**** BEGIN LOGGING AT Thu Nov 30 03:00:03 2017 Nov 30 08:43:26 I did check meta-qt4 and there is openssl disabled by default while when qt4 was part of poky it was enabled, any reason for that? Nov 30 08:46:24 open-nandra: I think it probably happened during the time we upgraded to openssl 1.1 (and then shortly after reverted) Nov 30 08:46:51 open-nandra: you can easily reconfigure it via a bbappend or at the configuration level Nov 30 08:47:17 bluelightning: ok something like -with-openssl and add DEPENDS += "openssl" ? Nov 30 08:47:40 open-nandra: yes Nov 30 08:47:51 bluelightning: ok thanks for info Nov 30 09:55:18 Hello Nov 30 09:55:41 samoud: hi Nov 30 09:56:09 How are you doing @nayfe Nov 30 09:57:37 Soon week-end, so fine Nov 30 09:58:10 I have a question, I need the getconf command in my new yocto image with poky and oe. Nov 30 09:58:38 where should I add it ? local conf ? Nov 30 10:02:55 samoud: it should be ok to add it with IMAGE_INSTALL_append = " glibc-utils" in local.conf Nov 30 10:03:14 Samoud, or modify busybox config Nov 30 10:03:48 aurele: if you actually use busybox, that is :-P Nov 30 10:04:05 Thanks Nayfe and Aurele ! what's the best solution ? Nov 30 10:04:14 LetoThe2nd, obviously Nov 30 10:05:06 Samoud: adding it to the IMAGE_INSTALL of your target image recipe Nov 30 10:05:45 Ok, I'll try this one ! Thanks a lot Nov 30 10:08:31 LetoThe2nd, why would you add a complete package if he uses busybox, modify busybox configuration seems more consistent to me if you use busybox, but maybe there is another reason. Nov 30 10:10:03 aurele: as long as he/she doesn't state more details, anything is equally good (or bad) Nov 30 10:10:10 getconf is in busybox ? Nov 30 10:11:37 nayfe, this is the point Nov 30 10:12:07 LetoThe2nd, nayfe, never mind there is no such tool in busybox Nov 30 10:12:42 I thought the entire world was in busybox, but it seems some tools are not provided Nov 30 10:13:25 and there's more than wnough people not running busybox too. Nov 30 10:13:27 :) name looked like a good candidate Nov 30 10:16:42 Hi all, someone here has good information to start with real-time with yocto (for Zynq arch)? Nov 30 10:17:14 depends on your definition of "real time" :) Nov 30 10:18:39 i would like to test real-time app to evaluate the better solution for our application ( Xenomai, LinuxRT, Sysgo maybe ) Nov 30 10:19:07 but i'm stuck on patching xenomai manually (linux-xlnx not the mainline kernel) Nov 30 10:19:53 thats why i'm asking. well the yocto project does actually not include infrastructure for xenomai or comparable solutions so far Nov 30 10:20:29 I'm trying to remove "vi" package from my qemux86 image, I tried PACKAGE_EXCLUDE_pn-target_image = "vi" BUT vi is still available in the image.. anyidea? Nov 30 10:20:29 there is an actively maintained preempt-rt kernel, but once you're in the land of rather specific hardware you usually need to maintain your own patched kernel Nov 30 10:20:56 xtron: depending on your image, this probably comes from busybox, not being a seperate package Nov 30 10:21:39 the trouble is i would like to use Xilinx Kernel in order to keep FPGA part ;) Nov 30 10:22:04 LetoThe2nd: then what is purpose of and how to remove a package for example "vi" ? Nov 30 10:22:15 butterknife-user: yes, so its your fun part to marry $RTSOLUTION + xilinx vendor kernel Nov 30 10:23:03 @LetoThe2nd i'm not the only one who want do real time on zynq ? ).0 Nov 30 10:23:11 xtron: the thing is that you're mixing up "package" and "binary on the target". if vi is not a package of itself, then any package-based infrastructure does not apply. Nov 30 10:23:18 @xtron IMAGE_INSTALL_remove += "vi" ? Nov 30 10:23:19 butterknife-user: i can't comment on that. Nov 30 10:24:08 @LetoThe2nd thx for info :) Nov 30 10:24:58 xtron: so again: if you're using busybox, then it will offer a lot of, lets call it "commands" on your target. but those are not pulled in by package names / recipes, but the busybox config. so tinkering with IMAGE_INSTALL and friends will get you nowhere. Nov 30 10:27:02 (IIRC busybox calls them applets, but i might be mistaken) Nov 30 10:27:33 Hello Nov 30 10:27:35 butterknife-user: I think I