lkml.org 
[lkml]   [2012]   [Oct]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] RFC: Fix AMD Northbridge-ID contiguity assumptions
On Wed, Oct 03, 2012 at 08:50:51AM -0700, H. Peter Anvin wrote:
> On 10/03/2012 08:30 AM, Daniel Blueman wrote:
> > >
> > > Is that "on NumaScale systems"? If so, please say so rather than trying
> > > to make it sound generic; if it is not, can you give some other examples?
> >
> >It is for Numascale (NumaChip) systems for our purposes.
> >
> >Any other systems which interconnect Opterons via address space routing
> >(needed for >8 HT nodes) will get this benefit. I can't put my hand to
> >exactly what is out there, but can find out.
> >
>
> The reason I'm asking is because it is an important bit of the
> record of the code to know if this is a specific need or a general
> need. This may be obvious now, but 5-10 years from now someone will
> need to know why or what.
>
> The two paragraphs above is exactly what is needed, i.e. "NumaChip
> or any other design which shares these specific design features:
> ..."

Absolutely!

And it would be best to put that explanation in the code somewhere
around node_to_amd_index() so that it is there at a first glance.

Btw, I'll review the patch tomorrow since it is a holiday today here.

I have only two nits for now:

* node is u32, do you really have such big systems with 2^32-1 nodes? Or
can the max node number fit into a, say, u16 or u8?

* node_to_amd_index returns 0 in the unsuccessful case but node index 0
seems ok to me, i.e. the first element in the array of northbridges. It
probably should return a negative value rather to signal a failure...

Thanks.

--
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
GM: Alberto Bozzo
Reg: Dornach, Landkreis Muenchen
HRB Nr. 43632 WEEE Registernr: 129 19551


\
 
 \ /
  Last update: 2012-10-03 21:01    [W:0.052 / U:0.056 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site