Skip to contentSkip to conversation

gpio: omap: Remove custom PM calls and use cpu_pm instead

PATCH New Tony Lindgren 4 participants 8 messages Sep 11, 2018 – Sep 14, 2018

Subsystems (6)
lore
Share & tools
lore

Tony Lindgren

author maintainer
#0 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

For a long time the gpio-omap custom PM calls have been annoying me so let's replace them with cpu_pm instead. This will enable GPIO PM for deeper idle states on omap4. And we can handle GPIO PM for omap2/3/4 in the same way.

Note that with this patch we are also slightly changing GPIO PM to be less aggressive for omap3 and only will idle GPIO when PER context may be lost.

For omap2, we don't need to save context and don't want to remove any triggering so let's add a quirk flag for that.

Let's do this all in a single patch to avoid a situation where old custom calls still are used with new code.

Cc Aaro Koskinen <[email protected]>
Cc Grygorii Strashko <[email protected]>
Cc Keerthy <[email protected]>
Cc Ladislav Michl <[email protected]>
Cc Tero Kristo <[email protected]>

---

Note that this depends on previously posted patch:

[PATCH] gpio: omap: Add level wakeup handling for omap4 based SoCs

Linus, once people are happy with these, can you maybe set up an immutable branch with both patches in it?

---
 arch/arm/mach-omap2/pm24xx.c            |  7 +--
 arch/arm/mach-omap2/pm34xx.c            | 14 ++---
 drivers/gpio/gpio-omap.c                | 78 +++++++++++++++----------
 include/linux/platform_data/gpio-omap.h | 13 -----
 4 files changed, 55 insertions(+), 57 deletions(-)
arch/arm/mach-omap2/pm24xx.c +3 -4
@@ -18,6 +18,7 @@
  * published by the Free Software Foundation.
  */
 
+#include <linux/cpu_pm.h>
 #include <linux/suspend.h>
 #include <linux/sched.h>
 #include <linux/proc_fs.h>
@@ -29,8 +30,6 @@
 #include <linux/clk-provider.h>
 #include <linux/irq.h>
 #include <linux/time.h>
-#include <linux/gpio.h>
-#include <linux/platform_data/gpio-omap.h>
 
 #include <asm/fncpy.h>
 
@@ -87,7 +86,7 @@ static int omap2_enter_full_retention(void)
 	l = omap_ctrl_readl(OMAP2_CONTROL_DEVCONF0) | OMAP24XX_USBSTANDBYCTRL;
 	omap_ctrl_writel(l, OMAP2_CONTROL_DEVCONF0);
 
-	omap2_gpio_prepare_for_idle(0);
+	cpu_cluster_pm_enter();
 
 	/* One last check for pending IRQs to avoid extra latency due
 	 * to sleeping unnecessarily. */
@@ -100,7 +99,7 @@ static int omap2_enter_full_retention(void)
 			   OMAP_SDRC_REGADDR(SDRC_POWER));
 
 no_sleep:
-	omap2_gpio_resume_after_idle();
+	cpu_cluster_pm_exit();
 
 	clk_enable(osc_ck);
 
arch/arm/mach-omap2/pm34xx.c +5 -9
@@ -18,19 +18,18 @@
  * published by the Free Software Foundation.
  */
 
+#include <linux/cpu_pm.h>
 #include <linux/pm.h>
 #include <linux/suspend.h>
 #include <linux/interrupt.h>
 #include <linux/module.h>
 #include <linux/list.h>
 #include <linux/err.h>
-#include <linux/gpio.h>
 #include <linux/clk.h>
 #include <linux/delay.h>
 #include <linux/slab.h>
 #include <linux/omap-dma.h>
 #include <linux/omap-gpmc.h>
-#include <linux/platform_data/gpio-omap.h>
 
 #include <trace/events/power.h>
 
@@ -197,7 +196,6 @@ void omap_sram_idle(void)
 	int mpu_next_state = PWRDM_POWER_ON;
 	int per_next_state = PWRDM_POWER_ON;
 	int core_next_state = PWRDM_POWER_ON;
-	int per_going_off;
 	u32 sdrc_pwr = 0;
 
 	mpu_next_state = pwrdm_read_next_pwrst(mpu_pwrdm);
@@ -227,10 +225,8 @@ void omap_sram_idle(void)
 	pwrdm_pre_transition(NULL);
 
