From 19298fbf453c90a6cf72288155f80c6f55e9139d Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:08:51 +0200
Subject: posix-cpu-timers: Provide quick sample function for itimer

get_itimer() needs a sample of the current thread group cputime. It invokes
thread_group_cputimer() - which is a misnomer. That function also starts
eventually the group cputime accouting which is bogus because the
accounting is already active when a timer is armed.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Link: https://lkml.kernel.org/r/20190821192919.599658199@linutronix.de
---
 include/linux/sched/cputime.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include/linux/sched')

diff --git a/include/linux/sched/cputime.h b/include/linux/sched/cputime.h
index 53f883f5a2fd..6de7b384d2be 100644
--- a/include/linux/sched/cputime.h
+++ b/include/linux/sched/cputime.h
@@ -62,7 +62,7 @@ extern void cputime_adjust(struct task_cputime *curr, struct prev_cputime *prev,
  */
 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
-
+void thread_group_sample_cputime(struct task_struct *tsk, struct task_cputime *times);
 
 /*
  * The following are functions that support scheduler-internal time accounting.
-- 
cgit v1.2.3-59-g8ed1b


From c506bef424ca282f2ad357e86fee940c69018974 Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:08:54 +0200
Subject: posix-cpu-timers: Rename thread_group_cputimer() and make it static

thread_group_cputimer() is a complete misnomer. The function does two things:

 - For arming process wide timers it makes sure that the atomic time
   storage is up to date. If no cpu timer is armed yet, then the atomic
   time storage is not updated by the scheduler for performance reasons.

   In that case a full summing up of all threads needs to be done and the
   update needs to be enabled.

- Samples the current time into the caller supplied storage.

Rename it to thread_group_start_cputime(), make it static and fixup the
callsite.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Link: https://lkml.kernel.org/r/20190821192919.869350319@linutronix.de
---
 include/linux/sched/cputime.h  |  1 -
 kernel/time/posix-cpu-timers.c | 17 +++++++++++++++--
 2 files changed, 15 insertions(+), 3 deletions(-)

(limited to 'include/linux/sched')

diff --git a/include/linux/sched/cputime.h b/include/linux/sched/cputime.h
index 6de7b384d2be..2638fd0ab3f2 100644
--- a/include/linux/sched/cputime.h
+++ b/include/linux/sched/cputime.h
@@ -61,7 +61,6 @@ extern void cputime_adjust(struct task_cputime *curr, struct prev_cputime *prev,
  * Thread group CPU time accounting.
  */
 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
-void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
 void thread_group_sample_cputime(struct task_struct *tsk, struct task_cputime *times);
 
 /*
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index cb736787145b..def225ae069a 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -253,7 +253,20 @@ void thread_group_sample_cputime(struct task_struct *tsk,
 	sample_cputime_atomic(times, &cputimer->cputime_atomic);
 }
 
-void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times)
+/**
+ * thread_group_start_cputime - Start cputime and return a sample
+ * @tsk:	Task for which cputime needs to be started
+ * @iimes:	Storage for time samples
+ *
+ * The thread group cputime accouting is avoided when there are no posix
+ * CPU timers armed. Before starting a timer it's required to check whether
+ * the time accounting is active. If not, a full update of the atomic
+ * accounting store needs to be done and the accounting enabled.
+ *
+ * Updates @times with an uptodate sample of the thread group cputimes.
+ */
+static void
+thread_group_start_cputime(struct task_struct *tsk, struct task_cputime *times)
 {
 	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
 	struct task_cputime sum;
@@ -536,7 +549,7 @@ static int cpu_timer_sample_group(const clockid_t which_clock,
 {
 	struct task_cputime cputime;
 
-	thread_group_cputimer(p, &cputime);
+	thread_group_start_cputime(p, &cputime);
 	switch (CPUCLOCK_WHICH(which_clock)) {
 	default:
 		return -EINVAL;
-- 
cgit v1.2.3-59-g8ed1b


From 2b69942f9021bf75bd1b001f53bd2578361fadf3 Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:09:04 +0200
Subject: posix-cpu-timers: Create a container struct

Per task/process data of posix CPU timers is all over the place which
makes the code hard to follow and requires ifdeffery.

Create a container to hold all this information in one place, so data is
consolidated and the ifdeffery can be confined to the posix timer header
file and removed from places like fork.

As a first step, move the cpu_timers list head array into the new struct
and clean up the initializers and simplify fork. The remaining #ifdef in
fork will be removed later.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Link: https://lkml.kernel.org/r/20190821192920.819418976@linutronix.de
---
 include/linux/init_task.h      | 11 -----------
 include/linux/posix-timers.h   | 34 ++++++++++++++++++++++++++++++++++
 include/linux/sched.h          |  4 +++-
 include/linux/sched/signal.h   |  5 +++--
 kernel/fork.c                  | 11 ++++-------
 kernel/time/posix-cpu-timers.c | 20 ++++++++++----------
 6 files changed, 54 insertions(+), 31 deletions(-)

(limited to 'include/linux/sched')

diff --git a/include/linux/init_task.h b/include/linux/init_task.h
index 6049baa5b8bc..2c620d7ac432 100644
--- a/include/linux/init_task.h
+++ b/include/linux/init_task.h
@@ -36,17 +36,6 @@ extern struct cred init_cred;
 #define INIT_PREV_CPUTIME(x)
 #endif
 
-#ifdef CONFIG_POSIX_TIMERS
-#define INIT_CPU_TIMERS(s)						\
-	.cpu_timers = {							\
-		LIST_HEAD_INIT(s.cpu_timers[0]),			\
-		LIST_HEAD_INIT(s.cpu_timers[1]),			\
-		LIST_HEAD_INIT(s.cpu_timers[2]),			\
-	},
-#else
-#define INIT_CPU_TIMERS(s)
-#endif
-
 #define INIT_TASK_COMM "swapper"
 
 /* Attach to the init_task data structure for proper alignment */
diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index 033374b99767..cdef89750b2c 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -62,6 +62,40 @@ static inline int clockid_to_fd(const clockid_t clk)
 	return ~(clk >> 3);
 }
 
+#ifdef CONFIG_POSIX_TIMERS
+/**
+ * posix_cputimers - Container for posix CPU timer related data
+ * @cpu_timers:		List heads to queue posix CPU timers
+ *
+ * Used in task_struct and signal_struct
+ */
+struct posix_cputimers {
+	struct list_head	cpu_timers[CPUCLOCK_MAX];
+};
+
+static inline void posix_cputimers_init(struct posix_cputimers *pct)
+{
+	INIT_LIST_HEAD(&pct->cpu_timers[0]);
+	INIT_LIST_HEAD(&pct->cpu_timers[1]);
+	INIT_LIST_HEAD(&pct->cpu_timers[2]);
+}
+
+/* Init task static initializer */
+#define INIT_CPU_TIMERLISTS(c)	{					\
+	LIST_HEAD_INIT(c.cpu_timers[0]),				\
+	LIST_HEAD_INIT(c.cpu_timers[1]),				\
+	LIST_HEAD_INIT(c.cpu_timers[2]),				\
+}
+
+#define INIT_CPU_TIMERS(s)						\
+	.posix_cputimers = {						\
+		.cpu_timers = INIT_CPU_TIMERLISTS(s.posix_cputimers),	\
+	},
+#else
+struct posix_cputimers { };
+#define INIT_CPU_TIMERS(s)
+#endif
+
 #define REQUEUE_PENDING 1
 
