Re: IDE disk corruption - A7M266-D; which kernel patches? or other solution
On Sun, Jan 19, 2003 at 05:53:54PM -0700, Al Davis wrote:
> I was getting disk corruption with an older kernel (2.4.8 or
> 2.4.18, Mandrake). A colleague said it might have something
> to do with a hardware bug (south bridge VIA VT82C686). He also
> said there was a fix in recent kernels. I thought it was bad
> memory. It did have CONFIG_BLK_DEV_VIA82CXXX=y . The
> workaround was to disable DMA. As expected, it has now been
> reliable but slow.
>
> I have since installed Debian, with 2.4.20-bf2.4, and now I
> wonder if it is safe to re-enable DMA.
I have a VIA 82C686 southbridge, and 2.4.20 enables its "VIA
southbridge workaround" when it boots. So presumably the problem has
been addressed to some extent. So far I've had no corruption, but "so
far" is only about a week with this MB.
I only noticed that boot message today; it may therefore be the case
that earlier kernels had a workaround too; but maybe 2.4.20's has been
updated.
My previous MB had a VIA VT82C596B. Using the same HD, I had ONE
instance of disk corruption - something scribbled all over the boot
sector of the DOS partition while Linux was running. That was either
with kernel 2.2.12 or 2.4.8. It only happened the once, so I never
found out what actually caused it.
I think there are a lot of VIA chipsets out there but most people
don't have problems; presumably some other factor is needed to make
the bug show up. Wonder what?
Pigeon
Reply to: