**** BEGIN LOGGING AT Tue Jan 24 02:59:57 2012 Jan 24 07:57:00 good morning :) Jan 24 08:05:10 the ages of darkness are over. i found someon with a spare mini din6 to ipaq connector(came with a gpsmouse once). need go get it, fiddle a subd9 con onto it and hope that the +-6V i read about that the ipaq delivers is enough for the uart controller in my pc. to make the kernel paint beautiful letters onto my terminal. :) Jan 24 08:15:18 cu later Jan 24 08:53:34 good morning Jan 24 09:28:37 morning all Jan 24 09:50:14 any light weighted notepads Jan 24 10:24:18 hi bluelightning, all Jan 24 10:46:17 morning all Jan 24 10:57:25 good morning Jan 24 10:58:20 bluelightning: hi. I'm trying to compile meta-toolchain-qte and it fails on qt4-native (compile.test-lflags.patch & configure-lflags.patch don't apply) and then on qt4-tools-nativesdk with a "x86_64-angstromsdk-linux-ld: unrecognized option '-Wl,-rpath-link, ..." message Jan 24 10:59:17 bluelightning: did you manage to compile meta-toolchain-qte with qt 4.8 ? Jan 24 11:01:30 Hello! Jan 24 12:17:03 ant_work: any hint why /usr/bin/kexecboot is now created everytime on my host (not in rootfs/initramfs) ? Jan 24 12:17:35 do you have merged all pending patches? Jan 24 12:17:41 y Jan 24 12:18:26 probably is +MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS_append = " kexecboot-cfg " Jan 24 12:19:00 ant_work: to be cleaer it's really on my host not target image Jan 24 12:19:25 ant_work: bitbake user can write to system's /usr/bin/ because it's minimalistic chroot owned by same user.. Jan 24 12:20:05 ant_work: so normaly it would fail to write there unless someone is running bitbake under root, but it didn't create this until recently Jan 24 12:20:18 hm... Jan 24 12:20:31 ant_work: if you don't know .. I'll narrow it a bit (if it's during kexecboot build or initramfs build etc..) Jan 24 12:21:35 you see what I was aiming at Jan 24 12:22:31 as far I can rememember only kexecboot_cfg is built on each image build Jan 24 12:22:40 no, I don't see how +MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS_append is related.. but I'll try Jan 24 12:22:44 there musty be an error somewhere Jan 24 12:22:57 isn't RDEPENDING on kexecboot? Jan 24 12:23:29 ah, no Jan 24 12:23:36 the point is that it was building kexecboot before, just the binary wasn't left in system /usr/bin Jan 24 12:24:25 florian: hi, you at ew2012 again? Jan 24 12:24:26 I tested in this way: 1) bitbake zaurus-installer 2) bitbake core-image-sato Jan 24 12:24:38 on 2) I saw only kexecboot_cfg rebuilding Jan 24 12:24:44 ant_work: nwm.. Jan 24 12:25:15 I have a fresh build of tonight to inspection at home, just a core-image Jan 24 12:25:51 likewise: yes, that's very very likely :) Jan 24 12:26:54 ant_work: but you're not running bitbake under root or have /usr/bin writeble by everybody, are you? So we're talking about different issues.. Jan 24 12:27:38 florian: haha, same here. Another question: I am looking for a PEEDI or Abatron config file for the iMX28, do you happen to have one? Or do you know from what iMX the iMX28 is derived? Jan 24 12:28:12 florian: PEEDI/Abatron do not have iMX28 configs at their site. I think you did earlier work on iMX2x, right? Jan 24 12:31:40 likewise: I didn't use one of these... but yes, I did hacked on the TX-25 a little bit soem time ago. Jan 24 12:32:30 I've changed my kernel config to make musb a module. To get it into my rootfs I only need to add it to IMAGE_INSTALL ? Jan 24 12:35:12 tasslehoffwrk: this should work... depending on the distro you might have additional variables you can add it too. Jan 24 12:36:36 JaMa: no,I run as user Jan 24 12:46:33 JaMa: hmm, maybe is the postinst Jan 24 12:47:08 can you check the logs and see the exit status? Is it run on device? Jan 24 12:53:33 florian: cool. thanks Jan 24 13:12:05 hm. it seems my module was not built Jan 24 14:06:40 ericben: I have not tried I have to admit Jan 24 14:06:57 ericben: I would not expect those problems with qt4-native though, that built fine here last I tried Jan 24 15:02:09 is anyone currently building images for atmel boards with meta-atmel? http://gitorious.org/oe-atmel/meta-atmel Jan 24 15:09:12 florian: hi, locale is working^^ Jan 24 15:09:31 florian: sometimes it's good to reboot -.