# On the insecurity of Linux-based operating systems **URL:** https://discuss.privacyguides.net/t/on-the-insecurity-of-linux-based-operating-systems/26093 **Category:** Off Topic **Created:** 2025-03-24T16:09:17Z **Posts:** 52 ## Post 1 by @Sam1 — 2025-03-24T16:09:17Z Continuing the discussion from [PebbleOS smartwatches have opened pre-orders](https://discuss.privacyguides.net/t/pebbleos-smartwatches-have-opened-pre-orders/25907/22): > [@PebbleOS smartwatches have opened pre-orders](https://discuss.privacyguides.net/t/pebbleos-smartwatches-have-opened-pre-orders/25907/21): > > Question. What makes Linux “famously insecure”? This is probably a good read about Linux not being secure: [Linux | Madaidan's Insecurities](https://madaidans-insecurities.github.io/linux.html) A (very) short summary of a few points raised on that page would be: - Linux desktop OSes don’t have proper sandboxing (yes, flatpak is **something** , but it doesn’t have a robust permission model like others do). - X11 provides zero GUI isolation, so every app can look into what every other app is doing. - Linux (the kernel) is monolithic, meaning all modules and drivers run with the exact same permissions as the kernel itself, whereas MacOS and Windows use hybrid kernels, which segregate some processes to userspace. - Several exploit mitigations found in MacOS/Windows such as Control Flow Integrity, Arbitrary Code Guard, and Code Integrity Guard are scarcely used, not enabled by default, or completely nonexistent. > [@PebbleOS smartwatches have opened pre-orders](https://discuss.privacyguides.net/t/pebbleos-smartwatches-have-opened-pre-orders/25907/21): > > For me, this risk is way bigger than any risks I might run by daily driving Linux. Imo. A valid assertion, you are free to choose how you handle your threat model. --- ## Post 2 by @anon80779245 — 2025-03-24T16:28:37Z I dislike those at-charge questions. When you look at something, you should look at it in an holistic manner, looking at least at the platform strength and weaknesses. --- ## Post 3 by @anon21666177 — 2025-03-24T16:32:14Z His review is holistic, albeit somewhat dated since we’re quickly transitioning from X11 to Wayland. --- ## Post 4 by @anon80779245 — 2025-03-24T16:38:03Z True, the article is fine but a thread asking all the bad things about Linux isn’t great I was just pointing that out. I think the biggest security benefits of Linux compared to Windows is that you actually know what you give admin permission to. On Windows you have to click "Ok’ on the popup about wanting to use admin privileges and it’s not even specified what it is for. Since trivial tasks require it, you become used to it and it becomes essentially useless. --- ## Post 5 by @Sam1 — 2025-03-24T16:46:18Z > [@anon80779245](#): > > On Windows you have to click "Ok’ on the popup about wanting to use admin privileges and it’s not even specified what it is for. This is definitely a problem. That prompt is immensely vague. --- ## Post 6 by @anon63378630 — 2025-03-24T16:46:39Z > [@anon21666177](#): > > quickly transitioning Wayland has been stable for like 8+ years now imo: [GNOME 3.22 Release Notes](https://help.gnome.org/misc/release-notes/3.22/) --- ## Post 7 by @anon21666177 — 2025-03-24T17:15:04Z I agree the transition could’ve happened earlier but now it’s not just GNOME using Wayland. KDE and others are making the transition too. --- ## Post 8 by @ignoramous — 2025-03-24T17:59:48Z > [@Sam1](#): > > (very) short summary of a few points raised on that page would be Most have since been addressed. AOSP, in particular, seems to have covered all the points you list in your summary. Indications are, Android will be on desktops soon. Linux is fairly modular, and Android’s (_binderized_) HAL (_hardware abstraction layer_) already demonstrated that a “hybrid” architecture is not a limitation imposed by Linux, but by the ecosystem that preceded it. Apart from the drivers, the major source of concern remains the Kernel & its subsystems, which remain heavily configurable & are predominantly written in a memory unsafe language. _Fuschia_ (also by Google) offers a way out & may yet take hold in Android (though, chances look grim after a decade of getting no where except on _Nest_ devices). --- ## Post 9 by @Sam1 — 2025-03-24T18:24:55Z Being fair, I don’t think anyone (including Madaidan) has