* [PATCH v2] KVM: APIC: avoid instruction emulation for EOI writes
@ 2011-08-30 1:15 Tian, Kevin
2011-08-30 10:57 ` Avi Kivity
0 siblings, 1 reply; 3+ messages in thread
From: Tian, Kevin @ 2011-08-30 1:15 UTC (permalink / raw)
To: kvm@vger.kernel.org
Cc: Nakajima, Jun, Dong, Eddie, Marcelo Tosatti, Avi Kivity,
Jan Kiszka, Sasha Levin
[-- Attachment #1: Type: text/plain, Size: 4341 bytes --]
v2 changes:
define exit qualification fields for APIC-Access in vmx.h
use apic_reg_write instead of apic_set_eoi, to avoid breaking tracing
add fasteoi option to allow disabling this acceleration
----
commit 2a66a12cb6928c962d24907e6d39b6eb9ac94b4b
Author: Kevin Tian <kevin.tian@intel.com>
Date: Mon Aug 29 13:08:28 2011 +0800
KVM: APIC: avoid instruction emulation for EOI writes
Instruction emulation for EOI writes can be skipped, since sane
guest simply uses MOV instead of string operations. This is a nice
improvement when guest doesn't support x2apic or hyper-V EOI
support.
a single VM bandwidth is observed with ~8% bandwidth improvement
(7.4Gbps->8Gbps), by saving ~5% cycles from EOI emulation.
Signed-off-by: Kevin Tian <kevin.tian@intel.com>
<Based on earlier work from>:
Signed-off-by: Eddie Dong <eddie.dong@intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
diff --git a/arch/x86/include/asm/vmx.h b/arch/x86/include/asm/vmx.h
index 2caf290..31f180c 100644
--- a/arch/x86/include/asm/vmx.h
+++ b/arch/x86/include/asm/vmx.h
@@ -350,6 +350,18 @@ enum vmcs_field {
#define DEBUG_REG_ACCESS_REG(eq) (((eq) >> 8) & 0xf) /* 11:8, general purpose reg. */
+/*
+ * Exit Qualifications for APIC-Access
+ */
+#define APIC_ACCESS_OFFSET 0xfff /* 11:0, offset within the APIC page */
+#define APIC_ACCESS_TYPE 0xf000 /* 15:12, access type */
+#define TYPE_LINEAR_APIC_INST_READ (0 << 12)
+#define TYPE_LINEAR_APIC_INST_WRITE (1 << 12)
+#define TYPE_LINEAR_APIC_INST_FETCH (2 << 12)
+#define TYPE_LINEAR_APIC_EVENT (3 << 12)
+#define TYPE_PHYSICAL_APIC_EVENT (10 << 12)
+#define TYPE_PHYSICAL_APIC_INST (15 << 12)
+
/* segment AR */
#define SEGMENT_AR_L_MASK (1 << 13)
diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index 57dcbd4..52645f2 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -864,6 +864,15 @@ static int apic_mmio_write(struct kvm_io_device *this,
return 0;
}
+void kvm_lapic_set_eoi(struct kvm_vcpu *vcpu)
+{
+ struct kvm_lapic *apic = vcpu->arch.apic;
+
+ if (apic)
+ apic_reg_write(vcpu->arch.apic, APIC_EOI, 0);
+}
+EXPORT_SYMBOL_GPL(kvm_lapic_set_eoi);
+
void kvm_free_lapic(struct kvm_vcpu *vcpu)
{
if (!vcpu->arch.apic)
diff --git a/arch/x86/kvm/lapic.h b/arch/x86/kvm/lapic.h
index 52c9e6b..8287243 100644
--- a/arch/x86/kvm/lapic.h
+++ b/arch/x86/kvm/lapic.h
@@ -26,6 +26,7 @@ int kvm_get_apic_interrupt(struct kvm_vcpu *vcpu);
void kvm_lapic_reset(struct kvm_vcpu *vcpu);
u64 kvm_lapic_get_cr8(struct kvm_vcpu *vcpu);
void kvm_lapic_set_tpr(struct kvm_vcpu *vcpu, unsigned long cr8);
+void kvm_lapic_set_eoi(struct kvm_vcpu *vcpu);
void kvm_lapic_set_base(struct kvm_vcpu *vcpu, u64 value);
u64 kvm_lapic_get_base(struct kvm_vcpu *vcpu);
void kvm_apic_set_version(struct kvm_vcpu *vcpu);
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 5e8d411..9d0364b 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -71,6 +71,9 @@ module_param(vmm_exclusive, bool, S_IRUGO);
static int __read_mostly yield_on_hlt = 1;
module_param(yield_on_hlt, bool, S_IRUGO);
+static int __read_mostly fasteoi = 1;
+module_param(fasteoi, bool, S_IRUGO);
+
/*
* If nested=1, nested virtualization is supported, i.e., guests may use
* VMX and be a hypervisor for its own guests. If nested=0, guests may not
@@ -4540,6 +4543,24 @@ static int handle_xsetbv(struct kvm_vcpu *vcpu)
static int handle_apic_access(struct kvm_vcpu *vcpu)
{
+ if (likely(fasteoi)) {
+ unsigned long exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
+ int access_type, offset;
+
+ access_type = exit_qualification & APIC_ACCESS_TYPE;
+ offset = exit_qualification & APIC_ACCESS_OFFSET;
+ /*
+ * Sane guest uses MOV to write EOI, with written value
+ * not cared. So make a short-circuit here by avoiding
+ * heavy instruction emulation.
