**** BEGIN LOGGING AT Mon May 15 03:00:04 2017 May 15 05:28:44 good morning May 15 05:28:54 is there a way to check if a recipe exists? May 15 05:29:04 using bitbake, that is May 15 06:40:33 Hi, My yocto image created a partition /dev/sda and inside this partition is /dev/sda1 /dev/sda2 /dev/sda3 /dev/sda4. I used the .hddimg file to flash my sd card. My sd card is a 16 GB card. the problem I have is that /dev/sda is in total 2GO not 16 GO. May 15 06:40:49 I tryed everything to exapnd my SD card but I can't May 15 06:47:36 good morning May 15 06:48:25 hamdyaea: /dev/sda is a device, not a prtition May 15 06:57:11 mckoan, : now I try to install xfsprogs May 15 06:57:46 mckoan, : the 4 partitions in /dev/sda are only 2 go and I can't resize them May 15 07:00:52 hamdyaea: go means GB, doesn't it? May 15 07:01:13 mckoan, : Yes but it's not the issue May 15 07:01:35 hamdyaea: the issue is to understand what you need May 15 07:02:08 mckoan, : I need that my root file system is expand to use all my card May 15 07:02:29 mckoan, : I tryed many things but still I can't May 15 07:05:23 mckoan, : I want to try that https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Storage_Administration_Guide/xfsgrow.html May 15 07:05:52 mckoan, : But I don't know wich one is my root root partition /dev/sda1 /dev/sda2 /dev/sd3 /dev/sd4 May 15 07:06:11 hamdyaea: what is your MACHINE ? May 15 07:07:32 mckoan, : intel64bit generic and I'm on branch jethro May 15 07:08:18 hamdyaea: I never used it, but you should have a loot at IMAGE_ROOTFS_EXTRA_SPACE option May 15 07:08:56 hamdyaea: http://www.yoctoproject.org/docs/2.0/ref-manual/ref-manual.html May 15 07:11:28 mckoan : How can I know wich one is the root partition ? May 15 07:12:02 hamdyaea : by deduction ahah. May 15 07:14:06 I will show you the output of fdisk /dev/sda and df -h May 15 07:14:10 hamdyaea : What are the name of each partition you have ? May 15 07:15:00 hamdyaea : Ok nice. May 15 07:16:04 this is the fdisk result : https://pastebin.com/BAbxJyYH May 15 07:16:47 this is the df -h result May 15 07:19:44 you miss the link for the df -h May 15 07:20:01 ChrysD, : https://pastebin.com/LpW6cmQg May 15 07:21:53 xfsgrow?!? May 15 07:23:03 are you sure you are using XFS, or did you just pick it because it sounds like it will make your FS larger? May 15 07:23:28 LetoThe2nd, : yes It was a mistake I tried but I don't have an XFS partition May 15 07:24:03 hamdyaea : the result is to much weird of what i usually have, and as i'm to newbie, sorry but i can't figure out which one it is. Mostly the partition which is the root partition is the one where the "/" is mounted on. May 15 07:24:08 stupid workaround upon more stupid workarounds. May 15 07:24:51 LetoThe2nd, : I try everything I find to expand my partition May 15 07:25:17 which will all ultimately fail because of the realroot loopback mount that points into a FAT partition. May 15 07:25:34 a complete and utter waste of time. May 15 07:26:01 LetoThe2nd, : Maybe I have to try to add that : IMAGE_ROOTFS_EXTRA_SPACE = May 15 07:26:17 LetoThe2nd, : and the amount of space of the rest of the card May 15 07:26:36 that might give you *SOME* extra space until you hit that FAT limits again. then it will fail too. just another workaround. May 15 07:27:03 LetoThe2nd, : So what's the solution ? May 15 07:27:10 the real solution is: create a wks script and have wic build your image custom tailored to the hardware. May 15 07:27:25 :q May 15 07:29:31 Hi all, Sometimes i have some trouble during kernel module compilation: The module is not built successfully because of the kernel is not built first, I tried to add some DEPENDS rules but it's not better, if someone here has some tircks or advices please let me know :) May 15 07:32:18 PinkSnake: you have inherit module in your recipe, right? May 15 07:33:23 Hello! I have a recipe building a bunch of shared libraries, with a custom builder. To build and run them I need zlib. During the do_rootfs, i have an error saying I'm copying zlib in a shared area where zlib already exists. But if I do not copy my own zlib, i have an error saying that no package provides zlib. Do someone have a clue on why ? May 15 07:34:35 top21: well do you properly depend on zlib? May 15 07:34:54 LetoThe2nd : The simple fact of inhereting module and having in SRC_URI a makefile, does it build it ? May 15 07:35:07 top21: http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-core/zlib/zlib_1.2.11.bb?h=master May 15 07:35:50 ChrysD: i don't think so. but i'd expect it to properly set up the dependencies to the kernel being ready for a out of tree build first May 15 07:36:14 @LetoThe2nd yep if i build manualy virtual/kernel and rebuild module the issue is gone May 15 07:36:30 LetoThe2nd : When you check to the poky directory with the kernel module example in recipes-kernel hello mod, in the recipe there is nothing than SRC_URI and inherit.. May 15 07:36:32 @ChrysD normally yes May 15 07:37:24 PinkSnake: interesting. let me google it, but i'd actually expect it the work too. May 15 07:38:39 ChrysD: probably this does not work as expected: http://git.yoctoproject.org/cgit.cgi/poky/tree/meta/classes/module-base.bbclass#n3 May 15 07:39:06 ah crap, PinkSnake ^^^^^^ May 15 07:40:19 just pondering who might know and explain this... May 15 07:41:03 an extract of error message https://pastebin.com/9NPFbQhQ May 15 07:42:19 @ChrysD yes i think the same but dependencies are set : # This is instead of DEPENDS = "virtual/kernel" do_configure[depends] += "virtual/kernel:do_compile_kernelmodules" May 15 07:43:02 bruce probably would know, but no idea if he's hanging out here. May 15 07:43:36 PinkSnake: i'd suggest to put this in a condensed form onto the mailing list May 15 07:47:03 @LetoThe2nd I have a RDEPENDS_${PN} += " zlib" in my recipe, but i still have an error saying no package provides libz.so. May 15 07:49:09 top21: hmhm. that pops up during the build, or when? May 15 07:49:39 Only during the do_rootfs when i'm building my final image May 15 07:49:53 LetoThe2nd : Is it possible to do thins kind of function_append(){ function_append( ){somethings there ) } } ??? May 15 07:51:06 ChrysD: ho idea. May 15 07:52:54 @LetoThe2nd the weird thing is that if I do not put a RDEPENDS in my recipe, and I copy my own libz.so in the target sysroot, I still have the same error May 15 07:53:16 LetoThe2nd : thanks May 15 07:54:42 top21: randomly copying stuff back and forth is certainly wrong. May 15 07:55:13 top21: the quesiton still is: what triggers this message? when does it arise? i'm still trying to understand the problem. May 15 07:57:30 LetoThe2nd: During the do_rootfs of my image, when installing my package containing several shared libraries, it says : Can't install no package provides libz.so May 15 08:00:33 top21: hm, ok. did you check that zlib gets built and installed properly? i see that the zlib recipe does some path optimization, so that *might* be conencted to your problem. May 15 08:01:59 LetoThe2nd: Okay I'll look into that, thx May 15 08:02:20 @top21 poky/meta/recipes-core/zlib/zlib_1.2.8.bb is enabled in your project ? May 15 08:02:46 @LetoThe2nd you are too fast for me May 15 08:03:02 PinkSnake: if the recipe would be there, then the build would abort earlier during the dependency evaluation. May 15 08:03:35 top21: so additionally, i'd look into your code how it tries to find zlib. is there some hardcoded things? May 15 08:05:32 LetoThe2nd: Okay I'll do that too May 15 08:06:55 @LetoThe2nd +1 May 15 08:09:13 LetoThe2nd: Nothing harcoded in the sources. But maybe my custom build system is doing something wrong May 15 08:09:32 top21: have to admit, this would be my suspect #1 May 15 08:10:07 top21: you could use ldd on your libs to verify that. May 15 08:16:06 LetoThe2nd: it says needed shared library: libz.so May 15 08:16:22 IMAGE_ROOTFS_EXTRA_SPACE = "11943040" I just add it to the local.conf to put more size to my sd card ? May 15 08:18:09 hamdyaea : IMAGE_ROOTFS_SIZE doesn't work ? May 15 08:19:42 ChrysD, : I have to try it again I did some branch mistakes.. May 15 08:32:45 ChrysD, : https://pastebin.com/9hwpyjPy May 15 08:33:16 ChrysD, : I have the problem that I need an hddimg and it's maximal 4 go but 4 go is enough for what I do May 15 08:33:58 hamdyaea : But why don't you do .wic ? May 15 08:35:10 ChrysD, : it don't work when I flash the .wic file the system don't boot May 15 08:36:04 hamdyaea: may I see your .wks file? Which MACHINE do you use? May 15 08:37:15 hamdyaea: if you provide those and other details in a bug I hope I can fix the issue. May 15 08:38:08 I'm working on this machine : https://www.syslogic.com/eng/sl-industrial-pccompact8-37135.shtml?c2=1&parentPageId=57548&total=1 May 15 08:38:31 ed2, : Where can I find the .wks file ? May 15 08:39:07 hamdyaea: which layers do you use? May 15 08:39:27 hamdyaea: cat conf/bblayers.conf May 15 08:39:52 ed2: cat abuser! May 15 08:39:52 hamdyaea: grep ^MACHINE conf/local.conf May 15 08:40:19 LetoThe2nd: ah come on! May 15 08:40:39 ed2: sorry. monday and i'm pre-fifth-coffee-mode. May 15 08:40:45 :) May 15 08:41:38 root@debian:~# grep ^MACHINE /home/hamdy/master/poky/build/conf/local.conf May 15 08:41:39 MACHINE = "intel-corei7-64" May 15 08:42:09 LetoThe2nd, : The friday you're the same :P May 15 08:42:10 oO( ... *weeps silently and leaves*) May 15 08:43:39 hamdyaea: $ grep WKS_FILE conf/machine/intel-corei7-64.conf May 15 08:43:39 WKS_FILE ?= "systemd-bootdisk-uuid.wks" May 15 08:45:12 hamdyaea: you can see .wks this way: $ cat scripts/lib/wic/canned-wks/systemd-bootdisk-uuid.wks May 15 08:45:24 LetoThe2nd: oops, I did it again :) May 15 08:46:12 hamdyaea: if you can show me kernel boot log I might be able to figure out what went wrong. May 15 08:46:13 ed2: i'm so shocked by the "root@..." that its fine. unless you meant to reference a certain USian pop singer in your last post. May 15 08:48:23 ed2, : there are only those file not the systemd-bootdisk-uuid.wks : directdisk-gpt.wks directdisk-multi-rootfs.wks directdisk.wks mkefidisk.wks mkgummidisk.wks mkhybridiso.wks qemux86-directdisk.wks sdimage-bootpart.wks May 15 08:48:52 ed2, : I will try to do a .wic again and burn it on my flash card May 15 08:50:00 hamdyaea: theoretically it should boot. I'm interested in finding the reason why it doesn't. May 15 08:51:17 ed2, : I just flash my card with the dd command, do I need something else ? May 15 08:51:35 hamdyaea: that should be enough May 15 08:52:16 Does someone know a good manual to use .bbappend ? I would like to override python functions. May 15 08:56:28 bluelightning: ^^^^^ May 15 09:00:27 most of time i found documentation for patching a source from a recipe, but now modifying a recipe itself with .bbappend. May 15 09:00:50 but not * May 15 09:00:55 ed2, : now I have this error while flashing the .wic ERROR: No kickstart files from WKS_FILES were found: core-image-x11.intel-corei7-64.wks core-image-x11.wks. Please set WKS_FILE or WKS_FILES appropriately. May 15 09:01:54 ChrysD: by and large there is nothing special about bbappends - think of them as being pasted on the end of the recipe as the recipe is processed by bitbake May 15 09:02:39 bluelightning : So if i would like to change a python function in a recipe, what shuold i write on my .bbappend ? May 15 09:02:53 hamdyaea: you can't have them while flashing I believe. you probably have them while building your image. May 15 09:03:25 ed2, : I am building the image it's an error when I build May 15 09:03:32 bluelightning : the question is more focused on modifying a