Re: unkown PCI device

Brian Gerst (bgerst@quark.vpplus.com)
Sun, 29 Nov 1998 05:03:26 -0500


Albert D. Cahalan wrote:
> Grrr... don't you even think of it!
>
> People paid money for those servers. How does /proc/pci hurt you?
> If you don't like it, you have the config option. Maybe there should
> never have been a /proc/pci, but it's there now and I'm addicted to it.
>
> One common complaint about Linux is that is changes too often.
> This wouldn't be just a new version freaking out a PHB, but a real
> incompatible change. It's not even a change we need for standards
> compliance. If /proc/pci somehow impedes development, please explain.

The problem with /proc/pci is that is requires maintenance of a list of
all known PCI cards inside the kernel. The kernel would need to be
recompiled every time a new PCI device was added before it could be
recognised. The point is that the kernel doesn't need the device name
strings to operate, and they just add to kernel bloat. Besides, there
is /proc/bus/pci, which although doesn't operate on name strings, could
still be parsed by the X server by the vendor and product numbers. This
change has been in the 2.1.x kernel for quite some time now, and the
impending removal of /proc/pci has been known ever since. Just as in
the case of the sound mmap fix, we can't let binary-only software force
us to retain buggy or unwanted code, just for the sake of backwards
compatability. There is still time for the commercial X servers to
remove their dependency on /proc/pci before it is actually removed (not
until at least 2.3.x). If they have been warned of the problem well in
advance and still drop the ball (like Real did with the sound fix), it's
their problem for not supporting their (paying) customers.

-- 

Brian Gerst

- 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/