Rating:

Here is a 3 solutions (choose your preferred one) write-up for CRAPSemu challenge

My teammate Shir0 start reversing Crapsemu,
I joined him because the end of the ctf was soon.. and time was running....

first we start studying the disassembly of the CRAPS program


0x0: add r29 = r0, 0x200
0x4: mov r0 = 0x20 // probably just a raw data -> b' \x00\x00\x01'
0x8: orcc r1 = r1, 0x3a
0xc: sub r29 = r29, 0x1
0x10: st r1 = r29, r0
0x14: "b'ord\x01'"
0x18: orcc r1 = r1, 0x77
0x1c: sub r29 = r29, 0x1
0x20: st r1 = r29, r0
0x24: "b'ass\x01'"
0x28: orcc r1 = r1, 0x50
0x2c: sub r29 = r29, 0x1
0x30: st r1 = r29, r0
0x34: add r1 = r0, 0x1
0x38: orcc r2 = r1, r0
0x3c: orcc r3 = r29, r0
0x40: add r4 = r0, 0xa
0x44: syscall
0x48: Unknown: 0x82184001
0x4c: orcc r2 = r1, r0
0x50: add r3 = r0, 0x100
0x54: add r4 = r0, 0x18
0x58: syscall
0x5c: add r29 = r0, 0x200
0x60: mov r0 = 0x321695 // probably just a raw data -> b'\x95\x162\x01'
0x64: orcc r1 = r1, 0x3e
0x68: sub r29 = r29, 0x1
0x6c: st r1 = r29, r0
0x70: mov r0 = 0x2740af // probably just a raw data -> b"\xaf@'\x01"
0x74: orcc r1 = r1, 0x39
0x78: sub r29 = r29, 0x1
0x7c: st r1 = r29, r0
0x80: "b'\xc5hw\x01'"
0x84: orcc r1 = r1, 0x3c
0x88: sub r29 = r29, 0x1
0x8c: st r1 = r29, r0
0x90: "b'\xc1jL\x01'"
0x94: orcc r1 = r1, 0x20
0x98: sub r29 = r29, 0x1
0x9c: st r1 = r29, r0
0xa0: mov r0 = 0x290dab // probably just a raw data -> b'\xab\r)\x01'
0xa4: orcc r1 = r1, 0x4e
0xa8: sub r29 = r29, 0x1
0xac: st r1 = r29, r0
0xb0: "b'\xa0vA\x01'"
0xb4: orcc r1 = r1, 0x5e
0xb8: sub r29 = r29, 0x1
0xbc: st r1 = r29, r0
0xc0: Unknown: 0x80180005
0xc4: bl
0xc8: orcc r6 = r6, 0xd
0xcc: subcc r0 = r29, 0x200
0xd0: Unknown: 0x2200000a
0xd4: ld r4 = r29, r0
0xd8: add r29 = r29, 0x1
0xdc: Unknown: 0x88190006
0xe0: ld r2 = r3, r0
0xe4: add r3 = r3, 0x1
0xe8: subcc r0 = r4, r2
0xec: Unknown: 0x23fffff8
0xf0: Unknown: 0x8a116001
0xf4: Unknown: 0x31fffff6
0xf8: subcc r0 = r5, r0
0xfc: Unknown: 0x22000019
0x100: orcc r29 = r29, 0x200
0x104: mov r0 = 0xa2e64 // probably just a raw data -> b'd.\n\x01'
0x108: orcc r1 = r1, 0x72
0x10c: sub r29 = r29, 0x1
0x110: st r1 = r29, r0
0x114: "b'swo\x01'"
0x118: orcc r1 = r1, 0x73
0x11c: sub r29 = r29, 0x1
0x120: st r1 = r29, r0
0x124: "b' pa\x01'"
0x128: orcc r1 = r1, 0x67
0x12c: sub r29 = r29, 0x1
0x130: st r1 = r29, r0
0x134: "b'ron\x01'"
0x138: orcc r1 = r1, 0x57
0x13c: sub r29 = r29, 0x1
0x140: st r1 = r29, r0
0x144: add r1 = r0, 0x1
0x148: orcc r2 = r1, r0
0x14c: orcc r3 = r29, r0
0x150: add r4 = r0, 0x10
0x154: syscall
0x158: add r2 = r0, 0x1
0x15c: Unknown: 0x3000001b
0x160: orcc r29 = r29, 0x200
0x164: add r1 = r0, 0xa
0x168: sub r29 = r29, 0x1
0x16c: st r1 = r29, r0
0x170: mov r0 = 0x21736e // probably just a raw data -> b'ns!\x01'
0x174: orcc r1 = r1, 0x6f
0x178: sub r29 = r29, 0x1
0x17c: st r1 = r29, r0
0x180: "b'ati\x01'"
0x184: orcc r1 = r1, 0x6c
0x188: sub r29 = r29, 0x1
0x18c: st r1 = r29, r0
0x190: "b'atu\x01'"
0x194: orcc r1 = r1, 0x72
0x198: sub r29 = r29, 0x1
0x19c: st r1 = r29, r0
0x1a0: "b'ong\x01'"
0x1a4: orcc r1 = r1, 0x43
0x1a8: sub r29 = r29, 0x1
0x1ac: st r1 = r29, r0
0x1b0: add r1 = r0, 0x1
0x1b4: orcc r2 = r1, r0
0x1b8: orcc r3 = r29, r0
0x1bc: add r4 = r0, 0x11
0x1c0: syscall
0x1c4: Unknown: 0x84188002
0x1c8: add r1 = r0, 0x2
0x1cc: syscall



