From: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
Cc: linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: [PATCH 10/18] KVM: VMX: Don't switch 64-bit msrs for 32-bit guests
Date: Thu, 26 Apr 2007 12:22:10 +0300 [thread overview]
Message-ID: <11775793383091-git-send-email-avi@qumranet.com> (raw)
In-Reply-To: <11775793382353-git-send-email-avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Some msrs are only used by x86_64 instructions, and are therefore
not needed when the guest is legacy mode. By not bothering to switch
them, we reduce vmexit latency by 2400 cycles (from about 8800) when
running a 32-bt guest on a 64-bit host.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
---
drivers/kvm/vmx.c | 58 ++++++++++++++++++++++++++++++++++++++--------------
1 files changed, 42 insertions(+), 16 deletions(-)
diff --git a/drivers/kvm/vmx.c b/drivers/kvm/vmx.c
index 3745e6c..6270df5 100644
--- a/drivers/kvm/vmx.c
+++ b/drivers/kvm/vmx.c
@@ -80,6 +80,9 @@ static const u32 vmx_msr_index[] = {
#ifdef CONFIG_X86_64
static unsigned msr_offset_kernel_gs_base;
+#define NR_64BIT_MSRS 4
+#else
+#define NR_64BIT_MSRS 0
#endif
static inline int is_page_fault(u32 intr_info)
@@ -301,6 +304,32 @@ static void vmx_inject_gp(struct kvm_vcpu *vcpu, unsigned error_code)
}
/*
+ * Set up the vmcs to automatically save and restore system
+ * msrs. Don't touch the 64-bit msrs if the guest is in legacy
+ * mode, as fiddling with msrs is very expensive.
+ */
+static void setup_msrs(struct kvm_vcpu *vcpu)
+{
+ int nr_skip, nr_good_msrs;
+
+ if (is_long_mode(vcpu))
+ nr_skip = NR_BAD_MSRS;
+ else
+ nr_skip = NR_64BIT_MSRS;
+ nr_good_msrs = vcpu->nmsrs - nr_skip;
+
+ vmcs_writel(VM_ENTRY_MSR_LOAD_ADDR,
+ virt_to_phys(vcpu->guest_msrs + nr_skip));
+ vmcs_writel(VM_EXIT_MSR_STORE_ADDR,
+ virt_to_phys(vcpu->guest_msrs + nr_skip));
+ vmcs_writel(VM_EXIT_MSR_LOAD_ADDR,
+ virt_to_phys(vcpu->host_msrs + nr_skip));
+ vmcs_write32(VM_EXIT_MSR_STORE_COUNT, nr_good_msrs); /* 22.2.2 */
+ vmcs_write32(VM_EXIT_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
+ vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
+}
+
+/*
* reads and returns guest's timestamp counter "register"
* guest_tsc = host_tsc + tsc_offset -- 21.3
*/
@@ -825,6 +854,7 @@ static void vmx_set_efer(struct kvm_vcpu *vcpu, u64 efer)
msr->data = efer & ~EFER_LME;
}
+ setup_msrs(vcpu);
}
#endif
@@ -988,7 +1018,6 @@ static int vmx_vcpu_setup(struct kvm_vcpu *vcpu)
struct descriptor_table dt;
int i;
int ret = 0;
- int nr_good_msrs;
extern asmlinkage void kvm_vmx_return(void);
if (!init_rmode_tss(vcpu->kvm)) {
@@ -1140,19 +1169,10 @@ static int vmx_vcpu_setup(struct kvm_vcpu *vcpu)
++vcpu->nmsrs;
}
- nr_good_msrs = vcpu->nmsrs - NR_BAD_MSRS;
- vmcs_writel(VM_ENTRY_MSR_LOAD_ADDR,
- virt_to_phys(vcpu->guest_msrs + NR_BAD_MSRS));
- vmcs_writel(VM_EXIT_MSR_STORE_ADDR,
- virt_to_phys(vcpu->guest_msrs + NR_BAD_MSRS));
- vmcs_writel(VM_EXIT_MSR_LOAD_ADDR,
- virt_to_phys(vcpu->host_msrs + NR_BAD_MSRS));
+ setup_msrs(vcpu);
+
vmcs_write32_fixedbits(MSR_IA32_VMX_EXIT_CTLS, VM_EXIT_CONTROLS,
(HOST_IS_64 << 9)); /* 22.2,1, 20.7.1 */
- vmcs_write32(VM_EXIT_MSR_STORE_COUNT, nr_good_msrs); /* 22.2.2 */
