13:01:52 <jki> #startmeeting CIP IRC weekly meeting
13:01:52 <collab-meetbot> Meeting started Thu Apr 11 13:01:52 2024 UTC and is due to finish in 60 minutes.  The chair is jki. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:01:52 <collab-meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:01:52 <collab-meetbot> The meeting name has been set to 'cip_irc_weekly_meeting'
13:01:59 <jki> #topic AI review
13:02:05 <jki> - prepare blog entry on SLTS kernel state and challenges [Jan]
13:02:14 <jki> no update (vacation...)
13:02:37 <jki> anything else from past meetings?
13:02:47 <jki> 5
13:02:49 <jki> 4
13:02:50 <jki> 3
13:02:52 <jki> 2
13:02:54 <jki> 1
13:02:57 <jki> #topic Kernel maintenance updates
13:03:17 <pave1> I'm doing reviews on 6.1.84..86
13:03:18 <uli> i released 4.4, now reviewing 6.1.83
13:03:21 <masami> This week reported 82 new CVEs and 16 updated CVEs.
13:03:40 <iwamatsu__> I reviewed 6.1.84.
13:04:02 <pave1> There's "Intel Atom" vulnerability -- not fixed in 4.19.
13:04:16 <jki> masami: do you also track later on rejected CVEs in the stats?
13:04:41 <masami> jki: not yet but I'll
13:04:41 <pave1> It is again speculative execution kind, so it will not matter for people not running untrusted code.
13:05:01 <jki> pavel: 5.10 got a fix?
13:05:11 <pave1> Yes.
13:05:30 <jki> would backporting be complex?
13:05:40 <jki> not suggesting that we try, just to get a cost feeling
13:06:06 <pave1> Not sure. It touches low level kernel entry/exit, so testing might be tricky.
13:07:01 <jki> well, unless that is rarely used entry/exit code, practical coverage might be still high
13:07:16 <jki> but it would also need a reproducer to validate to mitigation
13:07:40 <pave1> And affected system to run it on.
13:08:02 <jki> but there is no lower hw limited, is there?
13:08:35 <pave1> I believe there is...
13:08:50 <pave1> I'll make a note to take closer look?
13:08:54 <jki> we should highlight this at the next TSC
13:08:57 <jki> yes, please
13:09:39 <jki> anything else?
13:10:12 <jki> 5
13:10:14 <jki> 4
13:10:16 <jki> 3
13:10:18 <jki> 2
13:10:19 <jki> 1
13:10:22 <jki> #topic Kernel release status
13:10:34 <jki> 5.10-rt is late IIRC
13:10:54 <jki> waiting for upstream release?
13:11:12 <pave1> Yep. v5.10.214-rt106-rc1 is out there, so hopefully wait won't be too long.
13:13:09 <jki> ok
13:13:22 <jki> moving on...
13:13:25 <jki> 5
13:13:26 <jki> 4
13:13:28 <jki> 3
13:13:30 <jki> 2
13:13:32 <jki> 1
13:13:34 <jki> #topic Kernel testing
13:13:47 <patersonc> Hello all
13:13:54 <arisut> hi patersonc
13:13:59 <patersonc> Apologies for missing the E-TSC yesterday
13:14:13 <patersonc> Did I miss anything exciting? (I'll review the video once it's out)
13:15:00 <patersonc> We've had some issues with the auto-scaling gitlab runners this week, which I think I've resolved now. Our runner-man is on holiday so they can double check things when they return
13:15:05 <arisut> no idea but I see you at the KernelCI TSC
13:15:14 <patersonc> Please let me know if you spot any issues with the runners
13:15:40 <patersonc> arisut: Yea I try to attend the monthly ones just so I'm half-in the loop
13:15:46 <jki> patersonc: yes, my latest pipelines passed
13:15:59 <patersonc> Great
13:16:04 <arisut> I thought maybe we could try to deploy our own KernelCI instance
13:16:27 <patersonc> pave1: Today's stable-rcs have all gone through as well
13:17:13 <pave1> Umm. No, I had to retry.
13:17:30 <patersonc> Yes, but they have gone through now
13:17:30 <pave1> But this time it was test, not compile phase, so there may be changes there.
13:17:44 <pave1> Aha, yes. After enough retries it usually works :-).
