| Summary: | MUT: do_rootfs failed due to failed dependency on eglibc-locale | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Nitin Kamble <nitin.a.kamble> |
| Component: | core | Assignee: | Khem Raj <raj.khem> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium | CC: | bluelightning, corneliux.stoicescu, meta.mr.watcher, meta.watcher |
| Version: | unspecified | ||
| Target Milestone: | 1.3 M3 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | autobuilder | ||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | --- | |
|
Description
Nitin Kamble
2012-07-24 09:47:31 UTC
also happens on i586: http://autobuilder.yoctoproject.org:8010/builders/nightly-x86/builds/598/steps/shell_25/logs/stdio This is a MUT failure, probably triggered by my rootfs rework - the dependency issue is probably real though and needs to be looked at. I think this may actually be related to the 2.16 eglibc update, I believe I saw this also before paul's change. Khem may have a fix already. I have seen such issues when using rpm as packaging backend. When I used ipk then the issues disappeared. It may be that its showing up due to some latent issue. Mark Hatle touched the eglibc locale generation for his image SDK with external toolchain work. It could be that there is some empty package that is being requested through RDEPENDS, atleast that was one of problem I ran into recently where rpm did not give me useful message at all and message was very similar to what you are seeing. but opkg hinted rightly and I could fix it. It was an empty package being asked for in image and nothing related to locales. so now we have eglibc-locale which is separate recipe and then we generate ${PN}-locale out of it so it becomes eglibc-locale-locale-*, and then eglibc recipe will also spin out ${PN}-locale which will become eglibc-locale, something is not right here.
(In reply to comment #5) > so now we have eglibc-locale which is separate recipe and then we generate > ${PN}-locale out of it so it becomes eglibc-locale-locale-*, and then eglibc > recipe will also spin out ${PN}-locale which will become eglibc-locale, > something is not right here. I have a potential fix for this here http://git.openembedded.org/openembedded-core-contrib/commit/?h=kraj/misc-updates&id=eda847fa7f399073566e7e0404fd158bd685cda7 Please test and see if this helps. for the non-gplv3 build the issue is still present: Build Configuration: BB_VERSION = "1.15.3" TARGET_ARCH = "i586" TARGET_OS = "linux" MACHINE = "qemux86" DISTRO = "poky" DISTRO_VERSION = "1.2+snapshot-20120726" TUNE_FEATURES = "m32 i586" TARGET_FPU = "" meta meta-yocto = "stage/master_under_test:3713872e059ae9268165fb619a3201cfa43e102e" meta-qt3 = "master:f8475d5c5935695d4d1dfff1d7be30f49e162094" http://autobuilder.yoctoproject.org:8010/builders/nightly-non-gpl3/builds/175/steps/shell_23/logs/stdio its fixed by OE-Core rev: fc4b138948656d482603b80f8c8e86fdbb185df4 Hello All the attached autobuilder logs have expired, I can not access them thus I can not test this to verify the fix. Can someone please test the fix or give me more information on how to test this? Thank you. /Cornel If you can build core-image-lsb-sdk with one of the failing configurations and it does not fail at do_rootfs, that should verify that this issue is fixed; however if it fails with an error that is not in do_rootfs or not related to eglibc-locale, then it is a different bug. Verified under master, commit: 49649451cbe6ac3999aa337c4d9c2c622b76c178. Bitbaked core-image-lsb-sdk using qemuarm (like the original post) with a fresh build and it succeded with no error. |