mbox series

[v4,0/4] Add support for capturing the highest observable L2 TSC

Message ID 20191108051439.185635-1-aaronlewis@google.com (mailing list archive)
Headers show
Series Add support for capturing the highest observable L2 TSC | expand

Message

Aaron Lewis Nov. 8, 2019, 5:14 a.m. UTC
The L1 hypervisor may include the IA32_TIME_STAMP_COUNTER MSR in the
vmcs12 MSR VM-exit MSR-store area as a way of determining the highest
TSC value that might have been observed by L2 prior to VM-exit. The
current implementation does not capture a very tight bound on this
value.  To tighten the bound, add the IA32_TIME_STAMP_COUNTER MSR to the
vmcs02 VM-exit MSR-store area whenever it appears in the vmcs12 VM-exit
MSR-store area.  When L0 processes the vmcs12 VM-exit MSR-store area
during the emulation of an L2->L1 VM-exit, special-case the
IA32_TIME_STAMP_COUNTER MSR, using the value stored in the vmcs02
VM-exit MSR-store area to derive the value to be stored in the vmcs12
VM-exit MSR-store area.

v3 -> v4:
 - Squash the final commit with the previous one used to prepare the MSR-store
   area.  There is no need for this split after all.

v2 -> v3:
 - Rename NR_MSR_ENTRIES to NR_LOADSAVE_MSRS
 - Pull setup code for preparing the MSR-store area out of the final commit and
   put it in it's own commit (4/5).
 - Export vmx_find_msr_index() in the final commit instead of in commit 3/5 as
   it isn't until the final commit that we actually use it.

v1 -> v2:
 - Rename function nested_vmx_get_msr_value() to
   nested_vmx_get_vmexit_msr_value().
 - Remove unneeded tag 'Change-Id' from commit messages.

Aaron Lewis (4):
  kvm: nested: Introduce read_and_check_msr_entry()
  kvm: vmx: Rename NR_AUTOLOAD_MSRS to NR_LOADSTORE_MSRS
  kvm: vmx: Rename function find_msr() to vmx_find_msr_index()
  KVM: nVMX: Add support for capturing highest observable L2 TSC

 arch/x86/kvm/vmx/nested.c | 136 ++++++++++++++++++++++++++++++++------
 arch/x86/kvm/vmx/vmx.c    |  14 ++--
 arch/x86/kvm/vmx/vmx.h    |   9 ++-
 3 files changed, 131 insertions(+), 28 deletions(-)

Comments

Paolo Bonzini Nov. 15, 2019, 10:23 a.m. UTC | #1
On 08/11/19 06:14, Aaron Lewis wrote:
> The L1 hypervisor may include the IA32_TIME_STAMP_COUNTER MSR in the
> vmcs12 MSR VM-exit MSR-store area as a way of determining the highest
> TSC value that might have been observed by L2 prior to VM-exit. The
> current implementation does not capture a very tight bound on this
> value.  To tighten the bound, add the IA32_TIME_STAMP_COUNTER MSR to the
> vmcs02 VM-exit MSR-store area whenever it appears in the vmcs12 VM-exit
> MSR-store area.  When L0 processes the vmcs12 VM-exit MSR-store area
> during the emulation of an L2->L1 VM-exit, special-case the
> IA32_TIME_STAMP_COUNTER MSR, using the value stored in the vmcs02
> VM-exit MSR-store area to derive the value to be stored in the vmcs12
> VM-exit MSR-store area.
> 
> v3 -> v4:
>  - Squash the final commit with the previous one used to prepare the MSR-store
>    area.  There is no need for this split after all.
> 
> v2 -> v3:
>  - Rename NR_MSR_ENTRIES to NR_LOADSAVE_MSRS
>  - Pull setup code for preparing the MSR-store area out of the final commit and
>    put it in it's own commit (4/5).
>  - Export vmx_find_msr_index() in the final commit instead of in commit 3/5 as
>    it isn't until the final commit that we actually use it.
> 
> v1 -> v2:
>  - Rename function nested_vmx_get_msr_value() to
>    nested_vmx_get_vmexit_msr_value().
>  - Remove unneeded tag 'Change-Id' from commit messages.
> 
> Aaron Lewis (4):
>   kvm: nested: Introduce read_and_check_msr_entry()
>   kvm: vmx: Rename NR_AUTOLOAD_MSRS to NR_LOADSTORE_MSRS
>   kvm: vmx: Rename function find_msr() to vmx_find_msr_index()
>   KVM: nVMX: Add support for capturing highest observable L2 TSC
> 
>  arch/x86/kvm/vmx/nested.c | 136 ++++++++++++++++++++++++++++++++------
>  arch/x86/kvm/vmx/vmx.c    |  14 ++--
>  arch/x86/kvm/vmx/vmx.h    |   9 ++-
>  3 files changed, 131 insertions(+), 28 deletions(-)
> 

