The bulk of the MSR bitmap is either immutable, or can be copied from
the L1 bitmap. By initializing it at VMXON time, and copying the mutable
parts one long at a time on vmentry (rather than one bit), about 4000
clock cycles (30%) can be saved on a nested VMLAUNCH/VMRESUME.
The resulting for loop only has four iterations, so it is cheap enough
to reinitialize the MSR write bitmaps on every iteration, and it makes
the code simpler.
Suggested-by: Jim Mattson <[email protected]>
Signed-off-by: Paolo Bonzini <[email protected]>
---
arch/x86/kvm/vmx.c | 57 ++++++++++++++++++++++++++++--------------------------
1 file changed, 30 insertions(+), 27 deletions(-)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 1458cb52de68..ee214b4112af 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -5217,11 +5217,6 @@ static void nested_vmx_disable_intercept_for_msr(unsigned long *msr_bitmap_l1,
{
int f = sizeof(unsigned long);
- if (!cpu_has_vmx_msr_bitmap()) {
- WARN_ON(1);
- return;
- }
-
/*
* See Intel PRM Vol. 3, 20.6.9 (MSR-Bitmap Address). Early manuals
* have the write-low and read-high bitmap offsets the wrong way round.
@@ -7493,6 +7488,7 @@ static int enter_vmx_operation(struct kvm_vcpu *vcpu)
(unsigned long *)__get_free_page(GFP_KERNEL);
if (!vmx->nested.msr_bitmap)
goto out_msr_bitmap;
+ memset(vmx->nested.msr_bitmap, 0xff, PAGE_SIZE);
}
vmx->nested.cached_vmcs12 = kmalloc(VMCS12_SIZE, GFP_KERNEL);
@@ -10325,36 +10321,43 @@ static inline bool nested_vmx_merge_msr_bitmap(struct kvm_vcpu *vcpu,
/* This shortcut is ok because we support only x2APIC MSRs so far. */
if (!nested_cpu_has_virt_x2apic_mode(vmcs12))
return false;
+ if (WARN_ON_ONCE(!cpu_has_vmx_msr_bitmap()))
+ return false;
page = kvm_vcpu_gpa_to_page(vcpu, vmcs12->msr_bitmap);
if (is_error_page(page))
return false;
- msr_bitmap_l1 = (unsigned long *)kmap(page);
- memset(msr_bitmap_l0, 0xff, PAGE_SIZE);
+ msr_bitmap_l1 = (unsigned long *)kmap(page);
+ if (nested_cpu_has_apic_reg_virt(vmcs12)) {
+ /* Disable read intercept for all MSRs between 0x800 and 0x8ff. */
+ for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
+ unsigned word = msr / BITS_PER_LONG;
+ msr_bitmap_l0[word] = msr_bitmap_l1[word];
+ msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
+ }
+ } else {
+ for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
+ unsigned word = msr / BITS_PER_LONG;
+ msr_bitmap_l0[word] = ~0;
+ msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
+ }
+ }
- if (nested_cpu_has_virt_x2apic_mode(vmcs12)) {
- if (nested_cpu_has_apic_reg_virt(vmcs12))
- for (msr = 0x800; msr <= 0x8ff; msr++)
- nested_vmx_disable_intercept_for_msr(
- msr_bitmap_l1, msr_bitmap_l0,
- msr, MSR_TYPE_R);
+ nested_vmx_disable_intercept_for_msr(
+ msr_bitmap_l1, msr_bitmap_l0,
+ APIC_BASE_MSR + (APIC_TASKPRI >> 4),
+ MSR_TYPE_W);
+ if (nested_cpu_has_vid(vmcs12)) {
nested_vmx_disable_intercept_for_msr(
- msr_bitmap_l1, msr_bitmap_l0,
- APIC_BASE_MSR + (APIC_TASKPRI >> 4),
- MSR_TYPE_R | MSR_TYPE_W);
-
- if (nested_cpu_has_vid(vmcs12)) {
- nested_vmx_disable_intercept_for_msr(
- msr_bitmap_l1, msr_bitmap_l0,
- APIC_BASE_MSR + (APIC_EOI >> 4),
- MSR_TYPE_W);
- nested_vmx_disable_intercept_for_msr(
- msr_bitmap_l1, msr_bitmap_l0,
- APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
- MSR_TYPE_W);
- }
+ msr_bitmap_l1, msr_bitmap_l0,
+ APIC_BASE_MSR + (APIC_EOI >> 4),
+ MSR_TYPE_W);
+ nested_vmx_disable_intercept_for_msr(
+ msr_bitmap_l1, msr_bitmap_l0,
+ APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
+ MSR_TYPE_W);
}
kunmap(page);
kvm_release_page_clean(page);
--
1.8.3.1
On 13.12.2017 14:30, Paolo Bonzini wrote:
> The bulk of the MSR bitmap is either immutable, or can be copied from
> the L1 bitmap. By initializing it at VMXON time, and copying the mutable
> parts one long at a time on vmentry (rather than one bit), about 4000
> clock cycles (30%) can be saved on a nested VMLAUNCH/VMRESUME.
