hi

Targeting illumos based distributions, what's the recommended init system? - OpenRC? - Something else?

smf

see illumos.org/man/7/smf

Thanks

how do you switch threads in mdb (on a process core)? i thought maybe <tid>::context but that doesn't seem to be it

I never have. I just use the dcmds to address the thread.

What are you trying to get at?

really just ::stack

<tid>::findstack?

Just like a kernel crash dump.

Just you don't need the ulwp_t address just the thread id.

aha!

And you can do the same ::walk thread | ::stacks, etc.

Let me know if there's other stuff that'd help you out.

Was there another way to fetch the commits hooks from gerrit? scp sjorge⊙cio:hooks/commit-msg .git/hooks doesn't seem to work

I was messing about and messed up my repo just now and delete it and started fresh, but don't have the local copy anymore

got it, rtfm, but the gerrit included one not ours, just a curl from code.illumos.org/tools/hooks/commit-msg

Hmm seems I did not mess up my repo and it's something else? I had a `error: remote unpack failed: error badDate: object 8eec3430a514cd7c4e568f6dbe40d81b2f8f2d27: bad date` when pushing the cstyle fixes. I though I messed up git but started fresh and just git apply-patch my changes (so it keeps the commit/change id) but I get the same

jbk: I feel like ::context is supposed to do it though

maybe

that seemed the closest thing, but maybe I wasn't giving it the right thing?

unfortunately its help doesn't really tell you what it's expecting

[illumos-gate] 15735 revamp nvme and disk FMRIs, property groups, and related -- Robert Mustacchi <rm⊙fo>

Note to other distro maintainers: OpenIndiana has problems to build the actual firefox version with binutils-2.41 while 2.40 work fine. So I will go back to 2.40 until this has been sorted out.

Hm, or not: There already seems to be a patch for it according to an email I just got: bugzilla.mozilla.org/1846703

[illumos-gate] 15711 snoop capture into rotating output files -- Toomas Soome <tsoome⊙mc>

[illumos-gate] 15832 loader: chain command is really expecting MBR/VBR boot code -- Toomas Soome <tsoome⊙mc>

rmustacc: if there's a scenario w/ mac where you have > 1 rx ring, but only 1 shared interrupt, is there any harm in having the ring interrupt enable/disable function just do nothing?

the man page suggest those methods are mandatory for rx rings

but in that scenario, it seems like enabling/disabling interrupts in the mc_start/mc_stop method instead

would make more sense

(this is probably a corner case, but wanted to at least try to be reasonable about it)

jinni illumos 15834

jinni illumos#15834

fenix illumos#15834

BUG 15834: Update Intel microcode to 20230808 (New)

↳ illumos.org/issues/15834 | code.illumos.org/c/illumos-gate/+/3001

(jinni lives on #smartos, my bad.)

jbk: You mean there's no separate interrupt cause register for the rx rings that's separate from the MSI-X?

If you have more than 1 rx ring and you don't have separate interrupt generation for that, then I don't think you should expose multiple rings based on my memory.

There is definitely a problem. We call the interrupt disable path when we come in to poll.

So even if you're going to ignore the interrupts on that ring, polling will happen.