**** BEGIN LOGGING AT Tue Sep 10 03:00:00 2013 Sep 10 08:03:07 hi, is there any reason why the oe core contains procps, but not procps-ng? Sep 10 08:04:40 query lpapp Sep 10 08:25:41 hi all ! trying to build a poky-pure-wayland image, I'm trying to build the snd-hda-intel module in my kernel, does anyone know how to do this ? thank you Sep 10 08:28:55 pirut: replied. :-P Sep 10 08:39:16 hi everybody Sep 10 08:39:42 I've a problem with the inherit update-rc.d when I update a package using dpkg. Sep 10 08:40:10 I get a update-rc.d: /etc/init.d/samba exists during rc.d purge (use -f to force) Sep 10 08:40:23 The complete error is listed here: http://pastebin.com/Lch21KYx Sep 10 08:40:55 Can anybody give me some help on this problem please? Sep 10 09:19:50 hi. If I'm working on a new board, what is my KMACHINE? Sep 10 09:43:15 rburton: hi ! how are you ? I might need your help if you have the time for, still trying to have my poky-pure-wayland image, I'm trying to build the snd-hda-intel module in my kernel (because there is no sound for the moment when I launch a wave file after starting pulseaudio) Sep 10 09:45:08 what machine? Sep 10 09:46:38 rburton: you mean the target ? lenovo x230 thinkpad Sep 10 09:47:24 i mean what MACHINE did you set in your yocto build Sep 10 09:48:19 chiefriver Sep 10 09:49:00 clearly our hw testing needs to be improved Sep 10 09:52:17 * rburton looks Sep 10 09:54:08 elbc_: chiefriver has sound support compiled in Sep 10 09:54:26 CONFIG_SND_HDA_INTEL=y Sep 10 09:54:43 did you check that the output isn't muted? :) Sep 10 09:54:55 pulse will moan if it can't find any sound hardware Sep 10 09:56:17 rburton: ok thanks strange, output is no muted Sep 10 09:57:45 have you checked what alsamixer says? Sep 10 10:03:58 when i try to launch alsamixer, it sayx mixer: no such file or directory Sep 10 10:04:16 and there is no sound directory in /lib/modules directory Sep 10 10:05:01 the kernel configuration i'm got in front of me says its built directly into the kernel Sep 10 10:05:23 is there anything in /proc/asound? Sep 10 10:05:31 i look Sep 10 10:08:08 morning all Sep 10 10:08:49 does yocto follow oe bug-reports and patches? Sep 10 10:09:41 eglibc 2.18 is missing a patch that breaks binaries on non-NEON capable chipsets. Sep 10 10:13:26 rburton: you are right, it seems that it is built into the kernel, and it it is appearing in /proc/asound, but no sound ... Sep 10 10:14:23 maybe the kernel is still missing some config options Sep 10 10:14:27 * rburton is no kernel driver expert, sadly Sep 10 10:14:40 ok thanks anyway Sep 10 10:14:56 file a bug, definitely Sep 10 10:16:03 I have a fairly simple question, I think.. How do I enable the v4l2src plugin in gst-plugins-good package? I see there is "PACKAGECONFIG[v4l] ..." line in the recipe file but I can't seem to figure out where should I put the v4l flag to make it build the proper plugins into the image Sep 10 10:17:25 I suppose I should add something into my local.conf or make a recipe of my own in order to enable those. But I don't know what I should put there to alter the package configuration flags Sep 10 10:18:01 Any hints? Sep 10 10:21:49 Kaapeli: you can bbappend the gst-plugins-good recipe in your own layer and set PACKAGECONFIG += "v4l" there, or alternatively do PACKAGECONFIG_append_pn-gst-plugins-good = " v4l" (note leading space in value) in your local.conf or distro config Sep 10 10:26:22 bluelightning, Thanks, it is now fetching dependencies to build it with v4l support Sep 10 10:26:30 Kaapeli: np Sep 10 10:26:44 bluelightning, Where does the "pn" come to the variable name? Sep 10 10:27:54 I think I already tried the above line but without the "_pn_" part in the middle, so it didn't change the package flags at all Sep 10 10:28:32 Kaapeli: it needs to be _pn-recipename not _pn_recipename Sep 10 10:29:02 Oh yeah, indeed Sep 10 10:29:10 pn-recipename is an override that is defined within the context of each recipe Sep 10 10:33:59 hmm.. I still get error from gst-launch that "no element v4l2src" Sep 10 10:35:32 Kaapeli: you probably didn't install the package with the element in Sep 10 10:40:38 Yeah, I just run "bitbake core-image-minimal" and it seemed to do