**** BEGIN LOGGING AT Tue Jun 06 03:00:03 2017 Jun 06 03:03:14 kergoth: this is an iso image Jun 06 05:23:09 hello Jun 06 05:23:33 where can i get devmem2.c from? lartmaker.nl is not accessible Jun 06 05:36:32 where is the yocto project mirror located? Jun 06 05:53:38 cornel: did you try googling for "devmem2.c"? -> http://free-electrons.com/pub/mirror/devmem2.c Jun 06 05:54:12 nrossi, no, honestly, how do i know this is the right devmem2.c ? Jun 06 05:54:48 cornel: thats the one meta-oe points at http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/devmem2/devmem2.bb?h=master Jun 06 05:54:58 all right, i can check the sum Jun 06 05:55:02 thank you nrossi Jun 06 05:56:00 morning all Jun 06 05:56:54 hi CostinC Jun 06 07:20:57 nrossi, why is this necessary? os.rename("devmem2-new.c", "devmem2.c") Jun 06 07:22:34 cornel: tbh no idea, i thought it might be due to the patch but that doesn't seem to have been made against devmem2-new.c. But the rename is there because of the "downloadfile" in the src_uri Jun 06 07:23:39 nrossi, thank you Jun 06 07:23:46 is true that the checksums have changed Jun 06 07:23:49 so ..... Jun 06 07:24:11 i'm back to square one: is this the same file? Jun 06 07:24:19 need to do a digff, i suppose Jun 06 07:24:22 diff Jun 06 07:25:50 cornel: the devmem2.bb recipe has the expected checksum, just use the updated devmem2 recipe no? Jun 06 07:27:01 nrossi, sure Jun 06 07:30:46 oO( am I actually the only one who finds it a bit scary how many things seem to rely on devmem? ) Jun 06 07:31:54 LetoThe2nd, how do i find out how many? Jun 06 07:49:12 cornel: no thats not what i meant. devmem2 is actually a hacky thing to poke memory. i'm all fine with it using it for dev purposes, but IMHO it should not be product relevant in a proper design. the fact that this particular broken recipes creates such a stir tells me... otherwise. Jun 06 07:49:51 LetoThe2nd, i see, thqank you for clarifying it to me Jun 06 08:31:47 Hi, I'm having a strange issue with touchscreen input the touchscreen input is showing in evtest and in libinput-debug-events but if i start an weston-fower or a qt application the touchscreen input is not handled Jun 06 09:08:50 Hi everyone. I'm having an issue with installing python on my linux image. Even with a IMAGE_INSTALL += "python-core python-modules", I coudn't get a .py script to execute on my machine. I also put the '#!/usr/bin/env python' at the beginning of the file, but an 'unknown python' error appears. Does this ring a bell? Jun 06 09:10:46 can you enter python shell? Jun 06 09:11:42 Nop, and 'python --version' doesn't work either. Looks like it's just not present on the image... Jun 06 09:29:31 olivier: that's odd... I'd run through step-by-step - does bitbake -e your-image-name Jun 06 09:29:48 ... show that the value has taken effect? Jun 06 09:30:37 if yes, have a look at the package manifest next to the image in tmp/deploy/images/ list python? what about log.do_rootfs for the image, any mention of python in there? Jun 06 09:38:40 bluelightning: I can see in the manifest the python and its modules "libpython2.7-1.0 cortexa8hf-neon 2.7.11-r1 ... python-2to3 cortexa8hf-neon 2.7.11-r1 .. python-... etc" Jun 06 09:38:59 bluelightning: Also, in the do_rootfs, I see "Installing the following packages: python-modules ..." Jun 06 09:39:31 olivier: ok, that's a good sign Jun 06 09:40:08 olivier: now for a really stupid question, apologies in advance - you definitely flashed / copied the right image file to the device and you're sure it's the one being booted? Jun 06 09:42:07 bluelightning: No problem, and just to be sure, I'm doing it right now. Jun 06 09:44:47 @all morning, i have some issue with a custom wifi driver installation : configure: WARNING: bison not found. Please install before continuing. configure: error: Required packages are missing. Please install them and rerun ./configure make: *** [libnl_config] Error 1. bison is already added to DEPENDS/RDEPENDS but it still not working ... any adivces ? Jun 06 09:47:54 PinkSnake: it would need to be bison-native I think - is that what you added? Jun 06 09:49:21 bluelightning : +1 I show recipes that depends on bison-native but didn't found any recipes which depends of bison. Jun 06 09:49:30 bluelightning : I've see* Jun 06 09:52:20 bluelightning: It's updated, and the result is still the same. "python --version => python: command not found" Jun 06 09:53:17 bluelightning: But a "ls /usr/lib | grep python" => "libpython2.7.so.1.0" Jun 06 