**** BEGIN LOGGING AT Fri Aug 20 02:59:56 2010 Aug 20 14:02:44 Hello, does anyone know how to fix "configure: error: compiler support for __thread is required" while running task "openembedded.org/recipes/eglibc/eglibc-initial_2.11.bb, do_configure" in bitbake? Aug 20 14:03:38 Jordinar: clean whole toolchain and run again works for me Aug 20 14:04:25 cd oe_checkout; for i in recipes/binutils/*2.20.1*.bb recipes/gcc/gcc*4.5*bb recipes/eglibc/eglibc*2.12*.bb; do bitbake -c clean -b $i; done; bitbake -k favourite-image; Aug 20 14:04:31 update for your versions ofc Aug 20 14:20:47 hmm .. thanks for your advice, but this gives now another error when rebuilding the toolchain in recipes/binutils/binutils-cross_2.17.bb: "configure: error: cannot run C compiled programs." ... this has passed before. Aug 20 14:29:21 did you clean everythin? what's in config.log? Aug 20 14:34:56 i cleaned the whole tmp dir to try again ... , maybe it didn't work because I tampered in the toolchain to fix this error before .... just before that first error were also messages like "checking for string.h... no", it looks like a problem for me that include-header files aren't found, so directory-structure or environment variables are not correct.. Aug 20 14:35:36 the configure script that failed there runs fine when calling it manually - from outside bitbake. Aug 20 15:22:59 JaMa|Wrk: this is config.log - http://jordinar.dyndns.org/config.log - It looks like it doesn't find required files, e.g. wrong pathes? Aug 20 15:23:47 Jordinar: avr32-angstrom-linux-gcc: command not found Aug 20 15:23:54 seems like root of rproblem Aug 20 15:28:24 avr32-angstrom-linux-gcc is in at least 3 locations in $X../usr/avr32/bin/avr32-linux-gcc with $X like staging-pkg, image, ... which should I use and how to tell where to look for these files? Aug 20 15:29:04 this /usr/avr32/bin/... looks strange, shouldn't that be just ../usr/bin/avr32-linux-XXX ? Aug 20 15:29:29 Jordinar: are you building from org.openembedded.dev branch? and did you rebuild from scratch after staging->sysroots change and cross directory removal? Aug 20 15:31:31 yes, it's an up-to-date org.openembedded.dev and also built from scratch today, so I think that is after that change. Aug 20 15:33:05 I'm not building for avr32 but path should be similar to sysroots/x86_64-linux/usr/armv5te/bin/arm-angstrom-linux-gnueabi-gcc Aug 20 15:36:16 so it is org.openembedded.dev/work/avr32-angstrom-linux/gcc-cross-initial-4.2.2-r18.1/staging-pkg/sysroots/i686-linux/usr/avr32/bin/ ? Aug 20 15:36:26 (the avr32-linux-gcc is there) Aug 20 15:36:36 no.. Aug 20 15:36:54 work dir is just temporary for building Aug 20 15:37:07 after building the resulting stuff gets installed to sysroots directory Aug 20 15:37:20 and used from there since then Aug 20 15:38:24 Ok, in this case, the avr32-linux-* files are not copied to /org.openembedded.dev/sysroots Aug 20 15:38:44 are you sure? Aug 20 15:39:06 it will be sysroots/your_buildhost_architecture/usr/avr32* Aug 20 15:39:35 very sure. Aug 20 15:39:40 radi@shaman:~/oe/tmp/angstrom/org.openembedded.dev/sysroots$ find -name avr32-linux-gcc Aug 20 15:39:40 radi@shaman:~/oe/tmp/angstrom/org.openembedded.dev/sysroots$ Aug 20 15:40:17 Is it a good idea to copy these files manually from the working directory? I think the install-script from gcc-cross should have done this .. Aug 20 15:41:00 find -name avr32-angstrom-linux-gcc Aug 20 15:42:03 oh yes damn, it is there: ./sysroots/i686-linux/usr/avr32/bin/avr32-angstrom-linux-gcc Aug 20 15:44:27 the $PATH variable does not have this while running bitbake ... it is PATH=/home/radi/oe/tmp/angstrom/org.openembedded.dev/sysroots/i686-linux/usr/bin/avr32-angstrom-linux/bin:$a lot of other paths Aug 20 15:51:02 well looks like extra /bin before avr32-angstrom-linux.. but no idea why it should be broken now Aug 20 15:51:23 that's in those lot of other PATHs? Aug 20 15:52:06 * JaMa|Wrk leaving, bye Aug 20 15:53:13 I just added the real path, so it finds now avr32-angstrom-linux-gcc, but not avr32-angstrom-linux-g++, which has not been built yet... and the same error "compiler support for __thread is required" still there **** ENDING LOGGING AT Sat Aug 21 02:59:57 2010