automated, linux, ssuite: produce result file

This commit makes the ssuite produce a result file and send it to
lava. For each passed test case, the average value of the main
quantity measured is reported. Here is the format of possible lines in
the result file:

throughput-<scheduler1_name>--<workload_name> pass <real number> MB/s
throughput--<workload1_name>--<scheduler1_name> fail

<app_name>-startup--<workload_name>--<scheduler1_name> pass <real number> sec
<app_name>-startup--<workload_name>--<scheduler1_name> fail

Signed-off-by: Paolo Valente <paolo.valente@linaro.org>

automated/linux/ssuite/run-bench.sh

@@ -2,6 +2,7 @@
 
 . ../../lib/sh-test-lib
 OUTPUT="$(pwd)/output"
+RESULT_FILE="${OUTPUT}/result.txt"
 
 TESTS="throughput replayed-startup"
 TEST_DEV=sda
@@ -157,7 +158,19 @@ run_test() {
 	rm -f ${HOME_DIR}/.S-config.sh
 
 	cd "$S_PATH"/run_multiple_benchmarks/ || exit 1
-	./run_main_benchmarks.sh "$1" 2>&1 | tee -a "${OUTPUT}/log"
+	./run_main_benchmarks.sh "$1" "" "" "" "" "" 2 "${OUTPUT}" 2>&1 |\
+	    tee -a "${OUTPUT}/log"
+
+	# In the result file, the average value of the main quantity
+	# measured is reported. For each passed test case. Here is the
+	# format of possible lines in the result file:
+	#
+	# throughput-<scheduler1_name>--<workload_name> pass <real number> MB/s
+	# throughput--<workload1_name>--<scheduler1_name> fail
+	#
+	# <app_name>-startup--<workload_name>--<scheduler1_name> pass <real number> sec
+	#<app_name>-startup--<workload_name>--<scheduler1_name> fail
+	mv "${OUTPUT}/result_list.txt" "${RESULT_FILE}"
 }
 
 ! check_root && error_msg "This script must be run as root"

automated/linux/ssuite/ssuite-bench.yaml

@@ -73,3 +73,4 @@ run:
     steps:
         - cd ./automated/linux/ssuite/
         - ./run-bench.sh -t "${TESTS}" -d "${TEST_DEV}" -f "${FORMAT}" -v "${S_VERSION}"
+        - ../../utils/send-to-lava.sh ./output/result.txt