**** BEGIN LOGGING AT Thu Mar 14 02:59:58 2013 Mar 14 07:15:32 hi Mar 14 07:15:59 how to use if condition in .bb files Mar 14 07:16:01 ?? Mar 14 09:45:14 good morning Mar 14 11:10:45 in denzil at least, changing included files does not result in .bb recompilations. Is this fixed in newer yocto releases? Maybe I could cherry-pick the change.. Mar 14 11:46:39 Someone has any clue why this fails to build? http://autobuilder.yoctoproject.org:8010/builders/nightly-fsl-arm/builds/116/steps/shell_71/logs/stdio Mar 14 11:46:48 (avahi) Mar 14 11:52:12 mcfrisk: yes it does Mar 14 11:52:37 mcfrisk: but you can only really use it in upcoming 1.4 (master branch actually) Mar 14 11:52:53 mcfrisk: danny has initial work for it but it is not complete Mar 14 12:17:28 otavio: ok, thanks. Mar 14 13:47:57 hmm, I have image build failures with meta-ti where x-loader binary symlinks are not created by do_deploy from x-loader.inc. When should the do_deploy run when it has "inherit deploy" and "addtask deploy before do_package after do_install"? Mar 14 14:07:05 otavio: looks like the linker couldn't find the correct pthread libraries when building avahi (fails on do_compile: /srv/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-fsl-arm/build/build/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/avahi/0.6.31-r6.1/temp/log.do_compile.20324) Mar 14 14:07:55 otavio: looks like the sam problem occurred when building rpm: /srv/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-fsl-arm/build/build/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/rpm/5.4.9-r61/temp/log.do_compile.1353 Mar 14 14:25:03 looks like this build succeeded: http://autobuilder.yoctoproject.org:8010/builders/nightly-fsl-arm/builds/116/steps/shell_109/logs/stdio might be a good place to start looking for problems Mar 14 14:25:29 hi all! i've made a receipe where in do_install() I install a file with: "install -m 0755 swing-app.jar ${D}/${bindir}" and then change ownership with "chown -vR client:client ${D}${bindir}/swing-app.jar " Mar 14 14:25:35 the log.do_install stats that swing-app.jar changed from root:root to client:client, but in the rootfs is still root:root ownership...I'm doing wrong something? Mar 14 14:26:25 waynr: the problem this is not easy to reproduce Mar 14 14:26:40 waynr: for example I never been able to reproduce it in my build system Mar 14 14:27:35 soldoKyn: in the do_install task you are not yet actually in the rootfs that you are creating so running "chown" will not give you the result you are looking for...you might want to try a post install task which should happen on first boot if i remember correctly Mar 14 14:28:19 otavio: yeah i am running a similar build for one of my devices that does not fail. are you saying that if you were to re-run this build manually using the same configuration files and for the same actual image it might not fail? Mar 14 14:28:37 waynr: yes Mar 14 14:29:42 otavio: that sounds like it could be a problem where the avahi package does not explicitly depend on eglibc somehow but i'm not sure...if this were true, then sometimes eglibc would finish building before avahi, sometimes not Mar 14 14:32:36 waynr: ok, I'll search for some documentation about post install features...thank you! Mar 14 14:37:29 waynr: the problem is that we do not depends on libc in all packages, and all them work fine Mar 14 14:37:41 waynr: so something weird is happening there Mar 14 14:37:47 waynr: but I have no clue what Mar 14 14:38:05 otavio: maybe linker script is overwriting CLFAGS? Mar 14 14:38:24 mcfrisk: maybe; but when it happens sometimes? Mar 14 14:39:48 I've seen some bad inhouse code where only some CMakeLists.txt files were overwriting default CFLAGS, and this worked as long as default compiled options were ok. But not when soft/hard float changes came in. Mar 14 15:18:13 how to set a deploy dependency? image creation needs x-load binaries under tmp/deploy/images/ but they are not there after build. Mar 14 17:31:27 how can "inherit autotools" result in build failures, systemd configure script failing to find working . Are there some autotools dependencies to the build