[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180709090218.15342-5-Tianyu.Lan@microsoft.com>
Date: Mon, 9 Jul 2018 09:02:54 +0000
From: Tianyu Lan <Tianyu.Lan@...rosoft.com>
To: unlisted-recipients:; (no To-header on input)
CC: Tianyu Lan <Tianyu.Lan@...rosoft.com>,
KY Srinivasan <kys@...rosoft.com>,
Haiyang Zhang <haiyangz@...rosoft.com>,
Stephen Hemminger <sthemmin@...rosoft.com>,
"tglx@...utronix.de" <tglx@...utronix.de>,
"mingo@...hat.com" <mingo@...hat.com>,
"hpa@...or.com" <hpa@...or.com>, "x86@...nel.org" <x86@...nel.org>,
"pbonzini@...hat.com" <pbonzini@...hat.com>,
"rkrcmar@...hat.com" <rkrcmar@...hat.com>,
"devel@...uxdriverproject.org" <devel@...uxdriverproject.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"kvm@...r.kernel.org" <kvm@...r.kernel.org>,
"Michael Kelley (EOSG)" <Michael.H.Kelley@...rosoft.com>,
"vkuznets@...hat.com" <vkuznets@...hat.com>
Subject: [PATCH V2 4/5] KVM/x86: Add tlb_remote_flush callback support for vmx
Register tlb_remote_flush callback for vmx when hyperv capability of
nested guest mapping flush is detected. The interface can help to
reduce overhead when flush ept table among vcpus for nested VM. The
tradition way is to send IPIs to all affected vcpus and executes
INVEPT on each vcpus. It will trigger several vmexits for IPI
and INVEPT emulation. Hyper-V provides such hypercall to do
flush for all vcpus.
Signed-off-by: Lan Tianyu <Tianyu.Lan@...rosoft.com>
---
Change since v1:
Use ept_pointers_match to check condition of identical ept
table pointer and get ept pointer from struct vcpu_vmx->ept_pointer.
---
arch/x86/kvm/vmx.c | 27 ++++++++++++++++++++++++++-
1 file changed, 26 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 8142b2da430a..55fe14d1d4d4 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -4778,6 +4778,25 @@ static inline void __vmx_flush_tlb(struct kvm_vcpu *vcpu, int vpid,
}
}
+static int hv_remote_flush_tlb(struct kvm *kvm)
+{
+ int ret;
+
+ spin_lock(&to_kvm_vmx(kvm)->ept_pointer_lock);
+
+ if (!to_kvm_vmx(kvm)->ept_pointers_match) {
+ ret = -EFAULT;
+ goto out;
+ }
+
+ ret = hyperv_flush_guest_mapping(
+ to_vmx(kvm_get_vcpu(kvm, 0))->ept_pointer);
+
+out:
+ spin_unlock(&to_kvm_vmx(kvm)->ept_pointer_lock);
+ return ret;
+}
+
static void vmx_flush_tlb(struct kvm_vcpu *vcpu, bool invalidate_gpa)
{
__vmx_flush_tlb(vcpu, to_vmx(vcpu)->vpid, invalidate_gpa);
@@ -4968,7 +4987,7 @@ static void check_ept_pointer(struct kvm_vcpu *vcpu, u64 eptp)
u64 tmp_eptp = INVALID_PAGE;
int i;
- if (!kvm_x86_ops->tlb_remote_flush)
+ if (!kvm_x86_ops->hv_tlb_remote_flush)
return;
spin_lock(&to_kvm_vmx(kvm)->ept_pointer_lock);
@@ -7570,6 +7589,12 @@ static __init int hardware_setup(void)
if (enable_ept && !cpu_has_vmx_ept_2m_page())
kvm_disable_largepages();
+#if IS_ENABLED(CONFIG_HYPERV)
+ if (ms_hyperv.nested_features & HV_X64_NESTED_GUEST_MAPPING_FLUSH
+ && enable_ept)
+ kvm_x86_ops->hv_tlb_remote_flush = hv_remote_flush_tlb;
+#endif
+
if (!cpu_has_vmx_ple()) {
ple_gap = 0;
ple_window = 0;
--
2.14.3
Powered by blists - more mailing lists