[LTP] [PATCH V3] controllers/pids: Add new testcases

[LTP] [PATCH V3] controllers/pids: Add new testcases

[Cedric Hnyda]

Created new testcases to test the pids controller
which is used to stop any new tasks from being fork()'d
after a certain limit is reached.

Signed-off-by: Cedric Hnyda <chnyda@suse.com>
---
 runtest/controllers                            |  22 +++
 testcases/kernel/controllers/pids/.gitignore   |   2 +
 testcases/kernel/controllers/pids/Makefile     |  28 ++++
 testcases/kernel/controllers/pids/pids.sh      | 208 +++++++++++++++++++++++++
 testcases/kernel/controllers/pids/pids_task.c  |  63 ++++++++
 testcases/kernel/controllers/pids/pids_task2.c |  60 +++++++
 6 files changed, 383 insertions(+)
 create mode 100644 testcases/kernel/controllers/pids/.gitignore
 create mode 100644 testcases/kernel/controllers/pids/Makefile
 create mode 100644 testcases/kernel/controllers/pids/pids.sh
 create mode 100644 testcases/kernel/controllers/pids/pids_task.c
 create mode 100644 testcases/kernel/controllers/pids/pids_task2.c

diff --git a/runtest/controllers b/runtest/controllers
index c62bcce..254a0b0 100644
--- a/runtest/controllers
+++ b/runtest/controllers
@@ -277,3 +277,25 @@ cpuset_memory_spread	cpuset_memory_spread_testset.sh
 cpuset_regression_test cpuset_regression_test.sh
 
