Re: Regression: radeonfb: No synchronisation on CRT withlinux-2.6.13-rc5

[Benjamin Herrenschmidt]

> I'm having a similar issue with my shiny new 17" Powerbook G4.  The
> radeon chip works fine with framebuffer in 2.6.12.4 _with_ PREEMPT,
> but not in 2.6.13-rc5 _with_ PREEMPT (configs are virtually identical).
> I'll try your idea this afternoon when I get the chance.

Note that PREEMPT is known to cause problems on ppc32 ... I'm not sure
what's up yet.  (Random SIGILLs/SEGVs in userland typically)

> I wonder if perhaps some code in radeonfb is used under the BKL, which
> is now preemptable (Or maybe an ordinary spinlock changed or went
> away?), because I also set PREEMPT_BKL.

radeonfb should only rely on the console semaphore....

>   I've got an LCD, and on mine
> it looks like every third pixel-line gets shifted about 32-64 pixels to
> the left, and they move with display refresh.  My guess is that
> something is interrupting radeonfb during a critical time in display
> syncing and forcing the video card to wait too far into the next line
> before sending pixels.

radeonfb is mostly inactive after it has setup the framebuffer and
unless you actually draw something, in which case, accel code is called.

_However_ there is an unrelated problem with some panels, including some
of the 17": The panel doesn't always "sync" properly. This seem to be
related to some subtle timing issue in the LVDS code but I don't know
exactly what yet. You can usually get it back by repeately turning the
backlight all the way down (which shuts the panel off) and back up until
it "catches".

> One other data point, I've seen something like this, except not nearly
> as bad, is stock debian 2.6.8 vs. stock debian 2.6.11 on powerpc.  The
> former exhibits some similar (but not nearly as bad) symptoms.  (Same
> Powerbook), whereas 2.6.11 doesn't.  In that case, neither has PREEMPT.
> I'll run more tests this afternoon/evening, to try to track it down.
> 
> Cheers,
> Kyle Moffett
> 
> --
> There are two ways of constructing a software design. One way is to  
> make it so
> simple that there are obviously no deficiencies. And the other way is  
> to make
> it so complicated that there are no obvious deficiencies.  The first  
> method is
> far more difficult.
>    -- C.A.R. Hoare
> 

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/