#
# DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
#
# under the terms of the GNU General Public License version 2 only, as
# published by the Free Software Foundation.
#
# This code 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
# version 2 for more details (a copy is included in the LICENSE file that
# accompanied this code).
#
# You should have received a copy of the GNU General Public License version
# 2 along with this work; if not, write to the Free Software Foundation,
# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
#
# Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
# or visit www.oracle.com if you need additional information or have any
# questions.
#
# @test
# @bug 6942989
# @summary Check for WeakReference leak in Logger objects
# @author Daniel D. Daugherty
#
# @build SimpleApplication ShutdownSimpleApplication
# @build LoggerWeakRefLeak
# @run shell/timeout=240 LoggerWeakRefLeak.sh
# The timeout is: 2 minutes for infrastructure and 2 minutes for the test
#
TEST_NAME="LoggerWeakRefLeak"
TARGET_CLASS="java\.lang\.ref\.WeakReference"
# MAIN begins here
#
seconds="$1"
fi
# see if this version of jmap supports the '-histo:live' option
jmap_option="-histo:live"
set +e
status="$?"
set -e
if [ "$status" != 0 ]; then
# usage message doesn't show ':live' option
if $isWindows; then
# If SA isn't present, then jmap gives a different usage message
# that doesn't show the ':live' option. However, that's a bug that
# is covered by 6971851 so we try using the option just to be sure.
# For some reason, this problem has only been seen on OpenJDK6 on
# Windows. Not sure why.
set +e
# Note: Don't copy this code to try probing process 0 on Linux; it
# will kill the process group in strange ways.
status="$?"
set -e
if [ "$status" = 0 ]; then
# Usage message generated so flag the problem.
status=1
else
# No usage message so clear the flag.
status=0
fi
fi
if [ "$status" != 0 ]; then
echo "ERROR: 'jmap $jmap_option' is not supported so this test"
echo "ERROR: cannot work reliably. Aborting!"
exit 2
fi
fi
# Start application and use TEST_NAME.port for coordination
finished_early=false
loop_cnt=0
while true; do
# see if the target process has finished its run and bail if it has
set +e
status="$?"
set -e
if [ "$status" = 0 ]; then
break
fi
# Output format for 'jmap -histo' in JDK1.5.0:
#
# <#bytes> <#instances> <class_name>
#
# Output format for 'jmap -histo:live':
#
# <num>: <#instances> <#bytes> <class_name>
#
set +e
status="$?"
set -e
if [ "$status" != 0 ]; then
echo "INFO: jmap exited with exit code = $status"
# There are intermittent jmap failures; see 6498448.
#
# So far the following have been observed in a jmap call
# that was not in a race with target process termination:
#
# (Solaris specific, 2nd sample)
# <pid>: Unable to open door: target process not responding or HotSpot VM not loaded
# The -F option can be used when the target process is not responding
#
# (on Solaris so far)
#
# (on Solaris so far, 1st sample)
# <pid>: Permission denied
#
if [ "$loop_cnt" = "$loop_cnt_on_retry" ]; then
# loop count hasn't changed
else
# loop count has changed so remember it
fi
# This is '-ge' because we have the original attempt plus
# MAX_JMAP_TRY_CNT - 1 retries.
echo "INFO: jmap failed $MAX_JMAP_TRY_CNT times in a row" \
"without making any progress."
echo "FAIL: jmap is unable to take any samples." >&2
exit 2
fi
# short delay and try again
# Note: sleep 1 didn't help with "<pid>: Permission denied"
sleep 2
echo "INFO: retrying jmap (retry=$jmap_retry_cnt, loop=$loop_cnt)."
continue
fi
set +e
| sed '
# strip leading whitespace; does nothing in JDK1.5.0
s/^'"${PATTERN_WS}${PATTERN_WS}"'*//
# strip <#bytes> in JDK1.5.0; does nothing otherwise
s/^[1-9][0-9]*'"${PATTERN_WS}${PATTERN_WS}"'*//
# strip <num>: field; does nothing in JDK1.5.0
s/^[1-9][0-9]*:'"${PATTERN_WS}${PATTERN_WS}"'*//
# strip <class_name> field
s/'"${PATTERN_WS}"'.*//
'`
set -e
if [ -z "$instance_cnt" ]; then
echo "INFO: instance count is unexpectedly empty"
if [ "$loop_cnt" = 0 ]; then
echo "INFO: on the first iteration so no sample was found."
echo "INFO: There is likely a problem with the sed filter."
echo "INFO: start of jmap output:"
cat "$TEST_NAME.jmap"
echo "INFO: end of jmap output."
echo "FAIL: cannot find the instance count value." >&2
exit 2
fi
else
echo "INFO: instance_cnt = $instance_cnt"
else
# actually decreasing or the same
# For these particular WeakReference leaks, the count was
# always observed to be increasing so if we get a decreasing
# or the same count, then the leaks are fixed in the bits
# being tested.
echo "INFO: finishing early due to non-increasing instance count."
finished_early=true
break
fi
fi
# delay between samples
sleep 5
done
if [ $finished_early = false ]; then
stopApplication "$TEST_NAME.port"
fi
echo "INFO: $TEST_NAME has finished running."
echo "INFO: increasing_cnt = $increasing_cnt"
echo "INFO: decreasing_cnt = $decreasing_cnt"
echo "INFO: jmap_retry_cnt = $jmap_retry_cnt (in $loop_cnt iterations)"
fi
if [ "$loop_cnt" = 0 ]; then
echo "FAIL: jmap is unable to take any samples." >&2
exit 2
fi
if [ "$decreasing_cnt" = 0 ]; then
echo "INFO: is always increasing."
echo "FAIL: This indicates that there is a memory leak." >&2
exit 2
fi
echo "INFO: is not always increasing."
echo "PASS: This indicates that there is not a memory leak."
exit 0