I am not satisfied with Linux’s security and have been researching alternative open source OS for privacy and security So far only thing that’s ready to use is GrapheneOS (Based on Android) but that’s not available on desktop (Though when Android release Desktop mode it may become viable)

Qubes OS is wrapper around underlying operating systems, so it doesn’t really fix for example Linux’s security holes it just kinda sandbox/virtualize them

OpenBSD is more secure than Linux on a base level but lack mitigations and patches that are added to linux overtime and it’s security practices while good for it’s time is outdated now

RedoxOS (Written in Rust) got some nice ideas but sticks to same outdated practices and doesn’t break the wheel too much, and security doesn’t seems to be main focus of OS

Haiku and Serenity are outright worse than Linux, especially Haiku as it’s single user only

Serenity adopted Pledge and Unveil from OpenBSD but otherwise lacks basic security features

All new security paradigms seems to be happening in microkernels and these are the ones that caught my eyes

None of these are ready to be used as daily driver OS but in future (hopefully) it may change

Genode seems to be far ahead of game than everything else

Ironclad Written in ADA

Atmosphere And Mesosphere Open Source Re-implementation of Nintendo Switch’s Horizon OS, I didn’t expected this to be security-oriented but seems like Nintendo has done a very solid job

Then there are Managarm, HelenOS, Theseus but I couldn’t figure out how secure they are

Finally there is Kicksecure from creators of Whonix, Kicksecure is a linux distro that plans to fix Linux’s security problems

if you know of any other OS please share it here

  • Mike@sh.itjust.works
    link
    fedilink
    arrow-up
    6
    ·
    9 months ago

    Whew, there’s a lot to unpack here.

    First, microkernels being the future: This is a sentence that was said time and time again, but while microkernels definitely have some advantages in separating components which could yield better security, in practice it also introduces other security concerns, not present with monolithic kernels, mostly with the communication between the kernel services.

    Second, about the no secure Linux distros thing: As many others have mentioned, there are security-conscious Linux distros, mostly the “immutable” distros. You can use Fedore Silverblue (or even better, SecureBlue) as a daily driver, with Flatpak for your apps. That way, your main OS is read-only, thus harder to infect and all system updates are signed and verified. Using Flatpak helps enforce permissions on apps in a manner similar to Android permission (you can deny an app the right to see your files, for example).

    Third, I don’t really understand what you mean by “Linux’s security holes”. Of course it’s not bug free, but no kernel of this magnitude is. Also, GrapheneOS uses Linux as well, albeit with a hardening patchset, but you can also get that with desktop Linux distros. If you think Linux (being a monolithic kernel) is automatically less secure than microkernel and hybrid kernel based systems, take a look at Windows and macOS, which both use non-monolithic kernels, but most security experts will tell you that you’re better off using Linux.

    Fourth, about all the niche, mostly hobby OSes you listed: A big part of security is about having more eyes on the source code. Even if you write a kernel in a “safe” programming language, there will be bugs. Something as advanced as a kernel that’s ready for daily desktop use and provides advanced isolation between processes is going to be so complex that you won’t be able to see what bugs arised from the different parts interacting with each other. Safe programming languages make it easier to write safe code, but don’t stop you from messing up the logic that defines what apps have which permissions. Your best bet is to stick to software that has had time to mature and had more people and companies look through it. Linux is regularly audited by all tech giants, because all clouds use Linux to some extent. If it’s secure enough to isolate the workloads in Google Cloud, and Amazon’s AWS, it’s going to be secure enough for your desktop, provided you use it well (make use of it’s security features and don’t shoot yourself in the foot by disabling mitigations and the like). This is partly why I think the idea that OpenBSD is more secure than Linux is somewhat outdated. Yes, they advertise it as such, but it has seen much-much less auditing than Linux did in the cloud era.

    Of course, there’s nothing wrong with playing around with alternatives operating systems, just don’t think you’ll be more secure just because something is written in Rust, or is a microkernel. Those can help, but there’s much more to security than the guardrails a programming language or software architecture can provide, especially with something as complex as a modern kernel.