a lot of things related to gst-plugins-good Sep 10 10:55:58 Well, I found where all the rpm packages are. Now I need to get rpm on the target so I can install them. Looking good :) Sep 10 10:56:44 Kaapeli: or, just add the package name to IMAGE_INSTALL for your image Sep 10 10:57:02 Kaapeli: http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#usingpoky-extend-customimage Sep 10 10:57:53 hmm.. I have gst-plugins-good there but is that not enough to pull in all the dependencies? Sep 10 11:08:14 Kaapeli: I think the meta-package for all gst-plugins-* is gst-plugins-*-meta so you'd need to add gst-plugins-good-meta to install all built plugins Sep 10 11:10:21 or, of course, you can install just the ones you want individually Sep 10 11:13:35 Ah, that is it of course.. Sep 10 11:33:45 Yay, a lot of gsreamer plugins on the device now, cool Sep 10 11:34:04 Thanks for the help guys Sep 10 11:56:55 is there something like linux-yocto-tiny but for arm? Sep 10 12:44:06 n01: why wouldn't it work on arm as is? Sep 10 12:50:21 RP: I was scared about that COMPATIBLE_MACHINE = "(qemux86)" Sep 10 12:53:29 rburton: I solverd my problem, my user created by using /usr/sbin/useradd was not added in the audio group Sep 10 12:53:52 elbc_: aah Sep 10 12:54:23 rburton: so my question would be : how can I add a user at the build and this with the good groups, instead of each time recreating a user (the only one user by default is root) Sep 10 12:55:32 (and if you know how to have the french keyboard it would be useless too :) ) Sep 10 12:56:04 elbc_: if you'll always have the same user, you can create a recipe that uses useradd.bbclass to make a user Sep 10 12:56:57 ok I'll take a look to this clue Sep 10 12:58:02 n01: I guess it was only enabled/tested for that however you should be able to make it work for qemuarm Sep 10 12:59:37 RP: actually my idea was to use it for a new board I'm working with just 10MB RAM Sep 10 13:00:02 n01: that would seem reasonable too Sep 10 13:03:19 what I'm not understanding is why this kernel is particularly suited for tiny systems. Is it patched or does the saving come from the configuration stored in meta? Sep 10 13:07:48 n01: it's mostly about the config, just about everything that can be cut out has been cut out Sep 10 13:08:23 n01: this may be of interest - http://www.youtube.com/watch?v=WHLvI8j31vk Sep 10 13:09:28 (slides at http://elinux.org/images/2/2b/Elce11_hart.pdf ) Sep 10 13:10:18 Hi Sep 10 13:10:32 bluelightning: thank you, I was using that slides Sep 10 13:10:37 I have this warning. can anybody suggest a fix Sep 10 13:10:40 Use of ALTERNATIVE_NAME is deprecated, see update-alternatives.bbclass for more info. Sep 10 13:10:58 I am not using ALTERNATIVE... Sep 10 13:11:41 Guest42914: a recipe you are building must be though Sep 10 13:12:41 Guest42914, "grep -ri 'ALTERNATIVE_NAME'" if you wanna see where it is. Sep 10 13:12:44 ok the .inc file has lines like update-alternatives --remove Sep 10 13:14:15 thats in the inc file. Sep 10 13:16:26 so should I just inherit that update-alternatives.bbclass Sep 10 13:17:17 ok it is doing that Sep 10 13:17:18 ... Sep 10 13:17:29 Hi Sep 10 13:17:36 I'm having troubles with qmake Sep 10 13:17:48 I generated it using meta-toolchain-qt Sep 10 13:18:22 but for some reason qmake segfaults, so I want to recompile the native sdk qmake, so what package do I have to "bitbake -c clean <>"? Sep 10 13:25:09 fenrig: bitbake -c clean qt4-nativesdk I think Sep 10 13:25:21 fenrig: I don't think recompiling it alone is going to solve a segfault though Sep 10 13:25:46 bluelightning: I'm at least gonna try that first, I'm actually using angstrom v2013.06 Sep 10 13:25:52 and they use linaro Sep 10 13:26:04 and sometimes I (not much) I got a internal compiler error :o Sep 10 13:29:44 fenrig, can you pastebin the actual qmake message Sep 10 13:30:03 crofton|work: I'll do that Sep 10 13:30:28 crofton|work; this is the strace https://gist.github.com/fenrig/6504329 Sep 10 13:30:38 I need the actual messages Sep 10 13:30:42 I am looking for a pattern Sep 10 13:31:31 Crofton|work: well the only message I get is "segfault - made a memory dump" Sep 10 13:31:42 hmm Sep 10 13:31:43 (it's in dutch :/) Sep 