Friday, 2018-08-31

armpitsend a patch ?00:11
armpitotavio, did you want cmake update in 2.6 ?00:11
*** jkridner|pd <jkridner|pd!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto00:31
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has quit IRC00:35
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto00:36
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has joined #yocto00:36
*** jkridner_ <jkridner_!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto00:38
*** jkridner|pd <jkridner|pd!~jkridner@pdpc/supporter/active/jkridner> has quit IRC00:40
*** jkridner_ <jkridner_!~jkridner@pdpc/supporter/active/jkridner> has quit IRC00:40
*** jkridner|pd <jkridner|pd!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto00:40
*** nighty- <nighty-!> has joined #yocto00:41
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has quit IRC00:42
khemotavio: cmake upgrade could be a bomb so late in release cycle00:58
khembut I guess we can give it a shot00:58
*** blueness <blueness!~blueness@gentoo/developer/blueness> has quit IRC01:02
*** blueness <blueness!~blueness@gentoo/developer/blueness> has joined #yocto01:03
*** tgraydon <tgraydon!textual@nat/intel/x-vpwmmhdtihjoqzpv> has quit IRC01:19
armpitkhem, I won't use that work, the NAS is listening ; )01:43
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has quit IRC01:45
*** yizhao <yizhao!~zhaoyi@> has quit IRC01:46
*** yizhao <yizhao!~zhaoyi@> has joined #yocto01:51
*** jkridner|pd <jkridner|pd!~jkridner@pdpc/supporter/active/jkridner> has quit IRC02:03
*** jacques <jacques!~jacques@nslu2-linux/jacques> has quit IRC02:05
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto02:14
*** chandana73 <chandana73!~ckalluri@> has joined #yocto02:20
*** dc13ff <dc13ff!uid190567@gateway/web/> has quit IRC02:26
*** sarvesh <sarvesh!~sarvesh@> has joined #yocto02:31
*** sarvesh <sarvesh!~sarvesh@> has left #yocto02:36
*** peniwize <peniwize!~peniwize@> has quit IRC02:40
*** arielmr <arielmr!> has quit IRC02:54
*** chandana73 <chandana73!~ckalluri@> has quit IRC03:27
*** comptroller <comptroller!> has quit IRC04:16
*** comptroller <comptroller!> has joined #yocto04:26
*** JaMa <JaMa!~martin@> has joined #yocto04:38
*** morphis <morphis!> has joined #yocto05:01
*** thaytan <thaytan!> has quit IRC05:28
*** thaytan <thaytan!> has joined #yocto05:28
*** volestorm <volestorm!~volestorm@2400:8902::f03c:91ff:fe7f:f462> has joined #yocto05:33
*** stephano <stephano!stephano@nat/intel/x-ityxhfttbnuedzzs> has quit IRC05:34
*** gtristan <gtristan!~tristanva@> has joined #yocto05:41
*** joseppc <joseppc!~josep@linaro/joseppc> has joined #yocto06:01
*** AbleBacon <AbleBacon!~AbleBacon@unaffiliated/ablebacon> has quit IRC06:02
*** AbleBacon <AbleBacon!~AbleBacon@unaffiliated/ablebacon> has joined #yocto06:03
*** rob_w <rob_w!~rob@unaffiliated/rob-w/x-1112029> has joined #yocto06:10
*** yizhao <yizhao!~zhaoyi@> has quit IRC06:10
*** rob_w <rob_w!~rob@unaffiliated/rob-w/x-1112029> has quit IRC06:10
*** rob_w <rob_w!~rob@unaffiliated/rob-w/x-1112029> has joined #yocto06:10
*** pohly <pohly!> has joined #yocto06:12
*** yizhao <yizhao!~zhaoyi@> has joined #yocto06:15
*** yizhao <yizhao!~zhaoyi@> has quit IRC06:20
*** Bunio_FH <Bunio_FH!> has joined #yocto06:25
*** Bunio_FH1 <Bunio_FH1!> has joined #yocto06:28
*** yizhao <yizhao!~zhaoyi@> has joined #yocto06:28
*** Bunio_FH <Bunio_FH!> has quit IRC06:29
*** yizhao <yizhao!~zhaoyi@> has quit IRC06:43
*** yizhao <yizhao!~zhaoyi@> has joined #yocto06:53
*** Phreaknes <Phreaknes!~Phreaknes@> has joined #yocto06:56
PhreaknesHi all06:56
PhreaknesI'm going to take the dive and play around with linux this weekend, Can I run Ubuntu Linux inside Windows 10 or is it required to run Ubuntu on a dedicated box? I plan on playing with yocto, and qt automotive06:56
*** yann <yann!> has quit IRC07:00
*** eduardas_m <eduardas_m!~eduardas@> has joined #yocto07:00
khemPhreaknes: you can run ubuntu in a VM on Win1007:02
khemvirtual box is commonly used see
Phreaknes@ Khem I take it thats the preferred route07:03
khemI usually resort to virtual box, it runs on Mac and windows as well as on Linux07:05
*** lusus <lusus!~lusus@> has joined #yocto07:08
*** varjag <varjag!> has joined #yocto07:17
*** kanavin_ <kanavin_!~kanavin@> has joined #yocto07:31
*** kanavin <kanavin!~kanavin@> has quit IRC07:32
*** mckoan|away is now known as mckoan07:53
*** gtristan <gtristan!~tristanva@> has quit IRC07:54
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has joined #yocto08:05
*** florian <florian!~florian_k@Maemo/community/contributor/florian> has joined #yocto08:13
eduardas_mhello, has anyone of you ever successfully integrated USB-MTP functionality into their Yocto build (without any GPLv3/LGPLv3) ?08:19
