Xen 4.3 Release Notes
Xen 4.3.0 was released on July 9, 2013.
New Feature Highlights
- Early support for ARM 32- (ARMv7) and 64-bit (ARMv8) architectures
- qemu-upstream is now the default for VMs not using stub domains. You can change this by adding
device_model_version = "qemu-xen-traditional"to your config file.
- openvswitch hot-plug script support. See the Open vSwitch section of Xen Networking for more details.
- NUMA affinity for the scheduler
- xl can now accept several USB devices, rather than only one, using the following syntax:
usbdevice=["device1","device2",...]The old format is still supported.
- XSM improvements. XSM can now override all IS_PRIV checks in the hypervisor.
- As always, a number of stability, performance, and security enhancements "under the hood".
A more complete list of features can be found on the Xen 4.3 Feature List
The default installation prefix has been changed to /usr/local. Previously the default installation prefix was /usr To restore the previous behaviour, specify --prefix=/usr when running ./configure, e.g.:
$ ./configure --prefix=/usr
If you choose to accept the new default then be sure to cleanup any residual bits of the old installation in the old prefix, especially libraries and tools which you might pick up by mistake.
Due to this change it may also now be necessary to run /sbin/ldconfig as root after installing Xen.
Note also that Xen now requires gcc v4.1 or greater.
- Support for x86 32-bit and ia64 (itanium) hypervisors has been removed. 32bit VMs are still supported on the x86_64 64bit hypervisor.
- qemu-upstream does not yet work with stub domains. Stub domains will still be created with qemu-traditional.
- If a mac address is not specified in the config file, a new random one may be assigned on migrate or reboot; this can cause networking issues for some guests. The work-around for now is to specify a mac address.
- Passing through some PCI devices, particularly video cards with large amounts of memory, may have some issues with the default qemu-xen. If you experience problems, try using qemu-traditional instead.