[yocto] [AUH] Upgrade status: 2016-07-03
AUH finished upgrade batch the result patches/logs can be found at: https://logs.yoctoproject.org/auh//20160630215455.tar.gz, next are the statistics: Recipe upgrade statistics: * Failed(do_fetch): 6 python3-setuptools, 23.1.0, Alejandro Hernandezgit, 2.9.0, Robert Yang lttng-ust, 2.8.1+gitAUTOINC+514a87f3b6, Richard Purdie lttng-tools, 2.8.0+gitAUTOINC+a0f518a0b5, Richard Purdie lttng-modules, 2.8.0+gitAUTOINC+4a24ffe737, Richard Purdie babeltrace, 1.4.0, Alexander Kanavin * Failed(other errors): 22 bdwgc, 7.4.4, Alexander Kanavin slang, 2.3.0, Kai Kang lsbinitscripts, 9.66, Ross Burton perl, 5.24.0, Alejandro Hernandez python-setuptools, 23.1.0, Alejandro Hernandez gdb, 7.11.1, Richard Purdie fontconfig, 2.12.0, Jussi Kukkonen binutils, 2.26.1, Robert Yang cairo, 1.15.2, Jussi Kukkonen expat, 2.2.0, Jussi Kukkonen file, 5.28, Robert Yang u-boot-mkimage, v2016.05+gitAUTOINC+aeaec0e682, Denys Dmytriyenko nasm, 2.12.01, Richard Purdie gettext, 0.19.8.1, Wenzong Fan rpm, 5.4.17, Mark Hatle libpng, 1.6.23, Maxin B. John gnu-efi, 3.0.4, Robert Yang prelink, 20151030.+gitAUTOINC+927979bbd1, Mark Hatle cups, 2.1.4, Chen Qi wayland-protocols, 1.4, Jussi Kukkonen nss, 3.24, Alexander Kanavin bash-completion, 2.3, Alexander Kanavin * Succeeded: 47 libxfixes, 5.0.2, Jussi Kukkonen python3-git, 2.0.6, Alejandro Hernandez libgpg-error, 1.23, Jussi Kukkonen e2fsprogs, 1.43.1, Robert Yang busybox, 1.25.0, Chen Qi sysstat, 11.3.5, Chen Qi vala, 0.32.1, Alexander Kanavin xf86-input-evdev, 2.10.3, Jussi Kukkonen libevdev, 1.5.2, Jussi Kukkonen swig, 3.0.10, Alejandro Hernandez xkeyboard-config, 2.18, Jussi Kukkonen gnupg, 2.1.13, Hongxu Jia libgcrypt, 1.7.1, Hongxu Jia sudo, 1.8.17p1, Chen Qi libarchive, 3.2.1, Alexander Kanavin libsndfile1, 1.0.27, Tanu Kaskinen gsettings-desktop-schemas, 3.21.3, Jussi Kukkonen augeas, 1.5.0, Jussi Kukkonen at-spi2-core, 2.20.2, Jussi Kukkonen at-spi2-atk, 2.20.1, Jussi Kukkonen vte, 0.44.2, Jussi Kukkonen xf86-input-synaptics, 1.8.99.1, Jussi Kukkonen subversion, 1.9.4, Richard Purdie gstreamer1.0-plugins-bad, 1.8.2, Maxin B. John cronie, 1.5.1, Alejandro Hernandez libsolv, 0.6.22, Maxin B. John wayland, 1.11.0, Jussi Kukkonen settings-daemon, 2, Jussi Kukkonen libwnck3, 3.20.1, Alexander Kanavin gstreamer1.0-plugins-good, 1.8.2, Maxin B. John librsvg, 2.40.16, Jussi Kukkonen ethtool, 4.6, Maxin B. John ltp, 20160510, Robert Yang gstreamer1.0, 1.8.2, Maxin B. John matchbox-theme-sato, 2, Jussi Kukkonen gmp, 6.1.1, Robert Yang btrfs-tools, 4.6.1, Alexander Kanavin screen, 4.4.0, Maxin B. John
Re: [yocto] setcap using recipe
On 2 July 2016 at 11:37, Kumar, Shrawanwrote: > Post-installation scripts run immediately after installing a package on > the target *or during image creation when a package is included in an > image*. > > Does it not mean that we can set the file attributes(setcap) during > image creation ? > > > > I understand the delayed approach of executing it on the target, but my > requirement is to do it on the build host. We do not want “setcap” utility > to be present on the target. > > Then I suggest that you test the setcap patch that I pointed to earlier, you'll likely need the pseudo 1.8 upgrade that Joshua sent last night. This should ensure that setcap works in the pseudo environment. Then follow the instructions in image_types.bbclass to ensure that xattrs are preserved when the image is created, and check your kernel configuration to confirm that xattrs are supported. I recommend attempting to use fsetattr on the target directly to verify this. If all the pieces are in place then they should work. There's a number of bugs in bugzilla for tracking these issues (#9859 is the metabug tracking them). Your help in solving this would be much appreciated! Ross -- ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] setcap using recipe
Hello Ross, Post-installation scripts run immediately after installing a package on the target or during image creation when a package is included in an image. Does it not mean that we can set the file attributes(setcap) during image creation ? I understand the delayed approach of executing it on the target, but my requirement is to do it on the build host. We do not want “setcap” utility to be present on the target. Regards Shrawan From: Burton, Ross [mailto:ross.bur...@intel.com] Sent: Saturday, July 02, 2016 2:56 PM To: Kumar, Shrawan Cc: Daniel.; Mathieu Allard; yocto@yoctoproject.org Subject: Re: [yocto] setcap using recipe On 2 July 2016 at 10:22, Kumar, Shrawan> wrote: Can someone review the attached recipe and help solve the problem statement ? As has been said, you need to ensure the postinst is delayed so it runs on the target and not on the build host. http://www.yoctoproject.org/docs/2.1/dev-manual/dev-manual.html#new-recipe-post-installation-scripts Ross -- ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] setcap using recipe
On 2 July 2016 at 10:22, Kumar, Shrawanwrote: > Can someone review the attached recipe and help solve the problem > statement ? > As has been said, you need to ensure the postinst is delayed so it runs on the target and not on the build host. http://www.yoctoproject.org/docs/2.1/dev-manual/dev-manual.html#new-recipe-post-installation-scripts Ross -- ___ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
Re: [yocto] setcap using recipe
Dear All, The aim of my exercise is to apply "setcap " on an executable during rootfs creation. I should be able to view the set capability using getcap utility when this rootfs is mounted on the target. As I said earlier none of the suggested approach is working here.Currently I am qemux86 is my target. Can someone review the attached recipe and help solve the problem statement ? Regards Shrawan -Original Message- From: Daniel. [mailto:danielhi...@gmail.com] Sent: Friday, July 01, 2016 7:54 PM To: Mathieu Allard Cc: Kumar, Shrawan; yocto@yoctoproject.org Subject: Re: [yocto] setcap using recipe Hmmm I see, Well, I didn't note that. And yeah, that command should be ran at first boot, (that feature saved my life a bunch of times :) ) Regards, 2016-07-01 11:03 GMT-03:00 Mathieu Allard: > Hello, > > I think that the main issue here is that the pkg_postinst function runs its > action at the rootfs creation time, and not on the target as advised by Ross. > > The chapter 5.3.16, "post-installation scripts" in the mega-manual offers > some detailed explanations on how to make it run after the first boot. > > > Regards, > > Mathieu > > > - Original Message - > From: "Daniel." > To: "Kumar, Shrawan" > Cc: yocto@yoctoproject.org > Sent: Friday, July 1, 2016 3:54:15 PM > Subject: Re: [yocto] setcap using recipe > > Does your target filesystem support it? ubifs doesn't :( > http://www.linux-mtd.infradead.org/doc/ubifs.html#L_xattr > > 2016-07-01 9:53 GMT-03:00 Kumar, Shrawan : >> Hello Ross, >> >> >> >> None of the approach is working . I have attached the recipe where >> I am trying to execute postinst . It builds successfully , But when I >> run getcap on the target , does not return the set capabilities. >> >> >> >> Help will be highly appreciated . >> >> >> >> Regards >> >> Shrawan >> >> From: Burton, Ross [mailto:ross.bur...