Subject: Re: No consol cursor
To: Michael Lorenz <macallan@netbsd.org>
From: Heitzso <heitzso@growthmodels.com>
List: port-sparc
Date: 03/24/2006 09:11:45
If you get mods into current and want folks to test post
that to the list and I'll go through the update/compile sequence
to test.

>Hello,
>
>  
>
>>>>Hello there I have installed NetBSD 3.0 sparc on a SparcStation 20
>>>>and have updated to stable, built and installed a kernel.
>>>>However I have the  following problem even with the GENERIC
>>>>kernel, there is no consol cursor  so it's very difficult to edit
>>>>files at the console. The graphics  card is a TGX, has any one
>>>>come across this problem?
>>>>        
>>>>
>>>Hmm, I thought I fixed that a while ago.
>>>
>>>      
>>>
>>>>I have also done a  CVS update of the source but have yet to build
>>>>and install user land. Would this help?
>>>>        
>>>>
>>>No, this is a problem in the cgsix kernel driver. Userland has
>>>nothing to do with it.
>>>
>>>So, did you set the RASTECONSOLE option in your kernel?
>>>If you did please try the following:
>>>find this piece of code in src/sys/dev/sun/cgsix.c / cg6attach():
>>>			cg6_console_screen.sc = sc;
>>>			sc->sc_fb.fb_rinfo.ri_hw = &cg6_console_screen;
>>>			sc->sc_fb.fb_rinfo.ri_ops.copyrows = cg6_ras_copyrows;
>>>			sc->sc_fb.fb_rinfo.ri_ops.copycols = cg6_ras_copycols;
>>>			sc->sc_fb.fb_rinfo.ri_ops.erasecols = cg6_ras_erasecols;
>>>			sc->sc_fb.fb_rinfo.ri_ops.eraserows = cg6_ras_eraserows;
>>>			sc->sc_fb.fb_rinfo.ri_do_cursor = cg6_ras_do_cursor;
>>>			cg6_ras_init(sc);
>>>... comment out the cg6_ras_do_cursor; line and rebuild your kernel.
>>>This will disable using the blitter to draw the cursor ( the console
>>>driver will use inverted character attributes instead ). Please tell
>>>me if this fixes the problem.
>>>It shouldn't happen in -current but checking can't hurt.
>>>
>>>have fun
>>>Michael
>>>
>>>      
>>>
>>Michael,
>>         I had to remove the cg6_ras_do_cursor function and prototype
>>         to 
>>get this to compile but you'll be pleased to know after a successful 
>>compilation the problem is fixed!
>>    
>>
>
>I wonder what causes this - the exact same code is used to draw the
>cursor when using wscons which works fine. The only difference that
>matters here is the colour map. Since it apparently works on some
>machines ( like my U1 ) I suspect different firmware versions set up
>different colour maps.
>
>have fun
>Michael
>  
>