Return to BSD News archive
Xref: sserve comp.os.386bsd.questions:3284 comp.os.386bsd.bugs:973 Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!howland.reston.ans.net!usc!enterpoop.mit.edu!ai-lab!hal.gnu.ai.mit.edu!calvin From: calvin@hal.gnu.ai.mit.edu (Eric John Vette) Newsgroups: comp.os.386bsd.questions,comp.os.386bsd.bugs Subject: Re: Patchkit 0.2.4 breaks ... UDP problems. Date: 25 Jun 1993 19:48:39 GMT Organization: University Telecommunications, C&IT, Wayne State University Lines: 48 Distribution: world Message-ID: <20fkqnINNd56@life.ai.mit.edu> References: <20aubn$oeq@thumper.cc.utexas.edu> <20f90tINNkn@unidoct.Chemietechnik.Uni-Dortmund.DE> NNTP-Posting-Host: hal.ai.mit.edu > >Also, (this is the last bug, I promise) nothing that uses UDP sockets seems > >to work over slip; archie, xarchie and others all hang, FSP works if I set > >FSP_BUF_SIZE to 512 (or numbers lower than 1024). Any fix for this? > > Well I don't think UDP is the problem; I've played netrek with a UDP > client on 386bsd and it worked like a charm. It's probably something else. > > -Sven > While it may not necessarily be the UDP driver that's a problem, the problem does exist. UDP packets over SLIP are really buggy; in general, the SLIP driver is buggy. Any type of upload, as far as I've been able to see, kinda gets "lost". Most packets from the BSD box to it's outside connection get lost (although some do get through - it kinda reminds me of egg fertilisation by sperm... ). Since UDP packets are heavy on the outgoing, programs like Xarchie and Xtrek get fed up with the poor performance and give up. Can't say I blame them. Try ftp uploading a file somewhere, and see what kind off performance you get. I know that at one time, this was working on my machine, and another friend's machine that I helped to set up. It is now broken on both of our machines, at different times (local Merit is probably not the problem, although it would be a good candidate). Considering how little he experimented with his machine, I find it hard to believe that it's due to operator-malfunction. I've considered manipulating the MTU size in the slip driver (if_sl.c ?) and recompiling to see if it made a difference, but time and machine have not been kind enough to allow me to. calvin -- ***************************************************************************** * * Eric Vette (Calvin) * * The views expressed here are not * Computational Biochemist * * those of the FSF, my employer(s), * calvin@gnu.ai.mit.edu * * myself, or anyone else! * calvin@axe.cit.wayne.edu * * * evette@cms.cc.wayne.edu * * * evette@waynest1 * ***********************************************************.sig*virus******** !! Science is the search for more things to search for... !!