Linux 4.18 Lands Random Patch To Fix Slow Boot Times For Some Systemd-Based Boxes

[duby229]

Shower thought: If systemd is part of an intelligence op designed to get more control over and weaken the security of GNU/Linux systems (a popular refuge from the already compromised Windows operating systems), then what we might expect to see over time is:

1. Little or no effort made to standardize systemd APIs, or constantly evolving API standards that are hard to keep up with. This would make it hard to make competing compatible implementations.
2. Increasing the scope of systemd in order to make removal / replacement / opt-out of it harder and harder. This would make it more likely that targets (people who are trying to escape surveillance) will be using machines with known weaknesses (known to the intelligence groups).
3. Use of dominant market position to implement changes without much consideration of other projects attempting to remain compatible or offering alternatives.
4. Increased Integration of systemd with the kernel and the kernel with systemd, such that eventually 'reasons' can be presented for moving kernel functions into user space (into systemd).
5. Increasing complexity of systemd over time so that security sensitive bugs are more likely to naturally occur and so deliberately created bugs can more plausibly be blamed on being an oversight when changing a complex system.
6. Use of the usual tactics: e.g. suggest unsafe defaults for 'reasons' (but really because u know the defaults will likely get used making users vulnerable).

I don't know if this is has happened, is happening or will happen.

For those of you who are aware of the kind of shady stuff going on constantly, it's something to keep an eye out for.

Pretty sure some narrow, polarised minds will be triggered by my post. Go nuts!

I think your points 1, 2, 3, and 4 are definitely happening. But I don't think it's security related issue, it's actually an anti-competition issue. They believe that if systemd was the only option the world would be a better place or something like that. And so they have constantly taken the harshest anti-competitve attitudes that they can. They design their projects to make them incredibly difficult and in many cases impossible to fork and port out of systemd. LP himself has stated repeatedly that it is his goal to make certain udev can only work within systemd itself. And when he couldn't make that happen he stated he wanted to make sure that no future project could function outside of systemd.

EDIT: I mean really, just look at it and tell me how many intermediates there are.... It's a horribly tangled web and on purpose too, for sole -stated- reason to make it difficult or impossible to port.

[shmerl]

So will this replace the need for rng-tools?

[rtfazeberdee]

Seems like a few posters here subscribe fully to conspiracy theories

[Zan Lynx]

again, another work around to systemd, what a beautiful software /s

Do you want real random numbers or not?

In some cases it has been so bad that consumer IoT devices would boot up and on first boot there was such little entropy in the random pool they used that every device of that type would generate one of about 256 keys. Wow. Much security. Many bits!

[RealNC]

Will this be backported to LTS kernels?

[pcxmac]

This is non sense! Offer up one piece of software that is an example of perfection. If the kernel was perfect developers would not be working on it now. Same thing goes for user land apps all of which get bug fixes, feature updates or general improvements from time to time.

You may not like SystemD but you don't have to use it. Just like you don't have to use Python, Rust, Vim, sed, (the file system of your choice) or any other of the numerous options that can go into a Linux distro.

I mentioned the file system of your choice because tgere are so many to choose from. As such a distro or even a single installation can be tailored to a users needs. That is what Linux is all about, why SystemD gets such abuse is beyond me. Its high adoption rate indicates to me more good than bad.

If you don't care about what someone says, don't bother making a straw man argument(s). SystemD is Red Hat's attempt to 'control' the 'linux userspace'. You can deny that, but you can't make a credible argument against it. How long till they enrol their own package management facility. It's only consistent they do given how many aspects of the 'Operating System' they are attempting to consume with this 'proprietary' cancer.

[cybertraveler]

If you don't care about what someone says, don't bother making a straw man argument(s). SystemD is Red Hat's attempt to 'control' the 'linux userspace'. You can deny that, but you can't make a credible argument against it. How long till they enrol their own package management facility. It's only consistent they do given how many aspects of the 'Operating System' they are attempting to consume with this 'proprietary' cancer.

I'm pretty sure that in the context of software, "proprietary" simply means, not Free Software and not Open Source software. So systemd isn't proprietary software. It's many things, but not that

[rtfazeberdee]

If you don't care about what someone says, don't bother making a straw man argument(s). SystemD is Red Hat's attempt to 'control' the 'linux userspace'. You can deny that, but you can't make a credible argument against it. How long till they enrol their own package management facility. It's only consistent they do given how many aspects of the 'Operating System' they are attempting to consume with this 'proprietary' cancer.

ROFL -> en.wikipedia.org/wiki/Conspiracy_theory

[Guest]

If you don't care about what someone says, don't bother making a straw man argument(s). SystemD is Red Hat's attempt to 'control' the 'linux userspace'.

That's not necessarily a bad thing. 238.133 was a rock solid release. 239 is not good enough for my taste. But even as a side effect, systemd still is much better than the old arch linux initscripts for example.

Proprietary is not necessarily a bad thing either. What's bad is low quality software design, lack of proper maintenance, and lack of testing coverage. Windows would still be a much worse choice than Linux for custom operating systems such as Android even if they open sourced the Windows kernel.

[Weasel]

This actually doesn't have anything to do with systemd really. One question nobody seems to be asking though is whether RDRAND as a source of early entropy is good enough given that many people suspect Intel colluding with three letter agencies in USA. But if I put my tinfoil hat aside for a moment, this seems like a decent change.

I'm not sure what part of any user can write to /dev/urandom do you not understand? You don't have to be root to do it. Try it yourself.

Writing to /dev/urandom mixes in new seeds. It's world-writeable because this mixing process cannot ever do harm, but it can potentially do good, so why not?

Even if RDRAND returned zero every single time (or other predictable pattern) it would still not be worse than without using it.