yoctiNew news from stackoverflow: In devtool command line tool is finish subcommand removed or replaced? <>08:29
*** dv_ <dv_!> has quit IRC08:36
*** cslcm <cslcm!> has joined #yocto08:38
*** gtristan <gtristan!~tristanva@> has joined #yocto08:39
cslcmHi all - i've updated a layer but Yocto doesn't seem to have noticed and isn't pulling in the new recipes when I bake an image. How can I tell Yocto to rebuild the affected recipes?08:39
eduardas_mcslcm: well, I personally sometimes do a -c cleanall of a recipe, though that requires a complete rebuild afterwards, so be warned... however, did you check that your new recipes actually build separately via bitbake?08:47
eduardas_mcslcm: I mean does bitbake <new-recipe-name> work?08:48
eduardas_minstead of building the final image08:48
eduardas_mif not, then bitbake can not locate the recipe files08:48
eduardas_mcheck your layer config where it looks for recipes08:49
*** dv_ <dv_!~dv@> has joined #yocto08:50
*** yann <yann!~yann@> has joined #yocto08:50
*** T_UNIX <T_UNIX!uid218288@gateway/web/> has joined #yocto08:50
eduardas_mcslcm: something like a BBFILES += "${LAYERDIR}/recipes-*/*/*.bb in your layer layer.conf will restrict how recipe .bb files are found08:51
eduardas_mrecipe paths need to match the expression08:51
*** JaMa <JaMa!~martin@> has quit IRC08:54
nayfeHi, just in case, I'm trying to build two dependant projects with yocto sdk, these depend on gflags and it seems that gflags-config.cmake set gflags library name to gflags_shared and yocto gflags installs instead of, any idea?08:56
cslcmeduardas_m: it's a third party layer, i've just pulled the latest version08:57
cslcmbitbake <recipe name> does work, but it provides the cached version08:58
cslcmi don't really need to do a cleanall, surely? I thought this is what yocto is supposed to avoid?08:59
eduardas_mcslcm: did you try a rebuild after a -c cleanall for the recipe?08:59
cslcmthat will take 13 hours :(08:59
eduardas_mcslcm: even Qt base does not take that long... makes me wonder what that component is09:00
cslcmoh wait, just for the single recipe..09:00
eduardas_mat least not on an i7 Skylake09:00
cslcmmy bad09:00
eduardas_myeah, single recipe, not whole image09:01
cslcmso the command is "bitbake -c cleanall <recipe name>"?09:01
*** egavin <egavin!> has joined #yocto09:01
eduardas_mbitbake <recipe-name> -c cleanall09:01
bluelightningwhy -c cleanall? that also deletes the fetched source, I would imagine that's over the top09:01
eduardas_mbluelightning: yes, but I am extremely paranoid09:02
cslcmwell I want it do to that because the source has changed09:02
eduardas_mbluelightning: so that is what I ended up practicing09:02
bluelightningcslcm: how has it changed though?09:03
mckoanbluelightning: wouldn't be better -c cleansstate instead?09:03
cslcmheh, after a cleanall it stil thinks the package is the old version09:03
cslcmit obviously hasn't rescanned the layer09:03
eduardas_mcslcm: I suspect layer priorities then09:03
*** kaspter <kaspter!~Instantbi@> has quit IRC09:03
bluelightningcslcm: double-check that it's looking where you think it is and the recipe you think should be preferred is actually selected09:04
bluelightningbitbake -s or bitbake-layers show-recipes will indicate that09:04
bluelightningbrb switching networks09:05
cslcmit says -  nodejs:7.10.0-r1.4 , which isn't part of the layer any more -
*** kaspter <kaspter!~Instantbi@> has joined #yocto09:06
eduardas_mcslcm: I am not sure if it is an alternate recipe, a .bbappend or something else... if there are two recipes with exactly identical names, I believe layer priority will determine what is used09:06
cslcmthe layer has 7.10.1 at least09:06
eduardas_mcslcm: what version is getting built in your deploy directory then?09:07
cslcm7.10.0-r1.4, the old version09:07
*** bluelightning <bluelightning!~paul@pdpc/supporter/professional/bluelightning> has quit IRC09:08
*** bluelightning_ <bluelightning_!~paul@pdpc/supporter/professional/bluelightning> has joined #yocto09:08
*** bluelightning_ is now known as bluelightning09:08
*** gtristan <gtristan!~tristanva@> has quit IRC09:08
eduardas_mcslcm: how do you include the component into your image recipe? under what name?09:08
eduardas_mI believe one can specify the exact version, though I have never tried it09:09
cslcmCORE_IMAGE_EXTRA_INSTALL += "nodejs nodejs-npm"  in local.conf09:09
eduardas_mseems then that it completely ignores all the nodejs recipes in the layer09:09
eduardas_mcslcm: did you add the layer to your bblayers.conf?09:10
eduardas_mcslcm: in your build directory09:10