Android in mind when they say “linux-based operating system” (or any colloquial term referring to them). But yeah, AOSP definitely addresses several security concerns! --- ## Post 10 by @anon63117981 — 2025-03-24T18:56:14Z He has [another post about Android](https://madaidans-insecurities.github.io/android.html). So you’re right, he doesn’t mean Android when he writes about Linux. --- ## Post 19 by @anonfox — 2025-03-25T10:30:24Z Linux anecdotally provides better sandboxing than Windows in my experience. I’m not sure about the statistics, but it definitely feels like there are more Flatpaks proportionally than AppContainers, perhaps because UWP is in no longer in active development, or maybe because Flatpaks have a lot of other benefits (e.g. running on most distros). The point usually raised about how Flatpaks have invasive permissions is invalid imo because you can very easily change them through CLI (flatpak override) and GUI (Flatseal). For when they’re not suitable, bubblejail exists. --- ## Post 23 by @laugh.oysters — 2025-03-25T13:12:11Z it’s all relative. In android and iOS, the need for strong sandboxing is vital because the users are installing shitty user-hostile apps next to their banking apps by default. In android’s case, google wishes to protect its own access to your data as well as stop bad apps from going to town on your device. But In the linux context, you might have only one bad app - probably the browser since it is running anything you put through it. Also, there is no google trying to lock your data in for its own purposes. If your other apps are just open source types (gimp, vlc, libre office) installed from a debian repo or whatever, then I’m not sure sandboxing is quite as critical. There are also some sand boxing methods in linux, and if you wanted to go far with it, you could spin up VMs or containers à la qubes OS. The reason linux is secure on servers, is that it only runs the minimum of what it has to. ‘Apps’ aren’t the issue - more likely protecting access to RAM from processes and segregating network access. So if your linux box is basically a browser and a few open source apps, then who really cares about in-depth sandboxing? It’s not like you will be installing spyware that easily on linux anyway and your attack surface is probably very much reduced in comparison to a similar windows machine. --- ## Post 24 by @anonymous306 — 2025-03-25T13:14:27Z This argument is delusional. Linux is not secure. To elaborate: Open source has no relation to security. Grub, a core component of most distros, just had _73_ security issues posted last month except they didn’t even bother to make a new release forcing distros to each manually pull the fixes in. To [quote](https://gitlab.archlinux.org/archlinux/packaging/packages/grub/-/commit/2b77c1cab239c7ac3999d3f06f62bca0b4e24708) the Arch maintainer: > Countless vulnerabilities, but no release management, no maintenance > branch, or whatever. :woozy_face::face_with_peeking_eye: Grub maintainers aren’t even [bothering](https://lists.gnu.org/archive/html/grub-devel/2025-02/msg00024.html) to blacklist the binaries via the distributed secureboot dbx blocklist. > This time UEFI revocation list (dbx) will not be used and revocation of broken > artifacts will be done with SBAT only. And many distros take many days to ship browser updates. --- ## Post 25 by @laugh.oysters — 2025-03-25T13:17:54Z You’ve presented a flawless argument, and left me me… dare I say… bamboozled! I subsequently refute all my previous statements --- ## Post 27 by @laugh.oysters — 2025-03-25T13:23:05Z bro, please let me know when you’re done editing so I can explain to you how you’re wrong. It doesn’t matter what you have to say, I will find a way. --- ## Post 28 by @anonymous306 — 2025-03-25T13:28:52Z > [@laugh.oysters](#): > > I will find a way Keep digging your head into the sand, instead of accepting a long known truth. --- ## Post 29 by @laugh.oysters — 2025-03-25T13:41:25Z Ok tx bro, So for a start, android is linux and so is ChromeOS. Both incredibly highly secure linux distributions, with sandboxing and locked bootloaders. That alone shits on all your possible arguments. Now to be a little more in good faith, you also have less known variants - that run in RAM like TC, amnesic like TAILS or atomics like fedora silver blue… the list goes on