Messages in this thread | | | Subject | Re: /proc/pci unknown devices | From | Jes Sorensen <> | Date | 26 Oct 1999 12:29:28 +0200 |
| |
>>>>> "Peter" == Peter Samuelson <peter@wire.cadcamlab.org> writes:
Peter> [Jes Sorensen <Jes.Sorensen@cern.ch>] >> Besides, the table doesn't need to be a large static ever growing >> table, all you do is to copy the entries for the PCI devices you >> found to a new list and delete the rest.
Peter> Certainly that works today, since we know all PCI devices at Peter> boot time. This will not be true forever. I believe this has Peter> been discussed before here; what was the verdict on "knowing" Peter> the names of hot-swap devices?
Peter> Maybe when a module for a hot-swap device loads (responding to Peter> a "modprobe -k pci-dev-fe1c-ee05" or whatever) it can augment Peter> the /proc/pci table with its own __initdata. (Who knows, by Peter> then maybe __initdata in a module will mean something.)
Oh thats certainly a good option as well, there are two other ways as well a) list hot plug devices as "unknown or plugged in post boot" b) make the PCI table a loadable module that is loaded by the hot plug code, the device description is extracted from the big list and put in the /proc/pci list and the module unloaded again.
b) may sound a bit silly, but since you are unlikely to replace PCI hot-plug devices at a rate of 10000 devices/second, it is actually quite feasible.
Anyway I agree with Ted, this discussion is basically a "storm in a glass of water" (as we say in Danish).
Jes
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |