941 2011-03-31 04:46:59 +0000 gnu-configize fails to run 2011-06-09 19:04:59 +0000 1 1 7 Build System, Metadata & Runtime Meta-yocto meta-yocto unspecified x86 Multiple RESOLVED FIXED The patch series submitted "looks good to" RP and being reviewed by the community (01/Jun/2011) High major 1.1 M2 968 1 gary dexuan.cui dexuan.cui edwin.zhai poky.bs.watcher poky.watcher sgw --- --- --- oldest_to_newest 13065 0 gary 2011-03-31 04:46:59 +0000 Any recipe which uses gnu-configize may have problems since this step currently fails. Tested against master (6f4fbfe272f5d85cee95660a9278f3031533aec4) Looking at log.do_configure for 'bitbake bash' shows these errors: Perl lib version (5.12.2) doesn't match executable version (v5.10.1) at /home/local/p60_poky/tmp/sysroots/i686-linux/usr/lib/perl/5.12.2/Config.pm line 50. Compilation failed in require at /home/local/p60_poky/tmp/sysroots/i686-linux/usr/lib/perl/5.12.2/File/Copy.pm line 14. BEGIN failed--compilation aborted at /home/local/p60_poky/tmp/sysroots/i686-linux/usr/lib/perl/5.12.2/File/Copy.pm line 14. Compilation failed in require at /home/local/p60_poky/tmp/sysroots/i686-linux/usr/share/autoconf/Autom4te/FileUtils.pm line 166. BEGIN failed--compilation aborted at /home/local/p60_poky/tmp/sysroots/i686-linux/usr/share/autoconf/Autom4te/FileUtils.pm line 166. Compilation failed in require at /home/local/p60_poky/tmp/sysroots/i686-linux/usr/bin/autom4te line 41. BEGIN failed--compilation aborted at /home/local/p60_poky/tmp/sysroots/i686-linux/usr/bin/autom4te line 41. 13215 1 dexuan.cui 2011-04-11 21:52:23 +0000 Coul (In reply to comment #0) > Any recipe which uses gnu-configize may have problems since this step currently > fails. > Tested against master (6f4fbfe272f5d85cee95660a9278f3031533aec4) > Looking at log.do_configure for 'bitbake bash' shows these errors: > Perl lib version (5.12.2) doesn't match executable version (v5.10.1) at > /home/local/p60_poky/tmp/sysroots/i686-linux/usr/lib/perl/5.12.2/Config.pm line > 50. > Compilation failed in require at > /home/local/p60_poky/tmp/sysroots/i686-linux/usr/lib/perl/5.12.2/File/Copy.pm > line 14. > BEGIN failed--compilation aborted at > /home/local/p60_poky/tmp/sysroots/i686-linux/usr/lib/perl/5.12.2/File/Copy.pm > line 14. > Compilation failed in require at > /home/local/p60_poky/tmp/sysroots/i686-linux/usr/share/autoconf/Autom4te/FileUtils.pm > line 166. > BEGIN failed--compilation aborted at > /home/local/p60_poky/tmp/sysroots/i686-linux/usr/share/autoconf/Autom4te/FileUtils.pm > line 166. > Compilation failed in require at > /home/local/p60_poky/tmp/sysroots/i686-linux/usr/bin/autom4te line 41. > BEGIN failed--compilation aborted at > /home/local/p60_poky/tmp/sysroots/i686-linux/usr/bin/autom4te line 41. Could this bug be a duplicate of bug 968? Hi Gary, can you please check in the build log if perl-native's and gnu-config-native's do_populate_sysroot succeed or not before the first ERROR? 