Re: [PATCH] RISC-V: KVM: Add kvm-riscv.wfi_trap_policy to control VS-mode WFI trapping

["Yuhang.Chen" <yhchen312@gmail.com>]

Hi Anup,

Thanks for the review.

On Tue, 28 Jul 2026, Anup Patel <anup@brainfault.org> wrote:
> This would mean that for the "notrap" policy the VCPU would never
> sleep and increase host CPU utilization even when VCPU is idle.
> Is this expected behaviour ?

No, that was not intended. With "notrap", HSTATUS.VTW is cleared
unconditionally, so a VS-mode WFI never traps and the vCPU never reaches
kvm_vcpu_halt(); since WFI is only a hint, the vCPU thread can keep
spinning even while the guest is idle - the high host CPU you point out.

v2 drops "notrap" and replaces it with "auto": HSTATUS.VTW is cleared
only when the vCPU is the sole runnable task on its CPU
(single_task_running()), otherwise WFI traps and KVM blocks the vCPU
through kvm_vcpu_halt() as before. The default stays "trap", so there
is no regression, and the policy is re-evaluated on each vcpu_load() so
a vCPU that stops being the sole runnable task switches back to
trapping.

So to answer your question: the vCPU no longer stays busy at the
expense of other tasks - it either blocks through kvm_vcpu_halt(), or
runs WFI natively only when nothing else needs the CPU. On QEMU TCG,
host CPU stays around 5% while the vCPU is the sole task and rises back
to the trap level as soon as a competitor appears.

I have posted v2 as a new thread:

  [PATCH v2] RISC-V: KVM: Add kvm-riscv.wfi_trap_policy to control
  VS-mode WFI trapping

Regards,
Yuhang