**** BEGIN LOGGING AT Thu May 12 02:59:58 2016 May 12 10:05:49 hi, how hard would it be to create an image containing multiple kernels? May 12 10:13:30 I am thinking about fit image with eg. 3 zImages built in May 12 10:16:33 is some solution for this scenario available or must be implemented from scratch? May 12 10:25:03 (not sure if this is a yocto or uboot question) how is the default environement changed in u-boot ? is there a way to do that in a recipe ? May 12 10:34:28 Hello. I have one recipe that provides 2 packages. It is possible to specify that only one of these is native? May 12 10:49:45 how do I specify I need a given development tools to build, doesn't seem to fit with depends or rdepends. May 12 10:49:52 in this case unifdef. May 12 10:52:48 DEPENDS = "unifdef-native" does the trick May 12 11:12:24 Hello, Is there any way to undo the effect of a PROVIDES_remove="xxx" in a different layer?? May 12 11:13:06 shiftee_: you could do a PROVIDES_add .. as long the ordering was right, you would be ok ... feels like a bad way to fix this. May 12 11:14:38 i'd figure out why it's forcibly removed first... sounds like something fishy is happening May 12 11:15:56 They're using a kernel module to provide imx gpu support. I want to use linux-yocto kernel and no modules. May 12 11:16:33 I have tried PROVIDES_append_atlas = " dog" PROVIDES_remove_atlas = " dog" PROVIDES_append_atlas = " dog" but dog is still removed May 12 11:17:09 remove might have priority on everything else... check the bitbake documentation May 12 11:17:19 (and I think you change kernel with May 12 11:17:37 PREFERRED_PROVIDER_pn-virtual/kernel=linux-yocto May 12 11:17:43 but I have never tested that May 12 11:18:48 Bitbake docs dont seem to specify. Kernel builds fine - am using preferred-provider option, thanks May 12 11:24:08 Essentially _remove trumps _append regardless of order. Any ideas for adding something after this point?? May 12 11:38:39 shiftee have you tried XXX_remove_remove ? :P May 12 11:38:53 (probably won't work, but that's the only hting I can think of) May 12 11:40:37 I will try it, thanks - it's pretty crazy :) May 12 11:54:45 hi there. May 12 11:55:00 is this a proper list to ask about creating packages in bitbake? May 12 11:57:15 there are many places you can ask, but this will do for now :) May 12 11:57:48 Ok. I have a problem, because I'm trying to install a file into a package I created, but it fails during install with "opkg_install_cmd: Cannot install package XXX-aa" where XXX is name of my main package and aaa is a subpackage added with PACKAGES+="${PN}-aaa" May 12 11:58:07 now be patient :) May 12 11:58:32 lewiatan: check that the package didn't get renamed, or that it got generated in the first place May 12 11:58:46 but that's not everything :-) May 12 11:58:50 lewiatan: (have a look in tmp/deploy/ for the package feeds) May 12 11:59:47 I have 2 files in my git repository and I want to install them in the same path in 2 different packages May 12 12:00:37 less than trivial. why? :) May 12 12:00:47 have a single package that contains that shared file May 12 12:01:26 so I have added to do_install_append() { install -d ${PKGDEST}/${PN}-aaa/etc/XXX/aa.file ; install -m 0644 ${S}/file1.from.repo ${PKGDEST}/$PN}-aaa/etc/XXX } May 12 12:01:33 ewwwwwwwww May 12 12:01:55 I'm not sure if I have explained clearly. May 12 12:02:03 so that's lovely but i'd put money on do_package() deleting PKGDEST before it starts May 12 12:02:32 do you think I can put that in do_package_append()? May 12 12:02:39 what are you actually trying to achieve? May 12 12:03:18 basically my goal is to have one package, say XXX and two sub packages XXX-aa and XXX-bb with some config file which I will add to proper image May 12 12:03:26 because I also have two images May 12 12:03:33 both use package XXX May 12 12:03:46 but one needs config files from XXX-aa and the other from XXX-bb May 12 12:04:28 but thats the same config file name? May 12 12:04:55 you'll need a recipe for each one May 12 12:05:07 five-line recipe to grab the git repo and install the right config file to the right place May 12 12:05:15 in my repository it has two different names but I want to put it in the same path May 12 12:06:25 ok. This will do, but are you sure that I cannot do it inside one of my current recipes? May 12 12:06:27 yes May 12 12:07:15 ok. thanks May 12 12:18:04 @rburton: it worked, but with all AUTOREV, etc. it's 19 lines ;-) Thanks May 12 