Date: Mon, 21 Oct 1996 10:25:18 -0400 From: Garrett Wollman <wollman@lcs.mit.edu> To: Mikael Karpberg <karpen@ocean.campus.luth.se> Cc: current@FreeBSD.org Subject: Re: ISDN code removal, final warning. Message-ID: <9610211425.AA12165@halloran-eldar.lcs.mit.edu> In-Reply-To: <199610210016.CAA11690@ocean.campus.luth.se> References: <326A83CA.167EB0E7@whistle.com> <199610210016.CAA11690@ocean.campus.luth.se>
next in thread | previous in thread | raw e-mail | index | archive | help
<<On Mon, 21 Oct 1996 02:16:02 +0200 (MET DST), Mikael Karpberg <karpen@ocean.campus.luth.se> said: > Streams should be wonderful, no? So you could just plug a new protocol in, Well, only for values of ``wonderful'' meaning ``dreadfully slow''. Layerism really, really gets in the way of efficient networking implementations. If someone were to make a really fast TCP, it would sit right underneath sosend() and above ifp->if_output(). (Van Jacobson has allegedly done something like this, but nobody outside of LBL appears to have seen the actual code.) The actual buffering model used by STREAMS is even worse than the buffering model used by BSD, which is itself pretty horrid. As for the original question of ISDN: probably there needs to be some in-kernel support for the basic data-transfer functions. All of the signalling, however, should be done outside of the kernel. If this means you have to introduce a new isdnsrvr() system call, so be it. (Although I would prefer a generic kernel-user message-passing interface modeled along the lines of PF_ROUTE and PF_KEY, only completely general. Then it could be used to eliminate such things as /dev/klog.) -GAWollman -- Garrett A. Wollman | O Siem / We are all family / O Siem / We're all the same wollman@lcs.mit.edu | O Siem / The fires of freedom Opinions not those of| Dance in the burning flame MIT, LCS, ANA, or NSA| - Susan Aglukark and Chad Irschick
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9610211425.AA12165>