 /**
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 8dc1811487f5..fde844a3b86e 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -28,6 +28,7 @@
 #include <linux/signal_types.h>
 #include <linux/mm_types_task.h>
 #include <linux/task_io_accounting.h>
+#include <linux/posix-timers.h>
 #include <linux/rseq.h>
 
 /* task_struct member predeclarations (sorted alphabetically): */
@@ -878,8 +879,9 @@ struct task_struct {
 
 #ifdef CONFIG_POSIX_TIMERS
 	struct task_cputime		cputime_expires;
-	struct list_head		cpu_timers[3];
 #endif
+	/* Empty if CONFIG_POSIX_CPUTIMERS=n */
+	struct posix_cputimers		posix_cputimers;
 
 	/* Process credentials: */
 
diff --git a/include/linux/sched/signal.h b/include/linux/sched/signal.h
index efd8ce7675ed..88fbb3f1c375 100644
--- a/include/linux/sched/signal.h
+++ b/include/linux/sched/signal.h
@@ -9,6 +9,7 @@
 #include <linux/sched/task.h>
 #include <linux/cred.h>
 #include <linux/refcount.h>
+#include <linux/posix-timers.h>
 
 /*
  * Types defining task->signal and task->sighand and APIs using them:
@@ -151,9 +152,9 @@ struct signal_struct {
 	/* Earliest-expiration cache. */
 	struct task_cputime cputime_expires;
 
-	struct list_head cpu_timers[3];
-
 #endif
+	/* Empty if CONFIG_POSIX_TIMERS=n */
+	struct posix_cputimers posix_cputimers;
 
 	/* PID/PID hash table linkage. */
 	struct pid *pids[PIDTYPE_MAX];
diff --git a/kernel/fork.c b/kernel/fork.c
index d8ae0f1b4148..b6a135e4275b 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1523,6 +1523,7 @@ void __cleanup_sighand(struct sighand_struct *sighand)
  */
 static void posix_cpu_timers_init_group(struct signal_struct *sig)
 {
+	struct posix_cputimers *pct = &sig->posix_cputimers;
 	unsigned long cpu_limit;
 
 	cpu_limit = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur);
@@ -1531,10 +1532,7 @@ static void posix_cpu_timers_init_group(struct signal_struct *sig)
 		sig->cputimer.running = true;
 	}
 
-	/* The timer lists. */
-	INIT_LIST_HEAD(&sig->cpu_timers[0]);
-	INIT_LIST_HEAD(&sig->cpu_timers[1]);
-	INIT_LIST_HEAD(&sig->cpu_timers[2]);
+	posix_cputimers_init(pct);
 }
 #else
 static inline void posix_cpu_timers_init_group(struct signal_struct *sig) { }
@@ -1649,9 +1647,8 @@ static void posix_cpu_timers_init(struct task_struct *tsk)
 	tsk->cputime_expires.prof_exp = 0;
 	tsk->cputime_expires.virt_exp = 0;
 	tsk->cputime_expires.sched_exp = 0;
-	INIT_LIST_HEAD(&tsk->cpu_timers[0]);
-	INIT_LIST_HEAD(&tsk->cpu_timers[1]);
-	INIT_LIST_HEAD(&tsk->cpu_timers[2]);
+
+	posix_cputimers_init(&tsk->posix_cputimers);
 }
 #else
 static inline void posix_cpu_timers_init(struct task_struct *tsk) { }
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index b1c97664a730..849e2045fb6e 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -407,11 +407,11 @@ static void cleanup_timers_list(struct list_head *head)
  *
  * This must be called with the siglock held.
  */
-static void cleanup_timers(struct list_head *head)
+static void cleanup_timers(struct posix_cputimers *pct)
 {
-	cleanup_timers_list(head);
-	cleanup_timers_list(++head);
-	cleanup_timers_list(++head);
+	cleanup_timers_list(&pct->cpu_timers[CPUCLOCK_PROF]);
+	cleanup_timers_list(&pct->cpu_timers[CPUCLOCK_VIRT]);
+	cleanup_timers_list(&pct->cpu_timers[CPUCLOCK_SCHED]);
 }
 
 /*
@@ -421,11 +421,11 @@ static void cleanup_timers(struct list_head *head)
  */
 void posix_cpu_timers_exit(struct task_struct *tsk)
 {
-	cleanup_timers(tsk->cpu_timers);
+	cleanup_timers(&tsk->posix_cputimers);
 }
 void posix_cpu_timers_exit_group(struct task_struct *tsk)
 {
-	cleanup_timers(tsk->signal->cpu_timers);
+	cleanup_timers(&tsk->signal->posix_cputimers);
 }
 
 static inline int expires_gt(u64 expires, u64 new_exp)
@@ -446,10 +446,10 @@ static void arm_timer(struct k_itimer *timer)
 	struct cpu_timer_list *next;
 
 	if (CPUCLOCK_PERTHREAD(timer->it_clock)) {
-		head = p->cpu_timers;
+		head = p->posix_cputimers.cpu_timers;
 		cputime_expires = &p->cputime_expires;
 	} else {
-		head = p->signal->cpu_timers;
+		head = p->signal->posix_cputimers.cpu_timers;
 		cputime_expires = &p->signal->cputime_expires;
 	}
 	head += CPUCLOCK_WHICH(timer->it_clock);
@@ -773,8 +773,8 @@ static inline void check_dl_overrun(struct task_struct *tsk)
 static void check_thread_timers(struct task_struct *tsk,
 				struct list_head *firing)
 {
+	struct list_head *timers = tsk->posix_cputimers.cpu_timers;
 	struct task_cputime *tsk_expires = &tsk->cputime_expires;
-	struct list_head *timers = tsk->cpu_timers;
 	u64 expires, stime, utime;
 	unsigned long soft;
 
@@ -879,9 +879,9 @@ static void check_process_timers(struct task_struct *tsk,
 				 struct list_head *firing)
 {
 	struct signal_struct *const sig = tsk->signal;
+	struct list_head *timers = sig->posix_cputimers.cpu_timers;
 	u64 utime, ptime, virt_expires, prof_expires;
 	u64 sum_sched_runtime, sched_expires;
-	struct list_head *timers = sig->cpu_timers;
 	struct task_cputime cputime;
 	unsigned long soft;
 
-- 
cgit v1.2.3-59-g8ed1b


From 9eacb5c7e6607aba00a7322b21cad83fc8b101c8 Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:09:05 +0200
Subject: sched: Move struct task_cputime to types.h

For upcoming posix-timer changes to avoid include recursion hell.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lkml.kernel.org/r/20190821192920.909530418@linutronix.de
---
 include/linux/sched.h       | 17 +----------------
 include/linux/sched/types.h | 23 +++++++++++++++++++++++
 2 files changed, 24 insertions(+), 16 deletions(-)
 create mode 100644 include/linux/sched/types.h

(limited to 'include/linux/sched')

diff --git a/include/linux/sched.h b/include/linux/sched.h
index fde844a3b86e..37c39df9b186 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -25,6 +25,7 @@
 #include <linux/resource.h>
 #include <linux/latencytop.h>
 #include <linux/sched/prio.h>
+#include <linux/sched/types.h>
 #include <linux/signal_types.h>
 #include <linux/mm_types_task.h>
 #include <linux/task_io_accounting.h>
@@ -245,22 +246,6 @@ struct prev_cputime {
 #endif
 };
 
