Re: [PATCH] ACPI: ibm-acpi: improve backlight power handling (v2)

[Richard Purdie <rpurdie@rpsys.net>]

On Mon, 2007-02-26 at 22:25 +0100, Jiri Kosina wrote:
> On Mon, 26 Feb 2007, Henrique de Moraes Holschuh wrote:
> 
> >  static int brightness_update_status(struct backlight_device *bd)
> >  {
> > -	return brightness_set(bd->props.brightness);
> > +	return brightness_set(
> > +		(bd->props.fb_blank == FB_BLANK_UNBLANK &&
> > +		 bd->props.power == FB_BLANK_UNBLANK) ?
> > +				bd->props.brightness : 0);
> >  }
> 
> Are you sure about the '&&'? The original patch I submitted to you earlier 
> today was checking for (bd->props.fb_blank == FB_BLANK_UNBLANK || 
> bd->props.power == FB_BLANK_UNBLANK), and I tested it (to some extent) and 
> it worked well - no sudden unblanking without reason, no blinking, etc.
> 
> I also think that common sense implies that the condition should be 
> logical or - backlight layer could request blanking without requesting 
> powering the device off, right? We want to handle unblanking from such 
> situation properly, which doesn't necessairly mean we will get 
> bd->props.power == FB_BLANK_UNBLANK, right?

In the above context, && is correct, || isn't.

We want to blank (set to 0) if either fb_blank or power isn't set to
FB_BLANK_UNBLANK. This is the same as setting to brightness if both
fb_blank and power are set to FB_BLANK_UNBLANK. This is what the above
expression does.

Richard