original recipe into a .bbappend then the bbappend itself. May 15 09:03:49 hamdyaea: do you have meta-intel layer in your bblayers.conf? May 15 09:03:50 just write the function name again and your desired alternative contents May 15 09:04:15 bluelightning : It's what I have done without any result May 15 09:04:34 ed2, : yes I have the meta-intel May 15 09:04:44 bluelighting : I have write a .bbappend with the same name as my recipe, in the same kind of directories tree May 15 09:04:55 bluelighting : the exact same function name also May 15 09:05:12 ChrysD: is it listed by bitbake-layers show-appends ? May 15 09:06:00 bluelightning : yes of course May 15 09:06:28 bluelightning : I think i know what is happening.... May 15 09:06:37 right, ok so bitbake -e recipename | less and search for your function, what do you see? May 15 09:06:48 bluelightning : there is already another bbappend for the same recipe as I do May 15 09:07:01 hamdyaea: do you have WKS_FILE ?= "systemd-bootdisk-uuid.wks" line in meta-intel/conf/machine/intel-corei7-64.conf May 15 09:07:26 hamdyaea: which poky/oe release do you use? May 15 09:07:54 ChrysD: all appends will be applied if there are more than one, in ascending layer priority order May 15 09:08:24 bluelightning : yeah but still it doesn't explain May 15 09:08:31 let me check bitbake -e May 15 09:08:54 ed2, : No I don't this is what I have : https://pastebin.com/4y7sRhEr May 15 09:09:17 bluelightning : Because the other .bbappend does't affect the fucnction i wanted to change so it's not the problem May 15 09:09:52 hamdyaea: looks like you're using old release where this functionality is not implemented yet. May 15 09:10:04 hamdyaea: ok, let's do it another way. May 15 09:11:38 hamdyaea: can you add WKS_FILE = "mkefidisk.wks" to your local.conf and try to build an image again? May 15 09:11:51 ed2, : I'm on jethro branch bevause I need old softs May 15 09:12:01 ed2, : Yes I do it May 15 09:12:12 hamdyaea: and this one: IMAGE_FSTYPES += "wic" May 15 09:14:50 bluelightning : When i do bitbake -e , i see that the function i would like to overwrite have NONE inside it. May 15 09:15:26 bluelightning : But when I plug the sdcard, it works as it is in the recipe... and i have changed DIRECTLY the recipe, and it make the change into the system. May 15 09:17:00 ChrysD: er, how does the SD card come into it sorry? May 15 09:17:22 bluelightning : I wanted to patch a barebox environment May 15 09:17:35 bluelightning : and the recipe set the environment May 15 09:17:57 bluelightning : i would like to uncomment some lines from this files May 15 09:18:09 bluelightning : When i change the recipe directly to uncomment it works. May 15 09:18:26 bluelighting : but instead of changing directly the recipe, i would like in my own meta workspace, make the changes May 15 09:18:35 bluelightning : but instead of changing directly the recipe, i would like in my own meta workspace, make the changes May 15 09:19:09 bluelightning : But according to bitbake -e, the function is empty. May 15 09:19:25 ChrysD: can you pastebin your bbappend? May 15 09:22:13 bluelightning : maybe i'm getting the problem, if it is what i think it is, it's pretty stupid. May 15 09:22:53 ChrysD: well I'm still not sure what's going on, if you have an explanation then I'd be keen to hear it :) May 15 09:23:26 bluelightning : my file : https://pastebin.com/YE5fqbHL May 15 09:23:48 bluelightning : i think i had some problems with swap files and it didn't recored the changes into my file.... so very stupid. May 15 09:24:31 ChrysD: ah ok, at least you figured out the cause May 15 09:24:52 bluelightning : That was the solution T_T sorry for making you loosing time ! May 15 09:25:00 no problem May 15 09:25:38 ed2, : Now I have some error messages : https://pastebin.com/bL81BNJq May 15 09:26:00 bluelightning : the append is done after parsing the recipe right ? May 15 09:26:42 ChrysD: effectively the contents are appended to the end of the recipe and then the result is parsed May 15 09:26:46 bluelightning : if i would like to do modify some lines of a function instead of rewriting everything, could be a good way to do a python function which identify the line where to change and make the changes ? May 15 09:27:06 ChrysD: probably not worth the hassle May 15 09:27:47 ChrysD: honestly though having massive blocks of file contents pasted in like it appears to isn't ideal, a better way to have done it in the original recipe would be to have the template as a file next to the recipe May 15 09:27:56 bluelightning : Because pasting the same function, its work but maybe not readable, it could be " more beautiful " if i use python function to check the full line i would like to modify and give the result i want so that for somebody else understand what the bbappend do... May 15 09:28:24 bluelightning : but good to know that the resulst after it is parsed, so that it should work. May 15 09:28:34 bluelightning : thank you for your patience. May 15 09:28:56 hamdyaea: can you add IMAGE_FSTYPES += " hddimg" May 15 09:29:08 hamdyaea: and MACHINE_FEATURES_append = " efi" May 15 09:29:11 and try again May 15 09:29:13 ed2, : yes ok May 15 09:29:19 hamdyaea: i'm off for lunch. May 15 09:29:33 ed2, : thanks for the help May 15 09:30:57 ed2, : I leave the : IMAGE_FSTYPES += "wic" ' May 15 09:31:02 ed2, : ? May 15 09:32:27 I know that bitbake is made from python, but does we have all the " python " capabilities ? Like could we use " for example " a function from a module and import it if we need it for a recipe ? Can we code inside of bitbake as it is a full featured python ? May 15 09:33:52 ChrysD: certainly yes May 15 09:34:37 bluelightning : Ok nice, so i guess the best is to use python function like a " replace" function for doing my modifications =) May 15 09:37:03 ChrysD: bitbake functions are (bitbake) variables so yes you can manipulate them as strings from anonymous python if you need to May 15 09:37:22 depending on what you're doing it may be less obvious what is going on rather than more... May 15 09:38:46 bluelightning : So basically, doing a bbappend with a function to replace some text in a recipe file of the same name shuould works. May 15 09:42:49 ChrysD: yes, if you use anonymous python function to do d.setVar(d.getVar('do_taskname').replace()) or similar May 15 09:42:56 but it'll probably be ugly May 15 09:43:36 bluelightning : So if you consider it ugly, what should be done ? May 15 09:43:49 ChrysD: use the style from the original recipe May 15 09:43:51 bluelightning : In my case, i have a commented line which i would like to uncomment ? May 15 09:44:13 right, just repeat it with it uncommented May 15 09:44:32 bluelightning : yeah but its not readable May 15 09:44:32 as I mentioned above the original recipe probably should have done this differently May 15 09:44:42 it's as readable as the original recipe ;) May 15 09:44:59 bluelightning : the python function have 40 lines, if i copy 40 lines somebody else could not notice the difference May 15 09:45:19 hence why the original recipe shouldn't have been done like this May 15 09:45:22 bluelightning : but if i do a kind of replace (" the commented line", "by the uncommented line" ) you see what is done May 15 09:45:38 ChrysD: you do have that option yes May 15 09:45:55 bluelightning : because the recipe have been done with the purpose of changing it directly to the target May 15 09:46:05 blueligthing : but i would like to change it directly to the image that's why May 15 09:46:10 blueligtnhing : but i would like to change it directly to the image that's why May 15 09:46:37 right, I guess I'm just explaining why you're in this situation and why the system doesn't have a direct way of doing what you want - it's because it's not been done properly in the original recipe May 15 09:46:42 blueligthning : and for me is more readable to have a ligne with a python function like : "replace that line by that line ", instead of copying the full function again. May 15 09:47:06 ChrysD: you can certainly do the replace, that will work May 15 09:47:26 bluelightning : Yeah but for you it's not a best practice and i would like to know why ^^ May 15 09:48:19 bluelightning : For me the .bbappend should show only the changes made by the bbappend, so that's why the replace function, instead of copy paste the full function with the modification. May 15 09:48:55 ideal best practice would be for the original recipe author to have the file as an actual file that can be replaced easily May 15 09:49:04 then you'd just supply your modified file in the bbappend May 15 09:49:11 that would be commonly accepted practice May 15 09:49:36 bluelightning : https://pastebin.com/YE5fqbHL May 15 09:50:05 bluelightning : ah ok May 15 09:51:03 wait a sec May 15 09:51:18 couldn't you just env_add() just the uncommented lines? May 15 09:51:28 I'm assuming the order isn't that important May 15 09:52:00 bluelightning : i don't know what will be the effect of it May 15 09:52:23 bluelightning : you mean, doing the "python function_append" and writing in it the " env_add() ? May 15 09:52:25 this env_add() is completely specific to this recipe so I don't know either, but it would be worth trying as it would be the simplest way if it works May 15 09:52:30 ChrysD: yes May 15 09:52:41 bluelightning : yeah it worth a try May 15 09:54:31 bluelightning : in that case should be a do_env_append_phyboard....._append() ? May 15 09:55:33 bluelightning : but if i do a env_add() for the same file, i don't know if it will overwrite it, i will check May 15 09:55:36 ChrysD: no, the append bit is already there - it's conditional on phyboard... May 15 09:56:04 ChrysD: I'm assuming "env_add" means add this, rather than take this as the entire contents, but I haven't looked at the function May 15 09:56:20 bluelightning : yeah i will check it but thanks really for your help May 15 09:57:34 bluelightning : because when you see env_add, it's for adding a env file more than content that's why i'm wondering, but i'm bitbaking it May 15 09:57:53 Hello all May 15 09:58:46 I have been working on getting debian running in a chroot on my eink kindle paperwhite 3 and am happy to report that I can now successfully launch openbox May 15 09:58:57 But my progress has stalled due to the fact i have no keyboard input. Left and right click work out the box, I have installed and launched matchbox-keyboard from autostart but openbox doesn't seem to be listening to my virtual keyboard. Any ideas? May 15 09:59:31 pickles: to what extent is this related to the topic of this channel, if i might ask? May 15 10:02:33 Well, I was looking at using the matchbox window manager in place of openbox. I have been reading a bit about it and this seems to be closer to what I am actually needing, the problem is that it just produce a blank screen. I can get the keyboard to open under openbox, that is about where I stand May 15 10:03:35 pickles: you are aware that our discussion mainly evolves around custom distributions built with openembedded and the process of actually creating them? May 15 10:03:49 pickles: TBH, I'd say you should be better off in a debian channel. May 15 10:04:11 bluelightning : it overwrite it :s May 15 10:04:27 bluelightning : but i rewrite the entire file, it will be better May 15 10:07:14 Hey, does anyone know how i could install a binary .run file with a recipe? May 15 10:07:35 rj_: "depends" May 15 10:09:23 LetoThe2nd: as in, in the recipe file i should put it as dependency? May 15 10:09:50 rj_: no. it "depends". mainly on what the script actually