13249 2 gary 2011-04-12 16:58:59 +0000 I don't think it's related to bug #968 As for the ordering, here are the relevant time stamps -rw-rw-r-- 1 gthomas gthomas 0 Mar 30 06:23 tmp/stamps/i686-linux/gnu-config-native-0.1+cvs20080123-r2.do_populate_sysroot -rw-rw-r-- 1 gthomas gthomas 0 Mar 30 06:40 tmp/stamps/i686-linux/perl-native-5.12.2-r8.do_populate_sysroot -rw-rw-r-- 1 gthomas gthomas 0 Mar 31 05:16 tmp/stamps/armv7a-poky-linux-gnueabi/bash-4.1-r1.do_configure 13438 3 edwin.zhai 2011-04-27 01:59:47 +0000 gary, I can't reproduce it on 927d33c170f062c0a59d2a321679d5fe22e80a97 @ master by 'bitbake bash'. Seems it got fixed. Can you have a try, or I need do something else? Thanks, edwin 13439 4 dexuan.cui 2011-04-27 02:03:28 +0000 (In reply to comment #3) > gary, > I can't reproduce it on 927d33c170f062c0a59d2a321679d5fe22e80a97 @ master by > 'bitbake bash'. > Seems it got fixed. Can you have a try, or I need do something else? > Thanks, > edwin Edwin, please see the tmo/workdir/.../bash*/temp/log.do_configure that has the error msgs. I suppost this is What Gary meant. 13440 5 gary 2011-04-27 03:54:41 +0000 Yes, you only see this error message in log.do_configure A closer look shows that it seems to be running Perl scripts and libraries from sysroots, but the system host Perl is being called, thus causing the mismatch. On my system, here's the main host version: $ perl --version This is perl, v5.10.1 (*) built for i386-linux-thread-multi This is the one in sysroots that _should_ be used instead: $ tmp/sysroots/i686-linux/usr/bin/perl --version This is perl 5, version 12, subversion 2 (v5.12.2) built for i686-linux-thread-multi So however Perl is being executed in the gnu-configize, it's getting the host program (5.10.1), not the correct sysroots version (5.12.2) 13506 6 dexuan.cui 2011-04-29 01:40:45 +0000 (In reply to comment #5) > Yes, you only see this error message in log.do_configure > A closer look shows that it seems to be running Perl scripts and > libraries from sysroots, but the system host Perl is being called, > thus causing the mismatch. > On my system, here's the main host version: > $ perl --version > This is perl, v5.10.1 (*) built for i386-linux-thread-multi > This is the one in sysroots that _should_ be used instead: > $ tmp/sysroots/i686-linux/usr/bin/perl --version > This is perl 5, version 12, subversion 2 (v5.12.2) built for > i686-linux-thread-multi > So however Perl is being executed in the gnu-configize, it's getting > the host program (5.10.1), not the correct sysroots version (5.12.2) autoconf-native's do_configure detects perl from PATH and reflect the path into the scripts generated, e.g., ${STAGING_BINDIR_NATIVE}/autom4te. If perl-native's does populate_sysroot later than autoconf, the host's perl will be used -- this is not what we desire. I used the following patch to meta/recipes-devtools/autoconf/autoconf_2.65.bb -PR = "r2" +PR = "r3" -DEPENDS_virtclass-native = "m4-native gnu-config-native" +DEPENDS_virtclass-native = "m4-native gnu-config-native perl-native" This fixes the bug in my side. 13507 7 dexuan.cui 2011-04-29 01:43:29 +0000 (In reply to comment #6) > (In reply to comment #5) > > Yes, you only see this error message in log.do_configure > > A closer look shows that it seems to be running Perl scripts and > > libraries