 	/* PER */
-	if (per_next_state < PWRDM_POWER_ON) {
-		per_going_off = (per_next_state == PWRDM_POWER_OFF) ? 1 : 0;
-		omap2_gpio_prepare_for_idle(per_going_off);
-	}
+	if (per_next_state == PWRDM_POWER_OFF)
+		cpu_cluster_pm_enter();
 
 	/* CORE */
 	if (core_next_state < PWRDM_POWER_ON) {
@@ -295,8 +291,8 @@ void omap_sram_idle(void)
 	pwrdm_post_transition(NULL);
 
 	/* PER */
-	if (per_next_state < PWRDM_POWER_ON)
-		omap2_gpio_resume_after_idle();
+	if (per_next_state == PWRDM_POWER_OFF)
+		cpu_cluster_pm_exit();
 }
 
 static void omap3_pm_idle(void)
drivers/gpio/gpio-omap.c +47 -31

139 lines not on this page · raw message

include/linux/platform_data/gpio-omap.h +0 -13
@@ -205,17 +205,4 @@ struct omap_gpio_platform_data {
 	int (*get_context_loss_count)(struct device *dev);
 };
 
-#if IS_BUILTIN(CONFIG_GPIO_OMAP)
-extern void omap2_gpio_prepare_for_idle(int off_mode);
-extern void omap2_gpio_resume_after_idle(void);
-#else
-static inline void omap2_gpio_prepare_for_idle(int off_mode)
-{
-}
-
-static inline void omap2_gpio_resume_after_idle(void)
-{
-}
-#endif
-
 #endif
Signature · 1 line
2.18.0
#1 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

Tony,

Quoted 8 lines · Tony Lindgren

For a long time the gpio-omap custom PM calls have been annoying me so let's replace them with cpu_pm instead. This will enable GPIO PM for deeper idle states on omap4. And we can handle GPIO PM for omap2/3/4 in the same way.

Note that with this patch we are also slightly changing GPIO PM to be less aggressive for omap3 and only will idle GPIO when PER context may be lost.

I do not think it will make things any worse, but will run my favorite latency on test on this patch :)

Meanwhile see nit bellow.

Quoted 156 lines

For omap2, we don't need to save context and don't want to remove any triggering so let's add a quirk flag for that.

Let's do this all in a single patch to avoid a situation where old custom calls still are used with new code.

Cc Aaro Koskinen <[email protected]>
Cc Grygorii Strashko <[email protected]>
Cc Keerthy <[email protected]>
Cc Ladislav Michl <[email protected]>
Cc Tero Kristo <[email protected]>

---

Note that this depends on previously posted patch:

[PATCH] gpio: omap: Add level wakeup handling for omap4 based SoCs

Linus, once people are happy with these, can you maybe set up an immutable branch with both patches in it?

---
 arch/arm/mach-omap2/pm24xx.c            |  7 +--
 arch/arm/mach-omap2/pm34xx.c            | 14 ++---
 drivers/gpio/gpio-omap.c                | 78 +++++++++++++++----------
 include/linux/platform_data/gpio-omap.h | 13 -----
 4 files changed, 55 insertions(+), 57 deletions(-)
arch/arm/mach-omap2/pm24xx.c +3 -4
@@ -18,6 +18,7 @@
  * published by the Free Software Foundation.
  */
 
+#include <linux/cpu_pm.h>
 #include <linux/suspend.h>
 #include <linux/sched.h>
 #include <linux/proc_fs.h>
@@ -29,8 +30,6 @@
 #include <linux/clk-provider.h>
 #include <linux/irq.h>
 #include <linux/time.h>
-#include <linux/gpio.h>
-#include <linux/platform_data/gpio-omap.h>
 
 #include <asm/fncpy.h>
 
@@ -87,7 +86,7 @@ static int omap2_enter_full_retention(void)
 	l = omap_ctrl_readl(OMAP2_CONTROL_DEVCONF0) | OMAP24XX_USBSTANDBYCTRL;
 	omap_ctrl_writel(l, OMAP2_CONTROL_DEVCONF0);
 
-	omap2_gpio_prepare_for_idle(0);
+	cpu_cluster_pm_enter();
 
 	/* One last check for pending IRQs to avoid extra latency due
 	 * to sleeping unnecessarily. */
@@ -100,7 +99,7 @@ static int omap2_enter_full_retention(void)
 			   OMAP_SDRC_REGADDR(SDRC_POWER));
 
 no_sleep:
-	omap2_gpio_resume_after_idle();
+	cpu_cluster_pm_exit();
 
 	clk_enable(osc_ck);
 
arch/arm/mach-omap2/pm34xx.c +5 -9
@@ -18,19 +18,18 @@
  * published by the Free Software Foundation.
  */
 
+#include <linux/cpu_pm.h>
 #include <linux/pm.h>
 #include <linux/suspend.h>
 #include <linux/interrupt.h>
 #include <linux/module.h>
 #include <linux/list.h>
 #include <linux/err.h>
-#include <linux/gpio.h>
 #include <linux/clk.h>
 #include <linux/delay.h>
 #include <linux/slab.h>
 #include <linux/omap-dma.h>
 #include <linux/omap-gpmc.h>
-#include <linux/platform_data/gpio-omap.h>
 
 #include <trace/events/power.h>
 
@@ -197,7 +196,6 @@ void omap_sram_idle(void)
 	int mpu_next_state = PWRDM_POWER_ON;
 	int per_next_state = PWRDM_POWER_ON;
 	int core_next_state = PWRDM_POWER_ON;
-	int per_going_off;
 	u32 sdrc_pwr = 0;
 
 	mpu_next_state = pwrdm_read_next_pwrst(mpu_pwrdm);
@@ -227,10 +225,8 @@ void omap_sram_idle(void)
 	pwrdm_pre_transition(NULL);
 
 	/* PER */
-	if (per_next_state < PWRDM_POWER_ON) {
-		per_going_off = (per_next_state == PWRDM_POWER_OFF) ? 1 : 0;
-		omap2_gpio_prepare_for_idle(per_going_off);
-	}
+	if (per_next_state == PWRDM_POWER_OFF)
+		cpu_cluster_pm_enter();
 
 	/* CORE */
 	if (core_next_state < PWRDM_POWER_ON) {
@@ -295,8 +291,8 @@ void omap_sram_idle(void)
 	pwrdm_post_transition(NULL);
 
 	/* PER */
-	if (per_next_state < PWRDM_POWER_ON)
-		omap2_gpio_resume_after_idle();
+	if (per_next_state == PWRDM_POWER_OFF)
+		cpu_cluster_pm_exit();
 }
 
 static void omap3_pm_idle(void)
drivers/gpio/gpio-omap.c +6 -0
@@ -19,6 +19,7 @@
 #include <linux/err.h>
 #include <linux/clk.h>
 #include <linux/io.h>
+#include <linux/cpu_pm.h>
 #include <linux/device.h>
 #include <linux/pm_runtime.h>
 #include <linux/pm.h>
@@ -31,6 +32,7 @@
 #define OFF_MODE	1
 #define OMAP4_GPIO_DEBOUNCINGTIME_MASK 0xFF
 
+#define OMAP_GPIO_QUIRK_IDLE_REMOVE_TRIGGER	BIT(2)
 #define OMAP_GPIO_QUIRK_DEFERRED_WKUP_EN	BIT(1)
 
 static LIST_HEAD(omap_gpio_list);
@@ -72,6 +74,7 @@ struct gpio_bank {
 	raw_spinlock_t wa_lock;
 	struct gpio_chip chip;
 	struct clk *dbck;
+	struct notifier_block nb;
 	u32 mod_usage;
 	u32 irq_usage;
 	u32 dbck_enable_mask;
@@ -1308,6 +1311,38 @@ static int omap_gpio_chip_init(struct gpio_bank *bank, struct irq_chip *irqc)
 	return ret;
 }
 
+static int gpio_omap_cpu_notifier(struct notifier_block *nb,
+				  unsigned long cmd, void *v)
+{

drivers/gpio/gpio-omap.c static int omap_gpio_chip_init(struct gpio_bank *bank, struct irq_chip *irqc) · line 1326

Quoted 10 lines
drivers/gpio/gpio-omap.c +10 -0
+	struct gpio_bank *bank;
+	struct device *dev;
+	int error;
+
+	bank = container_of(nb, struct gpio_bank, nb);
+	dev = bank->chip.parent;
+
+	switch (cmd) {
+	case CPU_CLUSTER_PM_ENTER:
+		/* Gets cleard on runtime_suspend */

"Gets cleared" perhaps...

Quoted 130 lines
drivers/gpio/gpio-omap.c +31 -31
+		bank->power_mode = OFF_MODE;
+
+		error = pm_runtime_put_sync(dev);
+		if (error < 0)
+			dev_warn(dev, "CPU PM enter: %i\n", error);
+		break;
+	case CPU_CLUSTER_PM_ENTER_FAILED:
+	case CPU_CLUSTER_PM_EXIT:
+		error = pm_runtime_get_sync(dev);
+		if (error < 0) {
+			dev_err(dev, "CPU PM exit %i\n", error);
+			pm_runtime_put_noidle(dev);
+		}
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+
 static const struct of_device_id omap_gpio_match[];
 
 static int omap_gpio_probe(struct platform_device *pdev)
@@ -1445,6 +1480,9 @@ static int omap_gpio_probe(struct platform_device *pdev)
 
 	omap_gpio_show_rev(bank);
 
+	bank->nb.notifier_call = gpio_omap_cpu_notifier;
+	cpu_pm_register_notifier(&bank->nb);
+
 	pm_runtime_put(dev);
 
 	list_add_tail(&bank->node, &omap_gpio_list);
@@ -1456,6 +1494,7 @@ static int omap_gpio_remove(struct platform_device *pdev)
 {
 	struct gpio_bank *bank = platform_get_drvdata(pdev);
 
+	cpu_pm_unregister_notifier(&bank->nb);
 	list_del(&bank->node);
 	gpiochip_remove(&bank->chip);
 	pm_runtime_disable(&pdev->dev);
@@ -1622,37 +1661,7 @@ static int omap_gpio_runtime_resume(struct device *dev)
 
 	return 0;
 }
-#endif /* CONFIG_PM */
-
-#if IS_BUILTIN(CONFIG_GPIO_OMAP)
-void omap2_gpio_prepare_for_idle(int pwr_mode)
-{
-	struct gpio_bank *bank;
 
-	list_for_each_entry(bank, &omap_gpio_list, node) {
-		if (!BANK_USED(bank) || !bank->loses_context)
-			continue;
-
-		bank->power_mode = pwr_mode;
-
-		pm_runtime_put_sync_suspend(bank->chip.parent);
-	}
-}
-
-void omap2_gpio_resume_after_idle(void)
-{
-	struct gpio_bank *bank;
-
-	list_for_each_entry(bank, &omap_gpio_list, node) {
-		if (!BANK_USED(bank) || !bank->loses_context)
-			continue;
-
-		pm_runtime_get_sync(bank->chip.parent);
-	}
-}
-#endif
-
-#if defined(CONFIG_PM)
 static void omap_gpio_init_context(struct gpio_bank *p)
 {
 	struct omap_gpio_reg_offs *regs = p->regs;
@@ -1768,6 +1777,11 @@ static struct omap_gpio_reg_offs omap4_gpio_regs = {
 	.fallingdetect =	OMAP4_GPIO_FALLINGDETECT,
 };
 