12:36:05 the fewest the better :P May 12 13:16:49 Hello everybody, I'm building a custom AM335x hardware who works fine. Our OS came from yocto fido release + meta-ti layer. I've a strange issue with u-boot-ti-staging version 2015.07... the SPL (MLO file) doesn't work and doesn't display anything. Every works fine with the poky u-boot. Any idea ? May 12 14:17:06 Hi. I have a python script which depends on python-systemd --> https://github.com/systemd/python-systemd May 12 14:17:15 How do I package this while building image May 12 14:17:32 you write a recipe for it May 12 14:17:55 it uses setuptools etc etc, should be trivial May 12 14:23:27 WARNING: weston-1.10.0-r0 do_fetch: Failed to fetch URL http://wayland.freedesktop.org/releases/weston-1.10.0.tar.xz, attempting MIRRORS if available May 12 14:23:30 ERROR: weston-1.10.0-r0 do_fetch: Fetcher failure: Fetch command failed with exit code 4, no output May 12 14:23:40 master-next seems to point to a missing weston tarball May 12 14:25:18 @rburton Thanks. Will try that May 12 14:26:28 otavio: works here May 12 14:26:55 4 is network failure, so maybe some transient network woe? May 12 14:48:05 halstead: can we pune poky-contrib tree ? its gigabyte+ to clone May 12 14:49:13 @rburton Can you please help me with this? --> http://pastebin.com/u2AB2c7k May 12 14:51:03 kteza1: looks like the setuptools assumes that the headers are in /usr/include, when they are not for your build. you'll need to fix the build scripts May 12 15:18:20 can anyone submit to errors.yoctoproject.org ? May 12 15:18:39 yes May 12 15:19:16 cool May 12 15:20:31 * armpit has been "finding" h/w to build a build farm at work May 12 15:27:45 armpit: you'll regret that when you get the power bill :) May 12 15:27:55 hm how did i read "work" as "home" May 12 15:27:57 ignore me :) May 12 15:34:51 I had a bunch of high-end Dell workstations that had been retired.. I figured out why.. May 12 15:35:02 none of them would power up.. I finally got 5 machines and built one working one.. May 12 15:35:07 then about a week later it failed too.. May 12 15:35:25 (apparently dell must have designed in a 3 year life span into these machines.. they hit about that and started to all fail) May 12 15:35:39 armpit: we've build a build farm using lots of automation at WR. puppet and stuff for base configs, docker for testing distros, mesos for dispatching jobs to the cluster. May 12 15:35:41 shame too.. dual quad core 3 GHz xeons w/ 64 GB of ram each May 12 15:36:19 (they all ended up in the e-waste now... hard drives in the end were the only thing useful I got out of them..) May 12 15:36:30 fray: that's what the management engine is for May 12 15:36:40 everything from RAM to power supplies, to main boards were flaky.. (and I didn't have any other motherboards that could use the CPUs) :( May 12 15:39:11 (I stongly suspect that the machiens were made with really bad capacitors that just wore out) May 12 15:39:12 MV has their own build farm but they don't share. I am doing OE/Yocto builds for the branches I maintain. systems located at work so now electric bill May 12 15:39:26 :) May 12 15:39:38 that is what I was trying to do.. (I was going to play with fossology and such.. that all fell apart) May 12 15:39:51 I wonder the halls and pick up new system ; ) May 12 15:39:52 I think our next big workstation upgrade cycle is coming up.. so I may get to try again May 12 15:40:21 lol, my 'halls' are much smaller then yours.. (we've got 4 workstations in the little office I'm in.. and but there are only two of us in the office :) May 12 15:41:03 (and the ahlls going to get smaller.. work is not going to pay for my office much longer, so I have to move everything home.. which means -my- power bill, which means fewer machines..) May 12 15:41:24 I work from home most of the time so on hall here May 12 15:41:38 I share the office with the dogs May 12 15:42:00 my office is 2 miles from my house in a small town in southern MN.. but I work out of my house much of the time.. the office is really machine storage and server lab.. May 12 15:42:14 (I might end up paying fo rthe office out of pocket for a while to keep the machines out of my house).. May 12 15:42:29 I dont have air conditioning [other then window units] so machine heat is a concern until Fall.. :P May 12 15:42:45 MV is 1.5 hours from my house and go in 2x/week May 12 15:42:57 I'm starting ot investigate split units or a remodeled central air system.. (I've