Faulty UART baudrate divisor formula about rp-hal HOT 2 OPEN

Hi @KisImre!

Thanks for this great analysis. How did you even notice this? I'd guess that the difference is small enough that it doesn't impact the usability of the UART. (https://pdfserv.maximintegrated.com/en/an/AN2141.pdf)

Impressive that you didn't stop with pointing out the bug in the source code, but also confirmed it by measuring the actual timing on the wire.

Given the comment above that code ("Code inspired from the C SDK.") I wonder if the C SDK has the same issue. Did you check that? if it's present there as well, do you want to file a bug yourself, or should I do it?

from rp-hal.

I just noticed the affected line while I was tinkering with the UART and the (... + 0x7F) + 1 part seemed like a place where an overflow can happen. I put together the attached example to verify my theory and then tested it in the hardware just to see it in practice.

I guess it doesn't affect much on lower baudrates because the error is approx. 1 / divint. However, for a faulty baudrate in the 500000 baud range the error would be around 6% which is not good.

Now I tested the C SDK too, which indeed has the same bug. I will open bug report in their repo as well.

from rp-hal.