xfs: introduce object readahead to log recovery
authorZhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Wed, 14 Aug 2013 07:16:03 +0000 (15:16 +0800)
committerBen Myers <bpm@sgi.com>
Fri, 23 Aug 2013 19:32:50 +0000 (14:32 -0500)
commit00574da199291751bfaaee15ad0f42a7144276ad
treec1d4bb20d559b5e5d2a137347886c45dde7b3c46
parent8d1d40832b1c53ae73931f1b536ce1ab7375b3c8
xfs: introduce object readahead to log recovery

  It can take a long time to run log recovery operation because it is
single threaded and is bound by read latency. We can find that it took
most of the time to wait for the read IO to occur, so if one object
readahead is introduced to log recovery, it will obviously reduce the
log recovery time.

Log recovery time stat:

          w/o this patch        w/ this patch

real:        0m15.023s             0m7.802s
user:        0m0.001s              0m0.001s
sys:         0m0.246s              0m0.107s

Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Reviewed-by: Ben Myers <bpm@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
fs/xfs/xfs_log_recover.c