from sysroots, but the system host Perl is being called, > > thus causing the mismatch. > > On my system, here's the main host version: > > $ perl --version > > This is perl, v5.10.1 (*) built for i386-linux-thread-multi > > This is the one in sysroots that _should_ be used instead: > > $ tmp/sysroots/i686-linux/usr/bin/perl --version > > This is perl 5, version 12, subversion 2 (v5.12.2) built for > > i686-linux-thread-multi > > So however Perl is being executed in the gnu-configize, it's getting > > the host program (5.10.1), not the correct sysroots version (5.12.2) > autoconf-native's do_configure detects perl from PATH and reflect the path > into the scripts generated, e.g., ${STAGING_BINDIR_NATIVE}/autom4te. > If perl-native's does populate_sysroot later than autoconf, the host's perl > will be used -- this is not what we desire. > I used the following patch to meta/recipes-devtools/autoconf/autoconf_2.65.bb > -PR = "r2" > +PR = "r3" > -DEPENDS_virtclass-native = "m4-native gnu-config-native" > +DEPENDS_virtclass-native = "m4-native gnu-config-native perl-native" > This fixes the bug in my side. mised one more piece of patch: in meta/recipes-devtools/gnu-config/gnu-config/gnu-configize.in, should change the hardcoded "/usr/bin/perl" with "perl". -eval 'case $# in 0) exec /usr/bin/perl -S "$0";; *) exec /usr/bin/perl -S "$0" "$@";; esac' +eval 'case $# in 0) exec perl -S "$0";; *) exec perl -S "$0" "$@";; esac' 13510 8 gary 2011-04-29 07:11:01 +0000 Those two changes (which probably need to be separate in a series) do seem to fix the issue. Now when I run 'bitbake bash -c configure' there are no indications that gnu-configize failed in any way. 13584 9 dexuan.cui 2011-05-04 23:19:07 +0000 (In reply to comment #6) > (In reply to comment #5) > > Yes, you only see this error message in log.do_configure > > A closer look shows that it seems to be running Perl scripts and > > libraries from sysroots, but the system host Perl is being called, > > thus causing the mismatch. > > On my system, here's the main host version: > > $ perl --version > > This is perl, v5.10.1 (*) built for i386-linux-thread-multi > > This is the one in sysroots that _should_ be used instead: > > $ tmp/sysroots/i686-linux/usr/bin/perl --version > > This is perl 5, version 12, subversion 2 (v5.12.2) built for > > i686-linux-thread-multi > > So however Perl is being executed in the gnu-configize, it's getting > > the host program (5.10.1), not the correct sysroots version (5.12.2) > autoconf-native's do_configure detects perl from PATH and reflect the path > into the scripts generated, e.g., ${STAGING_BINDIR_NATIVE}/autom4te. > If perl-native's does populate_sysroot later than autoconf, the host's perl > will be used -- this is not what we desire. Actually, not only autoconf-native but also automake-native, quilt-native, glib-2.0-native, util-linux-native, quilt-native have the same issue (they all don't cause this bug) {quilt, glib-2.0, util-linux}-native depend on {autoconf,automake}-native, and {autoconf,automake}-native depend on gnu-config-native. So I think the proper fix should be making quilt-native