Re: kern/38791 (routing socket event header not cleared)

[Michael van Elst <mlelstv%serpens.de@localhost>]

The following reply was made to PR kern/38791; it has been noted by GNATS.

From: Michael van Elst <mlelstv%serpens.de@localhost>
To: "J.T. Conklin" <jtc%acorntoolworks.com@localhost>
Cc: gnats-bugs%NetBSD.org@localhost, kern-bug-people%NetBSD.org@localhost, 
netbsd-bugs%NetBSD.org@localhost,
        gnats-admin%NetBSD.org@localhost, mlelstv%NetBSD.org@localhost
Subject: Re: kern/38791 (routing socket event header not cleared)
Date: Mon, 2 Jun 2008 21:11:02 +0200

 On Mon, Jun 02, 2008 at 12:05:16PM -0700, J.T. Conklin wrote:
 > mlelstv%NetBSD.org@localhost writes:
 > > Synopsis: routing socket event header not cleared
 > >
 > > State-Changed-From-To: closed->open
 > > State-Changed-By: mlelstv%NetBSD.org@localhost
 > > State-Changed-When: Sun, 01 Jun 2008 16:51:01 +0000
 > > State-Changed-Why:
 > > The patch erases the whole message. Maybe clearing sizeof(rtm) is enough?
 > 
 > True, however this is essentially the same thing that callers to rt_msg2() 
 > do (to they structure they pass) when they build a route event.  Clearing
 > the whole message could conceivably fix problems where not all the event/
 > message payload data is being being zeroed too (althought this may not be
 > likely).
 > 
 > The total length of these messages are quite small, so any the extra cost
 > to zero the whole event should be similarly small.
 
 Let me clarify this: by erasing the message you erase the contents
 of the message that has been copied there two lines above with
 m_copyback().
 
 E.g. in case of RTM_IFINFO messages you overwrite the whole message
 with zeroes and thus break ifwatchd.
 
 Greetings,
 -- 
                                 Michael van Elst
 Internet: mlelstv%serpens.de@localhost
                                 "A potential Snark may lurk in every tree."