KVM x86 Enabling Nested Virtualization By Default, Other Virtualization Work
The initial round of Kernel-based Virtual Machine (KVM) updates has been sent in for the in-development Linux 4.20/5.0 kernel.
Notable on the KVM x86 front is that nested virtualization is enabled by default for running a KVM guest within a KVM guest. This support for allowing KVM guests to use VMX (VT-x) instructions is being enabled by default now that live migration support is squared away and other functionality that ensures for a stable user-space ABI. The capability has of nested KVM on x86_64 has been available for years but required the module parameter to enable the functionality.
Also on the nested KVM front, the POWER architecture code has nested HV KVM support for radix guests on POWER9 that leads to much better performance. Some of the other KVM changes this cycle include various ARM clean-ups, the POWER code adds a one-VM-per-core mode in order to avoid potential data leak scenario, there is also a POWER PCI pass-through optimization, initial AP crypto virtualization for s390, coalesced PIO accesses for x86, and Hyper-V IPI x86 hypercalls support.
A complete list of the latest KVM Linux changes can be found via this PR.
Notable on the KVM x86 front is that nested virtualization is enabled by default for running a KVM guest within a KVM guest. This support for allowing KVM guests to use VMX (VT-x) instructions is being enabled by default now that live migration support is squared away and other functionality that ensures for a stable user-space ABI. The capability has of nested KVM on x86_64 has been available for years but required the module parameter to enable the functionality.
Also on the nested KVM front, the POWER architecture code has nested HV KVM support for radix guests on POWER9 that leads to much better performance. Some of the other KVM changes this cycle include various ARM clean-ups, the POWER code adds a one-VM-per-core mode in order to avoid potential data leak scenario, there is also a POWER PCI pass-through optimization, initial AP crypto virtualization for s390, coalesced PIO accesses for x86, and Hyper-V IPI x86 hypercalls support.
A complete list of the latest KVM Linux changes can be found via this PR.
5 Comments