04:14:54 * ** alicef has quit IRC 04:15:56 * ** alicef has joined #cip 04:16:36 * ** alicef has quit IRC 04:16:36 * ** alicef has joined #cip 04:37:39 * ** dl9pf has quit IRC 04:49:55 * ** dl9pf has joined #cip 04:49:55 * ** dl9pf has joined #cip 06:31:34 * ** wagi has quit IRC 07:31:55 * ** robertmarshall has joined #cip 07:39:01 * robertmarshall runs a daily health on the iwg20m with an email to the list. 07:40:23 * ** robertmarshall is now known as rajm 07:41:15 <rajm> unfortunately the BBB got left at home - let me see if I can find an appropriate cable for a spare board... 07:47:21 * ** Ystk has joined #cip 07:56:01 * rajm runs a daily health on the BBB with an email to the list. 07:58:01 * ** patersonc has joined #cip 07:58:06 * ** wagi has joined #cip 07:59:29 * ** Tzongyen_Lin has joined #cip 08:00:06 <szlin> #startmeeting 08:00:13 <szlin> #roll call 08:00:20 <szlin> please say hi if you're around 08:00:26 <szlin> ._./ 08:00:28 <gavinlai> hi 08:00:31 <Tzongyen_Lin> hi 08:00:32 <bwh> hi 08:00:37 <patersonc> hi 08:00:49 <wagi> hi (partial around, also on a telco in parallel) 08:01:02 <szlin> #topic AI review 08:01:05 <rajm> hi 08:01:14 <szlin> - AI: Create a wiki page for collaboration (SZ Lin) 08:01:16 <szlin> * https://wiki.linuxfoundation.org/civilinfrastructureplatform/linux-4.4.y 08:01:47 <szlin> Any suggestions? 08:02:28 <patersonc> When you say "ACK: SZ Lin...", is it for the patches listed above? 08:02:50 <szlin> that's right 08:03:05 <patersonc> Okay 08:03:17 <szlin> ack the patch each by each 08:03:18 <iwamatsu> Hi 08:03:38 <szlin> because some patches are not in CIP kernel maintenance scope 08:04:02 <szlin> feel free to provide any idea afterwards. 08:04:15 <szlin> -AI: Discuss patchwork item in TSC meeting (Chris) 08:04:26 <szlin> *Example: https://patchwork.kernel.org/project/linux-renesas-soc/list/ 08:05:05 <patersonc> I think the outcome of the discussion about patchwork was that no one saw any issues with using patchwork, or proposed any alternitives 08:05:24 <patersonc> Does anyone here have any other views? 08:05:35 <patersonc> iwamatsu? (as the main maintainer going forward...) 08:06:32 <szlin> patersonc: which one do you prefer? https://patchwork.ozlabs.org/ or https://patchwork.kernel.org/ 08:07:21 <iwamatsu> patersonc: patchwork makes it easy to understand the state of the patch, I think. I agree using pathcwork. 08:07:46 <patersonc> I don't have a strong opinion, but kernel.org seems more 'legit', but only because I've never really heard of ozlabs until last week... 08:08:01 <patersonc> iwamatsu: Okay 08:08:16 <patersonc> szlin: Either would be fine 08:08:45 <szlin> Any objections? 08:09:56 <szlin> #agreed CIP kernel maintenance team will use patchwork 08:10:49 <szlin> patersonc: do you need any assistence with this AI? 08:11:15 <patersonc> Only if someone is keen to set it up. If no one is then I'll pick it up. 08:12:11 <szlin> anyone? 08:12:50 <patersonc> I'll email kernel.org first and see if they'll take us. If not I'll try ozlabs. 08:13:46 <szlin> #action patersonc will apply patchwork for CIP kernel maintenance team 08:13:51 <szlin> patersonc: thanks! 08:13:56 <patersonc> No worries 08:13:59 <szlin> #topic kernel maintenance updates 08:14:21 <szlin> Moxa UC-2100 dts was merged, I will check if we have any other in-house patches 08:14:22 <szlin> https://www.lkml.org/lkml/2018/9/28/880 08:15:03 <szlin> furthermore, I've sent CVE-2018-10880 kernel patch for 4.4.y 08:15:05 <szlin> https://www.spinics.net/lists/stable/msg260103.html 08:15:14 <patersonc> Great 08:15:25 <szlin> === 4.4.155 reviewing === 08:15:25 <szlin> I am taking my personal leave until next week, and thus I don't have any updates. 08:16:04 <wagi> v4.4-rt update: Peter Z recommended to use ticket spinlocks for -rt instead of qspinlocks. I retested v4.4-rt with ticket spinlocks and the problem are gone. 08:16:21 <szlin> wow... 08:16:43 <szlin> wagi: do you find out the root cause? 08:16:58 <wagi> it is still under investigation 08:17:04 <wagi> pretty nasty stuff 08:17:24 <szlin> wagi: true.. thanks for your update 08:17:25 <bwh> Ticket spinlocks are what mainline uses, right? 