**** BEGIN LOGGING AT Thu Jul 31 03:00:01 2014 Jul 31 06:11:46 hi Jul 31 06:13:34 anyone having experience enabling flexcan on imx6sl (wandboard with EDM1-CF-IMX6S10-R512-NI4G-L-2C-TECH) Jul 31 06:16:19 i've added the flexcan driver via menuconfig, but the device is not showing up after booting the image. Jul 31 06:40:52 could it be related to device tree config? Jul 31 06:50:08 Hi everyone Jul 31 06:50:44 just a basic question... if i am modifying conf/local.conf should I use "+=" to append a variable? Jul 31 06:50:57 and a simple "=" to overwrite it? Jul 31 06:53:59 andyj25: yes Jul 31 06:55:00 thanks : ) Jul 31 06:55:09 hi andyj25,you can use VAR += "something" to append to a variable with space and VAR_append = "something" to append without space Jul 31 06:55:43 so when using += i need to use a leading space? Jul 31 06:55:50 and with VAR_append, I don't? Jul 31 06:56:00 http://elinux.org/Bitbake_Cheat_Sheet, scroll down to .bb file syntax Jul 31 06:56:23 or use this: http://elinux.org/Bitbake_Cheat_Sheet#.bb_file_syntax Jul 31 06:56:48 nice link.. thanks : ) Jul 31 06:57:14 np Jul 31 07:27:10 hi all Jul 31 07:29:52 I am wondering why "toaster" reports that a lot of packages which are part of my image depend on busybox. That information is displayed in the column "Dependencies". Does anyone know why? Thanks. Jul 31 08:47:07 morning all Jul 31 08:49:45 morning :) Jul 31 08:52:44 moin Jul 31 09:41:42 bluelightning: Hi! When I build xterm in x86_64 host it builds fine...but same recipe in i686 host gives a fetch error.No such file 'lib32-xterm-277.tgz'. Jul 31 09:42:16 Why does it append "lib32-" to the source tar name ? Jul 31 09:46:33 lsb_tester: that's because that recipe has not been written to build properly for multilib Jul 31 09:46:49 lsb_tester: in practical terms that means that in SRC_URI it should be using BPN and not PN Jul 31 09:47:53 the lib32- prefix is the multilib variant that is being built Jul 31 09:48:27 we namespace the multilibs in order to keep them separate (as we do with native, cross, nativesdk etc.) Jul 31 09:48:48 it seemsw fixed in _303 with commit 05de0ca Jul 31 09:52:12 Hi, I am trying to add a new task like this https://gist.github.com/wfailla/864a2253e188900ffb47 but it is not working what am I missing? Is there a order of those commands I got wrong? Jul 31 09:52:24 ant_work: indeed, some cleanup of BPN / PN for multilib has been done recently Jul 31 09:52:30 fix is probably worth backport to daisy Jul 31 09:53:37 ant_work: could you please share the link, where I can find this fix Jul 31 09:54:14 lsb_tester: http://cgit.openembedded.org/meta-openembedded/commit/?id=05de0ca43d325d32faa6067013cd7c596f2001bf Jul 31 09:55:04 wfailla: you don't need the python bit, just call your shell function do_ova-create Jul 31 09:55:17 bluelightning: ok Jul 31 09:55:41 lsb_tester: specifically http://tinyurl.com/ouzp6hz Jul 31 09:56:33 lsb_tester: you only hit one ;) Jul 31 09:57:46 bluelightning: Also I am trying to add a new FSTYPE, can you give me a reference to a documentation hoe to do this. I was looking at the sdcard_image-rpi.bbclass in meta-raspberrypi but I had no luck so far. Jul 31 09:58:55 yes...I have only one such fetch error. Seems like others are not in use for us. Thanks for the quick fix :) Jul 31 10:28:53 hi, how does Yocto know which tag to grab from the git repository provided in a bitbake recipe? I cannot find any reference to one, yet it gets the latest tag for some reason, not the latest commit. Jul 31 10:32:00 should it not clone the latest commit by default when this is specified? SRC_URI = " git://meh/foo.git;branch=master;protocol=ssh? Jul 31 10:34:36 ah, it is SRCREV which defines it, I think! Jul 31 11:04:38 I am wondering why "toaster" reports that a lot of packages which are part of my image depend on busybox. That information is displayed in the column "Dependencies". Does anyone know why? Thanks. Jul 31 11:05:36 what would the best practice to share some files between the latest release and the git version of a package? Jul 31 11:13:09 ivanstojanovic: it'll probably be because those packages contain shell scripts that depend upon /bin/sh Jul 31 11:37:55 lpapp: include a common .inc containing i.e. SRC_URI += Jul 31 11:38:35 ant_work: yeah, that has to be managed, too, but I was thinking more about the directory share. Jul 31 11:38:51 both can point to a single dir Jul 31 11:39:23 i.e. you can share kernel patches btw 2 kernel recipes Jul 31 11:39:42 easy way is to use /files Jul 31 11:48:03 BPN is better than "files" when sharing between 2 versions of the same recipe (a bit faster lookup) Jul 31 11:52:11 Hi, is it possible to change ones recipe so that instead of a git uri it would