From: David Matlack <dmatlack@google.com>
To: Jing Zhang <jingzhangos@google.com>
Cc: KVM <kvm@vger.kernel.org>, Paolo Bonzini <pbonzini@redhat.com>,
Peter Shier <pshier@google.com>, Oliver Upton <oupton@google.com>,
Sean Christopherson <seanjc@google.com>,
Peter Feiner <pfeiner@google.com>,
Jim Mattson <jmattson@google.com>
Subject: Re: [PATCH] KVM: stats: x86: vmx: add exit reason stats to vt-x instructions
Date: Thu, 19 Aug 2021 11:43:29 -0700 [thread overview]
Message-ID: <CALzav=cP17YXD8dRJnYFe_qmox3CTtpVBtLbU42Ei9zea2w21Q@mail.gmail.com> (raw)
In-Reply-To: <20210819180012.744855-1-jingzhangos@google.com>
On Thu, Aug 19, 2021 at 11:00 AM Jing Zhang <jingzhangos@google.com> wrote:
>
> These stats will be used to monitor the nested virtualization use in
> VMs. Most importantly: VMXON exits are evidence that the guest has
> enabled VMX, VMLAUNCH/VMRESUME exits are evidence that the guest has run
> an L2.
This series is superseded by Peter Feiner's internal KVM patch that
exports an array of counts, one for each VM-exit reason ("kvm: vmx
exit reason stats"). This is better since it does not require
instrumenting every VM-exit handler in KVM and introducing a stat for
every exit.
Assuming upstream would want exit count stats I would suggest we drop
this patch and upstream Peter's instead. Although we need to sort out
AMD and other architectures as well.
>
> Original-by: David Matlack <dmatlack@google.com>
> Signed-off-by: Jing Zhang <jingzhangos@google.com>
> ---
> arch/x86/include/asm/kvm_host.h | 11 +++++++++++
> arch/x86/kvm/vmx/nested.c | 17 +++++++++++++++++
> arch/x86/kvm/x86.c | 13 ++++++++++++-
> 3 files changed, 40 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 09b256db394a..e3afbc7926e0 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -1257,6 +1257,17 @@ struct kvm_vcpu_stat {
> u64 directed_yield_attempted;
> u64 directed_yield_successful;
> u64 guest_mode;
> + u64 vmclear_exits;
> + u64 vmlaunch_exits;
> + u64 vmptrld_exits;
> + u64 vmptrst_exits;
> + u64 vmread_exits;
> + u64 vmresume_exits;
> + u64 vmwrite_exits;
> + u64 vmoff_exits;
> + u64 vmon_exits;
> + u64 invept_exits;
> + u64 invvpid_exits;
> };
>
> struct x86_instruction_info;
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index bc6327950657..8696f2612953 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -4879,6 +4879,7 @@ static int handle_vmon(struct kvm_vcpu *vcpu)
> const u64 VMXON_NEEDED_FEATURES = FEAT_CTL_LOCKED
> | FEAT_CTL_VMX_ENABLED_OUTSIDE_SMX;
>
> + ++vcpu->stat.vmon_exits;
> /*
> * The Intel VMX Instruction Reference lists a bunch of bits that are
> * prerequisite to running VMXON, most notably cr4.VMXE must be set to
> @@ -4964,6 +4965,7 @@ static inline void nested_release_vmcs12(struct kvm_vcpu *vcpu)
> /* Emulate the VMXOFF instruction */
> static int handle_vmoff(struct kvm_vcpu *vcpu)
> {
> + ++vcpu->stat.vmoff_exits;
> if (!nested_vmx_check_permission(vcpu))
> return 1;
>
> @@ -4984,6 +4986,7 @@ static int handle_vmclear(struct kvm_vcpu *vcpu)
> u64 evmcs_gpa;
> int r;
>
> + ++vcpu->stat.vmclear_exits;
> if (!nested_vmx_check_permission(vcpu))
> return 1;
>
> @@ -5025,6 +5028,7 @@ static int handle_vmclear(struct kvm_vcpu *vcpu)
> /* Emulate the VMLAUNCH instruction */
> static int handle_vmlaunch(struct kvm_vcpu *vcpu)
> {
> + ++vcpu->stat.vmlaunch_exits;
> return nested_vmx_run(vcpu, true);