09:54:04 bluelightning: (Nothing inside /usr/bin, /lib nor /bin) Jun 06 09:55:18 olivier: hmm, so python-core is the package that includes /usr/bin/python, was that mentioned as having been installed? Jun 06 10:18:07 ed2: https://autobuilder.yocto.io/builders/nightly-oe-selftest/builds/324/steps/Running%20oe-selftest/logs/stdio Jun 06 10:27:23 rburton: hm, strange. it worked on my system. I'll check it out. Jun 06 10:27:52 rburton: you can skip "wic rm" series if you're in a hurry. Jun 06 10:28:07 yeah might well do that Jun 06 10:31:32 hi, my touchscreen events are not handled. Jun 06 10:50:39 bluelightning: Yes, the python-core module is included (seen in log.do_rootfs file) : "Installing the following packages: ... python-core..." Then "... Installing python-core (2.7.11-r1) on root. ..." Jun 06 11:25:46 olivier: hmm, that is puzzling then :( Jun 06 11:32:18 bluelightning: Oh oh... I guess it means you're out of ideas... Jun 06 11:35:48 olivier: well, I suppose we could have a look at the generated image Jun 06 11:37:07 bluelightning: I see in the log a 'Breaking circular dependency on python-core for python-re'; but I'm not sure it's an error. Jun 06 11:37:43 I think that's just a warning Jun 06 11:38:20 can you have a look under IMAGE_ROOTFS for the image you're building? i.e. bitbake -e your-image | grep ^IMAGE_ROOTFS= Jun 06 11:45:02 bluelightning: That's troubling... I see a "/usr/bin/python" inside the rootfs folder... Jun 06 11:45:29 olivier: run it directly on the target Jun 06 11:45:51 olivier: if the linking has gone really wrong you'll get file not found if it can't find the loader, for example Jun 06 11:47:28 bluelightning: oh gosh, I realised my script for updating is pointing to a ancien folder for the update.... I guess the obvious wasn't enough checked.... I'm gonna check, but I'm almost sure the isuue is here; In that case, I'm sooo sorry for having watsed your time... Jun 06 11:48:00 olivier: no worries... I'm very honestly much happier that the mystery is resolved than worried about time :) Jun 06 11:53:43 bluelightning: Ok, after correcting the update script, python is present on the image and works. Many thanks, and sorry again for that mistake. Jun 06 12:40:10 Anyone familiar with gstreamer with Qt? Jun 06 12:52:45 olivier: no problem Jun 06 13:34:21 hello, I'm running into an IOError that occurred between two concurrent image builds. The image was compiling fine over the weekend, and when I did a clean/rebuild, I started getting this error: Exception: IOError: [Errno 13] Permission denied: '/blah/build-directory/blah/image-name.rootfs.manifest' Jun 06 13:54:31 twall: what do you mean by two concurrent builds? Jun 06 13:54:42 er, sorry, two builds right after each other Jun 06 13:54:51 i just realized that concurrent doesn't mean that at all Jun 06 13:57:05 essentially, the IOException error is coming out of a file that I had just written to, with nothing that would change it between image compilations, yet I'm receiving permission issues somehow Jun 06 14:57:03 YPTM in 3 minutes? Jun 06 14:59:15 YPTM armin on Jun 06 14:59:50 YPTM: Joshua joined Jun 06 14:59:55 standing by for host Jun 06 15:00:27 YPTM tlwoerner is on (holding) Jun 06 15:00:44 YPTM: Joshua Watt here. Jun 06 15:01:14 YPTM: Pascal Bach on Jun 06 15:01:33 * tlwoerner is Trevor Woerner Jun 06 15:01:37 YPTM: Stephe joined Jun 06 15:01:44 what is the current dial in, I don't have the new one Jun 06 15:01:56 1-800-262-0778/404-397-1527 Jun 06 15:02:07 88748961# Jun 06 15:02:09 YPTM ross in Jun 06 15:02:13 YPTM: Stephano joined Jun 06 15:02:26 * RP joined Jun 06 15:02:26 RP: ^ Jun 06 15:02:35 just in time :) Jun 06 15:02:37 YPTM: Michael here. Jun 06 15:02:40 rburton: indeed :) Jun 06 15:03:00 Juan joined Jun 06 15:03:24 YPTM: Paul Eggleton joined Jun 06 15:03:29 wow Jun 06 15:03:38 here now Jun 06 15:03:53 * bluelightning is in the European timezone for a while Jun 06 15:04:33 this is 2.2.2? Jun 06 15:04:50 https://wiki.yoctoproject.org/wiki/2.2_QA_Status#Summary_Report Jun 06 15:06:45 history has said 2.2.3 is likely as 2.2 EOLs near 2.4 release Jun 06 15:12:12 * fray doesn't have any suggestions.. sorry Jun 06 15:18:39 hi, when i use IMAGE_BOOT_FILES = "boot.scr" in my local.conf it does not copy the boot.scr file into the sdcard image ? what do i wrong? In the manual, they only say, list files there Jun 06 15:20:24 setting up YP AB and struggling, talk to me Jun 06 15:20:47 do i need to build a wic image for that Jun 06 15:21:21 container stuff