possible cg6 fix for pixels repeating at display edge or fading

possible cg6 fix for pixels repeating at display edge or fading

[Bob Breuer]

Has anyone else using a cg6 seen the color banding or fading and pixels
on the right side being repeated off to the edges of the monitor?

I think I've found the cause of the problem.  Contrary to the comment,
changing the bt->control register affects something other than the
cursor.  Also, it looks like cg6_bt_init() already sets up the DAC and
this chunk of the code may be redundant.

The following patch seems to work for me on a TGX+.  If you are seeing
problems, can you give this a try and report if it works or not?

Bob


--- linux-2.6.14/drivers/video/cg6.c.orig	2005-06-22 09:04:25.000000000 -0500
+++ linux-2.6.14/drivers/video/cg6.c	2005-10-29 15:25:19.000000000 -0500
@@ -658,7 +658,7 @@ static void cg6_chip_init(struct fb_info
 	/* Disable cursor in Brooktree DAC. */
 	sbus_writel(0x06 << 24, &par->bt->addr);
 	tmp = sbus_readl(&par->bt->control);
-	tmp &= ~(0x03 << 24);
+	tmp |= (0x03 << 24);
 	sbus_writel(tmp, &par->bt->control);
 }

Re: possible cg6 fix for pixels repeating at display edge or fading

[Weeve]

On Sat, 29 Oct 2005 16:49:12 -0500
Bob Breuer <breuerr@mc.net> wrote:

> Has anyone else using a cg6 seen the color banding or fading and
> pixels on the right side being repeated off to the edges of the
> monitor?

I'm seeing it as well with a cg6 (501-1672( in an Ultra 1.

> I think I've found the cause of the problem.  Contrary to the comment,
> changing the bt->control register affects something other than the
> cursor.  Also, it looks like cg6_bt_init() already sets up the DAC and
> this chunk of the code may be redundant.
> 
> The following patch seems to work for me on a TGX+.  If you are seeing
> problems, can you give this a try and report if it works or not?

Your patch works for me.  The odd coloration of text and pixels on the
edges of the screen (was seeing on both left and right here with
colorized init scripts) have gone away.  Display looks as I'd expect.

Cheers,
-- 
Jason Wever
Gentoo/Sparc Team Co-Lead

Re: possible cg6 fix for pixels repeating at display edge or fading

[David S. Miller]

From: Bob Breuer <breuerr@mc.net>
Date: Sat, 29 Oct 2005 16:49:12 -0500

> Has anyone else using a cg6 seen the color banding or fading and pixels
> on the right side being repeated off to the edges of the monitor?
> 
> I think I've found the cause of the problem.  Contrary to the comment,
> changing the bt->control register affects something other than the
> cursor.  Also, it looks like cg6_bt_init() already sets up the DAC and
> this chunk of the code may be redundant.
> 
> The following patch seems to work for me on a TGX+.  If you are seeing
> problems, can you give this a try and report if it works or not?

That register sets the overlay plane, we should just leave it
alone since as you correctly noted other code inits that register
already.

I'll delete those bogus lines, thanks Bob.

Re: possible cg6 fix for pixels repeating at display edge or fading

[David S. Miller]

From: Bob Breuer <breuerr@mc.net>
Date: Sat, 29 Oct 2005 16:49:12 -0500

> @@ -658,7 +658,7 @@ static void cg6_chip_init(struct fb_info
>  	/* Disable cursor in Brooktree DAC. */
>  	sbus_writel(0x06 << 24, &par->bt->addr);
>  	tmp = sbus_readl(&par->bt->control);
> -	tmp &= ~(0x03 << 24);
> +	tmp |= (0x03 << 24);
>  	sbus_writel(tmp, &par->bt->control);
>  }

BTW, I noticed that the 2.4.x cgsix.c driver does set the bits here
instead of clear them.  I have no idea why when I moved this code over
while writing the 2.6.x driver I changed it to clear the bits :-/