Re: [PATCH RFC] ata: Intel IDE-R support

[Tejun Heo <tj@kernel.org>]

Hello,

On 08/19/2010 03:08 PM, Kay Sievers wrote:
> Well, I wouldn't really call it broken. It's very helpful to have
> these priorities, and it solved a bunch of real problems because in
> most cases it produces predictable results, unlike it was before.
> 
> That compiled-in and and modules don't really mix here regarding your
> use case does not really matter in the general picture, it's still a
> predictable behavior, and we really need the priorities.

Yeah, in most usual cases, it has been fine.  I was just hoping that
it would work better so that fallback drivers can be handled with it.
For now, it seems there is no safe way to have a generic fallback
driver (be it pata_acpi or ata_generic).

> If we need finer-grained policy here, we need to move the driver
> binding to userspace. The driver core and udev can do that already.
> But nobody of us is so crazy to enable it, and handle all the fallout.
> But it's possible in theory ... :)

Right, userland already has enough mechanism and information to handle
the driver binding problem correctly, but given how little problem it
has been causing till now, I think that would be an overkill, for now
at least.

Thanks.

-- 
tejun