Re: [yocto] Empty source package when using devtool
Hello, I dig up this thread to mention that the problem is still in kirkstone but was fixed in langdale and greater. I have done some dissection and found that the 3 following commits have fixed it: - https://git.openembedded.org/openembedded-core/commit/?h=langdale=cbd6144a9769d21371ae0fe04db2adc05f6eed02 - https://git.openembedded.org/openembedded-core/commit/?h=langdale=c39b5020b8705d17e3745c41e38d0f99a1ac94cf - https://git.openembedded.org/openembedded-core/commit/?h=langdale=591cbe7140f01470b5799bc51e5b23f58bf95ffa WIth these 3 commits backported in kirkstone, I managed to having the source packages correctly populated with or without devtool activated on the components. -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#61876): https://lists.yoctoproject.org/g/yocto/message/61876 Mute This Topic: https://lists.yoctoproject.org/mt/83622035/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] Ownership issue during do_image
Hello, some news about this issue. Tim pointed me to the docker and the reason why crops container have some UID:GID manipulation: [see there](https://github.com/crops/extsdk-container/blob/master/usersetup.py). I used an ubuntu 20.04 VM to setup a build without docker and I fall into the exact same problem. But finally, I found that a workaround by creating and using another working directory inside my image creation command (though I clearly don't know why this work and the former way don't, despite it worked with yocto warrior) Before that, I have a working directory that was created in the bbclass : ``` SQSHDIR = "${S}/sqshimg/${MACHINE}" do_image_tapos_sqshimg[dirs] += "${SQSHDIR}" do_image_tapos_sqshimg[cleandirs] += "${SQSHDIR}" do_image_tapos_sqshimg[recrdeptask] += "do_deploy" IMAGE_CMD_tapos-sqshimg() { install -d ${SQSHDIR} ``` I simply replaced the creation command by: ``` IMAGE_CMD_tapos-sqshimg() { SQSHDIR_WORKDIR=$(mktemp -d) ``` and voilà ! -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#58810): https://lists.yoctoproject.org/g/yocto/message/58810 Mute This Topic: https://lists.yoctoproject.org/mt/94706311/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] #dunfell
A ticket has been open for backporting the fix: https://bugzilla.yoctoproject.org/show_bug.cgi?id=14971 -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#58644): https://lists.yoctoproject.org/g/yocto/message/58644 Mute This Topic: https://lists.yoctoproject.org/mt/94774777/21656 Mute #dunfell:https://lists.yoctoproject.org/g/yocto/mutehashtag/dunfell Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] Ownership issue during do_image
Yes I do, I build under docker (ubuntu 20.04), I didn't mention that in my original post , sorry. Do you think it is related to my issue ? On Tue, Nov 15, 2022 at 09:06 PM, Tim Orling wrote: > > Are you building with a docker/podman container? > > On Tue, Nov 15, 2022 at 9:24 AM Frederic Martinsons < > frederic.martins...@gmail.com> wrote: > > > Hello list, > > > > Doesn't ring a bell to anybody ? Even the slightest clue will help me to > > continue to track down this weird behavior , nobody have ever problem of > > files/directories rights during image generation ? > > > > > > > > > -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#58575): https://lists.yoctoproject.org/g/yocto/message/58575 Mute This Topic: https://lists.yoctoproject.org/mt/94706311/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] Ownership issue during do_image
Hello list, Doesn't ring a bell to anybody ? Even the slightest clue will help me to continue to track down this weird behavior , nobody have ever problem of files/directories rights during image generation ? -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#58572): https://lists.yoctoproject.org/g/yocto/message/58572 Mute This Topic: https://lists.yoctoproject.org/mt/94706311/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] #dunfell
On Thu, Nov 3, 2022 at 11:38 PM, Zoran wrote: > > >>> - for imageType in ${KERNEL_IMAGETYPES} ; do > >>> + for imageType in ${KERNEL_IMAGETYPE_FOR_MAKE} ; do > > Interesting... What is the difference between variables > ${KERNEL_IMAGETYPES} and ${KERNEL_IMAGETYPE_FOR_MAKE} ? > In my setup I have: * KERNEL_IMAGETYPES=fitImage (coming from KERNEL_IMAGETYPE of my machine configuration) * KERNEL_IMAGETYPE_FOR_MAKE=Image (used only internally by kernel.bbclass and setting to Image because of [this line](https://github.com/openembedded/openembedded-core/blob/dunfell/meta/classes/kernel-fitimage.bbclass#L29) -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#58486): https://lists.yoctoproject.org/g/yocto/message/58486 Mute This Topic: https://lists.yoctoproject.org/mt/94774777/21656 Mute #dunfell:https://lists.yoctoproject.org/g/yocto/mutehashtag/dunfell Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
[yocto] #dunfell
Hello, I'm currently migrating our system from warrior to dunfell and I have an issue on fitImage support for aarch64 architecture. We use KERNEL_IMAGETYPE = "fitImage" and INITRAMFS_IMAGE_BUNDLE = "1" in our configuration and the kernel do_deploy steps failed on not finding the initramfs file: ``` | lib/modules/4.19.255-rt113-sigfox/kernel/drivers/usb/serial/usb_wwan.ko | lib/modules/4.19.255-rt113-sigfox/kernel/drivers/usb/serial/option.ko | lib/modules/4.19.255-rt113-sigfox/kernel/drivers/usb/class/ | lib/modules/4.19.255-rt113-sigfox/kernel/drivers/usb/class/cdc-wdm.ko | lib/modules/4.19.255-rt113-sigfox/modules.order | lib/modules/4.19.255-rt113-sigfox/modules.builtin | install: cannot stat 'arch/arm64/boot/Image.initramfs': No such file or directory | WARNING: exit code 1 from a shell command. | ERROR: Execution of '/home/fmartinsons/TAPOS_build_for_dunfell/build-tapos/tmp/work/a3700-tapos-linux/linux-sbs/4.19.255+gitAUTOINC+5c7ccbe1aa-r4.17.1.1/temp/run.do_deploy.1460182' failed with exit code 1 ``` Doing some more research, I found that the problem came from this change https://git.openembedded.org/openembedded-core/commit/?id=526bdd88ccd758204452579333ba188e29270bde , I found a commit introduced in kirkstone that fix my issue (at least the part which revert the previous commit): https://git.openembedded.org/openembedded-core/commit/?id=10a4a132e87e835726bf5da81a60f6f509b90765 Can somebody know why the commit was not back-ported in dunfell ? Regards. -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#58476): https://lists.yoctoproject.org/g/yocto/message/58476 Mute This Topic: https://lists.yoctoproject.org/mt/94774777/21656 Mute #dunfell:https://lists.yoctoproject.org/g/yocto/mutehashtag/dunfell Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
[yocto] Ownership issue during do_image
Hello list, I currently migrate our project from warrior to dunfell and we experience some issues during the image generation step. We have a custom IMAGE_FSTYPES to meet our need and in one of our custom steps is to copy the rootfs (populated during yocto do_rootfs step) into a custom dir and act on it (reorganizing tree and making a squash file system). I'm experiencing an issue on file ownership preservation that I didn't have in warrior. We use rsync to copy the rootfs tree to another location (via multiple option, including the -a one which preserve ownership) and during that step, I see a lot of: ``` rsync: chown "" failed: Operation not permitted (1) ``` The files and directories are nevertheless copied (but of course. To investigate further I do a devshell and reproduce the issue via a simple 'cp -a' (which also preserve ownership of files) and see the same king of ownership preservation problems: ``` root@SGX-CMP-776:${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/tapos-image-1.0# ll ${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/rootfs/bin/arping -rwxr-xr-x 1 root root 22680 Mar 9 2018 ${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/rootfs/bin/arping* root@SGX-CMP-776 :${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/tapos-image-1.0# root@SGX-CMP-776:${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/tapos-image-1.0# cp -a ${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/rootfs/bin/arping . cp: failed to preserve ownership for './arping': Operation not permitted root@SGX-CMP-776:${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/tapos-image-1.0# ll arping -rwx-- 1 1000 1000 22680 Mar 9 2018 arping* root@SGX-CMP-776:${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/tapos-image-1.0# id uid=0(root) gid=0(root) groups=0(root) ``` During devshell'ed session, the rootfs populated by do_rootfs is seen as root user (while the real user is me with uid 1000) , I guess this is due to the fact that the images step is run under pseudo and fakes user/group. What I can't understand is why pseudo (or something else) doesn't do the same with my file copy ? ``` root@SGX-CMP-776:${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/tapos-image-1.0# ll ${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/rootfs/ total 72 drwxr-xr-x 17 root root 4096 Nov 1 05:38 ./ drwxr-xr-x 12 1000 1000 4096 Nov 1 05:38 ../ drwxr-xr-x 2 root root 4096 Mar 9 2018 bin/ drwxr-xr-x 2 root root 4096 Mar 9 2018 boot/ drwxr-xr-x 2 root root 4096 Mar 9 2018 dev/ drwxr-xr-x 38 root root 4096 Nov 1 05:38 etc/ lrwxrwxrwx 1 root root8 Nov 1 05:38 home -> /rw/home drwxr-xr-x 8 root root 4096 Mar 9 2018 lib/ lrwxrwxrwx 1 root root 19 Mar 9 2018 linuxrc -> /bin/busybox.nosuid drwxr-xr-x 2 root root 4096 Mar 9 2018 media/ drwxr-xr-x 2 root root 4096 Mar 9 2018 mnt/ dr-xr-xr-x 2 root root 4096 Mar 9 2018 proc/ drwxr-xr-x 2 