richlowe: I see there is a pgp key at illumos.org/docs/security, I should use it to sign my message, right?

all the logs are publicly available on the internet, though - log.omnios.org/illumos/2026-09-26

hope I sent it in a correct way

gemelen: thanks for trying that! :)

wiedi: np. it was an interesting exercise too

hi, i kind of need assistance. i've patched the I226 driver and MAC in the illumos kernel for my own appliances for parallel receive paths, but i've never committed anything to illumos before. how should i approach trying to upstream it?

illumos.org/docs/contributing has the basic docs here.

But basically, I would focus on getting the change up on gerrit (ttps://www.illumos.org/docs/contributing/gerrit/) and then we'll review. Feel free to tag me in it as I had done initial work on the driver.

I wouldn't expect that adding multiple receive rings would require changing the mac kernel module though.

so what's the plan. should i break it down into multiple smaller PRs?

I would just expect one change that has the totality of the work so we can review and integrate it coherently.

i could see the MAC change land fairly isolated, if i want to get something in in preparation without breaking too much

Why is MAC changing?

It may be reasonable, I just wouldn't expect it for adding RSS support to igc.

okay how do i describe that. i hit a bug that seems to affect multiple drivers (found some chelsio one on the bug tracker) as well as my x553 and i226, essentially, a network card's receive ring in polling mode with interrupts turned off after its queue empties

Ah, a bug in MAC versus new features to implement RSS.

yeah, but i'm not pushing a driver update, if it triggers a bug somewhere

Okay, I would probably structure that separately then.

And do one change with the mac bits and one with the igc changes on top of it.

I took your original message as having to add new functionality to mac.

that's related to setting maxbw as well, which i assume is used

If there are specific bugs there, then yeah, breaking that separately is reasonable.

Sorry, kind of hard to give specific advice without the context.

yeah, i guess i'll just try to post it, and see where it goes.

If helpful you can push a branch somewhere, send it to me, and I can make a more specific suggestion if that's helpful.

We appreciate you putting in the work, so least we can do to help in that regard.

let me pull it apart and prepare

@rmustacc github.com/siemato/illumos-gate/tree/mac-igc this is the fork. feel free to have a look at it, suggestions are welcome