Move a lot of common fuzz-test and randpkt-test code to its own file
(test-common.sh). Add build information to the error output as suggested in http://www.wireshark.org/lists/wireshark-dev/201212/msg00037.html svn path=/trunk/; revision=46446
This commit is contained in:
Gerald Combs
parent
5505655199
commit
739c90f0ad
|
|
@ -125,6 +125,7 @@ EXTRA_DIST = \
|
|||
rdps.py \
|
||||
runlex.sh \
|
||||
setuid-root.pl.in \
|
||||
test-common.sh \
|
||||
test-fuzzed-cap.sh \
|
||||
textify.sh \
|
||||
valgrind-wireshark.sh \
|
||||
|
|
|
|||
|
|
@ -9,24 +9,14 @@
|
|||
# each fuzzed file and checks for errors. The files are processed
|
||||
# repeatedly until an error is found.
|
||||
|
||||
# This needs to point to a 'date' that supports %s.
|
||||
DATE=/bin/date
|
||||
BASE_NAME=fuzz-`$DATE +%Y-%m-%d`-$$
|
||||
TEST_TYPE="fuzz"
|
||||
. `dirname $0`/test-common.sh
|
||||
|
||||
# Directory containing binaries. Default current directory.
|
||||
BIN_DIR=.
|
||||
|
||||
# Temporary file directory and names.
|
||||
# (had problems with this on cygwin, tried TMP_DIR=./ which worked)
|
||||
TMP_DIR=/tmp
|
||||
if [ "$OSTYPE" == "cygwin" ] ; then
|
||||
TMP_DIR=`cygpath --windows "$TMP_DIR"`
|
||||
fi
|
||||
TMP_FILE=$BASE_NAME.pcap
|
||||
ERR_FILE=$BASE_NAME.err
|
||||
|
||||
# Loop this many times (< 1 loops forever)
|
||||
MAX_PASSES=0
|
||||
# Sanity check to make sure we can find our plugins. Zero or less disables.
|
||||
MIN_PLUGINS=0
|
||||
|
||||
# Did we catch a signal?
|
||||
DONE=0
|
||||
|
|
@ -40,29 +30,6 @@ CONFIG_PROFILE=
|
|||
# Run under valgrind ?
|
||||
VALGRIND=0
|
||||
|
||||
# These may be set to your liking
|
||||
# Stop the child process if it's running longer than x seconds
|
||||
MAX_CPU_TIME=900
|
||||
# Stop the child process if it's using more than y * 1024 bytes
|
||||
MAX_VMEM=500000
|
||||
# Stop the child process if its stack is larger than than z * 1024 bytes
|
||||
# Windows XP: 2033
|
||||
# Windows 7: 2034
|
||||
# OS X 10.6: 8192
|
||||
# Linux 2.6.24: 8192
|
||||
# Solaris 10: 8192
|
||||
MAX_STACK=2033
|
||||
# Insert z times an error into the capture file (0.02 seems to be a good value to find errors)
|
||||
ERR_PROB=0.02
|
||||
# Trigger an abort if a dissector finds a bug.
|
||||
# Uncomment to enable
|
||||
# Note that if ABORT is enabled there will be no info
|
||||
# output to stderr about the DISSECTOR_BUG.
|
||||
# (There'll just be a core-dump).
|
||||
###export WIRESHARK_ABORT_ON_DISSECTOR_BUG="True"
|
||||
|
||||
# Sanity check to make sure we can find our plugins. Zero or less disables.
|
||||
MIN_PLUGINS=0
|
||||
|
||||
# To do: add options for file names and limits
|
||||
while getopts ":2b:C:d:e:gp:P:" OPTCHAR ; do
|
||||
|
|
@ -79,20 +46,6 @@ while getopts ":2b:C:d:e:gp:P:" OPTCHAR ; do
|
|||
done
|
||||
shift $(($OPTIND - 1))
