**** BEGIN LOGGING AT Wed Jan 23 02:59:59 2013 Jan 23 03:04:03 khem: you can push it to the branch, I did just start a MUT build with the 2 patches that where in the eglibc branch Jan 23 03:09:02 khem: i did but im having kernel panic right after mounting the nfs and kernel supose to start init Jan 23 03:09:22 do I need a TAP network bridge? Jan 23 03:11:12 hi, I use "Yocto 1.3 + jasperforest BSP", have successfully build and boot "core-image-minimal" and "core-image-basic" images, but after I build and boot "core-image-lsb-sdk" Jan 23 03:11:25 SYSLINUX 4.03 2010-10-22 CHS Copyright (C) 1994-2010 H. Peter Anvin et al Jan 23 03:11:26 No DEFAULT or UI configuration directive found! Jan 23 03:11:26 boot: Jan 23 03:12:12 kergoth: i had this Kernel panic - not syncing: Attempted to kill init! Jan 23 03:12:20 have tried to use different size USB flash disk (4GB, 8GB, 16GB), all same Jan 23 03:12:42 i think its a problem with the nfs Jan 23 03:12:56 i think this ip address is wrong.. can I manually set them? Jan 23 03:16:02 ftonello: randomly addressing people in the channel isn't proper irc etiquette Jan 23 03:24:43 kergoth: it was a mistaque.. i thought it was completed to your nick Jan 23 03:47:31 hi, one curious question: why there's core-image-lsb-sdk and core-image-sato-sdk bitbake image option but no "core-image-minimal-sdk" or "core-image-basic-sdk"? Jan 23 03:57:49 sgw1: ok refreshed Jan 23 04:00:04 EddyLaiTW: well, SDK targets should disappear in future since now there is populate_sdk task which can generate SDK for any image Jan 23 04:01:06 EddyLaiTW: so essentially you would do bitbake -cpopulate_sdk core-image-minimal and it should generate SDK which has dev libs and headers for packages in core-image-minimal Jan 23 08:34:24 good morning Jan 23 09:23:00 morning all Jan 23 09:48:21 hi everybody. I'm trying to test OpenGL ES on my Boundary Devices Nitrogen6X. I've got a Yocto build running accelerated X with EXA and video decoding. I've built es2_info and es2gears but I'm getting these errors: http://pastebin.com/raw.php?i=b9UDsNiN Jan 23 09:49:05 rburton: morning Ross, I tried to bisect the image with no success at all, a few revisions didn't compile and a couple more did odd things as they were clearly meant to be part of a patchset Jan 23 09:49:05 I'm not convinced it is the systemd changes now, but I think I'm going to wipe tmp, update to master again today and try again Jan 23 10:12:22 jackmitchell: ok, good luck. Jan 23 10:12:34 jackmitchell: what image was it again? Jan 23 10:13:00 it's a custom image, but loosely based on poky, core-image-minimal Jan 23 10:13:08 panda84kde: sounds like you should speak to whoever provided your GL stack Jan 23 10:13:19 jackmitchell: ok i'll do a build here too. sysvinit i presume? Jan 23 10:13:25 rburton: yes Jan 23 10:14:20 rburton: there is a but more too it though as I am also using meta-ti, but that is just for tune options as I use a kernel not compiled in bitbake Jan 23 10:14:37 rburton: thank you. I'll try in the Freescale forum Jan 23 10:14:59 I get a feeling it could maybe be udev related but we'll see... Jan 23 10:46:27 rburton: my latest image with a fresh tmp boots, and I can login as root but it still has issues Jan 23 10:46:45 it does actually seem to be init related as a lot of services don't start Jan 23 10:47:05 it seems to have an issue with the opkg postinstall service Jan 23 10:50:20 ah, the filesystem is read-only Jan 23 10:50:28 I bet that has something to do with it ;) Jan 23 10:51:21 nowhere is fstab can I see where it defines what the root device should be Jan 23 10:51:51 http://pastebin.com/YXfFnP9e Jan 23 10:56:36 f Jan 23 11:04:45 Test Jan 23 11:21:22 good morning Jan 23 11:23:02 good morning Jan 23 11:23:26 are there any known issues with image-mklibs in danny release tarball? can't find anything suitable in bugzilla. i get 2 warnings about