Trying to boot from a xen dom0 back into a xen dom0 via kexec fails with invalid argument. This happens if you try to load just the linux kernel or if you try to load the xen hypervisor.
I have set the xen use flag on kexec-tools and I have tried 2.0.14 and 2.0.16 with no success.
Steps to Reproduce:
1. kexec -l -t multiboot-x86 /boot/xen.gz --command-line="/xen.gz dom0_mem=48G,max:48G com1=115200,8n1 console=com1" --module="/boot/vmlinuz placeholder root=ZFS=pool/root ro console=hvc0 console=tty0" --module="/boot/initrd.img"
1. kexec -l /boot/vmlinuz --reuse-cmdline --initrd=/boot/initramfs.img
kexec_load failed: Invalid argument
Followed by other output about memory etc. Kernel is not loaded and a reboot goes all the way to the bios.
kexec should load new kernel making it ready for quick reboot.
Scouring the web for information on xen support in kexec implies it should work, but I have not been able to locate a central kexec group to ask about this issue. As such, I'm kind of assuming it is a gentoo issue, though it may really be an upstream kexec issue. If that is the case, a pointer where to report upstream would be helpful.
Also, the init.d file for kexec-tools does not really support xen. If someone wanted to use it for a xen system, they would need to roll their own.