>
> The resulting for loop only has four iterations, so it is cheap enough
> to reinitialize the MSR write bitmaps on every iteration, and it makes
> the code simpler.
>
> Suggested-by: Jim Mattson <[email protected]>
> Signed-off-by: Paolo Bonzini <[email protected]>
> ---
> arch/x86/kvm/vmx.c | 57 ++++++++++++++++++++++++++++--------------------------
> 1 file changed, 30 insertions(+), 27 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 1458cb52de68..ee214b4112af 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -5217,11 +5217,6 @@ static void nested_vmx_disable_intercept_for_msr(unsigned long *msr_bitmap_l1,
> {
> int f = sizeof(unsigned long);
>
> - if (!cpu_has_vmx_msr_bitmap()) {
> - WARN_ON(1);
> - return;
> - }
> -
> /*
> * See Intel PRM Vol. 3, 20.6.9 (MSR-Bitmap Address). Early manuals
> * have the write-low and read-high bitmap offsets the wrong way round.
> @@ -7493,6 +7488,7 @@ static int enter_vmx_operation(struct kvm_vcpu *vcpu)
> (unsigned long *)__get_free_page(GFP_KERNEL);
> if (!vmx->nested.msr_bitmap)
> goto out_msr_bitmap;
> + memset(vmx->nested.msr_bitmap, 0xff, PAGE_SIZE);
> }
>
> vmx->nested.cached_vmcs12 = kmalloc(VMCS12_SIZE, GFP_KERNEL);
> @@ -10325,36 +10321,43 @@ static inline bool nested_vmx_merge_msr_bitmap(struct kvm_vcpu *vcpu,
> /* This shortcut is ok because we support only x2APIC MSRs so far. */
> if (!nested_cpu_has_virt_x2apic_mode(vmcs12))
> return false;
> + if (WARN_ON_ONCE(!cpu_has_vmx_msr_bitmap()))
> + return false;
IMHO it would be nicer to always call nested_vmx_merge_msr_bitmap() and
make calling code less ugly:
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index ee214b4112af..d4f06fc643ae 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -10238,11 +10238,7 @@ static void nested_get_vmcs12_pages(struct
kvm_vcpu *vcpu,
(unsigned long)(vmcs12->posted_intr_desc_addr &
(PAGE_SIZE - 1)));
}
- if (cpu_has_vmx_msr_bitmap() &&
- nested_cpu_has(vmcs12, CPU_BASED_USE_MSR_BITMAPS) &&
- nested_vmx_merge_msr_bitmap(vcpu, vmcs12))
- ;
- else
+ if (!nested_vmx_merge_msr_bitmap(vcpu, vmcs12))
vmcs_clear_bits(CPU_BASED_VM_EXEC_CONTROL,
CPU_BASED_USE_MSR_BITMAPS);
}
@@ -10318,6 +10314,10 @@ static inline bool
nested_vmx_merge_msr_bitmap(struct kvm_vcpu *vcpu,
unsigned long *msr_bitmap_l1;
unsigned long *msr_bitmap_l0 = to_vmx(vcpu)->nested.msr_bitmap;
+ if (!cpu_has_vmx_msr_bitmap())
+ return false;
+ if (!nested_cpu_has(vmcs12, CPU_BASED_USE_MSR_BITMAPS))
+ return false;
/* This shortcut is ok because we support only x2APIC MSRs so
far. */
if (!nested_cpu_has_virt_x2apic_mode(vmcs12))
return false;
>
> page = kvm_vcpu_gpa_to_page(vcpu, vmcs12->msr_bitmap);
> if (is_error_page(page))
> return false;
> - msr_bitmap_l1 = (unsigned long *)kmap(page);
>
> - memset(msr_bitmap_l0, 0xff, PAGE_SIZE);
> + msr_bitmap_l1 = (unsigned long *)kmap(page);
Wouldn't it be easier to simply set everything to 0xff as before and
then only handle the one special case where you don't do that? e.g. the
complete else part would be gone.
