SunOS process and thread support

[gwr]

Read binary psinfo for System.Diagnostic.Process on SunOS (Solaris or illumos).
No failures in System.Diagnostic.Process.Tests (but lots of skip)

BTW, I tried rebasing on main from Mon. this week and ran into problems downloading stuff.
Not sure why, but it didn't seem to have anything to do with my changes.

[am11]

Lets drop the changes which are already submitted in other PRs.

[am11]

Good start! I've left initial feedback, which I anticipate the maintainers will point out. :)

[gwr]

Good start! I've left initial feedback, which I anticipate the maintainers will point out. :)

Thanks. I had not yet seen #105207 when I opened this. I think @AustinWise and I should figure out how to get the prerequisite fixes shown there (and here) all integrated, and then I'll rebase the last parts of this onto that.

[gwr]

Lets drop the changes which are already submitted in other PRs.

I'd love to, but I do not see a way to do "stacked" PRs, where one PR targets the branch of another PR.
If I drop the prerequisite changes then things break. Suggestions welcome.

[am11]

Create a local branch feature/illumos-port with all the miscellaneous changes. Then keep this branch only for System.Diagnostics.Port changes. You can git cherry-pick <spaces-separated commit hashes> across the branches.

[jeffhandley]

@gwr It looks like there are just a couple of comments still to resolve from Jan's review back in September. Can you address these comments please?

1. SunOS process and thread support #105403 (comment)
2. SunOS process and thread support #105403 (comment)

[gwr]

@gwr It looks like there are just a couple of comments still to resolve from Jan's review back in September. Can you address these comments please?

1. SunOS process and thread support #105403 (comment)
2. SunOS process and thread support #105403 (comment)

Yes, sorry, I've been busy with some other things. Will do.

[gwr]

This code didn't build for me anymore without updating to a later base.
I've rebased it onto main as of Dec 31 2025. (and squashed)

I also still need this Workaround #116929 but I've left it out of this PR.

[gwr]

All the new SunOS Interop.* functions now return void instead of bool.
There's still work to do in some consumers to adjust to that...
(I'm working on it.)

Hm. Changing all those Interop. functions from bool to void return means: (1) this code diverges from the Linux code we based it on, and (2) consumer call sites need to deal with exceptions for normal errors (like no such process) in addition to the usual exceptions for unexpected problems (bugs).
Maybe it would be easier to do the exception handling one level deeper in the call stack (in the SunOS Interop. functions) and let these return bool like Linux (ret. false if exception caught).
Thoughts on the design questions? @jkotas

[jkotas]

I have commented on method signature since it does not make sense to have internal bool-returning method that always returns true (without further explanation). If I am reading the code correctly, the equivalent Linux methods return false when the PID no longer exists.

If it helps with code sharing, it is fine to return bool to match Linux. The behavior should either match Linux (ie return false in situations where Linux return false), or there should be a comment why the behavior does not match and why it is fine.

[gwr]

Thanks.

The interop functions were coded to throw exceptions only becase some code I copied from FreeBSD was written that way. There's no need for that in this code;
it can just return false when there's an error instead of doing a "throw".
This seems to work.

[jkotas]

Thank you!

@am11 Could you please review as well?

[am11]

LGTM! 👍

[gwr]

Thanks for all the help! 🙏
@jkotas@am11@AustinWise