13:18:36 <arisut> still I think that the next set is to work on the tool for sending command to the main instance (like kernel builds and test executions)
13:18:42 <arisut> set/step
13:19:05 <patersonc> arisut: Yea. We should have a chat about what to try next.
13:19:46 <arisut> also send results to the CIP results analyzer tool that now I struggle to rember the name
13:20:47 <patersonc> Yes, otherwise most of the testing done in kernelci is being ignored atm. Need to get that sorted
13:21:40 <arisut> thanks
13:22:38 <jki> what is now the state of consolidating kernel configs to reduce builds (and costs...)?
13:24:16 <iwamatsu__> As for x86, I have  been able to confirm the boot.
13:25:13 <jki> ok, good
13:25:29 <jki> what is missing to finalize x86?
13:25:57 <jki> besides probably also patches to isar-cip-core
13:26:26 <iwamatsu__> I think the .config needs to be reviewed next.
13:26:35 <jki> by whom all?
13:27:05 <jki> I just found the diff for the ipc227e - that's on us, ok
13:27:06 <iwamatsu__> However, this is a merge of two .config files (OBS and IPC), so it may not be necessary.
13:27:46 <jki> CONFIG_NO_HZ on or off, CONFIG_PREEMPT_? - these are likely generic decisions
13:28:00 <iwamatsu__> jki: yes, please review.
13:28:01 <jki> which of those we would like to focus on
13:28:11 <jki> for testing
13:28:27 <pave1> I'd say CONFIG_NO_HZ=y.
13:28:46 <pave1> As NO_HZ is more complex/risky then the other.
13:29:01 <jki> ack
13:29:42 <pave1> CONFIG_PREEMPT... I feel PREEMPT_NONE is not sexy so may get less testing... Might be worth testing both but that goes against the goals.
13:30:33 <jki> well, we also have CONFIG_PREEMPT_RT(_FULL) in a different stream, I suppose
13:31:19 <pave1> Yep. If we are testing other PREEMPT options elsewhere, we might as well test NONE here.
13:31:26 <jki> maybe PREEMPT_VOLUNTARY? or majority vote of known configs
13:31:55 <iwamatsu__> Yes, we need to discuss them. However, the current diff has many items...
13:32:05 <jki> I'm seeing quite a view config switches now being off
13:32:15 <patersonc> Another option is to start building multiple configs in the same build environment. I assume a lot of files wouldn't need to be recompiled each time. At the moment we're very inefficient with a completely clean build env for each config
13:32:16 <jki> that should be changed unless there conflicts
13:32:16 <iwamatsu__> I extract the main features (CONFIG_) from the diff.
13:32:42 <pave1> I feel that we should have PREEMPT_VOLUNTARY in some config and PREEMPT_NONE in some other.
13:33:12 <jki> we still have different archs anyway and could use that to vary some generic switches
13:33:24 <pave1> Yes, that's what I meant.
13:35:01 <jki> ok, I will try to review the ipc thing, but it would be good to have more eyes/opinions
13:35:16 <iwamatsu__> Thank you.
13:35:35 <jki> and then maybe we just start with some version and ask for patches against it
13:35:44 <jki> nothing is set in stone by that
13:36:11 <jki> the key is to switch the mode and, thus, force to work with that central config
13:37:04 <jki> good - more testing topics?
13:37:25 <patersonc> some version could = the defconfig for each arch
13:37:36 <patersonc> jki: I've nothing else
13:38:53 <jki> 5
13:38:55 <jki> 4
13:38:56 <jki> 3
13:38:58 <jki> 2
13:38:59 <jki> 1
13:39:01 <jki> #topic AOB
13:39:28 <jki> I suppose I won't see anyone of you in Seattle next week, right?
13:40:00 <patersonc> Not me, unless you want to take a photo with you
13:40:06 <arisut> not me
13:40:27 <pave1> No plans on my side.
13:40:42 <iwamatsu__> Not me.
13:40:44 <masami> not me
13:40:47 <jki> just checked: I may oversleep our irc meeting next week (6am PST)
13:41:01 <jki> in that case, someone else should please take over
13:41:32 <jki> any other topics?
13:43:14 <jki> 5
13:43:15 <jki> 4
13:43:17 <jki> 3
13:43:18 <jki> 2
13:43:19 <jki> 1
13:43:21 <jki> #endmeeting