Re: Re: [PATCH] mach64 atari patch

["Antonino A. Daplas" <adaplas@gmail.com>]

James Simmons wrote:
>> Please, my patch (based from Jon's) is huge because it's removing > 200 lines
>> and it's moving softcursor.c from drivers/video to drivers/video/console where
>> it belongs. But it is _not_ intrusive.  In fact, the majority of the patch consists
>> of a single logical change which removes the line below from each driver
>>
>> - .fb_cursor = softcursor;
>>
>> and removes the line below from drivers/video/Kconfig.
>>
>> - select FB_SOFT_CURSOR
>>
>> Our proposal aims to make complex code simpler and big code smaller. Your patch,
>> on the other hand, although small, introduces another flag  FB_HWCURSOR_SOFTCURSOR,
>> which is redundant, and adds yet another level of complexity to an already too
>> complex code.
>>     
>
> I ask you the same question as Jon. Currently several fbdev drivers have a 
> optional flag to turn on and off the hardware cursor. Should this functionality
> be removed and force all fbdev drivers to always have the hardware cursor 
> avaiable. Thus leaving fbcon to control when to use the hardware cursor 
> or the software cursor. It all comes down to who controls when the 
> hardware driver will use the hardware cursor. As noted several drivers 
> have such a flag. You would have to strip that flag out of the drivers as 
> well.
>
> .
>
>   
BTW, the only drivers that have working hardware cursor support are 
rivafb, intelfb,
nvidiafb, and i810fb. Except for intelfb, I wrote all of them, if not 
the driver, the
hardware cursor support. And intelfb's cursor support was based on i810fb.

To answer your question, drivers can do either of two things:

1. If they don't want to use the hardware cursor, set 
info->fbops->fb_cursor = NULL
in fb_set_par();

2. Or, they can keep info->fbops->fb_cursor() as is, but the function 
will return
-ENODEV instead.

The end result is if info->fbops->fb_cursor == NULL, fbcon will call 
soft_cursor.
Or if the return value of info->fb_ops->fb_cursor is not equal to zero, 
fbcon will
call soft_cursor.

Either way, control remains with the driver, not fbcon.

Tony



-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf