**** BEGIN LOGGING AT Tue Nov 12 02:59:57 2019
Nov 12 06:29:01 <yocti> New news from stackoverflow: Kernel panic caused when flashing images to an imx6 board <https://stackoverflow.com/questions/58812919/kernel-panic-caused-when-flashing-images-to-an-imx6-board>
Nov 12 07:41:06 <nrossi> JPEW: think i have a solution that works... manually encode and decode the subtest results like extraresults does. Give it a test and let me know https://github.com/nathanrossi/openembedded-core/commit/e5c3fdc610ed33ee67c118bf39be9a530364d858
Nov 12 07:42:42 <nrossi> RP: things have improved in python with regards to I/O threads and the GIL. Based on what I have seen in selftest there are a majority of tests that are just calling and waiting on bitbake. It might be worth looking into if you think that tests themselves are generally not cpu bound (e.g. not processing a lot within the python process)
Nov 12 07:43:27 <nrossi> RP: also it might be worth making a smarter bitbake instance object either way since some tests are reproducing the concurrency behaviour already
Nov 12 08:42:11 <RP> nrossi: Given the nature of the tests and the fact they all run different builds its not really as straight forward as cpu or io bound :/
Nov 12 08:57:26 <nrossi> RP: i see, always tricky to figure out that sort of performance difference ;)
Nov 12 09:20:45 <RP> nrossi: Parallel selftest took us from 10 hours of selftest down to around 2 hours so was a huge win (numbers not exact but you get the idea, was a huge win)
Nov 12 09:24:53 <nrossi> RP: oh of course I fully understand the benefit, just looking for ways to help improve it :)
Nov 12 09:31:16 <RP> nrossi: really you need one bitbake instance per build directory and we end up with multiple parallel build directories with selftest
Nov 12 09:38:16 <nrossi> RP: i don't think i was clear enough with what i meant by bitbake instance objects. Basically the object would control the build directory, configuration and the bitbake process, and instead of a test just calling  "bitbake(...)", it would probably call "self.bitbake(...)". The instances could be reusable, and you might have more than one per "thread", or they might be temporary as opposed to reusable
Nov 12 10:24:46 <yann> hello - can we easily restrict appending to a postfuncs, only on class-target ?  The "variable flag syntax" bitbake doc does not leave me much hope :)
Nov 12 10:27:08 <RP> nrossi: isn't that memory resident bitbake?
Nov 12 10:27:46 <RP> yann: do postfuncs = "${VAR}" and then VAR="" VAR_class-target = "XYZ"
Nov 12 10:28:04 <RP> yann: just have to be slightly creative
Nov 12 10:29:39 <yocti> New news from stackoverflow: Is it possible to replace a file from one repository to another with android repo tool? <https://stackoverflow.com/questions/50210834/is-it-possible-to-replace-a-file-from-one-repository-to-another-with-android-rep>
Nov 12 10:32:40 <nrossi> RP: no more just referring to avoid this sort of duplication -> http://git.openembedded.org/openembedded-core/tree/meta/lib/oeqa/core/utils/concurrencytest.py#n258 http://git.openembedded.org/openembedded-core/tree/meta/lib/oeqa/selftest/cases/signing.py#n39
Nov 12 10:38:53 <RP> nrossi: you mean separate build directories?
Nov 12 10:40:16 <RP> nrossi: selftest is slow as it runs builds. We can't speed the builds up, much as we'd love to so the only way to speed up is to run in parallel. To run in parallel you need multiple build directories :/
Nov 12 10:41:17 <RP> nrossi: unless you mean teaching bitbake to run multiple builds in paralell with multiple configurations like multiconfig does but on a larger scale. That would get "interesting" though...
Nov 12 10:48:19 <nrossi> RP: I am thinking less complex, just making selftest a little bit smarter with how it manages multiple build directories :)
Nov 12 10:50:58 <RP> nrossi: sorry, think I must missing something but I'm really not seeing it :/
Nov 12 10:51:46 <nrossi> RP: yer i don't think i am explaining myself properly :(
Nov 12 10:51:46 <RP> nrossi: ultiamtely I hope that if we can have some standard setup mechanism, a lot of the complexity goes away. Each test may have its own builddir, maybe.