+/*
+ * Note that omap2 does not currently support idle modes with context loss so
+ * no need to add OMAP_GPIO_QUIRK_IDLE_REMOVE_TRIGGER quirk flag to save
+ * and restore context.
+ */
 static const struct omap_gpio_platform_data omap2_pdata = {
 	.regs = &omap2_gpio_regs,
 	.bank_width = 32,
@@ -1778,13 +1792,15 @@ static const struct omap_gpio_platform_data omap3_pdata = {
 	.regs = &omap2_gpio_regs,
 	.bank_width = 32,
 	.dbck_flag = true,
+	.quirks = OMAP_GPIO_QUIRK_IDLE_REMOVE_TRIGGER,
 };
 
 static const struct omap_gpio_platform_data omap4_pdata = {
 	.regs = &omap4_gpio_regs,
 	.bank_width = 32,
 	.dbck_flag = true,
-	.quirks = OMAP_GPIO_QUIRK_DEFERRED_WKUP_EN,
+	.quirks = OMAP_GPIO_QUIRK_IDLE_REMOVE_TRIGGER |
+		  OMAP_GPIO_QUIRK_DEFERRED_WKUP_EN,
 };
 
 static const struct of_device_id omap_gpio_match[] = {
include/linux/platform_data/gpio-omap.h +0 -13
@@ -205,17 +205,4 @@ struct omap_gpio_platform_data {
 	int (*get_context_loss_count)(struct device *dev);
 };
 
-#if IS_BUILTIN(CONFIG_GPIO_OMAP)
-extern void omap2_gpio_prepare_for_idle(int off_mode);
-extern void omap2_gpio_resume_after_idle(void);
-#else
-static inline void omap2_gpio_prepare_for_idle(int off_mode)
-{
-}
-
-static inline void omap2_gpio_resume_after_idle(void)
-{
-}
-#endif
-
 #endif
Signature · 1 line
2.18.0
author maintainer
#2 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

Quoted 8 lines · Ladislav Michl

Tony,

Quoted 5 lines · Tony Lindgren

For a long time the gpio-omap custom PM calls have been annoying me so let's replace them with cpu_pm instead. This will enable GPIO PM for deeper idle states on omap4. And we can handle GPIO PM for omap2/3/4 in the same way.

Quoted 6 lines
Quoted 3 lines

Note that with this patch we are also slightly changing GPIO PM to be less aggressive for omap3 and only will idle GPIO when PER context may be lost.

I do not think it will make things any worse, but will run my favorite latency on test on this patch :)

Thanks I was hoping you'd do that :)

Quoted 7 lines

Meanwhile see nit bellow.

drivers/gpio/gpio-omap.c static int omap_gpio_chip_init(struct gpio_bank *bank, struct irq_chip *irqc) · line 1326

Quoted 3 lines
drivers/gpio/gpio-omap.c +3 -0
+	switch (cmd) {
+	case CPU_CLUSTER_PM_ENTER:
+		/* Gets cleard on runtime_suspend */

"Gets cleared" perhaps...

Thanks will fix and post v2 after waiting few days for comments.

Regards,

Tony

author maintainer
#3 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

In reply to Tony Lindgren
Quoted 2 lines · Tony Lindgren

For omap2, we don't need to save context and don't want to remove any triggering so let's add a quirk flag for that.

...

drivers/gpio/gpio-omap.c static int omap_gpio_chip_init(struct gpio_bank *bank, struct irq_chip *irqc) · line 1327

Quoted 4 lines
drivers/gpio/gpio-omap.c +4 -0
+	switch (cmd) {
+	case CPU_CLUSTER_PM_ENTER:
+		/* Gets cleard on runtime_suspend */
+		bank->power_mode = OFF_MODE;

Oops I forgot to check for OMAP_GPIO_QUIRK_IDLE_REMOVE_TRIGGER here, this should be:

	if (bank->quirks & OMAP_GPIO_QUIRK_IDLE_REMOVE_TRIGGER)
		bank->power_mode = OFF_MODE;

Regards,

Tony

#4 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

In reply to Tony Lindgren
Quoted 30 lines · Tony Lindgren

For a long time the gpio-omap custom PM calls have been annoying me so let's replace them with cpu_pm instead. This will enable GPIO PM for deeper idle states on omap4. And we can handle GPIO PM for omap2/3/4 in the same way.

