Timekeeping items for 4.11

GIT PULL New John Stultz 1 participant 4 messages Jan 28, 2017

Subsystems (6)

MM core Scheduler exec and binfmt Proc filesystem +2 more subsystems

lore

Share & tools

lore

John Stultz

author

2017-01-28 01:08 UTC #0 lore raw

Hey Thomas, Ingo,
Just a few items I have queued for 4.11.

Let me know if you have any objections

thanks
-john

Cc Thomas Gleixner <[email protected]>

Cc Richard Cochran <[email protected]>

Cc Prarit Bhargava <[email protected]>

Cc Ingo Molnar <[email protected]>

The following changes since commit 0c744ea4f77d72b3dcebb7a8f2684633ec79be88:

Linux 4.10-rc2 (2017-01-01 14:31:53 -0800)

are available in the git repository at:

https://git.linaro.org/people/john.stultz/linux.git fortglx/4.11/time

for you to fetch changes up to b18b6a9cef7f30e9a8b7738d5fc8d568cf660855:

timers: Omit POSIX timer stuff from task_struct when disabled (2017-01-27 13:05:26 -0800)

----------------------------------------------------------------
Nicolas Pitre (1):
timers: Omit POSIX timer stuff from task_struct when disabled

Russell King (1):
delay: Add explanation of udelay() inaccuracy

Stephen Boyd (1):
timekeeping: Remove unused timekeeping_{get,set}_tai_offset()

 fs/proc/base.c            |  4 ++--
 include/linux/delay.h     | 11 +++++++++++
 include/linux/init_task.h | 40 +++++++++++++++++++++++++---------------
 include/linux/sched.h     | 13 ++++++++++---
 kernel/fork.c             | 10 +++++++++-
 kernel/sched/rt.c         |  4 ++++
 kernel/sched/stats.h      | 32 ++++++++++++++++++++------------
 kernel/time/timekeeping.c | 39 +--------------------------------------
 kernel/time/timekeeping.h |  2 --
 9 files changed, 82 insertions(+), 73 deletions(-)

Signature · 1 line

2.7.4

John Stultz

author

2017-01-28 01:07 UTC #1 lore raw

PATCH 1/3 timekeeping: Remove unused timekeeping_{get,set}_tai_offset()

From: Stephen Boyd <[email protected]>

The last caller to timekeeping_set_tai_offset() was in commit 0b5154fb9040 (timekeeping: Simplify tai updating from do_adjtimex, 2013-03-22) and the last caller to timekeeping_get_tai_offset() was in commit 76f4108892d9 (hrtimer:
Cleanup hrtimer accessors to the timekepeing state, 2014-07-16). Remove these unused functions now that we handle TAI offsets differently.

Cc Thomas Gleixner <[email protected]>

Cc Richard Cochran <[email protected]>

Cc Prarit Bhargava <[email protected]>

Cc Ingo Molnar <[email protected]>

Signed-off-by Stephen Boyd <[email protected]>

Signed-off-by John Stultz <[email protected]>

---
 kernel/time/timekeeping.c | 39 +--------------------------------------
 kernel/time/timekeeping.h |  2 --
 2 files changed, 1 insertion(+), 40 deletions(-)

kernel/time/timekeeping.c +1 ~~-38~~

index db087d7e1..95b258d 100644
@@ -1275,27 +1275,8 @@ int timekeeping_inject_offset(struct timespec *ts)
 }
 EXPORT_SYMBOL(timekeeping_inject_offset);
 
-
-/**
- * timekeeping_get_tai_offset - Returns current TAI offset from UTC
- *
- */
-s32 timekeeping_get_tai_offset(void)
-{
-	struct timekeeper *tk = &tk_core.timekeeper;
-	unsigned int seq;
-	s32 ret;
-
-	do {
-		seq = read_seqcount_begin(&tk_core.seq);
-		ret = tk->tai_offset;
-	} while (read_seqcount_retry(&tk_core.seq, seq));
-
-	return ret;
-}
-
 /**
- * __timekeeping_set_tai_offset - Lock free worker function
+ * __timekeeping_set_tai_offset - Sets the TAI offset from UTC and monotonic
  *
  */
 static void __timekeeping_set_tai_offset(struct timekeeper *tk, s32 tai_offset)

