ESP32-S3: USB-JTAG not working after esp32_reset() or Watchdog-Reset when TinyUSB CDC was initialized in application once (IDFGH-8354)

[rikaiot]

Answers checklist.

• I have read the documentation ESP-IDF Programming Guide and the issue is not addressed there.
• I have updated my IDF branch (master or release) to the latest version and checked that the issue is present there.
• I have searched the issue tracker for a similar issue and not found a similar issue.

IDF version.

v4.4.2

Operating System used.

Windows

How did you build your project?

Command line with idf.py

If you are using Windows, please specify command line type.

CMD

Development Kit.

ESP32-S3-DevKitM-1

Power Supply used.

USB

What is the expected behavior?

USB-JTAG is enabled by default, so when connecting the device via USB-Plug (not UART) to the PC the board is recognized as a JTAG-Interface in the device manager.

As soon as using TinyUSB CDC driver in the application by calling
tinyusb_driver_install()
and
tusb_cdc_acm_init()
the USB-JTAG is disabled and the device is recognized as a serial device.

When pressing the reset button or calling esp_reset() the board should again be recognized as USB-JTAG until the TinyUSB driver is loaded in the application.

What is the actual behavior?

When the TinyUSB has been loaded once, calling esp_reset() or forcing a watchdog reset does not make the device to be recognized as USB-JTAG any more.

Only when the reset button is pressed, the device is recognized as USB-JTAG again.

Steps to reproduce.

1. Connect the ESP32-S3-DevKitM-1 dev board to the pc via USB-Plug (not UART) - the device is recognized as USB-JTAG.

2. Initialize TinyUSB in the application by calling
   tinyusb_driver_install()
   and
   tusb_cdc_acm_init()
   after some delay.

3. The board is recognized as serial device.

4. Call esp_restart() or force and watchdog reset.

5. The board is not recognized as USB-JTAG again.

6. Press reset button - the board is recognized as USB-JTAG again until the TinyUSB initialization is done.

Debug Logs.

No response

More Information.

It seems that the USB-JTAG is disabled by TinyUSB and is not reactivated again when calling esp_restart().

[johnboiles]

Also running into this. I wonder if there's a bit in a register somewhere we could flip before calling esp_restart that would allow the USB-JTAG device to come back up. Or perhaps adding something like periph_module_disable(usb_otg_periph_signal.module); before rebooting. I'll report back if we're able to figure out a workaround to this.

[johnboiles]

Things i've tried (that haven't worked):

• periph_module_disable(usb_otg_periph_signal.module); before reboot
• Try to deep sleep for 5s (I was thinking maybe this would power cycle the USB peripheral)
• usb_del_phy before reboot.

I made a minimal repro case to demonstrate the issue: https://github.com/johnboiles/esp32-tinyusb-reset-bug

The only workaround we've found is to hook up a GPIO to the RESET pin and force the chip to reset itself. But it seems like there's probably just some register somewhere that needs to get reset at shutdown time.

I've been doing my testing on IDF 5.1 (cbce221).

The problem seems to be where in usb_new_phy since if you remove that line the USB/JTAG doesn't go away https://github.com/espressif/idf-extra-components/blob/master/usb/esp_tinyusb/tinyusb.c#L58

[johnboiles]

I think I got it! TL;DR is something like this:

#include <esp_private/usb_phy.h>
extern usb_phy_handle_t *phy_hdl_ptr;

usb_del_phy(*phy_hdl_ptr);
usb_phy_config_t phy_conf = {
    .controller = USB_PHY_CTRL_SERIAL_JTAG,
};
usb_phy_handle_t jtag_phy;
usb_new_phy(&phy_conf, &jtag_phy);

Where phy_hdl_ptr is a global from TinyUSB by adding this line to tinyusb.c (since you can't extern the static usb_phy_handle_t phy_hdl; from tinyusb.c)

usb_phy_handle_t *phy_hdl_ptr = &phy_hdl;

So now the question is how to we productionize this. Probably if there were a tinyusb_driver_uninstall in tinyusb.c that would be enough.

[johnboiles]

PR to propose tinyusb_driver_uninstall: espressif/idf-extra-components#229

[tore-espressif]

@rikaiot thank you for reporting and @johnboiles thank you for your thorough investigation!
I'll review your changes ASAP. I need to discuss with colleagues whether we want to change (fix) the behaviout of esp_restart(), too

[tore-espressif]

Hello @johnboiles here is some update:

The restart behavior you see is actually intentional. If a user wants to perform a software reset, he usually does not want to re-enumerate as different device (JTAG debug) during the process.

Moreover, changing this would be a functional breaking change, so we will fix this with the changes you proposed in esp_tinyusb component.

Even more info:

• Software reset (esp_restart()) and WDT restart are not considered the same. The software reset is deemed as regular software flow, while WDT timeout is considered panic abort
• You can read more about SW reset here and about WDT timeout here