On Fri, Jun 27, 2025 at 2:53 PM Stefan Sperling <stsp@stsp.name> wrote:

> Thanks for testing! So it seems no fix is needed for pcn(4) after all.
>

Yes.

What I actually managed to get is tons of "pcn0: overflow error"
(printf located at the string 1294, function pcn_rxintr() in if_pcn.c rev
1.50)
when I revert your iperf scenario:
I start 'iperf -s -u' on a machine with the pcn0 interface.
And an iperf client with '-u -l0 -t0' on another machine.
My UTP cable is okay (both sides show 100-FDX), connection is direct
between two hosts
and netstat shows collision errors in the same time.
But the pcn interface is still okay after the test.
Maybe exporting complete statistics via kstat will help to split hairs in
error flavors,
but pcn hardware is way too old for this.
I of course think it is completely orthogonal with your tx_intr
investigations.