> + if (nested_cpu_has_apic_reg_virt(vmcs12)) {
> + /* Disable read intercept for all MSRs between 0x800 and 0x8ff. */
> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
> + unsigned word = msr / BITS_PER_LONG;
> + msr_bitmap_l0[word] = msr_bitmap_l1[word];
> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
> + }
> + } else {
> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
> + unsigned word = msr / BITS_PER_LONG;
> + msr_bitmap_l0[word] = ~0;
> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
> + }
> + }
>
> - if (nested_cpu_has_virt_x2apic_mode(vmcs12)) {
> - if (nested_cpu_has_apic_reg_virt(vmcs12))
> - for (msr = 0x800; msr <= 0x8ff; msr++)
> - nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - msr, MSR_TYPE_R);
> + nested_vmx_disable_intercept_for_msr(
> + msr_bitmap_l1, msr_bitmap_l0,
> + APIC_BASE_MSR + (APIC_TASKPRI >> 4),
> + MSR_TYPE_W);
I'd vote for indenting the parameters properly (even though we exceed 80
chars by 1 then :) )
>
> + if (nested_cpu_has_vid(vmcs12)) {
> nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - APIC_BASE_MSR + (APIC_TASKPRI >> 4),
> - MSR_TYPE_R | MSR_TYPE_W);
> -
> - if (nested_cpu_has_vid(vmcs12)) {
> - nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - APIC_BASE_MSR + (APIC_EOI >> 4),
> - MSR_TYPE_W);
> - nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
> - MSR_TYPE_W);
> - }
> + msr_bitmap_l1, msr_bitmap_l0,
> + APIC_BASE_MSR + (APIC_EOI >> 4),
> + MSR_TYPE_W);
> + nested_vmx_disable_intercept_for_msr(
> + msr_bitmap_l1, msr_bitmap_l0,
> + APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
> + MSR_TYPE_W);
> }
> kunmap(page);
> kvm_release_page_clean(page);
>
--
Thanks,
David / dhildenb
> > - memset(msr_bitmap_l0, 0xff, PAGE_SIZE);
> > + msr_bitmap_l1 = (unsigned long *)kmap(page);
>
> Wouldn't it be easier to simply set everything to 0xff as before and
> then only handle the one special case where you don't do that? e.g. the
> complete else part would be gone.
It can change from one vmentry to the next, so I'm not sure how you'd
do that.
> > + nested_vmx_disable_intercept_for_msr(
> > + msr_bitmap_l1, msr_bitmap_l0,
> > + APIC_BASE_MSR + (APIC_TASKPRI >> 4),
> > + MSR_TYPE_W);
>
> I'd vote for indenting the parameters properly (even though we exceed 80
> chars by 1 then :) )
Ok, will do. Regarding the WARN, I'm undecided. I like the idea of not
calling a function that works on MSR bitmaps unless you have one...
Paolo
>
>>> + nested_vmx_disable_intercept_for_msr(
>>> + msr_bitmap_l1, msr_bitmap_l0,
>>> + APIC_BASE_MSR + (APIC_TASKPRI >> 4),
>>> + MSR_TYPE_W);
>>
>> I'd vote for indenting the parameters properly (even though we exceed 80
>> chars by 1 then :) )
>
> Ok, will do. Regarding the WARN, I'm undecided. I like the idea of not
> calling a function that works on MSR bitmaps unless you have one...
At least from a point of documentation it should be fine:
"return false to indicate that we do not use the hardware"
+gcc will inline this either way.
>
> Paolo
>
--
Thanks,
David / dhildenb
David Hildenbrand <[email protected]> writes:
...
