Re: Corruption Stats (fwd)

Linus Torvalds (
Wed, 29 Jul 1998 12:17:47 -0700 (PDT)

On Wed, 29 Jul 1998, Mark Lord wrote:
> This one is another SMP motherboard.

With a UP kernel. Which means that Linux won't even _touch_ the other CPU,
or even know that it is there at all.

> I wonder if there are any non-SMP reports at all?

Don't blame SMP if it isn't even enabled.

You can blame motherboard bugs, and that is exactly what I've been trying
to get you to do - disable DMA if there are bugs in the hardware, and
ADMIT that there are conditions where DMA is unreliable.

> WD has a non-announced bug with UDMA support on certain models,
> which I just found out about from reading some "semi-secret"
> material sent to me from a vendor of IDE-interface boards.

THANK YOU THANK YOU THANK YOU, for finally admitting what people have been
trying to tell you all along.

What I don't like is that you only acknowledge even the _possibility_ of
bugs like this once you get a paper saying so, ignoring the reports of

> I suppose we'll need to blacklist those models permanently.

Good. The second you send me a blacklist, the DMA code gets immediately
enabled again. The blacklist doesn't need to be complete - we can fill it
up as we hear about more problems (SCSI has had this kind of blacklist for
a long time, because even thought the specs clearly specify what a SCSI
device should do, there are buggy devices out there that react badly to
certain things).


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at