2026年8月25日,开发者 daniellionel01 开发了一款针对 Gleam 编译器的结构感知模糊测试工具,通过随机生成类型安全的 Gleam 程序并比较其在 Erlang 和 JavaScript 两个目标上的输出差异来发现编译器缺陷 1。Gleam 编译器使用 Rust 编写,可编译为 JavaScript 和 Erlang 两个目标 1。
该工具已发现 9 个问题,包括代码生成错误、编译器崩溃以及一个上报至 Erlang/OTP 的问题 1。其中,发现的 Gleam 编译器问题包括 #6179、#6180、#6181、#6182、#6187、#6192、#6212 和 #6213 1。另外 1 个问题被上报至 Erlang/OTP(issue #11494)1。此外,作者尝试使用 LLM 进行模糊测试,花费 20 美元发现 1 个问题(issue #5613)1。
On August 25, 2026, developer daniellionel01 developed a structure-aware fuzzing tool designed specifically for the Gleam compiler 1. The Gleam compiler is written in Rust and can compile programs to both JavaScript and Erlang targets 1. To identify compiler defects, the tool randomly generates type-safe Gleam programs and compares the output differences between the Erlang and JavaScript targets 1.
Through this fuzzing approach, the tool has uncovered nine distinct issues within the Gleam compiler, including code generation errors and compiler crashes 1. The specific compiler bugs identified are tracked under issue numbers #6179, #6180, #6181, #6182, #6187, #6192, #6212, and #6213 1. Furthermore, one of the discovered problems was reported to the Erlang/OTP project as issue #11494 1. In addition to the custom fuzzing tool, the developer also experimented with utilizing a Large Language Model (LLM) for fuzzing, spending $20 to find one additional compiler defect tracked as issue #5613 1.
评论
还没有评论,欢迎留下第一条。