[PATCH 0/4] FIO libnuma integration and 3 patches

[Yufei Ren <yufei.ren@stonybrook.edu>]

From: Yufei Ren <renyufei83@gmail.com>

Three patches and a numa integration is present as follows.

Yufei Ren (4):
  Current cpu ioengine always bruns out 100 percent cpu cycles     no
    matter what the cpuload value is. Since no data is transferred    
    with cpuio, bytes_done would be ZERO. Consequently, think_time    
    is omitted and loops keeps running.
  If `thread' option is enabled, resource usage should be thread    
    based instead of process based. For the following job,
  rdma ioengine improvement
  Two new options, numa_cpu_nodes and numa_mem_policy, are created    
    for a fine-grained job level numa control. Please refer HOWTO and  
      README for detailed description.     A example job,
    examples/numa, is added as well.

 HOWTO          |   18 +++++++
 README         |   14 ++++--
 backend.c      |   45 ++++++++++++++++++-
 engines/cpu.c  |    5 ++
 engines/rdma.c |   28 ++++++++---
 examples/cpuio |    8 +++
 examples/numa  |   21 +++++++++
 fio.1          |   22 +++++++++
 fio.h          |   18 +++++++
 options.c      |  138 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 os/os-linux.h  |    5 ++
 stat.c         |    4 ++
 12 files changed, 313 insertions(+), 13 deletions(-)
 create mode 100644 examples/cpuio
 create mode 100644 examples/numa

-- 
1.7.2.3