is on my To Do Jun 06 15:21:29 (it's pretty long though) Jun 06 15:23:31 joshuagl: containerizing the AB, or builds for containers, it the latter case, i'm willing to help Jun 06 15:24:15 lol Jun 06 15:24:23 LetoThe2nd: well, both. Containerising the AB and having the AB spin up containers of known distros for the associated release are higher priorities for me Jun 06 15:24:50 joshuagl: ah ok. nah thats unrelated then. i'm doing a bit of tinkering in the direction of super-slim container images. Jun 06 15:25:32 LetoThe2nd: sure, there are efforts in that direction. Patches on the list and bugs in bugzilla iirc Jun 06 15:26:55 joshuagl: setting up autobuilder locally: https://lists.yoctoproject.org/pipermail/yocto/2016-December/033598.html Jun 06 15:30:08 tlwoerner: looks like all of their questions were answered? Jun 06 15:30:54 YPTM is over Jun 06 15:32:00 tlwoerner: or were you more pointing at common bumps in the road? Jun 06 15:32:57 joshuagl: yes, the 2nd :-) just pointing out how someone else found the instructions lacking and what got them through Jun 06 15:35:54 tlwoerner: thanks. The instructions are indeed lacking Jun 06 15:57:55 do i need to add a mcopy line to copy the IMAGE_BOOT_FILES into the sdcard image? Jun 06 16:32:04 whats difference between yocto-layer create and yocto-bsp create ? Jun 06 16:32:53 yocto-bps creates BSP layers Jun 06 16:33:19 yocto-layer creates generic layers, so it can be used as a SW layer, for example Jun 06 17:28:04 hi, anyone knows how can I always fetch an http tar even when it didnt change Jun 06 17:28:29 I tried to reimplement do_fetch, calling fetch.clean() before fetch.download() Jun 06 17:28:54 but it didnt work, because the sstate I guess Jun 06 17:29:10 I tried to set do_fetch with nostamp, but it didnt work too Jun 06 17:45:38 igor: the bitbake fetcher writes .done in DL_DIR to indicate it's been completely fetched. even if do_fetch re-runs, it won't re-fetch the sources if it doesn't have to. use a postfunc or _append to remove the .done stamp at the end of do_fetch, along with nostamp, and it should work Jun 06 17:49:25 ok, I will try that Jun 06 17:49:30 thank you kergoth Jun 06 17:50:13 I tried to remove it manually, but it didnt work, the fetch task did not run at all Jun 06 17:55:27 kergoth: it did not work, I created a do_fetch_append that removes .done file, it works the first time I ran, but the second call of bitbake, it did not run do_fetch Jun 06 17:55:56 saying: Attempted 612 tasks of which 612 didn't need to be rerun and all succeeded. Jun 06 17:56:12 it did not detects that do_fetch needs to run again Jun 06 17:57:33 I'm trying to find out how bitbake knows when a task needs to run Jun 06 17:58:13 I thought it was on sstate class, but I removed this class from INHERIT and it didnt work too Jun 06 18:11:41 igor: you need to remove the file from DL_DIR *and* run a clean to get rid of the stamp files or make it nostamp Jun 06 18:12:46 RP: I set do_fetch[nostamp] = "1" but it is not working Jun 06 18:19:09 I'm using branch dizzy, so I dont know if it is a bug that was fixed in newer branches or if nostamp only works with some tasks Jun 06 18:19:33 whois seebs Jun 06 18:19:36 shit Jun 06 18:19:40 disregard me Jun 06 18:22:11 anyway, I've been running into some issues regarding how my git package is working on my target machine. I'd like for the submodule command to be enabled, however that doesn't appear to be happening. Jun 06 18:26:53 to the best of my knowledge, I'm downloading a git tarball from source, then doing a cross-compilation. Somewhere down the line I lose submodule functionality Jun 06 18:30:07 twall: most likely it's in a separate subpackage. the git recipe emits multiple packages, as do all of our recipes Jun 06 18:40:25 kergoth: could use some help with https://autobuilder.yocto.io/builders/nightly/builds/345/steps/BitbakeSelftest/logs/stdio :/ Jun 06 18:40:34 kergoth: also, could you send those docs to scott please? Jun 06 18:42:09 Would anyone have suggestions for following scenario: 2 targets, lets say foo & bar .. both rely on meta-baz layer which has deploy-something recipe. foo actually builds those binaries which bar uses then to deploy into real image. While all this is still fine and dandy - i only get warnings while building foo that files are missing. But if/when i want to create a source mirror, build fails because of Jun 06 18:42:16 missing stuff that was defined SRC_URI .. Jun 06 19:05:31 jani: better if you pastebin all steps and logs