1 files changed, 241 insertions, 120 deletions

diff --git a/kernel/time/sched_clock.c b/kernel/time/sched_clock.c
index a326f27d7f09..f39111830ca3 100644
--- a/kernel/time/sched_clock.c
+++ b/kernel/time/sched_clock.c
@@ -1,212 +1,333 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
- * sched_clock.c: support for extending counters to full 64-bit ns counter
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
+ * Generic sched_clock() support, to extend low level hardware time
+ * counters to full 64-bit ns values.
  */
 #include <linux/clocksource.h>
 #include <linux/init.h>
 #include <linux/jiffies.h>
+#include <linux/ktime.h>
 #include <linux/kernel.h>
+#include <linux/math.h>
 #include <linux/moduleparam.h>
 #include <linux/sched.h>
+#include <linux/sched/clock.h>
 #include <linux/syscore_ops.h>
-#include <linux/timer.h>
+#include <linux/hrtimer.h>
 #include <linux/sched_clock.h>
+#include <linux/seqlock.h>
+#include <linux/bitops.h>
+
+#include "timekeeping.h"
 
+/**
+ * struct clock_data - all data needed for sched_clock() (including
+ *                     registration of a new clock source)
+ *
+ * @seq:		Sequence counter for protecting updates. The lowest
+ *			bit is the index for @read_data.
+ * @read_data:		Data required to read from sched_clock.
+ * @wrap_kt:		Duration for which clock can run before wrapping.
+ * @rate:		Tick rate of the registered clock.
+ * @actual_read_sched_clock: Registered hardware level clock read function.
+ *
+ * The ordering of this structure has been chosen to optimize cache
+ * performance. In particular 'seq' and 'read_data[0]' (combined) should fit
+ * into a single 64-byte cache line.
+ */
 struct clock_data {
-	u64 epoch_ns;
-	u32 epoch_cyc;
-	u32 epoch_cyc_copy;
-	unsigned long rate;
-	u32 mult;
-	u32 shift;
-	bool suspended;
+	seqcount_latch_t	seq;
+	struct clock_read_data	read_data[2];
+	ktime_t			wrap_kt;
+	unsigned long		rate;
+
+	u64 (*actual_read_sched_clock)(void);
 };
 
-static void sched_clock_poll(unsigned long wrap_ticks);
-static DEFINE_TIMER(sched_clock_timer, sched_clock_poll, 0, 0);
+static struct hrtimer sched_clock_timer;
 static int irqtime = -1;
 
 core_param(irqtime, irqtime, int, 0400);
 
-static struct clock_data cd = {
-	.mult	= NSEC_PER_SEC / HZ,
-};
+static u64 notrace jiffy_sched_clock_read(void)
+{
+	/*
+	 * We don't need to use get_jiffies_64 on 32-bit arches here
+	 * because we register with BITS_PER_LONG
+	 */
+	return (u64)(jiffies - INITIAL_JIFFIES);
+}
 
-static u32 __read_mostly sched_clock_mask = 0xffffffff;
+static struct clock_data cd ____cacheline_aligned = {
+	.read_data[0] = { .mult = NSEC_PER_SEC / HZ,
+			  .read_sched_clock = jiffy_sched_clock_read, },
+	.actual_read_sched_clock = jiffy_sched_clock_read,
+};
 
-static u32 notrace jiffy_sched_clock_read(void)
+static __always_inline u64 cyc_to_ns(u64 cyc, u32 mult, u32 shift)
 {
-	return (u32)(jiffies - INITIAL_JIFFIES);
+	return (cyc * mult) >> shift;
 }
 
-static u32 __read_mostly (*read_sched_clock)(void) = jiffy_sched_clock_read;
+notrace struct clock_read_data *sched_clock_read_begin(unsigned int *seq)
+{
+	*seq = read_seqcount_latch(&cd.seq);
+	return cd.read_data + (*seq & 1);
+}
 
-static inline u64 notrace cyc_to_ns(u64 cyc, u32 mult, u32 shift)
+notrace int sched_clock_read_retry(unsigned int seq)
 {
-	return (cyc * mult) >> shift;
+	return read_seqcount_latch_retry(&cd.seq, seq);
 }
 
-static unsigned long long notrace sched_clock_32(void)
+static __always_inline unsigned long long __sched_clock(void)
 {
-	u64 epoch_ns;
-	u32 epoch_cyc;
-	u32 cyc;
+	struct clock_read_data *rd;
+	unsigned int seq;
+	u64 cyc, res;
 
