Date: Wed, 2 Sep 2009 20:23:29 GMT From: Chris Smowton <cs448@cam.ac.uk> To: freebsd-gnats-submit@FreeBSD.org Subject: kern/138485: BPF and raw-IP man pages missing important corner cases Message-ID: <200909022023.n82KNTOX046150@www.freebsd.org> Resent-Message-ID: <200909022030.n82KU8Bc018100@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 138485 >Category: kern >Synopsis: BPF and raw-IP man pages missing important corner cases >Confidential: no >Severity: non-critical >Priority: medium >Responsible: freebsd-bugs >State: open >Quarter: >Keywords: >Date-Required: >Class: doc-bug >Submitter-Id: current-users >Arrival-Date: Wed Sep 02 20:30:08 UTC 2009 >Closed-Date: >Last-Modified: >Originator: Chris Smowton >Release: Any >Organization: N/A >Environment: N/A >Description: Both the man pages concerning raw-IP and Berkeley Packet Filter are missing details of the following important corner cases: how should the kernel behave when faced with a packet which could be handled by more than one handler? In the raw-IP case, what if there exists a raw-IP socket live on some interface, as well as a TCP socket relevant to an incoming packet? Does the IP or the TCP socket receive that packet, or both? In the BPF case, it's mentioned that the filter program may request to "ignore" a packet, but it isn't specified whether "ignored" packets, non-"ignored" packets or both will then go on to be handled by the kernel's ordinary TCP packet handler. Both of these things can be discovered by experimentation, but it's tedious that every person facing these questions needs to do that when a couple of lines in the docs could save them the trouble. >How-To-Repeat: Read the man pages :) >Fix: Amend the man pages :) >Release-Note: >Audit-Trail: >Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200909022023.n82KNTOX046150>