Note that with this patch we are also slightly changing GPIO PM to be less aggressive for omap3 and only will idle GPIO when PER context may be lost.

For omap2, we don't need to save context and don't want to remove any triggering so let's add a quirk flag for that.

Let's do this all in a single patch to avoid a situation where old custom calls still are used with new code.

Cc Aaro Koskinen <[email protected]>
Cc Grygorii Strashko <[email protected]>
Cc Keerthy <[email protected]>
Cc Ladislav Michl <[email protected]>
Cc Tero Kristo <[email protected]>

---

Note that this depends on previously posted patch:

[PATCH] gpio: omap: Add level wakeup handling for omap4 based SoCs

Linus, once people are happy with these, can you maybe set up an immutable branch with both patches in it?

[...]

Quoted 6 lines
  }
drivers/gpio/gpio-omap.c +4 -0
+static int gpio_omap_cpu_notifier(struct notifier_block *nb,
+				  unsigned long cmd, void *v)
+{
+	struct gpio_bank *bank;

drivers/gpio/gpio-omap.c static int omap_gpio_chip_init(struct gpio_bank *bank, struct irq_chip *irqc) · line 1327

Quoted 10 lines
drivers/gpio/gpio-omap.c +10 -0
+	struct device *dev;
+	int error;
+
+	bank = container_of(nb, struct gpio_bank, nb);
+	dev = bank->chip.parent;
+
+	switch (cmd) {
+	case CPU_CLUSTER_PM_ENTER:
+		/* Gets cleard on runtime_suspend */
+		bank->power_mode = OFF_MODE;

It has to be reset somewhere.

Quoted 18 lines
drivers/gpio/gpio-omap.c +18 -0
+
+		error = pm_runtime_put_sync(dev);
+		if (error < 0)
+			dev_warn(dev, "CPU PM enter: %i\n", error);
+		break;
+	case CPU_CLUSTER_PM_ENTER_FAILED:
+	case CPU_CLUSTER_PM_EXIT:
+		error = pm_runtime_get_sync(dev);
+		if (error < 0) {
+			dev_err(dev, "CPU PM exit %i\n", error);
+			pm_runtime_put_noidle(dev);
+		}
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+

[...]

Signature · 2 lines
regards,
-grygorii
author maintainer
#5 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

Quoted 7 lines · Grygorii Strashko

drivers/gpio/gpio-omap.c static int omap_gpio_chip_init(struct gpio_bank *bank, struct irq_chip *irqc) · line 1327

Quoted 4 lines · Tony Lindgren
drivers/gpio/gpio-omap.c +4 -0
+	switch (cmd) {
+	case CPU_CLUSTER_PM_ENTER:
+		/* Gets cleard on runtime_suspend */
+		bank->power_mode = OFF_MODE;

It has to be reset somewhere.

Oh good catch omap_gpio_runtime_suspend() only clears it if (bank->power_mode != OFF_MODE).. This would certainly introduce some unexpected latencies..

We can just clear it unconditionally at the end of CPU_CLUSTER_PM_ENTER handler now.

Regards,

Tony

#6 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

In reply to Tony Lindgren
Quoted 2 lines · Tony Lindgren

Linus, once people are happy with these, can you maybe set up an immutable branch with both patches in it?

Sure thing, just awaiting the final ACKed version.

There are some more OMAP patches floating I see, so maybe a branch with all of them on for testing purposes and coordination?

Yours,
Linus Walleij

author maintainer
#7 lore raw

PATCH 1/1 gpio: omap: Remove custom PM calls and use cpu_pm instead

Quoted 6 lines · Linus Walleij
Quoted 2 lines · Tony Lindgren

Linus, once people are happy with these, can you maybe set up an immutable branch with both patches in it?

Sure thing, just awaiting the final ACKed version.

Looks like I need a bit more time though to sort out the suspend/resume though. And it seems we can now finally drop the pm_runtime_irq_safe too :)

Quoted 2 lines

There are some more OMAP patches floating I see, so maybe a branch with all of them on for testing purposes and coordination?

OK sounds good to me thanks.

Tony