Return to BSD News archive
Received: by minnie.vk1xwt.ampr.org with NNTP id AA841 ; Mon, 08 Feb 93 10:02:27 EST Path: sserve!manuel.anu.edu.au!munnari.oz.au!hp9000.csc.cuhk.hk!saimiri.primate.wisc.edu!sdd.hp.com!nigel.msen.com!yale.edu!spool.mu.edu!howland.reston.ans.net!paladin.american.edu!gatech!news.ans.net!cmcl2!sbcs.sunysb.edu!stark.UUCP!gene From: gene@stark.uucp (Gene Stark) Newsgroups: comp.unix.bsd Subject: Re: [386BSD] kernel on fixit-0.2 and Maxtor Date: 7 Feb 93 08:44:31 Organization: Gene Stark's home system Lines: 53 Message-ID: <GENE.93Feb7084431@stark.stark.uucp> References: <1993Feb3.160736.26695@cm.cf.ac.uk> <GENE.93Feb3190027@stark.stark.uucp> <44594@sdcc12.ucsd.edu> <44595@sdcc12.ucsd.edu> NNTP-Posting-Host: stark.uucp In-reply-to: cg18fbi@icogsci5.ucsd.edu's message of 6 Feb 93 02:50:22 GMT Rick Dante writes: >FOr some reason the beginning of my last post got cut off. What I said was >that I disabled my external cache today (which cut my CPU performance in HALF >:( on mmy 386/33) but my compiler woes went away. I can compile the kernel >from scratch without rebooting.. Since I've been using 386BSD since july, I >figured that if I haven't known the cache to be a potential problem with >system stability, then maybe others don't know as well. Hence this post. > >Gene: is it a surefire solution to replace the cache chips? Maybe you could >elaborate. All I really know is what happened to me, and here it is: I bought my system (486DX/33, 64k external cache, 4MB ram) from "Dee One Systems" (advertises in back of Computer Shopper) last August, with the specific intention of running 386BSD 0.1, if possible. I got my system and loaded up 386BSD. I expected some flakiness, but after I got familiar with the system I realized that my kernel compiles were often dying randomly with SIGBUS and SIGSEG errors. I added 4MB more memory somewhere along the way, that made things faster but didn't fix the problem. Then I read somebody's posting on the net suggesting that there might be a problem with sub-spec cache chips on some systems and that I should try disabling the external cache. I did, and it fixed the problem. Then, I took a look at the cache chips to try to figure out what was going on. As near as I could figure, they had installed 50ns cache chips when the motherboard manual they sent called for 30ns chips. I wasn't absolutely sure, because I couldn't easily get the databook for that chip manufacturer, but they did have a "-5" suffix. So, I figured it was worth risking a few bucks on new cache chips. I called up a memory chip outfit and for < $50 I got 128K worth of chips that were faster than what I needed (I forget the numbers). The hardest part was replacing the chips, since in my system they riveted the floppy drive bays in over the motherboard in such a way that the latter cannot be removed, and leaving only a small amount of clearance within which to stick a screwdriver and a pair of needle nose pliers. Well, I did manage to change the chips and jumpers without incident, but it was tedious. After the new chips were in, the system worked perfectly and I haven't had a similar problem since. I have a suspicion that some vendors out there may find that sub-spec cache chips don't cause any problem under DOS, so they figure they can install cheaper chips and nobody will ever know. It may be that the problems only show up when you run 386BSD and are doing a lot of cache flushing or something. Maybe other people that have had this problem will share their experiences so we can let vendors know that they aren't going to get away with this. - Gene Stark -- stark@cs.sunysb.edu