-	if (cd.suspended)
-		return cd.epoch_ns;
+	do {
+		seq = raw_read_seqcount_latch(&cd.seq);
+		rd = cd.read_data + (seq & 1);
+
+		cyc = (rd->read_sched_clock() - rd->epoch_cyc) &
+		      rd->sched_clock_mask;
+		res = rd->epoch_ns + cyc_to_ns(cyc, rd->mult, rd->shift);
+	} while (raw_read_seqcount_latch_retry(&cd.seq, seq));
 
+	return res;
+}
+
+unsigned long long noinstr sched_clock_noinstr(void)
+{
+	return __sched_clock();
+}
+
+unsigned long long notrace sched_clock(void)
+{
+	unsigned long long ns;
+	preempt_disable_notrace();
 	/*
-	 * Load the epoch_cyc and epoch_ns atomically.  We do this by
-	 * ensuring that we always write epoch_cyc, epoch_ns and
-	 * epoch_cyc_copy in strict order, and read them in strict order.
-	 * If epoch_cyc and epoch_cyc_copy are not equal, then we're in
-	 * the middle of an update, and we should repeat the load.
+	 * All of __sched_clock() is a seqcount_latch reader critical section,
+	 * but relies on the raw helpers which are uninstrumented. For KCSAN,
+	 * mark all accesses in __sched_clock() as atomic.
 	 */
-	do {
-		epoch_cyc = cd.epoch_cyc;
-		smp_rmb();
-		epoch_ns = cd.epoch_ns;
-		smp_rmb();
-	} while (epoch_cyc != cd.epoch_cyc_copy);
+	kcsan_nestable_atomic_begin();
+	ns = __sched_clock();
+	kcsan_nestable_atomic_end();
+	preempt_enable_notrace();
+	return ns;
+}
+
+/*
+ * Updating the data required to read the clock.
+ *
+ * sched_clock() will never observe mis-matched data even if called from
+ * an NMI. We do this by maintaining an odd/even copy of the data and
+ * steering sched_clock() to one or the other using a sequence counter.
+ * In order to preserve the data cache profile of sched_clock() as much
+ * as possible the system reverts back to the even copy when the update
+ * completes; the odd copy is used *only* during an update.
+ */
+static void update_clock_read_data(struct clock_read_data *rd)
+{
+	/* steer readers towards the odd copy */
+	write_seqcount_latch_begin(&cd.seq);
+
+	/* now its safe for us to update the normal (even) copy */
+	cd.read_data[0] = *rd;
+
+	/* switch readers back to the even copy */
+	write_seqcount_latch(&cd.seq);
 
-	cyc = read_sched_clock();
-	cyc = (cyc - epoch_cyc) & sched_clock_mask;
-	return epoch_ns + cyc_to_ns(cyc, cd.mult, cd.shift);
+	/* update the backup (odd) copy with the new data */
+	cd.read_data[1] = *rd;
+
+	write_seqcount_latch_end(&cd.seq);
 }
 
 /*
- * Atomically update the sched_clock epoch.
+ * Atomically update the sched_clock() epoch.
  */
-static void notrace update_sched_clock(void)
+static void update_sched_clock(void)
 {
-	unsigned long flags;
-	u32 cyc;
+	u64 cyc;
 	u64 ns;
+	struct clock_read_data rd;
 
-	cyc = read_sched_clock();
-	ns = cd.epoch_ns +
-		cyc_to_ns((cyc - cd.epoch_cyc) & sched_clock_mask,
-			  cd.mult, cd.shift);
-	/*
-	 * Write epoch_cyc and epoch_ns in a way that the update is
-	 * detectable in cyc_to_fixed_sched_clock().
-	 */
-	raw_local_irq_save(flags);
-	cd.epoch_cyc_copy = cyc;
-	smp_wmb();
-	cd.epoch_ns = ns;
-	smp_wmb();
-	cd.epoch_cyc = cyc;
-	raw_local_irq_restore(flags);
+	rd = cd.read_data[0];
+
+	cyc = cd.actual_read_sched_clock();
+	ns = rd.epoch_ns + cyc_to_ns((cyc - rd.epoch_cyc) & rd.sched_clock_mask, rd.mult, rd.shift);
+
+	rd.epoch_ns = ns;
+	rd.epoch_cyc = cyc;
+
+	update_clock_read_data(&rd);
 }
 
-static void sched_clock_poll(unsigned long wrap_ticks)
+static enum hrtimer_restart sched_clock_poll(struct hrtimer *hrt)
 {
-	mod_timer(&sched_clock_timer, round_jiffies(jiffies + wrap_ticks));
 	update_sched_clock();
+	hrtimer_forward_now(hrt, cd.wrap_kt);
+
+	return HRTIMER_RESTART;
 }
 
-void __init setup_sched_clock(u32 (*read)(void), int bits, unsigned long rate)
+void sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
 {
-	unsigned long r, w;
-	u64 res, wrap;
+	u64 res, wrap, new_mask, new_epoch, cyc, ns;
+	u32 new_mult, new_shift;
+	unsigned long r, flags;
 	char r_unit;
+	struct clock_read_data rd;
 
 	if (cd.rate > rate)
 		return;
 
-	BUG_ON(bits > 32);
-	WARN_ON(!irqs_disabled());
-	read_sched_clock = read;
-	sched_clock_mask = (1 << bits) - 1;
+	/* Cannot register a sched_clock with interrupts on */
+	local_irq_save(flags);
+
+	/* Calculate the mult/shift to convert counter ticks to ns. */
+	clocks_calc_mult_shift(&new_mult, &new_shift, rate, NSEC_PER_SEC, 3600);
+
+	new_mask = CLOCKSOURCE_MASK(bits);
 	cd.rate = rate;
 
-	/* calculate the mult/shift to convert counter ticks to ns. */
-	clocks_calc_mult_shift(&cd.mult, &cd.shift, rate, NSEC_PER_SEC, 0);
+	/* Calculate how many nanosecs until we risk wrapping */
+	wrap = clocks_calc_max_nsecs(new_mult, new_shift, 0, new_mask, NULL);
+	cd.wrap_kt = ns_to_ktime(wrap);
+
+	rd = cd.read_data[0];
+
+	/* Update epoch for new counter and update 'epoch_ns' from old counter*/
+	new_epoch = read();
+	cyc = cd.actual_read_sched_clock();
+	ns = rd.epoch_ns + cyc_to_ns((cyc - rd.epoch_cyc) & rd.sched_clock_mask, rd.mult, rd.shift);
+	cd.actual_read_sched_clock = read;
+
+	rd.read_sched_clock	= read;
+	rd.sched_clock_mask	= new_mask;
+	rd.mult			= new_mult;
+	rd.shift		= new_shift;
+	rd.epoch_cyc		= new_epoch;
+	rd.epoch_ns		= ns;
+
+	update_clock_read_data(&rd);
+
+	if (sched_clock_timer.function != NULL) {
+		/* update timeout for clock wrap */
+		hrtimer_start(&sched_clock_timer, cd.wrap_kt,
+			      HRTIMER_MODE_REL_HARD);
+	}
 
 	r = rate;
 	if (r >= 4000000) {
-		r /= 1000000;
+		r = DIV_ROUND_CLOSEST(r, 1000000);
 		r_unit = 'M';
-	} else if (r >= 1000) {
-		r /= 1000;
+	} else if (r >= 4000) {
+		r = DIV_ROUND_CLOSEST(r, 1000);
 		r_unit = 'k';
-	} else
+	} else {
 		r_unit = ' ';
+	}
+
+	/* Calculate the ns resolution of this counter */
+	res = cyc_to_ns(1ULL, new_mult, new_shift);
+
+	pr_info("sched_clock: %u bits at %lu%cHz, resolution %lluns, wraps every %lluns\n",
+		bits, r, r_unit, res, wrap);
 
-	/* calculate how many ns until we wrap */
-	wrap = cyc_to_ns((1ULL << bits) - 1, cd.mult, cd.shift);
-	do_div(wrap, NSEC_PER_MSEC);
-	w = wrap;
+	/* Enable IRQ time accounting if we have a fast enough sched_clock() */
+	if (irqtime > 0 || (irqtime == -1 && rate >= 1000000))
+		enable_sched_clock_irqtime();
 
-	/* calculate the ns resolution of this counter */
-	res = cyc_to_ns(1ULL, cd.mult, cd.shift);
-	pr_info("sched_clock: %u bits at %lu%cHz, resolution %lluns, wraps every %lums\n",
-		bits, r, r_unit, res, w);
+	local_irq_restore(flags);
 