well a bit ugly I know :)

we guess the program was outputting 'password: ', reading input, comparing to a static data transformed in a way or another,

maybe xor? and if password is correct it outputs 'Congratulations...'

but there were some instruction unknown, so I start reversing the first one with Ghidra.

0x48: Unknown: 0x82184001

it was this function, which is basically an xor, between to registers stored in a third destination register

or in the assembly output

I first launch a angr script to try to find an input that will print the 'Congratulations' string.. but as it was a bit slow, I forget it running in a terminal window :)

So I have the idea to put a breakpoint, at the xor instruction at 0x1cbc, (or 0x1cbe just after)

and to see what exactly was xored...?

I just launched gdb (gef variant) with the breakpoint..

gdb-gef -ex 'b main' -ex 'c' -ex 'pie breakpoint *0x1cbc' -ex 'pie run' ./crapsemu

the first xor, was 0xa with 0xa , not interesting...

then the second one...BINGO... the Flag start to appear.. complete after 6 breakpoint stops (6x 4bytes each time , 24chars)

echo -e '\x53\x50\x41\x52\x43\x5b\x3a\x3a\x2d\x31\x5d\x5f\x31\x35\x5f\x64\x34\x5f\x77\x34\x33\x65\x21\x21'

SPARC[::-1]_15_d4_w43e!!

ok got it..

then I wrote a qiling script to automatically resolve it.. (set up a correct rootfs path to run it)


import sys
sys.path.append("..")

from qiling import Qiling
from qiling.const import QL_VERBOSE

buff = b''
def xor_func(ql):
global buff
if (ql.reg.ecx):
buff += ql.reg.ecx.to_bytes(4,'little')
print('FLAG: '+buff.decode('UTF-8'))

if __name__ == "__main__":
ql = Qiling(["./crapsemu"], "rootfs/x8664_linux", verbose=QL_VERBOSE.OFF, stdin="pipo")
ql.run()


works great as you can see...

then after the CTF was finished...

I have a look to my terminal windows, where angr was running (forgot it totally...)

and guess what I see :)

Well angr resolved it also..but I didn't see it...

so here is the angr script to resolve it, for those interested...


import angr
import claripy

def main():

for i in range(16, 28):
input_len = i

flag_chars = [claripy.BVS('flag_%i' % i, 8) for i in range(input_len)]
flag = claripy.Concat( *flag_chars + [claripy.BVV(b'\n')]) # Add \n for scanf() to accept the input

st = proj.factory.full_init_state(
args='./crapsemu',
stdin=flag,
)

for byte in flag_chars:

sm = proj.factory.simulation_manager(st)
sm.run()

y = []