Actually it is "fixed" with recent Linux-VServer + 2.6.26+ versions. I still don't see this as a vserver bug. The bug is in NPTL / glibc, as stated above and described in the official documentation. There were lots of changes to clone(2) and the namespace handling in between vanilla kernel 2.6.19 and 2.6.26+. The more recent Linux-VServer kernel patches have adapted and don't trigger this specific NPTL bug anymore. It has nothing to do with util-vserver in the userspace.
Replacing the buggy caching getpid(2) wrapper with the direct syscall looks like the only way to guarantee SysV-Init's behaviour (which uses getpid(2) but is not linked against NPTL). Who knows what else will be affected. However, I'm not sure how that would affect portability of upstart. I'm fine with patching upstart for now and won't invest more time to trace down the circumstances in the old (current stable) patch triggering the NPTL bug. What about getting this fixed in glibc? Still no fix after the whole LKML conversation (see #5 and notice the date)? -- upstart incompatible with linux-vserver https://bugs.launchpad.net/bugs/482292 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