cslcmyes, i'm wondering if it might "help" if i renamed the layer09:10
cslcmmight defeat the cache09:10
bluelightningI very much doubt this is a cache issue09:11
bluelightninghave you tried the things I suggested?09:11
cslcm<cslcm> it says -  nodejs:7.10.0-r1.4 , which isn't part of the layer any more -
bluelightningok so what does bitbake-layers show-recipes nodejs say?09:12
cslcm(is there a preferred pastebin?)09:13
cslcm(i'm trying to avoid the 8.11.3 version in meta-oe btw)09:15
bluelightningcslcm: ah right, could you re-run it with -f ?09:15
cslcmah ffs i was on the wrong branch09:19
cslcmin the layer repo09:19
cslcmSorry for wasting your time09:19
bluelightningcslcm: no worries, I actually prefer this to finding a new bug ;)09:19
cslcmI'm just glad I didn't just give up and clean the whole image.. cos it would've still been broken 13 hours later09:20
cslcmthanks @:)09:20
*** nighty- <nighty-!> has quit IRC09:36
*** AbleBacon <AbleBacon!~AbleBacon@unaffiliated/ablebacon> has quit IRC09:37
*** aratiu <aratiu!~adi@> has quit IRC09:37
*** gtristan <gtristan!~tristanva@> has joined #yocto09:40
*** aratiu <aratiu!~adi@> has joined #yocto09:40
*** AbleBacon <AbleBacon!~AbleBacon@unaffiliated/ablebacon> has joined #yocto09:42
*** aratiu <aratiu!~adi@> has quit IRC09:46
*** rburton <rburton!> has joined #yocto09:52
*** morphis <morphis!> has quit IRC09:58
*** morphis <morphis!> has joined #yocto10:02
*** aratiu <aratiu!~adi@> has joined #yocto10:08
*** bluelightning_ <bluelightning_!~paul@pdpc/supporter/professional/bluelightning> has joined #yocto10:18
*** bluelightning <bluelightning!~paul@pdpc/supporter/professional/bluelightning> has quit IRC10:19
*** nighty- <nighty-!> has joined #yocto10:20
*** joaocfernandes <joaocfernandes!~Joao@> has joined #yocto10:25
*** aratiu <aratiu!~adi@> has quit IRC10:26
*** aratiu <aratiu!~adi@> has joined #yocto10:43
*** aratiu <aratiu!~adi@> has quit IRC10:44
*** aratiu <aratiu!~adi@> has joined #yocto10:48
*** BubuIIC <BubuIIC!bubuiicmat@gateway/shell/> has quit IRC10:52
*** bachp <bachp!bachpmatri@gateway/shell/> has quit IRC10:52
*** ant_home <ant_home!> has joined #yocto11:00
*** aratiu <aratiu!~adi@> has quit IRC11:00
*** kanavin__ <kanavin__!~kanavin@> has joined #yocto11:04
*** kanavin <kanavin!~kanavin@> has joined #yocto11:07
*** kanavin_ <kanavin_!~kanavin@> has quit IRC11:07
*** kanavin__ <kanavin__!~kanavin@> has quit IRC11:10
*** aratiu <aratiu!~adi@> has joined #yocto11:12
*** aratiu <aratiu!~adi@> has quit IRC11:20
*** aratiu <aratiu!~adi@> has joined #yocto11:23
RPHmm, why would core-image-sato work with testimage and core-image-sato-sdk hang11:30
RPrburton: seems the hang is -sdk image specific :/11:30
rburtonimage too big to boot?11:35
rburton(messed up image)11:36
*** xtron <xtron!~mentor@> has joined #yocto11:40
*** aratiu <aratiu!~adi@> has quit IRC11:40
RPrburton: 1.2G so should be ok11:54
RPzeddii, zeddii_home: around?11:54
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has quit IRC12:08
RP4.14 doesn't build perf either :/12:09
*** kanavin_ <kanavin_!~kanavin@> has joined #yocto12:18
*** kanavin <kanavin!~kanavin@> has quit IRC12:20
*** aratiu <aratiu!~adi@> has joined #yocto12:21
*** bluelightning_ <bluelightning_!~paul@pdpc/supporter/professional/bluelightning> has quit IRC12:22
*** gtristan <gtristan!~tristanva@> has quit IRC12:28
*** aratiu <aratiu!~adi@> has quit IRC12:31
*** yann <yann!~yann@> has quit IRC12:31
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has joined #yocto12:36
jofrIf I have a *.bbappend that contains e.g. a SRC_URI_append.. Is it possible to have one SRC_URI_append for one MACHINE and another SRC_URI_append for another? Is it perhaps just simply SRC_URI_append_machinename = "..."?12:44
zeddiiRP: just in now12:46
rburtonjofr: yes12:47
jofrWhat about if I would like an image to have a particular SRC_URI_append (no matter what MACHINE I'm building the particular image for)?12:48
RPzeddii: We're seeing hangs booting core-image-sato-sdk for qemuarm64 but not core-image-sato with 4.18. I tried 4.14 but perf doesn't build for qemuarm12:48
*** aratiu <aratiu!~adi@> has joined #yocto12:49
RPzeddii: I'm trying to narrow it down but struggling, I took perf/lttng/kernel-devsrc out the image and it still doesn't boot12:49
zeddiiand the qemuarm64 sato-sdk - (perf/lttng/kerneldevsrc) is master. perf is not building on 4.14 qemuarm6412:50
zeddiiI can start a build here.12:50
RPzeddii: I'm building master-next which has your patches in and a lot of other stuff12:50
jofrrburton: Does my question make sense?12:50
RPzeddii: news just in, a manual bisect of the image features suggests its something in "tools-testapps debug-tweaks ssh-server-openssh"12:50