10 13:31:44 weird Sep 10 13:31:56 on ubuntu I ahve heard of a dynamis linker issue Sep 10 13:32:17 are you running a 64 bit toolchain on a 32 bit machine or vice versa? Sep 10 13:33:39 Crofton|work: i'm running a 32 bit toolchain on a 64 bit machine :o Sep 10 13:34:00 Crofton|work: Strace clearly states it's using 32 bit compatibility mode Sep 10 13:34:11 does the toolchain work on the machine you built it onm? Sep 10 13:34:16 I saw Sep 10 13:34:43 Crofton|work: No it doesn't, actually I'm just using yocto at home :p so I only use this pc to experiment on Sep 10 13:35:09 ERROR: Nothing PROVIDES 'qt4-nativesdk' Sep 10 13:36:03 maybe I need more 32 bit libraries Sep 10 13:36:36 but I thought the sdk actually covers most of the dependencies Sep 10 13:38:33 fenrig: sorry, I got name wrong... it's nativesdk-qt4-tools Sep 10 13:40:02 bluelightning: It cleaned something, let's hope it cleaned that qmake :p Sep 10 13:48:18 bluelightning: It did not recompile qmake, it's still the qmake from last friday :/ Sep 10 13:53:28 fenrig: what did you change in the mean time? Sep 10 13:53:45 bluelightning: how do u mean? Sep 10 13:53:57 bluelightning: I've added networking capabilities in my image Sep 10 13:54:12 bluelightning: using connman and related tools, also I included linux-firmware Sep 10 13:54:43 well, you said that it's qmake from last friday, that implies that there's a change you expected in between then and now affecting qmake - what was that change? Sep 10 13:55:16 bluelightning: nothing, I just want to recompile qmake so I can test if it's a compiler thing or if it's something with the sources :) Sep 10 13:55:42 bluelightning: I only got to test qmake yesterday because I was not happy with my image until then :) Sep 10 13:56:51 fenrig: ok, well if you do use "-c cleansstate" rather than "-c clean" it will also remove the sstate packages; normally that is not necessary though Sep 10 13:59:17 bluelightning: now it's doing a lot more actually :p Sep 10 13:59:47 bluelightning: it's compiling nativesdk-qt4-tools, how do U actually know that that package contains qmake ? Sep 10 14:00:46 fenrig: well I just happen to know that :) but after it is built you can look in the workdir for the recipe to see what has been produced Sep 10 14:01:21 (I'm the maintainer of the Qt 4 recipes in OE-Core, FWIW) Sep 10 14:09:48 rburton: I put this in my core-image-weston.bb file : inherit useradd but at build I got the following error : inherits useradd but doesn't set USERADD_PACKAGES ... what I'm missing ? Sep 10 14:10:21 elbc_: see the class for the variables you need to set Sep 10 14:10:39 ok thks Sep 10 14:14:24 anyone here get qt5 working? Sep 10 14:15:01 reallife: JaMa is the maintainer of the meta-qt5 layer, what problem are you experiencing? Sep 10 14:16:37 whenever i try to run a qt example i get an error message regarding not being able to find the xcb platform. it insists i try the linuxfb and other buffers. i am using the core-image-x11 recipe on a wandboard Sep 10 14:16:42 bluelightning Sep 10 14:18:06 JaMa Sep 10 14:43:48 fenrig, are you trying to create a toolchain to use Qt or Qt-Embedded? If your trying to build for embedded you neem meta-toolchain-qte. Sep 10 14:48:34 cfo215: qt not qt-embedded Sep 10 14:49:12 fenrig, recently made that mistake ;) Sep 10 14:49:13 bluelightning: ah yes, I wish there was more transparancy in the build system :p Sep 10 14:56:06 YTPM: The Yocto Project Tech Meeting Con-Call will be starting shortly Sep 10 14:56:07 Dial-in number: 1.972.995.7777 / Participant passcode: 42001078 Sep 10 14:56:07 This call is open to all and the channel remains open to discuss any topic. Sep 10 14:58:32 okay the segmentation fault stays :/ Sep 10 14:58:37 even after recompilation -.