At 10/10/2012 05:10 PM, Peter Zijlstra Wrote:
> On Tue, 2012-10-09 at 16:27 -0700, David Rientjes wrote:
>> On Tue, 9 Oct 2012, Peter Zijlstra wrote:
>>
>>> Well the code they were patching is in the wakeup path. As I think Tang
>>> said, we leave !runnable tasks on whatever cpu they ran on last, even if
>>> that cpu is offlined, we try and fix up state when we get a wakeup.
>>>
>>> On wakeup, it tries to find a cpu to run on and will try a cpu of the
>>> same node first.
>>>
>>> Now if that node's entirely gone away, it appears the cpu_to_node() map
>>> will not return a valid node number.
>>>
>>> I think that's a change in behaviour, it didn't used to do that afaik.
>>> Certainly this code hasn't change in a while.
>>>
>>
>> If cpu_to_node() always returns a valid node id even if all cpus on the
>> node are offline, then the cpumask_of_node() implementation, which the
>> sched code is using, should either return an empty cpumask (if
>> node_to_cpumask_map[nid] isn't freed) or cpu_online_mask. The change in
>> behavior here occurred because
>> cpu_hotplug-unmap-cpu2node-when-the-cpu-is-hotremoved.patch in -mm doesn't
>> return a valid node id and forces it to return -1 so a kzalloc_node(...,
>> -1) fallsback to allocate anywhere.
>
> I think that's broken semantics.. so far the entire cpu<->node mapping
> was invariant during hotplug. Changing that is going to be _very_
> interesting and cannot be done lightly.
>
> Because as I said, per-cpu memory is preserved over hotplug, and that
> has numa affinity.
Hmm, if per-cpu memory is preserved, and we can't offline and remove
this memory. So we can't offline the node.
But, if the node is hot added, and per-cpu memory doesn't use the
memory on this node. We can hotremove cpu/memory on this node, and then
offline this node.
Before the cpu is hotadded, cpu's node is -1. We set cpu<->node mapping
when it is hotadded. So the entire cpu<->node mapping was not invariant
during hotplug.
So it is why I try to clear it when the cpu is hot-removed.
As we need the mapping to migrate a task to the cpu on the same node first,
I think we can clear the mapping when the node is offlined.
Thanks
Wen Congyang
>
> So for now, let me NACK that patch. You cannot go change stuff like
> that.
>
>>
>> But if you only need cpu_to_node() when waking up to find a runnable cpu
>> for this NUMA information, then I think you can just change the
>> kzalloc_node() in alloc_{fair,rt}_sched_group() to do
>> kzalloc(..., cpu_online(cpu) ? cpu_to_node(cpu) : NUMA_NO_NODE).
>
> That's a confusing statement, the wakeup stuff and the
> alloc_{fair,rt}_sched_group() stuff are unrelated, although both sites
> might need fixing if we're going to go ahead with this.
>
--
To unsubscribe from this list: send the line "unsubscribe linux-numa" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
