[PATCH 0/4] Implement PM runtime for musb-core based on session bit

[PATCH 0/4] Implement PM runtime for musb-core based on session bit

[Tony Lindgren]

Hi all,

Here's a series of patches to simplify musb PM runtime support further.

I finally figured out that we can get rid of most of the glue layer
specific workarounds by following the devctl session bit in musb core.

The series also adds PM runtime support to the musb dsps glue layer
for the PIO mode.

I've tested basic plugging of cables in peripheral and host mode
on omap3 and am335x using omap2plus_defconfig. Please test if you
can.

Regards,

Tony


Tony Lindgren (4):
  usb: musb: Implement session bit based runtime PM for musb-core
  usb: musb: Prepare dsps glue layer for PM runtime support
  usb: musb: Add PM runtime support for MUSB DSPS glue layer
  usb: musb: Simplify PM runtime for 2430 glue layer

 drivers/usb/musb/musb_core.c | 70 +++++++++++++++++++++++++++++++++++++
 drivers/usb/musb/musb_core.h |  1 +
 drivers/usb/musb/musb_dsps.c | 82 +++++++++++++++++++-------------------------
 drivers/usb/musb/omap2430.c  | 76 ++++++----------------------------------
 4 files changed, 117 insertions(+), 112 deletions(-)

-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 1/4] usb: musb: Implement session bit based runtime PM for musb-core

[Tony Lindgren]

We want to keep musb enabled always when the session bit is
set. This simplifies the PM runtime and allows making it more
generic across the various glue layers.

So far the only exception to just following the session bit is
host mode disconnect where the session bit stays set.

In that case, just allow PM and let the PM runtime autoidle
timeout deal with it.

Signed-off-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>
---
 drivers/usb/musb/musb_core.c | 70 ++++++++++++++++++++++++++++++++++++++++++++
 drivers/usb/musb/musb_core.h |  1 +
 2 files changed, 71 insertions(+)

diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c
index 74fc306..9c4e1a5 100644
--- a/drivers/usb/musb/musb_core.c
+++ b/drivers/usb/musb/musb_core.c
@@ -1831,11 +1831,81 @@ static const struct attribute_group musb_attr_group = {
 	.attrs = musb_attributes,
 };
 
+#define MUSB_QUIRK_B_INVALID_VBUS_91	(MUSB_DEVCTL_BDEVICE | \
+					 (2 << MUSB_DEVCTL_VBUS_SHIFT) | \
+					 MUSB_DEVCTL_SESSION)
+#define MUSB_QUIRK_A_DISCONNECT_19	((3 << MUSB_DEVCTL_VBUS_SHIFT) | \
+					 MUSB_DEVCTL_SESSION)
+
+/*
+ * Check the musb devctl session bit to determine if we want to
+ * allow PM runtime for the device. In general, we want to keep things
+ * active when the session bit is set except after host disconnect.
+ *
+ * Only called from musb_irq_work. If this ever needs to get called
+ * elsewhere, proper locking must be implemented for musb->session.
+ */
+static void musb_pm_runtime_check_session(struct musb *musb)
+{
+	u8 devctl, s;
+	int error;
+
+	devctl = musb_readb(musb->mregs, MUSB_DEVCTL);
+
+	/* Handle session status quirks first */
+	s = MUSB_DEVCTL_FSDEV | MUSB_DEVCTL_LSDEV |
+		MUSB_DEVCTL_HR;
+	switch (devctl & ~s) {
+	case MUSB_QUIRK_B_INVALID_VBUS_91:
+		if (musb->session)
+			break;
+		dev_dbg(musb->controller,
+			"Allow PM as device with invalid vbus: %02x\n",
+			devctl);
+		return;
+	case MUSB_QUIRK_A_DISCONNECT_19:
+		if (!musb->session)
+			break;
+		dev_dbg(musb->controller,
+			"Allow PM on possible host mode disconnect\n");
+		pm_runtime_mark_last_busy(musb->controller);
+		pm_runtime_put_autosuspend(musb->controller);
+		musb->session = false;
+		return;
+	default:
+		break;
+	}
+
+	/* No need to do anything if session has not changed */
+	s = devctl & MUSB_DEVCTL_SESSION;
+	if (s == musb->session)
+		return;
+
+	/* Block PM or allow PM? */
+	if (s) {
+		dev_dbg(musb->controller, "Block PM on active session: %02x\n",
+			devctl);
+		error = pm_runtime_get_sync(musb->controller);
+		if (error < 0)
+			dev_err(musb->controller, "Could not enable: %i\n",
+				error);
+	} else {
+		dev_dbg(musb->controller, "Allow PM with no session: %02x\n",
+			devctl);
+		pm_runtime_mark_last_busy(musb->controller);
+		pm_runtime_put_autosuspend(musb->controller);
+	}
+
+	musb->session = s;
+}
+
 /* Only used to provide driver mode change events */
 static void musb_irq_work(struct work_struct *data)
 {
 	struct musb *musb = container_of(data, struct musb, irq_work);
 
