All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register
@ 2020-01-07  1:06 Paul Cercueil
  2020-01-07  1:06 ` [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST Paul Cercueil
                   ` (3 more replies)
  0 siblings, 4 replies; 7+ messages in thread
From: Paul Cercueil @ 2020-01-07  1:06 UTC (permalink / raw)
  To: Daniel Lezcano, Thomas Gleixner; +Cc: od, linux-kernel, Paul Cercueil

Instead of issueing a warning if sched_clock_register is called from a
context where IRQs are enabled, the code now ensures that IRQs are
indeed disabled.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
---

Notes:
    v2: New patch

 kernel/time/sched_clock.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/kernel/time/sched_clock.c b/kernel/time/sched_clock.c
index dbd69052eaa6..e4332e3e2d56 100644
--- a/kernel/time/sched_clock.c
+++ b/kernel/time/sched_clock.c
@@ -169,14 +169,15 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
 {
 	u64 res, wrap, new_mask, new_epoch, cyc, ns;
 	u32 new_mult, new_shift;
-	unsigned long r;
+	unsigned long r, flags;
 	char r_unit;
 	struct clock_read_data rd;
 
 	if (cd.rate > rate)
 		return;
 
-	WARN_ON(!irqs_disabled());
+	/* 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);
@@ -233,6 +234,8 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
 	if (irqtime > 0 || (irqtime == -1 && rate >= 1000000))
 		enable_sched_clock_irqtime();
 
+	local_irq_restore(flags);
+
 	pr_debug("Registered %pS as sched_clock source\n", read);
 }
 
-- 
2.24.1


^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST
  2020-01-07  1:06 [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Paul Cercueil
@ 2020-01-07  1:06 ` Paul Cercueil
  2020-01-09 14:28   ` Thomas Gleixner
  2020-01-09 11:55 ` [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Daniel Lezcano
                   ` (2 subsequent siblings)
  3 siblings, 1 reply; 7+ messages in thread
From: Paul Cercueil @ 2020-01-07  1:06 UTC (permalink / raw)
  To: Daniel Lezcano, Thomas Gleixner
  Cc: od, linux-kernel, Maarten ter Huurne, Paul Cercueil,
	Mathieu Malaterre, Artur Rojek

From: Maarten ter Huurne <maarten@treewalker.org>

OST is the OS Timer, a 64-bit timer/counter with buffered reading.

SoCs before the JZ4770 had (if any) a 32-bit OST; the JZ4770 and
JZ4780 have a 64-bit OST.

This driver will register both a clocksource and a sched_clock to the
system.

Signed-off-by: Maarten ter Huurne <maarten@treewalker.org>
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Tested-by: Mathieu Malaterre <malat@debian.org>
Tested-by: Artur Rojek <contact@artur-rojek.eu>
---

Notes:
    v2: local_irq_save/restore were moved within sched_clock_register

 drivers/clocksource/Kconfig       |   8 ++
 drivers/clocksource/Makefile      |   1 +
 drivers/clocksource/ingenic-ost.c | 218 ++++++++++++++++++++++++++++++
 3 files changed, 227 insertions(+)
 create mode 100644 drivers/clocksource/ingenic-ost.c

diff --git a/drivers/clocksource/Kconfig b/drivers/clocksource/Kconfig
index 5fdd76cb1768..5be2f876f64f 100644
--- a/drivers/clocksource/Kconfig
+++ b/drivers/clocksource/Kconfig
@@ -697,4 +697,12 @@ config INGENIC_TIMER
 	help
 	  Support for the timer/counter unit of the Ingenic JZ SoCs.
 
