IRC logs for #cip for Thursday, 2019-12-12

*** masashi910 has joined #cip04:45
*** masashi910 has quit IRC06:52
*** masashi910 has joined #cip06:52
*** rajm has joined #cip07:01
*** helmut has quit IRC07:25
*** therisen has joined #cip07:43
wensyay, severe performance regression in 4.19.x on x86 :(07:44
wenssince 4.19.67 even, from
szlindoes it affect 4.19 only?07:49
wensnooooo, it's also in 4.14, 4.9, and 4.407:51
wensit severely affects fork performance if CONFIG_VMAP_STACK=y07:51
szlinhowa bout mainline?07:51
wensstill need to test CONFIG_VMAP_STACK=n, but based on the fix: it should be ok07:52
wensszlin: I would say yes, but PTT doesn't run mainline kernel :p07:52
wenssince the fix is still only in -next, mainline is likely to be affected07:53
szlinwens: I see, thanks. Maybe PTT can choose cip kernel in the future :p07:54
wensI'm using 4.19 for now, but we rarely update the kernel, as it's a pain to restart the machine07:55
wensanyway, that's my rant for the day08:12
*** fujita3 has joined #cip08:51
wenswell, CONFIG_VMAP_STACK=n only fixes part of it.08:59
*** pave1 has joined #cip08:59
*** pave1 has left #cip08:59
*** pave1 has joined #cip08:59
wensother parts of the kernel use vmalloc and still get hit. things such as tty08:59
masashi910#startmeeting CIP IRC weekly meeting09:00
brloggerMeeting started Thu Dec 12 09:00:02 2019 UTC and is due to finish in 60 minutes.  The chair is masashi910. Information about MeetBot at
brloggerUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.09:00
brloggerThe meeting name has been set to 'cip_irc_weekly_meeting'09:00
*** brlogger changes topic to " (Meeting topic: CIP IRC weekly meeting)"09:00
masashi910#topic rollcall09:00
*** brlogger changes topic to "rollcall (Meeting topic: CIP IRC weekly meeting)"09:00
masashi910please say hi if you're around09:00
masashi910#topic AI review09:00
*** brlogger changes topic to "AI review (Meeting topic: CIP IRC weekly meeting)"09:00
masashi9101. Test LTS (pre)releases directly - patersonc09:00
*** kazu has joined #cip09:00
*** suzuki98 has joined #cip09:00
patersoncFor all of my actions - no update09:01
masashi910ok, thanks09:01
masashi9102. Create a way/process to run LTP only for release tests - patersonc09:01
masashi910Chris-san, do you have updates?09:01
masashi910patersonc: ok, thanks09:02
masashi9103. Combine rootfilesystem with kselftest binary - Iwamatsu-san09:02
masashi910Quote from Iwamatsu-san: "No 4 does not update."09:02
masashi910Kudo's comment: "No 4 should be No 3". I will keep this AI as open09:02
masashi9104. Document a process on how to add tests to the CIP test setup - patersonc09:02
masashi910#topic kernel maintenance updates09:03
*** brlogger changes topic to "kernel maintenance updates (Meeting topic: CIP IRC weekly meeting)"09:03
masashi910== Quote from Iwamatsu-san ==09:03
masashi910I reviewed 4.4.202, and stable-rc on stable ML.09:03
masashi910I send some comments for 4.19.y and other.09:03
pave1I reviewed 4.19.88 and started reviewing
masashi910pave1: thanks!09:04
masashi910SZ-san, Wen-san, do you have updates?09:04
wensclassify-failed-patches now includes more patches from the stable mailing list09:04
wensfinal step would be to include all the stable review series09:05
wensalso, cip-kernel-sec has seen a few more CVEs09:05
wenswill do another round of updates for cip-kernel-sec tomorrow.09:05
bwhI have been working on kernel-sec, including reviewing wens's changes. Also reviewed some y2038 patches upstream.09:05
pave1Speaking of failed patches... I don't monitor that closely,09:06
pave1but if there are changes that should be backported, I guess09:06
pave1I can try to help...?09:06
