Thanks, after having "fixed" /tmp, I managed to play test_devtool_modify_git successfully.
Le mar. 28 mars 2023, 22:04, Alexander Kanavin <[email protected]> a écrit : > Not devtool's fault. The selftest is written so that devtool is > instructed to use /tmp/... to extract sources, and not a workspace in > a build: 'devtool modify -x /tmp/...'. > > Having /tmp with noexec is not common, that's why it isn't mentioned. > If the admin is not cooperative, you can hack the selftest definition > to use some other place. > > Alex > > On Tue, 28 Mar 2023 at 21:02, Frédéric Martinsons > <[email protected]> wrote: > > > > Oh my, your assumption is surely correct, my build host /tmp is mounted > with noexec so no chance to execute > > anything there: > > > > [ (8) ] jenkins@cibuilder2:~$ mount | grep "/tmp" > > tmpfs on /tmp type tmpfs (rw,nosuid,nodev,noexec,noatime,inode64) > > > > Tomorrow, I'll try with another machine (or talk to the admin of the > build hosts) with a /tmp capable of executing things. > > > > But the requirement to have /tmp directory capable of executing scripts > was not mentioned in the oe-selftest > > page (https://wiki.yoctoproject.org/wiki/Oe-selftest) or the quick > start one (https://docs.yoctoproject.org/brief-yoctoprojectqs/index.html) > > > > I would have thought that devtool, if it needs to execute some scripts, > will execute them in its working directory and not > > in /tmp of build hosts.... Is this something really expected ? > > > > > > On Tue, 28 Mar 2023 at 20:28, Alexander Kanavin <[email protected]> > wrote: > >> > >> Btw, I'm beginning to suspect something isn't right about /tmp on your > >> build host. The test requires being able to write executable scripts > >> into that, and then run them. I just ran the > >> devtool.DevtoolModifyTests.test_devtool_modify_git here, and it > >> succeeded. > >> > >> Alex > >> > >> On Tue, 28 Mar 2023 at 20:14, Alexander Kanavin via > >> lists.openembedded.org <[email protected]> > >> wrote: > >> > > >> > Would you be able to read the failing test, and replicate the steps it > >> > does by hand? If we can narrow down the precise point where it fails, > >> > that would already help. > >> > > >> > Alex > >> > > >> > On Tue, 28 Mar 2023 at 20:01, Frédéric Martinsons > >> > <[email protected]> wrote: > >> > > > >> > > I rebuilt a poky master from scratch on another machine (Ubuntu > 22.04 , 16 Xeon, 124 Gb RAM), it took > >> > > 2h30mn to finish (note that rust-llvm-native and llvm-native both > took around 1h each to be built). > >> > > > >> > > I have two questions for my curiosity: > >> > > - why rust things (cargo-native, librsvg) are the last > artifacts built on core-image-sato ? it seems strange at first look > >> > > (but may be explained by packages embedded by this image, I didn't > look on what are the dependencies of this image) > >> > > - my pristine BDIR took around 90 GB , maybe the page > https://docs.yoctoproject.org/dev/brief-yoctoprojectqs/index.html stating > >> > > that we need 50 GB of free space needs to be reviewed (unless it > was written with another image than core-image-sato in mind) ? > >> > > > >> > > Apart from that, I have the exact same error on > test_devtool_modify_git that I had on the other machine under /var build > dir: > >> > > > >> > > NOTE: Running setscene task 210 of 213 > (virtual:native:/tmp/devtoolqacvbsfqx2/core-copy/meta/recipes-devtools/gnu-config/gnu-config_git.bb: > do_populate_sysroot_setscene) > >> > > NOTE: recipe gnu-config-native-20230216+gitAUTOINC+63acb96f92-r0: > task do_populate_sysroot_setscene: Started > >> > > NOTE: recipe gnu-config-native-20230216+gitAUTOINC+63acb96f92-r0: > task do_populate_sysroot_setscene: Succeeded > >> > > NOTE: Running task 896 of 918 > (/tmp/devtoolqacvbsfqx2/core-copy/meta/recipes-core/psplash/psplash_git.bb: > do_prepare_recipe_sysroot) > >> > > NOTE: recipe psplash-0.1+git999-r0: task