Nov 12 10:52:40 <nrossi> RP: essentially the idea would be move the builddir/bitbake from concurrencytest into a class that can be used to manage a builddir and bitbake process
Nov 12 10:53:00 <nrossi> RP: then when a test goes to run it gets that object to use
Nov 12 10:54:33 <nrossi> RP: this would enable the ability to reuse and or have clean build dirs, but also make it easier to manage them so you could have say qemu be able to interrupt the build if qemu unexpectedly dies
Nov 12 11:18:54 <RP> nrossi: right, I think I understand now
Nov 12 11:19:13 <RP> nrossi: I've been holding off something like that until we think about the layer setup piece
Nov 12 11:20:08 <nrossi> RP: yer, i actually ran into bugs with it and how i have my build directory containing oe-core
Nov 12 11:20:13 <kanavin_> are the slides from the Lyon Yocto summit available somewhere?
Nov 12 11:20:23 <yann> thanks RP :)
Nov 12 11:24:17 <RP> kanavin_: they are, https://wiki.yoctoproject.org/wiki/YP_Summit_Lyon_2019
Nov 12 11:24:46 <RP> nrossi: I'm not surprised, that bit was "make it work" as the testing was valuable without fixing all the issues :/
Nov 12 11:24:51 <kanavin_> RP: thanks - I looked at the other pages, and none carries a link to this one
Nov 12 11:25:12 <RP> kanavin_: please do add one somewhere appropriate
Nov 12 11:25:28 <RP> nrossi: It needs to be solved with the wider layer setup piece :(
Nov 12 11:27:21 <nrossi> RP: makes sense, just throwing out ideas :) and happy to help where i can
Nov 12 11:29:50 <milloni> letothe2nd: did you ever find out why in one of your streams it was pulliing mesa etc as a dependency of strace?
Nov 12 11:30:17 <milloni> that looks really sketchy, strace definitely doesnt need the graphics stuff
Nov 12 11:47:26 <RP> nrossi: I appreciate that, I'm not trying to be negative. I really don't like to come over that way :/
Nov 12 11:49:34 <nrossi> RP: don't worry, i did not interpret it that way.
Nov 12 12:04:42 <kanavin_> RP: added to https://wiki.yoctoproject.org/wiki/Yocto_Project_Summit_2019
Nov 12 12:08:05 <RP> kanavin_: thanks
Nov 12 12:08:48 <kanavin_> RP: I can't edit https://www.yoctoproject.org/yocto-project-summit-2019-2/ though and not sure who to ask?
Nov 12 12:09:58 <RP> kanavin_: ndec can probably help?
Nov 12 12:10:25 <kanavin_> ndec, ^^^
Nov 12 12:10:34 <RP> ndec: context here is that people can't find the slides from the event so we need more links
Nov 12 12:15:51 <letothe2nd> milloni: nope, but i also didn't look into it.
Nov 12 12:18:35 <milloni> letothe2nd: looks like you've got a topic for the next stream then
Nov 12 12:18:44 <letothe2nd> milloni: nope :)
Nov 12 12:18:54 <milloni> hah
Nov 12 12:19:23 <letothe2nd> milloni: topic for today is already decided on. but actually it kinda fits what i might do in december, so... maybe
Nov 12 12:19:35 <milloni> what's the topic?
Nov 12 12:19:53 <letothe2nd> today will be building an image for the beaglebone black
Nov 12 12:20:02 <letothe2nd> and looking at the bsp + wic
Nov 12 12:21:58 <milloni> nice
Nov 12 12:23:21 <letothe2nd> :-)
Nov 12 13:30:07 <yocti> New news from stackoverflow: Bitbake Setting BBPATH: command not found <https://stackoverflow.com/questions/58819721/bitbake-setting-bbpath-command-not-found>
Nov 12 14:24:39 <JPEW> nrossi: Thanks for the patch. I'll give it a try. I have to fix some other issues I think, do you care if I squash that change into my original patch?
Nov 12 14:32:15 <yocti> New news from stackoverflow: update-rc.d on raspberrypi can not start service automatically on reboot <https://stackoverflow.com/questions/58820374/update-rc-d-on-raspberrypi-can-not-start-service-automatically-on-reboot>
Nov 12 14:45:08 <guerinoni> Hi guyz, I have an issue with serial ports on a iMX6ULL board, kernel rel_imx_4.19.35; when I try to echo to the serial I get those messages:
Nov 12 14:45:15 <guerinoni> ```
Nov 12 14:45:18 <guerinoni> `root@smart:~# while [ true ]; do echo -ne 'test' > /dev/ttymxc2; doneimx-sdma 20ec000.sdma: Timeout waiting for CH0 readyimx-uart 21ec000.serial: error in RX dma configuration.imx-sdma 20ec000.sdma: Timeout waiting for CH0 readyimx-uart 21ec000.serial: error in RX dma configuration.`
Nov 12 14:45:28 <guerinoni> Can someone help me? I'm in real trouble...