@intel.com] >> Sent: Friday, June 24, 2016 6:40 PM >> >> >> To: Kumar, Shrawan >> Cc: yocto@yoctoproject.org >> Subject: Re: [yocto] setcap using recipe >> >> >> >> Looks like using setcap directly is broken currently, there are two >> workarounds: >> >> >> >> 1) use a postinst to invoke setcap on the target instead >> >> 2) test the patch for pseudo that is on this list ([PATCH] Add capset >> pseudo function that always succeeds) and verify that it fixes the problem >> for you. >> >> >> >> Ross >> >> >> >> On 24 June 2016 at 13:31, Kumar, Shrawan wrote: >> >> I am using Yocto 2.0.2 >> >> >> >> Thanks and Regards >> >> Shrawan >> >> >> >> From: Burton, Ross [mailto:ross.bur...@intel.com] >> Sent: Friday, June 24, 2016 5:56 PM >> >> >> To: Kumar, Shrawan >> Cc: yocto@yoctoproject.org >> Subject: Re: [yocto] setcap using recipe >> >> >> >> What version of OE/Yocto are you using? Old versions of pseudo >> didn't support xattrs at all. >> >> >> >> Ross >> >> >> >> On 24 June 2016 at 13:23, Kumar, Shrawan wrote: >> >> Thanks Ross for your quick turn around , I am getting below error >> >> >> >> “Unable le to set CAP_SETFCAP effective capability: Operation not >> permitted.” >> >> >> >> But when I use# sudo setcap cap_net_raw+ep helloworldon command >> line I am able to set the cap. >> >> >> >> To achieve the sudo realization in recipe , I tried as below , but >> no luck…… Can you suggest something here ? >> >> >> >> fakeroot do_install() { >> >> install -d ${D}${bindir} >> >> install -m 0755 helloworld ${D}${bindir} >> >> install -d ${D}/lib/systemd/system >> >> install -m 0755 hello.service >> ${D}/lib/systemd/system/ >> >> setcap cap_net_raw+ep ${D}${bindir}/helloworld >> >> >> >> } >> >> >> >> Thanks and Regards >> >> Shrawan >> >> >> >> From: Burton, Ross [mailto:ross.bur...@intel.com] >> Sent: Friday, June 24, 2016 5:09 PM >> To: Kumar, Shrawan >> Cc: yocto@yoctoproject.org >> Subject: Re: [yocto] setcap using recipe >> >> >> >> Hi, >> >> >> >> On 24 June 2016 at 11:41, Kumar, Shrawan wrote: >> >> Is there a way to add a capability to a binary (cap_net_raw+ep),into >> a recipe? >> >> >> >> Example : >> >> do_install() { >> >>install -d ${D}${bindir} >> >>install -m 0755 helloworld ${D}${bindir} >> >>install -d ${D}/lib/systemd/system >> >>install -m 0755 hello.service ${D}/lib/systemd/system/ >> >>setcap cap_net_raw+ep ${D}${bindir}/helloworld >> >> } >> >> >> >> If yes is this correct approach to achieve the same from package >> recipe itself ? >> >> >> capabilities on files are just extended attributes, so assuming that >> you have a fairly recent Yocto and your host and target filesystems >> support extended attributes, yes this should work. >> >> >> >> Ross >> >> >> >> >> >> >> -- >>
Re: [yocto] meta-toolchain build errors
On 2016-07-01 14:21, Daniel. wrote: Did you tried "bitbake -fc cleanall imx6-demo-image && bitbake imx6-demo-image" !? I [effectively] did the same - started with an empty/new build and then % bitbake imx6-demo-image % bitbake imx6-demo-image -c populate_sdk Anyone know the innards enough to know what files I can compare to see where the mismatches are coming from? 2016-06-30 12:05 GMT-03:00 Khem Raj: On Thursday, June 30, 2016, Gary Thomas wrote: On 2016-06-29 17:27, Khem Raj wrote: On Jun 29, 2016 4:10 AM, "Gary Thomas" > wrote: > > I just [successfully] built a large image for my target (details): > BB_VERSION= "1.31.0" > BUILD_SYS = "x86_64-linux" > NATIVELSBSTRING = "Ubuntu-15.10" > TARGET_SYS= "arm-amltd-linux-gnueabi" > MACHINE = "teton-p0382" > DISTRO= "amltd" > DISTRO_VERSION= "2.1+snapshot-20160629" > TUNE_FEATURES = "arm armv7a vfp