Subject: Re: Driver to device bindings
To: Perry E. Metzger <>
From: Bill Studenmund <>
List: tech-kern
Date: 12/30/2003 12:59:58
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sun, Dec 28, 2003 at 05:59:29PM -0500, Perry E. Metzger wrote:
> True enough, but imagine someday we'd like to load drivers at run time
> based on the IDs. For that, we can't leave all the information in the
> driver itself because then we'd have to load and unload every driver
> we have...

This is getting a bit off-topic (of the original thread).

For LKMs, I can envision one of two solutions.

1) Each LKM includes some sort of info (PCI IDs, etc.) indicating what the=
driver (may) support. So if we find unconfigured PCI IDs, we know what=20
LKMs to try.

2) We compile in "Module" support, which would be a stub driver that=20
matches, then on attach calls out for the right driver to be loaded.

So you're right, we don't want to try _every_ driver. But we can figure=20
out how not to. :-)

Take care,


Content-Type: application/pgp-signature
Content-Disposition: inline

Version: GnuPG v1.2.3 (NetBSD)