|
||||
|
||||
# Tweak the following to your liking. Editcap must support "-E".
|
||||
TSHARK="$BIN_DIR/tshark"
|
||||
EDITCAP="$BIN_DIR/editcap"
|
||||
CAPINFOS="$BIN_DIR/capinfos"
|
||||
|
||||
if [ "$BIN_DIR" = "." ]; then
|
||||
export WIRESHARK_RUN_FROM_BUILD_DIRECTORY=1
|
||||
fi
|
||||
|
||||
# set some limits to the child processes, e.g. stop it if it's running longer then MAX_CPU_TIME seconds
|
||||
# (ulimit is not supported well on cygwin and probably other platforms, e.g. cygwin shows some warnings)
|
||||
ulimit -S -t $MAX_CPU_TIME -v $MAX_VMEM -s $MAX_STACK
|
||||
ulimit -c unlimited
|
||||
|
||||
### usually you won't have to change anything below this line ###
|
||||
|
||||
if [ $VALGRIND -eq 1 ]; then
|
||||
|
|
@ -159,70 +112,6 @@ echo ""
|
|||
trap "DONE=1; echo 'Caught signal'" HUP INT TERM
|
||||
|
||||
|
||||
##############################################################################
|
||||
### Set up environment variables for fuzz testing ###
|
||||
##############################################################################
|
||||
# Initialize (ep_ and se_) allocated memory to 0xBADDCAFE and freed memory
|
||||
# to 0xDEADBEEF
|
||||
export WIRESHARK_DEBUG_SCRUB_MEMORY=
|
||||
# Use canaries in se_ allocations (off by default due to the memory usage)
|
||||
export WIRESHARK_DEBUG_SE_USE_CANARY=
|
||||
# Verify that ep_ and se_ allocated memory is not passed to certain routines
|
||||
# which need the memory to be persistent.
|
||||
export WIRESHARK_EP_VERIFY_POINTERS=
|
||||
export WIRESHARK_SE_VERIFY_POINTERS=
|
||||
|
||||
# Turn on GLib memory debugging (since 2.13)
|
||||
export G_SLICE=debug-blocks
|
||||
# Cause glibc (Linux) to abort() if some memory errors are found
|
||||
export MALLOC_CHECK_=3
|
||||
# Cause FreeBSD (and other BSDs) to abort() on allocator warnings and
|
||||
# initialize allocated memory (to 0xa5) and freed memory (to 0x5a). see:
|
||||
# http://www.freebsd.org/cgi/man.cgi?query=malloc&apropos=0&sektion=0&manpath=FreeBSD+8.2-RELEASE&format=html
|
||||
export MALLOC_OPTIONS=AJ
|
||||
|
||||
# MacOS options; see http://developer.apple.com/library/mac/releasenotes/DeveloperTools/RN-MallocOptions/_index.html
|
||||
# Initialize allocated memory to 0xAA and freed memory to 0x55
|
||||
export MallocPreScribble=1
|
||||
export MallocScribble=1
|
||||
# Add guard pages before and after large allocations
|
||||
export MallocGuardEdges=1
|
||||
# Call abort() if heap corruption is detected. Heap is checked every 1000
|
||||
# allocations (may need to be tuned!)
|
||||
export MallocCheckHeapStart=1000
|
||||
export MallocCheckHeapEach=1000
|
||||
export MallocCheckHeapAbort=1
|
||||
# Call abort() if an illegal free() call is made
|
||||
export MallocBadFreeAbort=1
|
||||
|
||||
function exit_error() {
|
||||
echo -e "\n ERROR"
|
||||
echo -e "Processing failed. Capture info follows:\n"
|
||||
echo " Input file: $CF"
|
||||
|
||||
ERR_SIZE=$(du -sk $TMP_DIR/$ERR_FILE | awk '{ print $1 }')
|
||||
if [ $ERR_SIZE -ge 5000 ] ; then
|
||||
mv $TMP_DIR/$ERR_FILE $TMP_DIR/${ERR_FILE}.full
|
||||
head -n 2000 $TMP_DIR/${ERR_FILE}.full > $TMP_DIR/$ERR_FILE
|
||||
echo -e "\n\n[ Output removed ]\n\n" >> $TMP_DIR/$ERR_FILE
|
||||
tail -n 2000 $TMP_DIR/${ERR_FILE}.full >> $TMP_DIR/$ERR_FILE
|
||||
rm -f $TMP_DIR/${ERR_FILE}.full
|
||||
fi
|
||||
|
||||
if [ -d .svn ] ; then
|
||||
echo -e "\nSubversion revision" >> $TMP_DIR/$ERR_FILE
|
||||
svn log -l 1 >> $TMP_DIR/$ERR_FILE
|
||||
elif [ -d .git ] ; then
|
||||
echo -e "\nGit commit" >> $TMP_DIR/$ERR_FILE
|
||||
git log -1 >> $TMP_DIR/$ERR_FILE
|
||||
fi
|
||||
|
||||
echo -e "stderr follows:\n"