-/**
- * struct task_cputime - collected CPU time counts
- * @utime:		time spent in user mode, in nanoseconds
- * @stime:		time spent in kernel mode, in nanoseconds
- * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
- *
- * This structure groups together three kinds of CPU time that are tracked for
- * threads and thread groups.  Most things considering CPU time want to group
- * these counts together and treat all three of them in parallel.
- */
-struct task_cputime {
-	u64				utime;
-	u64				stime;
-	unsigned long long		sum_exec_runtime;
-};
-
 /* Alternate field names when used on cache expirations: */
 #define virt_exp			utime
 #define prof_exp			stime
diff --git a/include/linux/sched/types.h b/include/linux/sched/types.h
new file mode 100644
index 000000000000..2c5c28ddd9b2
--- /dev/null
+++ b/include/linux/sched/types.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_SCHED_TYPES_H
+#define _LINUX_SCHED_TYPES_H
+
+#include <linux/types.h>
+
+/**
+ * struct task_cputime - collected CPU time counts
+ * @utime:		time spent in user mode, in nanoseconds
+ * @stime:		time spent in kernel mode, in nanoseconds
+ * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
+ *
+ * This structure groups together three kinds of CPU time that are tracked for
+ * threads and thread groups.  Most things considering CPU time want to group
+ * these counts together and treat all three of them in parallel.
+ */
+struct task_cputime {
+	u64				utime;
+	u64				stime;
+	unsigned long long		sum_exec_runtime;
+};
+
+#endif
-- 
cgit v1.2.3-59-g8ed1b


From 3a245c0f110e2bfcf7f2cd2248a29005c78999e3 Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:09:06 +0200
Subject: posix-cpu-timers: Move expiry cache into struct posix_cputimers

The expiry cache belongs into the posix_cputimers container where the other
cpu timers information is.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Link: https://lkml.kernel.org/r/20190821192921.014444012@linutronix.de
---
 include/linux/posix-timers.h   | 22 +++++++++++++++++++++
 include/linux/sched.h          |  8 --------
 include/linux/sched/signal.h   |  3 ---
 kernel/fork.c                  | 25 +++--------------------
 kernel/sched/rt.c              |  6 ++++--
 kernel/time/posix-cpu-timers.c | 45 +++++++++++++++++++++++++-----------------
 6 files changed, 56 insertions(+), 53 deletions(-)

(limited to 'include/linux/sched')

diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index cdef89750b2c..a3731ba15bce 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -62,24 +62,43 @@ static inline int clockid_to_fd(const clockid_t clk)
 	return ~(clk >> 3);
 }
 
+/*
+ * Alternate field names for struct task_cputime when used on cache
+ * expirations. Will go away soon.
+ */
+#define virt_exp			utime
+#define prof_exp			stime
+#define sched_exp			sum_exec_runtime
+
 #ifdef CONFIG_POSIX_TIMERS
 /**
  * posix_cputimers - Container for posix CPU timer related data
+ * @cputime_expires:	Earliest-expiration cache
  * @cpu_timers:		List heads to queue posix CPU timers
  *
  * Used in task_struct and signal_struct
  */
 struct posix_cputimers {
+	struct task_cputime	cputime_expires;
 	struct list_head	cpu_timers[CPUCLOCK_MAX];
 };
 
 static inline void posix_cputimers_init(struct posix_cputimers *pct)
 {
+	memset(&pct->cputime_expires, 0, sizeof(pct->cputime_expires));
 	INIT_LIST_HEAD(&pct->cpu_timers[0]);
 	INIT_LIST_HEAD(&pct->cpu_timers[1]);
 	INIT_LIST_HEAD(&pct->cpu_timers[2]);
 }
 
+void posix_cputimers_group_init(struct posix_cputimers *pct, u64 cpu_limit);
+
+static inline void posix_cputimers_rt_watchdog(struct posix_cputimers *pct,
+					       u64 runtime)
+{
+	pct->cputime_expires.sched_exp = runtime;
+}
+
 /* Init task static initializer */
 #define INIT_CPU_TIMERLISTS(c)	{					\
 	LIST_HEAD_INIT(c.cpu_timers[0]),				\
@@ -94,6 +113,9 @@ static inline void posix_cputimers_init(struct posix_cputimers *pct)
 #else
 struct posix_cputimers { };
 #define INIT_CPU_TIMERS(s)
+static inline void posix_cputimers_init(struct posix_cputimers *pct) { }
+static inline void posix_cputimers_group_init(struct posix_cputimers *pct,
+					      u64 cpu_limit) { }
 #endif
 
 #define REQUEUE_PENDING 1
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 37c39df9b186..8cc8e323093f 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -246,11 +246,6 @@ struct prev_cputime {
 #endif
 };
 
-/* Alternate field names when used on cache expirations: */
-#define virt_exp			utime
-#define prof_exp			stime
-#define sched_exp			sum_exec_runtime
-
 enum vtime_state {
 	/* Task is sleeping or running in a CPU with VTIME inactive: */
 	VTIME_INACTIVE = 0,
@@ -862,9 +857,6 @@ struct task_struct {
 	unsigned long			min_flt;
 	unsigned long			maj_flt;
 
-#ifdef CONFIG_POSIX_TIMERS
-	struct task_cputime		cputime_expires;
-#endif
 	/* Empty if CONFIG_POSIX_CPUTIMERS=n */
 	struct posix_cputimers		posix_cputimers;
 
diff --git a/include/linux/sched/signal.h b/include/linux/sched/signal.h
index 88fbb3f1c375..729bd892ee45 100644
--- a/include/linux/sched/signal.h
+++ b/include/linux/sched/signal.h
@@ -149,9 +149,6 @@ struct signal_struct {
 	 */
 	struct thread_group_cputimer cputimer;
 
-	/* Earliest-expiration cache. */
-	struct task_cputime cputime_expires;
-
 #endif
 	/* Empty if CONFIG_POSIX_TIMERS=n */
 	struct posix_cputimers posix_cputimers;
diff --git a/kernel/fork.c b/kernel/fork.c
index b6a135e4275b..52bfe7c20ff6 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1527,12 +1527,9 @@ static void posix_cpu_timers_init_group(struct signal_struct *sig)
 	unsigned long cpu_limit;
 
 	cpu_limit = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur);
-	if (cpu_limit != RLIM_INFINITY) {
-		sig->cputime_expires.prof_exp = cpu_limit * NSEC_PER_SEC;
+	posix_cputimers_group_init(pct, cpu_limit);
+	if (cpu_limit != RLIM_INFINITY)
 		sig->cputimer.running = true;
-	}
-
-	posix_cputimers_init(pct);
 }
 #else
 static inline void posix_cpu_timers_init_group(struct signal_struct *sig) { }
@@ -1638,22 +1635,6 @@ static void rt_mutex_init_task(struct task_struct *p)
 #endif
 }
 
-#ifdef CONFIG_POSIX_TIMERS
-/*
- * Initialize POSIX timer handling for a single task.
- */
-static void posix_cpu_timers_init(struct task_struct *tsk)
-{
-	tsk->cputime_expires.prof_exp = 0;
-	tsk->cputime_expires.virt_exp = 0;
-	tsk->cputime_expires.sched_exp = 0;
-
-	posix_cputimers_init(&tsk->posix_cputimers);
-}
-#else
-static inline void posix_cpu_timers_init(struct task_struct *tsk) { }
-#endif
-
 static inline void init_task_pid_links(struct task_struct *task)
 {
 	enum pid_type type;
@@ -1932,7 +1913,7 @@ static __latent_entropy struct task_struct *copy_process(
 	task_io_accounting_init(&p->ioac);
 	acct_clear_integrals(p);
 