and gnu-config-native depend on perl-native (surely we still need to "bitbake -c clean" {autoconf,automake,glib-2.0,util-linux} and bitbake them again) > mised one more piece of patch: > in meta/recipes-devtools/gnu-config/gnu-config/gnu-configize.in, should change > the hardcoded "/usr/bin/perl" with "perl". > -eval 'case $# in 0) exec /usr/bin/perl -S "$0";; *) exec /usr/bin/perl -S "$0" > "$@";; esac' > +eval 'case $# in 0) exec perl -S "$0";; *) exec perl -S "$0" "$@";; esac' This change is still necessary. 13586 10 gary 2011-05-05 03:48:30 +0000 Do you have a patch which makes all of the changes for me to test? 13588 11 144 dexuan.cui 2011-05-05 06:05:11 +0000 Created attachment 144 gnu-config.patch 13589 12 dexuan.cui 2011-05-05 06:08:10 +0000 (In reply to comment #10) > Do you have a patch which makes all of the changes for me to test? Yes, please help to test the attached patch in comment #11. After applying the patch, at least we also need to do bitbake -c cleanall autoconf-native automake-native. (In reply to comment #11) > Created attachment 144 [details] > gnu-config.patch 13590 13 gary 2011-05-05 06:21:50 +0000 What about the previous change to autoconf-native? Is this not necessary (it's not in this patch)? 13591 14 dexuan.cui 2011-05-05 07:29:41 +0000 (In reply to comment #13) > What about the previous change to autoconf-native? Is this not necessary (it's > not in this patch)? Yes, I think it's not necessary as autoconf-native depends on gnu-config-native. If we add perl-native into gnu-config-native's DEPENDS, we can make sure autoconf-native also depends on perl-native. Certainly we need to re-build and re-populate autoconf-native. Does this make sense? 13592 15 gary 2011-05-05 07:31:52 +0000 I'll give it a try with only this patch. 13593 16 dexuan.cui 2011-05-05 07:33:40 +0000 (In reply to comment #9) > (In reply to comment #6) > > (In reply to comment #5) > > > Yes, you only see this error message in log.do_configure > > > A closer look shows that it seems to be running Perl scripts and > > > libraries from sysroots, but the system host Perl is being called, > > > thus causing the mismatch. > > > On my system, here's the main host version: > > > $ perl --version > > > This is perl, v5.10.1 (*) built for i386-linux-thread-multi > > > This is the one in sysroots that _should_ be used instead: > > > $ tmp/sysroots/i686-linux/usr/bin/perl --version > > > This is perl 5, version 12, subversion 2 (v5.12.2) built for > > > i686-linux-thread-multi > > > So however Perl is being executed in the gnu-configize, it's getting > > > the host program (5.10.1), not the correct sysroots version (5.12.2) > > autoconf-native's do_configure detects perl from PATH and reflect the path > > into the scripts generated, e.g., ${STAGING_BINDIR_NATIVE}/autom4te. > > If perl-native's does populate_sysroot later than autoconf, the host's perl > > will be used -- this is not what we desire. > Actually, not only autoconf-native but also automake-native, quilt-native, > glib-2.0-native, util-linux-native, quilt-native have the same issue (they all > don't cause this