Re: [Bugme-new] [Bug 40492] New: FB_BACKLIGHT should be set by

On Thu, 04 Aug 2011 14:04:46 +0200 David wrote:

On Thursday 04 August 2011 07:55:16 you wrote:

quoted

On Mit, 2011-08-03 at 17:55 -0700, Randy Dunlap wrote:

quoted

On Wed, 3 Aug 2011 18:55:41 GMT bugzilla-daemon@bugzilla.kernel.org wrote:

quoted

https://bugzilla.kernel.org/show_bug.cgi?id@492

           Summary: FB_BACKLIGHT should be set by
           BACKLIGHT_CLASS_DEVICE Product: Other
           Version: 2.5
    
    Kernel Version: 3.0.0
    
          Platform: All
        
        OS/Version: Linux
        
              Tree: Mainline
            
            Status: NEW
          
          Severity: normal
          Priority: P1
         
         Component: Configuration
        
        AssignedTo: other_configuration@kernel-bugs.osdl.org
        ReportedBy: StormByte@gmail.com
        Regression: No

Created an attachment (idg432)

 --> (https://bugzilla.kernel.org/attachment.cgi?idg432)

Kernel's config

I discovered a scenario in which menuconfig fails to set
FB_BACKLIGHT symbol when needed.

In my case, I have propietary nvidia drivers once loaded Xorg, but
in the meanwhile, I have configured an uvesafb framebuffer (thus,
no other framebuffer active nor compiled).

System works ok, but when trying to install nvidiabl[1] (to control
nvidia's backlight) I find that FB_BACKLIGHT is not set where it
should be because I have BACKLIGHT_CLASS_DEVICE set.

The result is that nvidiabl does not compile.

Sounds like an nvidiabl bug. I'm successfully using backlight control
with radeon KMS with

# CONFIG_FB_BACKLIGHT is not set
CONFIG_BACKLIGHT_CLASS_DEVICE=m

It does not seem a bug in nvidiabl, it needs this symbol in order to get 2 
defines (from include/linux/fb.h)
394 #ifdef CONFIG_FB_BACKLIGHT
395 /* Settings for the generic backlight code */
396 #define FB_BACKLIGHT_LEVELS     128
397 #define FB_BACKLIGHT_MAX        0xFF
398 #endif

I think nvidiabl uses those instead of having them hardcoded in its source, 
and thus not a bug really.

David,

You could propose a patch that makes those 2 defines not be conditional
on the CONFIB_FB_BACKLIGHT symbol.

Or you could propose a patch that makes the FB_BACKLIGHT symbol
user-configurable.  Currently it is not user-configurable since it
does not have a prompt string.  I think all it would take is
changing the "bool" line to include a prompt string and adding some
help text.  Something like (not tested, probably needs some
wordsmithing):

config FB_BACKLIGHT
	bool "Generic video backlight support"
	depends on FB
	select BACKLIGHT_LCD_SUPPORT
	select BACKLIGHT_CLASS_DEVICE
	default n
	help
	  This option provides a little video backlight support.
	  Most drivers that need it enable it automatically.

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***