[RFC PATCH 1/2] powerpc/numa: fix cpu_to_node() usage during boot
Nishanth Aravamudan
nacc at linux.vnet.ibm.com
Thu Jul 16 08:43:51 AEST 2015
On 15.07.2015 [16:35:16 -0400], Tejun Heo wrote:
> Hello,
>
> On Thu, Jul 02, 2015 at 04:02:02PM -0700, Nishanth Aravamudan wrote:
> > we currently emit at boot:
> >
> > [ 0.000000] pcpu-alloc: [0] 0 1 2 3 [0] 4 5 6 7
> >
> > After this commit, we correctly emit:
> >
> > [ 0.000000] pcpu-alloc: [0] 0 1 2 3 [1] 4 5 6 7
>
> JFYI, the numbers in the brackets aren't NUMA node numbers but percpu
> allocation group numbers and they're not split according to nodes but
> percpu allocation units. In both cases, there are two units each
> serving 0-3 and 4-7. In the above case, because it wasn't being fed
> the correct NUMA information, both got assigned to the same group. In
> the latter, they got assigned to different ones but even then if the
> group numbers match NUMA node numbers, that's just a coincidence.
Ok, thank you for clarifying! From a correctness perspective, even if
the numbers don't match NUMA nodes, should we expect the grouping to be
split along NUMA topology?
-Nish
More information about the Linuxppc-dev
mailing list