got hydronic / raditators now.. so no duct work) May 12 15:42:58 I have a pool : ) May 12 15:43:11 lol, I have 20 acres, a barn and a marsh.. :) May 12 15:43:27 I couldn't afford to even look at 20 acres in your area... ;) May 12 15:43:36 nope May 12 15:43:45 (not sure i could afford 1 acre) May 12 15:43:47 you could have a pond May 12 15:43:56 yup, I could.. May 12 15:44:31 (my neighbor has 3 ponds.. he's selling his land for $500k [100 acres].. seems a bit pricey to me.. and needless to say, I don't have $500k) May 12 15:45:01 that is cheep May 12 15:45:09 land in this area (that is GOOD for farming) is going for $3k an acre.. typical small property is around 5-7 acres here.. May 12 15:45:32 the land he has is aweful for farming, and really is only worth -maybe- $1k-$2k an acres for most of it.. May 12 15:45:48 of the 100 only about 20 acres of it is really good land.. everything else is marshy and not a good building site even May 12 15:46:14 huh, that's cool, don't think i knew you were from MN. most of my family is in the minneapolis suburbs May 12 15:46:29 (was talking to another neighboar.. he said the land is work -maybe- $350k... but my neighbor put it up for $500k looking for an idiot who wants a large property and has money to burn) May 12 15:46:40 some days i miss having an office to go to that's outside the house.. particularly now that my 1.5 year old comes running down the hall to bang on my office door on a regular basis :) May 12 15:46:47 I grew up in Bloomington, near the airport/Mall of America May 12 15:46:55 :) May 12 15:47:43 I'm south now closer to Northfield.. (i.e. the place that fought back against Jesse James many years ago).. May 12 15:47:54 (they've made movies about it, so some people have actually heard of it) May 12 15:49:38 kergoth: when we move we're converting the largish garage into 1/3rd office, so my working space isn't off the main living area May 12 15:49:55 about 8 years ago.. I had the choice [bad housing market] to either buy 1/3 acre space in Bloomington and do a big remodel to get what I wanted.. or move 'south' out of the cities and get 20 acres, no real neighbors and a bunch of 'projects'.. May 12 15:49:58 I voted the later.. May 12 15:50:10 8 years layer, I'm still working on projects.... :) May 12 15:50:21 and nobody complains if I'm doing construction at 10pm, or working on a car or... May 12 15:50:56 rburton: nice May 12 15:51:14 (in bloomignton I used to have a next door neighbor that yelled at me for taking out my trash at 9pm, cause I was too noisy.... ugh.. really?) May 12 15:51:29 one of the reasons I got the hell away from people.. May 12 15:51:33 that's pretty sad. 2am, i could see. 9, that's sad May 12 15:54:31 yup May 12 15:55:10 I admit I wasn't being 'quiet' about it.. but they didn't have any kids.. and I had to get all of the trash in the can and then wheel the can to the curb.. :P May 12 15:55:37 that and the neighbor who was behind me was an asshole.. (the one next to me was nice, I just didnt appreciate him to accepting 10 minutes of noise once a week) :P May 12 15:56:26 sitting in your bath robe until noon isn't as great as people would think May 12 15:56:35 HAHAHA May 12 15:57:06 Crofton|work: the trick is to never get out of it May 12 15:57:15 My wife gets unhappy May 12 15:57:25 rburton spoken like a true work from home employee.. :) May 12 15:57:32 need to go ride bike soon and put on real clothes May 12 15:57:34 and if you need video conferencing, just put on a clean shirt May 12 15:57:48 G+ default to cam off now May 12 15:57:53 thank god May 12 15:57:54 lol May 12 15:58:45 we're still old school and doing voice-only meetings May 12 15:59:03 so then you just have to remember to mute before flushing May 12 15:59:30 (pretty sure it was tf_ who learnt that the hard way) May 12 16:03:44 lol May 12 16:05:47 :) May 12 17:07:05 rburton: I sent the rename for https which fixed it for me May 12 18:03:26 seebs: ping May 12 18:10:35 ack May 12 18:10:41 it's user error, pseudo is fine May 12 18:10:46 hehe May 12 18:10:53 ... hey, it's worked a couple of times May 12 18:11:01 i mean, the tactic. not pseudo. pseudo never works. May 12 18:11:05 seebs: i sent a patch to the list about having pseudo obey LDFLAGS from the env, mind taking a look? it's on oe-core, but i forgot to cc you May 12 18:11:27 For the build or at runtime? May 12 18:11:51 build May 12 18:11:55 just a makefile patch, trivial May 12 18:12:06 That looks eminently