On Sat, 2012-01-28 at 14:56 +0100, Christoph Anton Mitterer wrote: > Hi Ben. > > On Sat, 2012-01-28 at 05:05 +0000, Ben Hutchings wrote: > > But they probably enabled 'Viridian compatibility mode' which makes > > XenServer look more like HyperV, probably to make Windows guests behave > > btter. Linux 3.1 now checks for Xen before HyperV, to avoid being > > fooled by this. That might be a problem if it causes the kernel to > > detect different emulated devices. (I don't know whether it could have > > this effect.) > I'll as them > > > > Which drivers are in the initramfs? Does > > /etc/initramfs-tools/initramfs.conf have MODULES=dep (which is > > vulnerable to configuration changes) or MODULES=most (which is safer)? > It is "dep", but that shouldn't matter here, should it? > Cause the detected modules are the same as in the (still booting) 3.0.x > kernel (I've compared the initramfs with lsinitramfs). It does matter if the kernel detects different devices due to interacting with the hypervisor in a different way. That could mean that a initramfs built while running Linux 3.0 has the wrong set of drivers for later versions. > And if I remember correctly, I even tried it with =most, when I first > stumbled across this when 3.1 came out. Can you try again, in case you remember wrongly? I assume it is not possible for you to interact with the shell that the initramfs runs after failing to mount the root filesystem? Ben. -- Ben Hutchings If you seem to know what you are doing, you'll be given more to do.
Attachment:
signature.asc
Description: This is a digitally signed message part