uac2_headset on STM32F401 - TU_BREAKPOINT() on USB reconnect

[elagil]

Operating System

Windows 11

Board

STM32F401 custom design

Firmware

Same as https://github.com/hathach/tinyusb/issues/1935 (CubeMX Makefile project on FREERTOS, using the new dcd_dwc2 driver)

Just to be sure, I tested without FREERTOS with a minimal application and nothing in the main loop but tinyusb. The behaviour is identical.

What happened ?

In general, I can connect the USB device and I can play music or loop it back to the microphone without issues. The looped back signal looks like the one I transmitted.

Now, there are two cases:

1. I pause/stop the music, disconnect USB, reconnect USB -> no problems
2. I leave the music playing, disconnect USB, reconnect USB -> I hit TU_BREAKPOINT() in dcd_dwc2.c:1085 with pktsts == 0 pktsts == 9 or pktsts == 11 (random).

Why could that happen, only when I reconnect when I previously had music playing? I do hit both the tud_umount_cb() when disconnecting, and the tud_mount_cb() when reconnecting, before I run into the breakpoint.

My tusb_config.h has __RX__ defined, so I limit the quality to 16 bit/48 kHz.

How to reproduce ?

See above (point 2.)

Debug Log as txt file

The log starts just before I disconnect (while playing music) and then reconnect. A lot of the configuration goes well, but it fails at getting the set volume: audio_log.txt

This is the log, when I reconnect without playing music: audio_log_no_error.txt

Please have a look at the difference between the two: https://diffonline.net/cHSN6gt3J5

When playing music while replugging, there are seemingly a lot more configuration messages exchanged - which might be normal. I have no experience with the protocol yet.

I have checked existing issues, dicussion and documentation

• [X] I confirm I have checked existing issues, dicussion and documentation.

[gabChouin]

I've hit the same issue with an STM32U575. I had similar behavior to what was described in https://github.com/hathach/tinyusb/issues/1935 but https://github.com/hathach/tinyusb/pull/1928/ allowed me to advance one step forward to this current bug (with the default settings of 96 kHz 24-bit). Do you have any updates on this issue ?

[elagil]

@gabChouin Unfortunately not. I gave up eventually and wrote my own UAC (1.0 only) driver.

[HiFiPhile]

Could you take a looks the version of dwc2 IP used in F401 ? Basically these registers : https://github.com/hathach/tinyusb/blob/d6b8b3515df4bb83b6038bd60260771e65af8de7/src/portable/synopsys/dwc2/dwc2_info.py#L6C4-L6C4

I'll see if it's reproducible on my F429 board.

[HiFiPhile]

Could you try #2333 ?

While I didn't hit breakpoint, sometimes after re-plug it NAK all OUT transfers (which will cause host to re-enumerate like in your log)

[HiFiPhile]

@rpascalsdl It's not the same issue. This one is specific to dwc2 IP, rp2040 use a different USB controller.

[rpascalsdl]

@HiFiPhile thanks. I removed the comment. Will try to create an issue when I have better repro steps.