+ */
+ if ((access_type == TYPE_LINEAR_APIC_INST_WRITE) &&
+ (offset == APIC_EOI)) {
+ kvm_lapic_set_eoi(vcpu);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ }
+ }
return emulate_instruction(vcpu, 0) == EMULATE_DONE;
}
Thanks
Kevin
[-- Attachment #2: 20110829_kvm_eoi_opt.patch --]
[-- Type: application/octet-stream, Size: 3999 bytes --]
commit 2a66a12cb6928c962d24907e6d39b6eb9ac94b4b
Author: Kevin Tian <kevin.tian@intel.com>
Date: Mon Aug 29 13:08:28 2011 +0800
KVM: APIC: avoid instruction emulation for EOI writes
Instruction emulation for EOI writes can be skipped, since sane
guest simply uses MOV instead of string operations. This is a nice
improvement when guest doesn't support x2apic or hyper-V EOI
support.
a single VM bandwidth is observed with ~8% bandwidth improvement
(7.4Gbps->8Gbps), by saving ~5% cycles from EOI emulation.
Signed-off-by: Kevin Tian <kevin.tian@intel.com>
<Based on earlier work from>:
Signed-off-by: Eddie Dong <eddie.dong@intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
diff --git a/arch/x86/include/asm/vmx.h b/arch/x86/include/asm/vmx.h
index 2caf290..31f180c 100644
--- a/arch/x86/include/asm/vmx.h
+++ b/arch/x86/include/asm/vmx.h
@@ -350,6 +350,18 @@ enum vmcs_field {
#define DEBUG_REG_ACCESS_REG(eq) (((eq) >> 8) & 0xf) /* 11:8, general purpose reg. */
+/*
+ * Exit Qualifications for APIC-Access
+ */
+#define APIC_ACCESS_OFFSET 0xfff /* 11:0, offset within the APIC page */
+#define APIC_ACCESS_TYPE 0xf000 /* 15:12, access type */
+#define TYPE_LINEAR_APIC_INST_READ (0 << 12)
+#define TYPE_LINEAR_APIC_INST_WRITE (1 << 12)
+#define TYPE_LINEAR_APIC_INST_FETCH (2 << 12)
+#define TYPE_LINEAR_APIC_EVENT (3 << 12)
+#define TYPE_PHYSICAL_APIC_EVENT (10 << 12)
+#define TYPE_PHYSICAL_APIC_INST (15 << 12)
+
/* segment AR */
#define SEGMENT_AR_L_MASK (1 << 13)
diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index 57dcbd4..52645f2 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -864,6 +864,15 @@ static int apic_mmio_write(struct kvm_io_device *this,
return 0;
}
+void kvm_lapic_set_eoi(struct kvm_vcpu *vcpu)
+{
+ struct kvm_lapic *apic = vcpu->arch.apic;
+
+ if (apic)
+ apic_reg_write(vcpu->arch.apic, APIC_EOI, 0);
+}
+EXPORT_SYMBOL_GPL(kvm_lapic_set_eoi);
+
void kvm_free_lapic(struct kvm_vcpu *vcpu)
{
if (!vcpu->arch.apic)
diff --git a/arch/x86/kvm/lapic.h b/arch/x86/kvm/lapic.h
index 52c9e6b..8287243 100644
--- a/arch/x86/kvm/lapic.h
+++ b/arch/x86/kvm/lapic.h
@@ -26,6 +26,7 @@ int kvm_get_apic_interrupt(struct kvm_vcpu *vcpu);
void kvm_lapic_reset(struct kvm_vcpu *vcpu);
u64 kvm_lapic_get_cr8(struct kvm_vcpu *vcpu);
void kvm_lapic_set_tpr(struct kvm_vcpu *vcpu, unsigned long cr8);
+void kvm_lapic_set_eoi(struct kvm_vcpu *vcpu);
void kvm_lapic_set_base(struct kvm_vcpu *vcpu, u64 value);
u64 kvm_lapic_get_base(struct kvm_vcpu *vcpu);
void kvm_apic_set_version(struct kvm_vcpu *vcpu);
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 5e8d411..9d0364b 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -71,6 +71,9 @@ module_param(vmm_exclusive, bool, S_IRUGO);
static int __read_mostly yield_on_hlt = 1;
module_param(yield_on_hlt, bool, S_IRUGO);
+static int __read_mostly fasteoi = 1;
+module_param(fasteoi, bool, S_IRUGO);
+
/*
* If nested=1, nested virtualization is supported, i.e., guests may use
* VMX and be a hypervisor for its own guests. If nested=0, guests may not
@@ -4540,6 +4543,24 @@ static int handle_xsetbv(struct kvm_vcpu *vcpu)
static int handle_apic_access(struct kvm_vcpu *vcpu)
{
+ if (likely(fasteoi)) {
+ unsigned long exit_qualification = vmcs_readl(EXIT_QUALIFICATION);
+ int access_type, offset;
+
+ access_type = exit_qualification & APIC_ACCESS_TYPE;
+ offset = exit_qualification & APIC_ACCESS_OFFSET;
+ /*
+ * Sane guest uses MOV to write EOI, with written value
+ * not cared. So make a short-circuit here by avoiding
+ * heavy instruction emulation.