wants to do. May 15 10:10:19 LetoThe2nd: Ah, it's a driver install script for a camera May 15 10:10:39 rj_: if in the worst case, it wants to unpack and do some native compilation, then you'd have to manually rip it apart and replicate the whole installation procedure in a platform-independent manner May 15 10:11:19 rj_: generally, about *ANYTHING* is superior to repackaging binary upstream scripts. May 15 10:12:32 LetoThe2nd: Hmm, what about the installer shell script that is inside the binary, would that be easily turned in to a recipe script? May 15 10:13:22 LetoThe2nd: or wouldnt that change anything as it is still doing the same thing? May 15 10:13:40 rj_: again, "depends" May 15 10:14:44 rj_: if it is only dumping a binary file at a known location, then there's no magic involved. but usually those scripts do nasty things, in nasty ways. read: not cross-compile aware, not image-building aware. May 15 10:15:54 LetoThe2nd: Alright, then i think it'll be a lot of work to get it to work as a recipe ... May 15 10:17:14 rj_: thats very well possible. May 15 10:18:18 LetoThe2nd:Although, now i'm looking at the install script. It's mostly install commands. It doesn't really seem to compile anything May 15 10:18:58 * LetoThe2nd hands rj_ a "depends" for the fourth time, if counted correctly. May 15 10:39:26 hi folks, anybody seen this: | ../ext/libsolvext.so.0: undefined reference to `db_create_rpmdb' May 15 10:39:58 when building libsolv; actually it was building meta-toolchain May 15 10:41:01 it happens with libsolv 0.6.26 May 15 10:49:06 hamdyaea: yes, you should keep 2 previous lines and add 2 more (hddimg and efi). May 15 10:49:21 hamdyaea: did it work for you? May 15 10:54:32 chris144|acurana: IIRC libsolv has a bunch of code specific to certain packaging systems; are you using RPM? May 15 10:56:48 neverpanic: actually yes, but let me check something, hold on.... May 15 10:59:02 neverpanic: ok, it should be "deb" in the conf file. I am just doing a new meta-toolchain, configuration in local.conf was not correct. May 15 10:59:18 neverpanic: thanks, you made my day :-) May 15 10:59:54 neverpanic: I'll file a bug some time. package_rpm in the local.conf should work too. May 15 11:00:39 neverpanic: BTW: I am using the master branch May 15 11:04:51 oe-core/poky master is about to get really "exciting" so if you mean that i'd switch to a release branch May 15 11:04:59 pyro released over the weekend \o/ May 15 11:05:09 saw the mail, yes.. May 15 11:05:12 * LetoThe2nd celebrates. May 15 11:05:19 DRINK! May 15 11:05:26 [x] May 15 11:06:00 https://www.youtube.com/watch?v=3GwjfUFyY6M May 15 11:06:02 [x] May 15 11:07:05 ed2, : No, I have a problem with the .wic https://pastebin.com/rj3yz3p9 May 15 11:09:21 ed2 : If I try that ? http://www.yoctoproject.org/docs/2.1/mega-manual/mega-manual.html#creating-partitioned-images May 15 11:10:21 hamdyaea: hm. can you show me the output of this command: find /home/hamdy/master/poky/build/tmp/work/intel_corei7_64-poky-linux/core-image-x11/1.0-r0/core-image-x11-1.0/hddimg/ May 15 11:11:08 hamdyaea: yes, you can try that, but it's essentially the same May 15 11:11:47 ed2, : find: `/home/hamdy/master/poky/build/tmp/work/intel_corei7_64-poky-linux/core-image-x11/1.0-r0/core-image-x11-1.0/hddimg/': No such file or directory May 15 11:12:04 hamdyaea: did you enable hddimg as I suggested earlier? May 15 11:13:24 ed2, : Yes and I add the hddimg image builded but an error with the .wic May 15 11:14:37 hamdyaea: can you show your local.conf? May 15 11:16:07 hamdyaea: wic is trying to use hddimg directory, which should be created if you have 'hddimg' in IMAGE_FSTYPES. I don't understand why it doesn't exist in your setup. May 15 11:16:36 chris144|acurana: always happy to help May 15 11:17:14 hamdyaea: wild guess: do you have rm_work enabled? May 15 11:17:26 neverpanic: thanks again! May 15 11:17:43 hamdyaea: do you have INHERIT += "rm_work" in your config? May 15 11:18:06 ed2, : Yes should I remove it ? May 15 11:18:26 yes, please, comment it out, remove tmp and rebuild May 15 11:19:54 ed2, : OK I try it May 15 11:25:49 ed2 : So now I will have hours to wait May 15 11:26:26 hamdyaea: no, you shouldn't. May 15 11:26:35 hamdyaea: if you're using sstate May 15 11:26:45 hamdyaea: everything will be taken from there May 15 11:27:17 ed2 : Yes now i'm a tasks 5157 of 5630 so it's ok May 15 11:31:32 ed2, : I have the same problem : https://pastebin.com/3jCzveTy May 15 11:32:35 ed2, : I don't know why I have this issue now with the wic format May 15 11:33:06 hamdyaea: don't worry. we'll find out why :) May 15 11:34:03 hamdyaea: can you remove wic from IMAGE_FSTYPES and try again? May 15 11:35:05 after build is done you should see .hddimg under tmp/deploy/… and hdd directory somewhere under tmp/work/ May 15 11:36:46 ed2, : OK I try it but I think it will work, the only problem is I can't expand the hddimg to use all my sd card May 15 11:40:18 halfhalo: did you try to set IMAGE_ROOTFS_EXTRA_SPACE ? May 15 11:40:48 hddimgs have a hard size limit May 15 11:41:10 you can work around it by installing the hddimg from a usb stick to a SD on the target May 15 11:41:26 iirc the install writes a full-disk partition May 15 11:42:19 hamdyaea: or by using wic, which we're trying to do. May 15 11:42:46 yeah using wic makes everything better :) May 15 11:43:15 rburton: yeah, when it works :) May 15 11:43:32 no, döts and glitter make everything better. buts that more like heavy metal topic. May 15 11:44:33 ed2, : now I wait the build.. May 15 11:45:05 hamdyaea: what did you do? May 15 11:46:36 ed2, : Comment the IMAGE_FSTYPES wic May 15 11:47:04 halfhalo: it shouldn't take long to build May 15 11:47:11 ed2, : The build is finish I have an iso, a hddimg an ext4 May 15 11:47:56 ed2, : Now I put back the