set_level(227): GPIO output gpio_num error
10:46:35.084 -> E (234) gpio: gpio_set_level(227): GPIO output gpio_num error

Looks like this is a known thing -- espressif/arduino-esp32#6737 -- but doesn't have a direct solution.

Curious if others are seeing it. Makes debugging via serial output really a pain.

Serial flooding with gpio: gpio_set_level(227): GPIO output gpio_num error

See issue with esp32 board package: espressif/arduino-esp32#6737

I noticed there is a difference in AUTH_TYPE list versus JHewitt code. Should we align?

• Jhewitt uses:
  case WIFI_AUTH_WPA2_ENTERPRISE:
  authtype = "[WPA2]";

• AtomGPS uses:
  case WIFI_AUTH_WPA2_ENTERPRISE:
  return "[WPA2_ENTERPRISE]";

The ESP32 spec doesn't give a name for this particular value, just showing
enumerator WIFI_AUTH_WPA2_ENTERPRISE
authenticate mode : WiFi EAP security

https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/network/esp_wifi.html

Not sure if Wigle cares for their statistics - they point from the CSV format file to https://developer.android.com/reference/android/net/wifi/ScanResult#capabilities - which doesn't have any direct match-up information in the getSecurityTypes() routine doc.

Room for improvement here. SPI might not be needed at all and blocking code.

What:
Power saving, warmail feature or when your eyes can’t take any more LEDs.

• account for debounce in Atom button
• needs to be non-blocking logic that calls M5.update periodically as an async call while in the loop.
• testing for button press could be handled with the SmartButton library

Latest version doesn't see my SD Card unless I specify parameters for SD.begin()

while (!SD.begin(-1, SPI, 40000000))

This works but has not solved the gpio_num issue for me yet. I tested my SD Card successfully using the example code for SD.h.

I am simply copying the code into my Arduino IDE so I may have different libraries or something when compiling. I wanted to mention it in case others have the same issue.
Thanks

General observation - there are a number of WiFi AuthType values in the ESP32 docs that aren't in the code enumeration.
https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/network/esp_wifi.html)

Are all the types from <wifi.h> included in the code? Should they be?
Maybe not relevant if they aren't supported by the chipset?

I do not see this one in JHewitt's code either:

enumerator WIFI_AUTH_ENTERPRISE
authenticate mode : WiFi EAP security

He just has the WPA2 equivalent:
case WIFI_AUTH_WPA2_ENTERPRISE:
authtype = "[WPA2]";

A happy medium of 300-600ms should be enough. Input welcome on this for sure.

• When flashing using esptool most have issues because of the chip frequency and identification

Tested working: esptool.py --chip esp32 --port [PORT] --baud 115200 write_flash -z --flash_mode dio --flash_freq 80m --flash_size detect 0x10000 [FIRMWARE_FILE]

Notes:
https://stackoverflow.com/questions/66514659/how-to-flash-tasmota-or-anything-to-m5stack-atom-lite-esp32-pico-d4-using-es

Add a variable or definition to track the code version, and put that value into the Wigle CSV file header 'appRelease'.

// code release Version number
const String VERSION = "1.300000";

The down in the file header code:

if (dataFile) {
dataFile.println("WigleWifi-1.4,appRelease=" + VERSION + ",model=GPS Kit,release=1.100000F+00,device=M5ATOM,display=NONE,board=ESP32,brand=M5");

Currently the "model" value in the Wigle CSV header is "model=GPS Kit".

The CSV spec doesn't indicate how spaces in the values are handled, but the Stats output make it look like the value is being truncated to "gps".

Should the code be updated to "GPS-Kit" or "AtomGPS"? Do we pick a value or keep it generic to group the AtomGPS, AtomMatrix, etc M5Stack models together?

CSV spec - https://api.wigle.net/csvFormat.html
Stats page - https://wigle.net/stats#octetstats

Investigating how squashed and rebased commits are improperly combined into this broken state.

When I start the Atom without an SD card, the LED flashes red as expected.

If I pull the SD card mid-run, the Atom appears to continue scanning - the green light flashes periodically and you wouldn't know the card has been lost.

What: Weight some channels with a longer dwell time.

Why: Spending equal time per channel could be optimized.

How: Wrap it in a for loop, use longer time params for popular channels. Input on said channels welcome