modules.order and modules.builtin being missing and then python dies in tmp/sysroots/x86_64-linux/usr/bin/mklibs", line 343, in : default_lib_path = multiarch(["/lib/", "/usr/lib/", "/usr/X11R6/lib/"]) with "OSError: [Errno 2] No such file or directory". the only major thing i changed so far is building with tuni Jan 23 11:23:26 ng for core2, before it was x86_64 generic Jan 23 11:24:23 i don't exactly see which file/dir seems to be missing though using bitbake with highest debug level Jan 23 11:30:22 jackmitchell: how are you booting this? Jan 23 11:31:33 sdcard Jan 23 11:32:56 well i just found a problem with systemd/udev trying to replicate Jan 23 11:35:31 clean up udev persistances Jan 23 11:35:33 incase Jan 23 11:36:16 yeah, its because i've got both udev and systemd's udev in the same deploy Jan 23 11:44:21 jackmitchell: just as i realised what the problem is, marcin submitted a patch. see "[OE-core] [PATCH] bitbake.conf: unbreak all builds with custom DISTRO_FEATURES". Jan 23 11:44:52 rburton: yes, I have just noticed that and re-building now with sysvinit specifically defined :) Jan 23 11:46:00 the next thing though is that existing systemd users will still be broken after marcin's patch... Jan 23 11:46:51 bluelightning: only if they don't set the variable to flip to systemd Jan 23 11:47:01 sure, but out of the box they won't have... Jan 23 11:47:20 maybe since previously systemd was an add-on then we can live with that, not sure Jan 23 11:47:21 could meta-systemd override that variable? Jan 23 11:47:29 in layer.conf it could, yes Jan 23 12:05:28 what is the easiest way to build a x environment with yocto project? is adding x11-sato to my IMAGE_FEATURES? Jan 23 12:06:46 gbrennon: x11-base is a bit leaner Jan 23 12:06:59 okok Jan 23 12:17:11 rburton when building a stable application in my board, should i make all my recipes or use this premade packages? Jan 23 12:20:45 gbrennon: you'll probably want to use something home-grown but similar - this pulls in a terminal :) Jan 23 12:23:32 rburton: hi Jan 23 12:23:38 hi hrw Jan 23 12:23:52 rburton: my fix looks fine but does not fixed Jan 23 12:23:57 oh no! Jan 23 12:24:20 13:19 hrw@puchatek:build$ bitbake bash -e|grep "^DISTRO_FEATURES=" Jan 23 12:24:26 DISTRO_FEATURES="x11 alsa argp ext2 largefile usbgadget usbhost xattr nfs zeroconf ipv4 ipv6 libc-backtrace libc-big-macros libc-bsd libc-cxx-tests libc-catgets libc-charsets libc-crypt libc-crypt-ufc libc-db-aliases libc-envz libc-fcvt libc-fmtmsg libc-fstab libc-ftraverse libc-getlogin libc-idn libc-inet-anl libc-libm libc-libm-big libc-locales libc-locale-code libc-memusage libc-nis libc-nsswitch libc-rcmd libc-rtld-debug libc-spawn libc-streams libc-sunrpc libc-utm Jan 23 12:24:37 and image lacks initscripts Jan 23 12:24:41 but but but Jan 23 12:25:10 moment - need to look exactly at rootfs - so far tried to boot it only Jan 23 12:25:24 sory, they are there Jan 23 12:25:41 booting again Jan 23 12:25:59 INIT: version 2.88 booting Jan 23 12:25:59 Starting Bootlog daemon: bootlogd. Jan 23 12:25:59 hwclock: can't open '/dev/misc/rtc': No such file or directory Jan 23 12:25:59 Wed Jan 23 12:11:00 UTC 2013 Jan 23 12:25:59 hwclock: can't open '/dev/misc/rtc': No such file or directory Jan 23 12:26:02 INIT: Entering runlevel: 5 Jan 23 12:26:04 Stopping Bootlog daemon: bootlogd. Jan 23 12:26:07 hrw: nice to have you back on OE btw ;) Jan 23 12:26:07 INIT: no more processes left in this runlevel Jan 23 12:26:22 hm. Jan 23 12:26:29 rburton: it is a pleasure to see old faces Jan 23 12:26:57 rburton: update-rc.d misbehaves I think Jan 23 12:27:16 I have nfsserver and openssh installed but no /etc/rc*.d/ links Jan 23 12:27:22 hrw: the class is conditionalised Jan 23 12:27:48 I see Jan 23 12:28:01 but sysvinit is in D_E as you see ^^ Jan 23 12:28:23 yeah... Jan 23 12:28:30 ordering? the inhert happens before the backfill? Jan 23 