+	musb_pm_runtime_check_session(musb);
+
 	if (musb->xceiv->otg->state != musb->xceiv_old_state) {
 		musb->xceiv_old_state = musb->xceiv->otg->state;
 		sysfs_notify(&musb->controller->kobj, NULL, "mode");
diff --git a/drivers/usb/musb/musb_core.h b/drivers/usb/musb/musb_core.h
index b55a776..65288a5 100644
--- a/drivers/usb/musb/musb_core.h
+++ b/drivers/usb/musb/musb_core.h
@@ -378,6 +378,7 @@ struct musb {
 	u8			min_power;	/* vbus for periph, in mA/2 */
 
 	int			port_mode;	/* MUSB_PORT_MODE_* */
+	bool			session;
 	bool			is_host;
 
 	int			a_wait_bcon;	/* VBUS timeout in msecs */
-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/4] usb: musb: Implement session bit based runtime PM for musb-core

[Bin Liu]

On Thu, Aug 18, 2016 at 03:46:47PM -0700, Tony Lindgren wrote:
> We want to keep musb enabled always when the session bit is
> set. This simplifies the PM runtime and allows making it more
> generic across the various glue layers.
> 
> So far the only exception to just following the session bit is
> host mode disconnect where the session bit stays set.
> 
> In that case, just allow PM and let the PM runtime autoidle
> timeout deal with it.
> 
> Signed-off-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>
> ---
>  drivers/usb/musb/musb_core.c | 70 ++++++++++++++++++++++++++++++++++++++++++++
>  drivers/usb/musb/musb_core.h |  1 +
>  2 files changed, 71 insertions(+)
> 
> diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c
> index 74fc306..9c4e1a5 100644
> --- a/drivers/usb/musb/musb_core.c
> +++ b/drivers/usb/musb/musb_core.c
> @@ -1831,11 +1831,81 @@ static const struct attribute_group musb_attr_group = {
>  	.attrs = musb_attributes,
>  };
>  
> +#define MUSB_QUIRK_B_INVALID_VBUS_91	(MUSB_DEVCTL_BDEVICE | \
> +					 (2 << MUSB_DEVCTL_VBUS_SHIFT) | \
> +					 MUSB_DEVCTL_SESSION)
> +#define MUSB_QUIRK_A_DISCONNECT_19	((3 << MUSB_DEVCTL_VBUS_SHIFT) | \
> +					 MUSB_DEVCTL_SESSION)
> +
> +/*
> + * Check the musb devctl session bit to determine if we want to
> + * allow PM runtime for the device. In general, we want to keep things
> + * active when the session bit is set except after host disconnect.
> + *
> + * Only called from musb_irq_work. If this ever needs to get called
> + * elsewhere, proper locking must be implemented for musb->session.
> + */
> +static void musb_pm_runtime_check_session(struct musb *musb)
> +{
> +	u8 devctl, s;
> +	int error;
> +
> +	devctl = musb_readb(musb->mregs, MUSB_DEVCTL);
> +
> +	/* Handle session status quirks first */
> +	s = MUSB_DEVCTL_FSDEV | MUSB_DEVCTL_LSDEV |
> +		MUSB_DEVCTL_HR;
> +	switch (devctl & ~s) {
> +	case MUSB_QUIRK_B_INVALID_VBUS_91:
> +		if (musb->session)
> +			break;
> +		dev_dbg(musb->controller,
> +			"Allow PM as device with invalid vbus: %02x\n",
> +			devctl);
> +		return;
> +	case MUSB_QUIRK_A_DISCONNECT_19:
> +		if (!musb->session)
> +			break;
> +		dev_dbg(musb->controller,
> +			"Allow PM on possible host mode disconnect\n");
> +		pm_runtime_mark_last_busy(musb->controller);
> +		pm_runtime_put_autosuspend(musb->controller);
> +		musb->session = false;
> +		return;
> +	default:
> +		break;
> +	}
> +
> +	/* No need to do anything if session has not changed */
> +	s = devctl & MUSB_DEVCTL_SESSION;
> +	if (s == musb->session)
> +		return;
> +
> +	/* Block PM or allow PM? */
> +	if (s) {
> +		dev_dbg(musb->controller, "Block PM on active session: %02x\n",
> +			devctl);
> +		error = pm_runtime_get_sync(musb->controller);
> +		if (error < 0)
> +			dev_err(musb->controller, "Could not enable: %i\n",
> +				error);
> +	} else {
> +		dev_dbg(musb->controller, "Allow PM with no session: %02x\n",
> +			devctl);

