**** BEGIN LOGGING AT Wed May 11 02:59:58 2016 May 11 08:43:28 after problems with GCC 6 I found a cool solution for separating yocto build environment from host OS - docker container May 11 08:44:03 I wanted to create it but seems someone beat me by a month: http://bec-systems.com/site/1281/using-docker-for-oeyocto-builds May 11 08:51:47 Using a container for Yocto builds is hardly news. May 11 08:56:55 probably - I did not stumble upon this earlier, and this particular docker is quite fresh May 11 10:07:57 hi is there a standard way to use multiple 32 and 64 bit toolchain in the same yocto recipe May 11 10:10:06 niteshnarayanlal: yes http://www.yoctoproject.org/docs/2.1/mega-manual/mega-manual.html#combining-multiple-versions-library-files-into-one-image May 11 10:16:09 boucman_work, thanks for the reference May 11 10:17:37 np May 11 10:17:53 it makes building images horrendously long, though :P May 11 10:31:25 Hello. Maybe somebody can help with my issue. I still get this error "multiple providers are available for virtual/kernel (linux-yocto, linux-dummy)" even tough I have a linux-yocto_4.4.bbappend in my layer where may machine is added to compatible machines var. Does anyone knows why May 11 10:38:30 boucman_work, actually I have a part of package built using 32bit tool chain and rest using 64t bit tool chain May 11 10:38:34 that's the problem May 11 10:41:11 niteshnarayanlal: a "part" of a package ? that's tricky May 11 10:41:32 hanthings_: the bbappend doesn't influence the choice of kernel provider May 11 10:42:07 boucman_work, the kernel provider is configured in my machine.conf May 11 10:42:24 so you have aline like May 11 10:42:27 PREFERRED_PROVIDER_virtual/kernel ?= "linux-toradex" May 11 10:42:37 in your machine.conf right ? May 11 10:42:56 yes May 11 10:43:18 that's weird... you sure you don't have a typo in there ? May 11 10:43:37 this is exaclty what I have in my conf May 11 10:43:38 PREFERRED_PROVIDER_virtual/linux ?= "linux-yocto" May 11 10:43:38 33 PREFERRED_VERSION_linux-yocto = "4.4%" May 11 10:44:03 I have virtual/kernel not virtual/linux May 11 10:44:17 please try that... that could be it May 11 10:45:14 let's see May 11 10:46:08 boucman_work, nice catch :) thanks May 11 10:46:40 I always read linux and in my mind was casted to kernel :P May 11 10:54:59 yeah, that's the sort of mistakes I make too :P May 11 11:00:03 boucman_work, yes actually the recipe builds two packages May 11 11:00:25 such that one is dependent on another, now I cant make separate recipes for them May 11 11:00:58 so the prerequisite package is built using 32 bit tool chain and the later one uses 64 bit tool chain May 11 11:01:06 its weird but this is how it is :P May 11 11:54:41 Hi. Have some issues with starting a fitImage May 11 11:55:04 for some reason the board stops at "Starting kernel ..." May 11 11:58:03 Verifying the fitimage.its it looks that kernel is marked uncompresssed. However in the kernel-fitimage.bbclass I've found this comment "# Override KERNEL_IMAGETYPE_FOR_MAKE variable, which is internal May 11 11:58:03 11 # to kernel.bbclass . We have to override it, since we pack zImage May 11 11:58:03 12 # (at least for now) into the fitImage . May 11 11:58:03 13 d.setVar("KERNEL_IMAGETYPE_FOR_MAKE", "zImage") May 11 11:58:03 " May 11 11:58:32 sorry the paste didn't work as expected May 11 12:00:35 how is your linux console configured ? May 11 12:00:41 what is your kernel commandline ? May 11 12:01:41 you mean bootargs? May 11 12:01:42 bootargs=root=/dev/mmcblk0p2 rw rootwait console=ttymxc1,115200 mtdparts=mxc_nand:1024k(u-boot),512k(env1),512k(env2),14m(boot),240m(data),-@2048k(UBI) nohlt panic=1 May 11 12:03:06 ok, (just checking) you are sure about this part : console=ttymxc1,115200 May 11 12:03:24 it means, IIUC, that the kernel will print its messages only on the serial line, not on the screen May 11 12:05:23 that's correct May 11 12:05:36 I mean the configuration is correct May 11 12:06:17 I'm connected to the console using the serial line May 11 12:23:42 Does anyoane knows how can i debug a non booting fitImage? May 11 12:31:01 boucman_work, if I am doing a build for 64 bit platform but want 32 bit toolchain to be built with it May 11 12:31:40 is there any direct way for it? May 11 12:36:09 billr : are you there my friend ! May 11 12:36:27 folks , i have a simple