+	pr_debug("Registered %pS as sched_clock source\n", read);
+}
+EXPORT_SYMBOL_GPL(sched_clock_register);
+
+void __init generic_sched_clock_init(void)
+{
 	/*
-	 * Start the timer to keep sched_clock() properly updated and
-	 * sets the initial epoch.
+	 * If no sched_clock() function has been provided at that point,
+	 * make it the final one.
 	 */
-	sched_clock_timer.data = msecs_to_jiffies(w - (w / 10));
+	if (cd.actual_read_sched_clock == jiffy_sched_clock_read)
+		sched_clock_register(jiffy_sched_clock_read, BITS_PER_LONG, HZ);
+
 	update_sched_clock();
 
 	/*
-	 * Ensure that sched_clock() starts off at 0ns
+	 * Start the timer to keep sched_clock() properly updated and
+	 * sets the initial epoch.
 	 */
-	cd.epoch_ns = 0;
+	hrtimer_setup(&sched_clock_timer, sched_clock_poll, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD);
+	hrtimer_start(&sched_clock_timer, cd.wrap_kt, HRTIMER_MODE_REL_HARD);
+}
 
-	/* Enable IRQ time accounting if we have a fast enough sched_clock */
-	if (irqtime > 0 || (irqtime == -1 && rate >= 1000000))
-		enable_sched_clock_irqtime();
+/*
+ * Clock read function for use when the clock is suspended.
+ *
+ * This function makes it appear to sched_clock() as if the clock
+ * stopped counting at its last update.
+ *
+ * This function must only be called from the critical
+ * section in sched_clock(). It relies on the read_seqcount_retry()
+ * at the end of the critical section to be sure we observe the
+ * correct copy of 'epoch_cyc'.
+ */
+static u64 notrace suspended_sched_clock_read(void)
+{
+	unsigned int seq = read_seqcount_latch(&cd.seq);
 
-	pr_debug("Registered %pF as sched_clock source\n", read);
+	return cd.read_data[seq & 1].epoch_cyc;
 }
 
-unsigned long long __read_mostly (*sched_clock_func)(void) = sched_clock_32;
-
-unsigned long long notrace sched_clock(void)
+int sched_clock_suspend(void)
 {
-	return sched_clock_func();
+	struct clock_read_data *rd = &cd.read_data[0];
+
+	update_sched_clock();
+	hrtimer_cancel(&sched_clock_timer);
+	rd->read_sched_clock = suspended_sched_clock_read;
+
+	return 0;
 }
 
-void __init sched_clock_postinit(void)
+static int sched_clock_syscore_suspend(void *data)
 {
-	/*
-	 * If no sched_clock function has been provided at that point,
-	 * make it the final one one.
-	 */
-	if (read_sched_clock == jiffy_sched_clock_read)
-		setup_sched_clock(jiffy_sched_clock_read, 32, HZ);
-
-	sched_clock_poll(sched_clock_timer.data);
+	return sched_clock_suspend();
 }
 
-static int sched_clock_suspend(void)
+void sched_clock_resume(void)
 {
-	sched_clock_poll(sched_clock_timer.data);
-	cd.suspended = true;
-	return 0;
+	struct clock_read_data *rd = &cd.read_data[0];
+
+	rd->epoch_cyc = cd.actual_read_sched_clock();
+	hrtimer_start(&sched_clock_timer, cd.wrap_kt, HRTIMER_MODE_REL_HARD);
+	rd->read_sched_clock = cd.actual_read_sched_clock;
 }
 
-static void sched_clock_resume(void)
+static void sched_clock_syscore_resume(void *data)
 {
-	cd.epoch_cyc = read_sched_clock();
-	cd.epoch_cyc_copy = cd.epoch_cyc;
-	cd.suspended = false;
+	sched_clock_resume();
 }
 
-static struct syscore_ops sched_clock_ops = {
-	.suspend = sched_clock_suspend,
-	.resume = sched_clock_resume,
+static const struct syscore_ops sched_clock_syscore_ops = {
+	.suspend	= sched_clock_syscore_suspend,
+	.resume		= sched_clock_syscore_resume,
+};
+
+static struct syscore sched_clock_syscore = {
+	.ops = &sched_clock_syscore_ops,
 };
 
 static int __init sched_clock_syscore_init(void)
 {
-	register_syscore_ops(&sched_clock_ops);
+	register_syscore(&sched_clock_syscore);
+
 	return 0;
 }
 device_initcall(sched_clock_syscore_init);