@@ -1305,24 +1286,6 @@ static void __timekeeping_set_tai_offset(struct timekeeper *tk, s32 tai_offset)
 }
 
 /**
- * timekeeping_set_tai_offset - Sets the current TAI offset from UTC
- *
- */
-void timekeeping_set_tai_offset(s32 tai_offset)
-{
-	struct timekeeper *tk = &tk_core.timekeeper;
-	unsigned long flags;
-
-	raw_spin_lock_irqsave(&timekeeper_lock, flags);
-	write_seqcount_begin(&tk_core.seq);
-	__timekeeping_set_tai_offset(tk, tai_offset);
-	timekeeping_update(tk, TK_MIRROR | TK_CLOCK_WAS_SET);
-	write_seqcount_end(&tk_core.seq);
-	raw_spin_unlock_irqrestore(&timekeeper_lock, flags);
-	clock_was_set();
-}
-
-/**
  * change_clocksource - Swaps clocksources if a new one is available
  *
  * Accumulates current time interval and initializes new clocksource

kernel/time/timekeeping.h +0 -2

index 704f595..d091467 100644
@@ -11,8 +11,6 @@ extern ktime_t ktime_get_update_offsets_now(unsigned int *cwsseq,
 extern int timekeeping_valid_for_hres(void);
 extern u64 timekeeping_max_deferment(void);
 extern int timekeeping_inject_offset(struct timespec *ts);
-extern s32 timekeeping_get_tai_offset(void);
-extern void timekeeping_set_tai_offset(s32 tai_offset);
 extern int timekeeping_suspend(void);
 extern void timekeeping_resume(void);

Signature · 1 line

2.7.4

John Stultz

author

2017-01-28 01:07 UTC #2 lore raw

PATCH 2/3 delay: Add explanation of udelay() inaccuracy

In reply to John Stultz

From: Russell King <[email protected]>

There seems to be some misunderstanding that udelay() and friends will always guarantee the specified delay. This is a false understanding. When udelay() is based on CPU cycles, it can return early for many reasons which are detailed by Linus' reply to me in a thread in 2011:

http://lists.openwall.net/linux-kernel/2011/01/12/372

However, a udelay test module was created in 2014 which allows udelay() to only be 0.5% fast, which is outside of the CPU-cycles udelay() results I measured back in 2011, which were deemed to be in the "we don't care" region.

test_udelay() should be fixed to reflect the real allowable tolerance on udelay(), rather than 0.5%.

Cc David Riley <[email protected]>

Cc Thomas Gleixner <[email protected]>

Cc Richard Cochran <[email protected]>

Cc Prarit Bhargava <[email protected]>

Cc Ingo Molnar <[email protected]>

Signed-off-by Russell King <[email protected]>

Signed-off-by John Stultz <[email protected]>

---
 include/linux/delay.h | 11 +++++++++++
 1 file changed, 11 insertions(+)

include/linux/delay.h +11 -0

index a6ecb34..2ecb3c4 100644
@@ -5,6 +5,17 @@
  * Copyright (C) 1993 Linus Torvalds
  *
  * Delay routines, using a pre-computed "loops_per_jiffy" value.
+ *
+ * Please note that ndelay(), udelay() and mdelay() may return early for
+ * several reasons:
+ *  1. computed loops_per_jiffy too low (due to the time taken to
+ *     execute the timer interrupt.)
+ *  2. cache behaviour affecting the time it takes to execute the
+ *     loop function.
+ *  3. CPU clock rate changes.
+ *
+ * Please see this thread:
+ *   http://lists.openwall.net/linux-kernel/2011/01/09/56
  */
 
 #include <linux/kernel.h>

Signature · 1 line

2.7.4

John Stultz

author

2017-01-28 01:08 UTC #3 lore raw

PATCH 3/3 timers: Omit POSIX timer stuff from task_struct when disabled

In reply to John Stultz

From: Nicolas Pitre <[email protected]>

When CONFIG_POSIX_TIMERS is disabled, it is preferable to remove related structures from struct task_struct and struct signal_struct as they won't contain anything useful and shouldn't be relied upon by mistake. Code still referencing those structures is also disabled here.

Cc Thomas Gleixner <[email protected]>

Cc Richard Cochran <[email protected]>

Cc Prarit Bhargava <[email protected]>

Cc Ingo Molnar <[email protected]>

Signed-off-by Nicolas Pitre <[email protected]>

Signed-off-by John Stultz <[email protected]>

---
 fs/proc/base.c            |  4 ++--
 include/linux/init_task.h | 40 +++++++++++++++++++++++++---------------
 include/linux/sched.h     | 13 ++++++++++---
 kernel/fork.c             | 10 +++++++++-
 kernel/sched/rt.c         |  4 ++++
 kernel/sched/stats.h      | 32 ++++++++++++++++++++------------
 6 files changed, 70 insertions(+), 33 deletions(-)

fs/proc/base.c +2 -2

index 8e7e61b..03deeac 100644
@@ -2179,7 +2179,7 @@ static const struct file_operations proc_map_files_operations = {
 	.llseek		= generic_file_llseek,
 };
 
-#ifdef CONFIG_CHECKPOINT_RESTORE
+#if defined(CONFIG_CHECKPOINT_RESTORE) && defined(CONFIG_POSIX_TIMERS)
 struct timers_private {
 	struct pid *pid;
 	struct task_struct *task;

@@ -2936,7 +2936,7 @@ static const struct pid_entry tgid_base_stuff[] = {
 	REG("projid_map", S_IRUGO|S_IWUSR, proc_projid_map_operations),
 	REG("setgroups",  S_IRUGO|S_IWUSR, proc_setgroups_operations),
 #endif
-#ifdef CONFIG_CHECKPOINT_RESTORE
+#if defined(CONFIG_CHECKPOINT_RESTORE) && defined(CONFIG_POSIX_TIMERS)
 	REG("timers",	  S_IRUGO, proc_timers_operations),
 #endif
 	REG("timerslack_ns", S_IRUGO|S_IWUGO, proc_pid_set_timerslack_ns_operations),

include/linux/init_task.h +25 ~~-15~~

index 325f649..3a85d61 100644
@@ -42,6 +42,27 @@ extern struct fs_struct init_fs;
 #define INIT_PREV_CPUTIME(x)
 #endif
 
+#ifdef CONFIG_POSIX_TIMERS
+#define INIT_POSIX_TIMERS(s)						\
+	.posix_timers = LIST_HEAD_INIT(s.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]),								\
+	},
+#define INIT_CPUTIMER(s)						\
+	.cputimer	= { 						\
+		.cputime_atomic	= INIT_CPUTIME_ATOMIC,			\
+		.running	= false,				\
+		.checking_timer = false,				\
+	},
+#else
+#define INIT_POSIX_TIMERS(s)
+#define INIT_CPU_TIMERS(s)
+#define INIT_CPUTIMER(s)
+#endif
+
 #define INIT_SIGNALS(sig) {						\
 	.nr_threads	= 1,						\
 	.thread_head	= LIST_HEAD_INIT(init_task.thread_node),	\

@@ -49,14 +70,10 @@ extern struct fs_struct init_fs;
 	.shared_pending	= { 						\
 		.list = LIST_HEAD_INIT(sig.shared_pending.list),	\
 		.signal =  {{0}}},					\
-	.posix_timers	 = LIST_HEAD_INIT(sig.posix_timers),		\
-	.cpu_timers	= INIT_CPU_TIMERS(sig.cpu_timers),		\
+	INIT_POSIX_TIMERS(sig)						\
+	INIT_CPU_TIMERS(sig)						\
 	.rlim		= INIT_RLIMITS,					\
-	.cputimer	= { 						\
-		.cputime_atomic	= INIT_CPUTIME_ATOMIC,			\
-		.running	= false,				\
-		.checking_timer = false,				\
-	},								\
+	INIT_CPUTIMER(sig)						\
 	INIT_PREV_CPUTIME(sig)						\
 	.cred_guard_mutex =						\
 		 __MUTEX_INITIALIZER(sig.cred_guard_mutex),		\

