05:59:51 * ** lchlan has quit IRC 06:00:47 * ** lchlan has joined #cip 06:23:06 <szlin> iwamatsu: ping 06:43:12 <szlin> iwamatsu: would you use the same methodology to manage 4.19 kernel source? 06:44:04 <szlin> Both Ben and Daniel created *-rebase branch [1] 06:44:07 <szlin> [1] https://git.kernel.org/pub/scm/linux/kernel/git/cip/linux-cip.git/ 07:43:26 * ** patersonc has joined #cip 07:50:33 * ** Ystk has joined #cip 07:59:24 * ** sangorrin has joined #cip 07:59:29 * ** Tzongyen_Lin has joined #cip 07:59:32 <sangorrin> hi! 08:03:44 <patersonc> Mornin 08:13:09 <sangorrin> I forgot the meeting time changed, 48 minutes wait ;) 08:16:53 <szlin> :D 08:18:22 <sangorrin> I'll write a script to classify the failed patches I sent in the meantime 08:23:04 * ** helmut has quit IRC 08:34:05 <szlin> sangorrin: Before discussing the update mechanism, I suggest we need to define our scope of RFS. 08:40:26 <iwamatsu> szlin: yes, i will use same methodlogy to 4.19. 08:43:10 * ** rajm has joined #cip 08:43:35 * ** vidda has joined #cip 08:47:11 <sangorrin> szlin: yes, today I want to discuss a few questions to ask each member about software updates 08:47:39 <szlin> iwamatsu: thanks. 08:49:00 <rajm> the logger is present but https://irclogs.baserock.org/cip/ hasn't updated since Monday? ironfoot ? 08:49:32 <szlin> sangorrin: BTW, armel won't be an official architecture in Debian 10. 08:50:06 <szlin> The status for armhf support is unknown. 08:50:13 <szlin> [1] https://release.debian.org/buster/arch_qualify.html 08:51:23 <sangorrin> humm and armhf has a ? 08:51:29 <szlin> rajm: perhaps the service broke again. 08:51:34 <sangorrin> that sucks 08:52:43 <sangorrin> what does it mean official? 08:52:56 <sangorrin> https://packages.debian.org/buster/armel/ is there 08:53:27 <sangorrin> and https://packages.debian.org/buster/armhf/ here 08:54:51 <ironfoot> rajm: thanks for the heads up, will look into it 08:55:41 <rajm> ta! 09:00:10 <patersonc> Hi 09:00:17 <szlin> #startmeeting 09:00:22 <szlin> #roll call 09:00:27 <szlin> please say hi if you're here 09:00:38 <gavinlai> hi 09:00:45 <bwh> hi 09:00:48 <sangorrin> hi 09:00:54 <patersonc> hi 09:00:59 <rajm> hi 09:01:02 <iwamatsu> hi 09:01:15 <szlin> #topic kernel maintenance updates 09:01:30 <szlin> === 4.4.156 reviewing === 09:01:31 <szlin> ACK - af25dc4cf354 ("scripts: modpost: check memory allocation results") 09:01:31 <szlin> https://wiki.linuxfoundation.org/civilinfrastructureplatform/linux-4.4.y#section44156 09:01:34 <szlin> ========================= 09:01:36 <szlin> Nobuhiro created development branch of CIP kernel for 4.19.y on git.kernel.org 09:01:37 <szlin> https://git.kernel.org/pub/scm/linux/kernel/git/cip/linux-cip.git/log/?h=linux-4.19.y-cip 09:02:02 <szlin> Daniel sent out some failed patches in 4.4, and I had put them into wiki. 09:02:03 <szlin> https://wiki.linuxfoundation.org/civilinfrastructureplatform/linux-4.4-failed-patches 09:02:11 <vidda> hi 09:02:28 <sangorrin> yes, i just wrote a script to classify them.. will update the wiki soon 09:02:46 <szlin> sangorrin: I had categorized with those patches 09:02:57 <szlin> feel free to modify it :) 09:03:15 <sangorrin> yes, but the script is checking the git log to make sure the current status of those patches 09:03:41 <iwamatsu> I have a comment about 4.19 LTS. 09:03:44 <szlin> sangorrin: awesome, could you share the script via gitlab-cip? 09:03:44 <sangorrin> my previous list was based on the e-mails only, without any checks 09:03:54 <patersonc> sangorrin: Can you share the script? 