[PATCH] i810: Add workaround for bad BIOS mode on Fujitsu laptops

[Tony Lindgren]

* Keith Packard <keithp at keithp.com> [070129 10:16]:
> On Sun, 2007-01-28 at 23:32 -0800, Tony Lindgren wrote:
> > * Keith Packard <keithp at keithp.com> [070128 23:22]:
> > > On Sun, 2007-01-28 at 22:48 -0800, Tony Lindgren wrote:
> > > > Hi all,
> > > > 
> > > > Just hacking a bit to get the i810 modesetting driver to work on my
> > > > laptop :) Here are few patches.
> > > 
> > > We've started adding a list of 'quirks' to the EDID parsing stuff;
> > > perhaps we need a similar list of quirks for the BIOS data.
> > 
> > OK. Originally when trying to fix this I was thinking that the BIOS
> > mode should be verified against all the probed modes, and if the BIOS
> > mode does not match any VESA modes, then ignore the BIOS mode. But
> > then I noticed that the probed modes use BIOS mode for some
> > checks..
> 
> BIOS modes are often not standard VESA modes, and often use non-standard
> timing as well. I'm surprised that your BIOS mode is so broken; we've
> never seen that before...

Yeah I guess trying to use VESA modes would be a problem especially on
widescreen laptops. Are the some other checks that could be done on the
BIOS modes before accepting them?
 
> > Maybe there should be some option to ignore BIOS mode? And then have
> > some quirks for known bad BIOS modes.
> 
> We need a mode to use in place of the broken BIOS mode; perhaps the
> quirk should just be that we take the panel size from the BIOS and
> generate a mode using the GTF.

That's kind of what I was thinking too as the size is correct.
 
> > Yeah, the LVDS is streched out estimated 2x horizontally, vertical is
> > OK without the patch. Are you suggesting that the bug would be in the
> > BIOS mode parsing?
> 
> I don't know where the bug lies. It's possible we're mis-interpreting
> the BIOS data, it's also possible we should ignore the timing data and
> use GTF if it looks screwy

Sounds like it should work, do you have some patch available to try?
  
> > > Can the BIOS set the mode correctly?
> > 
> > I don't think so unless there's a bug in parsing the BIOS mode.
> > I guess I could try to dump the BIOS, or let me know what data
> > you need.
> 
> The question is whether the BIOS can set the native panel mode
> correctly. You can test this by using the master branch; if that works,
> then the BIOS is doing the right thing and we're doing the wrong thing
> somehow. If the BIOS can get it right, then we should be able to get the
> modesetting code working.

Last time I tried the master branch it had the same problem. Earlier I
got it to work with 915resolution. I haven't tried that for a while now,
so I'll verify that when I have a chance.

Regards,

Tony