Possible race condition bug in dummynet? (out of order packets)

I've discovered a problem with dummynet:

1) using io_fast dummynet option may lead to out of order packet delivery within one tcp or udp flow, and then degradation of tcp throughput or other errors.
2) it is unable to properly disable io_fast because this option is commented in dummynet source code.

Problem was described in pfSense bug tracker: https://redmine.pfsense.org/issues/11192

Since this may be serious problem not only for pfSense users, but everyone who use dummynet, I've decided to post it to FreeBSD community forum.
Maybe someone had encountered this problem on FreeBSD?
 
Back
Top