kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Stamatis, Ilias" <ilstam@amazon.com>
To: "kvm@vger.kernel.org" <kvm@vger.kernel.org>,
	"pbonzini@redhat.com" <pbonzini@redhat.com>,
	"mlevitsk@redhat.com" <mlevitsk@redhat.com>,
	"ilstam@mailbox.org" <ilstam@mailbox.org>
Cc: "jmattson@google.com" <jmattson@google.com>,
	"Woodhouse, David" <dwmw@amazon.co.uk>,
	"vkuznets@redhat.com" <vkuznets@redhat.com>,
	"joro@8bytes.org" <joro@8bytes.org>,
	"mtosatti@redhat.com" <mtosatti@redhat.com>,
	"zamsden@gmail.com" <zamsden@gmail.com>,
	"seanjc@google.com" <seanjc@google.com>,
	"wanpengli@tencent.com" <wanpengli@tencent.com>
Subject: Re: [PATCH 6/8] KVM: VMX: Make vmx_write_l1_tsc_offset() work with nested TSC scaling
Date: Tue, 11 May 2021 17:44:50 +0000	[thread overview]
Message-ID: <659ed5bcd9e4f43da17b6956603b21b9253eba77.camel@amazon.com> (raw)
In-Reply-To: <875fa1ff9a85ff601a05030eaa24a1db45a71f36.camel@redhat.com>

On Tue, 2021-05-11 at 15:44 +0300, Maxim Levitsky wrote:
> On Mon, 2021-05-10 at 16:08 +0000, Stamatis, Ilias wrote:
> > On Mon, 2021-05-10 at 16:54 +0300, Maxim Levitsky wrote:
> > > On Thu, 2021-05-06 at 10:32 +0000, ilstam@mailbox.org wrote:
> > > > From: Ilias Stamatis <ilstam@amazon.com>
> > > > 
> > > > Calculating the current TSC offset is done differently when nested TSC
> > > > scaling is used.
> > > > 
> > > > Signed-off-by: Ilias Stamatis <ilstam@amazon.com>
> > > > ---
> > > >  arch/x86/kvm/vmx/vmx.c | 26 ++++++++++++++++++++------
> > > >  1 file changed, 20 insertions(+), 6 deletions(-)
> > > > 
> > > > diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> > > > index 49241423b854..df7dc0e4c903 100644
> > > > --- a/arch/x86/kvm/vmx/vmx.c
> > > > +++ b/arch/x86/kvm/vmx/vmx.c
> > > > @@ -1797,10 +1797,16 @@ static void setup_msrs(struct vcpu_vmx *vmx)
> > > >               vmx_update_msr_bitmap(&vmx->vcpu);
> > > >  }
> > > > 
> > > > -static u64 vmx_write_l1_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
> > > > +/*
> > > > + * This function receives the requested offset for L1 as an argument but it
> > > > + * actually writes the "current" tsc offset to the VMCS and returns it. The
> > > > + * current offset might be different in case an L2 guest is currently running
> > > > + * and its VMCS02 is loaded.
> > > > + */
> > > 
> > > (Not related to this patch) It might be a good idea to rename this callback
> > > instead of this comment, but I am not sure about it.
> > > 
> > 
> > Yes! I was planning to do this on v2 anyway as the name of that function
> > is completely misleading/wrong IMHO.
> > 
> > I think that even the comment inside it that explains that when L1
> > doesn't trap WRMSR then L2 TSC writes overwrite L1's TSC is misplaced.
> > It should go one or more levels above I believe.
> > 
> > This function simply
> > updates the TSC offset in the current VMCS depending on whether L1 or L2
> > is executing.
> > 
> > > > +static u64 vmx_write_l1_tsc_offset(struct kvm_vcpu *vcpu, u64 l1_offset)
> > > >  {
> > > >       struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
> > > > -     u64 g_tsc_offset = 0;
> > > > +     u64 cur_offset = l1_offset;
> > > > 
> > > >       /*
> > > >        * We're here if L1 chose not to trap WRMSR to TSC. According
> > > > @@ -1809,11 +1815,19 @@ static u64 vmx_write_l1_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
> > > >        * to the newly set TSC to get L2's TSC.
> > > >        */
> > > >       if (is_guest_mode(vcpu) &&
> > > > -         (vmcs12->cpu_based_vm_exec_control & CPU_BASED_USE_TSC_OFFSETTING))
> > > > -             g_tsc_offset = vmcs12->tsc_offset;
> > > > +         (vmcs12->cpu_based_vm_exec_control & CPU_BASED_USE_TSC_OFFSETTING)) {
> > > > +             if (vmcs12->secondary_vm_exec_control & SECONDARY_EXEC_TSC_SCALING) {
> > > > +                     cur_offset = kvm_compute_02_tsc_offset(
> > > > +                                     l1_offset,
> > > > +                                     vmcs12->tsc_multiplier,
> > > > +                                     vmcs12->tsc_offset);
> > > > +             } else {
> > > > +                     cur_offset = l1_offset + vmcs12->tsc_offset;
> > > > +             }
> > > > +     }
> > > > 
> > > > -     vmcs_write64(TSC_OFFSET, offset + g_tsc_offset);
> > > > -     return offset + g_tsc_offset;
> > > > +     vmcs_write64(TSC_OFFSET, cur_offset);
> > > > +     return cur_offset;
> > > >  }
> > > > 
> > > >  /*
> > > 
> > > This code would be ideal to move to common code as SVM will do basically
> > > the same thing.
> > > Doesn't have to be done now though.
> > > 
> > 
> > Hmm, how can we do the feature availability checking in common code?
> 
> We can add a vendor callback for this.
> 
> Just a few thoughts about how I think we can implement
> the nested TSC scaling in (mostly) common code:
> 
> 
> Assuming that the common code knows that:
> 1. Nested guest is running (already the case)
> 
> 2. The format of the scaling multiplier is known
> (thankfully both SVM and VMX use fixed point binary number.
> 
> SVM is using 8.32 format and VMX using 16.48 format.
> 
> The common code already knows this via
> kvm_max_tsc_scaling_ratio/kvm_tsc_scaling_ratio_frac_bits.
> 
> 3. the value of nested TSC scaling multiplier
> is known to the common code.
> 
> (a callback to VMX/SVM code to query the TSC scaling value,
> and have it return 1 when TSC scaling is disabled should work)
> 

