Ronald G Minnich <[EMAIL PROTECTED]> writes: > On 12 Mar 2002, Eric W. Biederman wrote: > > > > > On the long term agenda, I want to build a table similiar to what > > > > pci_enumerate generates at compile time. This should allows > > > > us to call pci_enumerate much later or not at all if all hardware > > > > is onboard. Plus we can enumerate onboard devices that are not pci. > > > > > > now that's a nice idea. > > > > If I can find a way to implement it that doesn't require everyone > > to change immediately I'll go ahead with it. So far I have been > > keeping on the backburner while I have resolved other problems. > > bear in mind that config currently builds static tables for superios. > There is no reason it can't do more.
You may be right. The fundamental shift (in my head at least) is having some kind of device id to recognize everything by. But something like that is really added fluff if you are building the table statically. You can do all of the matching up at compile time and save yourself some code. That sparcs some interesting thoughts. The interesting piece that definitely needs to be addressed is a version of pci_configure that works for superio devices to. Right now we pretty much hard code things which isn't to hot. In any event I have some fairly clean code that allow direct LinuxBIOS support in the Linux kernel. Which provides a direct route to doing interesting things from that direction. Eric