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto12:51
rburtonjofr: that doesn't make sense: images don't have SRC_URI12:51
jofrrburton: No, I mean for the *.bbappend. So let's say I'm building my "foo" image .. I'd like it to use a particular SRC_URI_append for all machines, say, QEMU as well as the actual hardware. But if I'm building the "bar" image, it should have another SRC_URI_append12:53
rburtonjofr: impossible12:53
RPzeddii: its looking very like adding ssh-server-openssh to the image breaks the boot. It looked like a kernel hang but its probably early init :/12:54
RPzeddii: in which case its likely the ssl 1.1 changes in -next and not your kernel...12:54
jofrrburton: Ok. Not even by defining some sort of global variable in the machine definition that's readable by the *.bbappend?12:54
RPkanavin_: looks like we may have an openssl issue :/12:54
jofrrburton: I should probably just explain exactly what it is that I'm doing.. instead of trying to be generic..12:55
kanavin_RP: what issue specifically?12:55
rburtonjofr: you literally can't have a package be built differently depending on what image it goes into12:55
*** davenporten_ <davenporten_!8b55c117@gateway/web/freenode/ip.> has joined #yocto12:55
rburtonrecipes make packages, images are build from packages.  those packages are built once.12:55
davenporten_Hey, first time on here: how do you get bitbake to just make the rootfs? Thanks12:56
kanavin_davenporten_: bitbake -c do_rootfs <image:12:56
jofrrburton: So I'd have to solve that by creating another set of package which are image-dependent and include (IMAGE_INSTALL) the appropriate one for each image?12:56
jofrrburton: set of packages*12:56
davenporten_Thanks @kanavin, I appreciate it12:57
RPkanavin_: adding openssh instead of dropbear to core-image-sato causes it to hang at boot in testimage for qemuarm6412:57
kanavin_RP: I strongly suspect libressl :(12:57
kanavin_RP: openssh is reconfigured to use that12:57
rburtonjofr: pretty much.  like a config-debug package and config-production package, put the right one in each image12:57
RPkanavin_: still confirming that reverting the ssl changes fixes this. I'd agree that is suspense12:57
RPer, suspect12:57
*** BubuIIC <BubuIIC!bubuiicmat@gateway/shell/> has joined #yocto12:57
kanavin_libressl isn't used much in linux, but I picked that out of several not-great options12:58
kanavin_RP: I am not gonna send the dnf stack update in this cycle, they rewrote it in c++ and that's just asking for headache from all of us :)12:59
RPkanavin_: yes, too late for that I think! :)12:59
* zeddii reads12:59
rburtonkanavin_: no more py?12:59
zeddiiRP: ok. I'll start with 4.14  and fix perf, and then flip over to poking at boot as well.13:00
davenporten_kanavin: would it generate a .rootfs.cpio? Or something else?13:00
rburtondavenporten_: what do you actually want?13:00
jofrrburton: What I'm doing is installing pure-ftpd to my image. I currently have a pure-ftpd_%.bbappend in order to install the .pdb FTP-user-database file and an init-script, etc. Which would work fine if I wanted to have a single generic (virtual) FTP-user on all my images. But if I wanted to have a specific user-database for each image, I'd have to create multiple packages and install the appropriate one for each image, like I mentioned earlier?  :)13:00
RPzeddii: thanks. Obviously just trying to narrow down which patches are causing which issues...13:00
RPkanavin_: debian is still using openssl 1.0 for openssh on arm64?13:01
davenporten_rburton: I just want an archive or some kind that I could extract the rootfs from based off some build. If that makes sense13:01
kanavin_rburton: I didn't fully investigate, but at least libdnf is now c++13:01
rburtondavenporten_: set IMAGE_FSTYPES to "tar.gz" and it will make a tar.gz if you do bitbake imagename13:01
davenporten_rburton: ok, great, thanks a lot13:02
kanavin_rburton: is no more. It has ceased to be. It's an ex-git.13:02
kanavin_patch is coming... :)13:02
rburtoncheers :)13:02
kanavin_I actually wrote that into a commit13:03
rburtongood ;)13:03
*** TurBoss <TurBoss!turbossmat@gateway/shell/> has joined #yocto13:04
*** bachp <bachp!bachpmatri@gateway/shell/> has joined #yocto13:04
RPkanavin_: I have a patch13:05
* RP needs to post that13:05
kanavin_I have a patch too!13:06
* rburton feels left out13:06
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has quit IRC13:07
RPkanavin_: interesting to see if we match :)13:07
RPkanavin_: your commit message is better13:07
rburtonkanavin_'s has the best commit13:07
rburtonRP: just prepping a pango upgrade, there's a cve to sort13:08
