Date:   Wed, 8 Feb 2023 17:21:01 +0000
From:   Sean Christopherson <seanjc@google.com>
To:     Kevin Cheng <chengkev@google.com>
Cc:     Paolo Bonzini <pbonzini@redhat.com>, linux-kernel@vger.kernel.org,
        kvm@vger.kernel.org, linux-kselftest@vger.kernel.org,
        Shuah Khan <shuah@kernel.org>
Subject: Re: [PATCH] KVM: selftests: Added eBPF program and selftest to
 collect vmx exit stat
Message-ID: <Y+PZ/a9O7KgXdRf1@google.com>
References: <20230126004346.4101944-1-chengkev@google.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230126004346.4101944-1-chengkev@google.com>
Precedence: bulk

On Thu, Jan 26, 2023, Kevin Cheng wrote:
> diff --git a/tools/testing/selftests/kvm/kvm_vmx_exit_ebpf_kern.c b/tools/testing/selftests/kvm/kvm_vmx_exit_ebpf_kern.c
> new file mode 100644
> index 000000000000..b9c076f93171
> --- /dev/null
> +++ b/tools/testing/selftests/kvm/kvm_vmx_exit_ebpf_kern.c
> @@ -0,0 +1,73 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +#include <linux/bpf.h>
> +#include <stdint.h>
> +#include <bpf/bpf_helpers.h>
> +#include <bpf/bpf_tracing.h>
> +#include <bpf/bpf_core_read.h>
> +
> +struct kvm_vcpu {
> +	int vcpu_id;
> +};
> +
> +struct vmx_args {
> +	__u64 pad;
> +	unsigned int exit_reason;
> +	__u32 isa;
> +	struct kvm_vcpu *vcpu;
> +};
> +
> +struct stats_map_key {
> +	__u32 pid;
> +	__u32 vcpu_id;
> +	__u32 exit_reason;
> +};
> +
> +struct {
> +	__uint(type, BPF_MAP_TYPE_HASH);
> +	__uint(max_entries, 1024);
> +	__type(key, struct stats_map_key);
> +	__type(value, int);
> +} vmx_exit_map SEC(".maps");
> +
> +struct {
> +	__uint(type, BPF_MAP_TYPE_HASH);
> +	__uint(max_entries, 1);
> +	__type(key, __u32);
> +	__type(value, __u32);
> +} pid_map SEC(".maps");

Can you add comments to explain why maps are used?  From our internal discussions,
I think I know the answer, but it would be super helpful for others (and me) to
explain exactly what this code is doing, and more importantly, why.