Messages in this thread | | | Date | Mon, 3 Mar 2008 21:29:55 +0000 | From | Alan Cox <> | Subject | Re: Weirdness in pci_read_bases() |
| |
> The main question regarding 0 in a BAR is do we know of devices that > consider it as "disabled" and don't operate/decode 0 ? (That would be > gross and out of spec but who knows...)
In the PC world it never gets tested so it might be wise to be cautious.
> But that can be solved, using something akin to IORESOURCE_UNSET, like > we do on powerpc, generalizing it, and having an x86-specific quirk set > it on anything that comes up with resource->start == 0.
And add pci_resource_unassigned() to wrap it nicely. That would definitely be a good thing as we have confusion right now and it has turned up in the real world.
Alan
| |