 cgroup_xattr	cgroup_xattr
+
+pids_1_1 pids.sh 1 1
+pids_1_2 pids.sh 1 2
+pids_1_10 pids.sh 1 10
+pids_1_50 pids.sh 1 50
+pids_1_100 pids.sh 1 100
+pids_2_1 pids.sh 2 1
+pids_2_2 pids.sh 2 2
+pids_2_10 pids.sh 2 10
+pids_2_50 pids.sh 2 50
+pids_2_100 pids.sh 2 100
+pids_3_1 pids.sh 3 1
+pids_3_2 pids.sh 3 2
+pids_3_10 pids.sh 3 10
+pids_3_50 pids.sh 3 50
+pids_3_100 pids.sh 3 100
+pids_4_1 pids.sh 4 1
+pids_4_2 pids.sh 4 2
+pids_4_10 pids.sh 4 10
+pids_4_50 pids.sh 4 50
+pids_4_100 pids.sh 4 100
+pids_5_1 pids.sh 5 1
diff --git a/testcases/kernel/controllers/pids/.gitignore b/testcases/kernel/controllers/pids/.gitignore
new file mode 100644
index 0000000..2a86ab2
--- /dev/null
+++ b/testcases/kernel/controllers/pids/.gitignore
@@ -0,0 +1,2 @@
+/pids_task
+/pids_task2
diff --git a/testcases/kernel/controllers/pids/Makefile b/testcases/kernel/controllers/pids/Makefile
new file mode 100644
index 0000000..6500363
--- /dev/null
+++ b/testcases/kernel/controllers/pids/Makefile
@@ -0,0 +1,28 @@
+#
+#    Copyright (C) 2015, SUSE
+#
+#    This program is free software; you can redistribute it and/or modify
+#    it under the terms of the GNU General Public License as published by
+#    the Free Software Foundation; either version 2 of the License, or
+#    (at your option) any later version.
+#
+#    This program is distributed in the hope that it will be useful,
+#    but WITHOUT ANY WARRANTY; without even the implied warranty of
+#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+#    GNU General Public License for more details.
+#
+#    You should have received a copy of the GNU General Public License along
+#    with this program; if not, write to the Free Software Foundation, Inc.,
+#    51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+#
+# Author: Cedric Hnyda <chnyda@suse.com>
+#
+
+top_srcdir		?= ../../../..
+
+include $(top_srcdir)/include/mk/testcases.mk
+include $(abs_srcdir)/../Makefile.inc
+
+INSTALL_TARGETS		:= *.sh
+
+include $(top_srcdir)/include/mk/generic_leaf_target.mk
diff --git a/testcases/kernel/controllers/pids/pids.sh b/testcases/kernel/controllers/pids/pids.sh
new file mode 100644
index 0000000..53c9d20
--- /dev/null
+++ b/testcases/kernel/controllers/pids/pids.sh
@@ -0,0 +1,208 @@
+#!/bin/sh
+
+################################################################################
+##                                                                            ##
+## Copyright (c) 2015 SUSE                                                    ##
+##                                                                            ##
+## This program is free software;  you can redistribute it and#or modify      ##
+## it under the terms of the GNU General Public License as published by       ##
+## the Free Software Foundation; either version 2 of the License, or          ##
+## (at your option) any later version.                                        ##
+##                                                                            ##
+## This program is distributed in the hope that it will be useful, but        ##
+## WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY ##
+## or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License   ##
+## for more details.                                                          ##
+##                                                                            ##
+## You should have received a copy of the GNU General Public License          ##
+## along with this program;  if not, write to the Free Software               ##
+## Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301   ##
+## USA                                                                        ##
+##                                                                            ##
+## Author: Cedric Hnyda <chnyda@suse.com>                                     ##
+##                                                                            ##
+################################################################################
+
+# Usage
+# ./pids.sh caseno max
+#
+
+caseno=$1
+max=$2
+export TCID="pids_$1_$2"
+export TST_TOTAL=1
+mounted=1
+
+. test.sh
+
+usage()
+{
+	echo "usage: ./pids01 caseno max"
+	tst_resm TFAIL "invalid parameters"
+}
+
+cleanup()
+{
+	killall -9 pids_task2 > /dev/null
+
+	tst_resm TINFO "removing created directories"
+	rmdir $testpath
+	if [ "$mounted" -ne "1" ]; then
+		tst_resm TINFO "Umounting pids"
+		umount $mount_point
+		rmdir $mount_point
+	fi
+}
+
+setup()
+{
+	tst_require_root
+
+	exist=`grep -w pids /proc/cgroups | cut -f1`;
+	if [ "$exist" = "" ]; then
+		tst_brkm TCONF NULL "pids not supported"
+	fi
+
+	mount_point=`grep -w pids /proc/mounts | cut -f 2 | cut -d " " -f2`