- Sep 10 14:59:24 YPTM: Paul Eggleton joined Sep 10 14:59:30 YPTM: Denys is online Sep 10 14:59:37 YPTM: belen joined Sep 10 14:59:50 YPTM: Saul is here Sep 10 15:00:18 YPTM: MatthewW has joined Sep 10 15:00:26 YPTM: Beth Flanagan joined Sep 10 15:01:12 YPTM: Björn is on the call Sep 10 15:01:15 YPTM: LaurentiuP joined Sep 10 15:01:15 YPTM: Welcome to the technical team meeting. Please let me know who's on the bridge. Thanks! Sep 10 15:01:24 YPTM: Richard is on the call Sep 10 15:02:16 YPTM: nitin is on the call Sep 10 15:02:33 YPTM: Michael here. Sep 10 15:02:54 YPTM: Tom Z here Sep 10 15:03:08 YPTM: jzhang on the call Sep 10 15:03:20 YPTM: ross here Sep 10 15:03:32 YPTM: Scott Rifenbark is on the call Sep 10 15:04:14 YPTM: Any opens? Sep 10 15:04:29 YPTM: Opens: branching for dora and 1.5_M4 Sep 10 15:04:42 YPTM: open: mesa-gl. Sep 10 15:07:32 morning walter Sep 10 15:07:35 s Sep 10 15:07:54 hi rburton Sep 10 15:09:50 YPTM: polk joined Sep 10 15:14:05 * scottrif is away from phone for 5 minutes Sep 10 15:18:14 * scottrif is back Sep 10 15:28:39 pidge: your current state: amused by bug report. state in an hour: "how did this even work?!". Sep 10 15:29:38 rburton: amused that my code is doing "funny stuff" Jinja2 is weird. Sep 10 15:35:52 YPTM: thank you all for joining the call. Have a nice day/evening! Sep 10 16:31:24 davest appears to be drinking Sep 10 16:31:49 good strong start to the day Sep 10 16:32:21 as he's my manager i'll follow his example and bring that bottle of Laphroaig to my desk tomorrow Sep 10 16:32:45 he is making wild claims about new intel stuff on twitter Sep 10 16:33:07 heh Sep 10 16:33:27 so he is! Sep 10 16:34:17 happy hour Sep 10 16:40:36 has anyone had luck configuring GPIO for the congatec qmx6? i am unable to figure out which kernel GPIO pins are associated with the GPIO0-7 in their docs. Sep 10 16:40:58 no help from the support folks. Sep 10 16:48:16 Crofton|work: oh, *that*. that could be interesting. Sep 10 17:12:01 I have built my First yocto build and have run the image on qemu simulator, where/how can i find/build the SDK image, Sep 10 17:14:12 bitbake -c populate_sdk Sep 10 17:27:07 what is the relevance of the "image_name", it doesnt seem to accept just about any string... is there document that explains the same? Sep 10 17:27:21 I am trying to build an Image that has GUI Sep 10 17:27:39 you made an image to run on qemu? Sep 10 17:30:00 yes, that was the default one, i followed the quick start guide. Now i want to build one with a gui Sep 10 17:33:30 populate_sdk builds an sdk with contents corresponding to an image. so if you want an sdk that matche sup with console-image, tehn run it against console-image. Sep 10 17:35:15 what should i do... if i dont wan an SDK ,but a GUI only Sep 10 17:39:42 depending on platform try something like bitbake fsl-image-gui Sep 10 17:39:52 it will output to tmp/deploy/images Sep 10 17:40:38 if you dont want an sdk, why did you ask about an sdk? Sep 10 17:40:40 * kergoth rolls eyes Sep 10 17:45:11 hahaha... the QS guide said that booting the SDK image gives a GUI Sep 10 17:47:59 that's why... I am new to embedded Linux DEvelopment and automated build tools, have been more of a Device Drivers girl Sep 10 17:49:45 "core-image-minimal" "core-image-sato", what is the relevance of these names , w.r.t to the bitbake utility? Sep 10 17:55:50 @mzinri_ , i will rn my image on qemu simulator... would this work there too? Sep 10 17:56:13 GeekGirl: generally, those are build targets; less generally those are image recipes Sep 10 17:56:32 GeekGirl: there's a reference for the image recipes here FWIW : http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#ref-images Sep 10 18:11:11 bluelightning, thanks for that documentaion, ooks like lots of reading is underway Sep 10 18:18:46 morn' challinan Sep 10 18:32:20 in trying to build meta-toolchain-qt, i get a failure in nativesdk-dbus running do_package_qa. i see mention of a patch here but can't find the patch. http://lists.openembedded.org/pipermail/openembedded-core/2013-July/080796.html Sep 10 19:19:38 RP: fyi, layer-management in meta-ivi builds fine with cmake using SDK_OS, but fails when using TARGET_OS. Presumbly something different between 'linux' and 'linux-gnueabi', since the latter includes the abi/libc suffixes Sep 10 19:19:57 any cmake experts around? Sep 10 19:42:11 Bitbake can't find my .bbclass file. Sep 10 19:42:58 Hi all, in local.conf I have the line INHERIT += "check_security". Sep 10 19:43:25 In meta-security/classes I have a new file, hopefully syntactically correct, called check_security.bbclass. Sep 10 19:44:22 make sure the layer is in bblayers.conf, check