host? Mar 14 18:06:58 halstead, ping Mar 14 18:43:57 How I can use bitbake -S to dump the signatures of a package? Mar 14 18:44:22 run it. then look in the stamps directory Mar 14 18:45:09 otavio: maybe you'll find this useful too openembedded-core/scripts/sstate-diff-machines.sh Mar 14 18:45:32 I use it to create "backups" if I'm later wondering what invalidated a lot of checksums Mar 14 18:50:06 JaMa: I am checking why avahi is maybe rebuild Mar 14 18:50:10 but I fail to see why Mar 14 18:50:33 JaMa: sstate-diff-machines.sh --tmpdir=tmp --machines="imx53qsb imx6qsabresd" --targets=avahi Mar 14 18:50:44 JaMa: no output in the end of the processing Mar 14 18:51:21 otavio: see tmp/sstate-diff/*timestamp* Mar 14 18:51:41 JaMa: so this should print it in the end ;-) Mar 14 18:51:51 patches welcome :) Mar 14 19:37:06 something does not like me today, a lot of do_package tasks fails with NameError: global name 'pkgfiles' is not defined, in debian.bbclass and I don't remember updating something which should cause this, anyone else seing those too? Mar 14 21:53:53 so i want to fill out this page: https://wiki.yoctoproject.org/wiki/FAQ:How_do_I_set_or_change_the_root_password Mar 14 21:54:27 I have an idea of how to start but first i need to know what hashing algorithm passwd uses by default Mar 14 21:54:46 I think it might be DES, at least on the image i have built so far Mar 14 21:57:04 I see in base-passwd-3.5.26 that there is no root password initially Mar 14 21:58:27 and then if debug-tweaks or whatever is not installed it gets replaced with "*" which leaves the root user inaccessible on first boot Mar 14 21:59:52 i suppose i could just use a post install function for base-passwd that replaces /etc/passwd in the destination dir with a custom one that i create Mar 14 22:00:21 use a post install function that sed's the passwd file (or shadow) and changes the root PW.. that's what I've done Mar 14 22:01:09 does the hash function used to do this matter? Mar 14 22:01:38 it can.. but what I've done in the past is just cleared it.. booted the taget.. set the pw to something I wanted.. and then captured the result for the sed.. Mar 14 22:01:45 a bit of a roundabout way of doing it, but it worked Mar 14 22:02:58 yeah that is what i will do but it seems like it might be better in the long run and for documentation purposes to create a function that does this Mar 14 22:03:28 at one point we had discussed it (on the oe-core list) but I'm not sure if it was ever implemented.. Mar 14 22:03:44 I know adduser/addgroup can be done w/o a sed.. but I don't know about the password change Mar 14 22:05:09 i am thinking of a variable, say ROOT_PASSWD, that if present during base-passwd post install will be hashed and sed'd into /etc/passwd Mar 14 22:05:56 should be fairly easy to implement..... but you should check if the 'passwd' program supports the --root argument.. if it does.. it is -really- easy to implement Mar 14 22:06:09 otherwise, updating passwd is needed so both shadow and non-shadow are easily supported Mar 14 22:06:45 yeah i am logged into the filesystem now and I don't see the "--root" option Mar 14 22:08:08 not in the filesystem.. it's in the shadow-native Mar 14 22:08:55 Mar 14 22:08:55 Add a --root command option to the following utilties: Mar 14 22:08:56 * useradd Mar 14 22:08:56 * groupadd Mar 14 22:08:56 * usermod Mar 14 22:08:56 * groupmod Mar 14 22:08:56 * userdel Mar 14 22:08:57 * groupdel Mar 14 22:08:57 * passwd Mar 14 22:08:58 * gpasswd Mar 14 22:08:58 * pwconv Mar 14 22:08:59 * pwunconv Mar 14 22:08:59 * grpconv Mar 14 22:09:00 * grpunconv Mar 14 22:09:01 Mar 14 22:09:08 so yes.. passwd and gpasswd are already patched.. Mar 14 22:10:20 question is what is the right argument set to set the passwd.. Mar 14 22:10:31 looks like: Mar 14 22:10:44 echo | passwd --root $D root Mar 14 22:10:48 as a post-install will work Mar 14 22:11:08 cool Mar 14 22:11:18 I knew we discussed this, just wasn't sure how to do it.. Mar 14 22:11:23 oops.. the args are wrong.. try: Mar 14 22:11:31 echo | password --root $d --stdin root Mar 14 22:11:32 ah, that's way better than patching passwd/shadow. that'd