[Ezbench-dev] [PATCH 22/25] runner.sh: Don't mark test as completed unless it signals as completed
Petri Latvala
petri.latvala at intel.com
Fri Feb 24 11:19:23 UTC 2017
Signed-off-by: Petri Latvala <petri.latvala at intel.com>
---
runner.sh | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/runner.sh b/runner.sh
index 65872b8..132aad4 100755
--- a/runner.sh
+++ b/runner.sh
@@ -422,6 +422,14 @@ function execute_test {
"$execFuncName" > "$run_log_file" 2> /dev/null
fi
local exit_code=$?
+ if [ "$testHasExitCode" -eq 1 ]; then
+ # If the testing requested a reboot etc, we don't really want to mark it as completely tested
+ # The exit codes signaling a completed test are 0 (no error) and 19 (already complete)
+ if [ ! "$exit_code" -eq 0 && ! "$exit_code" -eq 19 ]; then
+ return "$exit_code"
+ fi
+ fi
+
write_to_journal tested "$version" "${testName}" "$run_log_file_name"
callIfDefined benchmark_run_post_hook
--
2.9.3
More information about the Ezbench-dev
mailing list