08:17:46 <wagi> mainline uses qspinlocks these days 08:17:53 <bwh> Oh I see 08:18:15 <wagi> qspinlocks scale easily up to hundreds of cpu 08:18:48 <wagi> while ticket spinlocks is usuless for system with more then 16 cpus (if all cpus spin on the same lock, obviously) 08:19:32 <szlin> some references about kernel lock - http://events17.linuxfoundation.org/sites/events/files/slides/linuxcon-2014-locking-final.pdf 08:19:43 <szlin> I'm not quite familiar with qspinlocks 08:20:25 <wagi> the concept is 'easy' but the implementation is really really hard stuff 08:20:50 * ** mungaip has joined #cip 08:21:01 <wagi> anyway for 4.4-rt it is solved 08:21:17 <szlin> wagi: that's interesting, I will take a look with it afterwards, thanks 08:21:28 <wagi> I guess for 4.9 fixex could be backported 08:21:36 <wagi> bigeasy has done it for 4.9-rt 08:22:21 <wagi> the problem is that there are quite a few dependencies (atomics API changes etc) 08:23:21 <wagi> I shut up if no one has question now :) 08:23:43 <szlin> wagi: do you have related URL with 4.9-rt? 08:24:12 <wagi> IIRC bigeasy has not posted anything, still WIP 08:25:06 <wagi> I suppose any fix will go via Greg KH's stable trees 08:25:28 <wagi> I'll ask bigeasy what he's plan are 08:26:01 <szlin> wagi: Please keep us posted on the status :) 08:26:10 <wagi> sure will do 08:26:19 <szlin> any other points? 08:26:44 <szlin> 3 08:26:45 <szlin> 2 08:26:46 <szlin> 1 08:26:53 <szlin> #topic kernel testing 08:27:39 <rajm> hi - there's a MR in for the change to b@d device types to get iwg20m - some discussion around whether using zImage directly is ok 08:28:11 <rajm> I'm running daily tests on both BBB and iwg20m when in the office 08:28:52 <rajm> still having issues with uboot and lava-docker will discuss later 08:28:54 <rajm> thanks 08:29:10 <szlin> rajm: thanks for the updates. 08:29:18 <szlin> any questions? 08:30:02 <szlin> 3 08:30:03 <szlin> 2 08:30:04 <patersonc> o/ Update from Renesas 08:30:04 <szlin> 1 08:30:11 <patersonc> We were getting somewhere with the CIP LAVA master setup, but then the admin rebuilt the docker container from scratch rather than saved the instance before a reboot, so now we need to create all the user accounts and exchange passwords over the phone again. 08:31:27 <szlin> patersonc: thanks. 08:31:42 <szlin> any other points? 08:31:48 <patersonc> Nope 08:31:53 <szlin> #topic AOB 08:32:31 <szlin> any points? 08:32:34 <bwh> The Debian kernel team had a meeting on Tuesday evening 08:32:48 <bwh> We agreed to use 4.19 for the buster release 08:33:22 <patersonc> Thanks bwh 08:33:49 <wagi> Is 4.19 going to be a LTS? 08:33:55 <bwh> yes 08:34:00 <patersonc> I assume there are therefore no outstanding reasons for CIP not to use LTS v4.19 for the next SLTS? 08:34:13 <szlin> #info Debian 10 (buster) selects 4.19 kernel 08:34:23 <patersonc> Is this something the maintainers should take to the TSC (and maybe announce to the world at ELC?) 08:34:25 * ** toscalix has joined #cip 08:34:56 <iwamatsu> debian kernel team meeting log: http://meetbot.debian.net/debian-kernel/2018/debian-kernel.2018-10-02-18.33.html 08:35:56 <szlin> iwamatsu: thank you for the link. 08:36:12 <toscalix> thanks 08:36:53 <szlin> patersonc: yes, I think 4.19 might be the version of SLTS 08:37:20 <patersonc> "Might"? Is there any reason it wouldn't be? 08:37:56 <szlin> patersonc: we need the official announcement 08:38:22 <toscalix> there won't be 08:38:23 <patersonc> szlin: Good point 08:38:59 <patersonc> So assuming v4.19 _is_ an LTS, we will take it? If for some reason Greg doesn't use v4.19 afterall we'll have a rethink 08:39:15 <bwh> right 08:39:30 <patersonc> bwh: Is this basically the same view for Buster? Or will you use v4.19.y regardless? 08:40:38 <bwh> We want to use an LTS version 08:40:47 <patersonc> Okay 08:41:13 <szlin> there is very little information about 4.19 08:42:56 <szlin> any other points? 08:42:57 <szlin> 3 08:42:58 <szlin> 2 08:43:00 <szlin> 1 08:43:09 <szlin> #endmeeting