|
||||
cat $TMP_DIR/$ERR_FILE
|
||||
|
||||
exit 1
|
||||
}
|
||||
|
||||
# Iterate over our capture files.
|
||||
PASS=0
|
||||
while [ \( $PASS -lt $MAX_PASSES -o $MAX_PASSES -lt 1 \) -a $DONE -ne 1 ] ; do
|
||||
|
|
|
|||
|
|
@ -8,28 +8,9 @@
|
|||
# content. It runs TShark on each generated file and checks for errors.
|
||||
# The files are processed repeatedly until an error is found.
|
||||
|
||||
# Tweak the following to your liking.
|
||||
TSHARK=./tshark
|
||||
RANDPKT=./randpkt
|
||||
TEST_TYPE="randpkt"
|
||||
. `dirname $0`/test-common.sh
|
||||
|
||||
# This needs to point to a 'date' that supports %s.
|
||||
DATE=/bin/date
|
||||
BASE_NAME=randpkt-`$DATE +%Y-%m-%d`-$$
|
||||
|
||||
# Temporary file directory and names.
|
||||
# (had problems with this on cygwin, tried TMP_DIR=./ which worked)
|
||||
TMP_DIR=/tmp
|
||||
TMP_FILE=$BASE_NAME.pcap
|
||||
ERR_FILE=$BASE_NAME.err
|
||||
|
||||
# Loop this many times (< 1 loops forever)
|
||||
MAX_PASSES=0
|
||||
|
||||
# These may be set to your liking
|
||||
# Stop the child process, if it's running longer than x seconds
|
||||
MAX_CPU_TIME=900
|
||||
# Stop the child process, if it's using more than y * 1024 bytes
|
||||
MAX_VMEM=500000
|
||||
# Trigger an abort if a dissector finds a bug.
|
||||
# Uncomment to disable
|
||||
WIRESHARK_ABORT_ON_DISSECTOR_BUG="True"
|
||||
|
|
@ -46,11 +27,6 @@ while getopts ":d:p:t:" OPTCHAR ; do
|
|||
done
|
||||
shift $(($OPTIND - 1))
|
||||
|
||||
# set some limits to the child processes, e.g. stop it if it's running longer then MAX_CPU_TIME seconds
|
||||
# (ulimit is not supported well on cygwin and probably other platforms, e.g. cygwin shows some warnings)
|
||||
ulimit -S -t $MAX_CPU_TIME -v $MAX_VMEM
|
||||
ulimit -c unlimited
|
||||
|
||||
### usually you won't have to change anything below this line ###
|
||||
|
||||
# TShark arguments (you won't have to change these)
|
||||
|
|
@ -84,33 +60,6 @@ echo ""
|
|||
|
||||
trap "MAX_PASSES=1; echo 'Caught signal'" HUP INT TERM
|
||||
|
||||
function exit_error() {
|
||||
echo -e "\n ERROR"
|
||||
echo -e "Processing failed. Capture info follows:\n"
|
||||
echo " Input file: $CF"
|
||||
|
||||
ERR_SIZE=$(du -sk $TMP_DIR/$ERR_FILE | awk '{ print $1 }')
|
||||
if [ $ERR_SIZE -ge 5000 ] ; then
|
||||
mv $TMP_DIR/$ERR_FILE $TMP_DIR/${ERR_FILE}.full
|
||||
head -n 2000 $TMP_DIR/${ERR_FILE}.full > $TMP_DIR/$ERR_FILE
|
||||
echo -e "\n\n[ Output removed ]\n\n" >> $TMP_DIR/$ERR_FILE
|
||||
tail -n 2000 $TMP_DIR/${ERR_FILE}.full >> $TMP_DIR/$ERR_FILE
|
||||
rm -f $TMP_DIR/${ERR_FILE}.full
|
||||
fi
|
||||
|
||||
if [ -d .svn ] ; then
|
||||
echo -e "\nSubversion revision" >> $TMP_DIR/$ERR_FILE
|
||||
svn log -l 1 >> $TMP_DIR/$ERR_FILE
|
||||
elif [ -d .git ] ; then
|
||||
echo -e "\nGit commit" >> $TMP_DIR/$ERR_FILE
|
||||
git log -1 >> $TMP_DIR/$ERR_FILE
|
||||
fi
|
||||
|
||||
echo -e "stderr follows:\n"
