[libvirt] [PATCHv3 0/4] Use nodeGetCPUCount/virNodeGetCPUMap where appropriate

Viktor Mihajlovski mihajlov at linux.vnet.ibm.com
Tue Nov 13 12:54:35 UTC 2012

This series concludes the introduction of the virNodeGetCPUMap API
by replacing the usage of virNodeGetInfo for computing the maximum number
of node CPUs with either nodeGetCPUCount or virNodeGetCPUMap.

It's basically a resend of the original 3-patch series plus the
nodeGetCPUCount fix. Event in the light of Peter Krempa's recent
CPU topology fix this series is still relevant.

1/4 is needed to support older kernels
2/4 is not strictly required but is more efficient
3/4 and 4/4 are required for correct function of vcpu pinning calls
of python and virsh clients when talking to libvirtd 1.0.0.

V3 Changes:
- pulled in the nodeGetCPUCount patch for older kernels
- changed commit message of 2/4

V2 Changes:
Rework based on Eric's feedback:
- Use nodeGetCPUCount instead of nodeGetCPUMap
- Avoid code bloat by computing node CPU count in a helper function

Viktor Mihajlovski (4):
  nodeinfo: enable nodeGetCPUCount for older kernels
  qemu, lxc: Change host CPU number detection logic.
  python: Use virNodeGetCPUMap where possible
  virsh: Use virNodeGetCPUMap if possible

 python/libvirt-override.c | 87 ++++++++++++++++++++++++++++-------------------
 src/lxc/lxc_controller.c  |  8 ++---
 src/nodeinfo.c            | 33 +++++++++++++++---
 src/qemu/qemu_driver.c    | 14 +++-----
 src/qemu/qemu_process.c   |  8 ++---
 tools/virsh-domain.c      | 32 ++++++++++++-----
 6 files changed, 114 insertions(+), 68 deletions(-)


More information about the libvir-list mailing list