+
+	if [ "$mount_point" = "" ]; then
+		mounted=0
+		mount_point=/dev/cgroup
+	fi
+
+	TST_CLEANUP=cleanup
+
+	testpath=$mount_point/ltp_$TCID
+
+	if [ "$mounted" -eq "0" ]; then
+		ROD mkdir -p $mount_point
+		ROD mount -t cgroup -o pids none $mount_point
+	fi
+	ROD mkdir -p $testpath
+}
+
+case1()
+{
+	for i in `seq 1 $max`; do
+		pids_task2 "$testpath/tasks" &
+	done
+
+	sleep 1
+
+	# should return 0 because there is no limit
+	pids_task "$testpath/tasks"
+	ret=$?
+
+	killall -9 pids_task2
+
+	if [ "$ret" -eq "2" ]; then
+		tst_resm TFAIL "fork failed unexpectedly"
+	elif [ "$ret" -eq "0" ]; then
+		tst_resm TPASS "fork didn't fail"
+	else
+		tst_resm TBROK "pids_task failed"
+	fi
+}
+
+case2()
+{
+	tmp=$((max - 1))
+	tst_resm TINFO "limit the number of pid to $max"
+	ROD echo $max > $testpath/pids.max
+
+	for i in `seq 1 $tmp`; do
+		pids_task2 "$testpath/tasks" &
+	done
+
+	sleep 1
+
+	# should return 2 because the limit of pids is reached
+	pids_task "$testpath/tasks"
+	ret=$?
+
+	killall -9 pids_task2
+
+	if [ "$ret" -eq "2" ]; then
+		tst_resm TPASS "fork failed as expected"
+	elif [ "$ret" -eq "0" ]; then
+		tst_resm TFAIL "fork didn't fail despite the limit"
+	else
+		tst_resm TBROK "pids_task failed"
+	fi
+}
+
+case3()
+{
+	tmp=$((max + 1))
+	tst_resm TINFO "limit the number of avalaible pid to $tmp"
+	ROD echo $tmp > $testpath/pids.max
+
+	tmp=$((max - 2))
+	for i in `seq 1 $tmp`; do
+		pids_task2 "$testpath/tasks" &
+	done
+
+	sleep 1
+
+	pids_task "$testpath/tasks"
+	ret=$?
+
+	killall -9 pids_task2
+
+	if [ "$ret" -eq "2" ]; then
+		tst_resm TFAIL "fork failed unexpectedly"
+	elif [ "$ret" -eq "0" ]; then
+		tst_resm TPASS "fork worked as expected"
+	else
+		tst_resm TBROK "pids_task failed"
+	fi
+}
+
+case4()
+{
+	tst_resm TINFO "limit the number of avalaible pid to 0"
+	ROD echo 0 > $testpath/pids.max
+
+	for i in `seq 1 $max`; do
+		pids_task2 "$testpath/tasks" &
+	done
+
+	sleep 1
+
+	# count the number of attached process
+	cnt=0
+	for job in `jobs -p`; do
+		cnt=$((cnt + 1))
+	done
+
+	killall -9 pids_task2
+
+	if [ "$cnt" -ne "$max" ]; then
+		cnt=$((max - cnt))
+		tst_resm TFAIL "$cnt process were not attached successfully"
+	else
+		tst_resm TPASS "all process were attached"
+	fi
+}
+
+case5()
+{
+	tst_resm TINFO "try to limit the number of avalaible pid to -1"
+	echo -1 > $testpath/pids.max
+
+	if [ "$?" -eq "0" ]; then
+		tst_resm TFAIL "managed to set the limit to -1"
+	else
+		tst_resm TPASS "didn't manage to set the limit to -1"
+	fi
+}
+
+setup
+
+if [ "$#" -ne "2" ]; then
+	usage
+else
+	case$caseno
+fi
+
+tst_exit
diff --git a/testcases/kernel/controllers/pids/pids_task.c b/testcases/kernel/controllers/pids/pids_task.c
new file mode 100644
index 0000000..a470ccf
--- /dev/null
+++ b/testcases/kernel/controllers/pids/pids_task.c
@@ -0,0 +1,63 @@
+/*
+ * Copyright (c) 2015 Cedric Hnyda <chnyda@suse.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+/*
+* Description:
+* Write the current process in argv[1]
+* and returns:
+*	-> 2 if fork fails with EAGAIN,
+*	-> 1 if there is another problem
+*	-> 0 if everything worked
+*/
+
+#include <stdio.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include <errno.h>
+#include <stdlib.h>
+#include <string.h>
+
+int main(int argc, char **argv)
+{
+	FILE *f;
+	int newpid;
+
+	if (argc != 2) {
+		fprintf(stderr, "Usage: %s /cgroup/.../tasks\n", argv[0]);
+		return 1;
+	}
+
+	f = fopen(argv[1], "a");
+	if (!f) {
+		perror("fopen failed");
+		return 1;
+	}
+
+	fprintf(f, "%i\n", getpid());
+	fclose(f);
+
+	newpid = fork();
+	if (newpid == -1 && errno == EAGAIN)
+		return 2;
+	if (newpid == -1) {
+		strerror(errno);
+		return 1;
+	}
+	return 0;
+}
diff --git a/testcases/kernel/controllers/pids/pids_task2.c b/testcases/kernel/controllers/pids/pids_task2.c
new file mode 100644
index 0000000..3ce63fe
--- /dev/null
+++ b/testcases/kernel/controllers/pids/pids_task2.c
@@ -0,0 +1,60 @@
+/*
+ * Copyright (c) 2015 Cedric Hnyda <chnyda@suse.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+/*
+* Description:
+* Write the current process in argv[1],
+* return 1 if there is a problem in attaching the process,
+* otherwise do an infinite loop
+*/
+
+#include <stdio.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include <errno.h>
+#include <stdlib.h>
+
+int main(int argc, char **argv)
+{
+	FILE *f;
+	int i;
+
+	i = 0;
+
+	if (argc != 2) {
+		fprintf(stderr, "Usage: %s /cgroup/.../tasks\n", argv[0]);
+		return 1;
+	}
+
+	f = fopen(argv[1], "a");
+	if (!f) {
+		perror("fopen failed");
+		return 1;
+	}
+
+	i = fprintf(f, "%i\n", getpid());
+
+	fclose(f);
+	if (i < 0)
+		return 1;
+
+	while (1) {
+	}
+	return 0;
+}
-- 
2.1.4

