Re: 2.1.111: IDE DMA disabled?

Kristofer T. Karas (
Wed, 29 Jul 1998 15:19:06 -0400

Linus writes:
>I don't need to "measure" it.
>I have a very clear report of a person who could reproduce the behaviour
>at will.

So can I, on the P6DLE motherboard now collecting dust as I have too
many paperweights as it is. I can make IDE-DMA fail within an hour of
moderate use. And if I plug in an AGP graphics board, I can make that
fail in seconds. Both use bus-master DMA. Should I then disable all
attempts to use the AGP port in configurable software?

If I replace a DIMM with a slower, weaker variety, I can make GCC fail
with error 11's without much effort. Does that mean I should disable
all use of GCC on my machine? Good heavens, no!

For each instance where you're seeing stodgy hardware, where a user
can reproduce an IDE-DMA bug, I'm sure there are an equal or greater
number of instances where users have no problems. In another post I
made to Mark, I noted that sometimes it takes a particular piece of
software to trigger a hardware defect; mingo experimentally found a
fix that works in his case for no obvious reason. My current P2B-DS
motherboard works just fine with IDE-DMA and SMP too, though it's only
been two weeks under test.

A greater than average number of hardware failures due to some piece
of functionality is a good reason to disable it "by default" in
software. But Mark is right that you're not going to be able to find
and squash subtler bugs until you get a sizeable population enabling
the feature and testing; and to do that, you must provide a means by
which people can enable it. If you've disabled the functionality in
the kernel as you have in 2.1.111+, then you can't enable it with
hdparm. This is at least partly what Mark's (rightfully) complaining


Kristofer Karas                           *    Senior systems engineer/SysAdmin
AMA/CCS DoD RF900RR HawkGT !car           * BI Deaconess Medical Center, Boston
"Build a system that even a fool can use, *
 and only a fool will want to use it."    *  Will design LISP machines for food

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