To test a USB implementation, I have implemented a USB loopback on a Nucleo-144 Development Board (MB1404C). Testing is showing a strange result that I have been unable to isolate:
The code is very basic: it creates a thread for USB data handling. In that thread, ux_device_class_cdc_acm_read is used to block until data is received from USB. When it is received, it is send back out USB using ux_device_class_cdc_acm_write.
This works great if I send up to 6 bytes of data. As soon as I send 7 or more bytes, the transmitted data is junk (though it is repeatable).
Things I have tried and seemingly ruled out:
- If I don't try to loop the received data out and instead just send data out, I can send 7+ bytes so it doesn't seem to be on the PC side of send/receive data or directly in the transmit code.
- I have stepped all the way through the receive and transmit code (down through the USB drivers) and the data is received correctly (even 7+ bytes) and it looks good the entire time on the transmit side.
- I doubled the size of all stacks in case it was some type of stack overflow.
- I also inserted unused variables/ rearranged the order of variable declarations to try to rule out some type of overflow issue.
Any ideas?
