Return to BSD News archive
Xref: sserve comp.os.386bsd.bugs:2084 comp.os.386bsd.questions:8550 Path: sserve!newshost.anu.edu.au!munnari.oz.au!uunet!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!news.intercon.com!udel!news.sprintlink.net!news.clark.net!news.clark.net!not-for-mail From: ack@clark.net (Eric S. Hvozda) Newsgroups: comp.os.386bsd.bugs,comp.os.386bsd.questions Subject: Re: Protected Mode Lockup Problems--Please Help. Date: 8 Feb 1994 23:47:47 -0500 Organization: Clark Internet Services, Inc. Lines: 27 Message-ID: <2j9ptj$qk4@explorer.clark.net> References: <199402060404.XAA21622@freenet.buffalo.edu> <2j7ogh$pv@klaava.helsinki.fi> <2j8vqf$q2a@explorer.clark.net> NNTP-Posting-Host: explorer.clark.net Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit In article <2j8vqf$q2a@explorer.clark.net>, Eric S. Hvozda <ack@clark.net> wrote: > >[..lock-ups with i386 with i387 installed..] > >It's probably the i387. There are a set of 4 conditions, if met, will lock >a i386 if a i387 is installed. Basically the 386 ends up waiting on the 387 >and vice versa. The gory details are covered in "Repairing and Upgrading PCs" >written by Mueller and pubished by Que. > >I will post the 4 conditions when I gte home from work... As promised, from the middle of page 428: Demand page virtual memory must be active A 387DX must be installed and in use DMA must be in progress The 386 must enter a wait state Supposedly, if all of these conitions happen on the B stepping version of the 386DX, the 386 and 387 endup waiting on eachother forever, hence the lockup. Intel published this problem as Errata 21. Usually this problem is only noticed when OS's like OS/2, Unix, etc are used. I'm pressed to imagine how DOS would tickel this nifty bugger... -- Ack! Creek, not creek; Pop not soda; Car needs washed...