layer.conf in the layer to makle sure it adds itself to BBPATH Sep 10 19:44:31 bitbake finds classes and .conf fiels via BBPATH Sep 10 19:44:34 My error is an ExpansionError "Failure expanding variable check_security" Sep 10 19:44:39 OK. Sep 10 19:44:51 So, I looked at my BBPATH. Sep 10 19:45:06 that error doesn't mean it didn't parse the recipe. Sep 10 19:45:21 it means you had a check_security variable that failed to expand Sep 10 19:46:08 for anyone interested, i worked around my failure in nativesdk-dbus. i added a blank do_package_qa in the dbus.inc file to get around it for now. Sep 10 19:46:11 But my BBPATH includes /home/mulhern/meta-security. Sep 10 19:48:47 OK! I think it means that it tried to parse the check_security.bbclass file and found a syntactically poor definition of a function, also called check_security. Sep 10 19:52:28 that makes sense. bitbake has to parse the functions, both shell an dpython, in order to track what variables they're using, which is necessary both to know what to emit at build time, and for calculating metadata checksums Sep 10 19:54:38 I'm banging my head against a wall trying to use config fragments from a .bbappend in a new layer I've added to my build - AFAICT it seems correct but I never see its changes getting pulled in?! Is there a simple way to trace how it gets pulled in and used / not used? Sep 10 19:55:45 I've bitbake -e' Sep 10 19:55:59 I've bitbake -e'd and I can see an append line apparently parsing it correctly... Sep 10 20:09:54 Ah great. It seems that to use fragments you need to be using the linux-yocto bbclass - but I'm pretty sure Freescales linux-boundary doesn't... Sep 10 20:09:57 Meh Sep 10 20:10:50 So I've got Freescales linux-boundary_3.0.35.bb defining the kernel build then the upstream BSP authors linux-boundary_3.0.35.bbappend appending to that - how can I now get my project specific layer to change defconfig? Sep 10 20:52:25 So, do you ever call bitbake in a function in a class? Sep 10 20:53:00 Perhaps I want to run an app and I want to be sure that the appropriate recipe has been baked first. Sep 10 20:53:13 Can I do that in the function? Sep 10 20:53:16 bitbake has plenty of ways to mark dependencies Sep 10 20:53:16 no Sep 10 20:53:28 only one bitbake instance can be run against a given build dir at one time Sep 10 20:54:48 What way would you suggest in this case? Sep 10 20:55:20 insufficient information Sep 10 20:57:09 function is called check_security. It invokes native version of buck-security. check_security is appended to ROOTFS_POSTPROCESS_COMMAND. All this in check_security.bbclass. Sep 10 20:57:49 How do I, automatically, make sure that buck-security-native has already been baked? Sep 10 20:58:02 EXTRA_IMAGEDEPENDS += "buck-security-native" Sep 10 20:58:39 Aha! Sep 10 21:14:41 What part of bitbake manages the soft links to log files in the ${WORKDIR}/temp directory? Sep 10 21:17:08 does bitbake -f also invalidate anything that could be pulled from sstate-cache? Sep 10 21:17:34 or is there another way to forcibly update (or ignore) sstate-cache if it's got things wrong? Sep 10 21:18:36 -f forces it to re-run that task, not pull its bits from sstate, and any task that depends on it will get rerun as well Sep 10 21:18:54 but you can always use -c cleansstate to wipe sstates on a per-recipe basis Sep 10 21:19:17 aha! That's what I was wondering. Sep 10 21:19:24 see also -C, which says 'force this task to be re-run, then run the default task, so all the subsequent tasks for that recipe get rerun as well' Sep 10 21:19:28 Drat, that'd have saved me a couple of hours last night :-D Sep 10 21:19:53 that is, bitbake -C configure foo == { bitbake -f -c configure foo; bitbake foo } Sep 10 21:20:12 Will -c cleanstate also update the sstate-cache contents to fix things as it were or do I need to scratch the whole cache contents? Sep 10 21:20:21 i don't understand the question Sep 10 21:20:30 it cleans a recipe and removes its sstates, thats all Sep 10 21:20:49 if the recipe is then rebuilt, it will hav eto build it from scratch, since the sstates wont exist Sep 10 21:22:12 OK - that's what I mean - i.e. it removes the sstate-cache entry as opposed to ignoring it