eaacc22f2d
Reduce the blocked duration in the hypervisor when destroying a VM by splitting a single hypercall into multiple calls. Previously, the hypervisor could be blocked for an extended period when destroying a VM, as the entire process was handled in a single hypercall. This could lead to performance degradation because the scheduler does not have chances to schedule other tasks when the hypercall is processing by the hypervisor. By splitting the destruction process into multiple smaller hypercalls, significantly reduce the blocked duration in the hypervisor. Additionally, making the amount of each call adjustable provides flexibility to optimize the process based on different workloads and system configurations. Signed-off-by: Jerry Wang <ze-yu.wang@mediatek.com> Signed-off-by: Liju Chen <liju-clr.chen@mediatek.com> Change-Id: I0b22854224de7d248d6a2558ff3ed5c02c735306 Bug: 379230831 Link: https://lore.kernel.org/lkml/20241114100802.4116-26-liju-clr.chen@mediatek.com/