+ */
+ if ((access_type == TYPE_LINEAR_APIC_INST_WRITE) &&
+ (offset == APIC_EOI)) {
+ kvm_lapic_set_eoi(vcpu);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ }
+ }
return emulate_instruction(vcpu, 0) == EMULATE_DONE;
}
^ permalink raw reply related [flat|nested] 3+ messages in thread* Re: [PATCH v2] KVM: APIC: avoid instruction emulation for EOI writes
2011-08-30 1:15 [PATCH v2] KVM: APIC: avoid instruction emulation for EOI writes Tian, Kevin
@ 2011-08-30 10:57 ` Avi Kivity
2011-08-31 0:12 ` Tian, Kevin
0 siblings, 1 reply; 3+ messages in thread
From: Avi Kivity @ 2011-08-30 10:57 UTC (permalink / raw)
To: Tian, Kevin
Cc: kvm@vger.kernel.org, Nakajima, Jun, Dong, Eddie, Marcelo Tosatti,
Jan Kiszka, Sasha Levin
On 08/30/2011 04:15 AM, Tian, Kevin wrote:
> v2 changes:
> define exit qualification fields for APIC-Access in vmx.h
> use apic_reg_write instead of apic_set_eoi, to avoid breaking tracing
> add fasteoi option to allow disabling this acceleration
>
> ----
>
> commit 2a66a12cb6928c962d24907e6d39b6eb9ac94b4b
> Author: Kevin Tian<kevin.tian@intel.com>
> Date: Mon Aug 29 13:08:28 2011 +0800
>
> KVM: APIC: avoid instruction emulation for EOI writes
>
> Instruction emulation for EOI writes can be skipped, since sane
> guest simply uses MOV instead of string operations. This is a nice
> improvement when guest doesn't support x2apic or hyper-V EOI
> support.
>
> a single VM bandwidth is observed with ~8% bandwidth improvement
> (7.4Gbps->8Gbps), by saving ~5% cycles from EOI emulation.
>
>
Thanks, applied. Please use git format-patch (and git send-email)
instead of git show in the future.
--
error compiling committee.c: too many arguments to function
^ permalink raw reply [flat|nested] 3+ messages in thread
* RE: [PATCH v2] KVM: APIC: avoid instruction emulation for EOI writes
2011-08-30 10:57 ` Avi Kivity
@ 2011-08-31 0:12 ` Tian, Kevin
0 siblings, 0 replies; 3+ messages in thread
From: Tian, Kevin @ 2011-08-31 0:12 UTC (permalink / raw)
To: Avi Kivity
Cc: kvm@vger.kernel.org, Nakajima, Jun, Dong, Eddie, Marcelo Tosatti,
Jan Kiszka, Sasha Levin
> From: Avi Kivity [mailto:avi@redhat.com]
> Sent: Tuesday, August 30, 2011 6:57 PM
>
> On 08/30/2011 04:15 AM, Tian, Kevin wrote:
> > v2 changes:
> > define exit qualification fields for APIC-Access in vmx.h
> > use apic_reg_write instead of apic_set_eoi, to avoid breaking tracing
> > add fasteoi option to allow disabling this acceleration
> >
> > ----
> >
> > commit 2a66a12cb6928c962d24907e6d39b6eb9ac94b4b
> > Author: Kevin Tian<kevin.tian@intel.com>
> > Date: Mon Aug 29 13:08:28 2011 +0800
> >
> > KVM: APIC: avoid instruction emulation for EOI writes
> >
> > Instruction emulation for EOI writes can be skipped, since sane
> > guest simply uses MOV instead of string operations. This is a nice
> > improvement when guest doesn't support x2apic or hyper-V EOI
> > support.
> >
> > a single VM bandwidth is observed with ~8% bandwidth improvement
> > (7.4Gbps->8Gbps), by saving ~5% cycles from EOI emulation.
> >
> >
>
> Thanks, applied. Please use git format-patch (and git send-email)
> instead of git show in the future.
>
OK, and sorry for the formatting issue.
Thanks
Kevin
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2011-08-31 0:12 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-08-30 1:15 [PATCH v2] KVM: APIC: avoid instruction emulation for EOI writes Tian, Kevin
2011-08-30 10:57 ` Avi Kivity
2011-08-31 0:12 ` Tian, Kevin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox