On Wednesday 04 July 2001 16:26, Stephen Poynter wrote:
> From my understanding the VIA corruction bug only effects the 686B
> southbridge, which would indeed include the KT133, KX133A, and
> Apollo Pro 133 chipsets. However, I am using one of the early
> revisions of the EPOX 7KXA board that uses the 686A southbridge,
> and not the 686B.
True, however, we were uncertain how far this bug penetrated. There
have been MANY workarounds for VIA bugs, most especially for ABIT
boards, and it is not limited to one Chipset. So it was very short
time to release when the bug was finally admitted, and the big gun of
our kernel hackers used a broad brush -- most VIA are running on PIO
regardless of what the kernel messages say. Try hdparm -t to get a
good read. The kernel is not crippling the hardware, it is crippling
itself.
> Now this brings me to a question. Can I safely
> pass the following option at the lilo prompt 'linux ide0=ata66' at
> boot time? Is this a acceptable solution? I have done this
> without any noticeable problems. This did improve my performance
> according to 'hdpam -Tt'. What also is weird is that when I do
> 'cat /proc/ide/via' the output says that my drive is using a 40w
> cable when I know for a fact that I installed a brand new 80w when
> I built the computer. I do get the following message in my dmesg
> log:
> VP_IDE: ATA-66/100 forced bit set (WARNING)!!
>
I suppose it is as safe as it was before the bug announcement for you
to do it now. You might get better results with kernel-linus, and
you definitely would have better resuts with 2.2.19-10mdk
Every Chipset has its problems, and even CPUs do as well, so it is
not that the others are much better.
Anyway, when a bug is publicly confirmed by a manufacturer it is
serious news. Lots of intermittent hardware bugs just float, their
behavior indistinguishable from the normal behavior of some popular
OSes. Intel admitted the f00f bug when they had a fix ready for it,
and the 6x86 coma bug was never admitted, AFAIK, by Cyrix, and the
AMD K6 bug which had no workarounds? As I say it is serious news and
what is admitted is rarely the full scope and extent.
First it was " this happens only with Creative Sound Cards", then it
was "This happens mostly with Mobos manufactured by companies whose
engineers misinterpreted our advisories on avoiding problems with
Creative Sound Cards"
What will we hear next? Is it too much to ask, "Yes, here is what we
know and we'll help you folks work out a fix."
Civileme