@@ -247,7 +264,7 @@ extern struct task_group root_task_group;
 	.blocked	= {{0}},					\
 	.alloc_lock	= __SPIN_LOCK_UNLOCKED(tsk.alloc_lock),		\
 	.journal_info	= NULL,						\
-	.cpu_timers	= INIT_CPU_TIMERS(tsk.cpu_timers),		\
+	INIT_CPU_TIMERS(tsk)						\
 	.pi_lock	= __RAW_SPIN_LOCK_UNLOCKED(tsk.pi_lock),	\
 	.timer_slack_ns = 50000, /* 50 usec default slack */		\
 	.pids = {							\

@@ -274,13 +291,6 @@ extern struct task_group root_task_group;
 }
 
 
-#define INIT_CPU_TIMERS(cpu_timers)					\
-{									\
-	LIST_HEAD_INIT(cpu_timers[0]),					\
-	LIST_HEAD_INIT(cpu_timers[1]),					\
-	LIST_HEAD_INIT(cpu_timers[2]),					\
-}
-
 /* Attach to the init_task data structure for proper alignment */
 #define __init_task_data __attribute__((__section__(".data..init_task")))

include/linux/sched.h +10 -3

index 4d19052..e8f6af5 100644
@@ -734,13 +734,14 @@ struct signal_struct {
 	unsigned int		is_child_subreaper:1;
 	unsigned int		has_child_subreaper:1;
 
+#ifdef CONFIG_POSIX_TIMERS
+
 	/* POSIX.1b Interval Timers */
 	int			posix_timer_id;
 	struct list_head	posix_timers;
 
 	/* ITIMER_REAL timer for the process */
 	struct hrtimer real_timer;
-	struct pid *leader_pid;
 	ktime_t it_real_incr;
 
 	/*

@@ -759,12 +760,16 @@ struct signal_struct {
 	/* Earliest-expiration cache. */
 	struct task_cputime cputime_expires;
 
+	struct list_head cpu_timers[3];
+
+#endif
+
+	struct pid *leader_pid;
+
 #ifdef CONFIG_NO_HZ_FULL
 	atomic_t tick_dep_mask;
 #endif
 
-	struct list_head cpu_timers[3];
-
 	struct pid *tty_old_pgrp;
 
 	/* boolean value for session group leader */

@@ -1681,8 +1686,10 @@ struct task_struct {
 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
 	unsigned long min_flt, maj_flt;
 
+#ifdef CONFIG_POSIX_TIMERS
 	struct task_cputime cputime_expires;
 	struct list_head cpu_timers[3];
+#endif
 
 /* process credentials */
 	const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */

kernel/fork.c +9 -1

index 11c5c8a..105c667 100644
@@ -1304,6 +1304,7 @@ void __cleanup_sighand(struct sighand_struct *sighand)
 	}
 }
 
+#ifdef CONFIG_POSIX_TIMERS
 /*
  * Initialize POSIX timer handling for a thread group.
  */

@@ -1322,6 +1323,9 @@ static void posix_cpu_timers_init_group(struct signal_struct *sig)
 	INIT_LIST_HEAD(&sig->cpu_timers[1]);
 	INIT_LIST_HEAD(&sig->cpu_timers[2]);
 }
+#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)
 {

@@ -1346,11 +1350,11 @@ static int copy_signal(unsigned long clone_flags, struct task_struct *tsk)
 	init_waitqueue_head(&sig->wait_chldexit);
 	sig->curr_target = tsk;
 	init_sigpending(&sig->shared_pending);
-	INIT_LIST_HEAD(&sig->posix_timers);
 	seqlock_init(&sig->stats_lock);
 	prev_cputime_init(&sig->prev_cputime);
 
 #ifdef CONFIG_POSIX_TIMERS
+	INIT_LIST_HEAD(&sig->posix_timers);
 	hrtimer_init(&sig->real_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 	sig->real_timer.function = it_real_fn;
 #endif

@@ -1425,6 +1429,7 @@ static void rt_mutex_init_task(struct task_struct *p)
 #endif
 }
 