take the source code from a local directory (so I don't need to commit my small changes while developing) Jul 31 11:52:15 ? Jul 31 11:56:57 Jeena: see externalsrc.bbclass Jul 31 12:00:40 ah thanks, this looks interesting Jul 31 12:01:16 hello, I'm trying to compile a specific branch of linux kernel, which task did the branch git branch stuff in bitbake ? Jul 31 12:02:31 after I do bb unpack task the source of my kernel aren't set to the right branch is there an other way than SRCBRANCH to do that ? Jul 31 12:23:32 Is it possible to have a bitbake recipe that depends on two image recipes being built? Efforts so far have been unsuccessful. Jul 31 12:25:22 you need a third parent image depending on both Jul 31 12:25:30 ahh! Jul 31 12:25:32 or recipe or whatever Jul 31 12:25:56 Hrm, well, you mean just like DEPENDS = "my-image-one my-image-two" ? Jul 31 12:25:56 I hav eon eexample, one mom Jul 31 12:27:18 http://tinyurl.com/mrhwyw5 Jul 31 12:27:30 here is an initramfs image but you get the idea Jul 31 12:28:36 Awesome, thank you very much. This should get me what I need. Jul 31 12:30:28 yw Jul 31 13:04:37 sweet. I got it working.. The trick ended up being do_rootfs[depends] = "image-one:do_rootfs image-two:do_rootfs" Jul 31 13:14:16 Hi, have any of you had problems booting zImage on imx6 solo when using daisy branch of fsl-community-bsp? Jul 31 13:16:05 u-boot hanging on "Starting kernel..." Jul 31 13:16:44 i've tried dora instead and it boot without any problemsn Jul 31 13:17:10 anyone with the same experience? Jul 31 13:21:54 stiandre, wandboard solo ? Jul 31 13:39:00 kroon, oh yes! Jul 31 13:43:46 stiandre, hmm ok. I haven't had any problems booing a wandboard solo with daisy Jul 31 13:44:24 stiandre, you're using the linux-wandboard kernel ? Jul 31 13:45:55 stiandre, only issue I had was a couple of months ago, in which hdmi-probing would hang the kernel on boot, similar to the prints you got Jul 31 16:04:36 ok, this seems odd. We have "init-ifupdown" which only contains metadata, and no ifup or ifdown. Jul 31 16:04:50 apparently it assumes busybox will provide ifup/ifdown Jul 31 16:05:29 on debian, "ifupdown" package provides those binaries, but I can't find any non-busybox variants in yocto. Jul 31 16:30:19 * paulg assumes he's going to have to create an ifupdown package... Jul 31 16:34:21 probably.. :/ Jul 31 16:34:36 paulg: FWIW there is a recipe in OE-Classic that you could migrate: http://layers.openembedded.org/layerindex/oe-classic/recipes/?q=ifupdown Jul 31 16:34:49 nothing in meta-oe? Jul 31 16:34:55 nope Jul 31 16:35:10 I swear this was there in 0.9, must have been me remembering oe-classic instead Jul 31 16:35:14 not unless those utilities are wrapped up in something else Jul 31 16:36:33 maybe I was just thinking of init-ifupdown? Jul 31 16:37:37 fray_, yeah, one could argue that init-ifupdown should have them bundled within. Jul 31 16:37:59 probably should be seperate though.. Jul 31 16:38:32 ooh. fray_ thanks for the hint to look at ancient crap. Jul 31 16:39:17 recipes/ifupdown/ifupdown_0.6.10.bb exists in the original openembedded repo Jul 31 16:39:40 there is also recipes/ifupdown/ifupdown-ubuntu_0.6.8.bb Jul 31 16:39:43 ya, thats the one bluelightning was mentioning above.. probably the place to start Jul 31 16:40:12 ah, yeah I missed bluelightning 's comment while stumbling off to the fridge. Jul 31 16:40:54 guess i'll revive and uprev one of those two bitrot recipes. Jul 31 16:42:51 bluelightning, thanks. Jul 31 16:43:36 paulg: you're welcome, thanks for reviving an old recipe - would be awesome if you could send it back to meta-oe when you're done ;) Jul 31 16:44:31 yeah, will do. Having an implicit dependency on busybox always being present just sucks. Jul 31 16:44:33 I'd argue anything required to remvoe busybox has a place in oe-core at this point Jul 31 16:44:43 I've more then once wanted to build a system w/o BB Jul 31 16:45:05 fray_, question is, how does one make that thing we were working through last night into a more user-friendly switch? Jul 31 16:45:22 i.e all those _remove actions etc. etc. Jul 31 16:45:25 (My assumption is this is a small number of additional recipes that simply provide the discrete versions of the code BB is providing.. no 'additional' functionality.. Jul 31 16:45:45 I'm thinking an image or distro flag.. but I'm not sure which is 'correct' in the case.. Jul 31 16:46:18 distro flag perhaps "no base-busybox" which could change the defaults away from BB in all of the package groups... we still may want BB for the initrd and such though Jul 31 16:46:27 RP, btw, fray_ and I were making use of stuff like... Jul 31 16:46:29 RDEPENDS_packagegroup-core-boot_remove = "busybox-hwclock" Jul 31 