I changed dev_dbg() to musb_dbg(). musb core already moves to
tracepoints for debug.

> +		pm_runtime_mark_last_busy(musb->controller);
> +		pm_runtime_put_autosuspend(musb->controller);
> +	}
> +
> +	musb->session = s;
> +}
> +
>  /* Only used to provide driver mode change events */
>  static void musb_irq_work(struct work_struct *data)
>  {
>  	struct musb *musb = container_of(data, struct musb, irq_work);
>  
> +	musb_pm_runtime_check_session(musb);
> +
>  	if (musb->xceiv->otg->state != musb->xceiv_old_state) {
>  		musb->xceiv_old_state = musb->xceiv->otg->state;
>  		sysfs_notify(&musb->controller->kobj, NULL, "mode");
> diff --git a/drivers/usb/musb/musb_core.h b/drivers/usb/musb/musb_core.h
> index b55a776..65288a5 100644
> --- a/drivers/usb/musb/musb_core.h
> +++ b/drivers/usb/musb/musb_core.h
> @@ -378,6 +378,7 @@ struct musb {
>  	u8			min_power;	/* vbus for periph, in mA/2 */
>  
>  	int			port_mode;	/* MUSB_PORT_MODE_* */
> +	bool			session;
>  	bool			is_host;
>  
>  	int			a_wait_bcon;	/* VBUS timeout in msecs */
> -- 
> 2.8.1
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/4] usb: musb: Implement session bit based runtime PM for musb-core

[Tony Lindgren]

* Bin Liu <b-liu-l0cyMroinI0@public.gmane.org> [160902 07:03]:
> On Thu, Aug 18, 2016 at 03:46:47PM -0700, Tony Lindgren wrote:
> > +	} else {
> > +		dev_dbg(musb->controller, "Allow PM with no session: %02x\n",
> > +			devctl);
> 
> I changed dev_dbg() to musb_dbg(). musb core already moves to
> tracepoints for debug.

OK nice :)

Thanks,

Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 2/4] usb: musb: Prepare dsps glue layer for PM runtime support

[Tony Lindgren]

We want to be polling the state when nothing is connected.
Let's change the polling logic in preparation for PM runtime
support.

Signed-off-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>
---
 drivers/usb/musb/musb_dsps.c | 24 +++++++++++++++++++-----
 1 file changed, 19 insertions(+), 5 deletions(-)

diff --git a/drivers/usb/musb/musb_dsps.c b/drivers/usb/musb/musb_dsps.c
index 2537179..ec0369b 100644
--- a/drivers/usb/musb/musb_dsps.c
+++ b/drivers/usb/musb/musb_dsps.c
@@ -247,6 +247,10 @@ static void otg_timer(unsigned long _musb)
 
 	spin_lock_irqsave(&musb->lock, flags);
 	switch (musb->xceiv->otg->state) {
+	case OTG_STATE_A_WAIT_VRISE:
+		mod_timer(&glue->timer, jiffies +
+				msecs_to_jiffies(wrp->poll_timeout));
+		break;
 	case OTG_STATE_A_WAIT_BCON:
 		musb_writeb(musb->mregs, MUSB_DEVCTL, 0);
 		skip_session = 1;
@@ -338,7 +342,8 @@ static irqreturn_t dsps_interrupt(int irq, void *hci)
 			MUSB_HST_MODE(musb);
 			musb->xceiv->otg->default_a = 1;
 			musb->xceiv->otg->state = OTG_STATE_A_WAIT_VRISE;
-			del_timer(&glue->timer);
+			mod_timer(&glue->timer, jiffies +
+				  msecs_to_jiffies(wrp->poll_timeout));
 		} else {
 			musb->is_active = 0;
 			MUSB_DEV_MODE(musb);
@@ -358,11 +363,17 @@ static irqreturn_t dsps_interrupt(int irq, void *hci)
 	if (musb->int_tx || musb->int_rx || musb->int_usb)
 		ret |= musb_interrupt(musb);
 
