[PATCH][2.6] JS20 support

[Benjamin Herrenschmidt]

On Wed, 2003-11-26 at 01:28, Jake Moilanen wrote:
> > Why is this needed on a JS20 blade?  My understanding is that this
> > patch is needed in a partitioned environment where we may get function
> > N of a PCI-PCI bridge assigned to a partition but not function 0.  The
> > JS20 blade isn't partitioned, so why do we need this patch?
>
> You are correct, the JS20 blade is not partitioned.  But on the 8111 the
> the second device is the LPC bus, IDE controller, and some other
> controllers we don't use.  On this second device the LPC bus is function
> 0 and function 1 is the IDE controller.
>
> Firmware wanted to keep the LPC bus hidden since it is not available for
> use and if AIX sees an ISA bus, it will assume it has a certain level of
> ISA functionality.  Because of this firmware does not have a function 0
> in the device-tree.  So without this patch we will not see the IDE
> controller.

So basically, we have this nice new hw doing perfectly normal thing, and
we now add junk to the firmware which itself requires an horrible hack
in the kernel just for the sake of not fixing AIX wrong assumptions ?

Nice....

Ben.


** Sent via the linuxppc64-dev mail list. See http://lists.linuxppc.org/