Subject: Re: Problems with HDs and VIA Chipsets
To: None <port-i386@netbsd.org>
From: Bernd Ernesti <netbsd@arresum.inka.de>
List: port-i386
Date: 04/13/2001 00:38:17
On Thu, Apr 12, 2001 at 03:59:50PM -0400, Thor Lancelot Simon wrote:
> On Thu, Apr 12, 2001 at 07:05:29PM +0200, wojtek@3miasto.net wrote:
> > > lead to data loss or corruption or even to system crashes (windows
> > > oriented sites). It seems like additional DMA traffic can enhance the
> > > problem, soundblaster cards are known to very problematic.
> > >
> > > It looks like deactivating "PCI Delay Transaction" and "PCI Master
> > > Read Caching" and setting "PCI Latency" to a value of 0 to 32 will
> > > eliminate the problem also, at the cost of some performance.
> >
> > this is hardware problem of VIA chip i think. with linux you need to do
> > the same.
>
> If so, I strongly suspect that deactivating *just* "PCI Master Read Caching"
> will fix the problem.
I don't know if there is an english tranlation available, but here is the german
webpage which documents the problem:
http://www.au-ja.de/review-kt133a-1.html
From the solution part of this article (http://www.au-ja.de/review-kt133a-4.html):
--------------
Um einen sicheren Betrieb der VIA Southbridges (insbesondere der 686B) zu garantieren,
müssen die Paramater folgendermaßen gesetzt werden:
PCI Delay Transaction = 0 (aus)
PCI Master Read Caching = 0 (aus)
PCI Latency = 0 (Werte von 0 bis 32 scheinen sicher zu sein, alles über 32 nicht!)
--------------
You have to set the following parameters to ensure a save operation of the VIA
Southbridges (in particular the 686B):
PCI Delay Transaction = 0 (off)
PCI Master Read Caching = 0 (off)
PCI Latency = 0 (Values from 0 till 32 seems to be save, all over 32 are not!)
Bernd