PCI Error Recovery API Proposal. (WAS:: [PATCH/RFC] PCI Error Recovery)

[Benjamin Herrenschmidt]

On Wed, 2005-03-16 at 21:43 +0100, Andi Kleen wrote:
> Hi Ben,
> 
> I have not studied everything in detail, but from a quick browse it looks
> reasonable. I would make an own callback for each state though, otherwise
> the driver has to do a switch with subfunctions anyways.

Well... I preferred not to bloat pci_driver for everybody, but I'm ok
with separate function approach in principle. We could even have pci_ops
just have an optional pointer to a separate pci_err_ops in fact to avoid
the blotage in the common case.

Again, my point in this definition wasn't the details like that, but the
actual detailed semantic of each step that, I hope, will fit everybody.

Ben.