root root 4096 Mar 9 2018 run/ drwxr-xr-x 5 root root 4096 Nov 1 05:38 .rw_skeleton/ drwxr-xr-x 2 root root 4096 Mar 9 2018 sbin/ dr-xr-xr-x 2 root root 4096 Mar 9 2018 sys/ -rw-r--r-- 1 root root 149 Mar 9 2018 .tapos.manifest drwxrwxrwt 2 root root 4096 Mar 9 2018 tmp/ drwxr-xr-x 10 root root 4096 Mar 9 2018 usr/ drwxr-xr-x 7 root root 4096 Nov 1 05:38 var/ fmartinsons@SGX-CMP-776:~$ ll ${TMPDIR}/work/aimb231-tapos-linux/tapos-image/1.0-r0/rootfs/ total 72 drwxr-xr-x 17 fmartinsons fmartinsons 4096 nov. 1 06:38 ./ drwxr-xr-x 12 fmartinsons fmartinsons 4096 nov. 1 06:38 ../ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 bin/ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 boot/ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 dev/ drwxr-xr-x 38 fmartinsons fmartinsons 4096 nov. 1 06:38 etc/ lrwxrwxrwx 1 fmartinsons fmartinsons8 nov. 1 06:38 home -> /rw/home drwxr-xr-x 8 fmartinsons fmartinsons 4096 mars 9 2018 lib/ lrwxrwxrwx 1 fmartinsons fmartinsons 19 mars 9 2018 linuxrc -> /bin/busybox.nosuid drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 media/ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 mnt/ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 proc/ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 run/ drwxr-xr-x 5 fmartinsons fmartinsons 4096 nov. 1 06:38 .rw_skeleton/ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 sbin/ drwxr-xr-x 2 fmartinsons fmartinsons 4096 mars 9 2018 sys/ -rw-r--r-- 1 fmartinsons fmartinsons 149 mars 9 2018 .tapos.manifest drwxr-xr-t 2 fmartinsons fmartinsons 4096 mars 9 2018 tmp/ drwxr-xr-x 10 fmartinsons fmartinsons 4096 mars 9 2018 usr/ drwxr-xr-x 7 fmartinsons fmartinsons 4096 nov. 1 06:38 var/ fmartinsons@SGX-CMP-776:~$ id uid=1000(fmartinsons) gid=1000(fmartinsons) groups=1000(fmartinsons) ``` Am I missing something about pseudo usage ? And if so, why did these copy operations work perfectly within warrior branch ? Has anyone already experienced this
Re: [yocto] Would COMPATIBLE_IMAGE make sense?
Hello, instead of creating a new variable (don't sure of its possible usefulness), you can use BBMASK or review dependencies between your layer PS: I did a similar answer on your stackoverflow question. On Tue, 29 Jun 2021 at 01:29, Jonas Vautherin wrote: > I was thinking about my issue described here [1], and discovered the > variables called COMPATIBLE_MACHINE and COMPATIBLE_HOST, which "you can use > to stop recipes from being built for machines (/hosts) with which the > recipes are not compatible". And so I wondered if it would make sense to > add COMPATIBLE_IMAGE, for a similar purpose. > > Let me explain my issue: I define different images in different layers > (say `first-project-image` and `second-project-image`), and in each of > those layers I create `.bbappends` to configure some packages. Typically > `hostapd` is used by both images, but with a different config file. > > The thing is that when I run `bitbake first-project-image`, because > `second-project-image` is part of my bblayers.conf, then the > hostapd_%.bbappend from `second-project-image` is used and may have an > impact on `first-project-image`, which I don't want. I really want this > bbappend to only affect the image `second-project-image`. > > One way I can see to deal with that is to realize that > `first-project-image` and `second-project-image` are two different > projects, and build them from two different BUILDDIRs. The thing I don't > like here is that all the packages are therefore downloaded and built > twice, which seems like a loss of time and space. > > That's where I thought about COMPATIBLE_IMAGE. In the hostapd_%.bbappend > of `first-project-image`, I would set "COMPATIBLE_IMAGE = > 'first-project-image'", and similarly for "COMPATIBLE_IMAGE = > 'second-project-image'". So that I could still share a BUILDDIR between > different projects. > > How bad of an idea is that? > > Thanks in advance, > Jonas > > [1]: https://stackoverflow.com/questions/68167244/image-specific-layers > > > > -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#54001): https://lists.yoctoproject.org/g/yocto/message/54001 Mute This Topic: https://lists.yoctoproject.org/mt/83858212/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] Empty source package when using devtool