-	/* Poll for ID change in OTG port mode */
-	if (musb->xceiv->otg->state == OTG_STATE_B_IDLE &&
-			musb->port_mode == MUSB_PORT_MODE_DUAL_ROLE)
+	/* Poll for ID change and connect */
+	switch (musb->xceiv->otg->state) {
+	case OTG_STATE_B_IDLE:
+	case OTG_STATE_A_WAIT_BCON:
 		mod_timer(&glue->timer, jiffies +
-				msecs_to_jiffies(wrp->poll_timeout));
+			  msecs_to_jiffies(wrp->poll_timeout));
+		break;
+	default:
+		break;
+	}
+
 out:
 	spin_unlock_irqrestore(&musb->lock, flags);
 
@@ -461,6 +472,9 @@ static int dsps_musb_init(struct musb *musb)
 		musb_writeb(musb->mregs, MUSB_BABBLE_CTL, val);
 	}
 
+	mod_timer(&glue->timer, jiffies +
+		  msecs_to_jiffies(glue->wrp->poll_timeout));
+
 	return dsps_musb_dbg_init(musb, glue);
 }
 
-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/4] usb: musb: Prepare dsps glue layer for PM runtime support

[Bin Liu]

On Thu, Aug 18, 2016 at 03:46:48PM -0700, Tony Lindgren wrote:
> We want to be polling the state when nothing is connected.
> Let's change the polling logic in preparation for PM runtime
> support.
> 
> Signed-off-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>
> ---
>  drivers/usb/musb/musb_dsps.c | 24 +++++++++++++++++++-----
>  1 file changed, 19 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/usb/musb/musb_dsps.c b/drivers/usb/musb/musb_dsps.c
> index 2537179..ec0369b 100644
> --- a/drivers/usb/musb/musb_dsps.c
> +++ b/drivers/usb/musb/musb_dsps.c
> @@ -247,6 +247,10 @@ static void otg_timer(unsigned long _musb)
>  
>  	spin_lock_irqsave(&musb->lock, flags);
>  	switch (musb->xceiv->otg->state) {
> +	case OTG_STATE_A_WAIT_VRISE:
> +		mod_timer(&glue->timer, jiffies +
> +				msecs_to_jiffies(wrp->poll_timeout));
> +		break;
>  	case OTG_STATE_A_WAIT_BCON:
>  		musb_writeb(musb->mregs, MUSB_DEVCTL, 0);
>  		skip_session = 1;
> @@ -338,7 +342,8 @@ static irqreturn_t dsps_interrupt(int irq, void *hci)
>  			MUSB_HST_MODE(musb);
>  			musb->xceiv->otg->default_a = 1;
>  			musb->xceiv->otg->state = OTG_STATE_A_WAIT_VRISE;
> -			del_timer(&glue->timer);
> +			mod_timer(&glue->timer, jiffies +
> +				  msecs_to_jiffies(wrp->poll_timeout));
>  		} else {
>  			musb->is_active = 0;
>  			MUSB_DEV_MODE(musb);
> @@ -358,11 +363,17 @@ static irqreturn_t dsps_interrupt(int irq, void *hci)
>  	if (musb->int_tx || musb->int_rx || musb->int_usb)
>  		ret |= musb_interrupt(musb);
>  
> -	/* Poll for ID change in OTG port mode */
> -	if (musb->xceiv->otg->state == OTG_STATE_B_IDLE &&
> -			musb->port_mode == MUSB_PORT_MODE_DUAL_ROLE)
> +	/* Poll for ID change and connect */
> +	switch (musb->xceiv->otg->state) {
> +	case OTG_STATE_B_IDLE:
> +	case OTG_STATE_A_WAIT_BCON:
>  		mod_timer(&glue->timer, jiffies +
> -				msecs_to_jiffies(wrp->poll_timeout));
> +			  msecs_to_jiffies(wrp->poll_timeout));

Unintentional change? I undo'ed it in my tree.

