Re: Talking of Failing Drives

["Bryan J.Smith" <b.j.smith@ieee.org>]

Herbie wrote:
> I'm having one go right now :(
> Why is it with so many IDE drives the first thing to go is DMA?

Actually, that's not typically the drive mechanics failing.
That is typically a kernel ATA issue.

Long story short, an overwhelming number of ATA chipset and HD vendors 
don't follow the specs.
The result is that the Linux kernel takes issue with that, because the 
developers are not privy to that information.

It's the #1 reason I use 3Ware cards for ATA.
Put the intelligent on a card and the issues are far less.
Which is why SCSI is so much easier at times.

> At work I have several machines with drives that attempt to, and fail 
> to, go into DMA and then fall back into non-DMA and run quite happily.

Of course.
Because your CPU drives them in PIO mode.
In DMA mode, the drive attempts to directly transfer to memory.
And ATA isn't intelligent.

> Is this a function of drive or chipset?

Drive.
The chipset is just a bus arbitrator.
Of course, if the chipset and drive don't adhere to the same specs, 
that's part of the problem.

Hendrick gets really frustrated dealing with ATA vendors for a reason.
-- 
Bryan J. Smith   mailto:b.j.smith@ieee.org
Currently Mobile   Treo 600

-
To unsubscribe, send email to majordomo@luci.org with
"unsubscribe luci-discuss" in the body.