Operating System
Windows 11
Commit SHA
Version bundled in framework-arduinoespressif32 version 3.20017.241212+sha.dcc1105b, "a0e5626bc" mentioned in docs
Board
Custom ESP32-S2 design, wifi inactive
Firmware
Custom firmware in PlatformIO env, sending 2x 64-byte raw HID packets every second. I haven't created a new minimal test case project yet, but I will try do so shortly.
What happened ?
Periodic HID output reports (from device to host PC) are sent in sets of 2, one right after the other, once per second. This works perfectly for between 30 seconds and 12 hours--maybe more sometimes, I haven't always been capturing log data. But then at some point, a Complete but not empty: 0/64 error is generated in TUSB:DCD code, and immediately afterwards, the next SendReport() call results in report 1 wait failed, and no more data can be sent out. Incoming data from the host PC to the device still flows normally. This condition is not fixed until I reset the MCU.
ASIDE FROM THE DATA IN THIS BUG REPORT, what I would most like to do is learn how to find and debug the code that generates the initial Complete but not empty: 0/64 error. I haven't been able to find it yet; it seems to be a message present only in the precompiled libarduino_tinyusb.a file in .platformio\packages\framework-arduinoespressif32\tools\sdk\esp32s2\lib\. I will gladly flesh out the bug report with more info if I can dig and test more. At the moment, I'm flying blind and only able to observe one (blocking) symptom of the problem.
How to reproduce ?
I don't have to do anything special in the custom firmware, just let it run until the problem appears. As described elsewhere, it is simply sending two 64-byte raw HID packets to the host every second; nothing really special. As soon as I get a minimal test project doing the same thing, I will post more detail.
Debug Log as txt file (LOG/CFG_TUSB_DEBUG=2)
Here is an example log capture, though this doesn't have as much as I would like to see. I have added the xSTres=%d output temporarily in USBHID.cpp to confirm the correct return code from the first xSemaphoreTake call in SendReport(), as I suspect some kind of race condition at the moment.
[2025-06-12 12:17:34.259] [1598029][I][USBHID.cpp:353] SendReport(): xSTres=0
[2025-06-12 12:17:34.259] [1598037][I][USBHID.cpp:353] SendReport(): xSTres=0
[2025-06-12 12:17:35.269] [1599040][I][USBHID.cpp:353] SendReport(): xSTres=0
[2025-06-12 12:17:35.269] [1599049][I][USBHID.cpp:353] SendReport(): xSTres=0
[2025-06-12 12:17:36.281] [1600051][I][USBHID.cpp:353] SendReport(): xSTres=0
[2025-06-12 12:17:36.281] E (1533379) TUSB:DCD: Complete but not empty: 0/64
[2025-06-12 12:17:36.281] [1600060][I][USBHID.cpp:353] SendReport(): xSTres=0
[2025-06-12 12:17:36.390] [1600170][E][USBHID.cpp:360] SendReport(): report 1 wait failed
...[no more outgoing SendReport() events]
The raw HID report uses Report ID 1. I have tried shrinking the raw HID report from 64 down to 32 bytes on a whim, but the problem still occurs.
Screenshots
No response
I have checked existing issues, dicussion and documentation
Operating System
Windows 11
Commit SHA
Version bundled in framework-arduinoespressif32 version 3.20017.241212+sha.dcc1105b, "a0e5626bc" mentioned in docs
Board
Custom ESP32-S2 design, wifi inactive
Firmware
Custom firmware in PlatformIO env, sending 2x 64-byte raw HID packets every second. I haven't created a new minimal test case project yet, but I will try do so shortly.
What happened ?
Periodic HID output reports (from device to host PC) are sent in sets of 2, one right after the other, once per second. This works perfectly for between 30 seconds and 12 hours--maybe more sometimes, I haven't always been capturing log data. But then at some point, a
Complete but not empty: 0/64error is generated inTUSB:DCDcode, and immediately afterwards, the nextSendReport()call results inreport 1 wait failed, and no more data can be sent out. Incoming data from the host PC to the device still flows normally. This condition is not fixed until I reset the MCU.ASIDE FROM THE DATA IN THIS BUG REPORT, what I would most like to do is learn how to find and debug the code that generates the initial
Complete but not empty: 0/64error. I haven't been able to find it yet; it seems to be a message present only in the precompiledlibarduino_tinyusb.afile in.platformio\packages\framework-arduinoespressif32\tools\sdk\esp32s2\lib\. I will gladly flesh out the bug report with more info if I can dig and test more. At the moment, I'm flying blind and only able to observe one (blocking) symptom of the problem.How to reproduce ?
I don't have to do anything special in the custom firmware, just let it run until the problem appears. As described elsewhere, it is simply sending two 64-byte raw HID packets to the host every second; nothing really special. As soon as I get a minimal test project doing the same thing, I will post more detail.
Debug Log as txt file (LOG/CFG_TUSB_DEBUG=2)
Here is an example log capture, though this doesn't have as much as I would like to see. I have added the
xSTres=%doutput temporarily in USBHID.cpp to confirm the correct return code from the firstxSemaphoreTakecall inSendReport(), as I suspect some kind of race condition at the moment.The raw HID report uses Report ID 1. I have tried shrinking the raw HID report from 64 down to 32 bytes on a whim, but the problem still occurs.
Screenshots
No response
I have checked existing issues, dicussion and documentation