wic and I try it ? May 15 11:48:17 hamdyaea: can you show me the output of this command: find ./tmp -name hddimg -type d May 15 11:49:13 ed2, : ./tmp/work/intel_corei7_64-poky-linux/core-image-x11/1.0-r0/core-image-x11-1.0/hddimg May 15 11:50:14 hamdyaea: interesting. May 15 11:50:40 hamdyaea: wic create mkefidisk -e core-image-x11 May 15 11:51:07 ed2, : I try it May 15 11:51:55 rburton: any chance to merge http://lists.openembedded.org/pipermail/openembedded-core/2017-May/136235.html ? May 15 11:52:14 yeah in the queue :) May 15 11:52:20 mut is "a bit" long after the freeze May 15 11:52:30 working through the last few snags of the next batch now May 15 11:52:57 rburton: ok, thanks. no rush here. I just thought it's somehow forgotten. May 15 11:53:32 ed2, : The wic create is finish without error May 15 11:54:23 ed2, : So now what can I do with it ? May 15 11:54:57 hamdyaea: ok, the problem may be in the order of image types. Can you try IMAGE_FSTYPES += 'hddimg wic', remove tmp and rebuild? May 15 11:55:26 ed2, : OK I try it May 15 11:55:29 hamdyaea: we're almost there. A bit of patience and we'll figure this out. May 15 12:03:44 what's the supposed way to share a .dts between kernel and barebox? May 15 12:06:32 I'd go and use CONFIG_BUILTIN_DTB... configuration fragment. I need to provide the corresponding dts but am trying to avoid a duplicate (i.e. kernel/files/foo.dts and barebox/files/foo.dts). May 15 12:09:32 ed2, : Now I just wait Currently 2 running tasks (5157 of 5630): May 15 12:11:40 ed2, : It finaly faild : https://pastebin.com/YgmCNUNd May 15 12:15:42 hamdyaea: looks like a bug to me. never mind. we'll find the way. give me a sec. May 15 12:18:09 hamdyaea: let's first check if wic image works for you, i.e. if you can increase the size to fill your sd card. May 15 12:19:00 hamdyaea: remove wic from IMAGE_FSTYPES, rebuild, then run wic create mkefidisk -e core-image-x11 May 15 12:19:13 ed2, : OK I do it May 15 12:19:24 hamdyaea: it will produce an image. you can flash it to sd and try to boot. May 15 12:21:17 halfhalo: if it boots I'll tell you how to proceed further May 15 12:30:33 ed2, : it's done, how can I flash my card now ? https://pastebin.com/MvEu1uGN with wich file ? May 15 12:31:27 hamdyaea: 1 Info: The new image(s) can be found here: May 15 12:31:27 2   /var/tmp/wic/build/mkefidisk-201705151428-sda.direct May 15 12:32:04 hamdyaea: dd if=/var/tmp/wic/build/mkefidisk-201705151428-sda.direct of=/dev/sdX …. May 15 12:32:09 ed2, : I saw it but I don't know what's a .direct. OK I flash it. May 15 12:32:58 hamdyaea: btw, which linux distro do you use? May 15 12:33:20 ed2, : Debian jessie May 15 12:34:11 ed2, : my card is flashed I test it May 15 12:35:07 hamdyaea: you can use bmap-tools instead of dd to speed up flashing. as you're going to have big images you can flash much faster than using dd. May 15 12:35:22 hamdyaea: https://packages.debian.org/jessie/bmap-tools May 15 12:35:53 hamdyaea: not now probably. Let's finish this excercise first :) May 15 12:36:39 ed2, : The system don't boot May 15 12:37:06 hamdyaea: can you show boot log? May 15 12:37:57 ed2, : OK May 15 12:38:29 ed2, : There is no logs, the system simply don't boot May 15 12:39:14 hamdyaea: can you show me what's on the boot partition of it? May 15 12:39:41 hamdyaea: mount /dev/sdX1 /mnt && find /mnt/ May 15 12:40:34 ed2, : But from wich computer ? it simply not booted May 15 12:40:52 ed2, : I connect the card to my computer and try this command ? May 15 12:40:57 hamdyaea: yep May 15 12:41:26 hamdyaea: you should replace /dev/sdX with your device name May 15 12:42:32 ed2, : this is the result : https://pastebin.com/NMAbiLL4 May 15 12:43:03 hamdyaea: looks good so far. cat /mnt/EFI/BOOT/grub.cfg May 15 12:45:07 hamdyaea: do you have access to the BIOS? does it support EFI boot? is it enabled? May 15 12:46:13 ed2, : this is the result : https://pastebin.com/g2BRAv1Y May 15 12:46:25 ed2, : I look at the bios May 15 12:47:31 hamdyaea: now i see what's going on. Can you show your .wks? cat ../scripts/lib/wic/canned-wks/mkefidisk.wks May 15 12:55:08 ed2, : In the bios I have foud nothing about the efi I think it's not supported May 15 12:56:02 hamdyaea: ok, then let's switch to non-efi .wks May 15 12:57:00 ed2, : this is the result of the cat : https://pastebin.com/R0ZPdnRW May 15 12:59:30 hamdyaea: wic create directdisk -e core-image-x11 May 15 13:01:17 ed2, : now I flash the .direct May 15 13:01:32 yes May 15 13:07:30 ed2, : it's booted but It don't seems to use all the card May 15 13:09:23 hamdyaea: good. now you can make it bigger. you can specify partition sizes using —size in .wks file May 15 13:09:47 hamdyaea: you can also create separate partition for data if you prefer not to use root partition for everything May 15 13:09:51 ed2, : the network card also don't connect to my network May 15 13:10:35 hamdyaea: I'd recommend creating your own .wks file May 15 13:10:58 ed2, : in my local.conf I added that to make work my network card : MACHINE = "intel-corei7-64" May 15 13:10:59 BSP_SUBTYPE="leafhill" May 15 13:12:11 hamdyaea: yep, wic used rootfs prepared by your build. whatever you configured should be there. May 15 13:13:38 hamdyaea: btw, you don't need to add efi to MACHINE_FEATURES as you're not using it May 15 13:21:38 hamdyaea: as for networking problem I'd suggest to use latest poky and meta-intel. wic images are built by default there and hopefully contain everything you need. May 15 13:28:10 Hi there - i am looking for a combination like this: low cost controller like vybrid from NXP with yocto linux (with preempt-patch) for the dual core running .NET Core onto the yocto linux OS to enable .Net Framework. The real time HMI Application will be written with C# and the HMI with XAML or HTML5 May 15 