+config INGENIC_OST
+	bool "Ingenic JZ47xx Operating System Timer"
+	depends on MIPS || COMPILE_TEST
+	depends on COMMON_CLK
+	select MFD_SYSCON
+	help
+	  Support for the OS Timer of the Ingenic JZ4770 or similar SoC.
+
 endmenu
diff --git a/drivers/clocksource/Makefile b/drivers/clocksource/Makefile
index 4dfe4225ece7..6bc97a6fd229 100644
--- a/drivers/clocksource/Makefile
+++ b/drivers/clocksource/Makefile
@@ -80,6 +80,7 @@ obj-$(CONFIG_ASM9260_TIMER)		+= asm9260_timer.o
 obj-$(CONFIG_H8300_TMR8)		+= h8300_timer8.o
 obj-$(CONFIG_H8300_TMR16)		+= h8300_timer16.o
 obj-$(CONFIG_H8300_TPU)			+= h8300_tpu.o
+obj-$(CONFIG_INGENIC_OST)		+= ingenic-ost.o
 obj-$(CONFIG_INGENIC_TIMER)		+= ingenic-timer.o
 obj-$(CONFIG_CLKSRC_ST_LPC)		+= clksrc_st_lpc.o
 obj-$(CONFIG_X86_NUMACHIP)		+= numachip.o
diff --git a/drivers/clocksource/ingenic-ost.c b/drivers/clocksource/ingenic-ost.c
new file mode 100644
index 000000000000..dbb1d9634771
--- /dev/null
+++ b/drivers/clocksource/ingenic-ost.c
@@ -0,0 +1,218 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * JZ47xx SoCs TCU Operating System Timer driver
+ *
+ * Copyright (C) 2016 Maarten ter Huurne <maarten@treewalker.org>
+ * Copyright (C) 2018 Paul Cercueil <paul@crapouillou.net>
+ */
+
+#include <linux/clk.h>
+#include <linux/clocksource.h>
+#include <linux/mfd/ingenic-tcu.h>
+#include <linux/mfd/syscon.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/pm.h>
+#include <linux/regmap.h>
+#include <linux/sched_clock.h>
+
+#define TCU_OST_TCSR_MASK	0xffc0
+#define TCU_OST_TCSR_CNT_MD	BIT(15)
+
+#define TCU_OST_CHANNEL		15
+
+struct ingenic_ost_soc_info {
+	bool is64bit;
+};
+
+struct ingenic_ost {
+	struct regmap *map;
+	struct clk *clk;
+
+	struct clocksource cs;
+};
+
+static struct ingenic_ost *ingenic_ost;
+
+static u64 notrace ingenic_ost_read_cntl(void)
+{
+	u32 val;
+
+	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTL, &val);
+
+	return val;
+}
+
+static u64 notrace ingenic_ost_read_cnth(void)
+{
+	u32 val;
+
+	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTH, &val);
+
+	return val;
+}
+
+static u64 notrace ingenic_ost_clocksource_read64(struct clocksource *cs)
+{
+	u32 val1, val2;
+	u64 count, recount;
+	s64 diff;
+
+	/*
+	 * The buffering of the upper 32 bits of the timer prevents wrong
+	 * results from the bottom 32 bits overflowing due to the timer ticking
+	 * along. However, it does not prevent wrong results from simultaneous
+	 * reads of the timer, which could reset the buffer mid-read.
+	 * Since this kind of wrong read can happen only when the bottom bits
+	 * overflow, there will be minutes between wrong reads, so if we read
+	 * twice in succession, at least one of the reads will be correct.
+	 */
+
+	/* Bypass the regmap here as we must return as soon as possible */
+	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTL, &val1);
+	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTHBUF, &val2);
+	count = (u64)val1 | (u64)val2 << 32;
+
+	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTL, &val1);
+	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTHBUF, &val2);
+	recount = (u64)val1 | (u64)val2 << 32;
+
+	/*
+	 * A wrong read will produce a result that is 1<<32 too high: the bottom
+	 * part from before overflow and the upper part from after overflow.
+	 * Therefore, the lower value of the two reads is the correct value.