09:03:58 <sangorrin> sure 09:04:02 <patersonc> Thanks 09:04:07 <szlin> sangorrin: thanks! 09:04:24 <szlin> iwamatsu: yes, please 09:04:56 <iwamatsu> We have not announced the next kernel yet. I think that we should talk about this in TSC. 09:05:21 <patersonc> Agreed. It would be good to get a news item/blog post on the website 09:05:32 <patersonc> Even if for now our branch=LTS. 09:06:33 <szlin> #action Discuss the official CIP kernel in TSC meeting 09:07:01 <szlin> and we can have the announcement afterwards. 09:07:09 <iwamatsu> And I could not find an announcement form Greg on this time's LTS version decision. Please let me know if you know. 09:08:03 <patersonc> I saw an article on linux.com. But nothing from Greg. 09:08:22 <szlin> Me neither 09:08:56 <sangorrin> https://www.kernel.org/category/releases.html 09:09:03 <iwamatsu> szlin patersonc: thanks 09:09:04 <sangorrin> if it's there it's official 09:09:12 <szlin> Nothing from his blog so far, he was used to announce the version of LTS via his blog. 09:09:17 * ** mungaip has joined #cip 09:09:46 <szlin> http://www.kroah.com/log/ 09:09:51 <iwamatsu> szlin: and G+ :) 09:10:13 <iwamatsu> but I could not find form G+. 09:10:14 <szlin> bwh: Did you get the related news? 09:10:17 <bwh> https://lwn.net/Articles/769110/ "... given that this is going to be one of 09:10:18 <bwh> the "Long Term" kernels ..." 09:10:41 * ** hungtran has joined #cip 09:10:56 <szlin> bwh: thanks. 09:11:41 <szlin> any other points? 09:11:55 <iwamatsu> bwh: thanks 09:12:04 <patersonc> o/ 09:12:27 <patersonc> For patches to be backported to the CIP Kernel, do they need to be in a full mainline release? Or is an RC okay? 09:13:00 <bwh> I would say an RC is OK, as that's consistent with stable 09:13:21 <patersonc> Thanks bwh. Is that okay with you iwamatsu? 09:13:56 <iwamatsu> me too.I think that there is no problem if it is merged into Linus's tree. 09:14:04 <patersonc> Thank you! 09:15:11 <szlin> any other points? 09:15:50 <szlin> 3 09:15:51 <szlin> 2 09:15:53 <szlin> 1 09:16:00 <szlin> #topic kernel testing 09:16:04 <rajm> Still running the BBB and iwg20m health checks regularly 09:16:37 <rajm> Tidied up a few more open issues and moved some of their content into 'Known issues' on the wiki 09:16:43 <rajm> thanks! 09:16:49 <patersonc> Thanks 09:16:52 <patersonc> rajm: Have you/or anyone built the CIP Kernel for QEMU? 09:17:09 <rajm> I haven't - and don't know any further 09:17:29 <patersonc> Anyone else? 09:17:41 <bwh> I regularly build for arm and x86 on QEMU 09:18:22 <sangorrin> patersonc: as part of CIP core, yes 09:18:23 <patersonc> Okay. Any complications in building for QEMU? Or should it be trivial? 09:19:02 <sangorrin> https://gitlab.com/cip-project/cip-core/tree/master/deby/poky/meta-cip-qemux86-64 09:19:07 <szlin> we build kernel on x86/64 (usually) or armhf/64 09:19:54 <patersonc> Okay. Thank you all. We'll ping if we get stuck. At the moment the lab is building v4.18. We obviously need to change this to CIP. 09:20:09 <bwh> Just a matter of matching up the QEMU model, the device tree and the drivers enabled 09:20:26 <bwh> I can give a recipe after the meeting 09:20:37 <patersonc> Thanks bwh! 09:20:59 <patersonc> I've got 2 more points... 