Nov 12 14:46:08 <ndec> RP: kanavin_ : i can fix that. I just need to edit the page and add "Slides are here: https://wiki.yoctoproject.org/wiki/YP_Summit_Lyon_2019"?
Nov 12 14:49:13 <zeddii> guerinoni .. that's going to be deep in the BSP for an issue. Best bet is to find a forum that is related to where you got the layer/BSP and ask there. folks here are unlikely to know the details.
Nov 12 14:49:29 <letothe2nd> guerinoni: this seems to be a driver issue, and maybe not even fixed upstream so far: https://www.spinics.net/lists/arm-kernel/msg716464.html
Nov 12 14:50:02 <letothe2nd> guerinoni: so, best chance is to ask whomever gave that board and its BSP to you
Nov 12 14:52:45 <jonmason> what's the best way to use a binary built in another recipe?  For example, I'm trying to use the fiptool build in atf to create a ddr phy binary
Nov 12 14:53:25 <letothe2nd> jonmason: DEPENDS="whatever-native"?
Nov 12 14:57:31 <jonmason> native pulls from the locally installed things, right?
Nov 12 14:57:36 <guerinoni> me and teammate are the developer of bsp and we don't touch the driver
Nov 12 14:57:58 * zeddii would be busy Streaming Disney+ today .. if only they had load tested their servers.
Nov 12 14:58:11 <jonmason> zeddii: Mandalorian?
Nov 12 14:58:11 <letothe2nd> jonmason: nope, -native suffix means that the suffixed recipe is built to run on the host
Nov 12 14:58:14 <zeddii> clearly not built with a minimal footprint and scalable architecture.
Nov 12 14:58:26 * zeddii suggests OE + meta-virt to them in his problem report :P
Nov 12 14:58:36 <zeddii> jonmason. yah! but alas, I have just a spinner.
Nov 12 14:59:07 <letothe2nd> guerinoni: then i guess its time to touch the driver, and read the datasheet of the SoC
Nov 12 14:59:28 <jonmason> zeddii: I should have a Yocto built image for the honeycomb lx2k today, once I get this fixed
Nov 12 14:59:45 <jonmason> their build scripts are nested hacks
Nov 12 14:59:46 <zeddii> cool. I'm waiting to hear!
Nov 12 15:00:02 <zeddii> I almost bought the pine64 book a few days ago .. still tempted.
Nov 12 15:00:11 <jonmason> right!  I want one too
Nov 12 15:00:31 <guerinoni> letothe2nd Are you talking about the serial driver or something else?
Nov 12 15:00:57 <zeddii> I could replace my rpi home automation with it at that price, and have a built in screen to debug, etc. easier than sshing to my random rpi's
Nov 12 15:01:07 <letothe2nd> guerinoni: hard to say. looks like some effect from serial and/or dma
Nov 12 15:01:20 <letothe2nd> guerinoni: in any case this is deep in the imx kernel land
Nov 12 15:01:40 <letothe2nd> guerinoni: best chance is probably #linux-imx
Nov 12 15:01:40 <jonmason> zeddii: this will make a sweet NAS.  Multiple 10G ehternet and 4 SATA ports
Nov 12 15:02:02 <guerinoni> letothe2ndmany thanks
Nov 12 15:02:08 <guerinoni> letothe2nd many thanks
Nov 12 15:02:14 <letothe2nd> guerinoni: have fun.
Nov 12 15:02:15 <jonmason> the fan noise sucks.  I've convined myself to wait for the reqorked boards (as the erratas are annoying)
Nov 12 15:02:49 * letothe2nd hereby announces "reqorked" is typo of the day!
Nov 12 15:02:50 <qschulz> guerinoni: if you read the thread in the mailing list there is apparently a HW errata for that issue
Nov 12 15:02:53 <zeddii> jonmason. yah. I'd like it for something like that as well. I have a single overloaded box in my basement doing NAS, Emby server and a bunch of things. would like to build out a few more dedicated bits.
Nov 12 15:03:20 <ndec> RP: kanavin_ : made an update. let me know if you need more than that.
Nov 12 15:03:39 <jonmason> zeddii: step 1, get this working with hacked up recipes.  step 2, get it working on upstream everything. step 3, profit
Nov 12 15:03:59 * letothe2nd heads off to pub, prepare for live drinking... erm, streaming.