+	 */
+
+	diff = (s64)(recount - count);
+	if (unlikely(diff < 0))
+		count = recount;
+
+	return count;
+}
+
+static u64 notrace ingenic_ost_clocksource_read32(struct clocksource *cs)
+{
+	return ingenic_ost_read_cnth();
+}
+
+static int __init ingenic_ost_probe(struct platform_device *pdev)
+{
+	const struct ingenic_ost_soc_info *soc_info;
+	struct device *dev = &pdev->dev;
+	struct ingenic_ost *ost;
+	struct clocksource *cs;
+	unsigned long rate;
+	int err;
+
+	soc_info = device_get_match_data(dev);
+	if (!soc_info)
+		return -EINVAL;
+
+	ost = devm_kzalloc(dev, sizeof(*ost), GFP_KERNEL);
+	if (!ost)
+		return -ENOMEM;
+
+	ingenic_ost = ost;
+
+	ost->map = device_node_to_regmap(dev->parent->of_node);
+	if (!ost->map) {
+		dev_err(dev, "regmap not found\n");
+		return -EINVAL;
+	}
+
+	ost->clk = devm_clk_get(dev, "ost");
+	if (IS_ERR(ost->clk))
+		return PTR_ERR(ost->clk);
+
+	err = clk_prepare_enable(ost->clk);
+	if (err)
+		return err;
+
+	/* Clear counter high/low registers */
+	if (soc_info->is64bit)
+		regmap_write(ost->map, TCU_REG_OST_CNTL, 0);
+	regmap_write(ost->map, TCU_REG_OST_CNTH, 0);
+
+	/* Don't reset counter at compare value. */
+	regmap_update_bits(ost->map, TCU_REG_OST_TCSR,
+			   TCU_OST_TCSR_MASK, TCU_OST_TCSR_CNT_MD);
+
+	rate = clk_get_rate(ost->clk);
+
+	/* Enable OST TCU channel */
+	regmap_write(ost->map, TCU_REG_TESR, BIT(TCU_OST_CHANNEL));
+
+	cs = &ost->cs;
+	cs->name	= "ingenic-ost";
+	cs->rating	= 320;
+	cs->flags	= CLOCK_SOURCE_IS_CONTINUOUS;
+
+	if (soc_info->is64bit) {
+		cs->mask = CLOCKSOURCE_MASK(64);
+		cs->read = ingenic_ost_clocksource_read64;
+	} else {
+		cs->mask = CLOCKSOURCE_MASK(32);
+		cs->read = ingenic_ost_clocksource_read32;
+	}
+
+	err = clocksource_register_hz(cs, rate);
+	if (err) {
+		dev_err(dev, "clocksource registration failed: %d\n", err);
+		clk_disable_unprepare(ost->clk);
+		return err;
+	}
+
+	if (soc_info->is64bit)
+		sched_clock_register(ingenic_ost_read_cntl, 32, rate);
+	else
+		sched_clock_register(ingenic_ost_read_cnth, 32, rate);
+
+	return 0;
+}
+
+static int __maybe_unused ingenic_ost_suspend(struct device *dev)
+{
+	struct ingenic_ost *ost = dev_get_drvdata(dev);
+
+	clk_disable(ost->clk);
+
+	return 0;
+}
+
+static int __maybe_unused ingenic_ost_resume(struct device *dev)
+{
+	struct ingenic_ost *ost = dev_get_drvdata(dev);
+
+	return clk_enable(ost->clk);
+}
+
+static const struct dev_pm_ops __maybe_unused ingenic_ost_pm_ops = {
+	/* _noirq: We want the OST clock to be gated last / ungated first */
+	.suspend_noirq = ingenic_ost_suspend,
+	.resume_noirq  = ingenic_ost_resume,
+};
+
+static const struct ingenic_ost_soc_info jz4725b_ost_soc_info = {
+	.is64bit = false,
+};
+
+static const struct ingenic_ost_soc_info jz4770_ost_soc_info = {
+	.is64bit = true,
+};
+
+static const struct of_device_id ingenic_ost_of_match[] = {
+	{ .compatible = "ingenic,jz4725b-ost", .data = &jz4725b_ost_soc_info, },
+	{ .compatible = "ingenic,jz4770-ost", .data = &jz4770_ost_soc_info, },
+	{ }
+};
+
+static struct platform_driver ingenic_ost_driver = {
+	.driver = {
+		.name = "ingenic-ost",
+#ifdef CONFIG_PM_SUSPEND
+		.pm = &ingenic_ost_pm_ops,
+#endif
+		.of_match_table = ingenic_ost_of_match,
+	},
+};
+builtin_platform_driver_probe(ingenic_ost_driver, ingenic_ost_probe);
-- 
2.24.1