>> vmx->nested.cached_vmcs12 = kmalloc(VMCS12_SIZE, GFP_KERNEL);
>> @@ -10325,36 +10321,43 @@ static inline bool nested_vmx_merge_msr_bitmap(struct kvm_vcpu *vcpu,
>> /* This shortcut is ok because we support only x2APIC MSRs so far. */
>> if (!nested_cpu_has_virt_x2apic_mode(vmcs12))
>> return false;
>> + if (WARN_ON_ONCE(!cpu_has_vmx_msr_bitmap()))
>> + return false;
>
> IMHO it would be nicer to always call nested_vmx_merge_msr_bitmap() and
> make calling code less ugly:
>
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index ee214b4112af..d4f06fc643ae 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -10238,11 +10238,7 @@ static void nested_get_vmcs12_pages(struct
> kvm_vcpu *vcpu,
> (unsigned long)(vmcs12->posted_intr_desc_addr &
> (PAGE_SIZE - 1)));
> }
> - if (cpu_has_vmx_msr_bitmap() &&
> - nested_cpu_has(vmcs12, CPU_BASED_USE_MSR_BITMAPS) &&
> - nested_vmx_merge_msr_bitmap(vcpu, vmcs12))
> - ;
> - else
> + if (!nested_vmx_merge_msr_bitmap(vcpu, vmcs12))
> vmcs_clear_bits(CPU_BASED_VM_EXEC_CONTROL,
> CPU_BASED_USE_MSR_BITMAPS);
> }
> @@ -10318,6 +10314,10 @@ static inline bool
> nested_vmx_merge_msr_bitmap(struct kvm_vcpu *vcpu,
> unsigned long *msr_bitmap_l1;
> unsigned long *msr_bitmap_l0 = to_vmx(vcpu)->nested.msr_bitmap;
>
> + if (!cpu_has_vmx_msr_bitmap())
> + return false;
> + if (!nested_cpu_has(vmcs12, CPU_BASED_USE_MSR_BITMAPS))
> + return false;
This looks good, otherwise the WARN_ON_ONCE just seems like an unnecessary
check since the function is only called once.
Bandan
> /* This shortcut is ok because we support only x2APIC MSRs so
> far. */
> if (!nested_cpu_has_virt_x2apic_mode(vmcs12))
> return false;
>
>
>
>>
>> page = kvm_vcpu_gpa_to_page(vcpu, vmcs12->msr_bitmap);
>> if (is_error_page(page))
>> return false;
>> - msr_bitmap_l1 = (unsigned long *)kmap(page);
>>
>> - memset(msr_bitmap_l0, 0xff, PAGE_SIZE);
>> + msr_bitmap_l1 = (unsigned long *)kmap(page);
>
>
> Wouldn't it be easier to simply set everything to 0xff as before and
> then only handle the one special case where you don't do that? e.g. the
> complete else part would be gone.
>
>> + if (nested_cpu_has_apic_reg_virt(vmcs12)) {
>> + /* Disable read intercept for all MSRs between 0x800 and 0x8ff. */
>> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
>> + unsigned word = msr / BITS_PER_LONG;
>> + msr_bitmap_l0[word] = msr_bitmap_l1[word];
>> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
>> + }
>> + } else {
>> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
>> + unsigned word = msr / BITS_PER_LONG;
>> + msr_bitmap_l0[word] = ~0;
>> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
>> + }
>> + }
>>
>> - if (nested_cpu_has_virt_x2apic_mode(vmcs12)) {
>> - if (nested_cpu_has_apic_reg_virt(vmcs12))
>> - for (msr = 0x800; msr <= 0x8ff; msr++)
>> - nested_vmx_disable_intercept_for_msr(
>> - msr_bitmap_l1, msr_bitmap_l0,
>> - msr, MSR_TYPE_R);
>> + nested_vmx_disable_intercept_for_msr(
>> + msr_bitmap_l1, msr_bitmap_l0,
>> + APIC_BASE_MSR + (APIC_TASKPRI >> 4),
>> + MSR_TYPE_W);
>
> I'd vote for indenting the parameters properly (even though we exceed 80
> chars by 1 then :) )
>
>>
>> + if (nested_cpu_has_vid(vmcs12)) {
>> nested_vmx_disable_intercept_for_msr(
>> - msr_bitmap_l1, msr_bitmap_l0,
>> - APIC_BASE_MSR + (APIC_TASKPRI >> 4),
>> - MSR_TYPE_R | MSR_TYPE_W);
>> -
>> - if (nested_cpu_has_vid(vmcs12)) {
>> - nested_vmx_disable_intercept_for_msr(
>> - msr_bitmap_l1, msr_bitmap_l0,
>> - APIC_BASE_MSR + (APIC_EOI >> 4),
>> - MSR_TYPE_W);
>> - nested_vmx_disable_intercept_for_msr(
>> - msr_bitmap_l1, msr_bitmap_l0,
>> - APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
>> - MSR_TYPE_W);
>> - }
>> + msr_bitmap_l1, msr_bitmap_l0,
>> + APIC_BASE_MSR + (APIC_EOI >> 4),
>> + MSR_TYPE_W);
>> + nested_vmx_disable_intercept_for_msr(
>> + msr_bitmap_l1, msr_bitmap_l0,
>> + APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
>> + MSR_TYPE_W);
>> }
>> kunmap(page);
>> kvm_release_page_clean(page);
>>
On Wed, Dec 13, 2017 at 5:30 AM, Paolo Bonzini <[email protected]> wrote:
> The bulk of the MSR bitmap is either immutable, or can be copied from
> the L1 bitmap. By initializing it at VMXON time, and copying the mutable
> parts one long at a time on vmentry (rather than one bit), about 4000
> clock cycles (30%) can be saved on a nested VMLAUNCH/VMRESUME.
