長話短說:如何透過 UART 將遠端輸出正確傳輸tcpdump
到本地wireshark
?
我嘗試捕獲流經嵌入式設備的資料包,但我無法在該設備上安裝任何東西。幸運的是,串口上打開了getty,並安裝了tcpdump。可惜的是,沒有 SSH、沒有 dumpcap、沒有 tshark。
直管式
我首先嘗試配置tty並透過管道將資料傳遞給wireshark。
stty -F /dev/ttyUSB0 raw
stty -F /dev/ttyUSB0 -echo -echoe -echok
cat /dev/ttyUSB0 | wireshark -k -i -
# On another terminal:
echo "tcpdump -U -s0 -i eth0 -w - 2>/dev/null" > /dev/ttyUSB0
Wireshark 抱怨輸入不是有效的 libpcap 格式,當然是因為指令被回顯,而我沒有設法擺脫它。
使用原始 PySerial
所以我決定創建一個 python 腳本來控制管道的工作方式:
import serial
import sys
import subprocess
import fcntl
def main(args):
with serial.Serial('/dev/ttyUSB0', 115200, timeout=0) as ser:
length = ser.write(b"tcpdump -U -s0 -i eth0 -w - 2> /dev/null\n") + 1
# Discard the echoed command line
while length > 0:
discard = ser.read(length)
length -= len(discard)
# Spawn wireshark
wireshark = subprocess.Popen(
["wireshark", "-k", "-i", "-"], stdin=subprocess.PIPE
)
# Pipe data from serial to wireshark's input
while True:
data = ser.read(256)
wireshark.stdin.write(data)
try:
wireshark.stdin.flush()
except BrokenPipeError as e:
break
if len(data) > 0: print(data)
# Send "Ctrl+C" to tcpdump
ser.write(b"\x03")
wireshark.wait()
return 0
if __name__ == '__main__':
import sys
sys.exit(main(sys.argv))
撇開腳本如何正確結束的一些問題不談,這並沒有我想像的那麼好。 Wireshark 高興了一段時間,但很快輸入就損壞了,記錄也停止了。我認為這是因為主機上的tty仍然轉換一些特殊字符,可能是換行或回車。
變得愚蠢:透過 PySerial 進行十六進位轉儲
所以我知道這很蹩腳,但由於我沒有其他想法,這就是我想出來的:
import serial
import sys
import subprocess
import binascii
def main(args):
with serial.Serial('/dev/ttyUSB0', 115200, timeout=5) as ser:
# Spawn tcpdump on the host and convert the raw output to stupid hex format
# We need hexdump -C because that's the only format that doesn't mess up with the endianess
length = ser.write(b"tcpdump -U -s256 -i eth0 -w - 2> /dev/null | hexdump -C\n")
# Discard command line that is echoed
discard = ser.readline()
# Spawn wireshark
wireshark = subprocess.Popen(
["wireshark", "-k", "-i", "-"], stdin=subprocess.PIPE
)
while True:
# Process each line separately
data = ser.readline().decode('ascii')
elements = data.split()
# Remove the address and ascii convertion of hexdump and spaces
hexa = "".join(elements[1:17])
# Convert back hex to binary
real_data = binascii.unhexlify(hexa)
# Feed to the shark
wireshark.stdin.write(real_data)
try:
wireshark.stdin.flush()
except BrokenPipeError as e:
break
# Stop tcpdump
ser.write(b"\x03")
wireshark.wait()
return 0
if __name__ == '__main__':
import sys
sys.exit(main(sys.argv))
唉,雖然它的工作時間比以前的版本長一點,但當框架有點太大時,wireshark會彈出一個問題,說框架太大,長度確實很荒謬(比如-1562980309832),然後再一次錄音停止。
請幫忙! :)
您可以注意到,我嘗試使用-s
tcpdump 選項,但它不起作用,即使數量很少。
我也嘗試過 picocom 的管道,但無濟於事。
所以,如果您有任何想法,任何可以工作的 UART 隧道軟體,對我(不稱職的)使用 stty 的任何評論,或者對我的 python 腳本的任何改進,我會非常高興!
Wireshark 是 2.2.5,tcpdump 是 4.5.0,libpcap 是 1.5.0。
答案1
最後,我終於讓它真正發揮作用了。這不是完美的設置,但至少它有效,所以也許它可以幫助將來的某人。
我在上面使用了一個Python腳本py串口透過 UART 啟動 tcpdump 並使用 hexdump,以便二進位資料可以遍歷鏈接,而不會被 tty 轉錄規則修改。然後,Python 腳本將資料轉換回並透過管道傳輸到wireshark。下面的腳本是結果,與問題中的腳本相比,我-v
為 hexdump 新增了選項,這樣它就不會嘗試壓縮相同的行。
import serial
import sys
import subprocess
import binascii
def main(args):
with serial.Serial('/dev/ttyUSB0', 115200, timeout=5) as ser:
# Spawn tcpdump on the host and convert the raw output to stupid hex format
# We need hexdump -C because that's the only format that doesn't mess up with the endianess
length = ser.write(b"tcpdump -U -s256 -i eth0 -w - 2> /dev/null | hexdump -Cv\n")
# Discard command line that is echoed
discard = ser.readline()
# Spawn wireshark
wireshark = subprocess.Popen(
["wireshark", "-k", "-i", "-"], stdin=subprocess.PIPE
)
while True:
# Process each line separately
data = ser.readline().decode('ascii')
elements = data.split()
# Remove the address and ascii convertion of hexdump and spaces
hexa = "".join(elements[1:17])
# Convert back hex to binary
real_data = binascii.unhexlify(hexa)
# Feed to the shark
wireshark.stdin.write(real_data)
try:
wireshark.stdin.flush()
except BrokenPipeError as e:
break
# Stop tcpdump
ser.write(b"\x03")
wireshark.wait()
return 0
if __name__ == '__main__':
import sys
sys.exit(main(sys.argv))