Re: [PATCH 1/5] backlight: Add GPIO-based backlight driver
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95 more lines

Hi Jingoo,

On Monday 26 November 2012 09:49:36 Jingoo Han wrote:

On Saturday, November 24, 2012 1:35 AM, Laurent Pinchart wrote

quoted

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

CC'ed Andrew Morton

It looks good. Also, I tested this patch with SMDK4210 board.
Acked-by: Jingoo Han <redacted>

Thank you.

Unless Richard wants to take 1/5 and 2/5, I plan to push them through either 
the Renesas ARM or Renesas SH tree with the board code patches.

-- 
Regards,

Laurent Pinchart

On 11/26/2012 10:49 AM, Jingoo Han wrote:

On Saturday, November 24, 2012 1:35 AM, Laurent Pinchart wrote

[...]

quoted

+static int gpio_backlight_check_fb(struct backlight_device *bl,
+				   struct fb_info *info)
+{
+	struct gpio_backlight *gbl = bl_get_data(bl);
+
+	return gbl->fbdev = info->dev;

I think it makes sense to return true if fbdev is NULL, to provide a simple
fallback for systems with only one framebuffer device.

quoted

+}
+

[...]

quoted

+#ifdef CONFIG_PM
+static int gpio_backlight_suspend(struct device *dev)
+{
+	struct backlight_device *bl = dev_get_drvdata(dev);
+	struct gpio_backlight *gbl = bl_get_data(bl);
+
+	gpio_set_value(gbl->gpio, !gbl->active);
+
+	return 0;
+}
+
+static int gpio_backlight_resume(struct device *dev)
+{
+	struct backlight_device *bl = dev_get_drvdata(dev);
+
+	backlight_update_status(bl);
+	return 0;
+}

If you use BL_CORE_SUSPENDRESUME you can get rid of the custom
suspend/resume handlers.

quoted

+
+static SIMPLE_DEV_PM_OPS(gpio_backlight_pm_ops, gpio_backlight_suspend,
+			 gpio_backlight_resume);
+
+#endif
+

[...]

Hi Lars-Peter,

On Monday 26 November 2012 11:44:04 Lars-Peter Clausen wrote:

On 11/26/2012 10:49 AM, Jingoo Han wrote:

quoted

On Saturday, November 24, 2012 1:35 AM, Laurent Pinchart wrote

[...]

quoted

quoted

+static int gpio_backlight_check_fb(struct backlight_device *bl,
+				   struct fb_info *info)
+{
+	struct gpio_backlight *gbl = bl_get_data(bl);
+
+	return gbl->fbdev = info->dev;

I think it makes sense to return true if fbdev is NULL, to provide a simple
fallback for systems with only one framebuffer device.

Agreed, I'll change that.

quoted

quoted

+}
+

[...]

quoted

quoted

+#ifdef CONFIG_PM
+static int gpio_backlight_suspend(struct device *dev)
+{
+	struct backlight_device *bl = dev_get_drvdata(dev);
+	struct gpio_backlight *gbl = bl_get_data(bl);
+
+	gpio_set_value(gbl->gpio, !gbl->active);
+
+	return 0;
+}
+
+static int gpio_backlight_resume(struct device *dev)
+{
+	struct backlight_device *bl = dev_get_drvdata(dev);
+
+	backlight_update_status(bl);
+	return 0;
+}

If you use BL_CORE_SUSPENDRESUME you can get rid of the custom
suspend/resume handlers.

Good point, I'll do that.

quoted

quoted

+
+static SIMPLE_DEV_PM_OPS(gpio_backlight_pm_ops, gpio_backlight_suspend,
+			 gpio_backlight_resume);
+
+#endif
+

-- 
Regards,

Laurent Pinchart