- Jan 24 15:12:00 xumbi: never tried Jan 24 15:15:10 mckoan: ah, well I'm going to give it a shot with at91sam9g20ek, the official atmel supported OE tree is ancient Jan 24 15:16:29 linux4sam.org says to use branch origin/stable/2009 Jan 24 15:26:19 xumbi: I use my branch mckoan/kaeilos-2011 Jan 24 15:27:31 xumbi: http://www.kaeilos.com/ Jan 24 15:28:27 ah yes, I hadn't thought to try kaeilos, I'll give it a try if this doesn't work out Jan 24 15:54:34 bluelightning: hi, I think this comes from the patch version in FC16 which is less tolerant than previous ones because here configure-lflags.patch fails Jan 24 15:54:55 bluelightning: btw this is in qt4-tools-nativesdk not native Jan 24 16:00:39 bluelightning: I'm launching a clean build I'll keep you informed of the result Jan 24 16:06:56 ericben: ah right, that makes more sense Jan 24 16:44:05 Hi, I have a problem with chmod while compiling Jan 24 16:44:26 chmod: ss_err.c: new permissions are r-xrwxrwx, not r-xr-xr-x Jan 24 16:44:26 make[2]: *** [ss_err.c] Error 1 Jan 24 16:44:58 it is related to e2fsprogs-1.41.14-r30 Jan 24 16:47:37 in logfile: when changing chmod for two *.c and *.h files comes error Jan 24 16:47:41 I'm using angstrom / oe on an atmel micro, and I'm looking to modify my kernel config. bitbake -c menuconfig virtual/kernel does it.. Jan 24 16:47:46 but this is blown away when I clean. Jan 24 16:48:13 What is the proper way to set up a custom kernel configuration? Jan 24 16:48:45 the recipe in atmel calls " yes '' | oe_runmake at91sam9x5ek_defconfig " Jan 24 16:50:10 fullstop, there are many ways Jan 24 16:50:29 GNUtoo: I assume so.. :) Jan 24 16:50:32 the one you described works fine but don't save the config Jan 24 16:50:42 else go in workdir of the kernel Jan 24 16:50:48 and do make ARCH=arm xconfig Jan 24 16:50:50 save the config Jan 24 16:50:51 I modified my kernel recipe to point at a custom location for the config install, and then used menuconfig to reconfigure, and save the config at the aforementioned location Jan 24 16:50:59 and put it in oe Jan 24 16:51:13 I can do that too, I suppose.. Jan 24 16:51:15 problem was any time I updated I had to go in and re-mod the recipe Jan 24 16:53:54 It seems unusually hard to do this task within oe/bitbake.. Jan 24 16:54:07 Keep in mind that I'm coming from a buildroot world Jan 24 16:54:47 fullstop, sorry I missed the context.. what are you having problems with? Jan 24 16:55:08 There is an option to "bitbake -c menuconfig virtual/kernel" Jan 24 16:55:09 we're trying to understand usage sceneros that people want that may either not be well defined or not implemented in bitbake/oe-core Jan 24 16:55:12 and this works great Jan 24 16:55:24 but these changes are lost when I clean Jan 24 16:55:49 so you want a version of menuconfig that also copies the new file back to the recipe source directory, right? Jan 24 16:55:52 ok.. that is something we've discussed a bit (at Wind River, my employer).. but I'm not sure how that should work in the context of oe.. Jan 24 16:56:04 I'm not saying that I want to copy it back.. Jan 24 16:56:09 I will bring it up though at the next Yocto project meeting.. Jan 24 16:56:18 but it's not clear (through a lot of google searches) on how to accomplish what I need to do. Jan 24 16:56:36 The easiest thing I can think of is telling menuconfig to save it in an alternate location, and then save in default Jan 24 16:56:57 I've modified my recipe to copy it from a directory rather than do what it is doing now.. Jan 24 16:57:01 