*BSD News Article 68063


Return to BSD News archive

Newsgroups: comp.unix.bsd.freebsd.misc
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.uwa.edu.au!disco.iinet.net.au!news.uoregon.edu!vixen.cso.uiuc.edu!newsfeed.internetmci.com!in2.uu.net!news.nkn.net!news.panther.net!nemesis!uhclem
From: uhclem@nemesis.lonestar.org (Frank Durda IV)
Subject: Re: Linux vs. FreeBSD ...
X-Newsreader: Tin 1.1 PL5
Organization: The Big Blue Box
Message-ID: <Dr4CIs.88L@nemesis.lonestar.org>
References: <4mplph$n1s@agate.berkeley.edu>
Date: Thu, 9 May 1996 03:32:51 GMT
Lines: 36

Nick Kralevich (nickkral@america.CS.Berkeley.EDU) wrote:
:[2]According to Terry, 3 out of 5 EIDE interfaces have bugs.  If the 
:[2]FreeBSD core team is going to take the moral high ground and not try to 
:[2]work around those bugs ("it's a hardware problem, let the hardware guys 
:[2]fix it"), then I'm sure the Linux community would be more than happy to 
:[2]help out those users.

Well, since Intels own erattas say there is no way to work-around
this problem OTHER than to not use the EIDE modes at all (they say to
simply run the chip in IDE mode), I am sure Intel and everyone else
(including Microsoft) would be delighted if you can come up with a
software correction for this data-corrupting bug.   I guarantee Microsoft
will hold the next Service Pack for Windows '95 for your correction.

Now whether you can fix the PCI/EIDE signal cross-talk flaw associated
with the CMD chip in software another matter... 


We do read the erattas.  They say these modes are broken and to not use
them unless you want to risk data corruption. End of song.    If that 
is considered "taking the high ground", then perhaps that is a good thing.
I like my data intact.

At least Intel was a bit more forthcoming about admitting that 70%+ of
their commodity MLBs have this defective chip than they were about the
Pentium math flaws.   Now, they haven't done much in the way of offering
to replace the defective CMD part or to completely stop making boards with
it.  Now it we could only get SMC to confess (and fix) the
can't-work-around flaws in the Super I/O chip family 16550 implementations.


Frank Durda IV <uhclem@nemesis.lonestar.org>|"The Knights who say "LETNi"
or uhclem%nemesis@rwsystr.nkn.net           | demand...  A SEGMENT REGISTER!!!"
					    |"A what?"
or ...letni!rwsys!nemesis!uhclem	    |"LETNi! LETNi! LETNi!"  - 1983