13:28:18 Is this allready possible? May 15 13:34:45 Guest11289: certainly nothing readymade. May 15 13:35:38 Guest11289: it doesn't sound impossible, but very special, so you'd probably find a som vendor that is willing to support it for you. May 15 13:36:02 hi everybody, would anybody help me in changing a unit systemd file in my Yocto build ? May 15 13:36:25 Guest11289: there's a .net layer at http://git.yoctoproject.org/cgit/cgit.cgi/meta-mono so thats the software bit sorted, you just need to sort out the hardware support May 15 13:36:37 I need to customize the command line option given to networkd-wait-online service May 15 13:37:31 Guest11289: especially "real time hmi application in c#" has a boatload of obstacles, doubt that anybody is willing to guarantee that OOTB. May 15 13:38:18 but I can't even find the file in my YOCTO tree .. I am naively searching for find ./ -regex '.*\.service' | grep online but I can't find anything May 15 13:39:17 Thx rburton May 15 13:41:04 @LetoThe2nd As i understood .NET Core, the compiler will create machine code not interpretable code. Based on my understanding, real time app with C# can be possible May 15 13:42:19 Guest11289: i did not say impossible :-) but just having native code and a preempt rt kernel does not automagically make anything realtime. and if there is custom drivers involved, things get even more fishy May 15 13:43:07 Guest11289: so your best bet is really to look for some system vendor that says "we sell you this RT-enabled .NET thing" and get it. May 15 13:43:53 LetoThe2nd: I'll try to, thanxs May 15 13:44:26 Having two boards that communicate, with one having a OS and another with a RTOS, is most of the time the best solution if realtime is very critical. May 15 13:44:49 ChrysD: it "depends" May 15 13:44:57 ed2, : I will try to connect with ssh and resize it May 15 13:45:12 LetoThe2nd : I don't think that xenomai for example is enough reliable when timing is critical. May 15 13:45:19 ChrysD: it "depends" May 15 13:46:17 LetoThe2nd : ye? May 15 13:47:34 ChrysD: absolutely. saying that something "is reliable" or "is not reliable" without knowing any context, and the assumptions people make based on the expression "realtime" is totally useless May 15 13:47:45 LetoThe2nd : I said " critical", but didn't say that impossible May 15 13:49:00 ChrysD: no. you said "LetoThe2nd : I don't think that xenomai for example is enough reliable when timing is critical." given that sence, you imply that any application that has any form of "critical" realtime constraints cannot be implemented using xenomai May 15 13:49:31 ChrysD: its all assumptions over assumptions. what do you define to be "critical". what do you define to be "realtime" May 15 13:51:55 LetoThe2nd : I know what you mean. May 15 13:52:20 folks, would you know what are the needed steps to customize a systemd-* unit file in a YOCTO build ? May 15 13:52:30 LetoThe2nd : You can screw even with a saw. May 15 13:53:18 ChrysD: so, without knowing the actual problem, my opinion is that anything one could say about suited for RT or no or whatever is totally worthless. May 15 13:55:18 pedr0_: by looking at the systemd recipes, my guess is that the file just does not come from the poky build, but from systemd source itself May 15 13:56:33 pedr0_: as networkd is generated as a subpackage of systemd. so you'd have to either patch it for your case, or modify on the fly, or something like that. May 15 13:57:11 recipes-core/systemd/ May 15 13:57:20 does it look like the right folder to look at ? May 15 13:57:41 is there a guide of this subpackaging or patching ? May 15 13:58:50 pedr0_: googling it a bit confirms that this comes from upstream. May 15 13:59:50 therefore I need to write a patch which modifies the downloaded files - don't I ? May 15 14:00:03 pedr0_: is there not a way to provide unit overrides in systemd using "/etc/systemd/.service.d/.conf" or similar? May 15 14:00:13 pedr0_: thats one option, exactly. May 15 14:01:01 the logic must be hidden somewhere in http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-core/systemd/systemd_232.bb May 15 14:01:12 but i don't have the time to dig through that at the moment May 15 14:01:46 sure, thanks a lot anyway May 15 14:05:19 pedr0_: yer, so if you don't mind the overrides being processed at run time you can use the "drop-in" files directories in /etc/, see https://www.freedesktop.org/software/systemd/man/systemd.unit.html and search for "drop-in". May 15 14:05:36 ed2, : Now my system use all the card (15go) But /dev/root only 1.7 GO : /dev/root 1.7G 1.4G 224M 86% / May 15 14:05:45 ed2, : How can I resize it ? May 15 14:06:04 if its a very simple change, maybe even a bbappend that postinstall-modifies the file suffices. like a sed. May 15 14:06:33 hamdyaea: you can use —size in .wks file to set partition size May 15 14:07:30 hamdyaea: or —extra-space or —overhead-factor May 15 14:07:43 hamdyaea: wic help kickstart May 15 14:08:12 ed2 : OK thanks May 15 14:13:22 nrossi: thanks I am having a look May 15 14:14:41 but I can't find the main recipe which specifies the URL where systemd is cloned from May 15 14:15:14 I have got a systemd_225.bb May 15 14:18:44 ed2, : I use like this ? wic kickstart --size 15G then wic create mkefidisk -e core-image-x11 ? May 15 14:20:47 ed2, : i modified the wks with nano to see now May 15 14:23:52 hamdyaea: yes, you should modify .wks. BTW, It's better to create separate .wks file May 15 14:24:38 hamdyaea: you can use it locally or add to your layer if you have one. May 15 14:27:51 ed2, : Thanks I try to do it now May 15 14:29:09 What does "install -d" do ? thanks. May 15 14:29:12 hamdyaea: if you want to use it locally you should run wic this way: wic create ./ -e core-image-x11 May 15 14:29:35 ed2, : yes I see