Messages in this thread | | | From | Kechen Lu <> | Subject | RE: [RFC PATCH v6 6/6] KVM: selftests: Add tests for VM and vCPU cap KVM_CAP_X86_DISABLE_EXITS | Date | Thu, 2 Feb 2023 20:17:04 +0000 |
| |
Hi Zhi,
> -----Original Message----- > From: Zhi Wang <zhi.wang.linux@gmail.com> > Sent: Thursday, February 2, 2023 7:09 AM > To: Kechen Lu <kechenl@nvidia.com> > Cc: kvm@vger.kernel.org; seanjc@google.com; pbonzini@redhat.com; > chao.gao@intel.com; shaoqin.huang@intel.com; vkuznets@redhat.com; > linux-kernel@vger.kernel.org > Subject: Re: [RFC PATCH v6 6/6] KVM: selftests: Add tests for VM and vCPU > cap KVM_CAP_X86_DISABLE_EXITS > > External email: Use caution opening links or attachments > > > On Sat, 21 Jan 2023 02:07:38 +0000 > Kechen Lu <kechenl@nvidia.com> wrote: > > It works on my box now. LGTM. I was curious if there is any other userspace > application using this? It would be interesting to see their strategies. > Thanks for testing this! Appreciated.
For userspace applications, we had a POC with the changes in QEMU, to do the static per-vCPU HLT disabling, depending on the users to have the flexibility run guest workloads on these none-hlt-exits vCPUs instead of disabling hlt exits for whole VM, we saw 3%+ performance benefits. Furthermore, we had the dynamically toggled hlt exits strategy applied on QEMU, which depending on the guest stats and heuristics to do the hlt exits toggling.
BR, Kechen
| |