reasonable, yes. May 12 18:12:14 k, thanks May 12 18:12:36 So I have this pending issue, which is that I have some large and intrusive changes to pseudo which I want to merge, but for which I need to find code review, and it turns out that the pool of likely reviewers for large complicated changes to pseudo is limited. May 12 18:12:48 I will probably have to actually do something about this, except I always have Other Tasks that I need to look at. May 12 18:13:17 i have that problem with bitbake and parts of oe-core, richard ends up the bottleneck there. and at work, any nearly core change basically has no review, it makes me sad :( May 12 18:15:22 Normally I'll just say "hah, whatever, I know this code pretty well", but the rework of how the client gets launched is above my self-confidence, and that is saying something. May 12 18:15:49 It's way better, though. Also I finally comprehended the extremely strange problem causing the daemon to get launched inside the pseudo environment, if and only if the client was in a bash process. May 12 18:15:56 Turns out bash redefines setenv(3). May 12 18:16:06 For, terrifyingly enough, sound technical reasons. May 12 18:16:26 Hello - I'm trying to make an image that is a PXE server that can serve another image. May 12 18:16:50 Right now, I'm getting " * opkg_prepare_url_for_install: Couldn't find anything to satisfy 'my-client-image'." May 12 18:17:13 How can I place output files of one core-image recipe in a location in the sysroot of another core-image? May 12 18:18:11 the best you can do is add do_rootfs[depends] += "some-other-image:do_image_complete" and then grab the other image file from DEPLOY_DIR_IMAGE, IMO May 12 19:04:35 hi yocto team, how to avoid running update-alternatives for a recipe? May 12 19:08:35 my yocto image for virtualbox is not booting because the /boot/bzImge is a symlnk to /etc/alternatives/bzImage May 12 19:09:02 that made by update-alternatives May 12 19:09:42 So i want to avoid run update-alternatives for linux recipe custom written May 12 19:57:41 i am trying to find the source of the /etc/network/interfaces layer. In my build src dir, I did a find . -name interfaces and it only shows up in my tmp/work dir. Is that normal? May 12 19:59:13 if you have package managent, use that to find out which package installs the file and then use oe-pkgdata-util lookup-recipe with the package name May 12 19:59:41 iirc its netbase May 12 19:59:58 but yes, it would be expected that you'll only find it in tmp/work if it's in some tarball May 12 20:00:00 davis: It's init-ifupdown May 12 20:00:15 you can search packages using oe-pkgdata-util, no need to involve the target May 12 20:00:18 You can find it in poky/meta/recipes-core/ May 12 20:02:01 many thanks May 12 20:04:41 kergoth: Trying to copy in that image - when I define do_rootfs_prepend() in my core-image recipe, I get a Syntax error, but it's hard to tell exactly where. May 12 20:07:10 kergoth: Otherwise, is there a way I can add DEPLOY_DIR_IMG as a source directory for another recipe and other_image:do_image_complete as a dependency for the same recipe? May 12 20:20:47 Circuitsoft: do_rootfs is a python task. you can't prepend shell to it. use prefuncs instead May 12 20:37:36 kergoth: What variables are available in do_rootfs[postfuncs] that point to the directory the rootfs is being built in? May 12 20:38:06 Or just point me to where they're called from (I can't find it). May 12 20:38:12 the current, or the one used by the other image? the rootfs is constructed in a path relative to WORKDIR, isolated to the recipe in question May 12 20:38:17 did you actually read the bbclasses? May 12 20:38:28 I'm reading some of them. May 12 20:38:47 I've been grepping them for postfuncs and do_rootfs, but I'm having a heck of a time trying to figure out what environment they're actually called with. May 12 20:39:07 The current image. May 12 20:39:23 in this case, you're having trouble finding it because the var is defined in bitbake.conf,a nd it's used in meta/lib/oe/ python modules, not directly in the bbclass May 12 20:39:32 it's IMAGE_ROOTFS, which is ${WORKDIR}/rootfs May 12 20:39:32 I've tried IMAGE_ROOTFS, but that doesn't seem to be it. May 12 20:39:36 Ah. May 12 20:39:37 no, that's it May 12 20:39:40 Huh. May 12 20:39:45 but prefuncs is before the task is run, not after May 12 20:39:49 hence "pre" May 12 20:40:01 use postfuncs to run something after th erootfs is constructed but