>
> The resulting for loop only has four iterations, so it is cheap enough
> to reinitialize the MSR write bitmaps on every iteration, and it makes
> the code simpler.
Thanks so much for doing this!
> Suggested-by: Jim Mattson <[email protected]>
> Signed-off-by: Paolo Bonzini <[email protected]>
> ---
> arch/x86/kvm/vmx.c | 57 ++++++++++++++++++++++++++++--------------------------
> 1 file changed, 30 insertions(+), 27 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 1458cb52de68..ee214b4112af 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -5217,11 +5217,6 @@ static void nested_vmx_disable_intercept_for_msr(unsigned long *msr_bitmap_l1,
> {
> int f = sizeof(unsigned long);
>
> - if (!cpu_has_vmx_msr_bitmap()) {
> - WARN_ON(1);
> - return;
> - }
> -
> /*
> * See Intel PRM Vol. 3, 20.6.9 (MSR-Bitmap Address). Early manuals
> * have the write-low and read-high bitmap offsets the wrong way round.
> @@ -7493,6 +7488,7 @@ static int enter_vmx_operation(struct kvm_vcpu *vcpu)
> (unsigned long *)__get_free_page(GFP_KERNEL);
> if (!vmx->nested.msr_bitmap)
> goto out_msr_bitmap;
> + memset(vmx->nested.msr_bitmap, 0xff, PAGE_SIZE);
> }
>
> vmx->nested.cached_vmcs12 = kmalloc(VMCS12_SIZE, GFP_KERNEL);
> @@ -10325,36 +10321,43 @@ static inline bool nested_vmx_merge_msr_bitmap(struct kvm_vcpu *vcpu,
> /* This shortcut is ok because we support only x2APIC MSRs so far. */
> if (!nested_cpu_has_virt_x2apic_mode(vmcs12))
> return false;
> + if (WARN_ON_ONCE(!cpu_has_vmx_msr_bitmap()))
> + return false;
>
> page = kvm_vcpu_gpa_to_page(vcpu, vmcs12->msr_bitmap);
> if (is_error_page(page))
> return false;
> - msr_bitmap_l1 = (unsigned long *)kmap(page);
>
> - memset(msr_bitmap_l0, 0xff, PAGE_SIZE);
> + msr_bitmap_l1 = (unsigned long *)kmap(page);
> + if (nested_cpu_has_apic_reg_virt(vmcs12)) {
> + /* Disable read intercept for all MSRs between 0x800 and 0x8ff. */
Aren't we actually adopting the read intercepts from VMCS12 and
*enabling* the *write* intercepts?
> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
> + unsigned word = msr / BITS_PER_LONG;
> + msr_bitmap_l0[word] = msr_bitmap_l1[word];
> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
The indexing above seems a bit obtuse, but maybe it will be clear
enough after the above comment is fixed up.
> + }
> + } else {
> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
> + unsigned word = msr / BITS_PER_LONG;
> + msr_bitmap_l0[word] = ~0;
> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
> + }
> + }
>
> - if (nested_cpu_has_virt_x2apic_mode(vmcs12)) {
> - if (nested_cpu_has_apic_reg_virt(vmcs12))
> - for (msr = 0x800; msr <= 0x8ff; msr++)
> - nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - msr, MSR_TYPE_R);
> + nested_vmx_disable_intercept_for_msr(
> + msr_bitmap_l1, msr_bitmap_l0,
> + APIC_BASE_MSR + (APIC_TASKPRI >> 4),
Perhaps you could #define X2APIC_MSR(reg) (APIC_BASE_MSR + ((reg) >>
4)) somewhere appropriate (e.g. arch/x86/include/asm/apicdef.h) and
use that here (and below) for brevity?
