13:01:44 <jki> #startmeeting CIP IRC weekly meeting 13:01:44 <collab-meetbot`> Meeting started Thu Apr 24 13:01:44 2025 UTC and is due to finish in 60 minutes. The chair is jki. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:44 <collab-meetbot`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:01:44 <collab-meetbot`> The meeting name has been set to 'cip_irc_weekly_meeting' 13:01:46 <patersonc> Hey 13:01:51 <jki> #topic AI review 13:01:54 <jki> - write docs for release scripts [iwamatsu-san] 13:02:30 <iwamatsu__> I am updating it , but not finish. 13:02:40 <jki> ok 13:02:46 <jki> no other AIs recorded 13:02:51 <jki> 5 13:02:53 <jki> 4 13:02:54 <jki> 3 13:02:56 <jki> 2 13:02:57 <jki> 1 13:02:59 <jki> #topic Kernel maintenance updates 13:03:12 <pave1> Hi! 13:03:16 <uli_> i pushed 4.4 13:03:18 <masami> This week reported 28 new CVEs and 4 updated CVEs. 13:03:23 <pave1> I'm reviewing 6.1.135. 13:03:59 <iwamatsu__> I reviewed 6.1.132 13:05:30 <jki> anything else to add? 13:05:44 <jki> 5 13:05:46 <jki> 4 13:05:48 <jki> 3 13:05:49 <jki> 2 13:05:51 <jki> 1 13:05:53 <jki> #topic Kernel release status 13:05:59 <jki> all on track 13:06:27 <jki> anything to add? 13:06:37 <jki> 5 13:06:40 <jki> 4 13:06:41 <jki> 3 13:06:43 <jki> 2 13:06:45 <jki> 1 13:06:47 <jki> #topic Kernel testing 13:07:54 <arisut> looking into why kernelci is not using cip configurations anymore looks like after moving from kernelci legacy to pipeline we loose the cip configuration settings 13:08:35 <patersonc> I updated the stable-rc testing so we're testing the v6.14 and 6.15 etc. 13:08:38 <arisut> I'm currently trying to restore the configuration to kernelci pipeline 13:09:52 <patersonc> Thanks Arisu-san 13:10:02 <arisut> patersonc, any idea on why https://github.com/kernelci/kernelci-pipeline/pull/840 is missing the cip:// configurations? 13:11:12 <patersonc> I guess the plan was to add them later but didn't 13:11:19 <arisut> ok thanks 13:11:35 <patersonc> No technical reason - I just didn't try 13:12:01 <arisut> I see, I will try to restore it on pipeline 13:13:06 <arisut> also I think we want to run exclusively cip:// configurations 13:13:44 <arisut> gitlab is testing only cip configurations I suppose 13:14:20 <arisut> Kernelci should do same 13:14:35 <patersonc> We also test some defconfigs in CIP 13:14:42 <arisut> oh really? 13:14:53 <patersonc> But technically we only support CIP configs 13:15:13 <arisut> I see so we build defconfigs and cip configs 13:15:22 <jki> problem is likely not testing - it's evaluating the results 13:15:40 <jki> will we look at defconfig errors? 13:15:44 <arisut> so less results is better ? 13:15:52 <pave1> We don't want to introduce regressions. 13:16:02 <arisut> curerntly we are looking at defconfig ? 13:16:07 <arisut> on gitlab 13:16:29 <pave1> If we introduced a bug so bad it affect defconfig, I'd really wanted us to investigate. 13:16:58 <jki> then let's test defconfigs if maintainers do bother - fine 13:16:59 <arisut> do you mean defconfig run only at needed ? 13:17:20 <arisut> but usually don't get tested ? 13:17:30 <patersonc> I would assume that generic users are more likely to try defconfig as a first step than CIP configs as they are in-tree 13:17:35 <pave1> We do look at gitconfigs on the gitlab. bbb_defconfig is normally only one that needs to be retried. 13:17:40 <patersonc> arisut: defconfig is always run 13:18:20 <arisut> we have generic users ? by policy cip is only supported on cip configurations afaik 13:19:07 <pave1> arisut: People are using our kernels, yes. We got some patches for 4.X because we broke audio on phones or something like that. 13:19:28 <arisut> i see thanks 13:19:39 <pave1> arisut: Catching bugs early is quite cheap. Catching it after it is in tree for a year is less fun. 13:20:06 <pave1> arisut: But yes, supported configs on supported boards is only thing we officially support. 