Re: RFC: bare pci configuration access functions ?

Greg KH (greg@kroah.com)
Mon, 4 Nov 2002 13:30:02 -0800


On Mon, Nov 04, 2002 at 12:17:45PM -0800, Lee, Jung-Ik wrote:
> Hi Greg,
>
> > What's wrong with the _existing_ pci_config_read() and
> > pci_config_write() function pointers that ia64 and i386 have?
> > Can't you
> > just look into if the other archs can set them to the proper
> > function in
> > their pci init functions too?
>
> Other architectures' PCI config access methods vary and require their own
> address mappings, etc.

Ah, so exporting those types of functions is not pratical? Oh well...

> There could be two ways to achieve bare pci config accesses for all
> architectures.

<snip>

Wait, again I'm confused. Let's go over the main points here:

- for 2.5 everyone uses the pci_bus_read_config* and
pci_bus_write_config* functions and is happy. Well ACPI isn't happy,
but the code there currently works, so let's leave it at that.

- for 2.4 we don't have the pci_bus* functions, so we need to do
something. I originally wanted to look into exporting the
pci_config_* function pointers, but you said that doesn't look
possible based on the different arch specific implementation.

- Because of this, you just proposed a patch, yet your patch uses the
pci_bus_* functions which are not present on 2.4. If they were,
everyone would be happy again, and not need such a patch, right?

Confused,

greg k-h
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/