^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register
  2020-01-07  1:06 [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Paul Cercueil
  2020-01-07  1:06 ` [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST Paul Cercueil
@ 2020-01-09 11:55 ` Daniel Lezcano
  2020-01-09 17:33 ` [tip: timers/core] time/sched_clock: Disable interrupts in sched_clock_register() tip-bot2 for Paul Cercueil
  2020-01-09 17:56 ` tip-bot2 for Paul Cercueil
  3 siblings, 0 replies; 7+ messages in thread
From: Daniel Lezcano @ 2020-01-09 11:55 UTC (permalink / raw)
  To: Paul Cercueil, Thomas Gleixner; +Cc: od, linux-kernel

On 07/01/2020 02:06, Paul Cercueil wrote:
> Instead of issueing a warning if sched_clock_register is called from a
> context where IRQs are enabled, the code now ensures that IRQs are
> indeed disabled.
> 
> Signed-off-by: Paul Cercueil <paul@crapouillou.net>

Acked-by: Daniel Lezcano <daniel.lezcano@linaro.org>


> ---
> 
> Notes:
>     v2: New patch
> 
>  kernel/time/sched_clock.c | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/time/sched_clock.c b/kernel/time/sched_clock.c
> index dbd69052eaa6..e4332e3e2d56 100644
> --- a/kernel/time/sched_clock.c
> +++ b/kernel/time/sched_clock.c
> @@ -169,14 +169,15 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
>  {
>  	u64 res, wrap, new_mask, new_epoch, cyc, ns;
>  	u32 new_mult, new_shift;
> -	unsigned long r;
> +	unsigned long r, flags;
>  	char r_unit;
>  	struct clock_read_data rd;
>  
>  	if (cd.rate > rate)
>  		return;
>  
> -	WARN_ON(!irqs_disabled());
> +	/* 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);
> @@ -233,6 +234,8 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
>  	if (irqtime > 0 || (irqtime == -1 && rate >= 1000000))
>  		enable_sched_clock_irqtime();
>  
> +	local_irq_restore(flags);
> +
>  	pr_debug("Registered %pS as sched_clock source\n", read);
>  }
>  
> 


-- 
 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog


^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST
  2020-01-07  1:06 ` [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST Paul Cercueil
@ 2020-01-09 14:28   ` Thomas Gleixner
  2020-01-09 18:26     ` Paul Cercueil
  0 siblings, 1 reply; 7+ messages in thread
From: Thomas Gleixner @ 2020-01-09 14:28 UTC (permalink / raw)
  To: Paul Cercueil, Daniel Lezcano
  Cc: od, linux-kernel, Maarten ter Huurne, Paul Cercueil,
	Mathieu Malaterre, Artur Rojek

Paul Cercueil <paul@crapouillou.net> writes:
> +static u64 notrace ingenic_ost_clocksource_read64(struct clocksource *cs)
> +{
> +	u32 val1, val2;
> +	u64 count, recount;
> +	s64 diff;
> +
> +	/*
> +	 * The buffering of the upper 32 bits of the timer prevents wrong
> +	 * results from the bottom 32 bits overflowing due to the timer ticking
> +	 * along. However, it does not prevent wrong results from simultaneous
> +	 * reads of the timer, which could reset the buffer mid-read.
> +	 * Since this kind of wrong read can happen only when the bottom bits
> +	 * overflow, there will be minutes between wrong reads, so if we read
> +	 * twice in succession, at least one of the reads will be correct.
> +	 */
> +
> +	/* Bypass the regmap here as we must return as soon as possible */

I have a hard time to understand this comment. "Bypass the regmap ..."
and then use a regmap function?

> +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTL, &val1);
> +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTHBUF, &val2);
> +	count = (u64)val1 | (u64)val2 << 32;
> +
> +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTL, &val1);
> +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTHBUF, &val2);
> +	recount = (u64)val1 | (u64)val2 << 32;
> +
> +	/*
> +	 * A wrong read will produce a result that is 1<<32 too high: the bottom
> +	 * part from before overflow and the upper part from after overflow.
> +	 * Therefore, the lower value of the two reads is the correct value.
> +	 */
> +
> +	diff = (s64)(recount - count);
> +	if (unlikely(diff < 0))
> +		count = recount;

Is this really the right approach here? What is the 64bit readout buying
you?

The timekeeping code can handle a 32bit counter perfectly fine and the
only advantage you get is that your maximum possible idle time will be
longer with a 64bit counter.

But is that really worth the overhead of four MMIO reads versus one in a
hotpath?