szlinmasashi910: none from me this week.09:07
wenspave1: I saw at least some ext4 patches, though I need to take a closer look09:08
masashi910wens, bwh, szlin: thank you for your updates!09:09
masashi910any other updates?09:09
masashi910#topic Kernel testing09:09
*** brlogger changes topic to "Kernel testing (Meeting topic: CIP IRC weekly meeting)"09:09
masashi910patersonc: the floor is yours09:10
patersoncThis week I've seen a few build issues on the stable-rc branches for 4.4 and 4.19.09:10
patersoncTroublesome patches identified and reported to Greg.09:10
patersoncI've enabled arm, arm64 and x86 defconfig builds for the stable-rc testing.09:10
patersoncI've also been working on enabling the Siemens boards in our CI.09:10
patersoncBuilds are working okay, but haven't got the x86 board booting yet.09:10
patersoncIf anyone is x86 savvy can has a few minutes after the meeting I'd be glad for a few minutes of your time.09:10
pave1Savvy, I don't know, but I guess I can try to help.09:11
wensI don't know why odd dts patches get backported, while the related dtsi patches get missed :(09:11
patersoncThat's it from me. Any questions?09:11
patersoncwens: Indeed09:11
pave1And sorry for -rt still being on old .yml versions, we can talk about that later, too.09:11
patersoncpave1: Thanks09:11
pave1wens: Basically Sasha tries to backport _everything_ without checking closely if it is needed or not :-(.09:12
masashi910patersonc: thanks for working with Greg. By installing the fix, the issues were all solved?09:12
wenspave1: is there a way to tell his bot to not select certain patches (by file path?)09:12
szlinwens: many ppl did that via email...09:13
patersoncmasashi910: Issues resolved by dropping patches.09:13
wensszlin: I suppose that didn't work?09:13
masashi910patersonc: thanks!09:13
pave1wens: I believe we can take a trip to U.S.A. and take away his computer :-).09:13
*** suzuki98 has quit IRC09:13
*** kazu has quit IRC09:13
wenshe'll just get a new one09:14
pave1wens: I guess we can talk to Greg, explain that this causes problems for us,09:14
*** kazu has joined #cip09:14
pave1wens: and try to get him to do more reasonable patch selection.09:14
pave1wens: We are not only ones with this problems -- SuSE is not happy, either.09:14
patersoncI guess it's a problem for everyone, not just us09:14
wensfor my platform at least, dts/dtsi patches go hand in hand with driver patches09:15
patersoncIt just shows the value in us testing the stable release candidates09:15
pave1wens: I tried to talk to Greg in person, but I guess I did not get the message accross.09:15
*** suzuki15 has joined #cip09:15
pave1wens: If you want to exclude few .dts files, it should be enough to promise you'll do the stable marking for them.09:15
wensthanks, I'll discuss this with my co-maintainer and see what we can do09:15
pave1wens: That should actually be quite easy to do. (And does not include forced entry into buildings :-) )09:16
masashi910thanks, folks, for working on difficult issues...09:16
masashi910any other topics?09:17
masashi910#topic CIP Core09:17
*** brlogger changes topic to "CIP Core (Meeting topic: CIP IRC weekly meeting)"09:17
masashi910kazu: the floor is yours09:17
kazuI've started the review phase of the 1st package proposal given from Security WG (in cip-dev)09:17
kazuno feedback yet though09:17
kazui guess some discussions will be needed for the packages09:18
kazutemporally set the due date to 23rd, but please ask me if you want to extend09:18
kazuand sorry for that no technical updates for a few weeks09:19
kazuthat's all from me in this week09:19
masashi910kazu: thank you09:19
masashi910any other topics?09:19