12:29:05 13:28 hrw@puchatek:build$ bitbake openssh -e|grep "^INIT_D_DIR=" Jan 23 12:29:05 13:29 hrw@puchatek:build$ Jan 23 12:29:08 looks like Jan 23 12:29:22 INIT_D_DIR is in update-rc.d_real.bbclass Jan 23 12:29:35 hm. Jan 23 12:30:55 mh. Jan 23 12:33:10 i've build my core-minimal-image with the x11-base Jan 23 12:33:14 and when i boot my imx28 Jan 23 12:33:18 i just have the tux Jan 23 12:33:22 doenst show any console Jan 23 12:33:33 or x server :( Jan 23 12:33:33 what could i have done wrong? Jan 23 12:33:42 gbrennon: wrong console option? Jan 23 12:34:01 gbrennon: if you can, try a plain core-image-sato to verify that -sato is working fine Jan 23 12:34:06 hm Jan 23 12:34:07 okok Jan 23 12:34:13 so, i'll remove the x11-base Jan 23 12:34:18 and build a core-image- sato, right? Jan 23 12:35:26 yeah Jan 23 12:35:27 rburton: ideas for fix? other then adding DISTRO_FEATURES_INITMAN to DISTRO_FEATURES because this one I already implemented but some people will be still affected Jan 23 12:35:44 hrw: no. RP? Jan 23 12:36:32 rburton, okok Jan 23 12:37:46 rburton, the sato is come kind of x environment like gnome, kde and e others? Jan 23 12:38:30 gbrennon: yes Jan 23 12:38:33 gbrennon: its matchbox desktop+panel+window manager basically Jan 23 12:38:49 gnome mobile style Jan 23 12:38:50 rather minimal but surprisingly usable on a small touchscreen Jan 23 12:39:00 hrw: gnome doesn't say gnome mobile anymore :) Jan 23 12:39:16 rburton: but we did 5y ago ;D Jan 23 12:39:25 hrw: those were the days! Jan 23 12:40:50 rburton: shame none adapted it for qvga Jan 23 12:41:35 ant_work: at the time we decided that qvga was on its way out. patches welcome, you just need a gtk theme that draws really really small :) Jan 23 12:42:29 I've run it for fun, needs only minor adjustments Jan 23 12:42:42 rburton: the bug above is caused by mklibs calling dpkg-architecture without checking for its existence first. manually installing dpkg-devel fixes this, but it is not listed in the docs as requirement for building yocto. also, for the apt recipe, there is a patch which discards dpkg-architecture and uses a workaround. this issue seems to be present in the git version of danny as, as the recipes for mklibs are the same and it worked before for generic x8 Jan 23 12:42:42 6_64 as said. is this worth filing a new bug? Jan 23 12:43:53 schramae: yes, and presumably it happens in master too Jan 23 12:44:24 ant_work: at Bug Labs we used sato on qvga screens. hardly useful indeed Jan 23 12:45:12 rburton: ok, opening a new one. there seems to be more going on, now it dies at installing libX11 Jan 23 12:45:45 hrw: you can exit only by reaching console :) Jan 23 12:46:16 ant_work: sato never had 'logout' option iirc Jan 23 12:46:36 hmm sato seems to be a nice environment! how it works with qt? Jan 23 12:47:09 schramae: log? Jan 23 12:47:16 hrw: ohh, I though it was just drawing the buttons out of screen.. Jan 23 12:47:35 gbrennon: it's just X, so qt apps will run Jan 23 12:48:03 hrw, ant_work: there's a "shutdown" package you can install if you want a shutdown button - the assumption was that the hardware has a power switch. Jan 23 12:48:15 the shutdown button is only installed in qemu images as they don't have a button Jan 23 12:48:20 rburton: I don't think we can fix this for everyone 100% Jan 23 12:48:46 rburton: -sato was my testbed to verify the interaction with xserver from meta-oe Jan 23 12:48:49 RP: the regression is if you set your own DISTRO_FEATURES, the backfill isn't enough to get initscripts. Jan 23 12:48:55 (no screen at the time :p) Jan 23 12:49:18 rburton: ah, hmm :/ Jan 23 12:49:35 rburton: talking about that, we still need xinput-calibrator in core imho Jan 23 12:49:52 ant_work: yes. want to submit a patch? :) Jan 23 12:50:01 RP: yes :/ Jan 23 12:50:32 rburton: sure, noted Jan 23 12:50:55 RP: the backfill appears to happen after the inherit update-rcd (which then checks D_F and only inherits update-rcd.real if sysvinit) Jan 23 12:52:46 rburton: right, its implemented in anonymous python Jan 23 12:52:59 rburton: http://pastebin.com/L5hRvHSa not yet sure what causes this, i did a clean rebuild from tarball release (was git version before) and tuned for core2_64 instead of generic x64. USER_CLASSES ?