nRF5340 UART Interrupt mode send example

[SergeySK]

Hi all!
I'm currently working on the UART on the nRF5340 board. The SDK describes three options:
-polling API,
-interrupt-driven API,
-asynchronous API.
With Polling and Asynchronous - everything is clear and they do not suit me for various reasons.
Interrupt-driven doesn't have any APIs for starting (sending the first byte).
Searched everything.
Please share a link to an example.

[ovrebekk]

Hi Sergey

Thanks for explaining the issue in more detail. 

The way we handle this issue using EasyDMA (and the issue of buffer overrun in general) is to make the DMA pointer double buffered. Once you have configured the first buffer through the RXD.PTR register (or by using a higher level driver to do the same), and started the RX reception, you can then proceed to write a second buffer to the RXD.PTR register. 

As soon as the first buffer fills up the hardware peripheral will seamlessly switch to the second buffer, meaning you can continue receiving without any gap in the communication. 

On completion of the first buffer you will get an interrupt from the peripheral, allowing you to set up the third buffer while the second buffer is being filled up. And so on. 

Based on the UART baudrate used and the number and length of interrupts in your system you can scale the RX buffers to ensure that an overrun will never happen. 

Best regards
Torbjørn

[SergeySK]

Hi Torbjørn
Of course, a double buffer is also a very good solution and of course I use it, but if the packets go one after the other, then two packets appear in one buffer, and often the packet is located simultaneously in two buffers, which is not very convenient.
Best regards, Sergey.

[ovrebekk]

Hi Sergey

Whether or not you use DMA this is a result of the asynchronous nature of UART communication. 

You need some kind of algorithm on the receiving side that goes through the incoming bytes one by one, and figures out what is the start and end of each packet, before the packet is sent to the application for processing. 

The only difference is whether each RX callback could contain 1 or multiple bytes. In the case of DMA you might get multiple bytes at once, which means you need to loop through each one and run the packetization algorithm like mentioned above. 

Best regards
Torbjørn

[SergeySK]

Hi Torbjørn

thanks for answers!

Best regards, Sergey.

[ovrebekk]

You welcome Sergey. 

The best of luck with your project