> +		break;
> +	default:
> +		break;
> +	}
> +
>  out:
>  	spin_unlock_irqrestore(&musb->lock, flags);
>  
> @@ -461,6 +472,9 @@ static int dsps_musb_init(struct musb *musb)
>  		musb_writeb(musb->mregs, MUSB_BABBLE_CTL, val);
>  	}
>  
> +	mod_timer(&glue->timer, jiffies +
> +		  msecs_to_jiffies(glue->wrp->poll_timeout));
> +
>  	return dsps_musb_dbg_init(musb, glue);
>  }
>  
> -- 
> 2.8.1
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/4] usb: musb: Prepare dsps glue layer for PM runtime support

[Tony Lindgren]

* Bin Liu <b-liu-l0cyMroinI0@public.gmane.org> [160902 07:05]:
> On Thu, Aug 18, 2016 at 03:46:48PM -0700, Tony Lindgren wrote:
> > +	case OTG_STATE_A_WAIT_BCON:
> >  		mod_timer(&glue->timer, jiffies +
> > -				msecs_to_jiffies(wrp->poll_timeout));
> > +			  msecs_to_jiffies(wrp->poll_timeout));
> 
> Unintentional change? I undo'ed it in my tree.

Yes that's not needed thanks.

Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 3/4] usb: musb: Add PM runtime support for MUSB DSPS glue layer

[Tony Lindgren]

We can now just use PM runtime autoidle support as musb core
keeps things enabled when the devctl session bit is set. And
there's no need for dsps_musb_try_idle() so let's just remove
it.

Note that as cppi41 dma is clocked by musb, this only makes
PM work for dsps glue layer if CONFIG_MUSB_PIO_ONLY=y and
cppi41.ko is unloaded. This will get fixed when cppi41.c has
PM runtime implemented.

Signed-off-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>
---
 drivers/usb/musb/musb_dsps.c | 58 +++++++++++++-------------------------------
 1 file changed, 17 insertions(+), 41 deletions(-)

diff --git a/drivers/usb/musb/musb_dsps.c b/drivers/usb/musb/musb_dsps.c
index ec0369b..dc43521 100644
--- a/drivers/usb/musb/musb_dsps.c
+++ b/drivers/usb/musb/musb_dsps.c
@@ -145,43 +145,6 @@ static const struct debugfs_reg32 dsps_musb_regs[] = {
 	{ "mode",		0xe8 },
 };
 
-static void dsps_musb_try_idle(struct musb *musb, unsigned long timeout)
-{
-	struct device *dev = musb->controller;
-	struct dsps_glue *glue = dev_get_drvdata(dev->parent);
-
-	if (timeout == 0)
-		timeout = jiffies + msecs_to_jiffies(3);
-
-	/* Never idle if active, or when VBUS timeout is not set as host */
-	if (musb->is_active || (musb->a_wait_bcon == 0 &&
-			musb->xceiv->otg->state == OTG_STATE_A_WAIT_BCON)) {
-		dev_dbg(musb->controller, "%s active, deleting timer\n",
-				usb_otg_state_string(musb->xceiv->otg->state));
-		del_timer(&glue->timer);
-		glue->last_timer = jiffies;
-		return;
-	}
-	if (musb->port_mode != MUSB_PORT_MODE_DUAL_ROLE)
-		return;
-
-	if (!musb->g.dev.driver)
-		return;
-
-	if (time_after(glue->last_timer, timeout) &&
-				timer_pending(&glue->timer)) {
-		dev_dbg(musb->controller,
-			"Longer idle timer already pending, ignoring...\n");
-		return;
-	}
-	glue->last_timer = timeout;
-
-	dev_dbg(musb->controller, "%s inactive, starting idle timer for %u ms\n",
-		usb_otg_state_string(musb->xceiv->otg->state),
-			jiffies_to_msecs(timeout - jiffies));
-	mod_timer(&glue->timer, timeout);
-}
-
 /**
  * dsps_musb_enable - enable interrupts
  */
@@ -206,7 +169,6 @@ static void dsps_musb_enable(struct musb *musb)
 			musb->port_mode == MUSB_PORT_MODE_DUAL_ROLE)
 		mod_timer(&glue->timer, jiffies +
 				msecs_to_jiffies(wrp->poll_timeout));
-	dsps_musb_try_idle(musb, 0);
 }
 
 /**
@@ -236,6 +198,11 @@ static void otg_timer(unsigned long _musb)
 	u8 devctl;
 	unsigned long flags;
 	int skip_session = 0;
+	int err;
+
+	err = pm_runtime_get_sync(dev);
+	if (err < 0)
+		dev_err(dev, "Poll could not pm_runtime_get: %i\n", err);
 
