[PATCH 1/2] fbdev: sh-mobile: implement MIPI DSI runtime PM support

[PATCH 1/2] fbdev: sh-mobile: implement MIPI DSI runtime PM support

[Guennadi Liakhovetski]

On SH-Mobile platforms using runtime PM with the MIPI DSI driver switches the
DSI Tx link clock on PM events.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
---
 drivers/video/sh_mipi_dsi.c |   15 ++++++++++++++-
 1 files changed, 14 insertions(+), 1 deletions(-)

diff --git a/drivers/video/sh_mipi_dsi.c b/drivers/video/sh_mipi_dsi.c
index 3f3d431..ef19cac 100644
--- a/drivers/video/sh_mipi_dsi.c
+++ b/drivers/video/sh_mipi_dsi.c
@@ -13,6 +13,7 @@
 #include <linux/init.h>
 #include <linux/io.h>
 #include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
 #include <linux/slab.h>
 #include <linux/string.h>
 #include <linux/types.h>
@@ -33,6 +34,7 @@ struct sh_mipi {
 	void __iomem	*base;
 	struct clk	*dsit_clk;
 	struct clk	*dsip_clk;
+	struct device	*dev;
 };
 
 static struct sh_mipi *mipi_dsi[MAX_SH_MIPI_DSI];
@@ -104,6 +106,7 @@ static void mipi_display_on(void *arg, struct fb_info *info)
 {
 	struct sh_mipi *mipi = arg;
 
+	pm_runtime_get_sync(mipi->dev);
 	sh_mipi_dsi_enable(mipi, true);
 }
 
@@ -112,6 +115,7 @@ static void mipi_display_off(void *arg)
 	struct sh_mipi *mipi = arg;
 
 	sh_mipi_dsi_enable(mipi, false);
+	pm_runtime_put(mipi->dev);
 }
 
 static int __init sh_mipi_setup(struct sh_mipi *mipi,
@@ -356,7 +360,9 @@ static int __init sh_mipi_probe(struct platform_device *pdev)
 		goto emap;
 	}
 
-	mipi->dsit_clk = clk_get(&pdev->dev, "dsit_clk");
+	mipi->dev = &pdev->dev;
+
+	mipi->dsit_clk = clk_get(NULL, "dsit_clk");
 	if (IS_ERR(mipi->dsit_clk)) {
 		ret = PTR_ERR(mipi->dsit_clk);
 		goto eclktget;
@@ -405,6 +411,9 @@ static int __init sh_mipi_probe(struct platform_device *pdev)
 
 	mipi_dsi[idx] = mipi;
 
+	pm_runtime_enable(&pdev->dev);
+	pm_runtime_resume(&pdev->dev);
+
 	ret = sh_mipi_setup(mipi, pdata);
 	if (ret < 0)
 		goto emipisetup;
@@ -416,11 +425,13 @@ static int __init sh_mipi_probe(struct platform_device *pdev)
 	pdata->lcd_chan->board_cfg.board_data = mipi;
 	pdata->lcd_chan->board_cfg.display_on = mipi_display_on;
 	pdata->lcd_chan->board_cfg.display_off = mipi_display_off;
+	pdata->lcd_chan->board_cfg.owner = THIS_MODULE;
 
 	return 0;
 
 emipisetup:
 	mipi_dsi[idx] = NULL;
+	pm_runtime_disable(&pdev->dev);
 	clk_disable(mipi->dsip_clk);
 eclkpon:
 	clk_disable(mipi->dsit_clk);
@@ -467,10 +478,12 @@ static int __exit sh_mipi_remove(struct platform_device *pdev)
 	if (ret < 0)
 		return ret;
 
+	pdata->lcd_chan->board_cfg.owner = NULL;
 	pdata->lcd_chan->board_cfg.display_on = NULL;
 	pdata->lcd_chan->board_cfg.display_off = NULL;
 	pdata->lcd_chan->board_cfg.board_data = NULL;
 
+	pm_runtime_disable(&pdev->dev);
 	clk_disable(mipi->dsip_clk);
 	clk_disable(mipi->dsit_clk);
 	clk_put(mipi->dsit_clk);
-- 
1.7.2.3

Re: [PATCH 1/2] fbdev: sh-mobile: implement MIPI DSI runtime PM support

[Paul Mundt]

On Wed, Dec 22, 2010 at 10:52:18AM +0100, Guennadi Liakhovetski wrote:
> @@ -356,7 +360,9 @@ static int __init sh_mipi_probe(struct platform_device *pdev)
>  		goto emap;
>  	}
>  
> -	mipi->dsit_clk = clk_get(&pdev->dev, "dsit_clk");
> +	mipi->dev = &pdev->dev;
> +
> +	mipi->dsit_clk = clk_get(NULL, "dsit_clk");

Why did you drop &pdev->dev from clk_get()?

