Re: [PATCH v2 0/4] KVM: VMX: enable LBR virtualization

[Jian Zhou <jianjay.zhou@huawei.com>]

On 2015/11/11 23:23, Paolo Bonzini wrote:
>
>
> On 23/10/2015 11:15, Jian Zhou wrote:
>> Changelog in v2:
>>    (1) move the implementation into vmx.c
>>    (2) migraton is supported
>>    (3) add arrays in kvm_vcpu_arch struct to save/restore
>>        LBR MSRs at vm exit/entry time.
>>    (4) add a parameter of kvm_intel module to permanently
>>        disable LBRV
>>    (5) table of supported CPUs is reorgnized, LBRV
>>        can be enabled or not according to the guest CPUID
>>
>> Jian Zhou (4):
>>    KVM: X86: Add arrays to save/restore LBR MSRs
>>    KVM: X86: LBR MSRs of supported CPU types
>>    KVM: X86: Migration is supported
>>    KVM: VMX: details of LBR virtualization implementation
>>
>>   arch/x86/include/asm/kvm_host.h  |  26 ++++-
>>   arch/x86/include/asm/msr-index.h |  26 ++++-
>>   arch/x86/kvm/vmx.c               | 245 +++++++++++++++++++++++++++++++++++++++
>>   arch/x86/kvm/x86.c               |  88 ++++++++++++--
>>   4 files changed, 366 insertions(+), 19 deletions(-)
>
> Thanks, this looks better!
>
> The reason why it took me so long to review it, is that I wanted to
> understand what happens if you're running this on CPU model x but using
> CPU model y for the guest.  I still haven't grokked that fully, so I'll
> apply your patches locally and play with them.

   Yes, that is a good question. I plan to write a kernel module in the
   guest to read/write the MSR_IA32_DEBUGCTLMSR and MSRs of LBR
   stack with host CPU model e.g. SandyBridge while using guest CPU
   model e.g. core2duo for the guest. (The address of MSRs recording
   last branch information between SandyBridge and core2duo is different)

> In the meanwhile, feel free to send v3 with: 1) the tweak I suggested to
> patch 3; 2) the fix for the problem that the buildbot reported on patch 1.

   Okay, will fix them in v3.

   Thanks,
   Jian

> Paolo
>
> .
>