be nice to have a var that enables that easily Mar 14 22:11:35 nifty Mar 14 22:11:38 (make that -$D) Mar 14 22:12:04 end of the day and apparently I can't type Mar 14 22:12:11 echo | password --root $D --stdin root Mar 14 22:12:19 there that should work as a post-image scirpt Mar 14 22:12:19 must be the carpal tunnels Mar 14 22:13:37 (you can also set group passwords and such in the same way, I think) Mar 14 22:30:38 is there a way to specify a function in base-passwd that runs after the full image sysroot has been populated? Mar 14 22:33:47 fray: i tried running the command you specified from both the do_install which is wrong because $D here does not have etc/passwd but usr/share/base-passwd/passwd.master and base_passwd_sstate_postinst which is wrong because passwd attempts to chroot into the directory specified by '--root' Mar 14 22:34:46 when you said "post-image" script is that what you were referring to? I suppose this does not have to be done from the base-passwd recipe, it could be done from an image Mar 14 22:41:16 sorry.. ya, under do_install it's $IMAGE_ROOTFS that always trips me up Mar 14 22:41:55 it's "IMAGE_POSTPROCESS_COMMAND" Mar 14 22:41:56 do_install for base-passwd? Mar 14 22:42:01 look at meta/classes/image.bbclass Mar 14 22:42:10 okay Mar 14 22:42:41 you should be able to do something like: IMAGE_POSTPROCESS_COMMAND_append = " echo password | password -root ${IMAGE_ROOTFS} --stdin root ;" Mar 14 22:42:54 (may have to be prepend..) Mar 14 22:43:08 BTW that goes in your local.conf file -- or you custom image recipe Mar 14 22:43:40 and not do_install, I -meant- do_rootfs Mar 14 22:43:45 man, I am just not awake today Mar 14 22:44:18 fray: WAKE UP! Mar 14 22:45:25 not working.. :P Mar 14 22:45:44 been up since 4:30a here.. Mar 14 22:46:11 not here.. drove the kid to school this morning at 7.. went back to bed at 7:30 and woke up late for my 10am meeting.. Mar 14 22:46:11 cool, i think I understand :) maybe Mar 14 22:46:14 been in a fog ever since Mar 14 22:47:03 shouldn't that be a rootfs postprocess, not an image postprocess? Mar 14 22:47:06 * kergoth yawns Mar 14 22:47:13 Hmm.. you may be right.. Mar 14 22:47:27 it needs to run against the rootfs before anything else has mangled it.. Mar 14 22:47:36 (but after the package post install scripting has run) Mar 14 22:47:40 soo...ROOTFS_POSTPROCESS_COMMAND? Mar 14 22:48:09 ya.. ROOTFS_POSTPROCESS_COMMAND looks like it'll work Mar 14 22:48:31 ya.. he's right.. IMAGE_POSTPROCESS_COMMAND is too late.. Mar 14 22:52:50 I'm pretty sure today is really a Monday in disguise. Mar 14 22:55:42 kergoth: every day since i've started working.... :) Mar 14 22:57:13 is each one worse thaan the one before it? Mar 14 22:57:24 oh yea Mar 14 22:57:43 so today is the worst day of your life? Mar 14 23:10:18 :\ no --stdin option for passwd, the build just hands Mar 14 23:10:23 hangs* Mar 14 23:10:42 doh Mar 14 23:11:29 back to the drawing board Mar 14 23:13:28 Crofton|work, pm? Mar 14 23:13:36 sure Mar 14 23:14:05 okay so hashing the contents of $ROOT_PASSWD then conditionally sed the $D/usr/share/base-passwd/passwd.master seems likely...only problem is, will need to use some hashing function available in the sysroot Mar 14 23:27:00 waynr: i'd think it'd be better to just patch the native passwd to allow stdin password input :) Mar 14 23:38:23 kergoth: i just came to that conclusion myself x.x gonna work on it now Mar 15 00:34:21 okay, compiles just need to verify works Mar 15 00:34:48 i went with adding an optional parameter "--password" since that seemed simpler than --stdin Mar 15 00:35:25 arg could be slightly more dangerous, as it could be seen in a ps if it was run at the right time, generally speaking Mar 15 00:35:36 nonissue in this case, but potentially a concern if you want to usptream Mar 15 00:39:40 okay Mar 15 00:44:23 this doesn't seem like the kind of functionality that upstream would want Mar 15 00:47:30 (or maybe i am lazy) Mar 15 01:30:55 ah yeah...it works Mar 15 01:31:13 * waynr eats a celebratory spoonful of peanut butter Mar 15 01:31:57 waynr: good choice **** ENDING LOGGING AT Fri Mar 15 02:59:58 2013