13:20:21 <arisut> but referring to jki "problem is not testing, it's evaluating the results" 13:21:03 <arisut> more things we tests more we have results to evaluate 13:22:10 <jki> we can obviously revisit configs if they stay in permant failures, and no one looks at those 13:22:19 <pave1> arisut: Yes. Current gitlab setup makes it reasonably easy. 13:22:20 <arisut> I'm ok on enabling defconfig and cip config 13:22:27 <patersonc> Sounds good Arisu 13:22:47 <arisut> pave1, current kernelci have a system for sending only regressions 13:23:25 <pave1> arisut: Yes, so regressions are what we really care about. If we break boot on arm64_defconfig, 13:23:30 <pave1> it is quite a big deal. 13:23:39 <arisut> that i shared weeks ago and I don't find the link now 13:24:05 <arisut> we are sending mail with only the board that have regression 13:24:34 <arisut> also dashboard is becoming better https://d.kernelci.org/tree 13:25:58 <jki> ok - anything to add regarding this or testing in general? 13:26:24 <jki> 5 13:26:26 <jki> 4 13:26:28 <jki> 3 13:26:29 <jki> 2 13:26:31 <jki> 1 13:26:33 <jki> #topic AOB 13:27:00 <jki> just a recap: we have green lights to release 6.12-cip when we think it is ready 13:27:06 <jki> TSC approved 13:27:34 <pave1> This may be good time. 13:28:01 <patersonc> Any objections to adding the 6.12 tree to KernelCI? 13:28:53 <jki> well, rather a precondition ;-) 13:29:33 <iwamatsu__> I will not be able to attend the next meeting. Japan is holiday season. 13:30:59 <jki> 4.19.124-cip27 and 4.19.216-cip61, rcu stall on one cpu, apparently over held hb spinlock in futex_wake 13:31:06 <jki> does that ring any bells by chance? 13:31:36 <pave1> "bh"? 13:31:40 <pave1> No, not really. 13:31:51 <jki> hash bucket lock 13:32:09 <pave1> Aha, so not a bottom half. Still no bells. 13:32:12 <jki> but there are no suspicious changes on kernel/futex.c, also in newer kernels 13:32:43 <jki> ok - we need to hope for better reproducibility then 13:32:59 <jki> anything else for today? 13:33:23 <jki> 5 13:33:25 <jki> 4 13:33:25 <pave1> So ... we wait for KernelCI 13:33:28 <pave1> with the 6.12? 13:34:04 <patersonc> Shouldn't be a big job to add 13:34:31 <jki> you likely mean our linux-6.12.y-cip branch, right? 13:34:38 <patersonc> I'll create an MR today 13:34:42 <patersonc> jki: Yes 13:34:47 <jki> perfect 13:35:50 <pave1> I guess I could do 6.12 release with current tree (no version update, yet) after the KernelCI is ready 13:35:54 <pave1> if everyone is ok with that? 13:36:22 <patersonc> Did we get the configs we need from members? 13:36:45 <iwamatsu__> I got some MR for 6.12. I am checking it. 13:37:34 <jki> that's for qemu and x86 targets 13:37:46 <patersonc> There's another for Renesas arm64 13:38:05 <jki> ah, right 13:39:00 <jki> all those should be considered for kernelci 13:40:43 <patersonc> I'll just add defconfig today - then Arisu can add the CIP configs at the same time as the other branches 13:41:42 <patersonc> We don't have a 6.12-cip-rt yet right? 13:42:08 <pave1> There won't be 6.12-cip-rt tree. 13:42:16 <pave1> But we should probably do some configs with rt enabled. 13:42:30 <jki> indeed 13:42:38 <arisut> patersonc, ok 13:43:15 <pave1> So I guess we revisit 6.12-cip1 next week? :-) 13:45:23 <jki> sure - except that we will need iwamatsu-san for the final decision, and he will be off 13:45:52 <jki> but let's prepare the testing and configs and see when that is ready 13:47:25 <jki> anything else? 13:47:51 <jki> 5 13:47:54 <jki> 4 13:47:55 <jki> 3 13:47:57 <jki> 2 13:47:59 <jki> 1 13:48:01 <jki> #endmeeting