09:21:10 <patersonc> 1) Does anyone have any feedback/queries about the test infrastructure proposal from the TSC this week? 09:21:28 <sangorrin> yes 09:21:53 <szlin> patersonc: WIP, I could have the discussion in these days. 09:22:12 <sangorrin> 1) will we use that infrastructure for the kernel only, or for cip core base layer as well? 09:22:49 <patersonc> sangorrin: I'm open for it being for more than just the Kernel 09:22:54 <patersonc> Especially for CIP Core 09:23:18 <patersonc> Now is a good time to define these requirements, as it may lead to a change in our plans/design 09:23:25 <sangorrin> 2) where will we share the test definitions (yaml) files? we can reuse a lot from linaro validation I think. 09:23:38 <patersonc> For example, can we build CIP Core in GitLab CI? Or do we need to use Jenkins etc.? 09:24:03 <patersonc> We have a repo on CIP we can use for test definitions. 09:24:24 <sangorrin> CIP Core: it depends which build tool the environment will be different 09:25:02 <sangorrin> I want to propose to add "debos" as one of the CIP core image generators, and use it for CIP core base layer testing 09:25:20 * ** toscalix has joined #cip 09:25:24 <sangorrin> debos is already used in kernelci according to the slides i saw at ELCE 09:25:55 <patersonc> sangorrin: Thanks. I'll take a look. Feel free to email the ML with info as well :) 09:26:06 <szlin> back to kernel testing, any other points? 09:26:15 <patersonc> 2) Would anyone like access to the LAVA master? 09:26:22 <patersonc> It would be great if others could start trying to run some test cases so we can be sure it’s all working. 09:26:40 <sangorrin> patersonc: yes 09:26:57 <sangorrin> I used to have it, until my account was removed ;_; 09:27:04 <patersonc> yes.... 09:27:05 <szlin> sangorrin: how come ? 09:27:18 <sangorrin> I want to add support for BBB 09:27:44 <sangorrin> szlin: long story 09:27:45 <patersonc> szlin: Users get lost on reboot. They need to be hardcoded in the docker/lava conf - don't ask... 09:28:05 <patersonc> sangorrin: BBB - in your own lab or at Renesas? 09:28:19 <sangorrin> in my own lab 09:28:26 <szlin> patersonc: oops.. 09:28:45 <szlin> sangorrin: TL;DR 09:28:55 <patersonc> sangorrin: Okay. We'll try and publish a lab setup wiki in the next couple of weeks. 09:29:46 <patersonc> One more topic: Should we use cip-dev for all testing related matters? Or should some things be in cip-members / a new ML? 09:29:58 <sangorrin> patersonc: thanks, that will help me. 09:30:09 <bwh> I think cip-dev is more appropriate 09:30:27 <szlin> +1 09:30:44 <patersonc> Okay :). Thanks. 09:30:49 * patersonc puts down the mike 09:30:56 <patersonc> mic 09:30:57 <sangorrin> About infrastructure: I want to make sure we can remove logs submitted by mistake (eg containing our public ips or a password) 09:31:36 <patersonc> sangorrin: Details of test runs can be hidden. 09:31:53 <patersonc> I don't know if they can be completely hidden yet though. I mean to check. 09:31:56 <sangorrin> yes, just in case of a mistake or a miss 09:32:51 <patersonc> and private testing :) 09:33:00 <mungaip> sangorrin: yes, you can delete a log of a test 09:33:00 <szlin> any other points? 