**** BEGIN LOGGING AT Fri Feb 27 02:59:58 2015 Feb 27 07:22:40 hi can anyone help me out here: I'm unable to view /proc on my arm system Feb 27 07:23:00 uboot message : mount: only root can mount proc on /proc Feb 27 07:23:27 i have enabled proc filesystem through the kernel configuration, but still it is not working Feb 27 09:32:01 morning all Feb 27 09:37:02 gm Feb 27 10:25:22 I'm trying to edit some sources that a recipe is downloading. In order to do this, instead of trying to edit the sources directly, i need to include a patch on the recipe? Feb 27 10:27:02 miandonmenmian: if you want to reproduce it later on, yes. Feb 27 10:27:16 otherwise you can just edit the cached tarball in your downloads directory Feb 27 10:27:19 at least currently. Feb 27 10:28:02 neverpanic: no, doubt i want to reproduce it Feb 27 10:28:26 neverpanic: the download directory is not sanity checked? Feb 27 10:28:43 if i edit, it wont rewrite it with current say, svn revision ? Feb 27 10:29:39 No, currently not. There's a bug open, and I'm working on a patch. Feb 27 10:29:57 The only sane way is using a patch Feb 27 10:30:06 everything else is a hack Feb 27 10:32:36 neverpanic: ok, in the downloads directory, the compressed files with the sources? i've seen other directories with the unpacked version Feb 27 10:32:49 but i think editing those gets the changes lost Feb 27 10:33:35 it depends on your type of source. For example, if it's a git repository, a folder is used. Feb 27 10:40:58 thanks much Feb 27 13:12:09 I want to build a yocto image with qt5 Feb 27 13:12:36 However, I end up in do_populate_root with something of a: "qtwebkit can not be found in the basefeed" Feb 27 13:12:43 and then a complete list of packages Feb 27 15:23:51 RP: fyi, it turns out that bbappends are sorted in layer priority order based on basename, but between basenames there's no defined order, because it goes into a dictionary. so if i have a systemd_%.bbappend and a systemd_216.bbappend, the order they're applied in isn't layer priority order, it's undefined Feb 27 15:24:10 RP: will look into a fix today most likely Feb 27 15:25:06 kergoth: good catch, we definitely want to fix that Feb 27 15:36:49 <_4urele_> Hi Feb 27 15:38:04 <_4urele_> does someone use a standalone prserver? Feb 27 15:38:49 <_4urele_> or a common database for each build? Feb 27 15:39:28 <_4urele_> I(build directory :) ) Feb 27 15:47:16 Hi, I am trying to get started with Yocto ADT but I am having trouble debugging the hello world example with qemu. Can anyone help? Feb 27 15:50:48 euan: I can try, can you outline the problem you're having? Feb 27 15:51:51 I have installed ADT as per the ADT manual and the eclipse plugin as per the dev manual. Build seems OK qemu is running and I have configured TCF but when I try to debug there is an error... Feb 27 15:52:37 TCF command exception: Command Filesystem open "/user/bin/test/, 26, null Exception no such file or directory Feb 27 15:54:07 error code 65538 Feb 27 15:55:48 I am using prebuilt image for qemux86 as first test Feb 27 15:57:21 todor: can you help euan? ^ Feb 27 15:59:31 euan: looks like the problem is the /user/bin/test/ path Feb 27 16:00:02 euan: shouldn't it be /usr/bin/test? Feb 27 16:00:54 yes it is /usr/bin/test , sorry i mistyped the error, I haven't setup shared clipboard with the VM Feb 27 16:01:58 should have been ... FileSystem open "/usr/bin/test", 26, null ... Feb 27 16:03:24 0 Feb 27 16:03:29 oops Feb 27 16:04:00 euan: looks like the binary hasn't made it to /usr/bin.. could you log in to the vm and double check? Feb 27 16:07:01 ok my path was incorrect, I used "user" instead of "usr" Feb 27 16:07:55 I changed my remote path to "/usr/bin/test-euan" and now it seems to run and I can see "test-euan" in qemu Feb 27 16:08:13 where do i see the "hello world" output? Feb 27 16:09:11 euan: it should be on the eclipse console Feb 27 16:10:11 the only console message after I resume from the breakpoint is: test_gdb_i586-poky....... Feb 27 16:12:04 euan: does it show in the console if you do a run instead of debug? Feb 27 16:15:06 euan: another thing to check is if the binary is executable when it reaches the target.. Feb 27 16:16:14 eclipse is stuck launching debug... will restart and try again Feb 27 16:16:37 is the "Commands to execute before application" field blank in the