Re: RFC: PCI quirks update for 2.6.16

[Chris Wedgwood <cw@f00f.org>]

On Sun, Dec 10, 2006 at 02:39:37PM -0800, Linus Torvalds wrote:

> They should be safe, and OBVIOUS.

Well, it's not clear to me that reverting to a quirk the pokes *all*
VIA pci devices on all machines is safe, it's not even clear if it was
a good idea to merge this.

All the same, I can retest the latest 2.6.16.x with that change
reverted but since it originally caused pain there has been a BIOS
upgrade (or two, I forget) that might affect things (for many poeple
the quirk isn't needed and CPI does the right thing).

> If there is a box that breaks with a 2.6.x.y release, then that .y
> release was clearly a mistake, and fundamentally broke the whole
> point of the 3Astable tree.

Well, I think the current 2.6.16.x release series is already broken on
some other subset of hardware.

There might be more of those than there are with the quirk-me-hard
approach --- in which case do we try to accommodate the (potential)
majority with something that is clearly wrong or so we leave them
broken for a bit longer until we can get some more coverage on Alan's
much cleaner and specific fix which I think is slated for 2.6.20 and
then backport that?