-	posix_cpu_timers_init(p);
+	posix_cputimers_init(&p->posix_cputimers);
 
 	p->io_context = NULL;
 	audit_set_context(p, NULL);
diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c
index da3e85e61013..d6678f773c96 100644
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -2305,8 +2305,10 @@ static void watchdog(struct rq *rq, struct task_struct *p)
 		}
 
 		next = DIV_ROUND_UP(min(soft, hard), USEC_PER_SEC/HZ);
-		if (p->rt.timeout > next)
-			p->cputime_expires.sched_exp = p->se.sum_exec_runtime;
+		if (p->rt.timeout > next) {
+			posix_cputimers_rt_watchdog(&p->posix_cputimers,
+						    p->se.sum_exec_runtime);
+		}
 	}
 }
 #else
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index 849e2045fb6e..3e29d1692437 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -20,11 +20,18 @@
 
 static void posix_cpu_timer_rearm(struct k_itimer *timer);
 
+void posix_cputimers_group_init(struct posix_cputimers *pct, u64 cpu_limit)
+{
+	posix_cputimers_init(pct);
+	if (cpu_limit != RLIM_INFINITY)
+		pct->cputime_expires.prof_exp = cpu_limit * NSEC_PER_SEC;
+}
+
 /*
  * Called after updating RLIMIT_CPU to run cpu timer and update
- * tsk->signal->cputime_expires expiration cache if necessary. Needs
- * siglock protection since other code may update expiration cache as
- * well.
+ * tsk->signal->posix_cputimers.cputime_expires expiration cache if
+ * necessary. Needs siglock protection since other code may update
+ * expiration cache as well.
  */
 void update_rlimit_cpu(struct task_struct *task, unsigned long rlim_new)
 {
@@ -447,10 +454,10 @@ static void arm_timer(struct k_itimer *timer)
 
 	if (CPUCLOCK_PERTHREAD(timer->it_clock)) {
 		head = p->posix_cputimers.cpu_timers;
-		cputime_expires = &p->cputime_expires;
+		cputime_expires = &p->posix_cputimers.cputime_expires;
 	} else {
 		head = p->signal->posix_cputimers.cpu_timers;
-		cputime_expires = &p->signal->cputime_expires;
+		cputime_expires = &p->signal->posix_cputimers.cputime_expires;
 	}
 	head += CPUCLOCK_WHICH(timer->it_clock);
 
@@ -774,7 +781,7 @@ static void check_thread_timers(struct task_struct *tsk,
 				struct list_head *firing)
 {
 	struct list_head *timers = tsk->posix_cputimers.cpu_timers;
-	struct task_cputime *tsk_expires = &tsk->cputime_expires;
+	struct task_cputime *tsk_expires = &tsk->posix_cputimers.cputime_expires;
 	u64 expires, stime, utime;
 	unsigned long soft;
 
@@ -785,7 +792,7 @@ static void check_thread_timers(struct task_struct *tsk,
 	 * If cputime_expires is zero, then there are no active
 	 * per thread CPU timers.
 	 */
-	if (task_cputime_zero(&tsk->cputime_expires))
+	if (task_cputime_zero(tsk_expires))
 		return;
 
 	task_cputime(tsk, &utime, &stime);
@@ -954,10 +961,10 @@ static void check_process_timers(struct task_struct *tsk,
 			prof_expires = x;
 	}
 
-	sig->cputime_expires.prof_exp = prof_expires;
-	sig->cputime_expires.virt_exp = virt_expires;
-	sig->cputime_expires.sched_exp = sched_expires;
-	if (task_cputime_zero(&sig->cputime_expires))
+	sig->posix_cputimers.cputime_expires.prof_exp = prof_expires;
+	sig->posix_cputimers.cputime_expires.virt_exp = virt_expires;
+	sig->posix_cputimers.cputime_expires.sched_exp = sched_expires;
+	if (task_cputime_zero(&sig->posix_cputimers.cputime_expires))
 		stop_process_timers(sig);
 
 	sig->cputimer.checking_timer = false;
@@ -1058,12 +1065,13 @@ static inline int fastpath_timer_check(struct task_struct *tsk)
 {
 	struct signal_struct *sig;
 
-	if (!task_cputime_zero(&tsk->cputime_expires)) {
+	if (!task_cputime_zero(&tsk->posix_cputimers.cputime_expires)) {
 		struct task_cputime task_sample;
 
 		task_cputime(tsk, &task_sample.utime, &task_sample.stime);
 		task_sample.sum_exec_runtime = tsk->se.sum_exec_runtime;
-		if (task_cputime_expired(&task_sample, &tsk->cputime_expires))
+		if (task_cputime_expired(&task_sample,
+					 &tsk->posix_cputimers.cputime_expires))
 			return 1;
 	}
 
@@ -1088,7 +1096,8 @@ static inline int fastpath_timer_check(struct task_struct *tsk)
 
 		sample_cputime_atomic(&group_sample, &sig->cputimer.cputime_atomic);
 
-		if (task_cputime_expired(&group_sample, &sig->cputime_expires))
+		if (task_cputime_expired(&group_sample,
+					 &sig->posix_cputimers.cputime_expires))
 			return 1;
 	}
 
@@ -1204,12 +1213,12 @@ void set_process_cpu_timer(struct task_struct *tsk, unsigned int clock_idx,
 	 */
 	switch (clock_idx) {
 	case CPUCLOCK_PROF:
-		if (expires_gt(tsk->signal->cputime_expires.prof_exp, *newval))
-			tsk->signal->cputime_expires.prof_exp = *newval;
+		if (expires_gt(tsk->signal->posix_cputimers.cputime_expires.prof_exp, *newval))
+			tsk->signal->posix_cputimers.cputime_expires.prof_exp = *newval;
 		break;
 	case CPUCLOCK_VIRT:
-		if (expires_gt(tsk->signal->cputime_expires.virt_exp, *newval))
-			tsk->signal->cputime_expires.virt_exp = *newval;
+		if (expires_gt(tsk->signal->posix_cputimers.cputime_expires.virt_exp, *newval))
+			tsk->signal->posix_cputimers.cputime_expires.virt_exp = *newval;
 		break;
 	}
 
-- 
cgit v1.2.3-59-g8ed1b


From 11b8462f7e1d25f639c88949a2746a9c2667a766 Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:09:07 +0200
Subject: posix-cpu-timers: Provide array based access to expiry cache

Using struct task_cputime for the expiry cache is a pretty odd choice and
comes with magic defines to rename the fields for usage in the expiry
cache.

struct task_cputime is basically a u64 array with 3 members, but it has
distinct members.

The expiry cache content is different than the content of task_cputime
because

  expiry[PROF]  = task_cputime.stime + task_cputime.utime
  expiry[VIRT]  = task_cputime.utime
  expiry[SCHED] = task_cputime.sum_exec_runtime

So there is no direct mapping between task_cputime and the expiry cache and
the #define based remapping is just a horrible hack.

Having the expiry cache array based allows further simplification of the
expiry code.

