| Summary: | Expansion of attack vectors via fuzzing was achieved by acquiring relevant seeds from a
large corpus in a short time span. We present a framework capable of extracting,
categorising and managing object streams from fuzzing seeds according to type and
relevance.
Heuristic function of the framework was built upon experimentation and investigation into
object streams. The result is a framework capable of processing a 78KB file with multiple
objects in 29 seconds, a vast improvement in comparison to similar average timing for
manual processing of only 1 object. 100 small to medium sized PDF (< 1000KB each) in 49
minutes.
14 seeds processed by framework were fuzzed using Failure Observation Engine (FOE) 2.1,
each with minimum 50k iterations. No crashes were produced.
|
|---|