oss-sec mailing list archives

Re: Thousands of vulnerabilities, almost no CVEs: OSS-Fuzz


From: Matthew Fernandez <matthew.fernandez () gmail com>
Date: Mon, 24 Jun 2019 16:44:03 -0700



On Jun 24, 2019, at 09:42, Bob Friesenhahn <bfriesen () simple dallas tx us> wrote:

On Mon, 24 Jun 2019, Stuart D. Gathman wrote:

Question: is fuzzing useful for languages like Java/python?  Obviously,
you eventually reach a native code module in both cases, but fuzzing the entire virtual machine is cumbersome.  
Maybe native code libraries
for "safe" languages should include fuzzing as part of testing.

There is nothing about languages like Java and Python which necessarily makes them safe.  Access outside of memory 
bounds is just one issue which often afflicts C/C++.  Java and Python can easily do something wrong such as use all 
available resources or never finish. In the case of Python, Python can easily make arbitrary calls into C code under 
control of the script.

With something like ctypes? I took Stuart’s point to be about fuzzing the language VM (e.g. CPython), rather than 
fuzzing extensions written in C. Fuzzing the FFI into native code seems like a reasonable idea for extension 
maintainers, but this is much less cumbersome than the VM fuzzing that Stuart’s comment seemed to be getting at.

Current thread: