RE: https://infosec.exchange/@andersonc0d3/116903287835170560 The fixes in the quoted post are a great example of LLM-assisted development. One issue was fixed, and Sashiko automatically noticed that another component had the same issue. This has likely happened several times in Linux kernel history. I remember one case I posted about years ago — it took 301 days for the fix in the other component to land upstream. In this era of LLM-assisted development, the similar issue was spotted in just 17 seconds. "An example I often mention is the issues below. The three mechanisms (message queues, semaphores, and shared memory) of the IPC subsystem in the Linux kernel share code. Knowing this, if there's a vulnerability affecting one mechanism, it’s worth checking if it also affects the other mechanisms. They fixed a vulnerability in the semaphore mechanism but forgot the message queues and shared memory mechanisms. Analyzing public fixed vulnerabilities can help you learn about the inner details of the affected subsystem and sometimes even lead to new vulnerabilities being discovered." ipc/sem.c: fully initialize sem_array before making it visible https://github.com/torvalds/linux/commit/e8577d1f0329d4842e8302e289fb2c22156abef4 Initialize msg/shm IPC objects before doing ipc_addid() https://github.com/torvalds/linux/commit/b9a532277938798b53178d5a66af6e2915cb27cf