I have few questions, does anyone has success running Xen or KVM on board?
I have opened below topic but didnt have answer yet. I was expecting at least I will see LPAE in CPU futures. Is that because of my kernel version (3.10) i see in Linus repo most of arm & xen support came after the version 3.11. Do you know l4t 19.3 have newer kernel?
In ubuntu trusty repo there is already KVM port but what I see in kvm-arm documentations most of cortex-A15 proof of concepts are done with a recompiled kernel for KVM. I believe U-boot support will be helpful to handle Xen. I will appreciate if anyone can guide me here to setup virtualization enviroment on right way.
The L4T R19.2 (shipping L4T) and R19.3 use version 3.10.24 as the base version kernel. There are some configuration differences, and firmware which are not part of the kernel.
Flashing to R19.3 gives you the option of u-boot (which I highly recommend), but you must specify this or else you get fastboot. Pretty much everything else is a kernel compile with your desired configuration (which is why I recomment u-boot…no need to flash a new kernel, just add zImage to /boot and edit the config to try it out). Kernel configure, compile, copy, edit extlinux.conf, and try it out. Make sure you have a serial console to catch the boot prompt with. Run command “boot” if you stop boot too soon (before kernel pick).
Thank you linuxdev I was hoping someone has already tried. Me and my friend is also trying to port kvm and xen. If I will have success I will just drop here some info but for any case feedbacks are welcome as I really need it:)
I also want to run kvm on jetson.
There is my idea.
L4T R19.3 is basic on linux kernel 3.10.24 and kernel 3.10.24 have LPAE and kvm can be choiced in meneconfig. I can build a linux linux kernel include LPAE and KVM.
Build a u-boot which can let cpu start at HYP mode.
I am porting KVM(Kernel-based Virtual Machine) on Tegra K1. Because there is some necessary address(the address where secondary cores jump to) which is board-dependent, I cannot port KVM. Could nVidia give me some help to let me get the address? (ref [U-Boot] [PATCH 4/6] ARM: add SMP support for non-secure switch )
I bought nVidia Tegra K1 and I’d like to run virtual machine on Tegra K1. According to my knowledge on ARM virtualization extension, if an user would like to boot VM with hardware virtualization support, Hyp mode should be opened. Because Hyp mode only can be opened when booting, I have to make sure that TK1’s U-boot do have switched on Hyp mode. Although TK1’s U-boot do have code pieces for switching on Hyp mode, but it’s not compilable. Because there are an critical board-dependent address (ref [U-Boot] [PATCH 4/6] ARM: add SMP support for non-secure switch ).
As a result, could nVidia help me to provide the address to let me port KVM on Tegra K1?
NVIDIA doesn’t currently plan on supporting virtualization officially, but there is always a possibility that someone in the community could eventually support it such as by looking at upstream U-Boot + Linux kernel with the PSCI implementation, discussed at [url]The U-Boot April 2015 Archive by thread in the entry:
[U-Boot] [PATCH v7 00/17] Add PSCI support for Jetson TK1/Tegra124 + CNTFRQ fix by Jan Kiszka