Bugzilla – Bug 156347
Distro upgrade: rpm segfaults on rpm -U xorg-x11
Last modified: 2006-03-17 18:12:32 UTC
On upgrading from 9.1 to 10.1, YaSTY fails on the package xorg-x11 (x86-64). It does not display anything useful except for rpm failing. On the commandline, I see that rpm -U segfaults! rpm --rebuilddb does not help. rpm -e XFree86 does fail, rpm -i xorg-x11 correctly displays the file conflicts. gdb backtrace of rpm -e: #0 0x2b2414fb43c7 in rpmdsNewDNEVR () from /usr/lib64/librpm-4.4.so #1 0x2b2414fb51d0 in rpmdsNext () #2 0x2b2414fb5a18 in rpmdsNew () #3 0x2b2414fc4307 in rpmteNew () #4 0x2b2414f982e4 in cpioHeaderWrite () #5 0x2b2414fbb5d9 in rpmErase ()
rpm -ql and -qi succeed and show nothing unusual. (Details on request.) Using the rpm (4.1.1) from the installed system, things work: rpm -e correctly displays the dependency conflict rather than segfaulting. I have saved /var/lib/rpm and /var/adm/backup/rpmdb for later reproduction of this bug.
rpm --rebuilddb inside the chroot does not help. After doing rpm -e --nodeps inside the chroot, I can continue the system upgrade.
hwconfig exposes the same problem as XFree86.
(Correction) hwinfo exposes the same problem as XFree86.
And aspell ... As rpm has been updated in the chroot meanwhile, I had to copy over the old rpm and rpmlibs from a SLES9 to make the manual rpm -e inside the chroot succeed.
And yast2-network.
RPM segfaults make me nervous, raising to critical.
Kurt, which host is this? Can I have the saved database?
Hi mls, I just have the same problem here during the update of my workstation. Not with X11 but with qt3. Update is from SUSE Linux 9.1 to 10.1 Beta7. Shall I file a separate bug report? Do you want to come step by and take a look at it? Raising this to blocker, as any manual workaround like Kurt did with copying an old rpm and such is not acceptable from a useability point of view. My workstation is i386 BTW, so this rules out a x86_64 specific problem.
My machine is lakatos.suse.de and currently in that state. As I booted from the Beta7 mini-ISO to do this update the network is up but ssh is not up and running. Tell me how to bring this up and I will do it or come and stop by to debug it locally.
snwint hinted to use "inst_setup ssh" to bring up ssh in the instsys, but all I see is a "starting hald..." and then it hangs. So you may want to come and debug locally ...
My fault, not "inst_setup ssh" but "inst_setup_ssh" does the trick. Now ssh up and ready for remote debugging.
mls, please tell me when you are done with debugging this as lakatos is my workstation and I am currently blocked with this (I can work from my laptop but this is less convienient of course).
Michael, host is f52, backup of rpm database in /var/lib/rpm,backup.
Fixed in STABLE (hopefully)