and forcing a re-build leaving the old entry in the cache Sep 10 21:22:30 kergoth: Nice one, thanks Sep 10 21:26:31 Another probably easy question - I'm rebuilding a kernel I've done some local test changes on via "bitbake linux-boundary -c compile -f -v" which rebuilds my uImage and then "-c deploy" which I understand copies the kernel to the images directory? Sep 10 21:26:55 However, there's also the -c compile_kernelmodules step which I'm doing but I can't see an analagous deploy for the modules? Sep 10 21:30:50 kergoth: I think sgw_ is looking at the cmake issue Sep 10 21:31:08 kergoth: it needs to return "Linux" regardless of what suffixes are in use Sep 10 21:31:13 RP, kergoth patch is on the mail list Sep 10 21:31:23 sgw_: not got that far yet., thanks Sep 10 21:31:29 I put out 3 patches today. Sep 10 21:32:29 RP: I built with hans's libtool patch, that one concerns me a little, while it might be correct, it might cause some sstate related issues, sending email Sep 10 21:50:03 sgw_: what kind of issues though? Sep 10 21:55:36 RP: I am still digging Sep 10 22:04:06 RP: it seems that file is not finding libmagic correctly during the compile it tries, it's pointing to /usr/lib now not our sysroot! Sep 10 22:04:10 libmagic.so.1 => /usr/lib/x86_64-linux-gnu/libmagic.so.1 (0x00007f82918e9000) Sep 10 22:04:32 RP: I am digging more. Sep 10 22:05:05 sgw_: this is file-native? Sep 10 22:05:23 sgw_: its sounding a lot like something broke the sstate dependencies :/ Sep 10 22:06:44 RP: yes, that's my concern also, I am trying a clean build while looking at the issue. Sep 10 22:06:58 clean meaning no-sstate Sep 10 22:08:49 sgw_: I suspect the recent useradd change Sep 10 22:09:39 sgw_: a build from the sstate with that change reverted would be interesting Sep 10 22:10:00 RP: huh? When I bisect'ed MUT it landed in the libtool change. Sep 10 22:10:15 sgw_: just luck Sep 10 22:10:23 (bad luck) Sep 10 22:10:35 this change won't bisect easily Sep 10 22:10:38 right, let me try that since I have the setup ready to go. Sep 10 22:17:35 RP: Nope, failure still happens with the useradd change reverted, this really seems like a libtool issue Sep 10 22:23:53 sgw_: I guess there could be a problem with -natives and this change. We might need to stop passing the libtool-sysroot option for natives Sep 10 22:40:28 So, I am wrestling with a PREFERRED_PROVIDERS thing, and it occurs to me that perhaps I am not the only person who runs into this. Sep 10 22:41:10 The underlying problem appears to be that some recipe I definitely don't want is being picked up because it PROVIDES something, and the recipe I do want doesn't. Sep 10 22:41:15 So I get a lot of warnings about two recipes with a given package in PROVIDES, and a suggestion that the reason is that one of them provides something the other doesn't, and should. Sep 10 22:41:45 Would it be useful to anyone but me if that message also indicated, for each of those two packages, what things it PROVIDES that the other does not? Sep 10 22:42:03 Because that would almost always tell me why I am seeing it. Sep 10 22:42:18 ya.. I've had the same desire, and couldn't figure out how to implement it.. Sep 10 22:43:23 I am gonna see whether blacklisting one helps, because that usually gives the requirements-chain. Sep 10 22:43:39 But I may end up implementing this just because I am otherwise-stuck. :) Sep 10 22:44:27 btw fray stop using bandwidth and CPU time my shell is being all slow and it is probably your fault for existing. Sep 10 22:52:03 Hmm. Did bitbake's handling of tar files in SRC_URI change recently? Sep 10 22:52:39 I ask because I'm working with a (moderately heavily modified) meta-sourcery, and it wants a tarball in SRC_URI, then is has an unpack_libc which tries to tar jxf the tarball. But do_fetch is *untarring* the tarball, not copying it. Sep 10 22:59:12 seebs: do_unpack has always untared tarballs Sep 10 22:59:18 Huh. Sep 10 22:59:31 seebs: you can tell do_unpack not to touch it with some parameter iirc Sep 10 22:59:35 I am clearly missing something about how meta-sourcery works, then. Sep 10 22:59:55 Which probably doesn't matter since I almost certainly want to change this anyway. Hmm. Sep 11 00:11:37 