Re: [PATCH 1/2] fbdev: sh-mobile: implement MIPI DSI runtime PM

[Guennadi Liakhovetski]

On Thu, 23 Dec 2010, Paul Mundt wrote:

> On Wed, Dec 22, 2010 at 10:52:18AM +0100, Guennadi Liakhovetski wrote:
> > @@ -356,7 +360,9 @@ static int __init sh_mipi_probe(struct platform_device *pdev)
> >  		goto emap;
> >  	}
> >  
> > -	mipi->dsit_clk = clk_get(&pdev->dev, "dsit_clk");
> > +	mipi->dev = &pdev->dev;
> > +
> > +	mipi->dsit_clk = clk_get(NULL, "dsit_clk");
> 
> Why did you drop &pdev->dev from clk_get()?

Because that clock is not associated with that device, it sisn't work 
correctly until now. You also cannot associate it with that device 
exclusively, because it serves two devices...

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/

Re: [PATCH 1/2] fbdev: sh-mobile: implement MIPI DSI runtime PM support

[Paul Mundt]

On Thu, Dec 23, 2010 at 01:08:34AM +0100, Guennadi Liakhovetski wrote:
> On Thu, 23 Dec 2010, Paul Mundt wrote:
> 
> > On Wed, Dec 22, 2010 at 10:52:18AM +0100, Guennadi Liakhovetski wrote:
> > > @@ -356,7 +360,9 @@ static int __init sh_mipi_probe(struct platform_device *pdev)
> > >  		goto emap;
> > >  	}
> > >  
> > > -	mipi->dsit_clk = clk_get(&pdev->dev, "dsit_clk");
> > > +	mipi->dev = &pdev->dev;
> > > +
> > > +	mipi->dsit_clk = clk_get(NULL, "dsit_clk");
> > 
> > Why did you drop &pdev->dev from clk_get()?
> 
> Because that clock is not associated with that device, it sisn't work 
> correctly until now. You also cannot associate it with that device 
> exclusively, because it serves two devices...
> 
I'm not sure I see what the problem is. If there is no special lookup
matching this clock to the device pointer then the lookup will resolve to
whatever generically provides that clock. There is also nothing that
stops someone from providing the same clock to multiple devices via the
same lookup string, this is largely what clock lookups are for in the
first place. Any move away from a device pointer to clk_get() is a step
in the wrong direction.

Re: [PATCH 1/2] fbdev: sh-mobile: implement MIPI DSI runtime PM

[Guennadi Liakhovetski]

On Fri, 24 Dec 2010, Paul Mundt wrote:

> On Thu, Dec 23, 2010 at 01:08:34AM +0100, Guennadi Liakhovetski wrote:
> > On Thu, 23 Dec 2010, Paul Mundt wrote:
> > 
> > > On Wed, Dec 22, 2010 at 10:52:18AM +0100, Guennadi Liakhovetski wrote:
> > > > @@ -356,7 +360,9 @@ static int __init sh_mipi_probe(struct platform_device *pdev)
> > > >  		goto emap;
> > > >  	}
> > > >  
> > > > -	mipi->dsit_clk = clk_get(&pdev->dev, "dsit_clk");
> > > > +	mipi->dev = &pdev->dev;
> > > > +
> > > > +	mipi->dsit_clk = clk_get(NULL, "dsit_clk");
> > > 
> > > Why did you drop &pdev->dev from clk_get()?
> > 
> > Because that clock is not associated with that device, it sisn't work 
> > correctly until now. You also cannot associate it with that device 
> > exclusively, because it serves two devices...
> > 
> I'm not sure I see what the problem is. If there is no special lookup
> matching this clock to the device pointer then the lookup will resolve to
> whatever generically provides that clock. There is also nothing that
> stops someone from providing the same clock to multiple devices via the
> same lookup string, this is largely what clock lookups are for in the
> first place. Any move away from a device pointer to clk_get() is a step
> in the wrong direction.

That device also has other clocks, so, previously, with the device pointer 
but with an unmatching device string I was getting a wrong clock - the one 
from the device... So, it is a step in the "right" direction in the sense, 
that I am now getting a correct clock. But yes, if I can associate one 
clock with several devices, that's even better! I'll try that.

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/