CAP6135: Malware and Software Vulnerability Analysis Buffer Overflow : Example of Using GDB to Check Stack Memory Cliff Zou Spring 2015

A Stack Frame Parameters Return Address Calling Stack Pointer Added Protection Local Variables BP SP+offset SP Addresses 00000000 SP: stack pointer BP: base/frame pointer Calling stack pointer: previous function’s SP

Using GDB to Check Stack GDB tutorial: http://sunsite.ualberta.ca/Documentation/Gnu/gdb-4.18/html_chapter/gdb_toc.html http://www.yolinux.com/TUTORIALS/GDB-Commands.html#GDB_COMMAND_LINE_ARGS When compile the c code, use “gcc –g …..” so that Gdb can match source code line number with code Some knowledge: http://en.wikipedia.org/wiki/X86_assembly_language Register eip: instruction pointer, the current position of next executable instruction Register ebp: stack pointer, the top of the current stack, used for addressing local variable

Related Gdb Commands: List: list the source code and each execution’s corresponding line number Break linenumber: set breakpoint at the linenumber Break test.c:foo break when program run in the foo() function in test.c file. Run argv: run the execution code with the parameter argv Next: execute the next line of code Backtrace: show trace of all function calls in stack Info frame: List address, language, address of arguments/local variables and which registers were saved in frame. This will show where the return address is saved Return address is in Register EIP Calling stack pointer is in Register EBP x &variable: show the address and value of a local variable (in hex format) x address: print binary representation of 4 bytes of memory pointed to by address.

Example of Using GDB #include <stdio.h> void foo(char * input){ int a1=11; int a2=22; char buf[7]; strcpy(buf, input); } void main(int argc, char **argv){ foo(argv[1]); Question: What does the stack look like before strcpy()?

7 Remove address randomization czou@eustis:~/buffer-code$ setarch i686 –R gdb ./gdb-example (gdb) list 1 #include <stdio.h> 2 void foo(char * input){ 3 int a1=11; 4 int a2=22; 5 char buf[7]; 6 strcpy(buf, input); 7 } 8 void main(int argc, char **argv){ 9 foo(argv[1]); 10 } (gdb) break 6 Breakpoint 1 at 0x8048459: file gdb-example.c, line 6. (gdb) run “what is this? a book” Starting program: /home/czou/buffer-code/gdb-example “what is this? a book" Breakpoint 1, foo (input=0xbffff838 "1234567890") at gdb-example.c: 6 strcpy(buf, input); Remove address randomization used in Unix (will talk in next lecture)

(gdb) info frame Stack level 0, frame at 0xbffff620: eip = 0x8048459 in foo (gdb-example.c:6); saved eip 0x8048497 called by frame at 0xbffff640 source language c. Arglist at 0xbffff618, args: input=0xbffff82d "what is this? a book" Locals at 0xbffff618, Previous frame's sp is 0xbffff620 Saved registers: ebp at 0xbffff618, eip at 0xbffff61c (gdb) x &a1 0xbffff5fc: 0x0000000b (gdb) x &a2 0xbffff600: 0x00000016 (gdb) x buf 0xbffff605: 0xf4000000

Two Techniques for Generating Stack Overflow Codes

NOPs Most CPUs have a No-Operation instruction – it does nothing but advance the instruction pointer. Usually we can put a bunch of these ahead of our program (in the string). As long as the new return-address points to a NOP we are OK.

(exec /bin/ls or whatever) Using NOPs new return address Real program (exec /bin/ls or whatever) Can point anywhere in here nop instructions

Estimating the stack size We can also guess at the location of the return address relative to the overflowed buffer. Put in a bunch of new return addresses!

Estimating the Location new return address new return address new return address new return address new return address new return address Real program nop instructions

Explanation of Project 1 Target.c code vulnerability: int foo(char* arg, short arglen) { char buffer[85]; int i, maxlen = 80; int len; if (arglen < maxlen) len = strlen(arg); strncpy(buffer, arg, len); If input to foo(*arg, Big_Value) where Big_Value overflows ‘short’, then arglen could be negative value and passes the if() security check.

Explanation of Project 1 In the exploit.c code: #define TARGET “/home/czou/cap6135-project1/targets/target” Need to be changed to point to your own target executable code Change args[1] = “…….."; args[1] needs to point to a large buffer that can cause overflow to target code You can define such a large buffer in exploit.c and make args[1] points to it. Your main task is to: Find out where in stack stores the return address Find out where is the starting address of ‘buffer’ in foo() in target code Fill the shellcode[] into the large buffer in your exploit code (which will fill the ‘buffer’ variable in target code) Assign the starting address of buffer to the right place in the large buffer in your exploit code in order to overwrite the return address, then CPU will run the shellcode you put at the start of buffer variable.

Several Tips on Project 1 Be sure to use the Makefile to generate executable of both exploit program and target program At both ./exploit and ./target directory, run “make” Be sure to use “setarch i686 -R” in front of every execution, including both Gdb and direct execution of ./exploit You can use “break target.c:foo” to set breakpoint upon entering foo() function. Fill the shell executable code (in the string array shellcode[]) byte-by-byte into the buffer for your modified return address to execute, do not use strcpy() because shellcode[] is not an ASCII string. Make sure the long string you create has no NULL byte except the last byte

Several Tips on Project 1 As an example, suppose we know that: The address of ‘buffer’ in target.c is: 0xbfff0000 The address of the function’s return address (eip) is 0xbfff0100 We put the shellcode[] at the beginning of ‘buffer’. How to Overwrite the return address to execute shellcode? 0xbfff0100 – 0xbfff0000 = 0x100 = 256 in decimal Since address in 32-bit machine is 4 bytes and Eustis is a little-endian machine: buffer[256] = 0x00; buffer[257] = 0x00; buffer[258] = 0xff; buffer[259] = 0xbf; In this way, we have changed the flow to the beginning of shellcode!