+#ifdef CONFIG_POSIX_TIMERS
 /*
  * Initialize POSIX timer handling for a single task.
  */

@@ -1437,6 +1442,9 @@ static void posix_cpu_timers_init(struct task_struct *tsk)
 	INIT_LIST_HEAD(&tsk->cpu_timers[1]);
 	INIT_LIST_HEAD(&tsk->cpu_timers[2]);
 }
+#else
+static inline void posix_cpu_timers_init(struct task_struct *tsk) { }
+#endif
 
 static inline void
 init_task_pid(struct task_struct *task, enum pid_type type, struct pid *pid)

kernel/sched/rt.c +4 -0

index 2516b8d..a688a82 100644
@@ -2246,6 +2246,7 @@ prio_changed_rt(struct rq *rq, struct task_struct *p, int oldprio)
 	}
 }
 
+#ifdef CONFIG_POSIX_TIMERS
 static void watchdog(struct rq *rq, struct task_struct *p)
 {
 	unsigned long soft, hard;

@@ -2267,6 +2268,9 @@ static void watchdog(struct rq *rq, struct task_struct *p)
 			p->cputime_expires.sched_exp = p->se.sum_exec_runtime;
 	}
 }
+#else
+static inline void watchdog(struct rq *rq, struct task_struct *p) { }
+#endif
 
 static void task_tick_rt(struct rq *rq, struct task_struct *p, int queued)
 {

kernel/sched/stats.h +20 ~~-12~~

index 34659a8..c69a987 100644
@@ -172,18 +172,19 @@ sched_info_switch(struct rq *rq,
  */
 
 /**
- * cputimer_running - return true if cputimer is running
+ * get_running_cputimer - return &tsk->signal->cputimer if cputimer is running
  *
  * @tsk:	Pointer to target task.
  */
-static inline bool cputimer_running(struct task_struct *tsk)
-
+#ifdef CONFIG_POSIX_TIMERS
+static inline
+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))
-		return false;
+		return NULL;
 
 	/*
 	 * After we flush the task's sum_exec_runtime to sig->sum_sched_runtime

@@ -200,10 +201,17 @@ static inline bool cputimer_running(struct task_struct *tsk)
 	 * clock delta is behind the expiring timer value.
 	 */
 	if (unlikely(!tsk->sighand))
-		return false;
+		return NULL;
 
-	return true;
+	return cputimer;
+}
+#else
+static inline
+struct thread_group_cputimer *get_running_cputimer(struct task_struct *tsk)
+{
+	return NULL;
 }
+#endif
 
 /**
  * account_group_user_time - Maintain utime for a thread group.

@@ -218,9 +226,9 @@ static inline bool cputimer_running(struct task_struct *tsk)
 static inline void account_group_user_time(struct task_struct *tsk,
 					   cputime_t cputime)
 {
-	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
+	struct thread_group_cputimer *cputimer = get_running_cputimer(tsk);
 
-	if (!cputimer_running(tsk))
+	if (!cputimer)
 		return;
 
 	atomic64_add(cputime, &cputimer->cputime_atomic.utime);

@@ -239,9 +247,9 @@ static inline void account_group_user_time(struct task_struct *tsk,
 static inline void account_group_system_time(struct task_struct *tsk,
 					     cputime_t cputime)
 {
-	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
+	struct thread_group_cputimer *cputimer = get_running_cputimer(tsk);
 
-	if (!cputimer_running(tsk))
+	if (!cputimer)
 		return;
 
 	atomic64_add(cputime, &cputimer->cputime_atomic.stime);

@@ -260,9 +268,9 @@ static inline void account_group_system_time(struct task_struct *tsk,
 static inline void account_group_exec_runtime(struct task_struct *tsk,
 					      unsigned long long ns)
 {
-	struct thread_group_cputimer *cputimer = &tsk->signal->cputimer;
+	struct thread_group_cputimer *cputimer = get_running_cputimer(tsk);
 
-	if (!cputimer_running(tsk))
+	if (!cputimer)
 		return;
 
 	atomic64_add(ns, &cputimer->cputime_atomic.sum_exec_runtime);

Signature · 1 line

2.7.4