and on. Each with their own protections and security that can go above and beyond whatever you have in mind as an alternative. Sure, maybe your average pedestrian linux distro like ubuntu is not the most secure option. But it is still one of the most private. And if you want a linux that is more secure, it is pretty simple to find yourself a secure distro that mitigates whatever brain worm problems with grub or otherwise that has you fixated on. Security on linux is a skill issue. It can be as secure as you want it to be. It could literally be designed to self-destruct if you want to go full psycho with it. Now if you are upset that it’s not all magically being done for you in the timely manner you want, then you’ll just have to accept whatever slop is given to you. But then you don’t really have a leg to stand on as far as complaining about it. --- ## Post 30 by @anonymous306 — 2025-03-25T13:44:49Z > [@laugh.oysters](#): > > android is linux and so is ChromeOS If you’ve been paying attention to the topic it is about traditional distros, not Android or ChromeOS. > [@laugh.oysters](#): > > that run in RAM like Because running in RAM provides security benefits? Since when? > [@laugh.oysters](#): > > own protections and security Thinking the outdated poorly patched base that is Debian of Tails provide security :rofl: > [@laugh.oysters](#): > > secure distro Please name one that is actively maintained. All the ones I know are dead. > [@laugh.oysters](#): > > skill issue Ah yes, let me go rewrite all of Linux in a secure language with my skills. ![224735](//forum-uploads.privacyguidesusercontent.com/original/2X/a/abd8ef1548e6177fe5f1cc0bd7829a7dd2313d64.jpeg) > [@laugh.oysters](#): > > if you are upset I’m not upset at Linux and co. I’m upset at users, like you, wrongly proclaiming Linux as secure when it isn’t. I’m not going to fruitlessly continue to argue with you. Have a good day. --- ## Post 31 by @laugh.oysters — 2025-03-25T14:01:59Z Actually, there is nothing in the OP that says this is about specifically consumer and hobbiest distros. The title is LINUX-BASED OPERATING SYSTEMS. I put it in caps so you can read it more easily. Regardless, the existence of android and chromeos prove that linux can be secure on the desktop. Of course a hackable and open OS such as ubuntu is more accessible. Of course, they’re not rushing to patch grub. I’m confused if you are being purposefully single-minded about that. If something is designed with an open model. It is going to be more open!? The main point I am making, is that in most average people’s usecase, that openess doesn’t impact negatively on their threat model. And for those that have the aspirations like you, then it can be made secure with some skill. Now btw… grub is actually not at all necessary for booting linux. If you really cared about security you might want to use an encrypted bootloader on a separate usb device anyway… if you were really worried, you might even want to hide it inside your bottom. just saying yo’ --- ## Post 32 by @anonymous306 — 2025-03-25T14:05:04Z > [@laugh.oysters](#): > > rushing to patch grub. Another unsubstantiated claim, they already did: See 2.12-5ubuntu8: [https://changelogs.ubuntu.com/changelogs/pool/main/g/grub2/grub2\_2.12-5ubuntu10/changelog](https://changelogs.ubuntu.com/changelogs/pool/main/g/grub2/grub2_2.12-5ubuntu10/changelog) --- ## Post 33 by @laugh.oysters — 2025-03-25T14:14:09Z jesus, would you kindly get lost with unnecessary nitpicking eh? I’ve made some pretty simple and generalized arguments in good faith and you haven’t really addressed any of those beyond bombarding specific examples that refer only to very specific distros and configurations that suit your stance. You’ve completely ignorred the highly secure linux-based projects like graphene-OS or chromeOS… And posting yet another unclickable link to some frankly uninteresting development on ubuntu’s stance or status on the grub fiasco is just missing the point further. --- ## Post 34 by @Sam1 — 2025-03-25T14:24:06Z > [@laugh.oysters](#): > > nothing in the OP that says this is about specifically consumer and hobbiest distros Given that this is a forum for a website that recommends software (among other things) to everyday people without a particular exclusion to those that may not have a lot of technical skill, I didn’t specify