Infinity loop when reading complex expression

[cynecx]

Hi,

I am currently trying to transform this expression into an arybo expression:

(((((((- (((((((((((((((- x) & 0xffffffff) + 0x2C) & 0xffffffff) & ((((((- x) & 0xffffffff) + 0x2C) & 0xffffffff) | 0xC4) & 0xffffffff)) & 0xffffffff) ^ ((x + 0xFFFFFFD4) & 0xffffffff)) & 0xffffffff) + 0x80403064) & 0xffffffff) ^ 0xED) & 0xffffffff) | 0x403053) & 0xffffffff) & 0xff)) & 0xff) + 0x7E) & 0xff) & 0xff) << 0) | ((0x0 & 0xffffff) << 8))

With these instructions:

from arybo.lib import MBA

def transform(x):
    return (((((((- (((((((((((((((- x) & 0xffffffff) + 0x2C) & 0xffffffff) & ((((((- x) & 0xffffffff) + 0x2C) & 0xffffffff) | 0xC4) & 0xffffffff)) & 0xffffffff) ^ ((x + 0xFFFFFFD4) & 0xffffffff)) & 0xffffffff) + 0x80403064) & 0xffffffff) ^ 0xED) & 0xffffffff) | 0x403053) & 0xffffffff) & 0xff)) & 0xff) + 0x7E) & 0xff) & 0xff) << 0) | ((0x0 & 0xffffff) << 8))

mba = MBA(32)
x = mba.var('x')

print transform(x)

But apparently, it somehow runs into an infinity loop as it run on 100% cpu for ~10 minutes until I killed the process.

[cynecx]

I have reduced the testcase to this small expression:

from arybo.lib import MBA

def transform(x):
    a = 1 - x
    b = x + 1
    c = a ^ b
    return c

mba = MBA(32)
x = mba.var('x')

print (transform(x))

[aguinet]

Hello!

If you try it on 16 bits this will work in a decent amount of time. I think the expression start to be become too big above 16 bits to handle.
By the way, I am working on a new tree-based storage for boolean expression, that can consume more memory but be really more efficient in some cases!

[aguinet]

By the way, for your original expression, you can use "lazy evaluation" using Arybo expressions. This is a less documented feature, I've got to put it in the final documentation.

Here is the example script:

from arybo.lib import MBA
from arybo.lib.mba_exprs import ExprBV, eval_expr

def transform(x):
        return (((((((- (((((((((((((((- x) & 0xffffffff) + 0x2C) & 0xffffffff) & ((((((- x) & 0xffffffff) + 0x2C) & 0xffffffff) | 0xC4) & 0xffffffff)) & 0xffffffff) ^ ((x + 0xFFFFFFD4) & 0xffffffff)) & 0xffffffff) + 0x80403064) & 0xffffffff) ^ 0xED) & 0xffffffff) | 0x403053) & 0xffffffff) & 0xff)) & 0xff) + 0x7E) & 0xff) & 0xff) << 0) | ((0x0 & 0xffffff) << 8))

mba = MBA(32)
x = mba.var('x')
x = ExprBV(x)

f = transform(x)
print(f)
print(eval_expr(f))

This outputs:

Vec([
1,
1,
((x0 * x1) + x0 + x1),
((x0 * x2) + (x1 * x2) + (x0 * x1 * x2) + x2 + 1),
((x0 * x1) + x0 + x1),
((x2 * x3) + (x0 * x2 * x3) + (x1 * x2 * x3) + (x2 * x3 * x4) + (x0 * x1 * x2 * x3) + (x0 * x2 * x3 * x4) + (x1 * x2 * x3 * x4) + (x0 * x1 * x2 * x3 * x4)),
0,
((x2 * x3 * x5 * x6) + (x0 * x2 * x3 * x5 * x6) + (x1 * x2 * x3 * x5 * x6) + (x2 * x3 * x4 * x5 * x6) + (x0 * x1 * x2 * x3 * x5 * x6) + (x0 * x2 * x3 * x4 * x5 * x6) + (x1 * x2 * x3 * x4 * x5 * x6) + (x0 * x1 * x2 * x3 * x4 * x5 * x6) + 1),
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0
])

[cynecx]

Thank you :) I guess that will do it for now.