linux-kernel - [RFC PATCH 3/3] Introduce task's 'off cpu' time

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <1630107736-18269-4-git-send-email-prakash.sangappa@oracle.com>
Date:   Fri, 27 Aug 2021 16:42:16 -0700
From:   Prakash Sangappa <prakash.sangappa@...cle.com>
To:     linux-kernel@...r.kernel.org
Cc:     prakash.sangappa@...cle.com
Subject: [RFC PATCH 3/3] Introduce task's 'off cpu' time

Add a task's 'off cpu' time in nanoseconds to sched_info, that represents
accumulated time spent either on run queue or blocked in the kernel.
Publish the off cpu time thru the shared structure. This will be used by
an application to determine cpu time consumed(time executing on a cpu) as
accurately as possible, by taking elapsed time and subtracting off cpu
time.

Signed-off-by: Prakash Sangappa <prakash.sangappa@...cle.com>
---
 include/linux/sched.h            |  6 +++++
 include/linux/task_shared.h      |  2 ++
 include/uapi/linux/task_shared.h |  1 +
 kernel/sched/stats.h             | 56 ++++++++++++++++++++++++++--------------
 4 files changed, 45 insertions(+), 20 deletions(-)

diff --git a/include/linux/sched.h b/include/linux/sched.h
index 237aa21..a63e447 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -311,6 +311,12 @@ struct sched_info {
 	/* When were we last queued to run? */
 	unsigned long long		last_queued;
 
+	/* When did we last leave cpu */
+	unsigned long long		last_depart;
+
+	/* Time spent off cpu */
+	unsigned long long		off_cpu;
+
 #endif /* CONFIG_SCHED_INFO */
 };
 
diff --git a/include/linux/task_shared.h b/include/linux/task_shared.h
index 62793e4..ce475c4 100644
--- a/include/linux/task_shared.h
+++ b/include/linux/task_shared.h
@@ -70,6 +70,8 @@ struct task_ushrd_struct {
 		if (shrdp != NULL && shrdp->kaddr != NULL) {		\
 			shrdp->kaddr->ts.run_delay =			\
 				 t->sched_info.run_delay;		\
+			shrdp->kaddr->ts.off_cpu =			\
+				 t->sched_info.off_cpu;			\
 			if (p) {					\
 				shrdp->kaddr->ts.pcount =		\
 					 t->sched_info.pcount;		\
diff --git a/include/uapi/linux/task_shared.h b/include/uapi/linux/task_shared.h
index 06a8522..c867c09 100644
--- a/include/uapi/linux/task_shared.h
+++ b/include/uapi/linux/task_shared.h
@@ -18,5 +18,6 @@ struct task_schedstat {
 	volatile u64	sum_exec_runtime;
 	volatile u64	run_delay;
 	volatile u64	pcount;
+	volatile u64	off_cpu;
 };
 #endif
diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h
index 6b2d69c..ee59994 100644
--- a/kernel/sched/stats.h
+++ b/kernel/sched/stats.h
@@ -159,17 +159,24 @@ static inline void psi_sched_switch(struct task_struct *prev,
  */
 static inline void sched_info_dequeue(struct rq *rq, struct task_struct *t)
 {
-	unsigned long long delta = 0;
+	unsigned long long now = rq_clock(rq), delta = 0, ddelta = 0;
 
-	if (!t->sched_info.last_queued)
-		return;
+	if (t->sched_info.last_queued) {
+		delta = now - t->sched_info.last_queued;
+		t->sched_info.last_queued = 0;
+		t->sched_info.run_delay += delta;
+
+		rq_sched_info_dequeue(rq, delta);
+	}
 
-	delta = rq_clock(rq) - t->sched_info.last_queued;
-	t->sched_info.last_queued = 0;
-	t->sched_info.run_delay += delta;
-	task_update_runq_stat(t, 0);
+	if (t->sched_info.last_depart) {
+		ddelta = now - t->sched_info.last_depart;
+		t->sched_info.last_depart = 0;
+		t->sched_info.off_cpu += ddelta;
+	}
 
-	rq_sched_info_dequeue(rq, delta);
+	if (delta || ddelta)
+		task_update_runq_stat(t, 0);
 }
 
 /*
@@ -179,20 +186,27 @@ static inline void sched_info_dequeue(struct rq *rq, struct task_struct *t)
  */
 static void sched_info_arrive(struct rq *rq, struct task_struct *t)
 {
-	unsigned long long now, delta = 0;
+	unsigned long long now = rq_clock(rq), delta = 0, ddelta = 0;
 
-	if (!t->sched_info.last_queued)
-		return;
+	if (t->sched_info.last_queued) {
+		delta = now - t->sched_info.last_queued;
+		t->sched_info.last_queued = 0;
+		t->sched_info.run_delay += delta;
+		t->sched_info.last_arrival = now;
+		t->sched_info.pcount++;
+
+		rq_sched_info_arrive(rq, delta);
+	}
+
+	if (t->sched_info.last_depart) {
+		ddelta = now - t->sched_info.last_depart;
+		t->sched_info.last_depart = 0;
+		t->sched_info.off_cpu += ddelta;
+	}
 
-	now = rq_clock(rq);
-	delta = now - t->sched_info.last_queued;
-	t->sched_info.last_queued = 0;
-	t->sched_info.run_delay += delta;
-	t->sched_info.last_arrival = now;
-	t->sched_info.pcount++;
-	task_update_runq_stat(t, 1);
+	if (delta || ddelta)
+		task_update_runq_stat(t, 1);
 
-	rq_sched_info_arrive(rq, delta);
 }
 
 /*
@@ -216,10 +230,12 @@ static inline void sched_info_enqueue(struct rq *rq, struct task_struct *t)
  */
 static inline void sched_info_depart(struct rq *rq, struct task_struct *t)
 {
-	unsigned long long delta = rq_clock(rq) - t->sched_info.last_arrival;
+	unsigned long long delta, now = rq_clock(rq);
 
+	delta = now - t->sched_info.last_arrival;
 	rq_sched_info_depart(rq, delta);
 
+	t->sched_info.last_depart = now;
 	if (task_is_running(t))
 		sched_info_enqueue(rq, t);
 }
-- 
2.7.4