KVM: arm64: sdev: Support virq bypass by INT/VSYNC command

virt inclusion
category: feature
bugzilla: https://gitee.com/openeuler/kernel/issues/I8K8HP


CVE: NA

--------------------------------

We already have the set_irqchip_state() callback which will issue
INT/VSYNC commands on the physical side when the handled IRQ is
forwarded to vcpu.  It was intended to be used to handle the guest
INT command targeting a VLPI.

Let's reuse this hack to set the virtio-pci function's VLPI pending,
instead of directly writing message data into GITS_TRANSLATER, which
should only be treated as the Message Address of PCI devices.

Signed-off-by: wanghaibin <wanghaibin.wang@huawei.com>
Signed-off-by: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Kunkun Jiang <jiangkunkun@huawei.com>
Signed-off-by: Dongxu Sun <sundongxu3@huawei.com>