Do we support the Chelsio T7 now?

Got a wonderful message from SM: "Chelsio just informed us today that the requested T62100-LP-CR is going EOL"

rzezeski: I see this branch, github.com/oxidecomputer/illumos-gate/commits/ry/stlouis-t7 Do you know what the state of the T7 is on illumos?

Smithx10: I have a version of the cxgbe driver that runs against an older version of the T7. The work was put on the back-burner for various reasons. There will be T7 support at some point, but the update to cxgbe is not yet in gate.

Think they said we won't be able to buy at T6 after 9/24/2026

So just to confirm, at the moment If I put a T7 card in.... that th@ng aint' tx'ing or rx'ing nothing :P

*naddddda*

correct

Sorry jbk but I'm obligated to ask.. how's the ice driver coming along?

nothing new to report really..

we had several other emergencies, so hadn't been able to do much work the past couple of weeks...

and we're also moving our lab, so been waiting for the dust to settle from that before picking it back up

i've not tried this (so caveat operator(?)) but supposedly ice can work at least to some degree as an i40e NIC... unfortunately with the ever-decreasing quality of intel's documentation, it's not clear to what degree

e.g. if you could bind the i40e driver to instances

(my suspicion is that the compatibility doesn't go _that_ far and that it'll fail and/or panic the system, but don't really know)

I think I'll skip that experiment :)

I don't see how you can make i40e and ice work as the same device. While they have similar I/O engine constraints, the rest of the firmware interface is entirely different.

there was something buried in the manual.. it might just be for the VF stuff, I can't remember now

(which if so, probably doesn't help much since we don't really use that)

has anyone ran the dtrace test suite recently, and/or would I expect to see failures that are unrelated to a change I'm making?

jperkin: this was quite a while ago, maybe a year or more, but I think there we certain network related tests where I ran into failures because certain assumptions were made around the local network

I think I filed issues too

rzezeski: thanks, the failures don't appear to be related but just wanted to check I wasn't holding it wrong

there are tests that depend on network weather, and some that depend silently on some other stuff

I think maybe the NFS things assume you have some?

danmcd: is dtest in your release script?

It is not.

(At least not yet?)

ah, shame, thought it might give jperkin an easy baseline

[illumos-gate] 18195 Convert mpstat(8) to mdoc(7) -- Andy Fiddaman <illumos⊙fn>