Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed,  5 Aug 2020 17:14:29 -0700
In-Reply-To: <20200806001431.2072150-1-jwadams@google.com>
Message-Id: <20200806001431.2072150-6-jwadams@google.com>
Mime-Version: 1.0
References: <20200806001431.2072150-1-jwadams@google.com>
Subject: [RFC PATCH 5/7] core/metricfs: expose scheduler stat information
 through metricfs
From:   Jonathan Adams <jwadams@google.com>
To:     linux-kernel@vger.kernel.org
Cc:     kvm@vger.kernel.org, Paolo Bonzini <pbonzini@redhat.com>,
        Jim Mattson <jmattson@google.com>,
        David Rientjes <rientjes@google.com>,
        Jonathan Adams <jwadams@google.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Add metricfs support for displaying percpu scheduler counters.
The top directory is /sys/kernel/debug/metricfs/stat (analogous
to /proc/stat).  Then there is a subdirectory for each scheduler
stat.  For example:

    cat /sys/kernel/debug/metricfs/stat/user/values

Signed-off-by: Jonathan Adams <jwadams@google.com>

---

jwadams@google.com: rebased to 5.8-pre6
	This is work originally done by another engineer at
	google, who would rather not have their name associated with this
	patchset. They're okay with me sending it under my name.
---
 fs/proc/stat.c | 57 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 57 insertions(+)

diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 46b3293015fe..deb378507b0b 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -13,6 +13,7 @@
 #include <linux/irqnr.h>
 #include <linux/sched/cputime.h>
 #include <linux/tick.h>
+#include <linux/metricfs.h>
 
 #ifndef arch_irq_stat_cpu
 #define arch_irq_stat_cpu(cpu) 0
@@ -237,3 +238,59 @@ static int __init proc_stat_init(void)
 	return 0;
 }
 fs_initcall(proc_stat_init);
+
+#ifdef CONFIG_METRICFS
+#define METRICFS_ITEM(name, field, desc) \
+static void \
+metricfs_##name(struct metric_emitter *e, int cpu) \
+{ \
+	int64_t v = kcpustat_field(&kcpustat_cpu(cpu), field, cpu); \
+	METRIC_EMIT_PERCPU_INT(e, cpu, v); \
+} \
+METRIC_EXPORT_PERCPU_COUNTER(name, desc, metricfs_##name)
+
+#define METRICFS_FUNC_ITEM(name, func, desc) \
+static void \
+metricfs_##name(struct metric_emitter *e, int cpu) \
+{ \
+	struct kernel_cpustat cpustat; \
+	int64_t v; \
+	kcpustat_cpu_fetch(&cpustat, cpu); \
+	v = func(&cpustat, cpu); \
+	METRIC_EMIT_PERCPU_INT(e, cpu, v); \
+} \
+METRIC_EXPORT_PERCPU_COUNTER(name, desc, metricfs_##name)
+
+METRICFS_ITEM(user, CPUTIME_USER, "time in user mode (nsec)");
+METRICFS_ITEM(nice, CPUTIME_NICE, "time in user mode niced (nsec)");
+METRICFS_ITEM(system, CPUTIME_SYSTEM, "time in system calls (nsec)");
+METRICFS_ITEM(irq, CPUTIME_IRQ, "time in interrupts (nsec)");
+METRICFS_ITEM(softirq, CPUTIME_SOFTIRQ, "time in softirqs (nsec)");
+METRICFS_ITEM(steal, CPUTIME_STEAL, "time in involuntary wait (nsec)");
+METRICFS_ITEM(guest, CPUTIME_GUEST, "time in guest mode (nsec)");
+METRICFS_ITEM(guest_nice, CPUTIME_GUEST_NICE,
+	"time in guest mode niced (nsec)");
+METRICFS_FUNC_ITEM(idle, get_idle_time, "time in idle (nsec)");
+METRICFS_FUNC_ITEM(iowait, get_iowait_time, "time in iowait (nsec)");
+
+static int __init init_stat_metricfs(void)
+{
+	struct metricfs_subsys *subsys;
+
+	subsys = metricfs_create_subsys("stat", NULL);
+	metric_init_user(subsys);
+	metric_init_nice(subsys);
+	metric_init_system(subsys);
+	metric_init_irq(subsys);
+	metric_init_softirq(subsys);
+	metric_init_steal(subsys);
+	metric_init_guest(subsys);
+	metric_init_guest_nice(subsys);
+	metric_init_idle(subsys);
+	metric_init_iowait(subsys);
+
+	return 0;
+}
+module_init(init_stat_metricfs);
+
+#endif
-- 
2.28.0.236.gb10cc79966-goog