09:33:10 <szlin> 3 09:33:11 <szlin> 2 09:33:12 <szlin> 1 09:33:14 <szlin> #topic CIP Core 09:33:43 <szlin> https://release.debian.org/buster/arch_qualify.html 09:34:06 <sangorrin> OK, so as you know we will support two profiles: tiny and generic 09:34:13 <szlin> As I mentioned before, armel won't be a candidate for Debian 10 09:34:35 <sangorrin> the packages are already available 09:34:37 <bwh> szlin: No? 09:35:16 <sangorrin> https://packages.debian.org/buster/armel/ 09:35:23 <sangorrin> https://packages.debian.org/buster/armhf/ 09:35:33 <sangorrin> szlin link: https://release.debian.org/buster/arch_qualify.html 09:35:44 <sangorrin> armhf has a question mark 09:35:50 <szlin> bwh: it marks no in architecture requalification status for buster 09:36:05 <szlin> concerns-dsa : do not want to keep the hardware running past 2020 09:36:38 <bwh> Both armel and armhf have the problem that the build systems are dev boards that are hard to use remotely 09:37:03 <bwh> The plan has been to switch them to arm64 hardware in 32-bit compatibility mode 09:37:06 <sangorrin> bwh: one of the initial goals of CIP was to support the cross-compile effort 09:37:17 <bwh> Yes, but Debian itself won't use that 09:38:14 <sangorrin> will there be support for armhf through arm64 in 32-bit compatibility mode then? 09:38:57 <bwh> Some arm64 processors support that, yes 09:39:44 <sangorrin> szlin: maybe we need some clarifications, because armhf is fundamental for most members 09:40:33 <szlin> #action Make sure armhf is available in Debian 10 09:41:08 <szlin> sangorrin: so both tiny and generic profiles depend on Debian ? 09:41:20 <sangorrin> yes 09:41:59 <sangorrin> if Debian does not support armhf, we will have to 09:42:15 <sangorrin> at least in the tiny profile 09:42:16 <szlin> iwamatsu: do you have any comments? 09:42:43 <szlin> sangorrin: thanks, could you please share the profile link? 09:43:15 <sangorrin> ok, so the tiny profile will be based on meta-debian v2 09:43:24 <sangorrin> meta-debian v2 is currently in development 09:43:31 <sangorrin> https://github.com/meta-debian/meta-debian/tree/master-ng 09:43:40 <szlin> sangorrin: thanks. 09:43:41 <iwamatsu> I think that there is no problem with armhf. 09:43:49 <sangorrin> I think it already builds 09:44:18 <iwamatsu> armel is aware that there is a problem with support on debian 10 because hardware necessary for build is not found. 09:44:36 <sangorrin> meta-debian v2 is still unstable but can be used 09:44:58 <sangorrin> It would be nice to receive contributions: recipes for the base layer components 09:45:01 <iwamatsu> yes, i am using and developing too. 09:45:04 <sangorrin> or bug fixes 09:45:12 <sangorrin> iwamatsu: thank you! 09:45:51 <szlin> any other points? 09:45:56 <sangorrin> new products using CIP are expected to use meta-debian v2 (unstable, will be stable in a few months) or meta-eid (still in prototype phase) 09:46:05 <sangorrin> yes 09:46:28 <sangorrin> so meta-debian v2 will try to be compatible with yocto recipes as much as possible 09:46:44 <sangorrin> we want vendors to support meta-debian v2 with minimal effort 09:47:12 <patersonc> sangorrin: Compatibility with yocto would be welcome. 09:47:18 <sangorrin> yes 09:47:35 <sangorrin> we need to define the base layer again 09:47:42 <sangorrin> have it working on each reference board 09:47:47 <sangorrin> and _test_ it 09:48:24 <szlin> sangorrin: thanks. 09:48:35 <szlin> any other points? 09:48:39 <szlin> 3 09:48:40 <sangorrin> so if possible, people that intends to use the tiny profile please start now, and contribute 09:48:41 <szlin> 2 09:49:02 <sangorrin> one more thing 09:49:21 <sangorrin> i also want to add reference rootfs using debos 09:49:38 <sangorrin> this tool uses debootstrap 09:49:48 * ** Ystk has quit IRC 09:49:53 <sangorrin> i want to know if someone is interested 09:50:08 <szlin> https://www.collabora.com/news-and-blog/blog/2018/06/27/introducing-debos/ 09:50:31 <sangorrin> basically, it will be pure debian 09:51:30 <sangorrin> I will ask in the mailing list to the members if they plan to use cip core or not 09:51:47 <szlin> sangorrin: ok, I will take a look on debos 09:52:02 <sangorrin> and also which profile or image generator they are interested in 09:52:15 <iwamatsu> https://wiki.debian.org/SystemBuildTools There are many similar tools. 