consumer distros because I thought the context was obvious. That’s my bad. > [@laugh.oysters](#): > > Security on linux is a skill issue. It **shouldn’t** be, though. That’s the whole point of recommending software with good defaults. The average person shouldn’t have to be technically skilled for them to get decent protection when doing their computing tasks. * * * As a side note, you’re strangely defensive about your position. I suggest you tone it down. Politeness can go a long way, and we’re all here because we care about things that are in alignment, I hope. --- ## Post 35 by @laugh.oysters — 2025-03-25T14:29:38Z > [@anonymous306](#): > > Because running in RAM provides security benefits? Since when? Now because I’m not only just having fun. Here is an actual answer to your only actual question. An OS that runs in RAM is inherently read-only unless there is some persistence expressly configured elsewhere. It doesn’t even need a hard disk installed. If you couple that with an encrypted and physically separate boot disk… Well good luck picking up any kind of persistent malware as you peruse the grub and cvc mailing lists! But that’s not it. It also provides protection from driveby evil maid attacks. No hard disk means no vector for controller malware or just regular malware on your system disk. A reboot, similar to an atomic distribution returns the OS to its original state. --- ## Post 36 by @anonymous306 — 2025-03-25T14:33:00Z > [@laugh.oysters](#): > > good luck picking up any kind of persistent malware You’re focused on persistence while entirely overlooking the fact that any data you processed/accessed can still be siphoned off by the malware or that the malware can attack adjacent devices on your network. > [@laugh.oysters](#): > > similar to an atomic distribution returns the OS to its original state. Current atomic distros provide zero security benefit as they all lack verified boot. --- ## Post 37 by @laugh.oysters — 2025-03-25T14:44:06Z Yeah this is the point the other guy should be making. The main issure wrt security is that linux isn’t a centralized OS like android and the fact that it is open and hackable lends it to these kinds of unwanted developments. Also, trickle down solutions that take a while to dissipate. If you’ve ever waited for new kernel developments to trickle down to your LTS using distro, you would understand exactly what I mean. We cannot have the blazing fast distribution of patches across all the distros when each variety has a different purpose and a different structure. But that’s also the point of linux. There are different flavors for different tastes. You make of it what you want. If you cannot make your own, you make do with what has been made available. The constant compromise is always security+privacy vs convenience. There are clearly secure linux options, as well as alternatives such as qubes OS. They maybe aren’t super convenient, or private though. But this is the same compromise we make elsewhere. Like using your degoogled android that has no play store, banking app, google pay etc. So to say that linux as a general os that someone might install on their grandma’s PC isn’t secure is disingenuous. Grandma has actually been using wayland without knowing it for ages, her only app is a browser, and everytime she installs driveby malware, it doesn’t even know what to make of her system, let alone worry about sandboxing. Honestly, I think overbearing centralization is antithetical what linux is. --- ## Post 38 by @anonymous306 — 2025-03-25T14:49:33Z > [@laugh.oysters](#): > > and everytime she installs driveby malware, it doesn’t even know what to make of her system Ah yes, the fallacy that Linux is some niche OS that has no malware targetting it. If I had a bingo card for your responses I think it’d be done for already. --- ## Post 39 by @laugh.oysters — 2025-03-25T14:52:00Z I’m not familiar with verified boot on secure distros, but that is surprising as I was already using this feature in fedora over a decade ago. Same goes for wayland btw. And full disk encryption. I’m not trying to provide a perfect solution. You were asking for one that is premade and widely distributed and I was just trying to convey that being absolutist on this wasn’t going to work in the context of a private and open OS. If you are going as far as running your