To avoid an all in one cleanup which is hard to review add a temporary
anonymous union into struct task_cputime which allows array based access to
it. That requires to reorder the members. Add a build time sanity check to
validate that the members are at the same place.

The union and the build time checks will be removed after conversion.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Link: https://lkml.kernel.org/r/20190821192921.105793824@linutronix.de
---
 include/linux/posix-timers.h   | 24 ++++++++++++++++++------
 include/linux/sched/types.h    |  4 ++--
 kernel/time/posix-cpu-timers.c | 12 +++++++++++-
 3 files changed, 31 insertions(+), 9 deletions(-)

(limited to 'include/linux/sched')

diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index a3731ba15bce..ed0f6ec30aa6 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -65,27 +65,39 @@ static inline int clockid_to_fd(const clockid_t clk)
 /*
  * Alternate field names for struct task_cputime when used on cache
  * expirations. Will go away soon.
+ *
+ * stime corresponds to CLOCKCPU_PROF
+ * utime corresponds to CLOCKCPU_VIRT
+ * sum_exex_runtime corresponds to CLOCKCPU_SCHED
+ *
+ * The ordering is currently enforced so struct task_cputime and the
+ * expiries array in struct posix_cputimers are equivalent.
  */
-#define virt_exp			utime
 #define prof_exp			stime
+#define virt_exp			utime
 #define sched_exp			sum_exec_runtime
 
 #ifdef CONFIG_POSIX_TIMERS
 /**
  * posix_cputimers - Container for posix CPU timer related data
- * @cputime_expires:	Earliest-expiration cache
+ * @cputime_expires:	Earliest-expiration cache task_cputime based
+ * @expiries:		Earliest-expiration cache array based
  * @cpu_timers:		List heads to queue posix CPU timers
  *
  * Used in task_struct and signal_struct
  */
 struct posix_cputimers {
-	struct task_cputime	cputime_expires;
-	struct list_head	cpu_timers[CPUCLOCK_MAX];
+	/* Temporary union until all users are cleaned up */
+	union {
+		struct task_cputime	cputime_expires;
+		u64			expiries[CPUCLOCK_MAX];
+	};
+	struct list_head		cpu_timers[CPUCLOCK_MAX];
 };
 
 static inline void posix_cputimers_init(struct posix_cputimers *pct)
 {
-	memset(&pct->cputime_expires, 0, sizeof(pct->cputime_expires));
+	memset(&pct->expiries, 0, sizeof(pct->expiries));
 	INIT_LIST_HEAD(&pct->cpu_timers[0]);
 	INIT_LIST_HEAD(&pct->cpu_timers[1]);
 	INIT_LIST_HEAD(&pct->cpu_timers[2]);
@@ -96,7 +108,7 @@ void posix_cputimers_group_init(struct posix_cputimers *pct, u64 cpu_limit);
 static inline void posix_cputimers_rt_watchdog(struct posix_cputimers *pct,
 					       u64 runtime)
 {
-	pct->cputime_expires.sched_exp = runtime;
+	pct->expiries[CPUCLOCK_SCHED] = runtime;
 }
 
 /* Init task static initializer */
diff --git a/include/linux/sched/types.h b/include/linux/sched/types.h
index 2c5c28ddd9b2..3c3e049224ae 100644
--- a/include/linux/sched/types.h
+++ b/include/linux/sched/types.h
@@ -6,8 +6,8 @@
 
 /**
  * struct task_cputime - collected CPU time counts
- * @utime:		time spent in user mode, in nanoseconds
  * @stime:		time spent in kernel mode, in nanoseconds
+ * @utime:		time spent in user mode, in nanoseconds
  * @sum_exec_runtime:	total time spent on the CPU, in nanoseconds
  *
  * This structure groups together three kinds of CPU time that are tracked for
@@ -15,8 +15,8 @@
  * these counts together and treat all three of them in parallel.
  */
 struct task_cputime {
-	u64				utime;
 	u64				stime;
+	u64				utime;
 	unsigned long long		sum_exec_runtime;
 };
 
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index 3e29d1692437..a38b6d04e8b5 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -18,13 +18,23 @@
 
 #include "posix-timers.h"
 
+static inline void temporary_check(void)
+{
+	BUILD_BUG_ON(offsetof(struct task_cputime, stime) !=
+		     CPUCLOCK_PROF * sizeof(u64));
+	BUILD_BUG_ON(offsetof(struct task_cputime, utime) !=
+		     CPUCLOCK_VIRT * sizeof(u64));
+	BUILD_BUG_ON(offsetof(struct task_cputime, sum_exec_runtime) !=
+		     CPUCLOCK_SCHED * sizeof(u64));
+}
+
 static void posix_cpu_timer_rearm(struct k_itimer *timer);
 
 void posix_cputimers_group_init(struct posix_cputimers *pct, u64 cpu_limit)
 {
 	posix_cputimers_init(pct);
 	if (cpu_limit != RLIM_INFINITY)
-		pct->cputime_expires.prof_exp = cpu_limit * NSEC_PER_SEC;
+		pct->expiries[CPUCLOCK_PROF] = cpu_limit * NSEC_PER_SEC;
 }
 
 /*
-- 
cgit v1.2.3-59-g8ed1b


From b7be4ef1365dcb56fdffc6689e41058b23f5996d Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:09:16 +0200
Subject: posix-cpu-timers: Switch thread group sampling to array

That allows more simplifications in various places.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Link: https://lkml.kernel.org/r/20190821192921.988426956@linutronix.de
---
 include/linux/sched/cputime.h  |   2 +-
 kernel/time/itimer.c           |  11 ++---
 kernel/time/posix-cpu-timers.c | 104 ++++++++++++++++++-----------------------
 3 files changed, 49 insertions(+), 68 deletions(-)

(limited to 'include/linux/sched')

diff --git a/include/linux/sched/cputime.h b/include/linux/sched/cputime.h
index 2638fd0ab3f2..eefa5dff16b4 100644
--- a/include/linux/sched/cputime.h
+++ b/include/linux/sched/cputime.h
@@ -61,7 +61,7 @@ extern void cputime_adjust(struct task_cputime *curr, struct prev_cputime *prev,
  * Thread group CPU time accounting.
  */
 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
-void thread_group_sample_cputime(struct task_struct *tsk, struct task_cputime *times);
+void thread_group_sample_cputime(struct task_struct *tsk, u64 *samples);
 
 /*
  * The following are functions that support scheduler-internal time accounting.
diff --git a/kernel/time/itimer.c b/kernel/time/itimer.c
index ae04bc259240..77f1e5635cc1 100644
--- a/kernel/time/itimer.c
+++ b/kernel/time/itimer.c
@@ -55,15 +55,10 @@ static void get_cpu_itimer(struct task_struct *tsk, unsigned int clock_id,
 	val = it->expires;
 	interval = it->incr;
 	if (val) {
-		struct task_cputime cputime;
-		u64 t;
+		u64 t, samples[CPUCLOCK_MAX];
 
-		thread_group_sample_cputime(tsk, &cputime);
-		if (clock_id == CPUCLOCK_PROF)
-			t = cputime.utime + cputime.stime;
-		else
-			/* CPUCLOCK_VIRT */
-			t = cputime.utime;
+		thread_group_sample_cputime(tsk, samples);
+		t = samples[clock_id];
 
 		if (val < t)
 			/* about to fire */
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index 9ac601abc4c4..e62139a89375 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -225,22 +225,14 @@ retry:
 	}
 }
 
