• 5ymm3trY@discuss.tchncs.de
        link
        fedilink
        arrow-up
        3
        ·
        16 hours ago

        I haven’t checked the scripts from OP, but i think there is a script that is provided by the CachyOS team that basically just contains a list of compromised packages and compares that to your pacman -Qm output. If it finds a match, it tells you that the compromized package X is on your system. That seems pretty reasonable.

        I get your point and as always, you should check the source of the script as well as the code inside of it. Never installing anything outside of official OS repositories is probably not an option for most people. There are always pros and cons. Like in my example maybe some OS maintainers know more about the affected packages than I do with a quick search. On the other hand, the script might be outdated because the number of packages changed a lot over the last few days.

        • HaraldvonBlauzahn@feddit.org
          link
          fedilink
          arrow-up
          1
          arrow-down
          14
          ·
          16 hours ago

          but i think there is a script that is provided by the CachyOS team that basically just contains a list of compromised packages and compares that to your pacman -Qm output.

          So, the CachyOS maintainers suggest running untrusted code?

          Noice. I don’t need to know more.

          • radamant@lemmy.world
            link
            fedilink
            English
            arrow-up
            10
            ·
            13 hours ago

            In which way is it untrusted? If you use CachyOS, you use their binaries, that could contain anything at all inside of them. Do you draw the line at a shelll script you can read yourself?

          • 5ymm3trY@discuss.tchncs.de
            link
            fedilink
            arrow-up
            3
            ·
            13 hours ago

            I don’t use CachyOS nor do I know anything about their team and I haven’t used the script either. My point was just that I would trust OS maintainers more than some random guy on the internet.

            I have checked again and it seems the script I was referring to was actually from a mod on their community forum. Not sure if this is a maintainer as well or not.

            My point still stands, if you trust the source and checked the code that nothing shady is going on, it is perfectly fine to run a script. Even if it is just an additional check after you cleaned it manually. Maybe you have missed something.

            • HaraldvonBlauzahn@feddit.org
              link
              fedilink
              arrow-up
              2
              arrow-down
              1
              ·
              edit-2
              12 hours ago

              You are right that the distribution as it provides binary code is a trust root. If you can’t trust them, you have nothing to stand on.

              I had the impression that CachyOS suggests to use AUR packages - maybe I am wrong here?

              And if CachyOS is (what I am just assuming) geared towards less technical users, can you really expect their average user to examine shell scripts from a forum post?

              How do their users even know that the post and its author is legitimate? Are they supposed to check PGP keys?

              You can call that paranoid but there is a reason why distributions use packkage signing, publish webs of trust, and why the Guix developers even worked hard to reduce the binary bootstrapping code for the distro down to 512 bytes - it is a consequence of the “trusting trust” problem posed by Ken Thompson that the more stuff is opaque, the more trust is needed.

              • 5ymm3trY@discuss.tchncs.de
                link
                fedilink
                arrow-up
                2
                ·
                9 hours ago

                I have no idea about the stance of CachyOS on AUR packages.

                I totally agree with you, establishing trust is not an easy problem. I don’t expect the average joe to understand shell scripts. I would put myself in that categorie as well. This one however was simple enough that it seemed okay to me. If I don’t understand what’s going on in a script I am really careful and try to avoid it, if possible. I still wouldn’t consider them universally bad. For some things it is even the recommended install option. I vaguely remember some things in the Raspberry Pi universe ( IIRC this was even the case for Docker in the past).

                There are multiple factors which can lead to trust. Maybe you know the CachyOS forum and how well it is maintained. How old is the account etc… But as you said, there are always risks. The account could be compromized as well. But most of that isn’t specific to shell scripts or Linux in general. You shouldn’t install an application from some shady website in Windows either.

                What is your recommended way to deal with the current situation?

                • HaraldvonBlauzahn@feddit.org
                  link
                  fedilink
                  arrow-up
                  1
                  ·
                  edit-2
                  1 hour ago
                  What is your recommended way to deal with the current situation?
                  
                  • don’t tolerate malware. Get out the big hammer. This is an attack on Linux.
                  • be frugal on what you install
                  • slowing down. Not everything needs to be bleeding edge.
                  • perhaps use automated cooldown times of 2/4/8/12 weeks for changed packages, depending on software trustworthiness, and users experience. More cooldown for owner change.
                  • each PKGBUILD in an own repo, with a single owner
                  • removing all AUR recommendations from the Arch wiki
                  • gather the user community to help with review and testing of AUR packages, transforming the most important ones into extra packages
                  • score all packages by trustworthiness
                  • reward quality and sane practices when trusting / scoring packages - fire up a competition for quality
                  • build a web of trust, possibly with actually user-friendly software instead of GnuPG
                  • construct a robust voting/package reputation system for normal users (this is very hard because by experience, reputation systems can easily be gamed, it will by all experience NOT work purely electronically)
                  • ultimately trust in people, not personas or algorithms