16:46:41 'dirty' to say the least.. Jul 31 16:46:56 I'd love this to be a more standard use.. Jul 31 16:46:58 and we found it wouldn't regen the RPM with that change in a conf/local.conf of a layer ; we had to purge the RPM with a cleanall Jul 31 16:47:17 fray_ mentioned that I should mention it to you as a possible checksum oversight. Jul 31 16:47:53 paulg: I'm guessing that forcing a reparse (rm tmp/cache) would force a rebuild Jul 31 16:47:59 paulg: so cache bug, not checksum Jul 31 16:48:06 could very well be Jul 31 16:48:30 ok, fair enough ; I don't know enough of the underlying mechanism to get the terminology correct just yet. :) Jul 31 16:49:52 anyway, once I nuked packagegroup-core-boot, the do_rootfs stopped nagging me about hwclock, so I was happy. Jul 31 16:50:14 nuke via -f -c cleanall, that is. Jul 31 16:51:18 paulg, is an alternative hwclock provided by util-linux or coreutils? (I thinkit might be disabled there) which means a packageconfig setting is probably wanted Jul 31 16:53:02 fray_, yeah it is with util-linux. The util-linux is fragged into about 20pkgs. So I created a packagegroup-util-linux with all of 'em present. Jul 31 16:53:13 Figuring on sending that on to yocto / oe-core as well. Jul 31 16:53:31 no point in everyone who wants busybox gone reinventing the wheel on that. Jul 31 16:53:50 in the past we've done things like created a util-linux- that would bring in all of the sub packages in one big set Jul 31 16:54:21 ok, so apparently I've already reinvented the wheel. But I didn't find anything with a git-grep in currrent git repos. Jul 31 16:54:31 I didn't think to go looking in the abandonware. Jul 31 16:54:51 (as I didn't think to for ifupdown either.....) Jul 31 16:54:54 I'm saying for -other- recipes we've done an all type package.. Jul 31 16:55:00 AFAIK, not for util-linux Jul 31 16:55:06 oh, ok. Jul 31 16:55:20 but I'm completely blanking on the name we've used before.. Jul 31 16:56:20 http://pastebin.com/EEsb335r Jul 31 16:56:34 ^^^ <--- util-linux packagegroup Jul 31 16:57:05 once I've got ifup sorted and testing overall, I'll submit the above draft. Jul 31 18:53:17 * paulg wonders where the obtuse "/home/root" default that yocto uses came from.... I know it can be changed, but why is the default not simply "/root" ? Jul 31 18:59:20 paulg: Same reason anything happens - the person who did it had a preference and since they committed the code, they win until you provide a patch Jul 31 19:03:57 nar, guess I'll be sending a patch shortly then. :) Jul 31 19:09:36 On some of my more unix beardy days I'd even go as far as to say root doesn't even need a home directory Jul 31 19:09:51 Since it shouldn't be a login account Jul 31 19:19:49 hi. I tried building the raspberry pi image from master/yocto master and gcc crashed while compiling eglibc - is that known? Jul 31 19:22:24 rburton: o/ Jul 31 19:33:11 mh. cleansstate and rebuild fixed it Jul 31 19:33:12 odd Jul 31 20:00:01 kroon: okay.. so, do you think my issue could be related to the hdmi prøving, if som, what was your solution? Jul 31 20:01:36 kroon: sorry, i meant: hdmi probing Jul 31 20:02:43 stiandre, I had to revert one of the kernel commits. This problem only happened when I had the hdmi-cable plugged into the wandboard. If you have no cable connected, but it still wont boot, I think we could rule that one out. If possible, best thing would be to use a jtag-device for debugging, and see where the kernel gets stuck Jul 31 20:03:23 stiandre, you could try with appending "debug" on the kernel command line for a start, and see if that gets you any info Jul 31 20:03:35 stiandre, and "earlyprintk" Jul 31 20:03:47 nar, yeah I recall having HOME=/ a ways back. And the FHS says /root is optional. Anyways, patch sent. We'll see what the replies are... Jul 31 20:05:26 stiandre, youre using "MACHINE=wandboard-solo", right ? Jul 31 20:08:19 i think i am fundamentally confused about the process i should use to build an app and ultimately a linux image with yocto Jul 31 20:08:29 i need to build an app with qt, libusb and avahi Jul 31 20:09:02 so i would define an image recipe that contains all of those, build, then build a toolchain Jul 31 20:09:12 extract that toolchain and its sysroot and develop using that? Jul 31 20:09:33 in order for example to have the qt headers + libs and avahi headers + libs Jul 31 20:10:40 imagine at some point i decide i need another library, would i have to go the toolchain extraction process again? or is the preferred way to do it to point to my tmp/ sysroot ? Jul 31 20:10:58 (i’m reading http://www.yoctoproject.org/docs/1.6.1/adt-manual/adt-manual.html#extracting-the-root-filesystem) **** ENDING LOGGING AT Fri Aug 01 03:00:00 2014