May 15 14:29:54 hamdyaea: btw, how are you going to use that extra space? May 15 14:31:09 hamdyaea: you can create empty partition and use that if you want. It's safer than using root partition. May 15 14:32:04 does install -d is doing to the target as mkdir is done on linux systems ? May 15 14:32:32 hamdyaea: and don't forget about bmap-tools. It will save you a lot of flashing time. May 15 14:32:33 ed2: which actually would create a couple between wks and recipes, if the fstab is modified accordly, right? May 15 14:32:52 *coupling May 15 14:33:55 ChrysD: the crucial point is that install can set owner and mode: https://linux.die.net/man/1/install May 15 14:34:05 LetoThe2nd: not sure I understood, sorry. can you elaborate? I don't see how it would create coupling. May 15 14:34:52 ed2: once you add an additional partition for data, you gotta tell the system how to use it - which usually is some form of fstab entry. May 15 14:34:53 LetoThe2nd: wic modifies fstab to get partitions mounted automatically, yes. May 15 14:35:06 ed2: aaaaah now that was the missing piece. thx. May 15 14:35:40 LetoThe2nd : Thanks. So we can copy with install ? Because in most example i see the use of cp May 15 14:36:19 ChrysD: for the sake of reproductible owners and rights, you should use install - thats my understanding. May 15 14:37:14 LetoThe2nd : yeah i know, it was i use to copy " a file " but everytime i see in internet example to copy a "directory " cp -r is prefered instead. So I wanted to know more about the "install " function. But thanks. May 15 14:38:04 ChrysD: the main difference is if you want to copy a directory recursively, or not. cp can do it, install can't AFAIK May 15 14:38:22 LetoThe2nd : ohhh so that's the point. May 15 14:38:54 (but don't call me out if that was wrong, just as far as i remember it) May 15 14:39:08 LetoThe2nd : ;) May 15 14:52:18 is there a recipe which uses systemd drop-in that I can use as an example ? May 15 15:14:27 I have nothing in my mkefidisk.wks is it normal ? May 15 16:11:15 is there a complete idiots guide to adapting your image to rpi-3 and creating bootable media? May 15 16:12:58 add meta-raspberrypi to you bitbake build? May 15 16:13:14 set machine = raspberrypi3 May 15 16:13:30 flash the sdimg to an sd card May 15 16:13:45 (thing will probably going wrong :) ) May 15 16:13:50 *things May 15 16:17:59 no sdimage :) May 15 16:22:40 CTtpollard, I can't believe tlwoerner would let anything go wrong May 15 16:24:13 Crofton|work: i'm not a meta-raspberrypi maintainer (yet?) May 15 16:24:27 you seem to pay attention :) May 15 16:24:48 in any case there should be a .rpi-sdimg in tmp/deploy/images/raspberrypi3 ? May 15 16:25:58 Crofton|work: thank you :-) May 15 16:29:19 IMAGE_FSTYPES_append_raspberrypi3 = " rpi-sdimg.xz" May 15 16:29:25 I had to put this in local.conf May 15 16:29:46 gnuradio-dev-image-raspberrypi3.rpi-sdimg.xz May 15 16:45:52 ok, now I just need an actual borad May 15 16:51:07 Crofton|work: for bonus points you could also try setting MACHINE to raspberrypi3-64 May 15 16:53:21 I need something to work first May 15 16:53:29 will 2d display work for that? May 15 18:12:37 ed2: your git ls series doesn't apply for me, does it depend on anything else? (alternatively i've something locally it conflcits with) May 15 18:14:35 rburton: yes, it depends on argparse series. May 15 18:15:30 oh its failing in selftest for some reason May 15 18:15:39 rburton: tests probably wouldn't apply without generic EFI series May 15 18:15:41 already got the argparse ones May 15 18:15:41 ah May 15 18:15:52 ah yes, there it is May 15 18:16:18 rburton: may i see how it failed tests? May 15 18:16:40 rburton: i run all wic tests before sending for review and they all passed. May 15 18:17:19 rburton: i have all my patches in ed/wic/wip. you can pick them up from there if you want. May 15 18:17:26 merging, i mean. no doubt because of the efi stuff. May 15 21:07:23 Hey friends. Can someone tell me, I am building a system for an ARM7 device. I want to include gcc to compile programs on/for the system. Should I be using the 'gcc' or 'gcc-arm-none-eabi-native' recipe? May 15 21:09:29 try the tools-sdk IMAGE_FEATURE :) May 15 21:19:12 was there any success at crossing for windows? I've found meta-mingw, but that's for creating an sdk May 15 21:33:08 Crofton|work: Thanks! May 15 22:35:00 Is there anyway to include a full rootfs in an SDK installer? Its my understanding that only dynamic libraries and other developement requirements are included in the SDK May 15 22:42:39 rschaefe: what the aim of that? May 15 22:46:21 To have everything needed to create a final image in one spot. I can build the application using the SDK, but then also need to grab the rootfs seperately in order to create the final image with the application included. May 15 22:47:43 rschaefe: well, the extensible SDK can build images including any application(s) that you're working on, maybe worth a try May 15 22:49:58 Thanks I'll give it a look! May 15 22:50:56 Also, I meant "create the final rootfs with the application included". Currently I'm building linux and u-boot outside of the yocto project May 15 23:20:29 rschaefe: ah ok, for the eSDK to be able to build an image I think you'd need to be building those things within the build system May 15 23:21:30 though it might depend on your partition setup, if the rootfs and kernel image are separate and the rootfs contains no kernel modules then you might get away with it May 15 23:23:54 What's the best way, if there is one, for using yocto only for the rootfs and sdk? May 15 23:25:22 bluelightning: I think eventually I'll include the kernel and u-boot in the build, but for now am only using it for the rootfs and sdk generation **** ENDING LOGGING AT Tue May 16 03:00:02 2017