A virtual machine system can run multiple conventional operating systems (OSs) in a single real host computer. A virtual machine is a logical computer with almost the same architecture as the host, and may contain several logical processors. A hypervisor is a control program to control this virtual machine system. Traditionally, the hypervisor has to receive an I/O interrupt pending for a waiting logical processor, and to simulate the I/O interrupt, consequently incurring the simulation overhead of the I/O interrupt. To avoid this overhead we present a new method which introduces a new self-wait state different from the conventional wait state, presents a new instruction for the hypervisor to detect the I/O interrupts pending for the logical processors in the self-wait state, and dispatches those logical processors on the detection ahead of the ready queue. This new method has eliminated the simulation overhead of those I/O interrupts, and enhanced the system performance to the near native.

This paper presents methods for increasing the efficiency of operating systems in plural virtual machines to a near-native performance level. The proposed direct execution methods support direct I/O execution for plural virtual machines, that is, the V = R virtual machine, and the V = Resi virtual machines, including both I/O instruction issuances and I/O interrupts. All V = Resi virtual machines have an entirely resident memory and their real addresses are translated into those of the host simply by adding a constant, α(≠ 0), which constitutes a starting address given to each V = Resi virtual machine. The V = R virtual machine has almost the same memory attribute as the V = Resi VM except for α = 0. Only one V = R virtual machine can be present in all virtual machines. The experimental results obtained confirm that a near-native performance level, that is, a level exceeding 90 percent of native performance, can be realized for the V = R virtual machine as well as for the V = Resi virtual machines.