thumb neon callconvention-hard cortexa9" > TARGET_FPU= "hard" > meta = "master:c6d50b272936b61b98a056ea2e1de5d0f4ef53ce" > ... platform specifics removed ... > > Then I tried 'bitbake meta-toolchain' immediately afterwards, only > to be assaulted by a barrage of messages like this: > > ERROR: nativesdk-libsdl-1.2.15-r3 do_populate_sysroot: Taskhash mismatch 8ac6cdc4e26ea7fb43e76f0f9ce3def0 verses b05a369a00869fc70fe9ba2cc28e7259 for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-graphics/libsdl/libsdl_1.2.15.bb.do_populate_sysroot > ERROR: Taskhash mismatch 8ac6cdc4e26ea7fb43e76f0f9ce3def0 verses b05a369a00869fc70fe9ba2cc28e7259 for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-graphics/libsdl/libsdl_1.2.15.bb.do_populate_sysroot > ERROR: nativesdk-bash-completion-2.1-r0 do_compile: Taskhash mismatch 2e063c5bac7bf922fbff59be5120a62e verses 0223876e7a4bc1e57f93a44631b4e4aa for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-support/bash-completion/bash-completion_2.1.bb.do_compile > ERROR: Taskhash mismatch 2e063c5bac7bf922fbff59be5120a62e verses 0223876e7a4bc1e57f93a44631b4e4aa for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-support/bash-completion/bash-completion_2.1.bb.do_compile > ERROR: nativesdk-bash-completion-2.1-r0 do_install: Taskhash mismatch a9932697b29f5fbd06feafedea44d7d4 verses 0b4e83ecda4ce5560e50891715d6a5d1 for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-support/bash-completion/bash-completion_2.1.bb.do_install > ERROR: Taskhash mismatch a9932697b29f5fbd06feafedea44d7d4 verses 0b4e83ecda4ce5560e50891715d6a5d1 for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-support/bash-completion/bash-completion_2.1.bb.do_install > ERROR: nativesdk-bash-completion-2.1-r0 do_package: Taskhash mismatch 3cdba66cd643484bed95676b5a4cdc21 verses fe3108a63774b61b3910e75495fb4d5d for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-support/bash-completion/bash-completion_2.1.bb.do_package > ERROR: Taskhash mismatch 3cdba66cd643484bed95676b5a4cdc21 verses fe3108a63774b61b3910e75495fb4d5d for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-support/bash-completion/bash-completion_2.1.bb.do_package > ERROR: nativesdk-qemu-2.6.0-r1 do_configure: Taskhash mismatch c8c576da0ae11886e518efa4bc406f18 verses ad6e2e5d7704e6002f3b4ec56bc2d92b for virtual:nativesdk:/local/poky-cutting-edge/meta/recipes-devtools/qemu/qemu_2.6.0.bb.do_configure >... > NOTE: Tasks Summary: Attempted 2047 tasks of which 865 didn't need to be rerun and all succeeded. > Summary: There were 1212 ERROR messages shown, returning a non-zero exit code. > > There were too many error lines to copy :-( > > What does this mean and how do I get my toolchain? Bitbake -cpopulate_sdk your-image I tried that and it seems to have worked, but I still got similar error messages: ERROR: imx6-demo-image-1.0-r0 do_populate_sdk: Taskhash mismatch 5313c03c1d0cdd396e6128d6400e41e7 verses 96756bc7f74f66f35df9cb2fe9a596d0 for /local/poky-cutting-edge/meta-teton-imx6-p0382/packages/images/imx6-demo-image.bb.do_populate_sdk ERROR: Taskhash mismatch 5313c03c1d0cdd396e6128d6400e41e7 verses 96756bc7f74f66f35df9cb2fe9a596d0 for /local/poky-cutting-edge/meta-teton-imx6-p0382/packages/images/imx6-demo-image.bb.do_populate_sdk What file(s) can I look at to see why the mismatch happened? There could be some time related variable coming into checksum. I have seen such errors in other cases. You need to see why signature data contains these variables. One way is delete your tmp and rebuild -- Gary Thomas | Consulting for the MLB Associates |Embedded world -- ___ yocto mailing list