-static void update_gt_cputime(struct task_cputime_atomic *cputime_atomic, struct task_cputime *sum)
+static void update_gt_cputime(struct task_cputime_atomic *cputime_atomic,
+			      struct task_cputime *sum)
 {
 	__update_gt_cputime(&cputime_atomic->utime, sum->utime);
 	__update_gt_cputime(&cputime_atomic->stime, sum->stime);
 	__update_gt_cputime(&cputime_atomic->sum_exec_runtime, sum->sum_exec_runtime);
 }
 
-/* Sample task_cputime_atomic values in "atomic_timers", store results in "times". */
-static inline void sample_cputime_atomic(struct task_cputime *times,
-					 struct task_cputime_atomic *atomic_times)
-{
-	times->utime = atomic64_read(&atomic_times->utime);
-	times->stime = atomic64_read(&atomic_times->stime);
-	times->sum_exec_runtime = atomic64_read(&atomic_times->sum_exec_runtime);
-}
-
 /**
  * thread_group_sample_cputime - Sample cputime for a given task
  * @tsk:	Task for which cputime needs to be started
@@ -252,20 +244,19 @@ static inline void sample_cputime_atomic(struct task_cputime *times,
  *
  * Updates @times with an uptodate sample of the thread group cputimes.
  */
-void thread_group_sample_cputime(struct task_struct *tsk,
-				struct task_cputime *times)
+void thread_group_sample_cputime(struct task_struct *tsk, u64 *samples)
 {
 	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
 
 	WARN_ON_ONCE(!cputimer->running);
 
-	sample_cputime_atomic(times, &cputimer->cputime_atomic);
+	proc_sample_cputime_atomic(&cputimer->cputime_atomic, samples);
 }
 
 /**
  * thread_group_start_cputime - Start cputime and return a sample
  * @tsk:	Task for which cputime needs to be started
- * @iimes:	Storage for time samples
+ * @samples:	Storage for time samples
  *
  * The thread group cputime accouting is avoided when there are no posix
  * CPU timers armed. Before starting a timer it's required to check whether
@@ -274,14 +265,14 @@ void thread_group_sample_cputime(struct task_struct *tsk,
  *
  * Updates @times with an uptodate sample of the thread group cputimes.
  */
-static void
-thread_group_start_cputime(struct task_struct *tsk, struct task_cputime *times)
+static void thread_group_start_cputime(struct task_struct *tsk, u64 *samples)
 {
 	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
-	struct task_cputime sum;
 
 	/* Check if cputimer isn't running. This is accessed without locking. */
 	if (!READ_ONCE(cputimer->running)) {
+		struct task_cputime sum;
+
 		/*
 		 * The POSIX timer interface allows for absolute time expiry
 		 * values through the TIMER_ABSTIME flag, therefore we have
@@ -299,7 +290,15 @@ thread_group_start_cputime(struct task_struct *tsk, struct task_cputime *times)
 		 */
 		WRITE_ONCE(cputimer->running, true);
 	}
-	sample_cputime_atomic(times, &cputimer->cputime_atomic);
+	proc_sample_cputime_atomic(&cputimer->cputime_atomic, samples);
+}
+
+static void __thread_group_cputime(struct task_struct *tsk, u64 *samples)
+{
+	struct task_cputime ct;
+
+	thread_group_cputime(tsk, &ct);
+	store_samples(samples, ct.stime, ct.utime, ct.sum_exec_runtime);
 }
 
 /*
@@ -313,28 +312,18 @@ static u64 cpu_clock_sample_group(const clockid_t clkid, struct task_struct *p,
 				  bool start)
 {
 	struct thread_group_cputimer *cputimer = &p->signal->cputimer;
-	struct task_cputime cputime;
+	u64 samples[CPUCLOCK_MAX];
 
 	if (!READ_ONCE(cputimer->running)) {
 		if (start)
-			thread_group_start_cputime(p, &cputime);
+			thread_group_start_cputime(p, samples);
 		else
-			thread_group_cputime(p, &cputime);
+			__thread_group_cputime(p, samples);
 	} else {
-		sample_cputime_atomic(&cputime, &cputimer->cputime_atomic);
+		proc_sample_cputime_atomic(&cputimer->cputime_atomic, samples);
 	}
 
-	switch (clkid) {
-	case CPUCLOCK_PROF:
-		return cputime.utime + cputime.stime;
-	case CPUCLOCK_VIRT:
-		return cputime.utime;
-	case CPUCLOCK_SCHED:
-		return cputime.sum_exec_runtime;
-	default:
-		WARN_ON_ONCE(1);
-	}
-	return 0;
+	return samples[clkid];
 }
 
 static int posix_cpu_clock_get(const clockid_t clock, struct timespec64 *tp)
@@ -889,9 +878,7 @@ static void check_process_timers(struct task_struct *tsk,
 {
 	struct signal_struct *const sig = tsk->signal;
 	struct posix_cputimer_base *base = sig->posix_cputimers.bases;
-	u64 utime, ptime, virt_expires, prof_expires;
-	u64 sum_sched_runtime, sched_expires;
-	struct task_cputime cputime;
+	u64 virt_exp, prof_exp, sched_exp, samples[CPUCLOCK_MAX];
 	unsigned long soft;
 
 	/*
@@ -911,30 +898,29 @@ static void check_process_timers(struct task_struct *tsk,
 	 * Collect the current process totals. Group accounting is active
 	 * so the sample can be taken directly.
 	 */
-	sample_cputime_atomic(&cputime, &sig->cputimer.cputime_atomic);
-	utime = cputime.utime;
-	ptime = utime + cputime.stime;
-	sum_sched_runtime = cputime.sum_exec_runtime;
-
-	prof_expires = check_timers_list(&base[CPUCLOCK_PROF].cpu_timers,
-					 firing, ptime);
-	virt_expires = check_timers_list(&base[CPUCLOCK_VIRT].cpu_timers,
-					 firing, utime);
-	sched_expires = check_timers_list(&base[CPUCLOCK_SCHED].cpu_timers,
-					  firing, sum_sched_runtime);
+	proc_sample_cputime_atomic(&sig->cputimer.cputime_atomic, samples);
+
+	prof_exp = check_timers_list(&base[CPUCLOCK_PROF].cpu_timers,
+				     firing, samples[CPUCLOCK_PROF]);
+	virt_exp = check_timers_list(&base[CPUCLOCK_VIRT].cpu_timers,
+				     firing, samples[CPUCLOCK_VIRT]);
+	sched_exp = check_timers_list(&base[CPUCLOCK_SCHED].cpu_timers,
+				      firing, samples[CPUCLOCK_SCHED]);
 