= "buildstats image-mklibs image-prelink" is the default in local.conf Jan 23 12:58:25 Who is the rpm guru? Jan 23 12:58:45 schramae: never actually used image-mklibs, and i suspect it's not tested greatly. Jan 23 12:59:55 schramae: is "don't set MKLIBS_OPTIMIZED_IMAGES" an answer you'll be happy with? :) Jan 23 13:01:28 rburton: as my deadline is friday, it might become one tomorrow. ;) but i'll investigate a little. wondering why this option is enabled per default then Jan 23 13:03:31 schramae: the class is enabled but doesn't do anything unless you set that variable Jan 23 13:04:03 sounds like it needs to be fixed or removed as a default though Jan 23 13:04:07 hi! Jan 23 13:04:11 what is the correct way to have a recipe depending on libncurses? I have an autotools based application and tried to add 'ncurses' to its DEPENDS with no luck. I mean my application gets compiled and a corresponding libncurses ipk gets correctly built, but when I deploy my recipe to an image, the application complains for missing library. Jan 23 13:04:42 it can be fixed by manually installing libconfig package, but I can't get it automatically brought in by the deps chain Jan 23 13:04:59 what am I missing? Jan 23 13:06:09 btw, I'm building with denzil here... Jan 23 13:06:24 gizero: sounds like the automatic link-time dependency finding isn't working, or you're breaking it. can you share your recipe? Jan 23 13:08:05 gizero: libconfig? that's not ncurses. Jan 23 13:09:23 ;-) sorry.... that's another dependency... I meant 'manually installing libncurses fixes it' Jan 23 13:09:40 sounds like you're wiping the runtime dependencies Jan 23 13:10:06 are you setting RDEPENDS yourself? Jan 23 13:10:42 I'm appending "screen" to it via += Jan 23 13:11:07 I can share the recipe in a while Jan 23 13:11:26 RDEPENDS_${PN} += "screen" should work fine Jan 23 13:12:19 rburton: grepped the whole dir for MKLIBS_OPTIMIZED_IMAGES - does not show up and i don't set anything else related to this explicitly as far as i'm aware of Jan 23 13:12:57 rburton: in fact I'm appending like this. see http://pastebin.com/a7j2Jk5A Jan 23 13:13:03 schramae: that's very clever, i presumed you'd set it in local.conf. remove it from the classes to make sure it goes. Jan 23 13:13:52 gizero: maybe the binaries are being installed somewhere where the dep finding code isn't looking? Jan 23 13:14:02 oh! Jan 23 13:14:06 oh. no. Jan 23 13:19:25 rburton: do you mean the application binaries? they end up being installed in /usr/bin Jan 23 13:20:50 rburton: as you can see the recipe already depends on other libraries which get correctly picked up. this seem to be ncurses related in this context Jan 23 13:21:55 rburton: simply removing it works just fine... Jan 23 13:26:09 gizero: I have an ncurses linked application and just having depends in ncurses allows me to pick it up properly Jan 23 13:27:29 gizero: AC_SEARCH_LIBS([initscr, curs_set, getmaxyx, newwin],[ncurses], , AC_MSG_WARN(["ERROR: Could not find nCurses library!"])) Jan 23 13:27:38 gizero: is my autoconf pickup line Jan 23 13:39:41 I don't know why we need PROVIDES, can someone help me to explain? eg: perf_3.4.bb:PROVIDES = "virtual/perf", I think why we need this, if not have this, what's the side effect Jan 23 13:41:41 lyang0_: PROVIDES satisfies DEPENDS (i.e. build-time dependencies) Jan 23 13:42:33 Why we not directly use DEPENDS +="perf" directly ? Jan 23 13:42:39 Crofton|work: fray or myself I guess Jan 23 13:42:41 what's what I confused Jan 23 13:43:15 lyang0_: I think for perf it is to handle where it may be built as part of the kernel (as it was previously) Jan 23 13:43:30 bluelightning, I posted to the list Jan 23 13:44:15 I don't understand Jan 23 13:44:17 Crofton|work: that is certainly unusual... Jan 23 13:44:59 yes Jan 23 13:45:03 lyang0_: where we have a "virtual/something" build-time dependency relationship, it is because we want to allow for more than one thing to satisfy that dependency and allow it to be selected at build time Jan 23 13:45:15 I kept hearing from a guy that uhd wouldn't build for him ..... Jan 23 13:45:27 the rpm bit was a surprise :) Jan 23 13:45:50 I figure there is some corner case that is not expected, so worth figuring out Jan 23 13:46:06 lyang0_: specifically for perf it comes as part of the linux kernel source; some may choose to build it as part of the kernel build still and the virtual dependency allows for that Jan 23 13:46:27 rburton: i used the wrong file mask for grepping, MKLIBS_OPTIMIZED_IMAGES was/is enabled Jan 23 13:48:58 jackmitchell: here configure uses PKG_CHECK_MODULES([ncurses], [ncurses]) Jan 23 13:49:43 bluelightning: if more than one thing to provide the same dependency,why not directly depend it but depend on virtual eg: A.bb PROVDES="virtual/something" B.bb PROVDES="virtual/something", I think you mean we use Jan 23 13:49:49 I was assuming ncurses was pkg-config friendly... could this be the cause for the dependency not being picked up? Jan 23 13:50:11 PREFERRED_PROVIDER to select. but Why not directly depends on B or A Jan 23 13:50:28 this is the question 1 Jan 23 13:51:09 lyang0_: I'm not sure if that will allow switching the dependency properly when both are present Jan 23 13:52:11 what do you mean Jan 23 13:53:31 it seems the PROVIDES can help us easy to swith in conf but not in bb.that's maybe the better place Jan 23 13:55:00 ? Jan 23 13:55:46 bluelightning question2: some may choose to build it as part of the kernel build still and the virtual dependency allows for that, do you mean PROVDES="virtual/perf" is not the same function as PROVIDES="perf" Jan 23 13:56:01 kergoth: ?? Jan 23 13:56:38 I don't know "virtual" has the relation with "kernel" Jan 23 13:58:00 lyang0_: each recipe that could provide perf would have PROVIDES = "virtual/perf" and then you'd set PREFERRED_PROVIDER_virtual/perf to select which one to use Jan 23 13:58:16 yes Jan 23 13:58:18 lyang0_: setting the latter in distro configuration is the right place to do it Jan 23 13:58:54 right Jan 23 13:59:36 this is match what I think "it seems the PROVIDES can help us easy to swith in conf but not in bb.that's maybe the better place" when we depend it we don't need to change in bb but in conf Jan 23 14:00:15 the distro configuration is where decisions like this are supposed to be made Jan 23 14:00:23 i.e. policy decisions Jan 23 14:00:57 A.bb PROVDES="virtual/something" B.bb PROVDES="virtual/something" , DEPENDS="aa" or DEPENDS="bb" is a bad choice but good with PREFERRED_PROVIDER + DEPENDS="virtual/something" right? Jan 23 14:01:04 gizero: possibly not, I can't say for sure I'm afraid Jan 23 14:01:45 bluelightning, how about question2 Jan 23 14:01:58 some may choose to build it as part of the kernel build still and the virtual dependency allows for that, do you mean PROVDES="virtual/perf" is not the same function as PROVIDES="perf" Jan 23 14:01:59 lyang0_: correct, if there is a selection to be made Jan 23 14:02:12 lyang0_: it is not the same no Jan 23 14:02:36 lyang0_: note that "virtual/perf" has no special meaning, it's just a convention Jan 23 14:03:30 you could use "virtual/bob" or "cricketbats" in its place and it would work, all that matters is that there is a value in PROVIDES that satisfies the value in DEPENDS Jan 23 14:04:48 so can I change all the virtual/perf to "perf" in PROVIDES, and in DEPENDS="virtual/perf" to DEPENDS="perf" Jan 23 