| Summary: | on qemux86-64, "bitbake poky-image-minimal" will fail if eglibc is used. | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] Meta-yocto | Reporter: | Dexuan Cui <dexuan.cui> |
| Component: | meta-yocto | Assignee: | Dexuan Cui <dexuan.cui> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | major | ||
| Priority: | High | CC: | jessica.zhang, ke.yu, kevin.tian, poky.bs.watcher, poky.watcher, richard.purdie, sgw |
| Version: | 0.9 | ||
| Target Milestone: | 0.9 M4 | ||
| Hardware: | x86 | ||
| OS: | x86_64 | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Program Management | Documentation change: | --- |
|
Description
Dexuan Cui
2010-09-17 20:50:46 UTC
The log file's size is >1000k and the bugzilla doesn't allow such a big file.
So let me paste the part of error when building sdk.
...
NOTE: package eglibc-2.12-r6: task do_compile: Succeeded
NOTE: Running task 2475 of 4790 (ID: 468, /distro/dcui/dexuan/meta/recipes-core/eglibc/eglibc_2.12.bb, do_install)
NOTE: package eglibc-2.12-r6: task do_install: Started
NOTE: package eglibc-2.12-r6: task do_install: Succeeded
NOTE: Running task 2476 of 4790 (ID: 469, /distro/dcui/dexuan/meta/recipes-core/eglibc/eglibc_2.12.bb, do_populate_sysroot)
NOTE: Running task 2477 of 4790 (ID: 475, /distro/dcui/dexuan/meta/recipes-core/eglibc/eglibc_2.12.bb, do_package)
NOTE: package eglibc-2.12-r6: task do_populate_sysroot: Started
NOTE: package eglibc-2.12-r6: task do_package: Started
NOTE: preparing tree for binary locale generation
ERROR: Task failed: ('function do_prep_locale_tree failed', '/distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/temp/log.do_prep_locale_tree.23861')
ERROR: Logfile of failure stored in: /distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/temp/log.do_prep_locale_tree.23861
Log data follows:
| cp: cannot stat `/distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/package/lib/*': No such file or directory
| NOTE: preparing tree for binary locale generation
NOTE: package eglibc-2.12-r6: task do_package: Failed
ERROR: Task 475 (/distro/dcui/dexuan/meta/recipes-core/eglibc/eglibc_2.12.bb, do_package) failed with 1
Waiting for 1 active tasks to finish:
1: eglibc-2.12-r6 do_populate_sysroot (pid 23860)
ERROR: Task failed: ('function do_siteconfig failed', '/distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/temp/log.do_siteconfig.23860')
ERROR: Logfile of failure stored in: /distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/temp/log.do_siteconfig.23860
Log data follows:
| configure: creating cache eglibc_cache
| checking for x86_64-poky-linux-gcc... x86_64-poky-linux-gcc -march=core2 -m64 --sysroot=/distro/dcui/dexuan-qemux86-64/build/tmp/sysroots/x86_64-poky-linux
| checking whether the C compiler works... no
| configure: error: in `/distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/site_config':
| configure: error: C compiler cannot create executables
| See `config.log' for more details.
