On Tue, Jun 27, 2000, David Brownell <[EMAIL PROTECTED]> wrote:
> I thought you were going to mention needing to
> integrate the current PCI HotSwap framework
> more fully within Linux 2.4!
>
> Examples:
>
> - USB HCDs should use 'struct pci_driver'
> to support Cardbus, HotSwap PCI, and
> eventually better configuration tools;
>
> - Heck, Cardbus bridges should use that API.
> Improve that suspend/resume ordering!
I'm not worried. I guess for 2.4 it's a good idea, but it so far from a
proper solution that using it now would just give us incremental
improvement over current functionality while not solving any of the
real problems.
> - And shouldn't USB device drivers also have
> an opportunity to hook PCI suspend/resume
> processing at the right points too? They
> already have hooks mirroring probe/remove
> functionality in the PCI framework ...
I wasn't worried about Power Management at all. That is a different problem
and is being worked on seperately.
> Maybe such Linux 2.4 PCI-specific and USB-specific
> hotplugging stuff fits under "4.3 Any more?".
>
> I would also either look more at networking. Or else
> motivate a new definition of hotswapping which doesn't
> include someone adding a new PC to the office LAN, and
> then having it autoconfigure so invisibly that you were
> reading files from it before you knew the box existed!
> (The systems problems are the same: classic. :-)
Networking setup is completely userspace. I was focussing on the kernel to
userspace interaction.
I think you may have missed the intent of the paper.
JE
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