I suppose you mean return kvm_default_tsc_scaling_ratio

> 
> Then the common code can do the whole thing, and only
> let the SVM/VMX code write the actual multiplier.
> 
> As far as I know on the SVM, the TSC scaling works like that:
> 
> 1. SVM has a CPUID bit to indicate that tsc scaling is supported.
> (X86_FEATURE_TSCRATEMSR)
> 
> When this bit is set, TSC scale ratio is unconditionally enabled (but
> can be just 1), and it is set via a special MSR (MSR_AMD64_TSC_RATIO)
> rather than a field in VMCB (someone at AMD did cut corners...).
> 
> However since the TSC scaling is only effective when a guest is running,
> that MSR can be treated almost as if it was just another VMCB field.
> 
> The TSC scale value is 32 bit fraction and another 8 bits the integer value
> (as opposed to 48 bit fraction on VMX and 16 bits integer value).
> 
> I don't think that there are any other differences.
> 
> I should also note that I can do all of the above myself if
> I end up implementing the nested TSC scaling on AMD
> so I don't object much to the way that this patch series is done.
> 

That's fine, I will add the callbacks and move everything to common code. And
then you can fill the svm-specific bits if you want.

Thanks,
Ilias

  reply	other threads:[~2021-05-11 17:45 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-06 10:32 [PATCH 0/8] KVM: VMX: Implement " ilstam
2021-05-06 10:32 ` [PATCH 1/8] KVM: VMX: Add a TSC multiplier field in VMCS12 ilstam
2021-05-06 14:50   ` kernel test robot
2021-05-06 17:36   ` Jim Mattson
2021-05-10 13:42   ` Maxim Levitsky
2021-05-06 10:32 ` [PATCH 2/8] KVM: X86: Store L1's TSC scaling ratio in 'struct kvm_vcpu_arch' ilstam
2021-05-10 13:43   ` Maxim Levitsky
2021-05-06 10:32 ` [PATCH 3/8] KVM: X86: Pass an additional 'L1' argument to kvm_scale_tsc() ilstam
2021-05-10 13:52   ` Maxim Levitsky
2021-05-10 15:44     ` Stamatis, Ilias
2021-05-06 10:32 ` [PATCH 4/8] KVM: VMX: Adjust the TSC-related VMCS fields on L2 entry and exit ilstam
2021-05-06 11:32   ` Paolo Bonzini
2021-05-06 17:35     ` Stamatis, Ilias
2021-05-10 14:11       ` Paolo Bonzini
2021-05-10 13:53   ` Maxim Levitsky
2021-05-10 14:44     ` Stamatis, Ilias
2021-05-11 12:38       ` Maxim Levitsky
2021-05-11 15:11         ` Stamatis, Ilias
2021-05-06 10:32 ` [PATCH 5/8] KVM: X86: Move tracing outside write_l1_tsc_offset() ilstam
2021-05-10 13:54   ` Maxim Levitsky
2021-05-06 10:32 ` [PATCH 6/8] KVM: VMX: Make vmx_write_l1_tsc_offset() work with nested TSC scaling ilstam
2021-05-10 13:54   ` Maxim Levitsky
2021-05-10 16:08     ` Stamatis, Ilias
2021-05-11 12:44       ` Maxim Levitsky
2021-05-11 17:44         ` Stamatis, Ilias [this message]
2021-05-06 10:32 ` [PATCH 7/8] KVM: VMX: Expose TSC scaling to L2 ilstam
2021-05-10 13:56   ` Maxim Levitsky
2021-05-06 10:32 ` [PATCH 8/8] KVM: selftests: x86: Add vmx_nested_tsc_scaling_test ilstam
2021-05-10 13:59   ` Maxim Levitsky
2021-05-11 11:16     ` Stamatis, Ilias
2021-05-11 12:47       ` Maxim Levitsky
2021-05-11 14:02         ` Stamatis, Ilias
2021-05-06 17:16 ` [PATCH 0/8] KVM: VMX: Implement nested TSC scaling Jim Mattson
2021-05-06 17:48   ` Stamatis, Ilias
2021-05-10 13:43     ` Maxim Levitsky
2021-05-10 14:29   ` Paolo Bonzini

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=659ed5bcd9e4f43da17b6956603b21b9253eba77.camel@amazon.com \
    --to=ilstam@amazon.com \
    --cc=dwmw@amazon.co.uk \
    --cc=ilstam@mailbox.org \
    --cc=jmattson@google.com \
    --cc=joro@8bytes.org \
    --cc=kvm@vger.kernel.org \
    --cc=mlevitsk@redhat.com \
    --cc=mtosatti@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=seanjc@google.com \
    --cc=vkuznets@redhat.com \
    --cc=wanpengli@tencent.com \
    --cc=zamsden@gmail.com \
    --subject='Re: [PATCH 6/8] KVM: VMX: Make vmx_write_l1_tsc_offset() work with nested TSC scaling' \
    /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

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).