14:06:13 lyang0_: I don't think so no Jan 23 14:06:20 it seems if change perf.bb to perfxxx.bb,we can do that, when PROVIDES 's value is the same as PN, seems not, right? Jan 23 14:06:39 correct Jan 23 14:06:46 why would you want to change it though? Jan 23 14:08:01 eg: perf.bb--> PROVDES="virtual/perf" right, PROVDES="perf" ,wrong. but perfxxx.bb--> PROVDES="virtual/perf" right, PROVDES="perf" right Jan 23 14:08:07 what do you think Jan 23 14:08:48 that will work, but we would not make that change Jan 23 14:08:54 I just want to learn it not really change/ Jan 23 14:08:57 ok Jan 23 14:09:02 then yes Jan 23 14:09:08 rburton: now i have my x with sato! thx dude! but my x failed to some services :( Jan 23 14:09:21 question3: Jan 23 14:09:21 rburton so X only loaded somekind of bash terminal Jan 23 14:10:58 gettext_0.16.1.bb:PROVIDES = "virtual/libintl virtual/gettext", does DEPENDS="virtual/libintl" == DEPENDS="virtual/gettext" bluelightning Jan 23 14:11:06 rburton what should be the screen that i should expect with the core-image-sato package? Jan 23 14:11:31 lyang0_: no, those two are separate Jan 23 14:13:17 I don't see the different effect for DEPENDS="virtual/libintl" and DEPENDS="virtual/gettext" Jan 23 14:13:40 if only gettext_0.16.1.bb provide them Jan 23 14:13:43 I mean only Jan 23 14:15:23 sure, if there is only one provider Jan 23 14:15:41 the purpose is to handle when there are (or might be) multiple providers Jan 23 14:16:00 thanks, from the QA, I think I have known it Jan 23 14:16:00 thanks Jan 23 14:18:24 ok, first go at a systemd image and I get the following: Jan 23 14:18:40 [ 1.891670] EXT4-fs (mmcblk0p2): mounted filesystem with ordered data mode. Opts: (null) Jan 23 14:18:40 [ 1.900335] VFS: Mounted root (ext4 filesystem) readonly on device 179:2. Jan 23 14:18:40 [ 1.912877] devtmpfs: mounted Jan 23 14:18:40 [ 1.916474] Freeing init memory: 272K Jan 23 14:18:40 [ 2.631549] systemd[1]: Failed to mount /run: No such file or directory Jan 23 14:19:38 bluelightning when we do a new bb, Is there a quick method to know it already has mulit provider Jan 23 14:21:01 lyang0_: I think you can assume none exists most of the time Jan 23 14:21:43 hello bluelightning, can you help me with sato environment? Jan 23 14:21:58 i've got some errors when starting x Jan 23 14:22:34 gbrennon: possibly, can you pastebin the errors? Jan 23 14:22:40 okok Jan 23 14:22:48 this is my errors, in the board Jan 23 14:22:48 http://pastebin.com/B1XkwsGg Jan 23 14:23:16 the errors are from the x environment. all processes from the bitbake build have completed sucessfully Jan 23 14:23:55 bluelightning: do you mean if multiprovder provide the samething it will trigger error for us? Jan 23 14:24:10 then we can create fix with PROVIDES? Jan 23 14:26:05 lyang0_: I mean usually it's pretty obvious when you're adding something that provides exactly the same library or other functionality that is already there - it's not a very common situation in practice Jan 23 14:26:32 oh Jan 23 14:27:24 gbrennon: what if you put something valid in /usr/lib/X11/xinit/xinitrc ? Jan 23 14:30:35 :q Jan 23 14:31:30 hmm so many file conflicts in world-image :/ Jan 23 14:33:35 buelightning, question4. a_git.bb and a_1.0.bb we use PREFERRED_VERSION_a ="1.0" to use 1.0, if we want to use git version, can we use PREFERRED_VERSION_a ="git"? Jan 23 14:34:07 git is not number, I don't know if it can be used here Jan 23 14:35:40 bluelightning valid? what should i put? when i built my image with the bitbake it didn't made me a stable x? Jan 23 14:35:55 and the command to start sato, should be startx like x? Jan 23 14:44:16 lyang0_: it would need to match up with whatever PV is set to for the git recipe... which traditionally is something like "1.1+git${SRCPV}" Jan 23 14:44:44 lyang0_: you can use % as a wildcard to match the value of SRCPV, e.g. PREFERRED_VERSION_a = "1.1+git%" Jan 23 14:45:13 gbrennon: have you installed any X applications/environment in your image? Jan 23 14:45:23 thanks Jan 23 14:45:28 I will try it Jan 23 14:50:03 bluelightning: sorry, got some dc Jan 23 14:50:24 bluelightning: at the first time i've built my img with the sato environment and got this error. now i'm trying to build it with only x11 and got the same error Jan 23 14:52:25 gbrennon: AFAIK in order for X to stay running you'll need some kind of environment or application to run Jan 23 14:52:37 gbrennon: I don't know why sato wouldn't work out of the box Jan 23 14:53:52 bluelightning: this environment or application that u keep saying could be some kind of window management stuff like gnome, kde, lxfce, bla bla? Jan 23 14:54:01 gbrennon: yes Jan 23 14:54:34 hm ok Jan 23 14:55:17 bluelightning: wtf is going then? this sato is going crazy haha it didn't ran with x hm have u used sato or some other x with yocto b4? Jan 23 14:56:20 gbrennon: I've built and booted images with sato many times for testing Jan 23 14:56:41 gbrennon: in fact I just built one today from master and ran it now (for qemux86), and it starts up fine Jan 23 14:57:34 bluelightning: u just built with core-image-sato? Jan 23 14:57:39 yep Jan 23 14:58:29 hm Jan 23 14:58:40 what could be wrong? i made the same hm Jan 23 14:59:35 have u ever tested yocto on a board or only on qemu? Jan 23 15:00:50 gbrennon: I have yes, though not on imx28 Jan 23 15:01:10 bluelightning: hm nice. which boards have u built the yocto? Jan 23 15:02:18 too many to fully list... but intel black sand and beagleboard most recently Jan 23 15:03:08 bluelightning oh nice! do u work for intel or something? Jan 23 15:03:12 I do yes Jan 23 15:06:52 bluelightning are u graduate in some engineering? in which sector do u work? Jan 23 15:07:18 gbrennon: I have a CS degree... I'm a developer Jan 23 15:08:44 bluelightning what is ur main activity as a developer in intel? ur focus are in embedded systems or everything? in which platform do u develop for intel? Jan 23 15:09:15 gbrennon: I work only on the Yocto Project, so I'm not working on a specific platform Jan 23 15:10:34 bluelightning, hmm now i got it. so do u work developing yocto port to the intel boards? Jan 23 15:10:59 bluelightning, does opie work :) Jan 23 15:11:50 Crofton, opie? what could be "opie work"? i don't know this expression Jan 23 15:11:56 gbrennon: no, I work on the build system itself, i.e. bitbake and the core classes, occasionally recipes Jan 23 15:12:07 Crofton|work: last I checked :) Jan 23 15:12:11 bluelightning, nice Jan 23 15:12:17 and opie :) Jan 23 15:12:28 bluelightning, i'm out for lunch dude, cya Jan 23 15:12:28 gbrennon, he's been around OE for a very long time Jan 23 15:12:36 Crofton|work: touchscreen seems broken with modern kernels, I haven't had a chance to debug it yet Jan 23 15:12:42 bummer Jan 23 15:13:05 you should get something running opie for FOSDEM Jan 23 15:13:39 Crofton|work: heh... maybe Jan 23 15:13:54 it is a fairly attractive ui Jan 23 15:14:04 it could use a new theme Jan 23 15:14:26 yeah, but it would be fun to see who recognized it and who thought it was some cool new thing Jan 23 15:14:30 I wouldn't necessarily call it attractive, it's not totally awful though Jan 23 15:14:36 bluelightning, is opie still maintained or you switched to gta04? Jan 23 15:14:44 hi btw Jan 23 15:14:48 denisATeukrea: hey Jan 23 15:14:56 denisATeukrea: it's still maintained although I haven't done much work on it in the last few months Jan 23 15:15:02 ok Jan 23 15:15:26 problems such as the touchscreen thing really sap my motivation :/ Jan 23 15:15:36 ah that's fixable Jan 23 15:15:41 easily even Jan 23 15:15:47 think so? Jan 23 15:15:50 it's only JaMa that wants the proper fix Jan 23 15:15:59 so he didn't merge any workarrounds Jan 23 15:16:04 why, have you dealt with this same problem? Jan 23 