I think copying it back to the source dir is the only way to accomplish that without a lot of extra infrastructure. OE doesn't, at present, have any other way to do real persistent data storage. Jan 24 16:57:03 and then when after you clean, run menuconfig, and reload your alternate config from where you saved it. Jan 24 16:57:15 and, to be clear, this could be some funny stuff that the at91 folks are doing. Jan 24 16:57:16 If you wanted to add some sort of other persistent store then you could certainly do that too. Jan 24 16:57:44 They build the .config in a recipe dynamically. Jan 24 16:58:14 bitbake, to me, feels more like a nightly build system. Jan 24 16:58:29 it kind of is. Jan 24 16:58:40 although it's more oe in general than bitbake. Jan 24 16:59:21 So, for iterative development, it is kind of painful. Jan 24 16:59:36 Especially since I've not found a way to recompile the kernel without doing a clean first. Jan 24 16:59:46 Even passing -f to bitbake doesn't seem to do it. Jan 24 17:01:01 ya, our (non-bitbake system) we have a mechanism to export changes to things that use menuconfig and re-use them on the next compile.. Jan 24 17:01:11 but clean would remove any temporary changes if you hadn't exported first Jan 24 17:01:22 chmod: ss_err.c: new permissions are r-xrwxrwx, not r-xr-xr-x Error 1 Jan 24 17:01:47 that is very odd.. Jan 24 17:01:49 can anyone tell me how to make this work? Jan 24 17:01:53 which step is this happening under? Jan 24 17:01:55 bluelightning: ok same thing one more time ins aqt4-tools-nativesdk : x86_64-angstromsdk-linux-ld: unrecognized option '-Wl,-rpath-link,/... Jan 24 17:02:11 fray: I guess it's oe-classic Jan 24 17:02:34 fullstop, you should be able to compile w/o a clean.. at least it used to work Jan 24 17:02:44 ant_work -- ahh if thats the case, it's likely a bug in fakeroot behavior.. Jan 24 17:02:59 ibanez2006: still org.openembedded.dev ? Jan 24 17:03:26 ericben: right, the problem there is that it's taking LD from the environment (and overriding our own choice of ${CXX} as the linker), that was a 4.8 change in the configure script Jan 24 17:03:26 yes.. Jan 24 17:04:59 when i compile e2fsprogs-1.41.14-r30 this happens Jan 24 17:05:06 fray: maybe you can not with the kernel? Jan 24 17:05:28 no you should be able to.. if I remember right it's the bitbake -c menuconfig ... and then bitbake -f -c compile Jan 24 17:05:29 looks like chmod is not allowed from this function Jan 24 17:05:55 it's likely fakeroot isn't running -- or fakeroot is doing something with the modes/mask on the file in question Jan 24 17:06:02 (or fakeroot is broken.. all very possible) Jan 24 17:06:08 bluelightning: ok that's why you do a unset LD in do_configure ? Jan 24 17:06:14 ericben: yes Jan 24 17:06:20 bluelightning: ok thanks Jan 24 17:15:49 fray: one more thing -- having the generated in tmp/deploy/... is probably the _last_ place I would look for an image. Jan 24 17:16:54 ya.. it's one of those things that oe seems to do.. I agree, it wasn't where I was expecting them either.. Jan 24 17:17:03 the whole notion of calling your project files "tmp" is odd to me Jan 24 17:17:14 bluelightning: ok now g++ is used to link and I get anew error : | /usr/bin/ld: cannot find /usr/local/angstrom-eglibc-x86_64-armv7a/sysroots/x86_64-angstromsdk-linux/lib/libc.so.6 Jan 24 17:17:17 It seems entirely wrong. Jan 24 17:17:39 disk space is cheap, copy those things to a bin directory. Jan 24 17:18:03 bluelightning: seems that LEFLAGS is creating troubles here Jan 24 17:20:20 fullstop: er, tmp/deploy *is* a copy... Jan 24 17:20:27 bluelightning: gah! Jan 24 17:20:35 why put it in tmp? Jan 24 17:20:44 unless tmp stands for something else.. :) Jan 24 17:20:50 fullstop: tmp is just what we call the output dir Jan 24 17:20:55 fullstop: it's not /tmp Jan 24 17:21:04 (unless you've set it that way) Jan 24 17:21:15 why tmp then? Jan 