Nov 12 15:04:07 <jonmason> letothe2nd: good luck!
Nov 12 15:04:08 <zeddii> jonmason: 4) build a reference binary package feed to keep it up to date. ;)
Nov 12 15:04:08 <guerinoni> qschulz can you provide that link?
Nov 12 15:04:20 <fray> ohh I better get onto twitch then..
Nov 12 15:04:24 <jonmason> zeddii: exactly!
Nov 12 15:04:30 <qschulz> guerinoni: what letothe2nd sent earlier
Nov 12 15:04:44 <qschulz> https://www.spinics.net/lists/arm-kernel/msg716464.html
Nov 12 15:05:07 <qschulz> specifically https://www.spinics.net/lists/arm-kernel/msg716582.html
Nov 12 15:07:51 <guerinoni> I'll try with this patch
Nov 12 15:08:43 <qschulz> (knock on #linux-imx's door though. Worth talking to some people before pulling too many of your hair)
Nov 12 15:19:10 <RP> ndec: thanks
Nov 12 15:29:04 <Dave_Elec> Have anyone built yocto on arch?
Nov 12 15:29:26 <Dave_Elec> Poky?
Nov 12 15:30:28 <kergoth> ye
Nov 12 15:30:32 <kergoth> yes, even
Nov 12 15:30:46 <Dave_Elec> I have issues with gnulib
Nov 12 15:30:48 <Dave_Elec> Please port gnulib freadahead.c to your platform! Look at the definition of fflush, fread, ungetc on your system, then report this to bug-gnulib.
Nov 12 15:34:52 <Dave_Elec> kergoth: Had this issue before?
Nov 12 15:35:10 <kergoth> nope, cant' say that i have. been a hwile since i built on arch, though, could be a recent change
Nov 12 15:40:27 <Dave_Elec> I was trying to build from branch morty.. As the tutorial I was following suggested. I am a complete noob to yocto project.
Nov 12 15:40:38 <Dave_Elec> I am trying zeus now
Nov 12 15:47:47 <RP> Dave_Elec: right, use something modern, arch will need that. Morty is old
Nov 12 15:51:00 <letothe2nd> live session on twitch will be starting in 10 minutes :)
Nov 12 15:56:33 <qschulz> letothe2nd: give the link! :) https://www.twitch.tv/yocto_project/
Nov 12 15:56:55 <letothe2nd> qschulz: :)
Nov 12 16:02:07 <armpit> YPTM: armin is on
Nov 12 16:11:34 <letothe2nd> fray: stream is busted, it seems
Nov 12 16:17:00 <fray> I'm not getting drops here.. so it might be others or a local geo problem
Nov 12 16:18:56 <milloni> works for me
Nov 12 16:19:45 <fray> I'm watching at 480p BTW.. twitch itself might be having issues..
Nov 12 16:20:54 <fray> I just lagged..
Nov 12 16:30:57 <armpit> YPTM: is over
Nov 12 16:32:40 <yocti> New news from stackoverflow: Yocto: Nothing provides python-bsddb needed by python-db-2.7.16-r1.cortexa7t2hf-neon <https://stackoverflow.com/questions/58822557/yocto-nothing-provides-python-bsddb-needed-by-python-db-2-7-16-r1-cortexa7t2hf>
Nov 12 16:43:06 <letothe2nd> fray: *sigh*
Nov 12 16:43:21 <Crofton|work> shit happens
Nov 12 16:43:40 <letothe2nd> i walked into the usual pub, asked if i can sit in my usual corner, and guy was like "uhmno, we're closed!"
Nov 12 16:44:09 <Crofton|work> weird
Nov 12 16:44:14 <Crofton|work> how un German
Nov 12 16:44:37 <letothe2nd> actually i like this place more (quieter), but the wifi seems to have the usual "hotel security enhancements"
Nov 12 16:45:25 <Dave_Elec> I tried zeus on arch.. now I got sdl issues
Nov 12 16:45:56 <letothe2nd> any opinion on the lightning sessions?
