Please, Mark, drop this theory of yours.
Yes, the timings are the same _on_the_wire_.
That's like saying that because 10Mbps ethernet has the same wire speed
regardless of the controller, you have the same amount of corruption,
regardless of whether the controller has it's own bugs or regardless of
whether how the data is read from the controller.
The timings in the _system_ are very different when you use DMA. Very
different indeed.
> 4. If for some reason PIO works, but DMA produces corruption
> (there has never been a verified case of this yet),
Well, your definition of "verified" seems to be fairly strange.
> then perhaps the only possible way this might occur would be due
> to the IDE DMA controller performing burst PCI accesses during the
> data transfer. Thus, the error would be on the PCI bus, rather
> than in the IDE subsystem itself. Turning off IDE DMA may alleviate
> the problem, but any other PCI burst activity will likely also result
> in memory corruption for the same reason, eventually corrupting
> filesystems.
BINGO!
I've _never_ blamed the driver. I've blamed the fact that we do DMA.
Whether the bug is on the motherboatd, the IDE controller, the memory
settings etc is irrelevant.
> > You should have noted that I at no time blamed the driver itself. I don't
>
> Quoted from linux-2.1.111:
>
> "# Either the DMA code is buggy or the DMA hardware is unreliable.
> FIXME."
Did you notice the "or the DMA hardware is unreliable" part at all?
My personal suspicion is that it's the hardware. I'm willing to entertain
the notion that the Linux driver is buggy, but I'm also very willing to
take your word for that not being the case.
> "We" do not have any such reports yet.
> Perhaps "you" have some, thoughm, please pass them along.
I've seen at least three emails that you were Cc'd on that either said
that they had documented corruption with DMA (one of them was Doug
Ledford, whom I trust and know personally) or said that there are windows
drivers that actually blacklist some devices from using DMA with the
standard timings.
Why do you chose to ignore those?
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html