Make unpredictableSeed use getrandom (syscall) on Linux

[0xEAB]

This patch changes unpredictableSeed to use the getrandom syscall on Linux.

Currently, unpredictableSeed calls arc4random() on applicable BSD systems;
for everything else it executes RDRAND on InlineAsm_X86_Any-compatible targets or falls back to a homebrew solution.

[dlang-bot]

Thanks for your pull request, @0xEAB!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub run digger -- build "master + phobos#10623"

[rikkimax]

That function needs a warning.

It is possible that the kernel may not have enough entropy stored to give a value.

You want to call it sparingly and only after the system has been booted fully.

[0xEAB]

and only after the system has been booted fully.

I don’t think this is accurate.
Quoting random(4):

When read during early boot time, /dev/urandom may return data prior to the entropy pool being initialized.

And also:

If this is of concern in your application, use getrandom(2) […] instead.

[rikkimax]

Yes, during booting it may be empty. Once initialized the chance of it to be empty depends upon if it has been misused.

In any case, a warning is needed ;)

[0xEAB]

@rikkimax
I’ve added further notes and a randomnes-quality warning. Please check whether you consider those sufficient.

@everyone
Please spell/grammar check those new paragraphs of technobabble.

[rikkimax]

@rikkimax
I’ve added further notes and a randomnes-quality warning. Please check whether you consider those sufficient.

You have gone above and beyond what I was wanting!

Good job.

[thewilsonator]

Use "(cryptographically secure) pseudo-random number generator" and introduce the acronym, and then use it throughout (rather than using the acronym first, and then spelling it out later).

[0xEAB]

Updated to:

This function utilizes the system cryptographically-secure pseudo-random
number generator (CSPRNG) or pseudo-random number generator (PRNG)
where available and implemented (currently arc4random on applicable BSD
systems or getrandom on Linux) to generate “high quality” pseudo-random
numbers – if possible.

[0xEAB]

You know what, I’ll take this chance that this hasn’t been merged yet to ask a question or propose a change here.

[0xEAB]

• Are we fine that this assert might not make it into user code because of -release?
• Should we rather do if (status != buffer.sizeof) { assert(false); }?
• Should we rather do if (status != buffer.sizeof) { return fallbackSeed(); }?

[rikkimax]

Don't worry about it.

You get what you get when you use it.

[4dacowgirl]

there is enforce for that. Is this optional by the way? so I can set it by a flag , 0, this, 1 getrandom, 2, something else kinda thing?

[0xEAB]

Calling enforce() would make this nothrow @nogc function, well, throw and @gc.

[0xEAB]

This function is just a way to generate a "good" seed for initializing random number engines – not a CSPRNG. (And if it were one, we definitely shouldn’t overengineer it. It could probably return an “optional”-type result then – but make it configurable – with options not applicable to all operating system –, rather nope. Also keep in mind, this thing has already been utilizing arc4random on applicable targets.)