13:00:47 <jki> #startmeeting CIP IRC weekly meeting
13:00:47 <collab-meetbot> Meeting started Thu Mar 24 13:00:47 2022 UTC and is due to finish in 60 minutes.  The chair is jki. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:47 <collab-meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:47 <collab-meetbot> The meeting name has been set to 'cip_irc_weekly_meeting'
13:00:47 <brlogger`> Meeting started Thu Mar 24 13:00:47 2022 UTC and is due to finish in 60 minutes.  The chair is jki. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:47 <brlogger`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:47 <brlogger`> The meeting name has been set to 'cip_irc_weekly_meeting'
13:00:51 <jki> hi all
13:00:55 <iwamatsu> hello
13:00:58 <masami> hi
13:00:59 <josiah> Hello
13:00:59 <uli> hello
13:01:03 <pave1> hi
13:05:06 <jki> ok, test crew still missing, but let's get started anyway
13:05:24 <jki> #topic AI review
13:05:34 <jki> 1. Resolve/filter irrelevant failures of KernelCI for 4.4-cip - patersonc & alicefm
13:05:48 <jki> no one around to answer this, I assume
13:06:21 <jki> other AIs?
13:06:41 <jki> 3
13:06:43 <jki> 2
13:06:45 <jki> 1
13:06:47 <jki> #topic Kernel maintenance updates
13:07:13 <pave1> I have reviewed patches for 5.10.108.
13:07:15 <uli> reviewed 5.10.106
13:07:18 <masami> This week reported 8 new CVEs.
13:07:38 <masami> They are not so critical vulnerability.
13:07:50 <iwamatsu> I reviewed 5.10.108.
13:08:30 <pave1> I did work on 4.4-stable branch. It should be up-to-date with 4.9.305, and I believe we should do a -cip release to get more eyes on it and test the process.
13:09:03 <jki> sounds good to me
13:10:19 <pave1> I do believe we should trim down kernel configs. There's a lot of stuff that does not make sense because someone copied defconfig, and we have enabled it as modules.
13:10:52 <jki> how to approach that?
13:10:55 <iwamatsu> pave1: I also replied to your email, do you have any thoughts on the -st tree release tags?
13:11:59 <pave1> jki: Dunno. Ask for explanation for things that look unneeded in the config, and disable option if explanation is not given?
13:12:35 <pave1> jki: Other possibility would be to ask if we have tests to cover specific config option, and disable it if not.
13:13:42 <jki> I'm afraid the test coverage widely relies on kernelci & Co.
13:14:05 <pave1> iwamatsu: I have added localversion-st to the -st tree. You probably want to remove that before releasing -cip.
13:14:06 <jki> what worked best so far was asking if certain switches could be dropped
13:15:05 <pave1> jki: I'm pretty sure kernelci does not have the crazier options enabled.
13:16:00 <jki> make it concrete, then we can iterate over that
13:16:23 <pave1> jki: But yes, we can continue asking. See the ...Disabling XEN in our configs (used by QEMU and Rene... thread on cip-members.
13:17:19 <pave1> jki: I guess I'd like to get consensus that yes, we need configuration to be trimmed down, that it is okay to do and that we'll continue doing it.
13:17:54 <jki> I wouldn't disagree with that
13:18:23 <jki> I can add this as an item to present /remind folks of on our ext-TSC agenda
13:18:48 <pave1> jki: Good :-). Can you state that on the members list? I don't feel comfortable arguing there.
13:19:26 <iwamatsu> pave1: Well, are you planning to tag the linux-4.4.y-st tree for release? For example linux-4.4.y-st-st5.
13:19:27 <pave1> jki: But maybe ext-TSC agenta makes sense, too.
13:19:55 <jki> pavel: we can do both, but let's start with that meeting
13:20:04 <pave1> jki: Ok, thanks!
13:20:32 <pave1> iwamatsu: Yes, I guess I should do the tags, too.
13:20:58 <pave1> iwamatsu: I propose tagging it as "v4.4-st5" ...?
13:21:50 <patersonc[m]> Is my IRC broken? Or has this meeting stalled?
13:21:59 <iwamatsu> pave1: I got it. I think it will probably be such a tag.
13:22:18 <pave1> Ok. Let me do tags in v4.4-stX form.
13:22:26 <iwamatsu> thank you!
13:22:43 <uli> patersonc[m]: i think it's your irc
13:22:57 <alicefm> Hi
13:23:13 <alicefm> Same here
13:23:17 <pave1> (To be clear. I'll do v4.4-st5, not v4.4.302-st5, as .302 would just be repeated everywhere).
13:24:28 <alicef> PatersonMatrix bridge is stalled
13:25:32 <jki> how would 4.4-cip be tagged, what numbers will be incremented there?
13:25:53 <pave1> I guess I'll also do -st-next as doing development on -st branch is confusing.
13:26:24 <pave1> jki: I thought it would be v4.4-cipX or v4.4.302-cipX.
13:27:16 <jki> can be risky to change the naming scheme
13:27:19 <patersonc[m]> Or -st-rc?