already have tried this with no success. Nov 30 10:29:39 butterknife-user: you can look at the meta-eldk layer by denx, there is some xenomai-inspiration there. any other solutions, i don't have any experience. Nov 30 10:29:46 xtron: type ls -l $(which vi) and if it points to busybox then you have to remove it from busybox config Nov 30 10:30:08 marex-cloud: maybe you can comment here? -> butterknife-user Nov 30 10:31:17 Has someone experiences with "socketcan"? I am desperately trying to add drivers for my can stack to the kernl of my yocto. Everything works without errors, but I don't find the can stack on my imx6 board :( Nov 30 10:31:49 xtron: you can look here for example on how to tweak busybox http://bec-systems.com/site/1145/modifying-the-busybox-config-in-openembedded Nov 30 10:33:24 eldk is ancient/dead ... Nov 30 10:33:54 Marex: \o/ Nov 30 10:33:59 cfr xenomai website, do you have valid reason to use xenomai or does preempt-rt suffice ? Nov 30 10:34:23 afair xenomai people suggest to use preempt-rt unless you have a very good reason to use xenomai Nov 30 10:35:25 @LetoThe2nd thx i 'm going to check that Nov 30 10:41:25 Tim234: you problem is in kernelspace or userspace ? Nov 30 10:43:00 Tim234: i'd rather start with "everything works fine",,, which is what, exactly? Nov 30 10:43:32 (plus, http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-extended/socketcan/libsocketcan_0.0.10.bb?h=master) Nov 30 10:43:42 @Marex i would like just to compare Xenomai and preempt_rt Linux kernel patch :) so i will start with preempt_rt patch Nov 30 10:49:49 @nayfe: I'm not sure where exactly the probleme is. Maybe the log can help: Nov 30 10:49:50 root@phyboard-mira-imx6-6:~# dmesg | grep 'can\|CAN' [ 0.201433] i2c i2c-0: can't use DMA [ 0.202324] i2c i2c-1: can't use DMA [ 0.203751] i2c i2c-2: can't use DMA [ 1.345409] CAN device driver interface [ 2.603712] can: controller area network core (rev 20120528 abi 9) [ 2.614666] can: raw protocol (rev 20120528) [ 2.619086] can: broadcast manager protocol (rev 20120528 t) [ 2.624840] can: netlink gateway (rev Nov 30 10:50:13 uh this looks ugly... :| Nov 30 10:51:06 I am expacting something like: flexcan imx6q-flexcan.0: device registered Nov 30 10:51:24 Tim234> which mx6 board you have N Nov 30 10:51:26 ? Nov 30 10:52:51 nayfe: its a phyboard mira dualLite Nov 30 10:53:39 which kernel do you use ? Nov 30 10:54:19 I dont know, sorry Nov 30 10:54:49 Tim234: please try to be a little more precise what you have done already, and what you are using (software-sie) Nov 30 10:56:49 Tim234: bitbake -e virtual/kernel | grep ^SRC_URI Nov 30 10:57:49 LetoThe2nd: ok, I am using the yogurt distribution from phytec and added libsocketcan and canutils in the local.conf file. compiling worked without errors. Nov 30 11:00:55 nayfe: SRC_URI="git://git.phytec.de/linux-mainline;branch=v4.1.336-phy file://mtd-tests.cfg" Nov 30 11:05:00 brb Nov 30 11:07:10 strange, v4.1.336-phy branch's missing, ain't v4.1.36-phy ? Nov 30 11:08:38 nayfe: yes its v4.1.36 Nov 30 11:19:22 nayfe: uname -a on the board would be far more useful ... Nov 30 11:20:55 Tim234: ifconfig -a, look for can node, do you see it? Nov 30 11:21:27 If so, ip link set can0 up type can bitrate 125000 ... or something Nov 30 11:21:36 marex-cloud: no I dont Nov 30 11:22:00 ip link set can0 ... thus doesnt change anything Nov 30 11:22:49 Tim234: hmm, zgrep CAN /proc/config.gz , look for flexcan Nov 30 11:22:57 Is it enabled? Nov 30 11:25:17 marex-cloud: no zgrep on the board, sry Nov 30 11:27:01 and cat /proc/config.gz | grep CAN doesnt hit anything Nov 30 11:27:37 shall I try to install zgrep? Nov 30 11:27:43 No Nov 30 11:32:45 Use zcat and grep, config.gz is compressed Nov 30 11:36:05 marex-cloud: now I get something like: CONFIG_IMX_HAVE_PLATFORM_FLEXCAN=y Nov 30 11:36:14 Hi all Nov 30 11:36:14 C'ONFIG:CAN=y Nov 30 11:36:42 #CONFIG_CANVCAN is not set Nov 30 11:37:04 CONFIG_CAN_FLEXCAN0y Nov 30 11:37:08 =y Nov 30 11:38:20 I suggest you to contact pythec support, maybe a device tree problem Nov 30 11:41:14 nayfe: yes, I'll try to contact them. Nov 30 11:41:37 thank you for your help Nov 30 11:42:53 Tim234: check your dt if the flexcan is