13:01:12 <jki> #startmeeting CIP IRC weekly meeting
13:01:12 <brlogger`> Meeting started Thu Dec  9 13:01:12 2021 UTC and is due to finish in 60 minutes.  The chair is jki. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:01:12 <brlogger`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:01:12 <brlogger`> The meeting name has been set to 'cip_irc_weekly_meeting'
13:01:16 <jki> hi all!
13:01:27 <pav31> hi!
13:01:28 <uli> hello
13:01:33 <alicefm> Hi
13:01:35 <masami> hi
13:02:21 <iwamatsu> hi
13:02:48 <josiah> Hi
13:04:07 <jki> #topic AI review
13:04:10 <jki> 1. Combine root filesystem with kselftest binary (finishing) - alicef
13:04:27 <alicefm> Not yet done
13:04:45 <jki> 2. Perform initial comparison of KernelCI results 5.10 LTS vs. CIP - iwamatsu & alicef
13:05:16 <alicefm> Send the analysis on the announce thread
13:05:22 <alicefm> Sent
13:05:29 <iwamatsu> thanks
13:05:50 <alicefm> I didn’t find any worth regression
13:06:12 <alicefm> The two regression looks like caused by board problems
13:06:42 <iwamatsu> https://lore.kernel.org/cip-dev/TYAPR01MB62524C420EECB609794B0587926B9@TYAPR01MB6252.jpnprd01.prod.outlook.com/T/#m3ce69fea19de4537cd422243e7b0cd3d0792a205
13:07:52 <jki> that's good news, indeed - thanks, alicefm
13:09:32 <jki> means, we can now use the kernelci diff between to runs of the same branch in order to find regressions, right?
13:09:55 <alicefm> Yes
13:10:16 <jki> great
13:11:07 <jki> 3. Propose tweet on KernelCI-CIP collaboration progress - alicef
13:12:08 <alicefm> Not yet done. I’m currently busy with open source conference Japan 2021 presentation
13:12:27 <alicefm> Skip
13:12:30 <jki> ok, no problem!
13:12:47 <jki> we had a tweet this week on the 5.10 release
13:13:19 <jki> anything else for AIs?
13:13:30 <jki> 3
13:13:33 <jki> 2
13:13:35 <jki> 1
13:13:36 <jki> #topic Kernel maintenance updates
13:14:14 <uli> finished 5.10.80 review
13:14:16 <masami> There is three new CVEs. They  have already been fixed in the mainline.
13:14:54 <pav31> I was reviewing 5.10.84.
13:14:55 <pav31> Also 5.10-rt is now released.
13:14:56 <iwamatsu> I am reviewing 5.10.84
13:15:29 <iwamatsu> thanks for release -rt tree!
13:15:59 <pav31> You are welcome, and thanks for 5.10-cip :-).
13:16:18 <pav31> Timing worked well.
13:16:31 <iwamatsu> ;-)
13:16:35 <jki> indeed :)
13:18:02 <jki> anything else here?
13:18:15 <jki> 3
13:18:17 <jki> 2
13:18:20 <jki> 1
13:18:23 <jki> #topic Kernel testing
13:19:41 <alicefm> I’m currently enabling cip 5.10 rt on KernelCI but we have problem finding a machine that can run preemptrt test
13:20:20 <jki> what are the restrictions?
13:21:16 <alicefm> The only board that we have enabled is failing the boot now we are looking some different machine
13:21:33 <alicefm> Not sure what you mean by restriction
13:22:06 <alicefm> We have enabled for preempt-rt
13:22:11 <jki> thanks, that explains it
13:22:20 <pav31> You may want to try socfpga based boards.
13:22:49 <pav31> They have reasonable max latencies.
13:23:04 <alicef> you can see the problem explanation here https://github.com/kernelci/kernelci-core/pull/908
13:23:45 <alicef> second topic about automation of check with cip and lts results and regression.
13:24:20 <patersonc[m]> Can't KernelCI add the r8a779x_usb3_v3.dlmem firmware to the filesystem for the Renesas boards?
13:24:23 <alicef> looks like is something that is needed also for some other tree like rc kernel versions
13:24:57 <alicef> patersonc[m]: it should already build that module afaik
13:24:58 <patersonc[m]> yep
13:25:42 <alicef> anyway going back to results comparison
13:26:36 <alicef> unfortunatly currently KernelCI is rewriting the KernelCI API and that from KernelCI Technical Sterring Committee point of view have priority over such work
13:27:53 <alicef> and because the comparison system would be dependent of the KernelCI API we should implement it on the new API that is currently not finished yet
13:28:24 <alicef> implementing on the old API is mostly a lost of time as that will be deprecated in the near future
13:29:09 <jki> what is the rough timeline for this rewrite?
