Re: 2.1.111: IDE DMA disabled...BLAH...BLAH...

Rogier Wolff (
Thu, 30 Jul 1998 00:51:17 +0200 (MEST)

Linus Torvalds wrote:
> On Wed, 29 Jul 1998, Mark Lord wrote:
> >
> > These problems are probably solvable by removing "SMP=1".
> > The exact cause is probably not the IDE stuff, but rather some
> > as-yet-undebugged interaction with SMP.
> You still choose to ignore the other report that said that corruption
> happend with UP also?


Mark has a few reports now. He's working on it. Don't think that's
resolved in a few hours. That takes days. I hate to say it, but people
tend to be pretty unreliable in reporting stuff.

For the security list, I once tried having bugs verified by a team of
volunteers. Both the "real problems" and the "not an issue" came back
with about a 50/50 of "Yes I can reproduce this" and "Nope script
doesn't work on my machine". Really people run the exploit script as
root to start with and report that it broke root on their machine.....

That's why an isolated (I remember you saying: "Just one is all it
takes") report shouldn't be taken too seriously. All reports should
however be examined to the fullest possible extent. Some of them
will not be resolved ("Sorry, have to run production, machine is now
running freeBSD").

Mark is doing that right now. Alan is collecting statistics.
Diagnosing vague hardware-related statistical problems is very hard.
I trust YOU, Alan and Mark to be able to do this kind of stuff right.
And just about everybody else needs guidance to do it right.

Your arguments are based on the "fact" that you got some reports that
disabeling DMA helped. Just those reports are not strong enough. They
need to be thoroughly investigated. Mark is working on it. Let him do
his "job". Yes, you got his attention by now.

Mark, Alan, I think you're approaching this from the right angle.
Linus: You have every right to say: "And now it's been enough, this
issue needs resolving NOW." You've prodded a few people into action,
you've gotten a few people to report their problems. Lets cool down
and find the real cause of this problem.

About the technical stuff: If indeed there is something with cache/DMA
controller/PCI bus going wrong, "AFTER" the CRC, then I'd say that
this would be likely to occur on any DMA. In that case turning off
IDE-DMA would only reduce the frequency of the problems. Does that
sound like an acceptable way of "fixing" linux to you? It doesn't to
me. I don't buy the possibility that several different modern UDMA
chipsets are all broken. Those have had a pretty serious workout
by now: the bugs should be eliminated nowadays.


Roger Wolff.

Actor asks a collegue: "To what do you owe your success in acting?"
Answer: "Honesty. Once you've learned how to fake that, you've got it made."
-------- Custom Linux device drivers for sale! Call for a quote. ----------
Email: || Tel: +31-15-2137555 || FAX: +31-15-2138217

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