24 17:21:35 not great terminology perhaps but that's the way it's always been Jan 24 17:21:45 Maybe we could change it? :) Jan 24 17:22:01 Is it the build dir and the deploy munged together? Jan 24 17:22:15 If so, it would be far nicer to copy the deploy images elsewhere. Jan 24 17:22:23 It took me about 15 minutes to find my bin the first time. Jan 24 17:23:46 bluelightning: forget it I missed something in do_compile Jan 24 17:23:58 the deploy dir contains images and packages, plus anything else that is explicitly put there Jan 24 17:23:58 if you have just a bare binary that has been copied in there I'm assuming that was explicit... Jan 24 17:24:25 I'd be happy if deploy was pulled out of "tmp" honestly. Jan 24 17:25:50 fullstop: if you really don't like it you can put it elsewhere simply by setting DEPLOY_DIR in your local.conf Jan 24 17:30:26 bluelightning: OK thanks I'll post a patch the problem seems solved Jan 24 17:30:40 ericben: great, thanks Jan 24 17:30:50 bluelightning: LD need to be forced in several places Jan 24 17:31:12 ericben: really? can you not just unset it at configure time? Jan 24 17:31:14 I'll post the patch after testing the resulting sdk just in case I'm out of luck ;-) Jan 24 17:31:44 bluelightning: thanks! I was unaware of that setting Jan 24 17:31:51 bluelightning: it seems to take g++ from the host when I only unset it Jan 24 17:32:14 ericben: hmm, that's not good Jan 24 17:32:35 ericben: I hope there's no other less obvious host intrusion creeping in as well :/ Jan 24 17:32:59 bluelightning: it hangs latter on a compilation error. I'm out of luck ;-) Jan 24 17:42:13 ericben: ok, I'll have a look at it tomorrow Jan 24 17:42:20 bluelightning: I'm on it Jan 24 17:42:26 seems a problem with dbus Jan 24 17:42:27 ok Jan 24 17:45:44 gm Jan 24 17:45:51 ericben: whats broken ? Jan 24 17:46:05 khem: hi, qt4-tools-nativesdk-4.8.0 Jan 24 17:47:35 hmmm nativesdk Jan 24 18:08:23 pb_: ping Jan 24 18:08:54 pb_: who should initialize runtime_script_required in classes/rootfs_ipk.bbclass: if [ $runtime_script_required -eq 0 ]; then ? Jan 24 18:09:03 pb_: it's only place where this variable is used and results in run.do_rootfs.6328: line 235: [: -eq: unary operator expected Jan 24 18:12:11 hm, I thought RP_ had nuked that code Jan 24 18:12:34 was added in 87780fc09b066525e47d0f50ee5497db54d304cd Jan 24 18:12:58 it should be gone in 2feba313c991170747381c7cf821a45c2cd04632 Jan 24 18:14:15 if there's still some vestige of it then that would be a bug Jan 24 18:15:11 JaMa: hmm, ant reported that a few weeks ago, I meant to follow up on it... Jan 24 18:16:15 ah OK, I've just noticed it in log.rootfs when looking for some other issue Jan 24 19:13:15 hiZagain flo Jan 24 20:00:36 Sorry to keep asking beginner questions, but I have one about kernel modules. Jan 24 20:00:53 In my .config for the kernel, I have a few things set up to be modules. Jan 24 20:01:14 These were not included in the .ubi.bin image, but rather as their own .ipk files. Jan 24 20:01:36 But a few of the modules (ppp related) _were_ included in the image. Jan 24 20:01:53 What controls which modules go where? Jan 24 20:10:42 fullstop: I did the same today, and included it by adding IMAGE_INSTALL += kernel-module-musb-hdrc in my rootfs.bb Jan 24 20:30:43 tasslehoff: and I must do that for each module / ipkg that I want included? Jan 24 20:33:19 fullstop: not an expert, but I've had to do it for the two I need Jan 24 20:33:33 I don't appear to have a rootfs.bb :) Jan 24 20:35:58 fullstop: that's my own. started with console-image.bb and added stuff I need Jan 24 20:36:47 good to know Jan 24 20:36:54 I have my own one of those.. :) Jan 24 21:02:20 is there a list of valid MACHINE_FEATURES ? Jan 25 00:04:28 ath9k_htc from linux-firmware oe-core not working Jan 25 00:05:25 firmware download fails w error -22 **** ENDING LOGGING AT Wed Jan 25 02:59:56 2012