13:29:34 <alicef> mmm that's a good question
13:30:05 <alicef> they already working on it and the repository looks going on with steady pace
13:30:34 <pav31> Could we get some improvements to our gitlab testing?
13:30:51 <alicef> this is the project panel about KernelCI new api https://github.com/orgs/kernelci/projects/10
13:31:06 <pav31> It looks like it will be useful even with kernelci in place.
13:31:13 <alicef> pav31: sorry, what do you mean ?
13:31:51 <pav31> The tests we are running with each gitlab push...
13:32:04 <alicef> the comparison results use KernelCI results as the format is standardized
13:32:40 <alicef> currently working on gitlab is mostly duplicating the work as KernelCI is already covering most of the work
13:32:45 <pav31> Yes, Im talking about the old system.
13:33:22 <pav31> I dont see kernelci as a replacement.
13:33:34 <alicef> I don't think we have reasons to go on adding new tests to the old system
13:34:26 <pav31> By running tests on our systems we can run them as needed, without competing with kernelci stuff.
13:34:31 <alicef> why not ? what we are missing ? we are using the same boards as the old system and the work on KernelCI is following your descriptions
13:35:02 <pav31> Old system was getting results in half an hour.
13:35:22 <alicef> as patersonc[m] said we can your branch to kernelci and you can get your results
13:35:39 <pav31> Thats needed for development and not feasible with kernelci afaict.
13:36:26 <alicef> you can send build requests also to KernelCI
13:37:42 <iwamatsu> We haven't done any pre-release testing on kernelci with our test branch right now, so if we can do this, it's okay, I think.
13:37:49 <alicef> is probably something to be discussed with KernelCI TSC but as also patersonc[m] said is something possible
13:38:43 <alicef> developing both for KernelCI and the old system is not a feasible solution in the long run
13:39:18 <alicef> and using KernelCI we can use KernelCI resources
13:39:39 <alicef> both development and hardware related
13:39:40 <pav31> I'm not sure kernelci is suitable for all our needs.
13:40:11 <jki> what is needed to find that out?
13:40:19 <pav31> We may need a lot of hardware resources if we'll need to chase a bug.
13:41:01 <alicef> all the requirement I got for moving to KernelCI from CIP testing team are currently implemented
13:42:16 <jki> pav31: is there anything concrete that should be tried out with kernelci? or explained how to do that?
13:42:53 <alicef> I got a list from patersonc[m] about all the requirement that KernelCI should satisfy and currently that list is almost done
13:43:13 <pav31> I am worried about latency between submitting test tree and getting results back.
13:43:16 <alicef> if you want you can ask to patersonc[m]
13:44:19 <jki> pav31: is anything still missing for you to actually test that?
13:44:37 <jki> numbers would likely help with this discussion
13:44:50 <alicef> sorry I don't have much time open to discussion today. if you have request, please add a issue to kernelci-core with the label cip
13:45:28 <pav31> I guess we'll find out when we try to debug real problem.
13:45:33 <alicef> pav31: KernelCI build can be limited to only a laboratory
13:46:28 <alicef> but if you want a account on the jenkins staging for pushing new build you need to ask to request for a account
13:47:24 <alicef> and you can make your own push request with only your rt kernels and only lab cip laboratory for example
13:48:20 <alicef> ok
13:48:38 <alicef> but if need something just send a request
13:48:59 <iwamatsu> If my underdtanding is correct, I thought the issue was that the flow from push a branch to sending test results from KernelCI was unclear.
13:49:11 <alicef> or I don't know what you need
13:49:12 <iwamatsu> Will push be hooked on kernelCI immediately?
13:50:03 <alicef> in case that you can get a account on kernelCI staging yes
13:51:16 <alicef> that is a possibility
13:52:10 <alicef> I will ask to TSC for ideas on how I can get you an account
13:52:27 <alicef> if is needed
13:53:16 <pav31> I guess at least me and iwamatsu will need an account...
13:53:21 <jki> ok, thanks in advance!
13:53:30 <iwamatsu> I see.
13:53:58 <jki> and a guided test run might also help, I think
13:54:00 <iwamatsu> Same test as now, I want the test result in a few hours after pushing.
13:55:29 <alicef> I will try to ask
13:55:58 <alicef> but please make a list of what you need
13:56:00 <iwamatsu> OK, thanks.
13:56:29 <alicef> because from patersonc[m] requirements list the work on kernelCI is pratically done
13:56:39 <alicef> and I already worked on that
13:58:01 <jki> good - anything else regarding testing?
13:58:22 <jki> 3
13:58:23 <jki> 2
13:58:26 <jki> 1
13:58:29 <jki> #topic AOB
13:59:27 <jki> anyone any business?
13:59:38 <alicef> no
13:59:49 <jki> 3
13:59:52 <jki> 2
13:59:53 <jki> 1
14:00:00 <jki> #endmeeting