RPkanavin_: merged your commit message to my patch13:09
* kanavin_ is back with yet another patchbomb
kanavin_(wip, look but don't touch)13:09
kanavin_RP: cheers :=)13:09
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto13:10
* RP is ignoring more patches until the issues with -next are resolved13:11
RPkanavin_: any idea what to do about the aarch64 ressl issue?13:11
RPndec: is this something linaro are aware of?13:11
kanavin_RP: drop the patch that sets openssh to use libressl instead of openssl1013:12
kanavin_(optionally drop libressl introduction patch as well)13:12
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has quit IRC13:15
RPkanavin_: right, that would be a good test13:15
kanavin_RP: was that confirmed by the AB?13:15
RPkanavin_: the breakage was spotted by the AB, I'm confirming locally13:19
kanavin_and x86 etc are fine?13:20
RPkanavin_: so it would seem13:20
kanavin_I thought openbsd folks would pay more attention to portability...13:20
kanavin_RP: there might also be a possibility that 7.7p1->7.8p1 update is to blame13:21
kanavin_because we actually ran the openssl 1.1 patchset before, and it was all green13:22
kanavin_rburton: blast from the past, did openedhand care about Asus EEEpc at some point some 10 years ago?13:24
RPkanavin_: also true. I will try and isolate it13:24
RPkanavin_: yes!13:24
RPkanavin_: well, the people at Intel from OH did13:25
kanavin_RP: eee-acpi-scripts will be removed by the remover-in-chief :)13:25
RPkanavin_: I'm not surprised ;-)13:26
RPThe gconv data for arm and x86 looks similar enough that it should give similar optimisation results13:27
RPfor this python pgo issue13:27
*** jkridner <jkridner!~jkridner@pdpc/supporter/active/jkridner> has joined #yocto13:28
davenporten_Thanks for all your help everyone, see you later13:28
*** davenporten_ <davenporten_!8b55c117@gateway/web/freenode/ip.> has quit IRC13:28
yoctiNew news from stackoverflow: Cross compiling scipy for open-embedded <>13:31
RPkanavin_: the openssh version upgrade alone seems to work, continuing to narrow down...13:32
kanavin_RP: I was hoping it would be that :)13:33
RPkanavin_: definitively bisected down to the ressl change13:40
RPzeddii: its definitely the libressl change, kernel is ok, sorry for the false alarm13:41
kanavin_RP: :-( then I guess we stick with openssl 1.0 for openssh for now? not an ideal situation13:41
RPkanavin_: well, we can move over except for openssh13:42
zeddiiRP: no worries. I'm fixing perf on 4.14 anyway, so a good thing to have built.13:42
kanavin_RP: yes, but openssh is kind of important :) btw, there's no point raising this with them, it's been discussed to death, with all parties ending up where they started13:42
RPkanavin_: right, someone will need to look into what is breaking in openssh/libressl13:43
RPkanavin_: I'd rather move over the pieces we can than not do anything at all13:43
kanavin_RP: so it's only on aarch64? I can try to look into it, the workstation should be ready any day now13:44
RPkanavin_: yes, only aarch6413:45
RPI have another bug to track down from next, no clue what that one is from yet (in a selftest)13:45
RPkanavin_: error is "Postinstall scriptlets of ['busybox'] have failed. If the intention is to defer them to first boot, then please place them into pkg_postinst_ontarget_${PN} ()" :/13:47
RPkanavin_: you're haunting me :)13:47
kanavin_RP: yes, everyone will hate me for it13:48
kanavin_it was so easy and nice to just ignore the issue :)13:48
RPkanavin_: the autobuilder now highlights warnings fwiw13:49
RP(meaning we'll be able to pay more attention to the warnings now)13:51
RP file /etc/ssl/openssl.cnf conflicts between attempted installs of openssl10-conf-1.0.2p-r0.i586 and openssl-conf-1.1.1+pre9-r0.i58613:53
RPkanavin_: can we install 1.0 and 1.1 together? :/13:53
kanavin_RP: libraries, yes, other stuff, no13:54
kanavin_RP: I still can't believe how badly upstream has botched the transition for distros13:55
RPkanavin_: I'll try just removing the recommends for the openssl10-conf13:56
kanavin_RP: yeah, I'm trying to track down where and why that line was introduced13:57
kanavin_it happened while I was not looking in the summer :) I guess Andre McCurdy did it13:57
kanavin_RP: ah, I suppose the recommends should recommend openssl-conf, my messup probably13:59
*** [RLA]Eske <[RLA]Eske!c0e1ba63@gateway/web/freenode/ip.> has joined #yocto14:01
RPkanavin_: the ssl changes break one of the selftests too :/14:04
*** kyle <kyle!kylematrix@gateway/shell/> has joined #yocto14:07
*** morphis <morphis!> has quit IRC14:09
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has quit IRC14:09