OS in RAM and have hidden your secure boot disk as far up your butt as you can, then I’m sure you can go to the trouble of installing your browser in a VM. --- ## Post 40 by @anonymous306 — 2025-03-25T14:53:53Z > [@laugh.oysters](#): > > verified boot on secure distros, but that is surprising as I was already using this feature in fedora over a decade ago. Fedora Atomic Workstation aka Silverblue is not a decade old. Nor does any Linux distro offer verified boot. Please do not confuse verified boot with the joke that is (EFI) secure boot. --- ## Post 41 by @laugh.oysters — 2025-03-25T14:54:14Z It is a niche OS. It has 1-3% market share if you discount chromeos and android, as you are happily already doing. That is not a fallacy, it is a fact. --- ## Post 42 by @laugh.oysters — 2025-03-25T14:55:24Z I havent used the fedora atomics yet… when i was using fedora a decade ago, i was using verified boot already… so i am confused that it would not be available now in the atomic versions [Edit].. oh I see you discount EFI secure boot as a verified boot. Would that also fail your inspection if you had generated your own keys? Aren’t all these problems solved by using something like a core-booted laptop anyway? I genuinely don’t know since the last couple of laptops were chromebooks anyway since I don’t have the energy or time for home made solutions anymore. --- ## Post 43 by @laugh.oysters — 2025-03-25T15:07:15Z > [@Sam1](#): > > As a side note, you’re strangely defensive about your position. I suggest you tone it down. Politeness can go a long way, and we’re all here because we care about things that are in alignment, I hope. When I made my first comment in this thread, which was just neutral and polite. Your man there, replied with (before editing a bunch of times): “This argument is delusional. Linux is not secure.” That did not seem very polite to me, nor helpful. Since then their arguments seemed to me, mostly provided in bad faith - arguing details against generalized points, and giving general statements as facts as if their word is law. Not taking the time to meet me any length of the way… And so I am merely responding in kind. --- ## Post 44 by @Sam1 — 2025-03-25T15:12:03Z > [@laugh.oysters](#): > > and everytime she installs driveby malware, it doesn’t even know what to make of her system This is security through obscurity. It may work for now, but it won’t the moment a malicious site has `if (os == "linux") { ... }` for their drive-by malware distribution mechanism. > [@laugh.oysters](#): > > I think overbearing centralization is antithetical what Linux is. This is a valid belief, but it’s also very dependent on what Linux as a kernel/OS family/community means to you (and each of us) personally. I am, of course, not advocating for every distro to be tightly locked down and have them all be just slightly different variations of “FOSS MacOS”. However, people should be aware of the tradeoffs they’re accepting when using Linux as their OS, and that is that most distros aren’t set up to be secure by default. Linux (the kernel) may have many security features already set up in the code, but that doesn’t mean much if your distro doesn’t enable it out of the box. If you argue that the risk of anything actually infecting grandma’s computer on Mint (as an example) is lower than the risk of Microsoft’s or Apple’s malicious treatment of users affecting her, that is a very fair assessment. But you’re making it with the knowledge of the compromise it entails. Others should be able to, as well. > [@laugh.oysters](#): > > When I made my first comment in this thread, which was just neutral and polite. Your man there, replied with (before editing a bunch of times): > > “This argument is delusional. > Linux is not secure.” > > That did not seem very polite to me This is fair. @anonymous306, please be respectful when disagreeing with someone in a public forum. Otherwise it all devolves into name-calling and such. > [@laugh.oysters](#): > > And so I am merely responding in kind. This is not great either. --- ## Post 45 by @anonymous306 — 2025-03-25T15:19:43Z > [@laugh.oysters](#): > > you discount EFI secure boot as a verified boot I don’t discount it, the industry does because they are absolutely not the same thing. > [@laugh.oysters](#): > > Aren’t all these problems solved by using something like