On Fri, Mar 12, 2010 at 08:45:23AM -0500, Chris Lalancette wrote:
On 03/12/2010 06:32 AM, Daniel Veillard wrote:
> On Thu, Mar 11, 2010 at 06:00:56PM -0500, Chris Lalancette wrote:
>> The nodeGetInfo code was always assuming that machine had a
>> single NUMA node, which is not correct. The good news is that
>> libnuma gives us this information pretty easily, so let's
>> properly report it.
>
> okay
>
>> NOTE: With recent hardware starting to support CPU hot-add
>> and hot-remove, both this code and the nodeCapsInitNUMA()
>> code are quickly going to become obsolete. We'll have to
>> think of a more dynamic solution for dealing with NUMA
>> nodes and CPUs that can come and go at will.
>
> well it makes little sense to refresh all the time,
> I would expect some kind of signal we can hook on to detect
> change in topology or capacities, though I don't see anything
> on signal.h for this
Well, it's actually not horrible to refresh all of the time. The
way we get this information is all in memory (from /proc or /sys),
so it's a very fast operation. That being said, if we can get
a signal when things change, and just react to that, that's probably
the best way to go. This stuff is all very new, so things are
still being put into place; we'll just have to figure out how to
be more dynamic in the future.
Jes Sorensen suggested that the hotplug daemon could somehow
be tweaked to advertize when things changed. But this is all
suppositions at this point, and might be hard to make portable
accros distros :-)
Daniel
--
Daniel Veillard | libxml Gnome XML XSLT toolkit
http://xmlsoft.org/
daniel(a)veillard.com | Rpmfind RPM search engine
http://rpmfind.net/
http://veillard.com/ | virtualization library
http://libvirt.org/