I thought the idea was to remove that functionality out of the kernel
completely. In our case, we're only adding the "names" for the cards
that "our" driver supports. We have a 16 or 20-byte structure per card
already, and if we want, we can declare it all to be init_data.
Anyway, if you don't like the strings-in-the-table approach, we should
move to an exported function from oldproc which returns the name of a
PCI device.
It's not that I don't like the strings-in-the-table approach as much as
I don't like the idea of strings-in-two-tables approach. If we have a
PCI number->string database in oldproc, keeping a separate databases in
the serial driver just seems like a bad idea, besides potentially
wasting space.
As far as the approach of declaring the pci table to be init_data, that
may work for now, but it won't work in the long run once we support
hot-pluggable PCI. (I currently use register_serial() to register
individual PCI serial ports. This wasn't an accident, since I was
thinking about hot pluggable PCI support in the future.)
Linus, what is your opinion of oldproc? Is it going to stay or go, in
the long run?
- Ted
-
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/