- vmcs_write32(VM_EXIT_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
- vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
-
/* 22.2.1, 20.8.1 */
vmcs_write32_fixedbits(MSR_IA32_VMX_ENTRY_CTLS,
@@ -1769,9 +1789,11 @@ again:
fx_restore(vcpu->guest_fx_image);
#ifdef CONFIG_X86_64
- save_msrs(vcpu->host_msrs + msr_offset_kernel_gs_base, 1);
+ if (is_long_mode(vcpu)) {
+ save_msrs(vcpu->host_msrs + msr_offset_kernel_gs_base, 1);
+ load_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
+ }
#endif
- load_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
asm (
/* Store host registers */
@@ -1915,8 +1937,12 @@ again:
}
++kvm_stat.exits;
- save_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
- load_msrs(vcpu->host_msrs, NR_BAD_MSRS);
+#ifdef CONFIG_X86_64
+ if (is_long_mode(vcpu)) {
+ save_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
+ load_msrs(vcpu->host_msrs, NR_BAD_MSRS);
+ }
+#endif
fx_save(vcpu->guest_fx_image);
fx_restore(vcpu->host_fx_image);
--
1.5.0.6
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
next prev parent reply other threads:[~2007-04-26 9:22 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-04-26 9:22 [PATCH 00/18] KVM updates for 2.6.22 Avi Kivity
2007-04-26 9:22 ` [PATCH 02/18] KVM: Fix overflow bug in overflow detection code Avi Kivity
2007-04-26 9:22 ` [PATCH 03/18] KVM: Initialize cr0 to indicate an fpu is present Avi Kivity
2007-04-26 9:22 ` [PATCH 04/18] KVM: Handle partial pae pdptr Avi Kivity
[not found] ` <11775793382353-git-send-email-avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
2007-04-26 9:22 ` [PATCH 01/18] KVM: Use kernel-standard types Avi Kivity
2007-04-26 9:22 ` [PATCH 05/18] KVM: Use slab caches to allocate mmu data structures Avi Kivity
2007-04-26 9:22 ` [PATCH 06/18] KVM: Retry sleeping allocation if atomic allocation fails Avi Kivity
2007-04-26 9:22 ` [PATCH 07/18] KVM: SVM: Report hardware exit reason to userspace instead of dmesg Avi Kivity
2007-04-26 9:22 ` [PATCH 08/18] KVM: Handle guest page faults when emulating mmio Avi Kivity
2007-04-26 9:22 ` [PATCH 09/18] KVM: VMX: Reduce unnecessary saving of host msrs Avi Kivity
2007-04-26 9:22 ` Avi Kivity [this message]
2007-04-26 9:22 ` [PATCH 11/18] KVM: Fold drivers/kvm/kvm_vmx.h into drivers/kvm/vmx.c Avi Kivity
2007-04-26 9:22 ` [PATCH 12/18] KVM: VMX: Only save/restore MSR_K6_STAR if necessary Avi Kivity
2007-04-26 9:22 ` [PATCH 13/18] KVM: MMU: Avoid heavy ASSERT at non debug mode Avi Kivity
2007-04-26 9:22 ` [PATCH 14/18] KVM: VMX: Avoid unnecessary vcpu_load()/vcpu_put() cycles Avi Kivity
2007-04-26 9:22 ` [PATCH 15/18] KVM: Per-vcpu statistics Avi Kivity
2007-04-26 9:22 ` [PATCH 16/18] KVM: Allow passing 64-bit values to the emulated read/write API Avi Kivity
2007-04-26 9:22 ` [PATCH 17/18] KVM: Lazy FPU support for SVM Avi Kivity
2007-04-26 9:22 ` [PATCH 18/18] KVM: Don't complain about cpu erratum AA15 Avi Kivity
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=11775793383091-git-send-email-avi@qumranet.com \
--to=avi-atkuwr5tajbwk0htik3j/w@public.gmane.org \
--cc=kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org \
--cc=linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox