Skip to content

Comment on Linux DRM for games is now here

You are viewing one comment and its replies. View all comments · View the comment above

Comments

  • No, BPF is root. It’s like super ultra root. Bpf programs running in the kernel can do everything root does, and more, most notably, intercepting with and tampering with what root does/sees. There are plenty of examples of BPF being used as super root and backdoors:

    You do Kernel access without having root access.

    Okay, technically, you can do unprivileged bpf access. But the specific bpf lsm outlined in the post is not that. It is super-root.

    8
    • Alright then. I need to read more about this, maybe I was misinformed. But I’m not super convinced yet, maybe you are misinformed (no accusation here, just saying it could be).

      -2
      • I don’t know your level of programming knowledge, but I will back up the previous poster.

        With eBPF you can add code that runs right before various kernel functions, in kernel space (what they’re calling super root)

        E.g. https://eunomia.dev/tutorials/19-lsm-connect/

        Some of the hook points currently supported by LSM include:
        
        * File open, creation, deletion, and movement;
        * Filesystem mounting;
        * Operations on tasks and processes;
        * Operations on sockets (creating, binding sockets, sending and receiving messages, etc.);
        

        The rules around eBPF are more for preventing a process from locking up a kernel thread than what it can access

        It was originally created for high speed routing where making decisions about where network traffic goes was too slow in userspace, but they didn’t want a completely custom kernel module

        9

Sign in to comment. Sign in