Re: [PATCH] x86, AMD, NB: Cleanup and improve NB descriptors handling
From: Daniel J Blueman
Date: Tue Oct 16 2012 - 10:12:52 EST
On 16/10/2012 22:06, Borislav Petkov wrote:
> From: Borislav Petkov <borislav.petkov@xxxxxxx>
> First of all, the double loop of first counting the northbridges on
> the system, then allocating them and then going over them again was
> bugging me for a long time now. Fix it by moving the allocation of the
> descriptors in the initial iteration path.
> Then, modify the enclosing _info structure to contain two NB descriptor
> structures: and array of 8 NB descs which contains the first 8 possible
> NB descriptors for all single-board AMD systems out there. This is no
> different from the current state of affairs and is done this way so that
> lookup of those descriptors remains fast.
> Then, there's a linked list of further NB descs which are for large,
> confederated systems like NumaScale where more than 8 nodes are present
> in the fabric and are visible to the single system Linux image. We keep
> those unique by introducing the PCI domain ID into the node ID used for
> descriptor lookup.
> Rest of the code is cleanup and readability improvements.
I recall there were some other areas that needed fixing, so I'll rebase
on top of this tomorrow and present my change so far.
Daniel J Blueman
Principal Software Engineer, Numascale Asia
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/