Thanks,

        tglx

^ permalink raw reply	[flat|nested] 7+ messages in thread

* [tip: timers/core] time/sched_clock: Disable interrupts in sched_clock_register()
  2020-01-07  1:06 [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Paul Cercueil
  2020-01-07  1:06 ` [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST Paul Cercueil
  2020-01-09 11:55 ` [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Daniel Lezcano
@ 2020-01-09 17:33 ` tip-bot2 for Paul Cercueil
  2020-01-09 17:56 ` tip-bot2 for Paul Cercueil
  3 siblings, 0 replies; 7+ messages in thread
From: tip-bot2 for Paul Cercueil @ 2020-01-09 17:33 UTC (permalink / raw)
  To: linux-tip-commits
  Cc: Paul Cercueil, Thomas Gleixner, Daniel Lezcano, x86, LKML

The following commit has been merged into the timers/core branch of tip:

Commit-ID:     479c296303e8d29711a5153f0fc4b67faa6d9866
Gitweb:        https://git.kernel.org/tip/479c296303e8d29711a5153f0fc4b67faa6d9866
Author:        Paul Cercueil <paul@crapouillou.net>
AuthorDate:    Tue, 07 Jan 2020 02:06:29 +01:00
Committer:     Thomas Gleixner <tglx@linutronix.de>
CommitterDate: Thu, 09 Jan 2020 18:28:34 +01:00

time/sched_clock: Disable interrupts in sched_clock_register()

Instead of issueing a warning if sched_clock_register() is called from a
context where IRQs are enabled, the code now ensures that IRQs are indeed
disabled.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://lore.kernel.org/r/20200107010630.954648-1-paul@crapouillou.net

---
 kernel/time/sched_clock.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/kernel/time/sched_clock.c b/kernel/time/sched_clock.c
index dbd6905..e4332e3 100644
--- a/kernel/time/sched_clock.c
+++ b/kernel/time/sched_clock.c
@@ -169,14 +169,15 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
 {
 	u64 res, wrap, new_mask, new_epoch, cyc, ns;
 	u32 new_mult, new_shift;
-	unsigned long r;
+	unsigned long r, flags;
 	char r_unit;
 	struct clock_read_data rd;
 
 	if (cd.rate > rate)
 		return;
 
-	WARN_ON(!irqs_disabled());
+	/* 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);
@@ -233,6 +234,8 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
 	if (irqtime > 0 || (irqtime == -1 && rate >= 1000000))
 		enable_sched_clock_irqtime();
 
+	local_irq_restore(flags);
+
 	pr_debug("Registered %pS as sched_clock source\n", read);
 }
 

^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [tip: timers/core] time/sched_clock: Disable interrupts in sched_clock_register()
  2020-01-07  1:06 [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Paul Cercueil
                   ` (2 preceding siblings ...)
  2020-01-09 17:33 ` [tip: timers/core] time/sched_clock: Disable interrupts in sched_clock_register() tip-bot2 for Paul Cercueil
@ 2020-01-09 17:56 ` tip-bot2 for Paul Cercueil
  3 siblings, 0 replies; 7+ messages in thread
From: tip-bot2 for Paul Cercueil @ 2020-01-09 17:56 UTC (permalink / raw)
  To: linux-tip-commits
  Cc: Paul Cercueil, Thomas Gleixner, Daniel Lezcano, x86, LKML

The following commit has been merged into the timers/core branch of tip:

Commit-ID:     2707745533d6d38fa7d3a2212f1fd599c3879491
Gitweb:        https://git.kernel.org/tip/2707745533d6d38fa7d3a2212f1fd599c3879491
Author:        Paul Cercueil <paul@crapouillou.net>
AuthorDate:    Tue, 07 Jan 2020 02:06:29 +01:00
Committer:     Thomas Gleixner <tglx@linutronix.de>
CommitterDate: Thu, 09 Jan 2020 18:50:18 +01:00

time/sched_clock: Disable interrupts in sched_clock_register()

Instead of issueing a warning if sched_clock_register() is called from a
context where IRQs are enabled, the code now ensures that IRQs are indeed
disabled.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://lore.kernel.org/r/20200107010630.954648-1-paul@crapouillou.net

---
 kernel/time/sched_clock.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/kernel/time/sched_clock.c b/kernel/time/sched_clock.c
index dbd6905..e4332e3 100644
--- a/kernel/time/sched_clock.c
+++ b/kernel/time/sched_clock.c
@@ -169,14 +169,15 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
 {
 	u64 res, wrap, new_mask, new_epoch, cyc, ns;
 	u32 new_mult, new_shift;
-	unsigned long r;
+	unsigned long r, flags;
 	char r_unit;
 	struct clock_read_data rd;
 
 	if (cd.rate > rate)
 		return;
 