|
||||
cat $TMP_DIR/$ERR_FILE
|
||||
|
||||
exit 1
|
||||
}
|
||||
|
||||
# Iterate over our capture files.
|
||||
PASS=0
|
||||
|
|
|
|||
|
|
@ -0,0 +1,144 @@
|
|||
#!/bin/bash
|
||||
#
|
||||
# $Id$
|
||||
|
||||
# Common variables and functions for fuzz and randpkt tests.
|
||||
|
||||
# This needs to point to a 'date' that supports %s.
|
||||
if [ -z "$TEST_TYPE" ] ; then
|
||||
echo "TEST_TYPE must be defined by the sourcing script."
|
||||
exit 1
|
||||
fi
|
||||
|
||||
DATE=/bin/date
|
||||
BASE_NAME=$TEST_TYPE-`$DATE +%Y-%m-%d`-$$
|
||||
|
||||
# Directory containing binaries. Default current directory.
|
||||
BIN_DIR=.
|
||||
|
||||
# Temporary file directory and names.
|
||||
# (had problems with this on cygwin, tried TMP_DIR=./ which worked)
|
||||
TMP_DIR=/tmp
|
||||
if [ "$OSTYPE" == "cygwin" ] ; then
|
||||
TMP_DIR=`cygpath --windows "$TMP_DIR"`
|
||||
fi
|
||||
TMP_FILE=$BASE_NAME.pcap
|
||||
ERR_FILE=$BASE_NAME.err
|
||||
|
||||
# Loop this many times (< 1 loops forever)
|
||||
MAX_PASSES=0
|
||||
|
||||
# These may be set to your liking
|
||||
# Stop the child process if it's running longer than x seconds
|
||||
MAX_CPU_TIME=900
|
||||
# Stop the child process if it's using more than y * 1024 bytes
|
||||
MAX_VMEM=500000
|
||||
# Stop the child process if its stack is larger than than z * 1024 bytes
|
||||
# Windows XP: 2033
|
||||
# Windows 7: 2034
|
||||
# OS X 10.6: 8192
|
||||
# Linux 2.6.24: 8192
|
||||
# Solaris 10: 8192
|
||||
MAX_STACK=2033
|
||||
# Insert z times an error into the capture file (0.02 seems to be a good value to find errors)
|
||||
ERR_PROB=0.02
|
||||
|
||||
# Tweak the following to your liking. Editcap must support "-E".
|
||||
TSHARK="$BIN_DIR/tshark"
|
||||
EDITCAP="$BIN_DIR/editcap"
|
||||
CAPINFOS="$BIN_DIR/capinfos"
|
||||
RANDPKT="$BIN_DIR/randpkt"
