* Gerrit Pape <[EMAIL PROTECTED]> [2008-12-11 04:30-0500]: > On Wed, Dec 10, 2008 at 04:11:14PM -0500, Micah Anderson wrote: > > As it turns out dietlibc-0.31 doesn't properly define the umount2 > > symbols on two architectures: alpha and ia64. This sadly results in a > > build regression for util-vserver, which used to build on these > > architectures, but is refusing to build now. This is holding back an > > important transition of the package into Lenny. In otherwords, if this > > package cannot be built on alpha/ia64, then it will not be usable for > > most cases in Lenny due to the previous version not functioning > > properly in two important respects. > > Hi, I'm surprised. Not that I question that there's possibly a bug, but > version 0.31-1 of dietlibc is in the archive since more than one year. > I wonder why the util-vserver package needs such changes that late in > the Debian release cycle.
Yeah, I was surprised too. However, it goes far up the chain... the newer kernels brought in some virtualization namespace changes, which only have begun to appear in the kernels that have now transitioned into Lenny. These new changes mean that util-vserver has to change the way it deals with unmounting and cleanup in the guest because the chroot barrier is being faded out in favor of the new namespaces and pivot_root. > Anyway, I'm sorry, I currently don't have the time to look at it or even > upload a new package, please NMU if you think that's the right thing. Ok, I've been recruiting testers on different arches and just have mips, arm and sparc left to test before we are certain that the change works right. thanks! micah
signature.asc
Description: Digital signature