fujita3kazu: You may considering to have a meeting09:19
kazufujita3 yes, sorry for long delay09:19
kazufujita3 i hope i can arrange in this year...09:20
fujita3kazu: Agreed :-)09:20
masashi910any other topics?09:21
masashi910#topic Software update09:21
*** brlogger changes topic to "Software update (Meeting topic: CIP IRC weekly meeting)"09:21
masashi910suzuki: the floor is yours09:21
suzuki15I have no update in this week, too. Sorry about that.09:22
suzuki15I'll put a license file into our WG's demo script repository in a couple of days.09:22
suzuki15That's it from me.09:22
masashi910Suzuki-san, that's ok. thanks for your report.09:22
masashi910any other topics?09:22
masashi910#topic AOB09:23
*** brlogger changes topic to "AOB (Meeting topic: CIP IRC weekly meeting)"09:23
masashi9101. Proposed move from Mailman to and kernel.org09:23
masashi910Does any one opposite to move mail server to
patersoncNope. assuming that it handles patches okay.09:24
masashi910patersonc: ok, thanks.09:24
kazumasashi910 i have no strong opinion about this if required features are supported :)09:25
patersoncmasashi910: Others may have an opinion though?09:25
masashi910patersonc: yes, so if there are objections, please speak up09:27
masashi910currently Neal-san is planning to move if there are no objections.09:27
szlinI think everyone is agree with it?09:29
masashi910szlin: thanks for your help.09:30
masashi910So, at this moment, let me assume that the kernel team does not have objections.09:30
masashi910ok, then let me move to the next09:30
masashi9102. When (Feb 27th-Thursday and 28th-Friday) do you prefer for F2F Kernel Team Meeting in Nurnberg during Embedded World 2020? - Doodle Poll09:30
masashi910for this topic, I created a doodle poll. I will send out the link to that by email after this IRC meeting.09:31
masashi910any other business?09:31
brloggerMeeting ended Thu Dec 12 09:32:16 2019 UTC.  Information about MeetBot at . (v 0.1.4)09:32
brloggerMinutes (text):
*** brlogger changes topic to "Civil Infrastructure Platform Project. Find the logs at"09:32
szlinmasashi910: thank you!!09:32
pave1thank you!09:32
fujita3thank you!09:32
masashi910thanks all!09:32
kazumasashi910 Thank you09:32
pave1wens: 4.19 regression...09:32
pave1wens: fortunately it is "only performance"...09:32
suzuki15thank you!09:33
wenspave1: I'll send out an email about this to the stable ML later09:33
*** suzuki15 has quit IRC09:33
bwhI notice 4.19 is also missing commit 9a62d20027da, which looks really bad for x86-32 (so I wonder why it wasn't found earlier)09:33
pave1wens: and unfortunately kpti is "new" (and forced to us by hw issues), so it might be hard.09:33
pave1wens: If you could cc me at
pave1bwh: I suspect I'm the last one running x86-32 :-).09:34
wenspave1: this server now runs with mitigations=off :)09:34
bwh$ git grep -l X86_32=y09:34
patersoncbwh: Currently we only build those configs. No testing09:35
pave1bwh: But that hardware is 64-bit capable, right?09:35
bwhI don't think so09:35
pave1bwh: aha, oops.09:35
pave1bwh: that may be some fun in future, because not many people are running x86-32 on next / recent mainline.09:36
pave1bwh: I do, but I suspect I'm the only one.09:37
pave1bwh: Testing mainline on our x86-32 boards might be good idea to keep ammount of regressions low.09:38
bwhMight be worthwhile, yes09:38
bwhThere are still Debian users that care about i386, but I think they mostly run stable09:38
* patersonc waves for x86 help...09:39
pave1bwh: Yes, I suspect quite a lot of users run x86-32 on various distribution kernels.09:39
pave1patersonc: Go ahead.09:39
patersoncAny suggestions as to what is going wrong here?