-	WARN_ON(!irqs_disabled());
+	/* 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);
@@ -233,6 +234,8 @@ sched_clock_register(u64 (*read)(void), int bits, unsigned long rate)
 	if (irqtime > 0 || (irqtime == -1 && rate >= 1000000))
 		enable_sched_clock_irqtime();
 
+	local_irq_restore(flags);
+
 	pr_debug("Registered %pS as sched_clock source\n", read);
 }
 

^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST
  2020-01-09 14:28   ` Thomas Gleixner
@ 2020-01-09 18:26     ` Paul Cercueil
  0 siblings, 0 replies; 7+ messages in thread
From: Paul Cercueil @ 2020-01-09 18:26 UTC (permalink / raw)
  To: Thomas Gleixner
  Cc: Daniel Lezcano, od, linux-kernel, Maarten ter Huurne,
	Mathieu Malaterre, Artur Rojek

Hi Thomas,


Le jeu., janv. 9, 2020 at 15:28, Thomas Gleixner <tglx@linutronix.de> a 
écrit :
> Paul Cercueil <paul@crapouillou.net> writes:
>>  +static u64 notrace ingenic_ost_clocksource_read64(struct 
>> clocksource *cs)
>>  +{
>>  +	u32 val1, val2;
>>  +	u64 count, recount;
>>  +	s64 diff;
>>  +
>>  +	/*
>>  +	 * The buffering of the upper 32 bits of the timer prevents wrong
>>  +	 * results from the bottom 32 bits overflowing due to the timer 
>> ticking
>>  +	 * along. However, it does not prevent wrong results from 
>> simultaneous
>>  +	 * reads of the timer, which could reset the buffer mid-read.
>>  +	 * Since this kind of wrong read can happen only when the bottom 
>> bits
>>  +	 * overflow, there will be minutes between wrong reads, so if we 
>> read
>>  +	 * twice in succession, at least one of the reads will be correct.
>>  +	 */
>>  +
>>  +	/* Bypass the regmap here as we must return as soon as possible */
> 
> I have a hard time to understand this comment. "Bypass the regmap ..."
> and then use a regmap function?

Ah, sorry, it's a leftover from a previous version of the patch. It 
used to bypass the regmap in order to complete as fast as possible.

> 
>>  +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTL, &val1);
>>  +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTHBUF, &val2);
>>  +	count = (u64)val1 | (u64)val2 << 32;
>>  +
>>  +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTL, &val1);
>>  +	regmap_read(ingenic_ost->map, TCU_REG_OST_CNTHBUF, &val2);
>>  +	recount = (u64)val1 | (u64)val2 << 32;
>>  +
>>  +	/*
>>  +	 * A wrong read will produce a result that is 1<<32 too high: the 
>> bottom
>>  +	 * part from before overflow and the upper part from after 
>> overflow.
>>  +	 * Therefore, the lower value of the two reads is the correct 
>> value.
>>  +	 */
>>  +
>>  +	diff = (s64)(recount - count);
>>  +	if (unlikely(diff < 0))
>>  +		count = recount;
> 
> Is this really the right approach here? What is the 64bit readout 
> buying
> you?
> 
> The timekeeping code can handle a 32bit counter perfectly fine and the
> only advantage you get is that your maximum possible idle time will be
> longer with a 64bit counter.
> 
> But is that really worth the overhead of four MMIO reads versus one 
> in a
> hotpath?

The timer is 64-bit so I thought it made sense to register it as such. 
Using it as just a 32-bit counter sounds better indeed.

Thanks,
-Paul




^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2020-01-09 18:26 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-01-07  1:06 [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Paul Cercueil
2020-01-07  1:06 ` [PATCH v2 2/2] clocksource: Add driver for the Ingenic JZ47xx OST Paul Cercueil
2020-01-09 14:28   ` Thomas Gleixner
2020-01-09 18:26     ` Paul Cercueil
2020-01-09 11:55 ` [PATCH v2 1/2] time/sched_clock: Disable interrupts in sched_clock_register Daniel Lezcano
2020-01-09 17:33 ` [tip: timers/core] time/sched_clock: Disable interrupts in sched_clock_register() tip-bot2 for Paul Cercueil
2020-01-09 17:56 ` tip-bot2 for Paul Cercueil

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.