 	/*
 	 * Check for the special case process timers.
 	 */
-	check_cpu_itimer(tsk, &sig->it[CPUCLOCK_PROF], &prof_expires, ptime,
-			 SIGPROF);
-	check_cpu_itimer(tsk, &sig->it[CPUCLOCK_VIRT], &virt_expires, utime,
-			 SIGVTALRM);
+	check_cpu_itimer(tsk, &sig->it[CPUCLOCK_PROF], &prof_exp,
+			 samples[CPUCLOCK_PROF], SIGPROF);
+	check_cpu_itimer(tsk, &sig->it[CPUCLOCK_VIRT], &virt_exp,
+			 samples[CPUCLOCK_PROF], SIGVTALRM);
+
 	soft = task_rlimit(tsk, RLIMIT_CPU);
 	if (soft != RLIM_INFINITY) {
-		unsigned long psecs = div_u64(ptime, NSEC_PER_SEC);
+		u64 softns, ptime = samples[CPUCLOCK_PROF];
 		unsigned long hard = task_rlimit_max(tsk, RLIMIT_CPU);
-		u64 x;
+		unsigned long psecs = div_u64(ptime, NSEC_PER_SEC);
+
 		if (psecs >= hard) {
 			/*
 			 * At the hard limit, we just die.
@@ -961,14 +947,14 @@ static void check_process_timers(struct task_struct *tsk,
 				sig->rlim[RLIMIT_CPU].rlim_cur = soft;
 			}
 		}
-		x = soft * NSEC_PER_SEC;
-		if (!prof_expires || x < prof_expires)
-			prof_expires = x;
+		softns = soft * NSEC_PER_SEC;
+		if (!prof_exp || softns < prof_exp)
+			prof_exp = softns;
 	}
 
-	base[CPUCLOCK_PROF].nextevt = prof_expires;
-	base[CPUCLOCK_VIRT].nextevt = virt_expires;
-	base[CPUCLOCK_SCHED].nextevt = sched_expires;
+	base[CPUCLOCK_PROF].nextevt = prof_exp;
+	base[CPUCLOCK_VIRT].nextevt = virt_exp;
+	base[CPUCLOCK_SCHED].nextevt = sched_exp;
 
 	if (expiry_cache_is_zero(&sig->posix_cputimers))
 		stop_process_timers(sig);
-- 
cgit v1.2.3-59-g8ed1b


From 244d49e30653658d4e7e9b2b8427777cbbc5affe Mon Sep 17 00:00:00 2001
From: Thomas Gleixner <tglx@linutronix.de>
Date: Wed, 21 Aug 2019 21:09:24 +0200
Subject: posix-cpu-timers: Move state tracking to struct posix_cputimers

Put it where it belongs and clean up the ifdeffery in fork completely.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lkml.kernel.org/r/20190821192922.743229404@linutronix.de
---
 include/linux/posix-timers.h   |  8 +++++
 include/linux/sched/cputime.h  |  9 ++++--
 include/linux/sched/signal.h   |  6 ----
 init/init_task.c               |  2 --
 kernel/fork.c                  |  6 ----
 kernel/time/posix-cpu-timers.c | 73 +++++++++++++++++++++++-------------------
 6 files changed, 54 insertions(+), 50 deletions(-)

(limited to 'include/linux/sched')

diff --git a/include/linux/posix-timers.h b/include/linux/posix-timers.h
index 3ea920e8fe7f..a9e3f69d2db4 100644
--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -77,15 +77,23 @@ struct posix_cputimer_base {
 /**
  * posix_cputimers - Container for posix CPU timer related data
  * @bases:		Base container for posix CPU clocks
+ * @timers_active:	Timers are queued.
+ * @expiry_active:	Timer expiry is active. Used for
+ *			process wide timers to avoid multiple
+ *			task trying to handle expiry concurrently
  *
  * Used in task_struct and signal_struct
  */
 struct posix_cputimers {
 	struct posix_cputimer_base	bases[CPUCLOCK_MAX];
+	unsigned int			timers_active;
+	unsigned int			expiry_active;
 };
 
 static inline void posix_cputimers_init(struct posix_cputimers *pct)
 {
+	pct->timers_active = 0;
+	pct->expiry_active = 0;
 	pct->bases[0].nextevt = U64_MAX;
 	pct->bases[1].nextevt = U64_MAX;
 	pct->bases[2].nextevt = U64_MAX;
diff --git a/include/linux/sched/cputime.h b/include/linux/sched/cputime.h
index eefa5dff16b4..6c9f19a33865 100644
--- a/include/linux/sched/cputime.h
+++ b/include/linux/sched/cputime.h
@@ -70,7 +70,7 @@ void thread_group_sample_cputime(struct task_struct *tsk, u64 *samples);
  */
 
 /**
- * get_running_cputimer - return &tsk->signal->cputimer if cputimer is running
+ * get_running_cputimer - return &tsk->signal->cputimer if cputimers are active
  *
  * @tsk:	Pointer to target task.
  */
@@ -80,8 +80,11 @@ struct thread_group_cputimer *get_running_cputimer(struct task_struct *tsk)
 {
 	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
 
-	/* Check if cputimer isn't running. This is accessed without locking. */
-	if (!READ_ONCE(cputimer->running))
+	/*
+	 * Check whether posix CPU timers are active. If not the thread
+	 * group accounting is not active either. Lockless check.
+	 */
+	if (!READ_ONCE(tsk->signal->posix_cputimers.timers_active))
 		return NULL;
 
 	/*
diff --git a/include/linux/sched/signal.h b/include/linux/sched/signal.h
index 729bd892ee45..88050259c466 100644
--- a/include/linux/sched/signal.h
+++ b/include/linux/sched/signal.h
@@ -57,18 +57,12 @@ struct task_cputime_atomic {
 /**
  * struct thread_group_cputimer - thread group interval timer counts
  * @cputime_atomic:	atomic thread group interval timers.
- * @running:		true when there are timers running and
- *			@cputime_atomic receives updates.
- * @checking_timer:	true when a thread in the group is in the
- *			process of checking for thread group timers.
  *
  * This structure contains the version of task_cputime, above, that is
  * used for thread group CPU timer calculations.
  */
 struct thread_group_cputimer {
 	struct task_cputime_atomic cputime_atomic;
-	bool running;
-	bool checking_timer;
 };
 
 struct multiprocess_signals {
diff --git a/init/init_task.c b/init/init_task.c
index 7ab773b9b3cd..d49692a0ec51 100644
--- a/init/init_task.c
+++ b/init/init_task.c
@@ -30,8 +30,6 @@ static struct signal_struct init_signals = {
 	.posix_timers = LIST_HEAD_INIT(init_signals.posix_timers),
 	.cputimer	= {
 		.cputime_atomic	= INIT_CPUTIME_ATOMIC,
-		.running	= false,
-		.checking_timer = false,
 	},
 #endif
 	INIT_CPU_TIMERS(init_signals)
diff --git a/kernel/fork.c b/kernel/fork.c
index 52bfe7c20ff6..f1228d9f0b11 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1517,7 +1517,6 @@ void __cleanup_sighand(struct sighand_struct *sighand)
 	}
 }
 
-#ifdef CONFIG_POSIX_TIMERS
 /*
  * Initialize POSIX timer handling for a thread group.
  */
@@ -1528,12 +1527,7 @@ static void posix_cpu_timers_init_group(struct signal_struct *sig)
 
 	cpu_limit = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur);
 	posix_cputimers_group_init(pct, cpu_limit);
-	if (cpu_limit != RLIM_INFINITY)
-		sig->cputimer.running = true;
 }
-#else
-static inline void posix_cpu_timers_init_group(struct signal_struct *sig) { }
-#endif
 
 static int copy_signal(unsigned long clone_flags, struct task_struct *tsk)
 {
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index ef39a7a4a95c..52f4c99c1d60 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -23,8 +23,10 @@ static void posix_cpu_timer_rearm(struct k_itimer *timer);
 void posix_cputimers_group_init(struct posix_cputimers *pct, u64 cpu_limit)
 {
 	posix_cputimers_init(pct);
-	if (cpu_limit != RLIM_INFINITY)
+	if (cpu_limit != RLIM_INFINITY) {
 		pct->bases[CPUCLOCK_PROF].nextevt = cpu_limit * NSEC_PER_SEC;
+		pct->timers_active = true;
+	}
 }
 
