From mboxrd@z Thu Jan  1 00:00:00 1970
From: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
Date: Fri, 2 Aug 2019 18:12:58 +0800
Subject: [LTP] [PATCH] cgroup_regression_test.sh: fix test_5 possible
 mount failure because of cgroup hierarchy
In-Reply-To: <5D2D681D.1050909@cn.fujitsu.com>
References: <1560250815-2308-1-git-send-email-xuyang2018.jy@cn.fujitsu.com>
 <5D2D681D.1050909@cn.fujitsu.com>
Message-ID: <5D440CAA.9080701@cn.fujitsu.com>
List-Id: <ltp.lists.linux.it>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: ltp@lists.linux.it


Hi
Ping. :-)
> Hi 
>
> Ping.  :-) 
>
>> Currently, if systems doesn't mount subsys1,subsys2 and the hierarchy is not equal to 0, running it
>> reports the following error:
>>
>> mount: xxx is already mounted or /tmp/ltp-wPw08anmTI/LTP_cgroup_regression_test.V4jf0qrS7z/cgroup busy
>> cgroup_regression_test 5 TFAIL: mount net_prio and pids failed
>>
>> It fails because libcgroup doesn't permmit destroy cgroup subsystem hierarchies.
>> Simple umnout does not destroy the hierarchies. They still live inside kernel!
>>
>> When  hierarchy is equal to 0 in /proc/cgroups, we can mount them together on
>> a new mountpoint.
>>
>> I add a check for subsystem hierarchy and get subsystem from head.
>>
>> Notice:
>> more information about"Bug 612805 - cgroup: mount: none already mounted or /cgroups busy"
>>
>> Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
>> ---
>>  .../cgroup/cgroup_regression_test.sh           | 18 +++++++++++++-----
>>  1 file changed, 13 insertions(+), 5 deletions(-)
>>
>> diff --git a/testcases/kernel/controllers/cgroup/cgroup_regression_test.sh b/testcases/kernel/controllers/cgroup/cgroup_regression_test.sh
>> index e197f5d3f..38cb760c2 100755
>> --- a/testcases/kernel/controllers/cgroup/cgroup_regression_test.sh
>> +++ b/testcases/kernel/controllers/cgroup/cgroup_regression_test.sh
>> @@ -252,8 +252,10 @@ test5()
>>  		return
>>  	fi
>>  
>> -	local subsys1=`tail -n 1 /proc/cgroups | awk '{ print $1 }'`
>> -	local subsys2=`tail -n 2 /proc/cgroups | head -1 | awk '{ print $1 }'`
>> +	local subsys1=`head -2 /proc/cgroups | tail -n 1 | awk '{ print $1 }'`
>> +	local subsys1_hierarchy=`head -2 /proc/cgroups | tail -n 1 | awk '{ print $2 }'`
>> +	local subsys2=`head -3 /proc/cgroups | tail -n 1 | awk '{ print $1 }'`
>> +	local subsys2_hierarchy=`head -3 /proc/cgroups | tail -n 1 | awk '{ print $2 }'`
>>  
>>  	# Accounting here for the fact that the chosen subsystems could
>>  	# have been already previously mounted at boot time: in such a
>> @@ -267,10 +269,16 @@ test5()
>>  	if [ -z "$mounted" ]; then
>>  		mntpoint=cgroup
>>  		failing=$subsys1
>> -		mount -t cgroup -o $subsys1,$subsys2 xxx $mntpoint/
>> +		mount -t cgroup -o $subsys1,$subsys2 xxx $mntpoint/ 2>/dev/null
>> +		# Even subsystem has not been mounted, it still live in kernel.
>> +		# So we will get EBUSY when both mount subsys1 and subsys2 if
>> +		# hierarchy isn't equal to 0.
>>  		if [ $? -ne 0 ]; then
>> -			tst_res TFAIL "mount $subsys1 and $subsys2 failed"
>> -			return
>> +			if [ "$subsys1_hierarchy" = 0 -a "$subsys2_hierarchy" = 0 ]; then
>> +				tst_res TFAIL "mount $subsys1 and $subsys2 failed"
>> +				return
>> +			fi
>> +			failing=$subsys1,$subsys2
>>  		fi
>>  	else
>>  		# Use the pre-esistent mountpoint as $mntpoint and use a