[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-devel] [PATCH 0/3] Automatically derive soft affinity from vnuma information



Hi,

This small series is meant at achieving the following two goals:
 - when building a vnuma enabled guest, if any vcpu hard or soft affinity is
   specified, and that does not match with the vnuma configuration supplied
   (e.g., vcpu X, belonging to vnode Y on pnode W, has affinity with pcpus
   outside of pnode W), we want to warn the user about it;
 - when building a vnuma enabled guest, if no _soft_ affinity is specified, we
   want to automatically build it up, basing on the vnuma configuration supplied
   (e.g., vcpu X, belonging to vnode Y on pnode W will have soft affinity with
   pcpus of pnode W).

IOW, with the following in the config file:

 vcpus  = '4'
 memory = '2000'
 vnuma  = [ [ "pnode=0","size=1000","vcpus=0-1","vdistances=10,20"  ],
            [ "pnode=1","size=1000","vcpus=2-3","vdistances=20,10"  ] ]

We'll, automatically, end up with this:

 root@Zhaman:~# xl vcpu-list 1
 Name                                ID  VCPU   CPU State   Time(s) Affinity 
(Hard / Soft)
 test-pv                              1     0    7   -b-       8.1  all / 0-7
 test-pv                              1     1    3   -b-       6.2  all / 0-7
 test-pv                              1     2   10   -b-       0.1  all / 8-15
 test-pv                              1     3    8   -b-       0.1  all / 8-15

While at it, I'm taking the chance (in patch #3) for doing some trivial
cleanups.

Thanks and Regards,
Dario

---
Dario Faggioli (3):
      libxl: check whether vcpu affinity and vnuma info match
      libxl: automatically set soft affinity after vnuma info
      libxl: cleanup some misuse of 'cpumap' as parameter

 tools/libxl/libxl_dom.c   |   46 ++++++++++++++++++++++++++++++++++++++
 tools/libxl/libxl_utils.h |    6 +++--
 tools/libxl/libxl_vnuma.c |   54 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 102 insertions(+), 4 deletions(-)

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.