Common use of Sandboxing Clause in Contracts

Sandboxing. The goal of sandboxing is to enable the safe execution of untrusted, potentially malicious code. This is achieved by ensuring that the untrusted code is confined to a set of tightly controlled resources. Here we focus on one important aspect: preventing code from reading outside of its own subset of the address space. To achieve this, just-in-time compilers enforce access-control policies by inserting checks to ensure that all memory accesses happen within the sandbox’s bounds. We describe sandboxes using policies π, where memory out- side of the sandbox is declared high. To account for programs that may escape the sandbox by exploiting speculation across access-control checks, we make the following distinction: • Traditional sandboxing approaches [24], [25] check/en- force vanilla sandboxing: A program p is vanilla-sandboxed