 	/*
 	 * We poll because DSPS IP's won't expose several OTG-critical
@@ -279,6 +246,9 @@ static void otg_timer(unsigned long _musb)
 		break;
 	}
 	spin_unlock_irqrestore(&musb->lock, flags);
+
+	pm_runtime_mark_last_busy(dev);
+	pm_runtime_put_autosuspend(dev);
 }
 
 static irqreturn_t dsps_interrupt(int irq, void *hci)
@@ -634,7 +604,6 @@ static struct musb_platform_ops dsps_ops = {
 	.enable		= dsps_musb_enable,
 	.disable	= dsps_musb_disable,
 
-	.try_idle	= dsps_musb_try_idle,
 	.set_mode	= dsps_musb_set_mode,
 	.recover	= dsps_musb_recover,
 };
@@ -798,6 +767,8 @@ static int dsps_probe(struct platform_device *pdev)
 
 	platform_set_drvdata(pdev, glue);
 	pm_runtime_enable(&pdev->dev);
+	pm_runtime_use_autosuspend(&pdev->dev);
+	pm_runtime_set_autosuspend_delay(&pdev->dev, 200);
 
 	ret = pm_runtime_get_sync(&pdev->dev);
 	if (ret < 0) {
@@ -809,11 +780,15 @@ static int dsps_probe(struct platform_device *pdev)
 	if (ret)
 		goto err3;
 
+	pm_runtime_mark_last_busy(&pdev->dev);
+	pm_runtime_put_autosuspend(&pdev->dev);
+
 	return 0;
 
 err3:
-	pm_runtime_put(&pdev->dev);
+	pm_runtime_put_sync(&pdev->dev);
 err2:
+	pm_runtime_dont_use_autosuspend(&pdev->dev);
 	pm_runtime_disable(&pdev->dev);
 	return ret;
 }
@@ -825,7 +800,8 @@ static int dsps_remove(struct platform_device *pdev)
 	platform_device_unregister(glue->musb);
 
 	/* disable usbss clocks */
-	pm_runtime_put(&pdev->dev);
+	pm_runtime_dont_use_autosuspend(&pdev->dev);
+	pm_runtime_put_sync(&pdev->dev);
 	pm_runtime_disable(&pdev->dev);
 
 	return 0;
-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 4/4] usb: musb: Simplify PM runtime for 2430 glue layer

[Tony Lindgren]

With musb core now blocking PM based on the devctl status
bit, we can remove related quirks from the 2430 glue layer
and simplify PM runtime further.

Lets's also use musb->controller instead of dev to make it
clear we make the PM runtime calls for the core, not the
glue layer.

And we can now also lower the autoidle timeout.

Signed-off-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>
---
 drivers/usb/musb/omap2430.c | 76 ++++++---------------------------------------
 1 file changed, 10 insertions(+), 66 deletions(-)

diff --git a/drivers/usb/musb/omap2430.c b/drivers/usb/musb/omap2430.c
index 0b4cec9..1ab6973 100644
--- a/drivers/usb/musb/omap2430.c
+++ b/drivers/usb/musb/omap2430.c
@@ -49,9 +49,6 @@ struct omap2430_glue {
 	enum musb_vbus_id_status status;
 	struct work_struct	omap_musb_mailbox_work;
 	struct device		*control_otghs;
-	bool			cable_connected;
-	bool			enabled;
-	bool			powered;
 };
 #define glue_to_musb(g)		platform_get_drvdata(g->musb)
 
@@ -141,45 +138,6 @@ static inline void omap2430_low_level_init(struct musb *musb)
 	musb_writel(musb->mregs, OTG_FORCESTDBY, l);
 }
 
-/*
- * We can get multiple cable events so we need to keep track
- * of the power state. Only keep power enabled if USB cable is
- * connected and a gadget is started.
- */
-static void omap2430_set_power(struct musb *musb, bool enabled, bool cable)
-{
-	struct device *dev = musb->controller;
-	struct omap2430_glue *glue = dev_get_drvdata(dev->parent);
-	bool power_up;
-	int res;
-
-	if (glue->enabled != enabled)
-		glue->enabled = enabled;
-
-	if (glue->cable_connected != cable)
-		glue->cable_connected = cable;
-
-	power_up = glue->enabled && glue->cable_connected;
-	if (power_up == glue->powered) {
-		dev_warn(musb->controller, "power state already %i\n",
-			 power_up);
-		return;
-	}
-
-	glue->powered = power_up;
-
-	if (power_up) {
-		res = pm_runtime_get_sync(musb->controller);
-		if (res < 0) {
-			dev_err(musb->controller, "could not enable: %i", res);
-			glue->powered = false;
-		}
-	} else {
-		pm_runtime_mark_last_busy(musb->controller);
-		pm_runtime_put_autosuspend(musb->controller);
-	}
-}
-
 static int omap2430_musb_mailbox(enum musb_vbus_id_status status)
 {
 	struct omap2430_glue	*glue = _glue;
@@ -203,21 +161,15 @@ static int omap2430_musb_mailbox(enum musb_vbus_id_status status)
 static void omap_musb_set_mailbox(struct omap2430_glue *glue)
 {
 	struct musb *musb = glue_to_musb(glue);
-	struct device *dev = musb->controller;
-	struct musb_hdrc_platform_data *pdata = dev_get_platdata(dev);
+	struct musb_hdrc_platform_data *pdata =
+		dev_get_platdata(musb->controller);
 	struct omap_musb_board_data *data = pdata->board_data;
 	struct usb_otg *otg = musb->xceiv->otg;
-	bool cable_connected;
-
-	cable_connected = ((glue->status == MUSB_ID_GROUND) ||
-			   (glue->status == MUSB_VBUS_VALID));
-
-	if (cable_connected)
-		omap2430_set_power(musb, glue->enabled, cable_connected);
 
+	pm_runtime_get_sync(musb->controller);
 	switch (glue->status) {
 	case MUSB_ID_GROUND:
-		dev_dbg(dev, "ID GND\n");
+		dev_dbg(musb->controller, "ID GND\n");
 
 		otg->default_a = true;
 		musb->xceiv->otg->state = OTG_STATE_A_IDLE;
@@ -230,7 +182,7 @@ static void omap_musb_set_mailbox(struct omap2430_glue *glue)
 		break;
 
 	case MUSB_VBUS_VALID:
-		dev_dbg(dev, "VBUS Connect\n");
+		dev_dbg(musb->controller, "VBUS Connect\n");
 
 		otg->default_a = false;
 		musb->xceiv->otg->state = OTG_STATE_B_IDLE;
@@ -240,7 +192,7 @@ static void omap_musb_set_mailbox(struct omap2430_glue *glue)
 
 	case MUSB_ID_FLOAT:
 	case MUSB_VBUS_OFF:
-		dev_dbg(dev, "VBUS Disconnect\n");
+		dev_dbg(musb->controller, "VBUS Disconnect\n");
 
 		musb->xceiv->last_event = USB_EVENT_NONE;
 		if (musb->gadget_driver)
@@ -253,12 +205,10 @@ static void omap_musb_set_mailbox(struct omap2430_glue *glue)
 			USB_MODE_DISCONNECT);
 		break;
 	default:
-		dev_dbg(dev, "ID float\n");
+		dev_dbg(musb->controller, "ID float\n");
 	}
-
-	if (!cable_connected)
-		omap2430_set_power(musb, glue->enabled, cable_connected);
-
+	pm_runtime_mark_last_busy(musb->controller);
+	pm_runtime_put_autosuspend(musb->controller);
 	atomic_notifier_call_chain(&musb->xceiv->notifier,
 			musb->xceiv->last_event, NULL);
 }
@@ -376,8 +326,6 @@ static void omap2430_musb_enable(struct musb *musb)
 	if (!WARN_ON(!musb->phy))
 		phy_power_on(musb->phy);
 
-	omap2430_set_power(musb, true, glue->cable_connected);
-
 	switch (glue->status) {
 
 	case MUSB_ID_GROUND:
@@ -419,8 +367,6 @@ static void omap2430_musb_disable(struct musb *musb)
 	if (glue->status != MUSB_UNKNOWN)
 		omap_control_usb_set_mode(glue->control_otghs,
 			USB_MODE_DISCONNECT);
-
-	omap2430_set_power(musb, false, glue->cable_connected);
 }
 
 static int omap2430_musb_exit(struct musb *musb)
@@ -571,7 +517,7 @@ static int omap2430_probe(struct platform_device *pdev)
 
 	pm_runtime_enable(glue->dev);
 	pm_runtime_use_autosuspend(glue->dev);
-	pm_runtime_set_autosuspend_delay(glue->dev, 500);
+	pm_runtime_set_autosuspend_delay(glue->dev, 100);
 
 	ret = platform_device_add(musb);
 	if (ret) {
@@ -591,11 +537,9 @@ err0:
 static int omap2430_remove(struct platform_device *pdev)
 {
 	struct omap2430_glue *glue = platform_get_drvdata(pdev);
-	struct musb *musb = glue_to_musb(glue);
 
 	pm_runtime_get_sync(glue->dev);
 	platform_device_unregister(glue->musb);
-	omap2430_set_power(musb, false, false);
 	pm_runtime_put_sync(glue->dev);
 	pm_runtime_dont_use_autosuspend(glue->dev);
 	pm_runtime_disable(glue->dev);
-- 
2.8.1

