Re: Testing required: SVN build segfaults on stripping chapter6
El Sábado, 24 de Febrero de 2007 00:36, Bruce Dubbs escribió: Manuel, I've only been following this casually. What was different about the chroot command that would cause a segfault? The first issue was that chapter06/115-strippingagain script was using #!/bin/bash as the shabang, thus after stripping libreadline and/or libhistory the subshell hangs, leading to a make: Error 126 That has been fixed changing that script shabang to #!/tools/bin/bash The second issue, not fixed yet, is that when resuming a failed chroot build after installing Chapte06 Bash, $LFS/bin/bash exist and is the final bash binary (not the link to /tools/bin/bash created at the begining of the chapter), thus make mk_CHROOT will use $LFS/bin/bash as the $SHELL instead using /tools/bin/bash (like does on a clean build), segfaulting when running the strip command. To fix that we need to force the mk_CHROOT target to allways use /tools/bin/bash as their $SHELL, but withiut messing the shell used by sub-make process launched by build scripts :-/ -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
M.Canales.es wrote: El Viernes, 23 de Febrero de 2007 20:26, M.Canales.es escribió: That should corfim that the issue is in jhalfs, not related with the book or the GCC/Glibc planned updates. And fixed. After knowing what was causing the segfault in both a full build and later in my partial build tests, the fix was terribly obvious :-/ Does this mean that LFS LiveCD 6.3-pre2 has to use jhalfs from SVN trunk? -- Alexander E. Patrakov -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
El Sábado, 24 de Febrero de 2007 13:02, Alexander E. Patrakov escribió: Does this mean that LFS LiveCD 6.3-pre2 has to use jhalfs from SVN trunk? We want to release jhalfs-2.2 in a week or so. Plus that fixes it will have also other fixes for CLFSx books and add support for BLFS-6.2. -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
M.Canales.es wrote: OK, thanks for the explanation. I understand what was happening now. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Testing required: SVN build segfaults on stripping chapter6
Hi folks.
At least Manuel and I are experiencing segfaults when using jhalfs. It occurs
during chapter06/stripping-again. I've narrowed the problem down
to /lib/lib{history,readline}.so.5.2. Can anyone else confirm that
performing the following command fails (either segfaults, or causes the shell
to hang):
/tools/bin/find /{,usr/}{bin,lib,sbin} -type f \
-exec /tools/bin/strip --strip-debug '{}' ';'
Could you also confirm that the following command succeeds:
/tools/bin/find /{,usr/}{bin,lib,sbin} -type f \
-name -not libhistory.so.5.2 -name -not libreadline.so.5.2 \
-exec /tools/bin/strip --strip-debug '{}' ';'
If anyone can confirm the above, could you also suggest an explanation as to
why the above two libraries are problematic? I initially thought it was a
problem specific to jhalfs, because of the way it executes scripts
under /bin/bash (therefore /lib/lib* would be in use and stripping may cause
problems). I think Manuel's done tests to force the stripping to be executed
under /tools/bin/bash and had the same result though, which rules that
explanation out.
Thanks,
Matt.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
On 2/23/07, Matthew Burgess [EMAIL PROTECTED] wrote: If anyone can confirm the above, could you also suggest an explanation as to why the above two libraries are problematic? I initially thought it was a problem specific to jhalfs, because of the way it executes scripts under /bin/bash (therefore /lib/lib* would be in use and stripping may cause problems). I think Manuel's done tests to force the stripping to be executed under /tools/bin/bash and had the same result though, which rules that explanation out. I think the issue in jhalfs may be because progress-bar.sh is using /bin/bash. Just a guess, but I saw the same issue before and that was what initially struck me. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
El Viernes, 23 de Febrero de 2007 19:39, Matthew Burgess escribió: If anyone can confirm the above, could you also suggest an explanation as to why the above two libraries are problematic? I initially thought it was a problem specific to jhalfs, because of the way it executes scripts under /bin/bash (therefore /lib/lib* would be in use and stripping may cause problems). I think Manuel's done tests to force the stripping to be executed under /tools/bin/bash and had the same result though, which rules that explanation out. I still think that, at least for current LFS-SVN, the issue is jhalfs-specific due the recent Makefile $SHELL changes. But Matthew confirmed that while testing the updates to GCC-4.1.2 and glibc branch_update patch, running the stripping command manually from inside the chroot jail hangs the shell, thus maybe there is other issues involved here. -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
El Viernes, 23 de Febrero de 2007 19:53, Dan Nicholson escribió: I think the issue in jhalfs may be because progress-bar.sh is using /bin/bash. Just a guess, but I saw the same issue before and that was what initially struck me. Not, reducing the Makefile target to just: 116-strippingagain: 114-vim source envars $(crCMDSDIR)/chapter06/$@ segfaults also. -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
On Friday 23 February 2007 19:08, M.Canales.es wrote: But Matthew confirmed that while testing the updates to GCC-4.1.2 and glibc branch_update patch, running the stripping command manually from inside the chroot jail hangs the shell, thus maybe there is other issues involved here. Ah, slight clarification here! I'm actually running the command via the following (from chapter06/revisedchroot.html): sudo chroot $LFS /usr/bin/env -i \ HOME=/root TERM=\$TERM\ PS1='\u:\w\$ ' \ PATH=/bin:/usr/bin:/sbin:/usr/sbin \ /bin/bash --login Therefore, /bin/bash is linked to /lib/* which is why it exhibits the same behavior, I'd guess. -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
El Viernes, 23 de Febrero de 2007 20:18, Matthew Burgess escribió: Ah, slight clarification here! I'm actually running the command via the following (from chapter06/revisedchroot.html): sudo chroot $LFS /usr/bin/env -i \ HOME=/root TERM=\$TERM\ PS1='\u:\w\$ ' \ PATH=/bin:/usr/bin:/sbin:/usr/sbin \ /bin/bash --login Therefore, /bin/bash is linked to /lib/* which is why it exhibits the same behavior, I'd guess. Well, that is not the chroot command found in chapter06/strippingagain.html :-/ That should corfim that the issue is in jhalfs, not related with the book or the GCC/Glibc planned updates. -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
El Viernes, 23 de Febrero de 2007 20:26, M.Canales.es escribió: That should corfim that the issue is in jhalfs, not related with the book or the GCC/Glibc planned updates. And fixed. After knowing what was causing the segfault in both a full build and later in my partial build tests, the fix was terribly obvious :-/ -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Testing required: SVN build segfaults on stripping chapter6
M.Canales.es wrote: El Viernes, 23 de Febrero de 2007 20:26, M.Canales.es escribió: That should corfim that the issue is in jhalfs, not related with the book or the GCC/Glibc planned updates. And fixed. After knowing what was causing the segfault in both a full build and later in my partial build tests, the fix was terribly obvious :-/ Manuel, I've only been following this casually. What was different about the chroot command that would cause a segfault? -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