enabled there Nov 30 11:43:18 what is dt? Nov 30 11:43:35 Tim234: find /proc/device-tree -name \*can\* ... and check status node, it should say okay Nov 30 11:47:48 dont find anything on that Nov 30 11:48:54 so thats my problem?! Nov 30 11:53:09 am I right? what am I supposed to find in/proc/device-tree/ ? Nov 30 11:53:15 Without flexcan device tree node enabled you will not have CAN Nov 30 11:54:22 do you know how to enable the flexcan? Nov 30 11:56:46 You have to search selected dts for flexcan0 or flexcan1 nodes Nov 30 11:59:24 do I have to configure the dt before compiling yocto? Nov 30 12:17:31 Tim234: basically check if flexcan is enabled in the DT Nov 30 12:18:19 Tim234: you should have the flexcan nodes in DT though, so not finding them in /proc/devicetree is suspicious Nov 30 12:19:13 Then again, this is some trashy vendorkernel you're using, built on top of that outdated nxp garbage, so nothing would be surprising here ... Nov 30 13:04:40 Tim234: in https://git.phytec.de/linux-mainline/tree/arch/arm/boot/dts/imx6qp-phytec-mira-rdk-nand.dts?h=v4.1.36-phy &can1 status is okay , (imx6qp) but in https://git.phytec.de/linux-mainline/tree/arch/arm/boot/dts/imx6dl-phytec-mira-rdk-nand.dts?h=v4.1.36-phy it is not set Nov 30 13:08:11 linux-mainline .... and if you look at the history, you see all those MLK commits indicating it's NXP vendorkernel whcih has nothing to do with mainline ... Nov 30 13:08:36 nayfe: but yes, that's likely the issue Nov 30 13:08:49 nayfe: still, if there's a can node, it should've showed up in his /proc/device-tree anyway ... Nov 30 13:09:40 Marex: SRC_URI pointed to this kernel, so maybe it needs to remove use-mainline to its machine Nov 30 13:13:42 nayfe: I'm referring to the git repo URL, there's "linux-mainline" in it while it's obviously not true ... Nov 30 13:14:27 marex: Ah yes sure, it's mainline + a LOT of ptaches :) Nov 30 13:19:43 Tim234: http://www.phytec.de/documents/l-814e-3-imx6-bsp-manual/#CAN_Bus Nov 30 13:52:18 I'm looking for the wiki entry which shows how to build the host dependencies for not supported linux build hosts. Nov 30 13:53:47 zzeroo: buildtools? you can't build them if you don't have the host deps ;) there's a tarball on the yoctoproject site Nov 30 13:55:23 zzeroo: what build host are you using? Nov 30 13:57:14 iirc it was arch, building morty Nov 30 13:57:56 Hi all, after patching a 4.9 (mainline) kernel with xenomai 3 patch i got this error : cpu cpu0: dev_pm_opp_set_rate: failed to find current OPP for freq 666666660 (-34) any suggestion ? Cheers Nov 30 13:57:58 Current I use yocto rocko inside a docker image. Nov 30 13:58:45 My buildsystem itself is arch-linux Nov 30 13:59:02 rocko should build on arch unless they've broken something recently Nov 30 13:59:07 (which is entirely possible) Nov 30 14:00:08 There was an wiki which shows how I setup a system, isolated from my host system.... Maybee I'm tired and miss something. Nov 30 14:00:24 https://www.yoctoproject.org/downloads/tools has the buildtools tarballs, grab the one for the release you want to target. Nov 30 14:05:49 rburton: Thank you, looks good. I think that was it 'll try it Nov 30 14:11:16 hi everyone Nov 30 14:11:48 still stuck on my problem with samba. update-rc.d runs, but the links in rc*.d are apparently not created :/ Nov 30 14:12:13 running update-rc.d manually on the booted system works, though Nov 30 14:14:58 peacememories: hi, i wasn't there at first time, can you resume which yocto version, which samba recipe? Nov 30 14:15:37 morty (pulled yesterday), and the samba recipe in meta-openembedded/meta-networking Nov 30 14:16:11 my system (right now it's a raspberrypi recipe, but the final system will be a xilinx based platform) uses sysv as init system, samba inherits update-rc.d Nov 30 14:17:05 the configuration works for things like avahi-daemon, so there's probably nothing wrong with the init system at large, but i can't find anything obviously wrong with the recipe either Nov 30 14:18:06 the update-rc.d script gets run (i tested by making it add a file to /tmp for every package it processes^^'), and the exact thing the script does works on the booted system (namely, run `update-rc.d samba.sh start . stop .