Jun 06 19:14:42 lsandov: yeah well .. sort of cant do that as the parts of the layer & recipes are "internal only". but if i rephrase my question: Is there any way to temporary override SRC_URI for a single recipe ? Jun 06 19:16:05 jani: you can create a bbappend recipe and set it there Jun 06 19:16:06 so essentially the issue is: doing bitbake myfpgatarget -c fetchall will fail because there's a recipe in the layer that has SRC_URI=file://system.dtb (and few others which are generated later on in the build) Jun 06 19:17:48 jani: that looks to me that those dependencies should be in the Makefile instead of recipes Jun 06 19:18:02 jani: if your recipe uses autotools of course Jun 06 19:19:37 there's nothing really building, static firmware files. Jun 06 19:20:41 bbappend ofcourse would do but it would need to be generated at build time so that it wont affect the normal builds .. Jun 06 19:23:39 jani: there's a url parameter to rename the file in the download dir, then you can rename it back in do_unpack. see devmem2 in meta-oe as an example Jun 06 19:32:37 kergoth: thanks .. i dont quite see the logic there thought :D Jun 06 19:33:54 i think i found something, BBMASK and i just block the recipe. Jun 06 19:54:11 jani: nice, good find Jun 06 19:55:00 probably would have been easy to spot that but im rather noob yocto. Jun 06 19:57:30 but yeah, in my case, i already have a separate config file just for source mirroring, i can just add the mask there and be ignore the "bad" recipe .. Jun 06 20:03:58 kergoth and RP: I got it!! Jun 06 20:04:27 I create a annonymous python function and I call bb.build.write_taint('do_fetch', d) Jun 06 20:05:08 so I created a do_fetch_prepend removing the downloaded files by calling clean function from Fether class Jun 06 20:05:17 Fetch* Jun 06 20:05:27 Thank you guys for your help Jun 06 21:45:59 rburton: I think the prior version of -pie patch got merged Jun 06 21:46:30 rburton: the final patch is this http://git.openembedded.org/openembedded-core-contrib/commit/?h=kraj/master&id=b06033c5c42584145487333456b43dc63a52b9ff Jun 06 21:46:46 rburton: this fixed the case for both python2 and python3 in one go Jun 06 22:38:02 khem: damnit Jun 06 22:39:16 khem: can you send an updated patch to fix things please? Jun 06 22:39:41 RP: revert and cherry-pick in mut now Jun 06 22:40:21 rburton: ok Jun 06 22:48:55 Hello everyone! Is there a way to enable verbose in u-boot? Jun 06 23:31:56 RP: it was already sent to ml last week Jun 06 23:36:44 the patch which was merged is fine too it just fixes half the problem Jun 06 23:38:23 austin: during build ? Jun 07 00:26:31 Hello. I just upgrade Yocto to 2.3. When building an image, do_rootfs tells me that /usr/lib/perl conflicts between some of my recipes. I have multiple recipes which puts libs in this dir, and in the recipe do_install function, I create this dir using install -m xxx -d ${D}${libdir}/perl. How do I avoid this error? Jun 07 00:27:16 "file /usr/lib/perl conflicts between attempted installs of recipe-a and recipe-b" Jun 07 01:08:29 kspr: yes thats a problem you need to decide on which one should install it Jun 07 01:09:00 and if you need both versions then you have to either rename one of it or move it to a different subdir Jun 07 01:09:22 khem, well.. I have multiple Perl modules, in this case. All is installed under /usr/lib/perl. Each installed by different recipes. Jun 07 01:09:30 /usr/lib/perl is a directory Jun 07 01:10:01 yes that should not be a problem unless some package is packaging an empty /usr/lib/perl Jun 07 01:10:17 as long as files are unique it should work ok Jun 07 01:10:50 It could seem like I have added /usr/lib/perl to the FILES_${PN} array in multiple recipes.. in previous versions this was not an issue though Jun 07 01:15:04 it should tell you which two packages are in contention Jun 07 01:15:51 Yes, it does. Jun 07 01:16:49 It seems to fix the issue, removing the directory from the FILES_ array and only including the exact files I need. A little more work, to have all the files in the FILES_ array rather than just do it recursively, but it works :-) Jun 07 01:17:42 almost. hmm Jun 07 01:18:16 Ok, I need to remove -m xxxx from install command when creating directories, otherwise I get this error. Jun 07 01:26:34 usually install -d is enough for dirs Jun 07 01:27:18 Yes, okay. Well, that seems to be the issue - different file modes on the directories. Jun 07 01:27:27 thanks for your time :-) **** ENDING LOGGING AT Wed Jun 07 03:00:03 2017