NOTE: package eglibc-2.12-r6: task do_populate_sysroot: Failed
ERROR: Task 469 (/distro/dcui/dexuan/meta/recipes-core/eglibc/eglibc_2.12.bb, do_populate_sysroot) failed with 1
ERROR: '/distro/dcui/dexuan/meta/recipes-core/eglibc/eglibc_2.12.bb' failed
ERROR: '/distro/dcui/dexuan/meta/recipes-core/eglibc/eglibc_2.12.bb' failed
I suspect the files are in /distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/package/lib64/ and there is a patch we apply to glibc which needs applying to eglibc too for this problem. Can you confirm the lib64 directory exists and contains the expected files? (In reply to comment #2) > I suspect the files are in > /distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/package/lib64/ Hi Richard, thanks for the suggestion! Yes, the .../lib64 does exist: $ ls /distro/dcui/dexuan-qemux86-64/build/tmp/work/x86_64-poky-linux/eglibc-2.12-r6/package/lib64 ld-2.12.so libanl.so.1 libdl.so.2 libnss_compat-2.12.so libnss_hesiod-2.12.so libpcprofile.so librt.so.1 ld-linux-x86-64.so.2 libc-2.12.so libm-2.12.so libnss_compat.so.2 libnss_hesiod.so.2 libpthread-2.12.so libthread_db-1.0.so libBrokenLocale-2.12.so libc.so.6 libm.so.6 libnss_dns-2.12.so libnss_nis-2.12.so libpthread.so.0 libthread_db.so.1 libBrokenLocale.so.1 libcrypt-2.12.so libmemusage.so libnss_dns.so.2 libnss_nis.so.2 libresolv-2.12.so libutil-2.12.so libSegFault.so libcrypt.so.1 libnsl-2.12.so libnss_files-2.12.so libnss_nisplus-2.12.so libresolv.so.2 libutil.so.1 libanl-2.12.so libdl-2.12.so libnsl.so.1 libnss_files.so.2 libnss_nisplus.so.2 librt-2.12.so > and there is a patch we apply to glibc which needs applying to eglibc too for > this problem. Can you confirm the lib64 directory exists and contains the > expected files? Could you please explain more info about the patch? I looked at every file in /distro/dcui/dexuan-qemux86-64/meta/recipes-core/glibc/glibc-2.10.1/, but couldn't find any file or patch that seems be able to fix the issue. (In reply to comment #3) > Yes, the .../lib64 does exist: Thanks, this confirms the problem at least. > > and there is a patch we apply to glibc which needs applying to eglibc too for > > this problem. Can you confirm the lib64 directory exists and contains the > > expected files? > > Could you please explain more info about the patch? > I looked at every file in > /distro/dcui/dexuan-qemux86-64/meta/recipes-core/glibc/glibc-2.10.1/, but > couldn't find any file or patch that seems be able to fix the issue. My memory isn't correct and there is no patch we use for this, sorry. There must be some other difference between the glibc and eglibc recipe that is causing this to happen. I think the lib directory is pulled from the gcc multilib configuration but I need to look into this further to find out exactly where. I think the difference is the following which is in the glibc recipe but not the eglibc one:
export libc_cv_slibdir = "${base_libdir}"
(this defaults to /lib64 for qemux86-64)
I'm testing this change to see if it resolves the problem.
I've pushed a fix in d8137c014f74d32dcdd2e148523640842a6ac4b1 (In reply to comment #6) > I've pushed a fix in d8137c014f74d32dcdd2e148523640842a6ac4b1 Great!It did fix the eglibc build failure in my side, too. I'm now continuing building qemux86-64. (In reply to comment #7) > (In reply to comment #6) > > I've pushed a fix in d8137c014f74d32dcdd2e148523640842a6ac4b1 > Great!It did fix the eglibc build failure in my side, too. > I'm now continuing building qemux86-64. Thanks to Richard's fix, poky-image-sdk can also build and boot to X on qemux86-64 now. Since Richard's patch has been in poky.git's master, can I mark the bus as VERIFIED? BTW: Richard's patch has not been in poky-contrib.git. (In reply to comment #8) > (In reply to comment #7) > > (In reply to comment #6) > > > I've pushed a fix in d8137c014f74d32dcdd2e148523640842a6ac4b1 > > Great!It did fix the eglibc build failure in my side, too. > > I'm now continuing building qemux86-64. > Thanks to Richard's fix, poky-image-sdk can also build and boot to X on > qemux86-64 now. > Since Richard's patch has been in poky.git's master, can I mark the bus as > VERIFIED? > BTW: Richard's patch has not been in poky-contrib.git. If Richard's patch also fix the issue in your side, you can mark it as verified, since you are the reporter. BTW, you can also cherry-pick the patch into your contrib branch dcui/eglibc, so that people can reuse your branch to continue the work. Regards Ke (In reply to comment #9) > (In reply to comment #8) > > (In reply to comment #7) > > > (In reply to comment #6) > > > > I've pushed a fix in d8137c014f74d32dcdd2e148523640842a6ac4b1 > > > Great!It did fix the eglibc build failure in my side, too. > > > I'm now continuing building qemux86-64. > > Thanks to Richard's fix, poky-image-sdk can also build and boot to X on > > qemux86-64 now. > > Since Richard's patch has been in poky.git's master, can I mark the bus as > > VERIFIED? > > BTW: Richard's patch has not been in poky-contrib.git. > If Richard's patch also fix the issue in your side, you can mark it as Yes. > verified, since you are the reporter. OK. So let me VERIFY the bug. > BTW, you can also cherry-pick the patch into your contrib branch dcui/eglibc, > so that people can reuse your branch to continue the work. Will do. |