okay so :o Sep 11 00:11:42 Does anyone have a pointer to any quick crib-sheet on how to use guilt to create patches when working with yocto? It looks like it's the defacto tool but I can't see anything... Sep 11 00:12:00 pev: quilt? Sep 11 00:12:07 No, guilt Sep 11 00:12:11 I just generated "meta-toolchain-qt" for qemuarm and I have the exact same segfault on qmake :o Sep 11 00:13:15 tlwoerner: Well, if quilt's supported that'd be just fine too but from what I read it seems to have changed to guilt and also looking at the tree there are all sorts of hooks and guilt gets built locally within yocto as a devtool to use Sep 11 00:13:36 tlwoerner: so I assume that's what I should be using! Sep 11 00:15:02 pev: i've used quilt with OE/Yocto to help create patchsets and it works for me :-) Sep 11 00:15:47 Do yo have a quick cut/paste crib of the command sequence you use? I'd tried using quilt previously and had it complain biterly! Sep 11 00:19:28 pev: does this help? https://collaborate.linaro.org/pages/viewpage.action?pageId=6717644#OE/YoctoSkillsWorkshop-fixingoraddingviaquilt Sep 11 00:19:39 I just generated "meta-toolchain-qt" for qemuarm and I have the exact same segfault on qmake :o Sep 11 00:20:47 tlwoerner: Interesting, I'm pretty sure thats what I was doing before (had lots of experience with quilt working on OpenWRT)... Will give it another go! Sep 11 00:23:42 pev: both "guilt" and "quilt" are available in meta-poky and openembedded-core meta/recipes-devtools Sep 11 00:24:01 tlwoerner: Ah, ok that's what it was...! So if I do something like : 1) bitbake -c clean 2) bitbake -c devshell 3) ls -l patches : it looks like the build system has already set up quilt. Sep 11 00:24:02 and both have "-native" recipes Sep 11 00:25:22 tlwoerner: In fact if you have something (like my kernel) that has several patches you can "quilt pop -a" and it looks like it's doing the right thing (hoorah) - however, a simple "quilt series" borks. Then if you do a "quilt push " it looks like it moves to the right patch and does the right thing, but then a "quilt files" seems to think theres nothing there again Sep 11 00:25:48 tlwoerner: So kind of working but kind of not... I think I was a bit concerned to use it in case there was some broken magic internally somewhere Sep 11 00:26:33 tlwoerner: doing a "which quilt" looks like it's found one built by Yocto so that's right... Sep 11 00:32:07 fenrig: cause qt stinks like a butt Sep 11 00:32:13 and qmake never friggin works Sep 11 00:32:45 reallife: ironically I had it working a month ago on the emtrion I.MX53 :p Sep 11 00:36:07 fenrig: man ive been trying for weeks to get qt5 working on my imx6 wandboard Sep 11 00:36:26 reallife: Oh yeah but qt5 is still in development in yocto ;) Sep 11 00:36:43 fenrig: tell that to my BOSSSSSSSSSSSSS Sep 11 00:36:55 reallife: oh too bad :/ Sep 11 00:37:34 reallife: do u need Qt5 for technical reasons? Sep 11 00:37:35 fenrig: honestly, i think ill have better luck using Arch and trying to load the imx6 gpu blobs on there Sep 11 00:38:15 reallife: :o yeah yocto isn't perfect Sep 11 00:38:23 fenrig: yeah, the software devs have already built an application using its features. i think downgrading is possible but i have yet to figure out how to convince them Sep 11 00:39:16 reallife: hhhmmm much luck with that :/ Sep 11 00:39:25 fenrig: thank you Sep 11 00:39:36 reallife: but I spoke to the maintainer of meta-qt5 a month ago Sep 11 00:40:02 reallife: you can find this person in this channel under the name "bluelightning" Sep 11 00:40:41 reallife: he told me that sdk generation is not supported at this moment Sep 11 00:40:53 reallife: and it doesn't have priority :/ Sep 11 00:41:10 tlwoerner: This looks like a bug... http://www.yoctoproject.org/docs/1.0/poky-ref-manual/poky-ref-manual.html#structure-build-tmp-work "Within this directory, the source is unpacked to linux-2.6.20 and then patched by quilt (see Section 3.5.1). Within the linux-2.6.20 directory, standard quilt directories linux-2.6.20/patches and linux-2.6.20/.pc are created, and standard quilt commands can be used." Sep 11 00:41:31 fenrig: yeah i tried messaging blue but no luck. i'm pretty noobie so i dont really