*** aratiu <aratiu!~adi@> has quit IRC14:11
RPkanavin_: one of the tests installs socat but expects its dependencies to be already installed, libssl1.1 being one of them. With the changes, this is no longer the case14:16
*** aratiu <aratiu!~adi@> has joined #yocto14:16
*** joseppc <joseppc!~josep@linaro/joseppc> has quit IRC14:16
*** eduardas_m <eduardas_m!~eduardas@> has quit IRC14:18
kanavin_RP: could be because openssh is no longer dependent on openssl? just a wild guess14:19
*** OnkelUlla <OnkelUlla!> has quit IRC14:19
khemkanavin_: openssl update regressions see
khemmost of these 39 packages failing to build are due to openssl14:20
RPkanavin_: yes, exactly14:21
kanavin_khem: the recipes should be either upgraded to latest upstream, or changed to depend on openssl1014:21
otaviorburton: I couldn't reproduce your error in linux-firmware14:21
rburtonyeah i'll have a look maybe it was actually v114:21
*** eduardas_m <eduardas_m!~eduardas@> has joined #yocto14:23
*** ejoerns <ejoerns!~ejo@2001:67c:670:100:76d4:35ff:fee8:98b3> has quit IRC14:23
*** varjag <varjag!> has quit IRC14:24
RPrburton: v2 in -next and builds ok fwiw14:27
rburtonmust be something i did wrong14:29
RPkanavin_: I have a proposed fix for the conf issue14:32
RPkanavin_: - not sure I like it but...14:33
kanavin_RP: we can just fix the original openssl10 recipe to refer to openssl-conf14:33
RPkanavin_: no we can't :(14:34
RPkanavin_: think about something which depends on openssl10 and not openssl and what it would see in its sysroot14:34
RP(since you just impicitly made openssl10 depend on 1.1)14:34
RPkanavin_: initially I wanted to do that too14:35
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has quit IRC14:36
kanavin_RP: yes, then this is the way out I guess14:36
RPkanavin_: I'll try testing it14:36
kanavin_RP: we'd really benefit from having Andre involved in this14:38
*** marka <marka!~masselst@> has joined #yocto14:40
RPkanavin_: yes14:41
kanavin_RP: but wait - RRECOMMENDS still populates the sysroot? I thought you need DEPENDS for that, and RRECOMMENDS merely tweaks the packaging metadata.14:47
zeddiiSegmentation fault14:50
zeddiipmu-events/Build:13: recipe for target '/home/bruce/poky/build/tmp/work/qemuarm64-poky-linux/perf/1.0-r9/perf-1.0/pmu-events/pmu-events.c' failed14:50
zeddiimake[3]: *** [/home/bruce/poky/build/tmp/work/qemuarm64-poky-linux/perf/1.0-r9/perf-1.0/pmu-events/pmu-events.c] Error 13914:50
zeddiithat can't be good.14:50
RPzeddii: er, no :/14:51
zeddiilets just say, there's lots wrong with that 4.14 perf build. that's issue #4.14:51
*** pohly <pohly!> has quit IRC14:57
*** pohly <pohly!> has joined #yocto14:58
RPzeddii: :(15:00
*** tgoodwin <tgoodwin!> has quit IRC15:00
kanavin_RP: RRECOMMENDS_libcrypto10 += "openssl10-conf" ---> changing this to openssl-conf would only trigger do_package to re-do the packaging, is that not correct? I must be missing something here15:01
zeddiifinally. seems to have built.  now to make sure I didn't break other kernel versions and perf.15:02
RPkanavin_: what I want to avoid is when something does DEPENDS = "openssl10", only openssl10 ends up in the recipe-sysroot and not openssl (1.1)15:02
RPkanavin_: whilst that only causes packaging to rerun, I'm not 100% sure it won't mean openssl (1.1) gets pulled into the sysroots via dependencies15:03
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has joined #yocto15:03
*** varjag <varjag!> has joined #yocto15:04
*** stephano <stephano!~stephano@> has joined #yocto15:05
RPrburton: do you know if Anuj had on target benchmarks for python performance?15:05
rburtonRP: no15:05
*** mckoan is now known as mckoan|away15:07
rburtonthis py recipe is a bit of a mess15:07
rburtonwhy does it need to rebuild in install15:07
kanavin_RP: I'm still not grasping the issue, maybe the best is to get back to it next week :)15:07
RPrburton: PYTHON3_PROFILE_TASK has -n 10, I wonder if -n1 would give the same info15:07
rburtonand why does it not pass OPT=CFLAGS15:07
rburtonRP: you'd think so.  i'll check clear ;)15:07
RPits not like we care about the actual time, its just branch prediction as I understand it15:08
rburtonclear does -n2015:08
* RP isnt convinced15:08
rburtonme neither15:08
RPI'm going to have to write some tests aren't I :/15:09
rburtonRP: running pybench shoud be sufficient to demonstrate that pgo helps and test if -n1 is the same as -n1015:10
*** WillMiles <WillMiles!> has joined #yocto15:11
RPrburton: right, I just don't want to do it manually :)15:11
zeddiiwell crap. that's confusing.15:21
zeddiias long as I cleanall on perf, it builds fine. every time.15:21
zeddii1: perf-1.0-r9 do_package - 0s (pid 23028)15:22