bug) > {quilt, glib-2.0, util-linux}-native depend on {autoconf,automake}-native, and > {autoconf,automake}-native depend on gnu-config-native. > So I think the proper fix should be making quilt-native and gnu-config-native > depend on perl-native (surely we still need to "bitbake -c clean" > {autoconf,automake,glib-2.0,util-linux} and bitbake them again) > > mised one more piece of patch: > > in meta/recipes-devtools/gnu-config/gnu-config/gnu-configize.in, should change > > the hardcoded "/usr/bin/perl" with "perl". > > -eval 'case $# in 0) exec /usr/bin/perl -S "$0";; *) exec /usr/bin/perl -S "$0" > > "$@";; esac' > > +eval 'case $# in 0) exec perl -S "$0";; *) exec perl -S "$0" "$@";; esac' > This change is still necessary. Maybe doing sed replacement for gnu-config-native is a better way: sed -i -e 's,/usr/bin/perl,${STAGING_BINDIR_NATIVE}/perl,g' ${D}${bindir}/gnu-configize 13594 17 gary 2011-05-05 07:44:25 +0000 Seems to work with only the patch to gnu-config 13595 18 dexuan.cui 2011-05-05 07:46:39 +0000 (In reply to comment #17) > Seems to work with only the patch to gnu-config Gary, thank you very much for the verification! I'll do more tests and try to send out the patch to oe-core mailing list tomorrow. 13607 19 dexuan.cui 2011-05-05 21:04:00 +0000 (In reply to comment #16) > (In reply to comment #9) > > (In reply to comment #6) > > > (In reply to comment #5) > > > > Yes, you only see this error message in log.do_configure > > > > A closer look shows that it seems to be running Perl scripts and > > > > libraries from sysroots, but the system host Perl is being called, > > > > thus causing the mismatch. > > > > On my system, here's the main host version: > > > > $ perl --version > > > > This is perl, v5.10.1 (*) built for i386-linux-thread-multi > > > > This is the one in sysroots that _should_ be used instead: > > > > $ tmp/sysroots/i686-linux/usr/bin/perl --version > > > > This is perl 5, version 12, subversion 2 (v5.12.2) built for > > > > i686-linux-thread-multi > > > > So however Perl is being executed in the gnu-configize, it's getting > > > > the host program (5.10.1), not the correct sysroots version (5.12.2) > > > autoconf-native's do_configure detects perl from PATH and reflect the path > > > into the scripts generated, e.g., ${STAGING_BINDIR_NATIVE}/autom4te. > > > If perl-native's does populate_sysroot later than autoconf, the host's perl > > > will be used -- this is not what we desire. > > Actually, not only autoconf-native but also automake-native, quilt-native, > > glib-2.0-native, util-linux-native, quilt-native have the same issue (they all > > don't cause this bug) > > {quilt, glib-2.0, util-linux}-native depend on {autoconf,automake}-native, and > > {autoconf,automake}-native depend on gnu-config-native. > > So I think the proper fix should be making quilt-native and gnu-config-native > > depend on perl-native (surely we still need to "bitbake -c clean" > > {autoconf,automake,glib-2.0,util-linux} and bitbake them again) > > > mised one more piece of patch: > > > in meta/recipes-devtools/gnu-config/gnu-config/gnu-configize.in, should change > > > the hardcoded "/usr/bin/perl" with "perl". > > > -eval 'case $# in 0) exec /usr/bin/perl -S "$0";; *) exec /usr/bin/perl -S "$0" > > > "$@";; esac' > > > +eval 'case $# in 0) exec perl -S "$0";; *) exec perl -S "$0" "$@";; esac' > > This change is still necessary. > Maybe doing sed replacement for gnu-config-native is a better way: > sed -i -e 's,/usr/bin/perl,${STAGING_BINDIR_NATIVE}/perl,g' > ${D}${bindir}/gnu-configize RP doesn't think this is ok. We're discussining. 