Subject: Re: ahc bug in current? (was: ccd changed in current?)
To: Chris G. Demetriou <cgd@pa.dec.com>
From: None <tober@albino.ir.bbn.com>
List: current-users
Date: 08/29/1997 16:49:26
> > While this may or may not be the problem, you should know that the DECchip
> > 21050 is a first-generation PCI-to-PCI bridge and contains a serious design
> > flaw.  It is susceptible to deadlock under certain conditions.  Under certain
> > cirumstances where a downstream and an upstream transaction are trying to
> > proceed through a 21050 simultaneously, neither transaction will ever proceed.
> 
> Actually, I was under the impression (from discussions i'd seen about
> the bug over a year ago) that the problem was specific to the Intel
> Triton chipset, and was in fact a Triton bug which had been worked
> around in later revs of the PPB...
> 
> However, my understanding could well be out of date, since it was
> pre-PCI2.1...
>
Well, I'm not qualified to say what deadlock problems the Triton chipset may
or may not have (it probably has some bugs), but I can say for certain that
we experienced crippling deadlock problems when using 21050s with the
AXPpci33 (Digital Alpha 21066 motherboard, which certainly doesn't use the
Triton chipset :) ).  Does anyone know what PPB Adaptec is now using on the
3940UW?  One would hope that's it's no longer the 21050...
-ben