do_prepare_recipe_sysroot: > Started > >> > > NOTE: recipe psplash-0.1+git999-r0: task do_prepare_recipe_sysroot: > Succeeded > >> > > NOTE: Running task 910 of 918 > (/tmp/devtoolqacvbsfqx2/core-copy/meta/recipes-core/psplash/psplash_git.bb: > do_configure) > >> > > NOTE: recipe psplash-0.1+git999-r0: task do_configure: Started > >> > > ERROR: psplash-0.1+git999-r0 do_configure: no configure script > found at ../../../../../../../../../../../tmp/devtoolqamf2smwsy/configure > >> > > ERROR: psplash-0.1+git999-r0 do_configure: > ExecutionError('/home/jenkins/yocto-poky-master/poky/build-st/tmp/work/core2-64-poky-linux/psplash/0.1+git999-r0/temp/run.do_configure.1967375', > 1, None, None) > >> > > ERROR: Logfile of failure stored in: > /home/jenkins/yocto-poky-master/poky/build-st/tmp/work/core2-64-poky-linux/psplash/0.1+git999-r0/temp/log.do_configure.1967375 > >> > > NOTE: recipe psplash-0.1+git999-r0: task do_configure: Failed > >> > > ERROR: Task > (/tmp/devtoolqacvbsfqx2/core-copy/meta/recipes-core/psplash/psplash_git.bb:do_configure) > failed with exit code '1' > >> > > NOTE: Tasks Summary: Attempted 910 tasks of which 904 didn't need > to be rerun and 1 failed. > >> > > > >> > > Full test log joined (I would like to join also the log of > do_configure step but it seems that the test do a cleanall of the recipe > during its tear down method) > >> > > > >> > > I'm kind of stuck here since I cannot develop more oe-selftest > tests if I didn't even manage to make the existing ones succeed. > >> > > Can someone give me the setup of this autobuilder where Alexander > talked about ? > >> > > > >> > > I could try to get as close as possible to the autobuilder setup if > the oe-selftest is ok there. > >> > > > >> > > On Tue, 28 Mar 2023 at 17:36, Frédéric Martinsons < > [email protected]> wrote: > >> > >> > >> > >> Note that I managed to build libtirpc-native and shadow-native > with bitbake (and the same setup). > >> > >> > >> > >> I currently try the build on another machine and under /home. > >> > >> > >> > >> I'll tell you the results soon > >> > >> > >> > >> Le mar. 28 mars 2023, 15:13, Richard Purdie < > [email protected]> a écrit : > >> > >>> > >> > >>> On Tue, 2023-03-28 at 14:40 +0200, Frédéric Martinsons wrote: > >> > >>> > I just finished a pristine build of core-image-sato on poky > master, > >> > >>> > and set up SSTATE_DIR in my local.conf. > >> > >>> > Tests are now ok, I can start to write some now. > >> > >>> > > >> > >>> > If anyone is interested , please find attached the full log of > the > >> > >>> > two first failures. > >> > >>> > >> > >>> Having the errors helps, thanks. > >> > >>> > >> > >>> The short summary is these two lines: > >> > >>> > >> > >>> libtirpc-native > >> > >>> | chown: changing ownership of > '/var/lib/jenkins/YOCTO_POKY_MASTER_TEST/poky/build-st/tmp/work/x86_64-linux/libtirpc-native/1.3.3-r0/image/var/lib/jenkins/YOCTO_POKY_MASTER_TEST/poky/build-st/tmp/work/x86_64-linux/libtirpc-native/1.3.3-r0/recipe-sysroot-native/etc/netconfig': > Invalid argument > >> > >>> > >> > >>> shadow-native > >> > >>> | chown: changing ownership of > '/var/lib/jenkins/YOCTO_POKY_MASTER_TEST/poky/build-st/tmp/work/x86_64-linux/shadow-native/4.13-r0/image/var/lib/jenkins/YOCTO_POKY_MASTER_TEST/poky/build-st/tmp/work/x86_64-linux/shadow-native/4.13-r0/recipe-sysroot-native/var/spool/mail': > Invalid argument > >> > >>> > >> > >>> > >> > >>> I'm wondering if the fact the build is under /var might be > triggering > >> > >>> something somehow... > >> > >>> > >> > >>> Cheers, > >> > >>> > >> > >>> Richard > >> > > >> > > >> > >
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#179263): https://lists.openembedded.org/g/openembedded-core/message/179263 Mute This Topic: https://lists.openembedded.org/mt/97900549/21656 Group Owner: [email protected] Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]] -=-=-=-=-=-=-=-=-=-=-=-