13660 20 dexuan.cui 2011-05-10 18:02:10 +0000 A workaround patch has been in poky master: http://git.pokylinux.org/cgit/cgit.cgi/poky/commit/?id=605141a93443df042634b2219a8628a9004be023 And I'm working to tidy the usages of perl-native/perl-native-runtime to get a real fix. 13792 21 sgw 2011-05-13 10:49:24 +0000 Work-around for bernard http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=5d3bfbbd187c4eb9fbfe66f0a9f3f22e5ce94427 14241 22 sgw 2011-06-09 19:04:59 +0000 Patch series has been applied to master and oe-core 144 2011-05-05 06:05:11 +0000 2011-05-05 06:05:11 +0000 gnu-config.patch gnu-config.patch text/plain 1503 dexuan.cui ZGlmZiAtLWdpdCBhL21ldGEvcmVjaXBlcy1kZXZ0b29scy9nbnUtY29uZmlnL2dudS1jb25maWcv Z251LWNvbmZpZ2l6ZS5pbiBiL21ldGEvcmVjaXBlcy1kZXZ0b29scy9nbnUtY29uZmlnL2dudS1j b25maWcvZ251LWNvbmZpZ2l6ZS5pbgppbmRleCBkNDkwOGZjLi5hZGY5YzBlIDEwMDc1NQotLS0g YS9tZXRhL3JlY2lwZXMtZGV2dG9vbHMvZ251LWNvbmZpZy9nbnUtY29uZmlnL2dudS1jb25maWdp emUuaW4KKysrIGIvbWV0YS9yZWNpcGVzLWRldnRvb2xzL2dudS1jb25maWcvZ251LWNvbmZpZy9n bnUtY29uZmlnaXplLmluCkBAIC0xLDcgKzEsNyBAQAogIyEgL3Vzci9iaW4vZW52IHBlcmwKICMg LSotIHBlcmwgLSotCiAKLWV2YWwgJ2Nhc2UgJCMgaW4gMCkgZXhlYyAvdXNyL2Jpbi9wZXJsIC1T ICIkMCI7OyAqKSBleGVjIC91c3IvYmluL3BlcmwgLVMgIiQwIiAiJEAiOzsgZXNhYycKK2V2YWwg J2Nhc2UgJCMgaW4gMCkgZXhlYyBwZXJsIC1TICIkMCI7OyAqKSBleGVjIHBlcmwgLVMgIiQwIiAi JEAiOzsgZXNhYycKICAgICBpZiAwOwogCiAjIGdudS1jb25maWdpemUgLSBpbnN0YWxsIHRoZSBH TlUgY29uZmlnLmd1ZXNzIC8gY29uZmlnLnN1YiBpbiBhIGRpcmVjdG9yeSB0cmVlCmRpZmYgLS1n aXQgYS9tZXRhL3JlY2lwZXMtZGV2dG9vbHMvZ251LWNvbmZpZy9nbnUtY29uZmlnXzIwMDgwMTIz LmJiIGIvbWV0YS9yZWNpcGVzLWRldnRvb2xzL2dudS1jb25maWcvZ251LWNvbmZpZ18yMDA4MDEy My5iYgppbmRleCBlMGE4MTU1Li5jMjFlYWE0IDEwMDY0NAotLS0gYS9tZXRhL3JlY2lwZXMtZGV2 dG9vbHMvZ251LWNvbmZpZy9nbnUtY29uZmlnXzIwMDgwMTIzLmJiCisrKyBiL21ldGEvcmVjaXBl cy1kZXZ0b29scy9nbnUtY29uZmlnL2dudS1jb25maWdfMjAwODAxMjMuYmIKQEAgLTMsMTIgKzMs MTQgQEAgREVTQ1JJUFRJT04gPSAiVG9vbCB0aGF0IGluc3RhbGxzIHRoZSBHTlUgY29uZmlnLmd1 ZXNzIC8gY29uZmlnLnN1YiBpbnRvIGEgZGlyZWMKIFNFQ1RJT04gPSAiZGV2ZWwiCiBMSUNFTlNF ID0gIkdQTHYxKyIKIExJQ19GSUxFU19DSEtTVU0gPSAiZmlsZTovL2NvbmZpZy5ndWVzcztlbmRs aW5lPTM5O21kNT1hMDg5OTg3YWY0YTI1Y2IwNDE5ZDFjMmZkNmQ0OTVlMyIKLURFUEVORFMgPSAi IgorCitERVBFTkRTX3ZpcnRjbGFzcy1uYXRpdmUgPSAicGVybC1uYXRpdmUiCisKIElOSElCSVRf REVGQVVMVF9ERVBTID0gIjEiCiAKIEZJWEVEU1JDREFURSA9ICIke0BiYi5kYXRhLmdldFZhcign RklMRScsIGQsIDEpLnNwbGl0KCdfJylbLTFdLnNwbGl0KCcuJylbMF19IgogUFYgPSAiMC4xK2N2 cyR7RklYRURTUkNEQVRFfSIKLVBSID0gInIyIgorUFIgPSAicjMiCiAKIFNSQ19VUkkgPSAiY3Zz Oi8vYW5vbnltb3VzQGN2cy5zdi5nbnUub3JnL2N2c3Jvb3QvY29uZmlnO21vZHVsZT1jb25maWc7 bWV0aG9kPXBzZXJ2ZXI7ZGF0ZT0ke0ZJWEVEU1JDREFURX0gXAogCSAgIGZpbGU6Ly9jb25maWct Z3Vlc3MtdWNsaWJjLnBhdGNoIFwK