--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/4] Implement PM runtime for musb-core based on session bit

[Andreas Kemnade]

Hi,

On Thu, 18 Aug 2016 15:46:46 -0700
Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org> wrote:

> Hi all,
> 
> Here's a series of patches to simplify musb PM runtime support
> further.
> 
> I finally figured out that we can get rid of most of the glue layer
> specific workarounds by following the devctl session bit in musb core.
> 
> The series also adds PM runtime support to the musb dsps glue layer
> for the PIO mode.
> 
> I've tested basic plugging of cables in peripheral and host mode
> on omap3 and am335x using omap2plus_defconfig. Please test if you
> can.
> 
Testing on GTA04 hardware (dm3730 + twl4030 as usb-phy), device mode,
charging: it does not make things worse than it is without it. Same
problems,
phy->power_count goes down to -1, so no gadget, no charging.
Fixing that (I have some nearly-ready things in my queue) causes
device mode and charging to work (as it does without
this patch-series).

Regards,
Andreas
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/4] Implement PM runtime for musb-core based on session bit

[Tony Lindgren]

* Andreas Kemnade <andreas-cLv4Z9ELZ06ZuzBka8ofvg@public.gmane.org> [160821 22:29]:
> Hi,
> 
> On Thu, 18 Aug 2016 15:46:46 -0700
> Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org> wrote:
> 
> > Hi all,
> > 
> > Here's a series of patches to simplify musb PM runtime support
> > further.
> > 
> > I finally figured out that we can get rid of most of the glue layer
> > specific workarounds by following the devctl session bit in musb core.
> > 
> > The series also adds PM runtime support to the musb dsps glue layer
> > for the PIO mode.
> > 
> > I've tested basic plugging of cables in peripheral and host mode
> > on omap3 and am335x using omap2plus_defconfig. Please test if you
> > can.
> > 
> Testing on GTA04 hardware (dm3730 + twl4030 as usb-phy), device mode,
> charging: it does not make things worse than it is without it. Same
> problems,

OK good to hear and thanks for testing.

> phy->power_count goes down to -1, so no gadget, no charging.
> Fixing that (I have some nearly-ready things in my queue) causes
> device mode and charging to work (as it does without
> this patch-series).

OK. I'll try to test that phy patch you sent today.

Regards,

Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/4] Implement PM runtime for musb-core based on session bit

[Bin Liu]

On Thu, Aug 18, 2016 at 03:46:46PM -0700, Tony Lindgren wrote:
> Hi all,
> 
> Here's a series of patches to simplify musb PM runtime support further.
> 
> I finally figured out that we can get rid of most of the glue layer
> specific workarounds by following the devctl session bit in musb core.
> 
> The series also adds PM runtime support to the musb dsps glue layer
> for the PIO mode.
> 
> I've tested basic plugging of cables in peripheral and host mode
> on omap3 and am335x using omap2plus_defconfig. Please test if you
> can.
> 
> Regards,
> 
> Tony
> 
> 
> Tony Lindgren (4):
>   usb: musb: Implement session bit based runtime PM for musb-core
>   usb: musb: Prepare dsps glue layer for PM runtime support
>   usb: musb: Add PM runtime support for MUSB DSPS glue layer
>   usb: musb: Simplify PM runtime for 2430 glue layer
> 
>  drivers/usb/musb/musb_core.c | 70 +++++++++++++++++++++++++++++++++++++
>  drivers/usb/musb/musb_core.h |  1 +
>  drivers/usb/musb/musb_dsps.c | 82 +++++++++++++++++++-------------------------
>  drivers/usb/musb/omap2430.c  | 76 ++++++----------------------------------
>  4 files changed, 117 insertions(+), 112 deletions(-)
 
Signed-off-by: Bin Liu <b-liu-l0cyMroinI0@public.gmane.org>
with minor tweaks in patch 1/4 and 2/4.

Regards,
-Bin.

> -- 
> 2.8.1
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html