Subject: Re: Cache Chip Bug
To: Chris Torek <torek@BSDI.COM>
From: David Brownlee <abs@netbsd.org>
List: port-sparc
Date: 07/07/2000 10:05:20
On Fri, 7 Jul 2000, Chris Torek wrote:
> >from sys/arch/sparc/sparc/cpu.c
> >
> >/* Machines with "buserr-type" 1 have a bug in the cache chip that affects
> >traps. */
> >
> >The "cache chip bug; trap page uncached" message is the kernel activating
> >a workaround for this bug.
>
> For whatever it is worth, this comment and code is mine. I put it in
> on, as I recall, advice from John Gilmore, who said that some sun4c
> CPUs had this bug. The "buserr-type" test was my best guess at how
> such CPUs were identified. I have never seen any official document
> saying whether this is actually right, but it did seem to work.
How much of a performance penalty is the workaround and what sort
of symptoms does the bug exhibit otherwise?
(Just curious if its worth testing without the workaround on some
sun4c boxes :)
David/absolute
-- www.netbsd.org: No hype required --