*** stephano <stephano!~stephano@> has quit IRC15:23
*** stephano <stephano!~stephano@> has joined #yocto15:28
*** ant_home <ant_home!> has quit IRC15:36
*** davenporten <davenporten!8b55c117@gateway/web/freenode/ip.> has joined #yocto15:38
davenportenWhen I run bitbake -c do_rootfs core-image-minimal it creates the rootfs, but I don't get a tar.gz or anything like that of it15:39
davenportenHow can I get it to do that?15:39
kanavin_davenporten: bitbake core-image-minimal15:40
davenportenSo I don't need the -c do_rootfs option?15:41
kanavin_do_rootfs only installs packages into the target rootfs directory on your machine, but doesn't package that directory into a flashable image15:41
davenportenGotcha, thanks kanavin, much appreciated15:41
*** tgoodwin <tgoodwin!> has joined #yocto15:41
kanavin_pay attention to the tasks that are being executed as an image is being built :) rootfs is followed by several other things15:42
davenportenkanavin: will do, really thanks15:43
RPzeddii_home: 'contamination' from the kernel build?15:45
*** florian <florian!~florian_k@Maemo/community/contributor/florian> has quit IRC15:47
*** davenporten <davenporten!8b55c117@gateway/web/freenode/ip.> has quit IRC15:52
*** eduardas_m <eduardas_m!~eduardas@> has quit IRC15:53
*** eduardas_m <eduardas_m!~eduardas@> has joined #yocto15:53
kanavin_yessss python 3.7 patch15:53
*** Son_Goku <Son_Goku!~King_InuY@fedora/ngompa> has quit IRC15:54
*** eduardas_m <eduardas_m!~eduardas@> has quit IRC16:01
*** egavin <egavin!> has quit IRC16:18
*** chandana73 <chandana73!~ckalluri@> has joined #yocto16:33
*** T_UNIX <T_UNIX!uid218288@gateway/web/> has quit IRC16:37
*** joaocfernandes <joaocfernandes!~Joao@> has quit IRC16:41
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has joined #yocto16:50
*** arielmr <arielmr!> has joined #yocto17:00
*** Bunio_FH1 <Bunio_FH1!> has quit IRC17:03
khemkanavin_: I guess I have to blacklist these recipes and wait for folks to fix them17:04
khemah kernel branches got rebased17:04
*** boucman_work <boucman_work!~jrosen@wesnoth/developer/boucman> has quit IRC17:06
mcfriskdid someone already try to switch poky from gzip to pigz, form tar -cz to tar -c | pigz -c ... ?17:21
*** stephano <stephano!~stephano@> has quit IRC17:29
khemmcfrisk: yes we use pigz for many functions e.g. image creation and sstate too now17:30
khemsee for sstate17:30
khem for image17:31
*** stephano <stephano!~stephano@> has joined #yocto17:33
*** georgem_home <georgem_home!uid210681@gateway/web/> has joined #yocto17:57
*** khem <khem!~khem@unaffiliated/khem> has quit IRC18:09
*** aratiu <aratiu!~adi@> has quit IRC18:14
*** khem <khem!~khem@unaffiliated/khem> has joined #yocto18:14
khemRP: I am seeing Exception: FileExistsError: [Errno 17] File exists: '/mnt/a/oe/build/tmp/sysroots-components/riscv64/openssl10/usr/include/openssl/ssl23.h' -> '/mnt/a/oe/build/tmp/work/riscv64-bec-linux/openssh/7.8p1-r0/recipe-sysroot/usr/include/openssl/ssl23.h'18:16
khemwhen I build another image in same tmp/18:17
khemrelated to the latest push ?18:17
khemmaster-next btw.18:17
*** aratiu <aratiu!~adi@> has joined #yocto18:18
*** jae1 <jae1!~jaewon@> has joined #yocto18:38
*** fdanis_away is now known as fdanis18:44
*** fdanis is now known as fdanis_away18:45
*** tgoodwin <tgoodwin!> has quit IRC19:05
*** rewitt1 is now known as rewitt19:18
*** tgoodwin <tgoodwin!> has joined #yocto19:25
*** ant_home <ant_home!> has joined #yocto19:42
*** pohly <pohly!> has quit IRC19:44
*** AbleBacon <AbleBacon!~AbleBacon@unaffiliated/ablebacon> has quit IRC19:50
*** georgem_home <georgem_home!uid210681@gateway/web/> has quit IRC20:07
*** BCMM <BCMM!~BCMM@unaffiliated/bcmm> has joined #yocto20:21
*** BCMM <BCMM!~BCMM@unaffiliated/bcmm> has quit IRC20:22
*** marka <marka!~masselst@> has quit IRC20:31
*** sveinse <sveinse!> has quit IRC20:35
aehs29RP: rburton Id still like to see if running the python pgo on qemu has the same results (or similar) as running the profile on target20:36
*** likewise <likewise!> has joined #yocto20:36
aehs29and Im not sure we have to do it on every build for qemu, I understand that we'd want to do it for hw targets, but perhaps in the qemu case just hold a profile  and use it, and we can refresh the profile every release or something?20:37
RPaehs29: having looked into it, its statistical branch analysis so it should be arch independent21:06
aehs29RP: wouldnt that mean the profile is specific to the application?21:07
RPaehs29: the profile is specific to the load you test with, yes21:11
aehs29so were profiling for something without knowing what it is...21:11