[LTP] [PATCH V3] controllers/pids: Add new testcases

[Alexey Kodanev]

Hi,
On 11/20/2015 04:34 PM, Cedric Hnyda wrote:
> Created new testcases to test the pids controller
> which is used to stop any new tasks from being fork()'d
> after a certain limit is reached.
>
> Signed-off-by: Cedric Hnyda <chnyda@suse.com>
...
> index 0000000..6500363
> --- /dev/null
> +++ b/testcases/kernel/controllers/pids/Makefile
> @@ -0,0 +1,28 @@
> +#
> +#    Copyright (C) 2015, SUSE
> +#
> +#    This program is free software; you can redistribute it and/or modify
> +#    it under the terms of the GNU General Public License as published by
> +#    the Free Software Foundation; either version 2 of the License, or
> +#    (at your option) any later version.
> +#
> +#    This program is distributed in the hope that it will be useful,
> +#    but WITHOUT ANY WARRANTY; without even the implied warranty of
> +#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +#    GNU General Public License for more details.
> +#
> +#    You should have received a copy of the GNU General Public License along
> +#    with this program; if not, write to the Free Software Foundation, Inc.,
> +#    51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
You could write the last paragraph as follows:

# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.

or, I guess, we may skip it. The main reason - the address can be changed.

> +
> +	while (1) {
> +	}

Is it necessary to add a busy loop to completely load CPU? If we only 
wait, I would add a sleep, something like

while (1)
      usleep(10);


Thanks,
Alexey

[LTP] [PATCH V3] controllers/pids: Add new testcases

[Cedric Hnyda]

Hi,

On 11/25/2015 10:20 AM, Alexey Kodanev wrote:
>> +
>> +    while (1) {
>> +    }
>
> Is it necessary to add a busy loop to completely load CPU? If we only 
> wait, I would add a sleep, something like
>
> while (1)
>      usleep(10);
>

You are absolutely right, it is not needed at all to completely load CPU.
Maybe it would also be better to use sleep rather than usleep.

Best regards,
Cedric

-- 
Cedric Hnyda

[LTP] [PATCH V3] controllers/pids: Add new testcases

[Cyril Hrubis]

Hi!
> > Is it necessary to add a busy loop to completely load CPU? If we only 
> > wait, I would add a sleep, something like
> >
> > while (1)
> >      usleep(10);
> >
> 
> You are absolutely right, it is not needed at all to completely load CPU.
> Maybe it would also be better to use sleep rather than usleep.

Or we can call pause(2) which will block the process until signal
arrives.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [PATCH V3] controllers/pids: Add new testcases

[Cyril Hrubis]

Hi!
> --- /dev/null
> +++ b/testcases/kernel/controllers/pids/.gitignore
> @@ -0,0 +1,2 @@
> +/pids_task
> +/pids_task2

I would have renamed the pids_task to pids_task1 just for consistency.