debug configuration? Feb 27 16:18:30 yes, it's blank Feb 27 16:19:23 starting qemu..... Feb 27 16:19:37 euan: chmod 755 /usr/bin/test-euan Feb 27 16:19:40 Hi Feb 27 16:20:31 euan: please enter this command in that filed Feb 27 16:20:41 in qemu terminal? Feb 27 16:21:13 in "Commands to execute before application" Feb 27 16:21:33 ok Feb 27 16:22:26 s/filed/field Feb 27 16:24:44 qemu now has a different ip address, updated TCF setup but now it says gdbserver: not found... Feb 27 16:26:03 ok running now... Feb 27 16:26:16 euan: cool Feb 27 16:26:37 stopped on breakpoint in main() but shows no source available Feb 27 16:27:48 when i hit resume the application terminates but i dont see the output in the console Feb 27 16:27:59 usually means your debugger doesn't have the right references to the -dbg package contents Feb 27 16:28:56 euan: i think this is now to do with the renaming you did earlier.. the binary compiled by eclipse is probably called still test and you transfer it on the target as test-euan Feb 27 16:29:15 euan: if you change that path to /usr/bin/test it will probably work Feb 27 16:29:54 same result Feb 27 16:30:28 hmmm Feb 27 16:31:23 success! Feb 27 16:31:35 I see the output if I click run rather than debug Feb 27 16:31:38 \0/ Feb 27 16:33:00 how do I debug to step through the source? Feb 27 16:34:40 still shows no source for main()? Feb 27 16:34:49 euan: i think that there is a mismatch somewhere between the binary name configure in your project and the binary name sent to the target Feb 27 16:35:14 i'd delete the debug configuration and create a new one Feb 27 16:35:51 brb Feb 27 16:52:33 deleted the debug config but when I create a new one it seems to be missing some starting config, any way to recover the previous template? Feb 27 17:02:06 this is frustrating! create a new hello world project but it won't build, "make: *** No rule to make target 'all'. Stop." Feb 27 17:06:09 Is it possible to rewrite a package depends ? like PACKAGECONFIG_depends_pn_NAME = "depends1, depends2, depends3"? Feb 27 17:06:31 that line doesn't make a great deal of sense. can you cl arify what you're trying to accomplish? Feb 27 17:10:15 kergoth, I got this message: "WARNING: Package busybox, set to be excluded, is in core-image-minimal-initramfs PACKAGE_INSTALL" Feb 27 17:10:40 what's pulling it in? Feb 27 17:11:41 kergoth, I just don't need busybox on what I am doing, and I specified on my local conf PACKAGE_EXCLUDE but then i got this warning that I'd like to get rid of Feb 27 17:11:43 also why are you removing it from an innitramfs image? Feb 27 17:12:03 sounds like you should fix your PAKCAGE_EXCLUDE to apply only to the image you want to build, so it doesn't apply to any initramfs images, only regular ones Feb 27 17:12:12 e.g. PACKAGE_EXCLUDE_pn-some-image = "busybox" Feb 27 17:12:25 nice Feb 27 17:12:26 but it'll still get pulled in via dependencies from packagegroups Feb 27 17:12:31 so you'll have to go remove it from those Feb 27 17:12:44 likely via RDEPENDS__remove = "busybox" in a bbappend Feb 27 17:13:00 removing busybox is non-trivial, you'll have a heck of a lot to add back in to fill in the gaps Feb 27 17:13:03 so be aware Feb 27 17:13:09 but i wouldn't like to change the core recipes, is there any other way around ? Feb 27 17:13:13 kergoth, i will ;) Feb 27 17:13:29 you were just talking about changing the metadta of the core recipes Feb 27 17:13:33 what do you think _pn-foo does? Feb 27 17:13:36 it overrides their variables Feb 27 17:13:47 whether you do so from local.conf or a bbappend really has no practical difference Feb 27 17:14:13 that makes sense, thanks! Feb 27 17:14:32 np Feb 27 17:15:11 long term using a layer of your own with appends is likely going to be easier to maintain than local.conf tweaks anyway, since you can source control it, and collect all the busybox-removal-related-bits in one place Feb 27 17:26:19 without some requirements i can only suggest local.conf Feb 27 17:29:13 Hey! I'm trying to compile software from a third party. This software makes use of the `create_proc_entry` function. However, this function got decrepated in Linux 3.10. Is there a way to support Poky on 3.9 or lower? Feb 27 17:32:33 shouldn't be an issue if the recipe/version is there Feb 27 17:32:55 beaglebone still uses 3.8 Feb 27 17:35:21 looks like lowest