*** WillMiles <WillMiles!> has quit IRC21:12
RPaehs29: we're assuming that pybench stresses the python interpreter in ways which make it faster in general to optimise it that way21:13
RPaehs29: if you test bitbake parsing speed it is faster on a pgo optimised python21:13
RPeven though bitbake parsing isn't what it was profiled with21:14
aehs29RP: ok I understand, but then the profile were creating wont change much right?, so we dont actually have to create a profile everytime21:14
RPaehs29: That was my point...21:14
RPaehs29: why not generate it for python-native and then use the results everywhere?21:15
aehs29haha ok that was mine too21:15
*** ant_home <ant_home!> has quit IRC21:18
*** ant_home <ant_home!> has joined #yocto21:20
*** rob_w <rob_w!~rob@unaffiliated/rob-w/x-1112029> has quit IRC21:20
*** peniwize <peniwize!~peniwize@> has joined #yocto21:24
peniwizeHi all.  I'm new to Yocto and I'm trying to create a simple recipe that creates a system image based on "core-image-x11".  I've created a recipe that duplicates "poky/meta/recipes-graphics/images/".  It works when it's a verbatim copy.  Next I tried to add the Intel video driver by cloning "" ad adding 'IMAGE_INSTALL += "xf86-video-intel"' to my recipe.  This results in an image that boots to sh21:31
peniwizeand recognizes no commands (even ls).  Any ideas what I’m doing wrong?21:31
peniwizeI also can't run dmesg since no commands are recognized and there are no obvious errors on the console.21:32
peniwizeOh, I'm running this OS image on an Intel board via USB flash drive that is populated by dd'ing "tmp/deploy/images/intel-corei7-64/sockeye-gm-intel-corei7-64.wic" to it.21:33
rburtonpeniwize: you've overridden the default value of IMAGE_INSTALL which is set with ?=21:33
peniwizeOh, I thought += would preserve the default.21:34
peniwizeWhat is the proper way to preserve it?21:34
rburton+=  is append but core-image.bbclass does IMAGE_INSTALL ?= ...21:35
rburtonsee core-image.bbclass, easiest thing is to set CORE_IMAGE_EXTRA_INSTALL instead21:35
peniwizeOh, ok.  I was about to ask where IMAGE_INSTYALL was initially configured.21:36
rburtonwhen in doubt, bitbake -e myimage would show you the values it is using21:36
rburtonand you'll see how IMAGE_INSTALL is just set to xf86-video-intel21:36
rburtonof course, core-image-x11 with a meta-intel MACHINE will pull that driver in automatically21:36
rburtonso you don't need to do that21:37
peniwizerburton: Thanks much for the help.  I'll make some changes and see if it starts working.  Yocto and BitBake are pretty awesome, but it's a LOT of info to absorb and retain.  This channel is a fantastic asset.21:37
rburtonnote that in yocto master we don't use that driver on recent hardware, because its better to use the integrated modesettings driver instead21:38
peniwizeOh, that's good to know.  Much appreciated.21:39
*** rburton <rburton!> has quit IRC21:40
khemRP: too many variables to remeber a release21:44
khemmay be we should just stick to codenames all along21:44
*** varjag <varjag!> has quit IRC21:52
*** georgem_home <georgem_home!uid210681@gateway/web/> has joined #yocto22:02
*** [RLA]Eske <[RLA]Eske!c0e1ba63@gateway/web/freenode/ip.> has quit IRC22:05
armpitkhem, we should us emoji's22:23
halsteadmoto-timo, Can you tell if is an infra problem or a code issue? Seems to be happening on our Ubuntu workers.22:25
moto-timohalstead: looking22:25
*** mattsm <mattsm!~mattsm@> has quit IRC22:26
moto-timohalstead: neon doesn't support Java 10. We only tested neon with JDK 822:28
*** mattsm <mattsm!~mattsm@> has joined #yocto22:28
halsteadmoto-timo, I suspected that might be the case. I'll downgrade.22:29
halsteadThanks again.22:30
moto-timohalstead: I haven't tried Ubuntu 18.04 yet either...22:30
moto-timohalstead: sure thing22:30
moto-timohalstead: IIRC oxygen should be ok on Java 10.22:30
halsteadmoto-timo, I don't know if there is an easy way to keep oxygen on 10.22:32
moto-timohalstead: not worth differentiating, I guess we just downgrade until we can live with 10 across the board22:34
moto-timoof course if anybody in the community wants to stay on Neon but also wants Java 10, patches are welcome :)22:35
khemhalstead: ubuntu has dash as sh may be thats important ?23:07
halsteadkhem: perhaps. I think we need to test that default. Right?23:10
*** ant_home <ant_home!> has quit IRC23:37
*** stephano <stephano!~stephano@> has quit IRC23:46
*** ant_home <ant_home!> has joined #yocto23:53
*** ntl <ntl!> has quit IRC23:56
aehs29I think theres a bug on our process to launch a terminal, it gets launched, but it immediately exists, e.g. devshell, menuconfig, etc23:58

Generated by 2.11.0 by Marius Gedminas - find it at!