question May 11 12:37:37 i wanted to add custom splash screen to core-image-minimal , i am following a cookbook of yocto , after downloading yocto from Git , i simply issued May 11 12:38:04 Shell# source oe-init-build-env qemuarm May 11 12:38:24 the script created a folder called qemuarm inside "Poky" directoy May 11 12:39:09 now , i wanted to add psplash to that qemuarm image , so i created recpies-core directory inside "qemuarm" folder May 11 12:39:21 niteshnarayanlal: once you have configured multilib as described in the link I posted, you can add package like lib32- in your IMAGE_INSTALL and it will install the 32bit version May 11 12:39:34 so it will build the toolchain along the way May 11 12:39:34 inside that recipes-core directory , i have placed the folder called "psplash" May 11 12:39:43 (i'm not sure what you are asking for exactly) May 11 12:40:07 boucman_work, i want to add psplash to the image i am building May 11 12:40:25 snouto: i'm not sure exactly how to do that, but what you are doing sounds wrong May 11 12:40:31 boucman_work: i just wanted to add psplash to the image i am creating May 11 12:40:37 # source oe-init-build-env qemuarm May 11 12:40:55 ^ this will create a build directory called qemuarm (instead of the more common "build") May 11 12:41:03 yes May 11 12:41:15 what is wrong with that ? May 11 12:41:15 from that directory you can edit a file "conf/local.conf" May 11 12:41:20 nothing May 11 12:41:30 ok May 11 12:41:36 but you can't add recipe in your build directory, yocto won't look for them there May 11 12:42:00 where can i place them and how can i attach them and let yocto feels them ? May 11 12:42:00 snouto: you have bulid core-image-minimal already, right ? May 11 12:42:05 yes May 11 12:43:08 boucman_work: yes , i have built core-image-minimal already , but i wanted to customize it by adding a splash screen during the booting process instead of the more common linux black screen May 11 12:43:52 ok, add the following in your local.conf May 11 12:44:11 EXTRA_IMAGE_FEATURES_append = "psplash" May 11 12:44:34 okk thanks May 11 12:44:36 that should add a splash screen to your image. rebuild the image, check that it works, and then you'll look into customizing it May 11 12:44:59 boucman_work: where exactly psplash exists that yocto will look for ? May 11 12:45:31 boucman_work: thank you so much my friend , but can you answer , where yocto grab that folder called "psplash" in this case ? May 11 12:46:40 if you look into the file qemuarm/conf/bblayers.conf you will see a list of layer directories May 11 12:46:50 oh May 11 12:46:51 yeah May 11 12:47:06 each of them contains a conf/layer.conf file that explains where to find all the recipes ( *.bb and *.bbappend files) May 11 12:47:44 psplash is in the layer called "meta" under recipes-core May 11 12:48:32 boucman_work: you are awesome , thank you for enlighting me , ignorance is very bad , thank god , that ignorance , is not written in genes :) May 11 12:49:39 boucman_work: there are some errors May 11 12:49:40 ERROR: core-image-minimal was skipped: 'debug-tweakspsplash' in IMAGE_FEATURES (added via EXTRA_IMAGE_FEATURES) is not a valid image feature. Valid features: allow-empty-password dbg-pkgs debug-tweaks dev-pkgs doc doc-pkgs eclipse-debug empty-root-password hwcodecs nfs-client nfs-server package-management post-install-logging ptest-pkgs read-only-rootfs splash ssh-server-dropbear ssh-server-openssh staticdev-pkgs tools-debug tools-profile tools-sdk tools-testa May 11 12:49:40 pps x11 x11-base x11-sato May 11 12:50:15 Snert_: when using append you need to put a space before the value May 11 12:50:24 'debug-tweakspsplash' May 11 12:50:56 rburton: are you talking to me ? May 11 12:51:06 erm, yeah May 11 12:51:13 sorry, tab complete went wrong May 11 12:51:34 rburton: what should i write though ? May 11 12:52:21 rburton: something like EXTRA_IMAGE_FEATURES_append=" psplash" May 11 12:52:38 with a space before the value of psplash between quotes , that what you meant ? May 11 12:52:50 yes May 11 12:52:54 append doesn't add whitespace May 11 12:53:03 which is why you got debug-tweakspsplash May 11 12:54:06 rburton: i have made that May 11 12:54:15 but still receiving the same error sir May 11 12:54:31 yeah, sorry I always mess-up between append, += and co. May 11 12:54:42 snouto: paste as unless