On Mon, Sep 24, 2012 at 06:23:21PM +0400, Dmitry Monakhov wrote:

Perform various regression tests for ext4defrag  subsystem

 Test1: Defragment file while other task does direct AIO
 Test2: Perform defragmentation on file under buffered AIO
 	while third task does direct AIO to donor file
 Test3: Two defrag tasks use common donor file.
 Test4: Stress defragmentation. Several threads pefrorm
 	fragmentation at random position use inplace=1 will
	allocate and free blocks inside defrag event improve
	load pressure.

Please do these as 4 separate tests so that it is easy to
isolate/reproduce failures caused by a specific test that is being
run.

Same comments as previous two patches, as well.

+_workout()
+{
+	echo ""
+	echo " Start defragment activity "
+	echo ""
+	cat $tmp-$seq.fio >>  $seq.full
+	run_check $FIO_PROG $tmp-$seq.fio >>  $seq.full
+}
+
+# real QA test starts here
+_supported_fs generic
+_supported_os Linux
+_supported_fs ext4
+_need_to_be_root
+_require_scratch
+_require_fio $tmp-$seq.fio

Even though e4defrag is not used, it's probably still a good idea to
call _require_defrag to check for systems that have a recent enough
ext4 to run defrag on. e.g. a recent fio on an old distro might
result in fio saying "I understand this job file" but then not be
able to run defrag...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com