this feels like a weird statement, but stay with me: I think aok/aask were at their most useful before boot environments

something I just thought of though, you also can't stub assfail and friends in the debugger as you normally would, if they're truly noreturn and the compiler knows it. if you take away aok, you also take away the means to simulate aok

richlowe: I think a lot of stuff like that was more useful before BEs, or before VMs for testing, etc

It's pretty unusual that I boot totally unhinged WIP bits on the machine I use to edit and build things these days

And on the machines that we use for real work (like device control) temporary BE activation makes me a lot more comfortable with trying things out there

I'm totally OK with not being able to stub out assfail, tbh

If I need to doctor the program text I'll just do it with the conditional that lead there

I think the ergonomics of VERIFY() and panic() and so on are just generally better when the compiler can be sure we're not coming back

like if you VERIFY(1) I feel like gcc would be able to tell that we're definitely not reaching code after that

VERIFY(0) rather I guess sigh

it does, if noreturn, I think tsoome has fixed things based on this.

`if (constant) noreturn();` certainly didn't confuse anything

yeah

[illumos-gate] 16426 Want libnvme support for OCP log pages -- Robert Mustacchi <rm⊙fo>

[illumos-gate] 16427 want libnvme support for supported command and log page logs -- Robert Mustacchi <rm⊙fo>

Hi! I'm considering porting the Illumos kernel over to Itanium, and wanted to see the community's opinion on that. :)

ferrad: ok, I've got to ask, why port to a literally dead platform

Because Itanium is still used, even if not produced. It's also a very fascinating architecture to me :)

I also recently got an Itanium server unit, and have been planning to mess around with my own operating system kernel, as well as other kernels.

I think if it's something you'd find interesting and are doing it for yourself, then enjoy it. I think given the challenges we've head with SPARC, I think it's unlikely we'd probably want to maintain it in tree.

Oh thats absolutely fair, I do plan to have the source code available obviously, but I do not currently plan on trying to get it on the mainline kernel. I just like to use things like this to gauge the community and how comfortable I might feel sharing these projects with them :)

There are other porting efforts related to ARM and related that are ongoing, so if there's stuff that'd improve the system generally, I expect we'd be receptive.

But I also don't expect there to be many folks who can answer questions on Itanium related bits in here.

Yes, that's to be expected. Luckily I know someone who worked on Itanium 2 at intel, and they have been incredibly receptive to my questions. I've even been working on my own Itanium implementation using verilog with their help, although that is quite a ways off.

Sounds good. Good luck.