Date: Fri, 03 Mar 1995 21:14:09 -0800 From: David Greenman <davidg@Root.COM> To: Poul-Henning Kamp <phk@ref.tfs.com> Cc: CVS-commiters@freefall.cdrom.com, cvs-sys@freefall.cdrom.com Subject: Re: cvs commit: src/sys/net if_loop.c Message-ID: <199503040514.VAA00190@corbin.Root.COM> In-Reply-To: Your message of "Fri, 03 Mar 95 21:08:07 PST." <199503040508.VAA13824@ref.tfs.com>
next in thread | previous in thread | raw e-mail | index | archive | help
>> davidg 95/03/03 20:28:51 >> >> Modified: sys/net if_loop.c >> Log: >> Reduced loopback MTU to 16k to work around a miriad of problems with it >> set near or above 32k (likely caused by 16bit signed word overflow). 16k >> actually (surprizingly) has higher performance than other values I tested. > >Probably because it matches our default TCP-window, thus making it possible >to return to user-space and effectively run in parallel. An interesting thought...but I think it's much more complicated than that. Even with the MTU at 16k, there are still some anomolous things about the traffic patterns through the interface. The segment size varies from 2k to 14k, for instance - it never actually sends 16k. All of this needs to be looked at carefully; the above change was purely pragmatic. -DG
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199503040514.VAA00190>