you didn't save you won't get that exact error again May 11 12:54:44 I have a nice cheatsheet for that, but I'm usually too lazy to open it :P May 11 12:55:06 what do you mean sir ? May 11 12:55:08 oh hang on, psplash isn't an image feature May 11 12:55:38 yeah that what the error tells May 11 12:56:13 so… don't add psplash to IMAGE_FEATURES. if you're trying to add psplash to an image then you need to modify IMAGE_INSTALL or its deriviates. May 11 12:56:33 can you paste the command May 11 12:57:10 rburton: can you paste the command for me , i am just a newcomer to yocto May 11 12:57:39 snouto: http://www.yoctoproject.org/docs/2.1/mega-manual/mega-manual.html#usingpoky-extend-customimage-localconf May 11 12:58:17 rburton: psplas is an image feature, according to the ref-manual... it's not just a package May 11 12:58:29 (I assume it changes stuff on the kernel command line or something) May 11 12:59:05 boucman_work: probably , you are using an older version of yocto, this happens between packages in open source community May 11 12:59:15 boucman_work: never mind about this May 11 12:59:18 ah, the image feature is "splash" May 11 12:59:22 oh wait, the feature is "splash" not psplash May 11 12:59:24 my bad May 11 12:59:48 boucman_work: so what about psplash ? May 11 13:00:12 the spash image feature says "have a splash screen", the default implementation of that is with psplash May 11 13:00:42 all the splash image feature does is add a single package to the image, which defaults to psplash May 11 13:01:01 so what should i place then , May 11 13:01:12 either works May 11 13:01:19 they're literally identical at the end of the day May 11 13:02:55 ok , the last question , how can i modify that image May 11 13:02:57 rburton: I would have thought the feature would influence kernel CLI or something like that... no ? May 11 13:03:17 snouto: we will answer that once you have booted an image with the default splash screen :P May 11 13:03:21 one thing at a time May 11 13:03:26 rburton: it works , the splash screen shows yocto project and a progress bar May 11 13:03:35 boucman_work: yes it worked May 11 13:03:44 i can see it using runqemu qemuarm May 11 13:03:50 #runqemu qemuarm May 11 13:03:53 ok May 11 13:04:47 boucman_work,rburton: you knew , i can't find the image itself , i think it exists as an array of bytes in C code May 11 13:05:05 how can i create such array of bytes from my custom image May 11 13:05:16 i read yesterday that there is a tool called "make-image" May 11 13:05:20 but i could not find it May 11 13:05:33 snouto: the default image is psplash-poky-img.h in oe-core May 11 13:06:09 #define POKY_IMG_RLE_PIXEL_DATA ((uint8*) \ May 11 13:06:09 "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377" \ May 11 13:06:10 "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377" \ May 11 13:06:10 "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377" \ May 11 13:06:10 "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377" \ May 11 13:06:10 "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377" \ May 11 13:06:17 this is the array , i am talking about May 11 13:06:26 grab the psplash source from git.yoctoproject.org if you want a new image. it looks like a gdk-pixbuf dump so the tools in gdk-pixbuf can generate that. May 11 13:06:38 (iirc the psplash readme cover this) May 11 13:07:21 please check /openembedded-core/meta-intel/meta-tlk/recipes-core/psplash May 11 13:07:34 it gives an idea how to override the image with a bbappend May 11 13:07:52 ok May 11 13:08:01 rburton: apparently yocto can take an actual image and generate the patch, no need to do that anymore May 11 13:08:36 yes so it can May 11 13:08:47 so just put a PNG in the SRC_URI and it does magic May 11 13:09:04 (assuming that hasn't bitrotted) May 11 13:09:30 boucman_work: where can i place the bbappend file with "files" directory that contains my custom image May 11 13:09:42 inside the qemuarm build directory that i have created using oe-init-build-env May 11 13:10:28 ? May 11 13:19:14 in your own layer May 11 13:19:38 (use bitbake-layers to create one) May 11 13:50:36 hi folks, sent a patch into meta-openembedded, is there anyone I need to kick to reivew? May 11 16:12:40 When building my image with meta-qt, it always builds qtwebkit even though I am not including qtwebkit anywhere. I used depexp for