At first, serial console put us behind the schedule. As it was very unusual problem due to malfunctioned serial cable. And now some another bug has cost us some very precious time. The anomaly started while I was trying to boot Xen on my system, I found Xen stuck in an infinite loop of the line:
(XEN) traps.c:3290: GPF (0000): ffff82d0801af585 -> ffff82d080240e5a
(full log here)
After inspection it turned out to be dom0 kernel bug. We Located the hex-address by using
$ addr2line -e xen-syms ffff82d0801af585
And it returned ‘xen/arch/x86/xstate.c:387’ which is position for XSAVE in the file. Adding ‘xsave=0’ made xen boot just fine. This is further evidence that the error was caused by XSAVE. This bug is also reported by another Xen user using Skylake cpu. So, this bug can be due to Xen’s incompatibility with Skylake cpu (for now).
Using the workaround (add ‘xsave=0’ to xen line parameter while booting) I am able to run on Skylake cpu. Work is in progress to patch this bug. Let’s hope that it’ll be out soon.
Detailed conversation at Xen-devel mailing list