09:52:29 <sangorrin> yes, i have investigated most of them 09:52:34 <sangorrin> there are more 09:54:06 <szlin> #action Daniel will send email to cip-dev for cip-core 09:54:20 <szlin> 1 09:54:25 <szlin> #topic Software update 09:54:39 <sangorrin> ok, so this topic is related to cip core 09:54:48 <sangorrin> because the goal is to enable software updates for CIP core reference boards 09:55:06 <sangorrin> I am going to ask members several questions 09:55:15 <sangorrin> 1) Do you agree with the goal of the CIP sw updates work group? 09:55:20 <sangorrin> 2) What profiles of CIP are you interested to have updates (tiny, generic)? 09:55:25 <sangorrin> 3) Are you using sw update tools in production? which? (custom, swupdate, ostree, mender..) 09:55:29 <sangorrin> 4) If a custom one, would you open source it? 09:55:34 <sangorrin> 5) What sw update tools would you like to see supported? 09:55:39 <sangorrin> 6) Do you have any preferable update method? (sector diff, A/B, file binary diff, file based) 09:55:58 <sangorrin> if you want to add some question, or have a comment on my questions let me know 09:57:59 <szlin> sangorrin: I suggest to ask/ answer these questions via cip-members ML. 09:58:08 <sangorrin> yes, i will do that 09:58:11 <szlin> Some questions need to have internal consensus 09:58:28 <sangorrin> 7) what do you want to update (bootloader, kernel, rootfs, app..) 09:58:54 <sangorrin> i will ask in the cip-members list, here i want to agree on the questions to ask 09:59:16 <szlin> Any objections? 09:59:41 <patersonc> Seem like sensible questions to me 10:00:35 <szlin> #action Daniel will send above questions via cip-members. 10:00:41 <sangorrin> thanks 10:00:46 <szlin> sangorrin: thanks! 10:01:29 <szlin> sangorrin: did you know deltadeb? 10:01:35 <szlin> https://wiki.debian.org/Teams/Dpkg/Spec/DeltaDebs 10:01:44 <sangorrin> no, thanks! 10:02:07 <sangorrin> i will check it out 10:02:20 <szlin> and also debdelta 10:02:21 <szlin> http://debdelta.debian.net/ 10:02:38 <sangorrin> nice thanks 10:02:58 <sangorrin> btw, debos already supports ostree 10:03:16 <sangorrin> for the first prototype i want to do as little effort as possible haha 10:03:47 <szlin> sangorrin: AFAIK, Endless OS is using ostree + flatpak 10:04:22 <sangorrin> thanks, ill check that. I didn't even know about such os 10:04:25 <szlin> note: Endless OS is a Debian derivative distribution 10:05:10 <szlin> any other points? 10:05:17 <sangorrin> btw, the reproducibility project is important for reducing deltas 10:05:46 <sangorrin> szlin: not from my side 10:06:00 <szlin> #topic AOB 10:06:07 <szlin> any other business? 10:06:24 <szlin> 3 10:06:26 <szlin> 2 10:06:27 <patersonc> o/ 10:06:40 <patersonc> Should we be creating working group pages on the wiki? 10:07:06 <sangorrin> Agustin proposed to send a report before the TSC meetings 10:07:25 <sangorrin> we could save those reports in the wiki 10:07:48 <szlin> yap, he sent the email yesterday 10:07:52 <patersonc> Okay. I think we need to make sure it looks like the project is active, so as many public updates as possible. 10:09:27 <szlin> 3 10:09:28 <szlin> 2 10:09:30 <szlin> 1 10:09:39 <szlin> #endmeeting