ESP32-S2 WiFi does not work (Scan returns no results)

[carterghill]

Describe the bug The WiFi interface seems altogether non-functional on the S2 chip. Running "wifi scan" in the shell returns no results, and any attempts to connect always fail, through the code or the shell.

[00:00:00.165,000] pca953x: pca9534@20 init ok [00:00:00.217,000] esp32_wifi: Failed to send packet [00:00:00.218,000] esp32_wifi: Failed to send packet [00:00:00.219,000] esp32_wifi: Failed to send packet [00:00:00.220,000] esp32_wifi: Failed to send packet Booting Zephyr OS build v3.7.0-rc1 [00:00:01.221,000] esp32_wifi: Failed to send packet [00:00:02.222,000] esp32_wifi: Failed to send packet uart:~$ wifi scan Scan requested Scan request done [00:00:07.386,000] esp32_wifi: No Wi-Fi AP found

Sometimes it will actually show one or two of the nearby SSIDs, but it doesn't happen often. I would say 99% of the time it returns no results at all.

To Reproduce Steps to reproduce the behavior:

1. cd zephyr/samples/net/wifi
2. west build -p -b esp32s2_v2duck && west flash --esp-device /dev/ttyACM0
3. minicom -b 921600 -D /dev/ttyUSB0
4. "wifi scan"

We are using a custom board that is most similar to the saola board, so for the purposes of reproducing the bug on your end I would point to esp32s2_saola or maybe esp32s2_devkitc. I believe the only difference between our board and the saola is the uart pin output

Expected behavior

When compiling the wifi sample code for another esp32 device, the expected output is shown:

uart:~$ wifi scan Scan requested

Num | SSID (len) | Chan (Band) | RSSI | Security | 1 | 10 | 6 (2.4GHz) | -51 | WPA2-PSK | 2 | 10 | 1 (2.4GHz) | -56 | WPA2-PSK | 3 | **** 6 | 5 (2.4GHz) | -64 | UNKNOWN | Scan request done uart:~$

Impact This is a show stopper, we need our devices to have functioning wifi.

Logs and console output

Given the inconsistent nature of the bug (the fact that it SOMETIMES shows an SSID) has made us think it might be a memory issue, so here's some kernel information that may or may not shed some light on the issue for you.

Max allocated does go up after the first scan, and then stays the same afterwards.

uart:~$ kernel heap free: 42160 allocated: 27536 max. allocated: 27720 uart:~$ wifi scan Scan requested Scan request done [00:00:13.918,000] esp32_wifi: No Wi-Fi AP found uart:~$ kernel heap free: 42084 allocated: 27600 max. allocated: 29592

uart:~$ kernel threads Scheduler: 71969 since last call Threads: 0x3ffcb618 rx_q[0]
options: 0x0, priority: -1 timeout: 0 state: pending, entry: 0x400a0688 stack size 2048, unused 1784, usage 264 / 2048 (12 %)

0x3ffcb278 wifi
options: 0x8, priority: 2 timeout: 0 state: pending, entry: 0x40034390 stack size 3584, unused 2464, usage 1120 / 3584 (31 %)

0x3ffcb540 net_mgmt
options: 0x0, priority: -1 timeout: 0 state: pending, entry: 0x4002a76c stack size 2048, unused 840, usage 1208 / 2048 (58 %)

*0x3ffcb050 shell_uart options: 0x0, priority: 14 timeout: 0 state: queued, entry: 0x40086380 stack size 2048, unused 456, usage 1592 / 2048 (77 %)

0x3ffcb1d0 esp_timer options: 0x8, priority: 2 timeout: 0 state: pending, entry: 0x40089464 stack size 4096, unused 3680, usage 416 / 4096 (10 %)

0x3ffcb8c0 sysworkq
options: 0x1, priority: -1 timeout: 0 state: pending, entry: 0x4002a76c stack size 1024, unused 256, usage 768 / 1024 (75 %)

0x3ffcaf70 logging
options: 0x0, priority: 14 timeout: 0 state: pending, entry: 0x4002cfb4 stack size 1024, unused 744, usage 280 / 1024 (27 %)

0x3ffcb780 idle
options: 0x1, priority: 15 timeout: 0 state: , entry: 0x4002a240 stack size 1024, unused 840, usage 184 / 1024 (17 %)

[github-actions[bot]]

Hi @carterghill! We appreciate you submitting your first issue for our open-source project. 🌟

Even though I'm a bot, I can assure you that the whole community is genuinely grateful for your time and effort. 🤖💙

[sylvioalves]

@marekmatej

[marekmatej]

hi @carterghill I see the same behavior on S2 boards and we are looking for the fix.

[carterghill]

Unfortunately, this has not fixed things on my end it seems. After "west update" (and then blobs), I verified that the changes made were present in clock_control_esp32.c, west.yml, and loader.c, yet after building and flashing the wifi sample code, I get the same results as above.

[sylvioalves]

@carterghill We confirmed that was an issue with ESP32-S2 Wi-Fi exactly as you described. However, the fix provide was good. Is there any other ESP32-S2 devkit board you could give a second try? Do you have any changes in prj.conf?

[carterghill]

@carterghill We confirmed that was an issue with ESP32-S2 Wi-Fi exactly as you described. However, the fix provide was good. Is there any other ESP32-S2 devkit board you could give a second try? Do you have any changes in prj.conf?

No changes to prj.conf. I tested several of our devices, and I actually managed to find one that was able to get wifi scan results. This was a bit surprising to me, because wifi had been confirmed to work on our boards a while ago using arduino. I will try to confirm that myself on the boards that are currently not getting results with zephyr.

Might be a silly question, was the fix tested on Saola boards specifically? I know there are different varieties of the S2 but I don't know if the differences would be significant.

In any case, probably the best thing to do is figure out what the difference is with our one functioning board. Thank you for your help!