Solar Designer

> I think supporting this (as described in OP) is a good idea but there would obviously need to be some conf setting for turning it off My suggested ways/conditions...

> This info is also available via sysfs (differently), which may be a more official way for programs to obtain it. ... and this is discussed further in the oss-security...

Optimal affinity settings may also need to consider (even) distribution across physical sockets, as seen in https://github.com/openwall/john/issues/5668#issuecomment-2790064384 - obviously, relevant only when using fewer than max hw threads.

As an experiment, I've just briefly tried changing `p_is_ed_task` to `return 1;` in all cases (on top of #391). I saw no issues. However, reverting #387 (that is otherwise unneeded...

If we continue to use conditional `p_is_ed_task` (not always true), then we'd likely want to review where we use vs. don't use it, and explain those choices (or a global...

In short: > PREEMPT_RT Currently unsupported, LKRG will refuse to build. > development kernel (RCs) They generally just work, but sometimes recent/development kernels (not only RC) require us to make...

Yes, I get your point, which is why I kept this issue open. I'm not sure what we want to actually do about it. For recent kernels requiring git LKRG,...

> Alternatively, until we've completed this implementation, we should temporarily `#if 0` the above pieces of code (and maybe the corresponding struct fields) to remove the (currently unjustified) performance overhead....

We should also add tracking and validation of supplementary groups.

We have native yescrypt code for Go https://pkg.go.dev/github.com/openwall/yescrypt-go https://github.com/openwall/yescrypt-go but in other languages you'd currently access yescrypt by calling the system-provided `crypt(3)` function (or its corresponding thread-safe alternatives), as long...