|
||||
|
||||
if [ "$BIN_DIR" = "." ]; then
|
||||
export WIRESHARK_RUN_FROM_BUILD_DIRECTORY=1
|
||||
fi
|
||||
|
||||
# set some limits to the child processes, e.g. stop it if it's running longer then MAX_CPU_TIME seconds
|
||||
# (ulimit is not supported well on cygwin and probably other platforms, e.g. cygwin shows some warnings)
|
||||
ulimit -S -t $MAX_CPU_TIME -v $MAX_VMEM -s $MAX_STACK
|
||||
ulimit -c unlimited
|
||||
|
||||
|
||||
##############################################################################
|
||||
### Set up environment variables for fuzz testing ###
|
||||
##############################################################################
|
||||
# Initialize (ep_ and se_) allocated memory to 0xBADDCAFE and freed memory
|
||||
# to 0xDEADBEEF
|
||||
export WIRESHARK_DEBUG_SCRUB_MEMORY=
|
||||
# Use canaries in se_ allocations (off by default due to the memory usage)
|
||||
export WIRESHARK_DEBUG_SE_USE_CANARY=
|
||||
# Verify that ep_ and se_ allocated memory is not passed to certain routines
|
||||
# which need the memory to be persistent.
|
||||
export WIRESHARK_EP_VERIFY_POINTERS=
|
||||
export WIRESHARK_SE_VERIFY_POINTERS=
|
||||
|
||||
# Turn on GLib memory debugging (since 2.13)
|
||||
export G_SLICE=debug-blocks
|
||||
# Cause glibc (Linux) to abort() if some memory errors are found
|
||||
export MALLOC_CHECK_=3
|
||||
# Cause FreeBSD (and other BSDs) to abort() on allocator warnings and
|
||||
# initialize allocated memory (to 0xa5) and freed memory (to 0x5a). see:
|
||||
# http://www.freebsd.org/cgi/man.cgi?query=malloc&apropos=0&sektion=0&manpath=FreeBSD+8.2-RELEASE&format=html
|
||||
export MALLOC_OPTIONS=AJ
|
||||
|
||||
# MacOS options; see http://developer.apple.com/library/mac/releasenotes/DeveloperTools/RN-MallocOptions/_index.html
|
||||
# Initialize allocated memory to 0xAA and freed memory to 0x55
|
||||
export MallocPreScribble=1
|
||||
export MallocScribble=1
|
||||
# Add guard pages before and after large allocations
|
||||
export MallocGuardEdges=1
|
||||
# Call abort() if heap corruption is detected. Heap is checked every 1000
|
||||
# allocations (may need to be tuned!)
|
||||
export MallocCheckHeapStart=1000
|
||||
export MallocCheckHeapEach=1000
|
||||
export MallocCheckHeapAbort=1
|
||||
# Call abort() if an illegal free() call is made
|
||||
export MallocBadFreeAbort=1
|
||||
|
||||
# Create an error report
|
||||
function exit_error() {
|
||||
echo -e "\n ERROR"
|
||||
echo -e "Processing failed. Capture info follows:\n"
|
||||
echo " Input file: $CF"
|
||||
|
||||
# Fill in build information
|
||||
echo -e "Input file: $CF\n" > $TMP_DIR/${ERR_FILE}.header
|
||||
echo -e "Build host information:" >> $TMP_DIR/${ERR_FILE}.header
|
||||
uname -a >> $TMP_DIR/${ERR_FILE}.header
|
||||
lsb_release -a >> $TMP_DIR/${ERR_FILE}.header 2> /dev/null
|
||||
|
||||
if [ -n "$BUILDBOT_BUILDERNAME" ] ; then
|
||||
echo -e "\nBuildbot information:" >> $TMP_DIR/${ERR_FILE}.header
|
||||
env | grep "^BUILDBOT_" >> $TMP_DIR/${ERR_FILE}.header
|
||||
fi
|
||||
|
||||
echo -e "\n" >> $TMP_DIR/${ERR_FILE}.header
|
||||
|
||||
if [ -d .svn ] ; then
|
||||
echo -e "\nSubversion revision" >> $TMP_DIR/${ERR_FILE}.header
|
||||
svn log -l 1 >> $TMP_DIR/${ERR_FILE}.header
|
||||
elif [ -d .git ] ; then
|
||||
echo -e "\nGit commit" >> $TMP_DIR/${ERR_FILE}.header
|
||||
git log -1 >> $TMP_DIR/${ERR_FILE}.header
|
||||
fi
|
||||
|
||||
echo -e "\n" >> $TMP_DIR/${ERR_FILE}.header
|
||||
|
||||
# Trim the stderr output if needed
|
||||
ERR_SIZE=$(du -sk $TMP_DIR/$ERR_FILE | awk '{ print $1 }')
|
||||
if [ $ERR_SIZE -ge 5000 ] ; then
|
||||
mv $TMP_DIR/$ERR_FILE $TMP_DIR/${ERR_FILE}.full
|
||||
head -n 2000 $TMP_DIR/${ERR_FILE}.full > $TMP_DIR/$ERR_FILE
|
||||
echo -e "\n\n[ Output removed ]\n\n" >> $TMP_DIR/$ERR_FILE
|
||||
tail -n 2000 $TMP_DIR/${ERR_FILE}.full >> $TMP_DIR/$ERR_FILE
|
||||
rm -f $TMP_DIR/${ERR_FILE}.full
|
||||
fi
|
||||
|
||||
cat $TMP_DIR/${ERR_FILE} >> $TMP_DIR/${ERR_FILE}.header
|
||||
mv $TMP_DIR/${ERR_FILE}.header $TMP_DIR/${ERR_FILE}
|
||||
|
||||
echo -e "stderr follows:\n"
|
||||
cat $TMP_DIR/$ERR_FILE
|
||||
|
||||
exit 1
|
||||
}
|
||||
Loading…
Reference in New Issue