Queued, but it would be good to have a testcase for this, either for
kvm-unit-tests or for tools/testing/selftests/kvm.

Paolo
Aaron Lewis Nov. 15, 2019, 2:25 p.m. UTC | #2
On Fri, Nov 15, 2019 at 2:23 AM Paolo Bonzini <pbonzini@redhat.com> wrote:
>
> On 08/11/19 06:14, Aaron Lewis wrote:
> > The L1 hypervisor may include the IA32_TIME_STAMP_COUNTER MSR in the
> > vmcs12 MSR VM-exit MSR-store area as a way of determining the highest
> > TSC value that might have been observed by L2 prior to VM-exit. The
> > current implementation does not capture a very tight bound on this
> > value.  To tighten the bound, add the IA32_TIME_STAMP_COUNTER MSR to the
> > vmcs02 VM-exit MSR-store area whenever it appears in the vmcs12 VM-exit
> > MSR-store area.  When L0 processes the vmcs12 VM-exit MSR-store area
> > during the emulation of an L2->L1 VM-exit, special-case the
> > IA32_TIME_STAMP_COUNTER MSR, using the value stored in the vmcs02
> > VM-exit MSR-store area to derive the value to be stored in the vmcs12
> > VM-exit MSR-store area.
> >
> > v3 -> v4:
> >  - Squash the final commit with the previous one used to prepare the MSR-store
> >    area.  There is no need for this split after all.
> >
> > v2 -> v3:
> >  - Rename NR_MSR_ENTRIES to NR_LOADSAVE_MSRS
> >  - Pull setup code for preparing the MSR-store area out of the final commit and
> >    put it in it's own commit (4/5).
> >  - Export vmx_find_msr_index() in the final commit instead of in commit 3/5 as
> >    it isn't until the final commit that we actually use it.
> >
> > v1 -> v2:
> >  - Rename function nested_vmx_get_msr_value() to
> >    nested_vmx_get_vmexit_msr_value().
> >  - Remove unneeded tag 'Change-Id' from commit messages.
> >
> > Aaron Lewis (4):
> >   kvm: nested: Introduce read_and_check_msr_entry()
> >   kvm: vmx: Rename NR_AUTOLOAD_MSRS to NR_LOADSTORE_MSRS
> >   kvm: vmx: Rename function find_msr() to vmx_find_msr_index()
> >   KVM: nVMX: Add support for capturing highest observable L2 TSC
> >
> >  arch/x86/kvm/vmx/nested.c | 136 ++++++++++++++++++++++++++++++++------
> >  arch/x86/kvm/vmx/vmx.c    |  14 ++--
> >  arch/x86/kvm/vmx/vmx.h    |   9 ++-
> >  3 files changed, 131 insertions(+), 28 deletions(-)
> >
>
> Queued, but it would be good to have a testcase for this, either for
> kvm-unit-tests or for tools/testing/selftests/kvm.
>
> Paolo
>

Agreed.  I have some test cases in kvm-unit-tests for this code that
I've been using to test these changes locally, however, they would
fail upstream without "[kvm-unit-tests PATCH] x86: Fix the register
order to match struct regs" being taken first.  I'll ping that patch
again.
Paolo Bonzini Nov. 15, 2019, 2:26 p.m. UTC | #3
On 15/11/19 15:25, Aaron Lewis wrote:
> Agreed.  I have some test cases in kvm-unit-tests for this code that
> I've been using to test these changes locally, however, they would
> fail upstream without "[kvm-unit-tests PATCH] x86: Fix the register
> order to match struct regs" being taken first.  I'll ping that patch
> again.
> 

You've just done that... :)

Paolo