Cellular Network Infrastructure » Radio Access Network » OsmoPCU

There is at least one corner cases, which I'ven't found in the 44.060: What happens when an EGPRS ACK/NACK does not have a Length field, but the uncompressed RBB is less the remaining rest_bits? This was possible, when less than 8 bits were remaining, so only happens when having huge packet loss.

... This is defined as spec compliant. Out of window bits should be ignored.

I've implemented a fix for this in my branch lynxis/gprs-ack-nack https://gerrit.osmocom.org/plugins/gitiles/osmo-pcu/+/lynxis/gprs-ack-nack
I will push this patch to gerrit and write test cases for the encoding.

However I want also validate the encoding of the URBB is correct at all.

lynxis wrote:

There is at least one corner cases, which I'ven't found in the 44.060: What happens when an EGPRS ACK/NACK does not have a Length field, but the uncompressed RBB is less the remaining rest_bits? This was possible, when less than 8 bits were remaining, so only happens when having huge packet loss.

... This is defined as spec compliant. Out of window bits should be ignored.

But those bits are still within the window. They won't be ignored. And bits which represent V(R) or higher must be set to 0 (44.060 9.1.10).

The uncompressed bitmap will be removed.

https://gerrit.osmocom.org/c/osmo-pcu/+/14697

The patch is now merged.