[Bug 502] New: Broken cursor when using neofb

[Bug 502] New: Broken cursor when using neofb

[Martin J. Bligh]

http://bugme.osdl.org/show_bug.cgi?id=502

           Summary: Broken cursor when using neofb
    Kernel Version: 2.5.66
            Status: NEW
          Severity: normal
             Owner: jsimmons@infradead.org
         Submitter: jochen@jochen.org


Distribution: Debian sarge
Hardware Environment: IBM Thinkpad 600

Problem Description:

I use neofb, the boot messages are:
Mar 25 20:04:58 gswi1164 kernel: neofb: mapped io at c680d000
Mar 25 20:04:58 gswi1164 kernel: Autodetected internal display
Mar 25 20:04:58 gswi1164 kernel: Panel is a 1024x768 color TFT display
Mar 25 20:04:58 gswi1164 kernel: neofb: mapped framebuffer at c6a0e000
Mar 25 20:04:58 gswi1164 kernel: neofb v0.4.1: 2048kB VRAM, using 1024x768, 48.361kHz, 60Hz
Mar 25 20:04:58 gswi1164 kernel: fb0: MagicGraph 128XD frame buffer device
Mar 25 20:04:58 gswi1164 kernel: Console: switching to colour frame buffer device 128x48

On a vc the line cursor looks like
****** ** ********
instead of
****************** (the normally continous line is broken).
Emacs uses a block cursor that is broken similar, the block
is broken by two vertical bars.

Re: [Bug 502] New: Broken cursor when using neofb

[Petr Vandrovec]

On 25 Mar 03 at 11:57, Martin J. Bligh wrote:
> Mar 25 20:04:58 gswi1164 kernel: Console: switching to colour frame buffer device 128x48
> 
> On a vc the line cursor looks like
> ****** ** ********
> instead of
> ****************** (the normally continous line is broken).
> Emacs uses a block cursor that is broken similar, the block
> is broken by two vertical bars.

Well, then it's not my bug then. Go to the drivers/video/console/fbcon.c,
and do (pseudopatch):

+ memset(data, 0xff, size);
  if (cursor->set & FB_CUR_SETSIZE) {
-    memset(data, 0xff, size);
     cursor->set |= FB_CUR_SETSHAPE;
  }
 
Correct cursor should appear after that. It is questionable whether this
layer should do this SETSIZE -> SETSHAPE propagation, as soft_cursor()
does same. I recommend leaving this here, and removing it from
soft_cursor() code.
                                                            Petr
                                                            

Re: [Bug 502] New: Broken cursor when using neofb

[Alex Goddard]

On Tue, 25 Mar 2003, Martin J. Bligh wrote:

> http://bugme.osdl.org/show_bug.cgi?id=502
> 
>            Summary: Broken cursor when using neofb
>     Kernel Version: 2.5.66
>             Status: NEW
>           Severity: normal
>              Owner: jsimmons@infradead.org
>          Submitter: jochen@jochen.org
> 
> 
> Distribution: Debian sarge
> Hardware Environment: IBM Thinkpad 600
> 
> Problem Description:
> 
> I use neofb, the boot messages are:
> Mar 25 20:04:58 gswi1164 kernel: neofb: mapped io at c680d000
> Mar 25 20:04:58 gswi1164 kernel: Autodetected internal display
> Mar 25 20:04:58 gswi1164 kernel: Panel is a 1024x768 color TFT display
> Mar 25 20:04:58 gswi1164 kernel: neofb: mapped framebuffer at c6a0e000
> Mar 25 20:04:58 gswi1164 kernel: neofb v0.4.1: 2048kB VRAM, using 1024x768, 48.361kHz, 60Hz
> Mar 25 20:04:58 gswi1164 kernel: fb0: MagicGraph 128XD frame buffer device
> Mar 25 20:04:58 gswi1164 kernel: Console: switching to colour frame buffer device 128x48
> 
> On a vc the line cursor looks like
> ****** ** ********
> instead of
> ****************** (the normally continous line is broken).
> Emacs uses a block cursor that is broken similar, the block
> is broken by two vertical bars.

The VESA fbcon does exactly this as well.

I can grab my boot messages if they would help.

-- 
Alex Goddard
agoddard@purdue.edu