 /*
@@ -248,8 +250,9 @@ static void update_gt_cputime(struct task_cputime_atomic *cputime_atomic,
 void thread_group_sample_cputime(struct task_struct *tsk, u64 *samples)
 {
 	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
+	struct posix_cputimers *pct = &tsk->signal->posix_cputimers;
 
-	WARN_ON_ONCE(!cputimer->running);
+	WARN_ON_ONCE(!pct->timers_active);
 
 	proc_sample_cputime_atomic(&cputimer->cputime_atomic, samples);
 }
@@ -269,9 +272,10 @@ void thread_group_sample_cputime(struct task_struct *tsk, u64 *samples)
 static void thread_group_start_cputime(struct task_struct *tsk, u64 *samples)
 {
 	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
+	struct posix_cputimers *pct = &tsk->signal->posix_cputimers;
 
 	/* Check if cputimer isn't running. This is accessed without locking. */
-	if (!READ_ONCE(cputimer->running)) {
+	if (!READ_ONCE(pct->timers_active)) {
 		struct task_cputime sum;
 
 		/*
@@ -283,13 +287,13 @@ static void thread_group_start_cputime(struct task_struct *tsk, u64 *samples)
 		update_gt_cputime(&cputimer->cputime_atomic, &sum);
 
 		/*
-		 * We're setting cputimer->running without a lock. Ensure
-		 * this only gets written to in one operation. We set
-		 * running after update_gt_cputime() as a small optimization,
-		 * but barriers are not required because update_gt_cputime()
+		 * We're setting timers_active without a lock. Ensure this
+		 * only gets written to in one operation. We set it after
+		 * update_gt_cputime() as a small optimization, but
+		 * barriers are not required because update_gt_cputime()
 		 * can handle concurrent updates.
 		 */
-		WRITE_ONCE(cputimer->running, true);
+		WRITE_ONCE(pct->timers_active, true);
 	}
 	proc_sample_cputime_atomic(&cputimer->cputime_atomic, samples);
 }
@@ -313,9 +317,10 @@ static u64 cpu_clock_sample_group(const clockid_t clkid, struct task_struct *p,
 				  bool start)
 {
 	struct thread_group_cputimer *cputimer = &p->signal->cputimer;
+	struct posix_cputimers *pct = &p->signal->posix_cputimers;
 	u64 samples[CPUCLOCK_MAX];
 
-	if (!READ_ONCE(cputimer->running)) {
+	if (!READ_ONCE(pct->timers_active)) {
 		if (start)
 			thread_group_start_cputime(p, samples);
 		else
@@ -834,10 +839,10 @@ static void check_thread_timers(struct task_struct *tsk,
 
 static inline void stop_process_timers(struct signal_struct *sig)
 {
-	struct thread_group_cputimer *cputimer = &sig->cputimer;
+	struct posix_cputimers *pct = &sig->posix_cputimers;
 
-	/* Turn off cputimer->running. This is done without locking. */
-	WRITE_ONCE(cputimer->running, false);
+	/* Turn off the active flag. This is done without locking. */
+	WRITE_ONCE(pct->timers_active, false);
 	tick_dep_clear_signal(sig, TICK_DEP_BIT_POSIX_TIMER);
 }
 
@@ -877,17 +882,17 @@ static void check_process_timers(struct task_struct *tsk,
 	unsigned long soft;
 
 	/*
-	 * If cputimer is not running, then there are no active
-	 * process wide timers (POSIX 1.b, itimers, RLIMIT_CPU).
+	 * If there are no active process wide timers (POSIX 1.b, itimers,
+	 * RLIMIT_CPU) nothing to check.
 	 */
-	if (!READ_ONCE(sig->cputimer.running))
+	if (!READ_ONCE(pct->timers_active))
 		return;
 
        /*
 	 * Signify that a thread is checking for process timers.
 	 * Write access to this field is protected by the sighand lock.
 	 */
-	sig->cputimer.checking_timer = true;
+	pct->timers_active = true;
 
 	/*
 	 * Collect the current process totals. Group accounting is active
@@ -933,7 +938,7 @@ static void check_process_timers(struct task_struct *tsk,
 	if (expiry_cache_is_inactive(pct))
 		stop_process_timers(sig);
 
-	sig->cputimer.checking_timer = false;
+	pct->expiry_active = false;
 }
 
 /*
@@ -1027,39 +1032,41 @@ task_cputimers_expired(const u64 *sample, struct posix_cputimers *pct)
  */
 static inline bool fastpath_timer_check(struct task_struct *tsk)
 {
+	struct posix_cputimers *pct = &tsk->posix_cputimers;
 	struct signal_struct *sig;
 
-	if (!expiry_cache_is_inactive(&tsk->posix_cputimers)) {
+	if (!expiry_cache_is_inactive(pct)) {
 		u64 samples[CPUCLOCK_MAX];
 
 		task_sample_cputime(tsk, samples);
-		if (task_cputimers_expired(samples, &tsk->posix_cputimers))
+		if (task_cputimers_expired(samples, pct))
 			return true;
 	}
 
 	sig = tsk->signal;
+	pct = &sig->posix_cputimers;
 	/*
-	 * Check if thread group timers expired when the cputimer is
-	 * running and no other thread in the group is already checking
-	 * for thread group cputimers. These fields are read without the
-	 * sighand lock. However, this is fine because this is meant to
-	 * be a fastpath heuristic to determine whether we should try to
-	 * acquire the sighand lock to check/handle timers.
+	 * Check if thread group timers expired when timers are active and
+	 * no other thread in the group is already handling expiry for
+	 * thread group cputimers. These fields are read without the
+	 * sighand lock. However, this is fine because this is meant to be
+	 * a fastpath heuristic to determine whether we should try to
+	 * acquire the sighand lock to handle timer expiry.
 	 *
-	 * In the worst case scenario, if 'running' or 'checking_timer' gets
-	 * set but the current thread doesn't see the change yet, we'll wait
-	 * until the next thread in the group gets a scheduler interrupt to
-	 * handle the timer. This isn't an issue in practice because these
-	 * types of delays with signals actually getting sent are expected.
+	 * In the worst case scenario, if concurrently timers_active is set
+	 * or expiry_active is cleared, but the current thread doesn't see
+	 * the change yet, the timer checks are delayed until the next
+	 * thread in the group gets a scheduler interrupt to handle the
+	 * timer. This isn't an issue in practice because these types of
+	 * delays with signals actually getting sent are expected.
 	 */
-	if (READ_ONCE(sig->cputimer.running) &&
-	    !READ_ONCE(sig->cputimer.checking_timer)) {
+	if (READ_ONCE(pct->timers_active) && !READ_ONCE(pct->expiry_active)) {
 		u64 samples[CPUCLOCK_MAX];
 
 		proc_sample_cputime_atomic(&sig->cputimer.cputime_atomic,
 					   samples);
 
-		if (task_cputimers_expired(samples, &sig->posix_cputimers))
+		if (task_cputimers_expired(samples, pct))
 			return true;
 	}
 
-- 
cgit v1.2.3-59-g8ed1b