version in yocto-bsp is 3.10 Feb 27 17:35:36 on master Feb 27 17:37:10 Seems so, so I have to use an older branch? Feb 27 17:37:42 kergoth, indeed, in a long term an custom image will be better Feb 27 17:39:20 or a different bsp or make a recipe... Feb 27 17:40:11 maybe try checking out earlier commit hash on master with 3,8 Feb 27 17:40:51 ugh, rip leonard nimoy :( Feb 27 17:41:05 wha?!? Feb 27 17:41:24 yep, this morning Feb 27 17:41:24 sucks Feb 27 17:41:25 Okay, I'll give it a shot. Thanks for your help! Feb 27 17:41:28 euan: if you created a new project and got this make error.. i would check that your paths under Preferences -> Windows -> Yocto Project ADT are correct Feb 27 17:41:55 Guest52802: most of it is in the inc file Feb 27 17:42:30 you could try just making a linux-blah_3.8.bb Feb 27 20:07:02 I have a cyclades console server with a PPC 8xx CPU running 2.6.11 from MontaVista. How impossible would it be to modernize this? Feb 27 20:53:36 fignew: that's a big question Feb 27 20:54:32 afaik ppc embedded should have good support, but i gave away my 8402 or whatever a long time ago... Feb 27 20:55:10 either find or make a bsp layer and try it and see... Feb 27 20:55:10 Yea, big question indeed. Feb 27 20:55:59 least it isn't 2.4... Feb 27 20:59:11 fignew: looked at layer index? is there oe-classic support? where'd you get the current runtime/sdk? Feb 27 21:01:41 actually having a very hard time finding any source or SDK... Wish I had a time machine Feb 27 21:03:07 The Cyclades copies root off of a CF card to ramdisk. I can't figure out what FS the CF card has Feb 27 21:03:54 cpio? Feb 27 21:04:00 cpio.gz/bz2? thats pretty common Feb 27 21:07:15 cpio would make sense.. -_- will try that Feb 27 21:07:18 is it possible remove binaries that was generated or mark it to not be generated ? i.e: Mark or remove binary: "dbus-monitor" not to be installed with dbus Feb 27 21:10:58 no go on the cpio :/ Feb 27 21:11:45 kergoth, kergoth_ do you know ? ^ Feb 27 21:13:55 realBigfoot: If it's in it's own package you can just remove that package. If it isn't, then you can make a bbappend for the recipe with a do_install_append() that deletes the file so that it doesn't get packaged. It may even be possible to pass something to EXTRA_OECONF for dbus to tell it to not build dbus-monitor Feb 27 21:14:47 rewitt, I see... what if i want to install systemd without systemctl.... would it be the same? Feb 27 21:17:18 Yeah. The only thing you have to be careful of is of other packages that are runtime dependent on the package you are changing. So if you remove a file it will build, but there could possibly be a failure by something else at runtime. Feb 27 21:18:47 rewitt, got it ... thanks Feb 27 21:19:18 So if package A contained files foo and bar, and package B was dependent on package A because it uses foo, but then no longer put foo in A, something in B might break Feb 27 21:19:50 rewitt, indeed i have to be extra careful Feb 27 22:11:29 nerdboy: http://sprunge.us/AhbT That's the boot log :) Feb 27 22:15:08 * nerdboy thinks he must've missed the "long time ago..." part Feb 27 22:16:09 that's not what you said Feb 27 22:16:23 that's 2.4.17 monte vista Feb 27 22:17:25 * nerdboy used to be ppc gentoo guy too Feb 27 22:17:53 it's cool and all, but you're not exactly swimming in resources Feb 27 22:18:33 can you maybe splurge $35 and get rpi2? Feb 27 22:23:22 (it's MontaVista BTW) Feb 27 22:23:34 and 2.4.17, I'm trying to rember which product that was.. Feb 27 22:23:49 (I probably ahve the product CDs somewhere in a box at home) Feb 27 22:24:31 mvl 2.1! Feb 27 22:24:34 wow.. that is old.. Feb 27 22:24:42 2001-ish Feb 27 22:25:16 this is actually a different unit than the one I ssh'd into.. older kernel Feb 27 22:25:20 same hardwar Feb 27 22:26:21 well if it works, I take credit for the distro.. if it doesn't, I won't admit to anything.. :) Feb 27 22:26:31 (I did a lot of the toolchain and core distribution work at MV back in 2001) Feb 27 22:30:13 fray: do you have any resources for me? please! :D Feb 27 22:30:47 I'm thinking I can get http://www.denx.de/wiki/ELDK-5/WebHome to work on it Feb 27 22:38:18 not really this stuff is really old.. Feb 27 22:38:31 MVL 2.1 was a public release, so there is probably a mirror of it "somewhere" **** ENDING LOGGING AT Sat Feb 28 02:59:59 2015