Nov 12 16:46:05 <Dave_Elec> cant compile recipes-graphics/libsdl2/libsdl2_2.0.10.bb:do_compile
Nov 12 16:46:18 <letothe2nd> Dave_Elec: i'm on arch too, and the usual simplest trick to just use a docker container
Nov 12 16:47:00 <fray> we just need to get you a 4G/5G access point to do the streaming.. :)
Nov 12 16:47:34 <letothe2nd> fray: i know its un-twitch-y, to just go on, stream like 5 or 10 minutes and then go off again. but i feel it might serve like short recordings that explain just one thing well
Nov 12 16:48:11 <Dave_Elec> letothe2nd: It's okay for me
Nov 12 16:48:19 <fray> na if it's just for the recording that is cool.. audience you need a 10 minute "before" to get people to join
Nov 12 16:49:05 <fray> BTW not sur eif you saw, but the channel is up to almost 900 views.. that's really good for a channel like this
Nov 12 16:49:51 <letothe2nd> fray: actually i have LTE built into the laptop, but didn't book a data rate for today, though wifi would be fine
Nov 12 16:50:10 <letothe2nd> so yeah i've got options, but didn't know i'd have to use any today
Nov 12 16:50:14 <fray> no worries.. streaming is rarely ever perfect.
Nov 12 16:50:22 <letothe2nd> fray: yeah and 66 followers :)
Nov 12 16:51:01 <fray> honestly it can take streamers well over a year to get past about 10 followers and 100 views
Nov 12 16:51:34 <Dave_Elec> letothe2nd: I'm not sure but I think asciinema has live streaming for terminals.. Hacky but streaming audio with other thing and terminal with asciinema..
Nov 12 16:51:58 <letothe2nd> fray: i guess the point is that for this highly specific topic, we have a small audience in comparison to the usual streamers, but on the other hand no competition too.
Nov 12 16:52:02 <Dave_Elec> That's not comfortable but..
Nov 12 16:52:44 <fray> most of the 'live' coding things I've seen are evenings in North America..
Nov 12 16:52:54 <fray> so this is a good time to do it.. as you said, little to no competition.. :)
Nov 12 16:53:16 <letothe2nd> Dave_Elec: interesting idea, but i actually like a lot to "talk with my hands and gestures"
Nov 12 16:53:49 <Dave_Elec> Sure
Nov 12 16:54:40 <letothe2nd> fray: and the topic is sufficiently "professional" people can watch it on the clock :)
Nov 12 16:54:46 <fray> yup
Nov 12 16:55:45 <letothe2nd> so yeah, i'm quite happy with the results so far. a lot of folks stepped up to me in lyon
Nov 12 16:57:04 <letothe2nd> anyways. off for "quality time" and licking my wounds :(
Nov 12 16:58:03 <fray> have a good evening
Nov 12 17:04:48 <zeddii> that was wierd. gmail classified half of a thread about bare metal application building as spam. I could only see Ross and PaulB's answers but not the question.
Nov 12 17:05:22 <fray> ha
Nov 12 17:05:34 <zeddii> which triggered me to dig it out of spam.
Nov 12 17:10:03 <paulbarker> Maybe gmail just doesn't like mail from yahoo.com
Nov 12 19:07:01 <tvm> what is proper way to enable coredumps on yocto based OS ?
Nov 12 19:46:00 <letothe2nd> in case anybody is still around from the fiasco earlier: i'll do an out-of-schedule session this friday on the same topic.
Nov 12 19:47:30 <letothe2nd> fray: ^^^^
Nov 12 20:12:18 <fray> letothe2nd thanks.. I should be able to be there
Nov 12 20:14:06 <letothe2nd> fray: :)
Nov 12 20:19:25 <khem> tvm: depends if you use systemd or sysvinit or other init system
Nov 12 20:21:59 <tvm> khem: this is systemd, i have already enabled DumpCore and DefaultLimitCore in /etc/systemd/systemd.conf, not sure if that's enough
Nov 12 20:30:05 <khem> tvm: you will need to add coredump to PACKAGECONFIG for systemd recipe as well
Nov 12 20:44:47 <tvm> khem: thanks
Nov 12 20:51:01 <khem> tvm: I hope you are building your image
Nov 12 21:03:59 <tvm> khem: yes
Nov 12 21:09:17 <khem> ok then you can control it fine
Nov 12 21:16:35 <tvm> hopefully strace will give me some hints before i resort to that though
Nov 12 21:16:43 <tvm> but thanks, this might be next step
Nov 12 21:58:30 <tgamblin> Is there a way to change the busybox config only if a particular ptest is to be built into an image?
Nov 12 21:59:19 <paulbarker> tgamblin: Not really, it's not known which image busybox will be installed into when it is built
Nov 12 22:01:12 <tgamblin> paulbarker: Alright, thanks
**** ENDING LOGGING AT Wed Nov 13 02:59:58 2019