> }
>
> @@ -5032,6 +5036,7 @@ static int handle_vmlaunch(struct kvm_vcpu *vcpu)
> static int handle_vmresume(struct kvm_vcpu *vcpu)
> {
>
> + ++vcpu->stat.vmresume_exits;
> return nested_vmx_run(vcpu, false);
> }
>
> @@ -5049,6 +5054,8 @@ static int handle_vmread(struct kvm_vcpu *vcpu)
> short offset;
> int len, r;
>
> + ++vcpu->stat.vmread_exits;
> +
> if (!nested_vmx_check_permission(vcpu))
> return 1;
>
> @@ -5141,6 +5148,8 @@ static int handle_vmwrite(struct kvm_vcpu *vcpu)
> */
> u64 value = 0;
>
> + ++vcpu->stat.vmwrite_exits;
> +
> if (!nested_vmx_check_permission(vcpu))
> return 1;
>
> @@ -5245,6 +5254,8 @@ static int handle_vmptrld(struct kvm_vcpu *vcpu)
> gpa_t vmptr;
> int r;
>
> + ++vcpu->stat.vmptrld_exits;
> +
> if (!nested_vmx_check_permission(vcpu))
> return 1;
>
> @@ -5311,6 +5322,8 @@ static int handle_vmptrst(struct kvm_vcpu *vcpu)
> gva_t gva;
> int r;
>
> + ++vcpu->stat.vmptrst_exits;
> +
> if (!nested_vmx_check_permission(vcpu))
> return 1;
>
> @@ -5351,6 +5364,8 @@ static int handle_invept(struct kvm_vcpu *vcpu)
> } operand;
> int i, r;
>
> + ++vcpu->stat.invept_exits;
> +
> if (!(vmx->nested.msrs.secondary_ctls_high &
> SECONDARY_EXEC_ENABLE_EPT) ||
> !(vmx->nested.msrs.ept_caps & VMX_EPT_INVEPT_BIT)) {
> @@ -5431,6 +5446,8 @@ static int handle_invvpid(struct kvm_vcpu *vcpu)
> u16 vpid02;
> int r;
>
> + ++vcpu->stat.invvpid_exits;
> +
> if (!(vmx->nested.msrs.secondary_ctls_high &
> SECONDARY_EXEC_ENABLE_VPID) ||
> !(vmx->nested.msrs.vpid_caps & VMX_VPID_INVVPID_BIT)) {
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 1a00af1b076b..c2c95b4c1a68 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -277,7 +277,18 @@ const struct _kvm_stats_desc kvm_vcpu_stats_desc[] = {
> STATS_DESC_COUNTER(VCPU, nested_run),
> STATS_DESC_COUNTER(VCPU, directed_yield_attempted),
> STATS_DESC_COUNTER(VCPU, directed_yield_successful),
> - STATS_DESC_ICOUNTER(VCPU, guest_mode)
> + STATS_DESC_ICOUNTER(VCPU, guest_mode),
> + STATS_DESC_COUNTER(VCPU, vmclear_exits),
> + STATS_DESC_COUNTER(VCPU, vmlaunch_exits),
> + STATS_DESC_COUNTER(VCPU, vmptrld_exits),
> + STATS_DESC_COUNTER(VCPU, vmptrst_exits),
> + STATS_DESC_COUNTER(VCPU, vmread_exits),
> + STATS_DESC_COUNTER(VCPU, vmresume_exits),
> + STATS_DESC_COUNTER(VCPU, vmwrite_exits),
> + STATS_DESC_COUNTER(VCPU, vmoff_exits),
> + STATS_DESC_COUNTER(VCPU, vmon_exits),
> + STATS_DESC_COUNTER(VCPU, invept_exits),
> + STATS_DESC_COUNTER(VCPU, invvpid_exits),
> };
>
> const struct kvm_stats_header kvm_vcpu_stats_header = {
>
> base-commit: 47e7414d53fc12407b7a43bba412ecbf54c84f82
> --
> 2.33.0.rc2.250.ged5fa647cd-goog
>
next prev parent reply other threads:[~2021-08-19 18:43 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-19 18:00 [PATCH] KVM: stats: x86: vmx: add exit reason stats to vt-x instructions Jing Zhang
2021-08-19 18:43 ` David Matlack [this message]
2021-08-19 19:01 ` Jing Zhang
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='CALzav=cP17YXD8dRJnYFe_qmox3CTtpVBtLbU42Ei9zea2w21Q@mail.gmail.com' \
--to=dmatlack@google.com \
--cc=jingzhangos@google.com \
--cc=jmattson@google.com \
--cc=kvm@vger.kernel.org \
--cc=oupton@google.com \
--cc=pbonzini@redhat.com \
--cc=pfeiner@google.com \
--cc=pshier@google.com \
--cc=seanjc@google.com \
/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;
as well as URLs for NNTP newsgroup(s).