> +cleanup()
> +{
> +	killall -9 pids_task2 > /dev/null
> +
> +	tst_resm TINFO "removing created directories"
> +	rmdir $testpath
> +	if [ "$mounted" -ne "1" ]; then
> +		tst_resm TINFO "Umounting pids"
> +		umount $mount_point
> +		rmdir $mount_point
> +	fi
> +}
> +
> +setup()
> +{
> +	tst_require_root
> +
> +	exist=`grep -w pids /proc/cgroups | cut -f1`;
> +	if [ "$exist" = "" ]; then
> +		tst_brkm TCONF NULL "pids not supported"
> +	fi
> +
> +	mount_point=`grep -w pids /proc/mounts | cut -f 2 | cut -d " " -f2`
> +
> +	if [ "$mount_point" = "" ]; then
> +		mounted=0
> +		mount_point=/dev/cgroup
> +	fi
> +
> +	TST_CLEANUP=cleanup
> +
> +	testpath=$mount_point/ltp_$TCID
> +
> +	if [ "$mounted" -eq "0" ]; then
> +		ROD mkdir -p $mount_point
> +		ROD mount -t cgroup -o pids none $mount_point
> +	fi
> +	ROD mkdir -p $testpath
> +}
> +
> +case1()
> +{
> +	for i in `seq 1 $max`; do
> +		pids_task2 "$testpath/tasks" &
> +	done
> +
> +	sleep 1

The sleep is here to make sure all tasks are attached to the subgroup
right?

If so we should rather watch the "$testpath/tasks" file if it has enough
lines and sleep only if it doesn't.

Something as:

while [ $(cat "$testpath/tasks" | wc -l) -lt $max ]; do
	sleep 1
done

> +	# should return 0 because there is no limit
> +	pids_task "$testpath/tasks"
> +	ret=$?
> +
> +	killall -9 pids_task2
> +
> +	if [ "$ret" -eq "2" ]; then
> +		tst_resm TFAIL "fork failed unexpectedly"
> +	elif [ "$ret" -eq "0" ]; then
> +		tst_resm TPASS "fork didn't fail"
> +	else
> +		tst_resm TBROK "pids_task failed"
> +	fi
> +}
> +
> +case2()
> +{
> +	tmp=$((max - 1))
> +	tst_resm TINFO "limit the number of pid to $max"
> +	ROD echo $max > $testpath/pids.max
> +
> +	for i in `seq 1 $tmp`; do
> +		pids_task2 "$testpath/tasks" &
> +	done
> +
> +	sleep 1

Here as well.

> +	# should return 2 because the limit of pids is reached
> +	pids_task "$testpath/tasks"
> +	ret=$?
> +
> +	killall -9 pids_task2
> +
> +	if [ "$ret" -eq "2" ]; then
> +		tst_resm TPASS "fork failed as expected"
> +	elif [ "$ret" -eq "0" ]; then
> +		tst_resm TFAIL "fork didn't fail despite the limit"
> +	else
> +		tst_resm TBROK "pids_task failed"
> +	fi
> +}
> +
> +case3()
> +{
> +	tmp=$((max + 1))
> +	tst_resm TINFO "limit the number of avalaible pid to $tmp"
> +	ROD echo $tmp > $testpath/pids.max
> +
> +	tmp=$((max - 2))
> +	for i in `seq 1 $tmp`; do
> +		pids_task2 "$testpath/tasks" &
> +	done
> +
> +	sleep 1

And here as well. We should probably create a function to start N
pids_task2 processes and to wait until they are attached.

> +	pids_task "$testpath/tasks"
> +	ret=$?
> +
> +	killall -9 pids_task2
> +
> +	if [ "$ret" -eq "2" ]; then
> +		tst_resm TFAIL "fork failed unexpectedly"
> +	elif [ "$ret" -eq "0" ]; then
> +		tst_resm TPASS "fork worked as expected"
> +	else
> +		tst_resm TBROK "pids_task failed"
> +	fi
> +}
> +
> +case4()
> +{
> +	tst_resm TINFO "limit the number of avalaible pid to 0"
> +	ROD echo 0 > $testpath/pids.max
> +
> +	for i in `seq 1 $max`; do
> +		pids_task2 "$testpath/tasks" &
> +	done
> +
> +	sleep 1
> +
> +	# count the number of attached process
> +	cnt=0
> +	for job in `jobs -p`; do
> +		cnt=$((cnt + 1))
> +	done

Hmm, shouldn't cat $testpath/tasks | wc -l or jobs -p | wc -l do the
same job?

