On Tue, Aug 18, 2009 at 09:21:22PM +0100, Daniel P. Berrange wrote: > On Tue, Aug 18, 2009 at 11:18:24PM +0300, Pasi K?rkk?inen wrote: > > On Tue, Aug 18, 2009 at 01:18:23PM +0300, Pasi Kärkkäinen wrote: > > > On Tue, Aug 18, 2009 at 10:25:28AM +0100, Daniel P. Berrange wrote: > > > > On Tue, Aug 18, 2009 at 11:42:00AM +0300, Pasi K?rkk?inen wrote: > > > > > Hello. > > > > > > > > > > While continuing to test pv_ops dom0 kernels I also rebuilt > > > > > xen-3.4.1-1 src.rpm on F11. > > > > > > > > > > I'm able to virt-install new centos 5.3 domU, but it seems virt-viewer > > > > > doesn't work. It segfaults on startup. vncviewer works though, and I'm able > > > > > to complete the installation with vncviewer. > > > > > > > > > > http://pasik.reaktio.net/fedora/virt-viewer-crash.txt > > > > > > > > > > virt-viewer works when used with the default xen-3.3.1 (which is included in F11). > > > > > > > > strace logs aren't much help with crashes. Try to run it under valgrind and > > > > capture the errors, or GDB and get a stack trace > > > > > > > > > > Yeah, I was going to try gdb next.. just didn't have time yet. Thanks. > > > > > > > (gdb) run test2vm > > Starting program: /usr/bin/virt-viewer test2vm > > [Thread debugging using libthread_db enabled] > > [New Thread 0xb7c64b70 (LWP 2802)] > > > > Program received signal SIGSEGV, Segmentation fault. > > strcmp () at ../sysdeps/i386/i686/strcmp.S:40 > > 40 ../sysdeps/i386/i686/strcmp.S: No such file or directory. > > in ../sysdeps/i386/i686/strcmp.S > > Current language: auto; currently asm > > (gdb) bt > > #0 strcmp () at ../sysdeps/i386/i686/strcmp.S:40 > > #1 0xb7f237d2 in virEnumFromString (types=0xb7fb7074, ntypes=2, type=0x0) at util.c:1481 > > #2 0xb7f3ce4f in virDomainGraphicsTypeFromString (type=0x0) at domain_conf.c:150 > > #3 0xb7f77490 in xenDaemonParseSxprGraphicsNew (conn=<value optimized out>, def=<value optimized out>, > > root=0x8089278) at xend_internal.c:2085 > > #4 0xb7f79c44 in xenDaemonParseSxpr (conn=<value optimized out>, root=<value optimized out>, > > xendConfigVersion=4, cpus=0x0) at xend_internal.c:2452 > > #5 0xb7f7d079 in xenDaemonDomainFetch (conn=0x8079ca0, domid=1, name=0x808b190 "test2vm", cpus=0x0) > > at xend_internal.c:3372 > > #6 0xb7f7d180 in xenDaemonDomainDumpXML (domain=0x808b168, flags=0, cpus=0x0) at xend_internal.c:3416 > > #7 0xb7f74394 in xenUnifiedDomainDumpXML (dom=0x808b168, flags=0) at xen_unified.c:1034 > > #8 0xb7f357cf in virDomainGetXMLDesc (domain=0x808b168, flags=0) at libvirt.c:2621 > > #9 0x0804c116 in viewer_extract_vnc_graphics (port=<value optimized out>, dom=<value optimized out>) > > at main.c:709 > > #10 viewer_start (port=<value optimized out>, dom=<value optimized out>) at main.c:906 > > #11 0x0804ca07 in main (argc=2, argv=0xbffff1d4) at main.c:1037 > > (gdb) > > > > So something has changed in Xen 3.4.x and now virEnumFromString() fails? > > Yeah there will be something unexpected in the SEXPR from Xen, which > then triggers a latent bug in libvirt code not doing proper sanity > checking, resulting in the crash. Can you file a bug against libvirt, > and include the output of 'xm list --long $GUESTNAME' and xenstore-ls > as attachments. It ought to be trivial to fix. > Yeah I can do that. Wondering if it's this bug: http://www.mail-archive.com/libvir-list@xxxxxxxxxx/msg14082.html At least that patch adds check for type == 0, and that's what I have here.. -- Pasi -- Fedora-xen mailing list Fedora-xen@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-xen