`) Nov 30 14:18:16 so my debug-foo is not really strong enough for this^^' Nov 30 14:19:54 the postinst is probably breaking, maybe something stupid like a bashism in the script that fails when run with dash on the target Nov 30 14:22:18 the postinst of samba or of update-rc.d.bbclass? Nov 30 14:22:49 i mean, the init script gets installed, and the update-rc.d apparently gets run, so i'm not sure what breaks^^' Nov 30 14:33:19 peacememories> maybe backport latest samba recipe ? Nov 30 14:33:44 nayfe never done that before. wouldn't that break a lot of things? Nov 30 14:35:18 peacememories> just copy http://cgit.openembedded.org/meta-openembedded/tree/meta-networking/recipes-connectivity/samba/* in custom layer and see how it goes :) Nov 30 14:36:06 so if i copy it into the devtool workspace it should override the current recipe because the workspace has a higher priority, right? :) Nov 30 14:36:24 id just run a diff on the two recipes and see if there's anything obvious Nov 30 14:37:04 well there's a lot obvious, for example the samba version seems to have switched from a "samba.sh" initscript name to "samba" ^^ Nov 30 14:45:14 oh, backporting that might prove to be a bit more difficult Nov 30 14:45:22 seems the recipe format changed? Nov 30 14:46:12 recipe format? nope Nov 30 14:46:55 'module Nov 30 14:46:57 oops Nov 30 14:47:04 'module' object has no attribute 'filter' Nov 30 14:47:08 is what i get as an error message Nov 30 14:47:45 yeah we added a new function to oe.utils Nov 30 14:48:13 its just a convenience function though. this is why i'd look for diffs... Nov 30 14:53:15 New news from stackoverflow: python error on adding meta-java in yocto Nov 30 15:11:29 i think i may be a bit too dumb to port this^^ Nov 30 15:25:41 okay that's interesting Nov 30 15:26:06 the update-rc.d command for samba that's run in the bbclass does not output anything Nov 30 15:32:21 Should libopkg or libopkg-dev should install opkg.h in the target? It doesn't seem like the case.. Nov 30 15:33:11 I found this patch that installs the OPKG header file, but it's quite dated and no longer is part of the OPKG recipe - https://patchwork.openembedded.org/patch/51767/ Nov 30 15:37:41 Hey guys, not sure who to tell this two but a quick heads up if you're building at91 with rocko: Nov 30 15:38:12 The linux makefile hardcodes the EXTRAVERSION with an underscore. Nov 30 15:44:09 rburton okay, i've at least found the problem. at the time update-rc.d is run, the samba initscript does not exist yet Nov 30 15:44:27 that would be a problem Nov 30 15:44:43 why does the init script not exist in the package Nov 30 15:44:46 question is, why? it's copied in do_install, isn't that earlier than the postinstall in update-rc.d.bbclass? Nov 30 15:44:55 yes, postinstall happens *on target* Nov 30 15:45:04 it does exist after the build is finished at least Nov 30 15:45:24 the updatercd class stuff happens after install too Nov 30 15:50:05 * peacememories is very confused Nov 30 15:52:06 i'd say it could be that it's not properly listed in the package contents, but it's present in the sysimage after all, so it must have been installed at some point Nov 30 16:15:20 so yeah, the file really does not exist during updatecrd_postinst Nov 30 16:16:06 i've sent an email to the meta-openembedded mailing list describing the problem, in case i can't find a solution^^' Nov 30 16:30:33 also a bit weird is that yocto complains that samba did not supply license files correctly Nov 30 16:30:43 i'm not sure if that was always the case^^' Nov 30 16:37:01 can you set the unpacking directory for a tgz file at SRC_URI? like destsuffix for git fetch? Nov 30 16:39:15 Tamis: https://www.yoctoproject.org/docs/latest/bitbake-user-manual/bitbake-user-manual.html#bb-the-unpack Nov 30 16:41:54 rburton: Thanks. But this sections it seems to be missing from the MEGA manual Nov 30 16:44:06 It says that subdir is for file:// .... Nov 30 16:53:38 New news from stackoverflow: Yocto complains about perl lib mistmatch Nov 30 17:30:44 Has anyone seen an issue where u-boot reports "cc: not found" but "which cc" returns a vaild path? Nov 30 17:39:28 austen: recent bitbake's remove host