a core-booted laptop anyway This has absolutely nothing to do with verfied boot, please stop. --- ## Post 46 by @laugh.oysters — 2025-03-25T15:22:05Z > [@Sam1](#): > > This is security through obscurity. It may work for now, but it won’t the moment a malicious site has `if (os == "linux") { ... }` for their drive-by malware distribution mechanism. Security through obscurity is still security if not perfect… but in this context I see it more as being a numbers game, it’s better when you’re grandma isn’t not part of the low-hanging fruit. Because out of 100 given fictional grannies, let’s be honest, the one and only that’s rocking linux mint is way less likely to be harboring some OS-level malware than the others. --- ## Post 47 by @Sam1 — 2025-03-25T15:27:10Z If that is the trade-off you are willing to make, go ahead, but I would think it irresponsible to call it a viable security measure when the workaround an attacker has to make is trivial. --- ## Post 48 by @laugh.oysters — 2025-03-25T15:27:57Z > [@anonymous306](#): > > This has absolutely nothing to do with verfied boot, please stop. It was a question as I had just admitted to not knowing. Would you care to explain how a core-booted laptop is not up to measure, or whether self generated keys on an EFI secure booted laptop is still not enough to keep grandma safe?.. or are you just going to continue being unhelpfully vile? --- ## Post 49 by @laugh.oysters — 2025-03-25T15:34:33Z I mean, in the context of an every day person, doing every day things. Then yes I think we can say that your average linux distro is secure enough. I think that for the type of utilization - open source style apps installed from the main repo - inherently brings less risks than installing shitty android apps, making the lack of standardized containerization in linux less problematic than other OS. There are fair points being made about the disadvantage of opensource models for finding, implementing counter measures and patching security issues. But it can go both ways, where the centralized model is slower to respond. Just look at router firmware for example. --- ## Post 50 by @anonymous306 — 2025-03-25T15:51:42Z @laugh.oysters I let my emotions get to me and as a result my above posts were rude and inappropriate, for that I sincerely apologize. --- ## Post 51 by @jonah — 2025-03-25T15:52:11Z --- ## Post 52 by @jonah — 2025-03-25T15:59:27Z Thank you for the apology @anonymous306. I will also give @laugh.oysters the benefit of the doubt and thank you in advance for accepting it since I locked this thread before you could respond. > [@laugh.oysters](#): > > Not taking the time to meet me any length of the way… And so I am merely responding in kind. This is the reason I am locking this post, for everyone’s information. I don’t think this discussion is going anywhere. This entire conversation could have gone: 1. “Linux can be secure” ([On the insecurity of Linux-based operating systems - #23 by laugh.oysters](https://discuss.privacyguides.net/t/on-the-insecurity-of-linux-based-operating-systems/26093/23)) 2. “Linux isn’t secure, FOSS maintainers do not always follow good security practices” ([On the insecurity of Linux-based operating systems - #24 by anonymous306](https://discuss.privacyguides.net/t/on-the-insecurity-of-linux-based-operating-systems/26093/24)) 3. “There are plenty of Linux-based systems like Android which are very secure” ([On the insecurity of Linux-based operating systems - #29 by laugh.oysters](https://discuss.privacyguides.net/t/on-the-insecurity-of-linux-based-operating-systems/26093/29)) 4. “I was only referring to traditional distros” ([On the insecurity of Linux-based operating systems - #30 by anonymous306](https://discuss.privacyguides.net/t/on-the-insecurity-of-linux-based-operating-systems/26093/30)) 5. “Okay” :handshake: The rest of this post to my eyes is mostly bickering back and forth with neither of you actually having a conversation about the other’s arguments. I’m also locking this post because nobody besides you two and the OP are really chiming in with any information, so I assume there’s not much to add. Correct me in a DM if I’m wrong, anyone. > [@laugh.oysters](#): > > giving general statements as facts as if their word is law. I _do_ agree with this issue, although it is not entirely one-sided in this conversation.