Fix exit status of `run` scripts executing more than one test.
For some ART run-tests, the `run` script executes more than one
instance of the test (e.g. test/004-ThreadStress/run executes the
ThreadStress test twice: the first time as a normal run, the second
time in locks-only mode with stack-dump lock profiling). However,
these tests were returning the exit status of last test executed,
which could hide a potential earlier failure. This change ensure we
return the first failing exit status, if any.
Test: art/test/testrunner/testrunner.py
Change-Id: I5e4e4cc7d9311fe15637ea2f5248a0e9f2432d61
diff --git a/test/018-stack-overflow/run b/test/018-stack-overflow/run
index 1a71a1a..7443bd7 100755
--- a/test/018-stack-overflow/run
+++ b/test/018-stack-overflow/run
@@ -17,7 +17,12 @@
# Run normal. This will be the debug build.
echo "libartd run."
${RUN} "${@}"
+return_status1=$?
# Run non-debug.
echo "libart run."
${RUN} "${@/#libartd.so/libart.so}"
+return_status2=$?
+
+# Make sure we don't silently ignore an early failure.
+(exit $return_status1) && (exit $return_status2)