13:27:31 <jki> so the second option might be safer
13:27:32 <patersonc[m]> To follow the same convention as other cip branches
13:27:45 <pave1> -st-rc works for me, too.
13:28:18 <pave1> jki: Yes, I guess we can stay with v4.4.302-cipX.
13:29:22 <iwamatsu> hmm, what should we do with the CIP release tag? v4.4-st5-cip69?
13:29:52 <pave1> iwamatsu: I'd just tag it as v4.4.302-cipX.
13:30:36 <pave1> iwamatsu: If someone wants to know what -stX it is based on, it will be in git history.
13:31:21 <jki> continuity, also in naming
13:33:21 <jki> ok, further topics here?
13:33:41 <iwamatsu> If we want to release it, merge -st tree into -cip tree and set the tag to 4.4.302-cipX+1. is this correct understanding?
13:34:09 <pave1> iwamatsu: Correct, AFAICT.
13:34:17 <iwamatsu> pave1: OK
13:34:48 <pave1> (You'll need to delete localversion-st, so that we get correct version info).
13:35:29 <iwamatsu> Yes, I think so.
13:36:46 <jki> so... moving on in...
13:36:50 <jki> 3
13:36:53 <jki> 2
13:36:55 <jki> 1
13:36:58 <jki> #topic Kernel testing
13:38:13 <jki> patersonc[m]: do we have news on the AI?
13:39:27 <alicef> The pull request i sent about preempt rt as been updated and take over with Guillermo
13:40:15 <alicef> With Gtucker pull request
13:40:23 <jki> alicef: what was that about?
13:41:09 <alicef> Adding preempt rt test on KernelCI
13:41:24 <alicef> Also for cip rt branches
13:41:32 <jki> ah, great
13:42:45 <jki> https://github.com/kernelci/kernelci-core/pull/1094 - right?
13:43:35 <alicef> Yes that's it
13:44:40 <jki> how sophisticated are rt tests in kernelci already?
13:45:30 <alicef> What do you mean by sophisticated?
13:46:21 <jki> what do they run, roughly, for how long? already recording latencies or "just" stability topics?
13:48:20 <jki> "it's fine if it boots" is surely not enough here ;)
13:50:12 <alicef> They use rt-test and do some round of cyclic test
13:50:48 <alicef> cyclictest measures event latency in Linux kernel by
13:50:48 <alicef> measuring the amount of time that passes between when a timer
13:50:48 <alicef> expires and when the thread which set the timer actually
13:50:48 <alicef> runs.
13:51:03 <jki> yes, sure
13:51:21 <jki> but you need to run that longer than usual tests, you need to run stress in parallel
13:51:36 <jki> and we eventually also need an I/O loop test
13:51:48 <pave1> cyclictest is kind of default for realtime testing, yes.
13:51:52 <pave1> And it is a bit of art:
13:52:09 <jki> I'm asking as I could try to push that topic via some colleagues, looking into status quo and possible improvements
13:52:29 <pave1> a) Not all systems are suitable; you'll get failures on many machines because firmware is broken.
13:52:43 <pave1> b) You have to load system a bit, but not unrealistically so.
13:53:17 <alicef> https://github.com/kernelci/kernelci-core/issues/474
13:53:25 <jki> "a bit" is relative to what you want to show - usually, you load a lot
13:53:37 <alicef> Is the discussion about rt tests implementation
13:53:38 <jki> and vary that
13:54:50 <pave1> jki: Well, you want to test basic functionality. You don't want to test OOM killer and serial console, for example.
13:55:32 <jki> pavel: if you want to test out numbers, you need to do more than that
13:55:46 <alicef> We are adding what is used on mainline to cip
13:55:49 <jki> pavel: for basic stability tests, that is fine, yes
13:56:11 <jki> that is definitely a valuable start!
13:58:32 <alicef> Maybe write in the KernelCI pr or open a new issue if there are suggestions
13:59:04 <jki> sure, will do - need an overview first
13:59:32 <jki> other testing topics?
14:00:09 <alicef> Not that I know about.
14:00:37 <jki> then move on in
14:00:40 <jki> 3
14:00:42 <jki> 2
14:00:44 <jki> 1
14:00:47 <jki> #topic AOB
14:01:08 <jki> I would need a substitute next Thursday (OOO)
14:02:58 <alicef> I asked patersonc[m] if he can help out with the already opened KernelCI pr as I'm busy in this days.
14:04:47 <alicef> If anyone is interested or want to help out this is the current KernelCI situation on cip https://github.com/orgs/kernelci/projects/11
14:06:15 <pave1> jki: I guess I can send the email and operate the chatbot.
14:06:20 <pave1> jki: Is anything else required?
14:06:48 <jki> pavel: that should be enough, TIA!
14:06:55 <pave1> jki: No problem :-).
14:08:49 <jki> ok - anything else for today?
14:09:07 <jki> 3
14:09:09 <jki> 2
14:09:11 <jki> 1
14:09:12 <jki> #endmeeting