15:16:19 yes I know the issue well Jan 23 15:16:26 (touchscreen of the gta04) Jan 23 15:16:30 it's a software issue Jan 23 15:16:41 oh, no this is a generic touchscreen problem although I experienced it on gta04 as well Jan 23 15:16:50 ok Jan 23 15:17:05 because on gta04 the pointers jumps a bit sometimes Jan 23 15:17:14 ok, that's a different problem Jan 23 15:17:20 ok Jan 23 15:17:35 my issue is it's reporting wacky values or qt/e is interpreting them incorrectly so it's basically useless Jan 23 15:17:50 affects many different devices Jan 23 15:17:51 did it work before? Jan 23 15:18:01 on some of them yes Jan 23 15:18:07 ok Jan 23 15:18:16 the annoying thing is it's using tslib and ts_test works fine Jan 23 15:18:23 oh Jan 23 15:18:31 I don't understand how that could be the case, but it is Jan 23 15:18:36 ok Jan 23 15:18:42 strange Jan 23 15:18:50 that's weird. ts_test and qt/e are using the exact same api Jan 23 15:18:56 I've ported an imx25 board to dt and I've the same issue with xorg Jan 23 15:19:06 I calibrate in ts_calibrate Jan 23 15:19:13 and xorg is still uncalibrated after Jan 23 15:19:16 kergoth: that's what I thought based on my examination of the code Jan 23 15:21:42 hmmm Jan 23 15:21:46 denisATeukrea: try xinput-calib from meta-oe Jan 23 15:21:58 yes I know it Jan 23 15:22:02 xinput-calibrator Jan 23 15:22:06 yeah, that :) Jan 23 15:22:11 i'm still missing something with a touchscreen to verify this stuff :( Jan 23 15:22:22 bluelightning, when I think about it....maybe it's the calibration Jan 23 15:22:28 qte has its own calibration apps Jan 23 15:22:35 denisATeukrea: it does, and that is custom code Jan 23 15:22:43 ok Jan 23 15:22:59 rburton, hrw may be giving some away at FOSDEM Jan 23 15:23:02 denisATeukrea: recently though I tried using calibration data produced from ts_calibrate and that didn't fix the problem :( Jan 23 15:23:49 bluelightning, I used to test on bug20, but I can't anymore, it has a 2.6.35 and that doens't boot anymore with systemd Jan 23 15:23:51 Crofton|work: i was hoping for something better than a 770 :) Jan 23 15:23:58 so I can't test opie anymore Jan 23 15:24:05 other devices are different Jan 23 15:24:14 like non-standard touchscreens etc... Jan 23 15:24:24 Crofton|work: I think that rburton used 770 year before world heard about them Jan 23 15:24:26 and you already have the gta04 Jan 23 15:24:40 hrw: still got a n810 that i can't seem to get rid of Jan 23 15:24:57 rburton: friend borrowed mine from me ~3 years ago Jan 23 15:25:21 denisATeukrea: no worries about not testing opie, but couldn't you just use sysvinit on bug20? Jan 23 15:25:44 rburton: he also has m gta01bv4 + debug board and ngw100 avr32 board I got at fosdem years ago Jan 23 15:25:46 what does KMETA in the linunx-yocto do? Jan 23 15:26:07 bluelightning, yes but what distro still supports sysvinit? Jan 23 15:26:27 Error: No calibratable devices found. #xinput_calibrator Jan 23 15:26:28 denisATeukrea: poky does... Jan 23 15:26:41 FWIW Jan 23 15:26:42 can angstrom works without systemd? Jan 23 15:26:53 theoratically Jan 23 15:27:03 denisATeukrea: thats annoying isn't it :( Jan 23 15:27:16 does systemd.bbclass go based on the distro feature yet? i need to get ca ught up on all of that Jan 23 15:27:32 rburton, what? the bug20 not booting? or the work stuff with the touchscreen? Jan 23 15:27:38 kergoth: its not in oecore yet - i wanted to review it first Jan 23 15:27:44 ah Jan 23 15:27:44 denisATeukrea: the calibrator Jan 23 15:27:48 fair enough Jan 23 15:27:48 yes Jan 23 15:37:13 i'm back Jan 23 15:40:46 Crofton|work: meta contains the fragments Jan 23 15:51:48 I have been having a battle with an evil proxy :/ most files are being downloaded but 1 is failing Jan 23 15:51:58 I have managed to download the proper file manually Jan 23 15:52:18 is there a way that I can "inject" it into the build process Jan 23 15:52:45