Hello, I dug further into yocto classes and I think I found what is going on. All seems to come from the fact that gcc debug-prefix-map and macro-prefix-map options are statically defined in bitbake.conf to > > DEBUG_PREFIX_MAP ?= > "-fmacro-prefix-map=${WORKDIR}=/usr/src/debug/${PN}/${EXTENDPE}${PV}-${PR} > \ > -fdebug-prefix-map=${WORKDIR}=/usr/src/debug/${PN}/${EXTENDPE}${PV}-${PR} > \ > -fdebug-prefix-map=${STAGING_DIR_HOST}= \ > -fdebug-prefix-map=${STAGING_DIR_NATIVE}= \ > " > > > I'll took an example to try to be clearer. Let's say that I devtool modify the glib-2.0 package. Without any configuration, sources will be extracted in ${BDIR}/workspace/sources/glib-2.0. Build directory is set by externalsrc.bbclass to ${BDIR}/tmp/work/${arch}-${distro}/glib-2.0/${PN}/${EXTENDPE}${PV}-${PR}/${BPN}/${BPN}-${PV} (the real path doesn't matter, what is matter is that B is not under devtool spaces). This lead to have pretty long relative path in gcc compilation line (sotheming like -c ../../../../../../../../worspace/sources/glib-2.0/glib/gmain.c for example). Hence the debug-prefix-map could not be appplied. After the compilation stage, package.bbclass (via splitdebuginfo), there is an extraction of sources path via dwarf format parsing. We ended up parsing something like - /workspace/sources/glib-2.0/glib/gmain.c instead of - /usr/src/debug/glib-2.0/1_2.58.3-r0.2/glib-2.58.3/glib/gmain.c I patch externalsrc.bbclass to dynamically caculate correct debug-prefix-map and prepend to DEBUG_PREFIX_MAP variable, I also patch the package.bbclass to add a condition when EXTERNALSRC is defined to change the way the sources are found and copied to packages-split/${pkg}-src. I would greatly appreciate if there is a yocto guru around there to tell me what he thinks about all of this (bug or wrong configuration ?) and if my assumptions are correct (I assume this was a compiling/packaging problem instead of a pure devtool one) -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#53936): https://lists.yoctoproject.org/g/yocto/message/53936 Mute This Topic: https://lists.yoctoproject.org/mt/83622035/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] Appending an existing machine .conf file
Hello, you can customize QB_OPT_APPEND by removing options you don't want, e.g. QB_OPT_APPEND_remove = "-device usb-tablet" and add your own after. But maybe a lazy operator in the original conf file would be more suitable. On Sat, 19 Jun 2021 at 06:41, Mike Frampton wrote: > Greetings Yocto, > > I'm interested in assigning custom config settings for machine type > qemuarm. > > I tried a method suggested in this thread > https://lists.yoctoproject.org/g/yocto/message/38359 by Ayoub Zaki, he > suggested defining the following in local.conf > > > include conf/machine/${MACHINE}-extra.conf > > However, this doesn't work for me because quemuarm.conf in poky/meta > **assigns** its variables, e.g., > > > QB_OPT_APPEND = "-device VGA,edid=on" > > QB_OPT_APPEND += "-device qemu-xhci -device usb-tablet -device usb-kbd" > > So that if I define a value for this, it will be overwritten. > > Is there another method for achieving this? If not I could submit a > patch to change poky's definitions to "??" assignments. > > Cheers, > --Mike > > > > -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#53917): https://lists.yoctoproject.org/g/yocto/message/53917 Mute This Topic: https://lists.yoctoproject.org/mt/83644679/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
Re: [yocto] Empty source package when using devtool
It it can help to understand, I rechecked the full content of source packages and glib-2.0 / networkamanager are not ok like I said earlire. There are not empty but they seem to contain only build generated files (for example glib/glibconfig.h which, in the glib source repository, is glib/glibconfig.h. *in* ). -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#53908): https://lists.yoctoproject.org/g/yocto/message/53908 Mute This Topic: https://lists.yoctoproject.org/mt/83622035/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
[yocto] Empty source package when using devtool
Hello, I currently used yocto warrior and I encounter an issue during packaging of recipes. I would like to embed some source package (to be able to run coverage on target) , this work without issues and places the source of the package in ${ROOTFS}/usr/src/debug, but when I 'devtool modify' a recipe to work on , the source package is totally empty. First, I suspected that was my various recipe customization so I tried the same method (devtool modify then packing it and look at content of -src package) for more standard package, here are my results: - glib-2.0 -> ok - networkmanager -> ok - dnsmasq -> empty source - zlib -> empty source I try to look at log.do_package but I didn't see any warning nor error there. Since warrior is EOL, I tried a fresh poky on dunfell and hardknott, I got the exact same results. Is anybody here have experienced this ? Is this bug or did I miss some configuration ? Thank you very much in advance for any help you can bring. -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#53907): https://lists.yoctoproject.org/g/yocto/message/53907 Mute This Topic: https://lists.yoctoproject.org/mt/83622035/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-