before the image is built May 12 20:40:09 I'm trying "do_rootfs[postfuncs] += my_function" right now May 12 20:40:10 or one of the rootfs/image construction hooks May 12 20:40:19 i.e. ROOTFS_POSTPROCESS_COMMAND, etc May 12 20:40:31 But my outside image is still way smaller than the inside image. May 12 20:40:44 So, clearly, the inside image rootfs is not showing up in the outside image yet. May 12 20:53:06 I just added a call to xmessage to my ROOTFS_POSTPROCESS_COMMAND and it's not getting called. May 12 21:01:04 ROOTFS_POSTPROCESS_COMMAND is a list of functions, you can't put actually shell in there May 12 21:04:26 Hi guys!!! May 12 21:04:51 I'm in the psplash recipe and sources. I just found out that we can specify the framebuffer device. May 12 21:05:00 "fbdev = getenv("FBDEV");" May 12 21:05:06 Extracted from the source code. May 12 21:05:19 The question is where should I set this FBDEV? May 12 21:05:31 I need to target /dev/fb1. May 12 21:05:40 * sri waves. May 12 21:06:00 that depends on the init you're using. i don't think sysvinit provdies a way to alter the environment of init scripts in a generic way, so the initscript would need altering to source /etc/default/psplash or so and you could put it in there May 12 21:06:08 systemd has its own configuration handling for services May 12 21:07:04 If I set FBDEV in a /etc/profile file it will be useless? May 12 21:07:45 I'm actually withsysvinit right now. May 12 21:07:52 But that can change. :P May 12 21:07:53 that won't do anything, as profile is sourced for user shells, not startup scripts May 12 21:07:59 afaik, anyway May 12 21:08:06 not only user shells, but user login shells May 12 21:08:16 s/only//; s/but/but only/ May 12 21:08:36 k May 12 21:08:59 i'd suggest reading the psplash startup script to see if it sources anything from /etc/ to configure it May 12 21:09:15 it's a common pattern for the startup script to source a file in /etc/default/ so the admin can set variable sthere to alter behavior May 12 21:09:37 more often variables used by the startup script to pass commandline arguments, but i don't see why you ccouldn't use it for this sort of thing too if you export it there May 12 21:09:56 Nope it doesn't. May 12 21:10:38 It only checks cmdline for psplash=false. May 12 21:10:51 !!!! May 12 21:10:55 You're right!!! May 12 21:11:05 /usr/bin/psplash --angle $rotation & May 12 21:11:23 I just need to get FBDEV as a parameter instead of an environment variable. May 12 21:11:57 /etc/fbdev :) May 12 21:12:37 Otherwise systemd would make it also? May 12 21:14:15 Let's just say that the command line parameter will be more important than the environment variable. May 12 21:15:56 Thanks kergoth! May 12 23:02:07 Hmm, wonder why wic bmap wasn't merged yet. would think everything could be merged but the bmaptool in DEPLOY_DIR_IMAGE thing, since that piece is more controversial May 12 23:03:55 shouldn't that just be a native? May 12 23:06:07 Crofton|work: it's a generated wrapper script for easier use. a native will be in the sysroot, which we encourage people not to run directly May 12 23:15:43 hmm May 12 23:16:17 So it turns out, if you don't have runtime package management listed in IMAGE_FEATURES, rootfs.py automatically and unconditionally removes a handful of packages, such as "shadow", on the grounds that they are "unneeded". May 12 23:16:39 This is biting me in this particular case because of a smallish filesystem on which we want to run the ptest tests for acl, which require gpasswd. May 12 23:16:59 But more generally, is it really a safe assumption that no one would ever need to modify users or groups unless they were going to be doing package management? May 12 23:17:39 sounds liek some comments on a systemd issue May 12 23:18:05 but I think they were related to ahving package management on May 12 23:18:36 My problem is that if I create a package and put RDEPENDS_${PN} += "shadow", I don't really *expect* to have the shadow package forcibly removed thereafter. May 12 23:18:38 I mean, it seems surprising. May 12 23:18:48 seebs: i've seen complaints about that before before. but yes, that does seem a bit strange to me as well May 12 23:18:49 * kergoth shrugs May 12 23:21:05 I'm not even convinced that the "remove_unneeded" thing is a good choice anyway. Or at least. I'd probably make it a configurable thing, rather than a hard-coded dependency on an only marginally-related feature. **** ENDING LOGGING AT Fri May 13 02:59:58 2016