esp32c3 i2c will cause cpu IDLE (IDFGH-6463)

[FanhuaCloud]

Environment

• Development Kit: none
• Module or chip used: esp32-c3
• IDF version (run git describe --tags to find it):
  // ESP-IDF v5.0-dev-797-g279c8aeb8a
• Build System: idf.py
• Compiler version (run xtensa-esp32-elf-gcc --version to find it):
  // riscv32-esp-elf-gcc-8.4.0.exe (crosstool-NG esp-2021r2) 8.4.0
• Operating System: Windows
• (Windows only) environment type: ESP Command Prompt
• Using an IDE?: Yes vscode esp-idf
• Power Supply: USB

Problem Description

When using i2c, there is a chance that the cpu will be idle
I think it might be because the i2c code took too long
//Detailed problem description goes here.

Expected Behavior

Actual Behavior

E (6907) task_wdt: Task watchdog got triggered. The following tasks did not reset the watchdog in time:
E (6907) task_wdt: - IDLE (CPU 0)
E (6907) task_wdt: Tasks currently running:
E (6907) task_wdt: CPU 0: main

Steps to reproduce

1. step1
2. ...

// If possible, attach a picture of your setup/wiring here.

Code to reproduce this issue

        i2c_cmd_handle_t cmd = i2c_cmd_link_create();
        i2c_master_start(cmd);
        i2c_master_write_byte(cmd, AHT_I2C_ADDRESS_GND << 1 | I2C_MASTER_READ, true);
        i2c_master_read(cmd, data, 6, I2C_MASTER_ACK);
        i2c_master_stop(cmd);
        i2c_master_cmd_begin(I2C_NUM_0, cmd, 1000 / portTICK_PERIOD_MS);
        i2c_cmd_link_delete(cmd);

// If your code is longer than 30 lines, GIST is preferred.

Debug Logs

Debug log goes here, should contain the backtrace, as well as the reset source if it is a crash.
Please copy the plain text here for us to search the error log. Or attach the complete logs but leave the main part here if the log is *too* long.

Other items if possible

• sdkconfig file (attach the sdkconfig file from your project folder)
• elf file in the build folder (note this may contain all the code details and symbols of your project.)
• coredump (This provides stacks of tasks.)

[FanhuaCloud]

I look the same ploblem #7812

[mythbuster5]

i2c_patch_1210 .txt

A patch may help here, change the *.txt to *.patch and git apply i2c_patch_1210.patch. Thanks

[o-marshmallow]

Hi @FanhuaCloud ,

Can you provide a way for us to reproduce this issue? Can you provide an sdkconfig? What is you FreeRTOS tick rate and task watchdog timeout period in your sdkconfig (FREERTOS_HZ and ESP_TASK_WDT_TIMEOUT_S) ?
Thank @mythbuster5 for providing a fix

In a more general note, we do provide wrappers to ease I2C transfers, in your case, you could use:

i2c_master_read_from_device(I2C_NUM_0, AHT_I2C_ADDRESS_GND, data, 6, 1000 / portTICK_PERIOD_MS)

[FanhuaCloud]

Hi @FanhuaCloud ,

Can you provide a way for us to reproduce this issue? Can you provide an sdkconfig? What is you FreeRTOS tick rate and task watchdog timeout period in your sdkconfig (FREERTOS_HZ and ESP_TASK_WDT_TIMEOUT_S) ? Thank @mythbuster5 for providing a fix

In a more general note, we do provide wrappers to ease I2C transfers, in your case, you could use:

i2c_master_read_from_device(I2C_NUM_0, AHT_I2C_ADDRESS_GND, data, 6, 1000 / portTICK_PERIOD_MS)

https://gist.github.com/FanhuaCloud/be12c7ad155f7a75221b99bd3e64ca62
This is my code, I am using aht10 to get the temperature.But when I use i2c to operate other devices, it is normal

[AxelLin]

i2c_patch_1210 .txt

A patch may help here, change the *.txt to *.patch and git apply i2c_patch_1210.patch. Thanks

@mythbuster5
The patch does not apply to current master (and v4.4) tree.

[AxelLin]

Just wondering how is the status, @mythbuster5 @o-marshmallow

I found:
In i2c_master_cmd_begin(),
If portBASE_TYPE evt_res = xQueueReceive(p_i2c->cmd_evt_queue, &evt, wait_time); timeout
then it calls
i2c_hw_fsm_reset
-> i2c_master_clear_bus(i2c_num); << This call never return

Then it keeps showing:
E (11445) task_wdt: Task watchdog got triggered. The following tasks did not reset the watchdog in time:
E (11445) task_wdt: - IDLE (CPU 0)
E (11445) task_wdt: Tasks currently running:
E (11445) task_wdt: CPU 0: main

[AxelLin]

Hi @FanhuaCloud ,

Thank @mythbuster5 for providing a fix

@o-marshmallow @mythbuster5
Caould you clarify what was fix? (Above patch does not apply)

[mythbuster5]

i2c_patch_0317.txt

@AxelLin Sorry, I saw the message just now

[AxelLin]

@mythbuster5

I test it by disconnect the SCL from my i2c sensor on pupose to simulate the TIMEOUT case.
Without the patch, I always got "task_wdt: Task watchdog got triggered.".
With the path, I no longer hit watchdog issue.
Thanks.

[AxelLin]

@o-marshmallow @mythbuster5

The fix is not available in any branch, when would you merge it?
The original patch (#8120 (comment)) was provided in last year,
no idea what are you waiting for.

[o-marshmallow]

@AxelLin This fix is simply a workaround, the goal is to fix the root cause first

[AxelLin]

@AxelLin This fix is simply a workaround, the goal is to fix the root cause first

The workaround fix appears in various places:
#8358 (comment)
#8543 (comment)
The feedback is the patch does not apply, however, there is no follow up.
It's a mess now.

[AxelLin]

Just wondering how is the status, @mythbuster5 @o-marshmallow

I found: In i2c_master_cmd_begin(), If portBASE_TYPE evt_res = xQueueReceive(p_i2c->cmd_evt_queue, &evt, wait_time); timeout then it calls i2c_hw_fsm_reset -> i2c_master_clear_bus(i2c_num); << This call never return

@o-marshmallow
FYI: Comment out this line fixes the "Task watchdog got triggered" issue:
https://github.com/espressif/esp-idf/blob/master/components/hal/esp32c3/include/hal/i2c_ll.h#L811

[Alvin1Zhang]

Thanks for reporting and sorry for slow turnaround, fix is available 0b00831, we are also back porting to release branches.

Feel free to reopen.