Date: Tue, 27 Jul 1999 08:10:29 -0700 From: Mike Smith <mike@smith.net.au> To: Brian Somers <brian@FreeBSD.org.uk> Cc: cvs-all@FreeBSD.org, cvs-committers@FreeBSD.org Subject: Re: cvs commit: src/release/sysinstall tcpip.c Message-ID: <199907271510.IAA03379@dingo.cdrom.com> In-Reply-To: Your message of "Tue, 27 Jul 1999 10:01:12 BST." <199907270901.KAA01833@keep.lan.Awfulhak.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> > > You can blame the 6-byte MAC header for most of the headaches. > > > > Actually, I blame the way that layered protocol information is prefixed > > rather than postfixed in the datagram's wire format. It doesn't strike > > me as being at all well designed from a performance perspective > > (although mbuf chaining helps a lot). It would be possible to reduce > > both copying and checksum overheads by preallocating trailing space in > > the buffer based on the down-stack path (or even just a worst-case > > assessment of potential buffer growth), and adding an incremental > > checksum field. ... > Of course there's nothing wrong with guessing how much preceeding > space might be required and setting up the mbuf with an appropriate > unused gap at the front. The user-land mbuf code in ppp does this so > that it already has room for the protocol and address & control > fields and doesn't have to go off and find another mbuf. It gets a > bit muddy on I-must-be-aligned type architectures, but not really > much more than it would anyway. Well, no, that's just it; if you are on a must-be-aligned architecture, you start with the payload already aligned, so no matter what the ensuing encapsulation it stays aligned. -- \\ The mind's the standard \\ Mike Smith \\ of the man. \\ msmith@freebsd.org \\ -- Joseph Merrick \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199907271510.IAA03379>