> + MSR_TYPE_W);
>
> + if (nested_cpu_has_vid(vmcs12)) {
> nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - APIC_BASE_MSR + (APIC_TASKPRI >> 4),
> - MSR_TYPE_R | MSR_TYPE_W);
> -
> - if (nested_cpu_has_vid(vmcs12)) {
> - nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - APIC_BASE_MSR + (APIC_EOI >> 4),
> - MSR_TYPE_W);
> - nested_vmx_disable_intercept_for_msr(
> - msr_bitmap_l1, msr_bitmap_l0,
> - APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
> - MSR_TYPE_W);
> - }
> + msr_bitmap_l1, msr_bitmap_l0,
> + APIC_BASE_MSR + (APIC_EOI >> 4),
> + MSR_TYPE_W);
> + nested_vmx_disable_intercept_for_msr(
> + msr_bitmap_l1, msr_bitmap_l0,
> + APIC_BASE_MSR + (APIC_SELF_IPI >> 4),
> + MSR_TYPE_W);
> }
> kunmap(page);
> kvm_release_page_clean(page);
> --
> 1.8.3.1
>
Should we also think about letting L1 control pass-through of some of
the more mundane MSRs, like FS_BASE, GS_BASE, and KERNEL_GS_BASE?
Reviewed-by: Jim Mattson <[email protected]>
On 19/12/2017 20:58, Jim Mattson wrote:
>> + /* Disable read intercept for all MSRs between 0x800 and 0x8ff. */
> Aren't we actually adopting the read intercepts from VMCS12 and
> *enabling* the *write* intercepts?
Yeah, the comment isn't the best. What I mean is that L0 doesn't care
about intercepting the reads, so the default all-set msr_bitmap_l0 is
changed as if the read intercept was disabled. This leaves
msr_bitmap_l1 as the result.
>> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
>> + unsigned word = msr / BITS_PER_LONG;
>> + msr_bitmap_l0[word] = msr_bitmap_l1[word];
>> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
>
> The indexing above seems a bit obtuse, but maybe it will be clear
> enough after the above comment is fixed up.
>
>> + }
>> + } else {
>> + for (msr = 0x800; msr <= 0x8ff; msr += BITS_PER_LONG) {
>> + unsigned word = msr / BITS_PER_LONG;
>> + msr_bitmap_l0[word] = ~0;
>> + msr_bitmap_l0[word + (0x800 / sizeof(long))] = ~0;
>> + }
>> + }
>>
>> - if (nested_cpu_has_virt_x2apic_mode(vmcs12)) {
>> - if (nested_cpu_has_apic_reg_virt(vmcs12))
>> - for (msr = 0x800; msr <= 0x8ff; msr++)
>> - nested_vmx_disable_intercept_for_msr(
>> - msr_bitmap_l1, msr_bitmap_l0,
>> - msr, MSR_TYPE_R);
>> + nested_vmx_disable_intercept_for_msr(
>> + msr_bitmap_l1, msr_bitmap_l0,
>> + APIC_BASE_MSR + (APIC_TASKPRI >> 4),
> Perhaps you could #define X2APIC_MSR(reg) (APIC_BASE_MSR + ((reg) >>
> 4)) somewhere appropriate (e.g. arch/x86/include/asm/apicdef.h) and
> use that here (and below) for brevity?
Good idea.
> Should we also think about letting L1 control pass-through of some of
> the more mundane MSRs, like FS_BASE, GS_BASE, and KERNEL_GS_BASE?
Rhetorical question? :) Yes, it probably gives a small performance
improvement on real-world multi-process (or even just multi-threaded)
workloads. It's a separate thing to do though.
The next obvious step, to me, is to split prepare_vmcs02 in two parts.
Most of the work can be skipped in the common case where we didn't get
any vmread/vmwrite exit and all guest accesses between VMRESUMEs go
through the shadow VMCS. This gives both an easy place to draw the line
between the two parts, and an easy way to detect the need to go down
slow path. Any takers? :)
Paolo