Re: 2.6.10-mm1

[David Howells <dhowells@redhat.com>]

Greg KH <greg@kroah.com> wrote:

> > [1] Should the code generating the warning be active without CONFIG_PM
> > being set?

I believe the code is attempting to make sure the device is properly powered
on in the first place. CONFIG_PM only governs later power management.

> > [2] Can you explain why the message is generated (why not silently ignore
> > the older hardware) or is there something in an init script (I am using
> > Fedora Core 2) that [incorrectly] assumes power management is available to
> > cause the message to be printed?
> 
> David, any ideas?  Should I just revert this change for now?

Please don't. A system with this Promise 20269 card in it hangs without this
patch, I can see the splat happen with a PCI analyser.

The function being altered is almost certainly _wrong_ for non-PM-version-2
cards, but I don't have an old enough PCI spec to check.

A better solution would be to drop the level of the printk() to KERN_DEBUG or
to delete it entirely, assuming the patch otherwise works for Mark.

David