[uae] Re: UAE, libSDL, and how to hunt down bugs ...

  • From: Richard Drummond <evilrich@xxxxxxxxxxxxxx>
  • To: uae@xxxxxxxxxxxxx
  • Date: Mon, 3 Jan 2005 22:00:59 -0500

Hi Byron

On Monday 03 January 2005 09:28 pm, Byron Q. Desnoyers Winmill wrote:
> If I told you, you would say, "oh, one of those people." ;-)  Crux/PPC.

LOL! Never tried Crux myself. I'm a die-hard Debian man. ;-)

> Anyhow, I found a simple program which managed to compile.  The
> results are: it scrolled up (as UAE does) and the palette didn't
> change back after returning from another virtual console.

I think this is a general problem with the framebuffer device under 2.6.x. I 
think SDL will need to be modified to do the palette restore itself upon 
switching back to its virtual console. I'll have a look at the SDL fbdev 
code.

> Based upon it's behaviour, I'm guessing that the origin of the
> screen moves down as the text scrolls up.  This is eventually
> nudging anything off screen.

Yes. Basically the linux console fills up the framebuffer with a virtual 
screen that's as wide as the physical screen and as high as your video memory 
will allow. Scrolling the console is performed by panning up and down this 
virtual screen. Perhaps the offb doesn't support panning, and that would 
explain the different behaviour (I don't remember off-hand - it's been a 
while since I've mucked about with the offb driver).

> This scrolling was never a problem  
> with MOL because MOL opens it's own virtual console.  Perhaps I
> can redirect stdout from your program to /dev/null to avoid that
> problem.  (Unlike offb, which only scrolls the text up and leaves
> the graphics intact, rivafb scrolls everything.)  I'm also guessing
> that it is some sort of problem with the kernel.  The palette
> problem is in the kernel or SDL.  I'm not sure where, but it clearly
> isn't your concern.

It sounds like a problem with the rivafb driver, but I could be utterly wrong. 

When I get some free time, I'll do some experimentation with SDL/fbdev on a 
2.6 kernel here, and see what I can find out. I don't have an Nvidia card 
myself, but I do have variety of other gfx chipsets that I can compare 
behaviour with. 

> Thank-you for the pointers,

You're welcome.

Cheers,
Rich

Other related posts:

  1. Home
  2. uae
  3. Archive
  4. 01-2005