paths from $PATH and only symlink in the bits we want. cc isn't part of that list, uboot should be using BUILD_CC anyway Nov 30 17:40:18 ie http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=21db71990bdaac6740873f591fc11b2e54caa217 Nov 30 17:40:22 It looks like u-boot's Makefile is using HOSTCC and $(TT) Nov 30 17:40:54 Cool, seems like I'm building an older version that hasn't been patched Nov 30 17:40:55 with master it must work, so its possible this is becase you have some BSP's custom uboot which is broken with a recent release of oe Nov 30 17:41:19 Yeah, that sounds like the problem. Let me do some quick monkey patching Nov 30 17:48:52 I have a device with an eMMC and I'm using WIC to create an image for the rootfs/boot/data/... partitions Nov 30 17:49:12 However, my SPL, U-Boot (incl. redundant copies) are located in the boot-partition of the eMMC Nov 30 17:49:29 is there any way to create two WIC images for one machine? Nov 30 18:38:24 Is it possible to disable license check for a recipe? Nov 30 19:52:33 Anyone know whats up with libgcc in recent versions? core-image-minimal (qemuarm) does not have one, core-image-sato does. It looks like libc.so.6 has some of the helpers (__aeabi_XXX etc). Does it use libgcc.a? Nov 30 20:35:52 So I'm using the meson.bbclass to try and compile a C project, but it seems that fundemental headers like sys/socket.h are missing. Does anyone happen to know what DEPENDS I need for this? Nov 30 20:47:23 overdamped: it's most likely that these are not missing (since they come from the C library which is in DEPENDS by default) - it's more likely that the build system for whatever you are building isn't looking in the right place Nov 30 20:59:36 bluelightning: I checked recipe-sysroot-native and they are indeed not there Nov 30 21:00:10 overdamped: is this a native recipe? or building for the target? Nov 30 21:00:36 Cross target Nov 30 21:00:56 then you wouldn't find them in recipe-sysroot-native, but recipe-sysroot Nov 30 21:01:07 systemd 235 to be specific. 234 doesn't work on my system and I figure I'd try upstream Nov 30 21:02:38 bluelightning: Thanks. I think you're right here. recipe-sysroot does contain the headers. I'll figure out what I'm passing in incorrectly. :) Nov 30 22:00:36 Hello, I just built an image with IMAGE_FSTYPE="container" . How do I use the build output to bring up a docker container? Nov 30 22:01:42 batman_: the image produced is a file system image, so point docker at it Nov 30 22:11:29 hello @everyone Nov 30 22:13:16 I am having issues getting Bro Network Security Monitor built within the Yocto environment Nov 30 22:15:06 I believe the issue revolves around the use of cmake. Nov 30 22:16:56 hi jdelmore Nov 30 22:17:03 what seems to be the problem? Nov 30 22:17:04 As CMake runs through the CMakeLists.txt file and checks dependencies it fails at the first one Nov 30 22:17:20 sed in this instance Nov 30 22:18:39 it is as if it is looking for sed in "/bin/" on the Target Nov 30 22:18:54 jdelmore: hmm, so the question is how is it detecting the presence or lack thereof of sed - it's one of the utilities we expect to be provided by the host and we do ensure it's in PATH Nov 30 22:19:22 the host machine has sed in path Nov 30 22:19:32 which sed returns /bin/sed Nov 30 22:20:12 with this kind of cmake issue usually you have to dig through the Find*.cmake files and see what it's doing... sometimes there will be some kind of override path you can set Nov 30 22:23:53 Do you know where Find*.cmake would live Nov 30 22:24:44 it should use find_program() to locate sed, so you should be able to get away with setting CMAKE_FIND_ROOT_PATH_MODE_PROGRAM to BOTH Nov 30 22:26:01 the CMakeLists.txt file is using find_program() and is failing Nov 30 22:29:25 What's the value of CMAKE_FIND_ROOT_PATH_MODE_PROGRAM in the CMakeToolchain file yocto generates? Nov 30 22:30:42 jdelmore: http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/classes/cmake.bbclass#n29 Nov 30 22:34:01 rburton: Thank you! I got it working Nov 30 22:35:36 neverpanic: perfect that is finding the dependencies now Nov 30 22:37:48 onto the next problem now! **** ENDING LOGGING AT Fri Dec 01 03:00:01 2017