patersoncRFS isn't mounting09:39
patersoncIt's using siemens_ipc227e_defconfig09:40
patersoncI don't know if it's a LAVA issue or a config issue09:40
patersoncMentor have been using vmlinuz images, but it would make things a little easier for me if I can get bzImage working09:41
pave1Do you have CONFIG_NFS and CONFIG_NFS_ROOT enabled?09:41
pave1Let me take a look.09:41
bwhI don't see any "IP-Config" message09:41
patersoncHere's a healthy job:
patersoncUsing vmlinuz09:43
patersoncWhich has Ip-Config09:43
pave1It seems to have dhcp enabled.09:45
patersoncSorry, I need to afk for 15 mins09:47
pave1I don't see driver for network card.09:47
pave1I believe driver for network card needs to be enabled in kconfig.09:48
*** toscalix has joined #cip09:51
bwhYeah, either you need IP_PNP, IP_PNP_DHCP, ROOT_NFS, and the net driver built-in *or* you must use an initramfs with nfsroot support09:52
pave1Hmm. Bad Siemens.09:53
pave1 specifications09:53
pave1It explains that ethernet is 2x gigabit, but not what chip it uses.09:54
pave1bwh: Yeah. I guess for our testing building driver in is easier.09:54
pave1welcome :-)10:02
pave1See above.10:02
pave1.yml in -cip-rt: They did not make it, yet. They should make it during the next cycle.10:02
pave1Would it help testing if I merge them early?10:02
patersoncpave1: It doesn't really matter tbh10:03
patersoncFor now it can wait10:03
pave1patersonc: Ok, good, easier for me :-).10:03
patersoncSo it looks like the MAC isn't being set properly:
patersoncApparently this is an issue with the version of LAVA we're using10:05
patersoncBut somehow it doesn't affect things when using vmlinuz10:05
pave1Well, the MAC address - not sure what LAVA is trying to do there.10:06
pave1But you really need driver for the ethernet, and you don't seem to have one.10:06
pave1MAC may be secondary problem (or not).10:06
patersoncSo how is NFS mounting for the vmlinuz case?10:07
patersoncDoes vmlinuz have more things built in?10:07
pave1Do you have a boot log where it works ok10:07
pave1Thanks, let me take a look.10:08
pave1[    3.327006] igb 0000:02:00.0: eth0: (PCIe:2.5Gb/s:Width x1) 20:87:56:b6:21:d010:08
pave1Can you enable CONFIG_IGB in kconfig?10:10
pave1It is =m, but needs to be =y.10:10
patersoncAh that's another point10:12
patersoncI haven't included the Kernel modules in LAVA10:12
patersoncs|I|the scripts10:12
pave1You should not _need_ kernel modules for the boot test. And IGB will need to be built-in anyway due to nfsroot.10:14
*** kazu has quit IRC10:15
patersoncI'll try a local build with CONFIG_IGB10:16
*** toscalix has quit IRC10:19
pave1(See you, if the boot problem persists, we can discuss over email).10:32
patersoncThanks pavel10:33
pave1You are welcome :_)10:38
*** masashi910 has quit IRC10:38
*** rajm has quit IRC11:07
*** fujita3 has quit IRC11:12
*** rajm has joined #cip11:30
patersoncI'm trying to test, but the internet d/l speed is crazy slow :(11:30
*** pave1 has quit IRC11:39
*** masashi910 has joined #cip11:58
*** masashi910 has quit IRC13:14
*** tpollard has quit IRC13:18
*** tpollard has joined #cip13:19
*** tpollard has quit IRC14:00
*** helmut has joined #cip14:04
*** tpollard has joined #cip14:05
patersoncIncluding CONFIG_IGB seems to work:
patersoncbwh: Are you around?14:50
bwhpatersonc: yes14:50
patersoncbwh: I'm having a go at generating the .sources list for the cip-kernel-config repo14:53
patersoncHow can I do it for just one config?14:53
bwhYou should be able to give that filename as the target for make14:54
patersoncit looks like it tries to go to a directory ../kernel14:56
patersoncDo I just hack the Makefile according to my setup?14:56
bwhyou can put KSRC=... on the make command line14:56
patersoncDo I also need to set the crosscompiler and arch?14:58
patersoncI guess so15:00
patersoncLooks like it's compiling now15:01
patersoncThanks bwh15:01
bwhThe ARCH is automatically set based on the filename, and CROSS_COMPILE is set automatically based on the ARCH (but it might not guess right)15:04
patersoncAh okay15:05
*** therisen has quit IRC16:15
*** tpollard has quit IRC17:05
*** masashi910 has joined #cip22:32
*** rajm has quit IRC23:06
*** Guest68068 has quit IRC23:10
*** jward has joined #cip23:11
*** jward is now known as Guest356323:11

Generated by 2.15.3 by Marius Gedminas - find it at!