On Sunday 22 December 2002 06:53 pm, Gwenole Beauchesne wrote: > Le dimanche 22 d�cembre 2002, � 09:46 PM, Jerome Hugues a �crit : > > got the same problem doin a urpmi --auto-select on my cooker box (on a > > Pentium IV), here is a extract of traceback of urpmi doing the upgrade : > > Thanks for the report. However, I can't see in the log initscripts > update. Do you happen to already have 6.91-18mdk prior to urpmi > --auto-select'ing? If your system is still crashing, hmm try > /lib/ld-linux.so.2 and see whether you get already something (that's the > runtime loader). > > BTW, I can reliably crash on 9.0 + glibc 2.3.1 ld.so provided that I > don't expand LD_LIBRARY_PATH to the temporary directory holding its > libraries. This would mean you either don't have updated glibc libraries > (unlikely to me), or your ld.so.cache got corrupted somehow? > > Hmm, what is the contents of /etc/ld.so.conf? Could try to rm -f > /etc/ld.so.cache once new glibc is installed? Or from a 9.0 urpmi > initscripts first, remove ld.so.cache then DURING_INSTALL=1 urpmi glibc? > Hmm, I don't think that'd would change anything.
I have 2 installations of 9.0 on my system. I've got a little script setup to mirror the working one to the other for the time being ;) Anyway - seems ld.so.conf.rpmnew is empty, but ld.so.cache was there, I regenerated it using the -r option to generate a new cache where the new glibc is. Even after making a new cache the system is still dead. I also cannot chroot to the system where new glibc is (no suprize) but I was wondering if having a working system with access to the filesystem of the fscked one might help somehow to find the problem? This is what I get when I chroot after updating glibc [root@kato pub]# LD_DEBUG=libs chroot /mnt/disk 01595: find library=libc.so.6; searching 01595: search cache=/etc/ld.so.cache 01595: trying file=/lib/i686/libc.so.6 01595: 01595: 01595: calling init: /lib/i686/libc.so.6 01595: 01595: 01595: initialize program: chroot 01595: 01595: 01595: transferring control: chroot 01595: Segmentation fault
