Serial Flush blocks forever but not only at programm start

[WMuellSika]

Hi there

I have a similar issue to #597, but not at the start of my program.

In my code, there is a function which looks like this:

void SlaveWaterCOM::send() {

    digitalWrite(2, HIGH);

    Serial1.write(outbuffer.dataarray, 28);
    
    Serial1.flush();

    digitalWrite(2, LOW);
}

The function communicates via the Serial1 over a MAX3075, and therefore it needs an enablepin for the driver.

In my program, this is the only point I use the Serial1.write or the Serial1.flush command.

Sadly, the program sometimes hangs at the Serial. Flush() command.

I tracked the issue down to the Sercom->USART.INTFLAG.bit.TXC flag, which seems to never be written.

Has anyone a clue why this could be?

[Duality4Y]

I have this exact same issue, it hangs on communication using a sercom module when calling flush() at some point,
I ALSO see with a debugger that the TXC bit is NOT set.
(JLink + gdb )

But In my specific case it seems to also be related to putting the SAMD to sleep, as when I disable this functionality it seems to run fine and stable.

It doesn't happen on the first wake-up and sleep cycle but somewhere down the road at some Time X
But that is in my case.

Here is an observation I made in my case the serial communication is with a modem,
I saw It had transmitted an A and it was about to transmit an T,
This was AT as it is a modem command it makes sense, what doesn't really make sense to me is where it went wrong,
the previous line was a println containing something like "AT+COPS" but somewhere between A and T the TXC bit was cleared from the interupt enable register and caused the following flush to hang.