know how to go about using the meta-qt5 patches and such Sep 11 00:41:46 which is a shame as that would be damn handy... (well, unless it's something broken with the distro I'm using which isn't beyond the realms of possibility) Sep 11 00:57:16 pev: i assume you're doing something like "bitbake -c devshell linux-yocto"? Sep 11 00:59:00 tlwoerner: Yep Sep 11 00:59:53 tlwoerner: Just checking quilt related files to see if the distro I've got has deviated significantly from 10.4 as I haven't got a 'clean' 10.4 built I can check against to see if it works (fails!) the same way... Sep 11 01:23:55 tlwoerner: Just had a quick test and it's broken the same way in 9.0.2... Sep 11 01:25:40 pev: so you're working on master, and when you do a "bitbake -c devshell linux-yocto" you end up with directories "patches" and ".pc" at the location in which the devshell dumps you? Sep 11 01:25:48 Yep Sep 11 01:26:08 tlwoerner: Which look correct but a simple "quilt series" or "quilt files" seems to not work Sep 11 01:26:46 Oh crap Sep 11 01:26:49 I just realised Sep 11 01:26:52 that's strange because when i perform that command i don't even get either of those directories i just see the linux sources Sep 11 01:27:00 kernel sources Sep 11 01:27:09 tlwoerner: Its the ****ing "PSEUDO_UNLOAD=1 bash" thing Sep 11 01:27:23 tlwoerner: Is that because your kernel doesn't have any patches applied? Sep 11 01:27:46 tlwoerner: What a ***** waste of an hour ..... Sep 11 01:28:00 i guess not... i haven't looked into it, just the basic qemu stuff Sep 11 01:28:03 works fine after setting that... Sep 11 01:28:39 yeah, i could replicate that with a stock 9.0.2 build by creating a couple of patches (with some content!) via quilt and then doing a quilt series to check Sep 11 01:29:32 looks like it should actually work really well - the files under "patches" are symlinks to the original location so going back with a quilt pop and an edit and a refesh should modify the original patch in its normal location... Sep 11 01:29:39 that'll save a lot of swearing... Sep 11 01:30:35 i'm not familiar with the pseudo unload thing. that's used when you don't want pseudo (aka fakeroot) in your devshell? Sep 11 01:30:58 yeah. I don't fully understand its usage but I know that's what's buggering up quilt... Sep 11 01:31:15 s Sep 11 01:39:43 tlwoerner: Ok - nearly great ; when you do a quilt refresh it overwrites the symlink with the new file but that's only really a single manual copy you can do at the end of editing so no big deal Sep 11 01:45:23 pev: so is the manual correct? or is there a bug that needs to be filed? Sep 11 01:46:46 The manual is correct to an extent - however if you're doing more operations than the add / refresh that it shows as examples I'd change it to note that you need to do the PSEUDO_UNLOAD in order to see the output of other quilt actions... That's crucial...! Sep 11 01:47:44 Although the manual I linked to is quite an old one no? I think the newer ones might not even have that much info in about quilt from memory Sep 11 01:48:23 which is a shame - i've just spotted a nice patch https://github.com/mrchapp/arago-oe-dev/blob/master/recipes/quilt/quilt-0.48/no_rm_symlinks.patch from earlier OE that fixes the symlink issue as well which I might port up Sep 11 01:52:31 yes, the documentation link you provided is ancient Sep 11 01:53:24 the latest would be: http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#using-a-quilt-workflow Sep 11 01:54:39 or http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#structure-build-tmp-work Sep 11 01:55:06 tlwoerner: Yeah, it would be great to update that to nod to full quilt docs but ESPECIALLY the ability to edit existing patches in place ; IMHO thats one of the most powerful things about it and mayboe one of the more frustrating things for other developers too...? Sep 11 01:57:35 pev: i agree. i would open a bug, and/or post a message to the yocto mailing list and perhaps CC the documentation maintainer Sep 11 02:08:35 tlwoerner: Ah, I hadn't noticed the xml for the docs were in the build - is there a friendly way to edit those? I'll see if I can update and submit a patch... **** ENDING LOGGING AT Wed Sep 11 02:59:59 2013