> +	killall -9 pids_task2
> +
> +	if [ "$cnt" -ne "$max" ]; then
> +		cnt=$((max - cnt))
> +		tst_resm TFAIL "$cnt process were not attached successfully"
> +	else
> +		tst_resm TPASS "all process were attached"
> +	fi
> +}
> +
> +case5()
> +{
> +	tst_resm TINFO "try to limit the number of avalaible pid to -1"
> +	echo -1 > $testpath/pids.max
> +
> +	if [ "$?" -eq "0" ]; then
> +		tst_resm TFAIL "managed to set the limit to -1"
> +	else
> +		tst_resm TPASS "didn't manage to set the limit to -1"
> +	fi
> +}
> +
> +setup
> +
> +if [ "$#" -ne "2" ]; then
> +	usage
> +else
> +	case$caseno
> +fi
> +
> +tst_exit
> diff --git a/testcases/kernel/controllers/pids/pids_task.c b/testcases/kernel/controllers/pids/pids_task.c
> new file mode 100644
> index 0000000..a470ccf
> --- /dev/null
> +++ b/testcases/kernel/controllers/pids/pids_task.c
> @@ -0,0 +1,63 @@
> +/*
> + * Copyright (c) 2015 Cedric Hnyda <chnyda@suse.com>
> + *
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public License as
> + * published by the Free Software Foundation; either version 2 of
> + * the License, or (at your option) any later version.
> + *
> + * This program is distributed in the hope that it would be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, write the Free Software Foundation,
> + * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
> + */
> +
> +/*
> +* Description:
> +* Write the current process in argv[1]
> +* and returns:
> +*	-> 2 if fork fails with EAGAIN,
> +*	-> 1 if there is another problem
> +*	-> 0 if everything worked
> +*/
> +
> +#include <stdio.h>
> +#include <sys/time.h>
> +#include <sys/types.h>
> +#include <unistd.h>
> +#include <errno.h>
> +#include <stdlib.h>
> +#include <string.h>
> +
> +int main(int argc, char **argv)
> +{
> +	FILE *f;
> +	int newpid;
> +
> +	if (argc != 2) {
> +		fprintf(stderr, "Usage: %s /cgroup/.../tasks\n", argv[0]);
> +		return 1;
> +	}
> +
> +	f = fopen(argv[1], "a");
> +	if (!f) {
> +		perror("fopen failed");
> +		return 1;
> +	}
> +
> +	fprintf(f, "%i\n", getpid());
> +	fclose(f);
> +
> +	newpid = fork();
> +	if (newpid == -1 && errno == EAGAIN)
> +		return 2;
> +	if (newpid == -1) {
> +		strerror(errno);
> +		return 1;
> +	}
> +	return 0;
> +}

Hi!

> +/*
> +* Description:
> +* Write the current process in argv[1],
> +* return 1 if there is a problem in attaching the process,
> +* otherwise do an infinite loop
> +*/
> +
> +#include <stdio.h>
> +#include <sys/time.h>
> +#include <sys/types.h>
> +#include <unistd.h>
> +#include <errno.h>
> +#include <stdlib.h>
> +
> +int main(int argc, char **argv)
> +{
> +	FILE *f;
> +	int i;
> +
> +	i = 0;
> +
> +	if (argc != 2) {
> +		fprintf(stderr, "Usage: %s /cgroup/.../tasks\n", argv[0]);
> +		return 1;
> +	}
> +
> +	f = fopen(argv[1], "a");
> +	if (!f) {
> +		perror("fopen failed");
> +		return 1;
> +	}
> +
> +	i = fprintf(f, "%i\n", getpid());
> +
> +	fclose(f);
> +	if (i < 0)
> +		return 1;
> +
> +	while (1) {
> +	}
> +	return 0;
> +}

Given that this process doesn't do anything else than attaching the
process to the cgroup, we can simplify it to call only the pause(2) and
attach the process to the cgroup in the main test script
(echo $! > $testpath/tasks). That way we do not have to wait until the
processes are attached in the main script and the code would be much
simpler.

-- 
Cyril Hrubis
chrubis@suse.cz