Give valgrind fuzzing a bit more time. Otherwise we have some large (~80MB)

captures in the menagerie which time out even under normal circumstances.

svn path=/trunk/; revision=54941

diff --git a/tools/fuzz-test.sh b/tools/fuzz-test.sh
index face22c7939aafe1f256a848027d57fccd11f619..1a8f54bf690bfedfe3f7466ba91b3dea03f5c160 100755 (executable)
--- a/tools/fuzz-test.sh
+++ b/tools/fuzz-test.sh
@@ -77,8 +77,9 @@ if [ $VALGRIND -eq 1 ]; then
     RUNNER="`dirname $0`/valgrind-wireshark.sh"
     COMMON_ARGS="-b $BIN_DIR $COMMON_ARGS"
     declare -a RUNNER_ARGS=("" "-T")
-    # Valgrind requires more resources, so bump our limits to 1.5 times normal
-    MAX_CPU_TIME=`expr 3 \* $MAX_CPU_TIME / 2`
+    # Valgrind requires more resources, so permit 1.5x memory and 2x time
+    # (1.5x time is too small for a few large captures in the menagerie)
+    MAX_CPU_TIME=`expr 2 \* $MAX_CPU_TIME`
     MAX_VMEM=`expr 3 \* $MAX_VMEM / 2`
 else
     # Not using valgrind, use regular tshark.