reverse dependencies but do not see anything requiring it. May 11 18:27:34 I'm working on a layer which holds mentor fixups for bsps for our use. It's not an ideal case, but it happens. There's a single meta-mentor-bsp repository, and within that is individual sub-layers for each bsp and possibly also common ones for bits shared amongst the bsps. These are not bsp layers, but only hold fixups for our distro for the bsps outside our control. The sub-layer layer names in BBFILE_COLLECTIONS are complete, i.e. mentor-bsp- e>, but I think naming the sub-layers meta-mentor-bsp-intel-corei7-64 is getting rather verbose and clunky, so I'm thinking the sub-layers won't include meta-, but the repo will. I.e. meta-mentor-bsp/common (this part is a pattern seen elsewhere, i.e. meta-amd, meta-intel), meta-mentor-bsp/intel-corei7-64, etc. Does that seem reasonable, even though it deviates from the meta- layer naming convention? May 11 18:50:14 kergoth: yes yes yes!! May 11 18:50:41 this is something i've been thinking about lately and brought up at the OEDAM at ELC May 11 18:50:50 :-) May 11 18:51:54 might not be approrpriate for this channel, but in the grand scheme of things, if you have a system which is wearing out SD cards, would it be safe to say that the driver is not performing wear leveling properly? May 11 18:52:52 kergoth: i wanted to write something to oe-arch but wanted to have a working concept in place and wait for the current dust to settle before opening a new can of worms ;-) May 11 19:21:30 kergoth: the sublayer is for you to enable it optionally? May 11 19:22:28 our setup scripts have this notion of "optional layers", where if it exists, it's pulled in, but it being missing is non-fatal, so we'll make it so when you source the setup scripts from meta-mentor, it'll check for mentor-bsp and mentor-bsp-$MACHINE and pull them in if so, along with their depss May 11 19:23:05 will have to figure out how to arrange that after more of our bits get integrated upstream with yours (which i still haven't had time to dig into, have a sick 1.5 year old at home now, and i work from home full time, and he's in a dada phase) May 11 19:30:26 kergoth: I see; and those fixes are not easy to upstream? May 11 19:30:42 kergoth: ideally those fixes should be merged and dropped May 11 19:30:56 yeah, usually they are, and in that case they go into meta-mentor-staging and go upstream eventually May 11 19:31:09 our goal is always to minimize our layers, not maintain workarounds indefinitely :) May 11 19:31:19 kergoth: regarding the tooling, it is fine. I hope you find some there soon ... but before, I wish your kid a fast recovery :) May 11 19:34:48 thanks :) May 11 20:08:41 Hello, how can I install qemu on the yocto image? (I do not want to run the kernel under yocto, I want a version of the yocto kernel with kvm and with qemu or something similar on the image). May 11 20:10:26 radens: you can look into meta-virtualization May 11 20:10:45 there must be a sample image for kvm/qemu May 11 20:12:45 radens: yes, you can build qemu for the target May 11 20:27:27 thanks May 12 00:39:14 Apologies if this is RTFM (spent 20mins searching) but what are some errors to expect if TMPDIR is on nfs? Also, I had to create an empty sanity.conf but how can I selectively just turn off the nfs check? May 12 00:40:28 (and greetings from an NVIDIA automotive team getting used to Yocto :) May 12 00:44:40 mfornander: usually nfs has issues and is generally not recommended May 12 00:46:26 mfornander: do you have any particular error messages to share May 12 00:54:18 Hi Khem, no it's working fine so far, just wondering why there isn't even an option to disable the check without disabling all sanity. May 12 01:04:33 it has had many issues reported in past and even 0 sized files May 12 01:04:38 etc. May 12 01:04:50 it was thought good to disable it May 12 01:11:41 @khem: So with a VM farm with little local VM storage (that is actually a low level NFS mount anyway) what do people do? I don't think I can convince the VM admins to give me 100+GB of storage per VM instance. May 12 01:42:11 mfornander: these are build VMs May 12 01:42:41 and you are building whole distro from source May 12 01:43:15 We usually spin VMs on demand e.g. May 12 01:43:33 which have 100G ephemeral storage **** ENDING LOGGING AT Thu May 12 02:59:58 2016