On Wed. 24 Nov. 2021 at 06:10, Oliver Hartkopp [off-list ref] wrote:

On 23.11.21 12:53, Vincent Mailhol wrote:

quoted

There are two common errors which are made when reporting the CAN RX
statistics:

   1. Incrementing the "normal" RX stats when receiving an Error
   frame. Error frames is an abstraction of Socket CAN and does not
   exist on the wire.

   2. Counting the length of the Remote Transmission Frames (RTR). The
   length of an RTR frame is the length of the requested frame not the
   actual payload. In reality the payload of an RTR frame is always 0
   bytes long.

This patch series fix those two issues for all CAN drivers.

Vincent Mailhol (2):
   can: do not increase rx statistics when receiving CAN error frames
   can: do not increase rx_bytes statistics for RTR frames

I would suggest to upstream this change without bringing it to older
(stable) trees.

It doesn't fix any substantial flaw which needs to be backported IMHO.

I fully agree. Bringing it to the stable trees would be a
considerable effort and was not my intent either (thus the
absence of "Fixes" tags).

Btw. can you please change 'error frames' to 'error message frames'?

We had a discussion some years ago that the 'error frames' are used as
term inside the CAN protocol.

ACK. Thanks for the clarification on the vocabulary.

Yours sincerely,
Vincent Mailhol

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel