mcexec: detect mismatch of mcexec -n and mpirun -ppn

Change-Id: I0ce1b2d48cda10713920cb88692e107b8c4d3bab
Refs: #929

test/issues/929/C929.sh

@@ -0,0 +1,145 @@
+#/bin/sh
+
+USELTP=0
+USEOSTEST=0
+
+. ../../common.sh
+
+issue="929"
+tid=01
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+TEST_CMD="mpirun -f ./hostfile -ppn 5 ${MCEXEC} -n 5 ./test_prog.sh"
+echo ${TEST_CMD}
+${TEST_CMD} &> ${tname}.txt
+mpi_ret=$?
+
+cat ./${tname}.txt
+started_num=`grep 'test_prog is started' ./${tname}.txt | wc -l`
+
+if [ ${mpi_ret} -eq 0 -a ${started_num} -eq 5 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+TEST_CMD="mpirun -f ./hostfile -ppn 5 ${MCEXEC} -n 3 ./test_prog.sh"
+echo ${TEST_CMD}
+${TEST_CMD} &> ${tname}.txt
+mpi_ret=$?
+
+cat ./${tname}.txt
+started_num=`grep 'test_prog is started' ./${tname}.txt | wc -l`
+
+if [ ${mpi_ret} -ne 0 -a ${started_num} -eq 3 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+TEST_CMD="mpirun -f ./hostfile -ppn 3 ${MCEXEC} -n 5 ./test_prog.sh"
+echo ${TEST_CMD}
+${TEST_CMD} &> ${tname}.txt
+mpi_ret=$?
+
+cat ./${tname}.txt
+started_num=`grep 'test_prog is started' ./${tname}.txt | wc -l`
+
+if [ ${mpi_ret} -ne 0 -a ${started_num} -eq 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+TEST_CMD="mpirun -f ./hostfile -ppn 6 ${MCEXEC} -n 3 ./test_prog.sh"
+echo ${TEST_CMD}
+${TEST_CMD} &> ${tname}.txt
+mpi_ret=$?
+
+cat ./${tname}.txt
+started_num=`grep 'test_prog is started' ./${tname}.txt | wc -l`
+
+if [ ${mpi_ret} -ne 0 -a ${started_num} -eq 3 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+TEST_CMD="mpirun -f ./hostfile -ppn 250 ${MCEXEC} -n 250 ./test_prog.sh"
+echo ${TEST_CMD}
+${TEST_CMD} &> ${tname}.txt
+mpi_ret=$?
+
+head -n 10 ./${tname}.txt
+echo "..."
+started_num=`grep 'test_prog is started' ./${tname}.txt | wc -l`
+
+if [ ${mpi_ret} -ne 0 -a ${started_num} -eq 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+ng=0
+TEST_CMD="mpirun -f ./hostfile -ppn 5 ${MCEXEC} -n 5 ./test_prog.sh"
+echo "** reboot mcrernel for check pe_list_len"
+mcreboot
+echo "** enable debug message in mcexec_get_cpuset"
+sudo sh -c "echo -n 'func mcexec_get_cpuset +p' > /sys/kernel/debug/dynamic_debug/control"
+echo ${TEST_CMD}
+for i in `seq 1 20`
+do
+	${TEST_CMD} &> ${tname}.txt
+	mpi_ret=$?
+	started_num=`grep 'test_prog is started' ./${tname}.txt | wc -l`
+	if [ ${mpi_ret} -eq 0 -a ${started_num} -eq 5 ]; then
+		echo "[OK] exec: $i"
+	else
+		echo "[NG] exec: $i"
+		let ng++
+	fi
+done
+echo "** check pe_list_len"
+dmesg --notime | grep "mcexec_get_cpuset: pe_list" | tail -n 20 | cut -f 2-3 -d ':' > ./pe_list_len.txt
+cat ./pe_list_len.txt | while read line
+do
+	len=`echo ${line} | cut -f 2 -d ':'`
+	if [ ${len} -ge 0 -a ${len} -le 5 ]; then
+		echo "[OK] ${line}"
+	else
+		echo "[NG] ${line}"
+		let ng++
+	fi
+done
+echo "** disable debug message in mcexec_get_cpuset"
+sudo sh -c "echo -n 'func mcexec_get_cpuset -p' > /sys/kernel/debug/dynamic_debug/control"
+
+if [ ${ng} -eq 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+

test/issues/929/Makefile

@@ -0,0 +1,11 @@
+CFLAGS=-g
+LDFLAGS=
+
+TARGET=
+
+all: $(TARGET)
+
+test: all
+	./C929.sh
+clean:
+	rm -f $(TARGET) *.o *.txt

test/issues/929/README

@@ -0,0 +1,36 @@
+【Issue#929 動作確認】
+□ テスト内容
+1. mpirunで指定する-ppnと、mcexecで指定する-n の指定状況ごとに
+  想定どおりの動作となることを確認
+C929T01:
+  -ppn == -n の場合に、プログラムが実行され、mpirunが成功する
+
+C929T02:
+  -ppn > -n の場合に、プログラムの一部が実行され、mpirunが失敗する
+
+C929T03:
+  -ppn < -n の場合に、プログラムが実行されず、mpirunが失敗する
+
+C929T04:
+  -ppn が -n の整数倍である場合に、プログラムの一部が実行され、mpirunが失敗する
+
+C929T05:
+  -ppn と -n がMcKernelに割り当てたCPU数よりも大きい場合に、
+  プログラムが実行されず、mpirunが失敗する
+
+C929T06:
+  -ppn == -n での正常実行を20回連続で行った場合に、
+  プログラムが実行され、mpirunが成功する
+  また、mcctrlで管理しているpart_exec_list の要素数が5を超えない
+
+□ 実行手順
+$ make test
+
+McKernelのインストール先や、OSTEST, LTPの配置場所は、
+$HOME/.mck_test_config を参照している
+.mck_test_config は、McKernelをビルドした際に生成されるmck_test_config.sample ファイルを
+$HOMEにコピーし、適宜編集する
+
+□ 実行結果
+x86_64_result.log aarch64_result.log 参照。
+すべての項目をPASSしていることを確認。

test/issues/929/aarch64_result.log

@@ -0,0 +1,99 @@
+*** C929T01 start *******************************
+mpirun -f ./hostfile -ppn 5 /home/satoken/ihk+mckernel/bin/mcexec -n 5 ./test_prog.sh
+test_prog is started.
+test_prog is started.
+test_prog is started.
+test_prog is started.
+test_prog is started.
+*** C929T01 PASSED ******************************
+
+*** C929T02 start *******************************
+mpirun -f ./hostfile -ppn 5 /home/satoken/ihk+mckernel/bin/mcexec -n 3 ./test_prog.sh
+getting CPU set for partitioned execution: Invalid argument
+getting CPU set for partitioned execution: Invalid argument
+test_prog is started.
+test_prog is started.
+test_prog is started.
+*** C929T02 PASSED ******************************
+
+*** C929T03 start *******************************
+mpirun -f ./hostfile -ppn 3 /home/satoken/ihk+mckernel/bin/mcexec -n 5 ./test_prog.sh
+getting CPU set for partitioned execution: Connection timed out
+getting CPU set for partitioned execution: Connection timed out
+getting CPU set for partitioned execution: Connection timed out
+*** C929T03 PASSED ******************************
+
+*** C929T04 start *******************************
+mpirun -f ./hostfile -ppn 6 /home/satoken/ihk+mckernel/bin/mcexec -n 3 ./test_prog.sh
+getting CPU set for partitioned execution: Invalid argument
+getting CPU set for partitioned execution: Invalid argument
+getting CPU set for partitioned execution: Invalid argument
+test_prog is started.
+test_prog is started.
+test_prog is started.
+*** C929T04 PASSED ******************************
+
+*** C929T05 start *******************************
+mpirun -f ./hostfile -ppn 250 /home/satoken/ihk+mckernel/bin/mcexec -n 250 ./test_prog.sh
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+...
+*** C929T05 PASSED ******************************
+
+*** C929T06 start *******************************
+** reboot mcrernel for check pe_list_len
+mcreboot.sh -c 1-6,29-34 -m 50G@0,50G@1 -r 1-6:0+29-34:28 -O ... done
+** enable debug message in mcexec_get_cpuset
+mpirun -f ./hostfile -ppn 5 /home/satoken/ihk+mckernel/bin/mcexec -n 5 ./test_prog.sh
+[OK] exec: 1
+[OK] exec: 2
+[OK] exec: 3
+[OK] exec: 4
+[OK] exec: 5
+[OK] exec: 6
+[OK] exec: 7
+[OK] exec: 8
+[OK] exec: 9
+[OK] exec: 10
+[OK] exec: 11
+[OK] exec: 12
+[OK] exec: 13
+[OK] exec: 14
+[OK] exec: 15
+[OK] exec: 16
+[OK] exec: 17
+[OK] exec: 18
+[OK] exec: 19
+[OK] exec: 20
+** check pe_list_len
+[OK] pe_list_len:0
+[OK] pe_list_len:1
+[OK] pe_list_len:2
+[OK] pe_list_len:3
+[OK] pe_list_len:4
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+[OK] pe_list_len:5
+** disable debug message in mcexec_get_cpuset
+*** C929T06 PASSED ******************************
+

test/issues/929/hostfile

@@ -0,0 +1 @@
+localhost

test/issues/929/test_prog.sh

@@ -0,0 +1,3 @@
+#!/bin/sh
+
+echo "test_prog is started."

test/issues/929/x86_64_result.log

@@ -0,0 +1,74 @@
+*** C929T01 start *******************************
+mpirun -f ./hostfile -ppn 5 /home/satoken/ihk+mckernel/bin/mcexec -n 5 ./test_prog.sh
+test_prog is started.
+test_prog is started.
+test_prog is started.
+test_prog is started.
+test_prog is started.
+*** C929T01 PASSED ******************************
+
+*** C929T02 start *******************************
+mpirun -f ./hostfile -ppn 5 /home/satoken/ihk+mckernel/bin/mcexec -n 3 ./test_prog.sh
+getting CPU set for partitioned execution: Invalid argument
+getting CPU set for partitioned execution: Invalid argument
+test_prog is started.
+test_prog is started.
+test_prog is started.
+*** C929T02 PASSED ******************************
+
+*** C929T03 start *******************************
+mpirun -f ./hostfile -ppn 3 /home/satoken/ihk+mckernel/bin/mcexec -n 5 ./test_prog.sh
+getting CPU set for partitioned execution: Connection timed out
+getting CPU set for partitioned execution: Connection timed out
+getting CPU set for partitioned execution: Connection timed out
+*** C929T03 PASSED ******************************
+
+*** C929T04 start *******************************
+mpirun -f ./hostfile -ppn 6 /home/satoken/ihk+mckernel/bin/mcexec -n 3 ./test_prog.sh
+getting CPU set for partitioned execution: Invalid argument
+getting CPU set for partitioned execution: Invalid argument
+getting CPU set for partitioned execution: Invalid argument
+test_prog is started.
+test_prog is started.
+test_prog is started.
+*** C929T04 PASSED ******************************
+
+*** C929T05 start *******************************
+mpirun -f ./hostfile -ppn 250 /home/satoken/ihk+mckernel/bin/mcexec -n 250 ./test_prog.sh
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+error: nr_processes can't exceed nr. of CPUs
+...
+*** C929T05 PASSED ******************************
+
+*** C929T06 start *******************************
+mpirun -f ./hostfile -ppn 5 /home/satoken/ihk+mckernel/bin/mcexec -n 5 ./test_prog.sh
+[OK] exec: 1
+[OK] exec: 2
+[OK] exec: 3
+[OK] exec: 4
+[OK] exec: 5
+[OK] exec: 6
+[OK] exec: 7
+[OK] exec: 8
+[OK] exec: 9
+[OK] exec: 10
+[OK] exec: 11
+[OK] exec: 12
+[OK] exec: 13
+[OK] exec: 14
+[OK] exec: 15
+[OK] exec: 16
+[OK] exec: 17
+[OK] exec: 18
+[OK] exec: 19
+[OK] exec: 20
+*** C929T06 PASSED ******************************
+