Re: [Unichrome-users] (no subject)

[Luc V. <lu...@mp...>]

On 2004.03.08 17:08, Nicolas CASSART wrote:
> This is the log with the
> via_drv.o-unichrome_r10-debian_sarge_xfree86-4.3.0-2
> It worked when i had the experimental debian xfree-4.3.0 builds
> installed
> on my sarge (because sid's were 4.2.1) .
> 
> My splash screen is a picture that is displayed on boot in  
> framebuffer
> 
> mode (bootsplash kernel patch and splashutils)
> I tried without splash and the result is the same .
> 
> The via_drv.o i use is from the today's unichrome-cvs built against
> the
> original xfree-4.3.0 tarball contained in debian-src package
>

> (II) LoadModule: "via"
> (II) Loading /usr/X11R6/lib/modules/drivers/via_drv.o
> (EE) LoadModule: Module via does not have a viaModuleData data  
> object.
> (II) UnloadModule: "via"
> (II) Unloading /usr/X11R6/lib/modules/drivers/via_drv.o

Aha. It seems that this is the vga register restoration problem!

This has been reported on netBSD, where the console implementation  
doesnt do its own restoration too (on top of what xfree86 already tries  
to do). Normal linux console does this itself np, effectively masking  
the problem with the via driver.

Now, i have never used directfb, never even been near it. But someone  
on undernet was willing to answer some questions about it. The splash  
screen apparently runs on top of directFB, directFB uses the hardware  
directly. And then you run xfree86 which, in this case, also uses the  
hardware directly. Apparently this isnt much of a problem, with a  
recent enough X and with a suited driver. (this amazes me though, as i  
would expect both to contend for the same resources)

Anyway, this might be where the VT restoration issues comes back at us  
again. This is imho great news, as this means that we can test this  
without installing netBSD :)

About the driver issue, hrm, apparently nothing has changed (in this  
respect) between -2 and -5.

Luc Verhaegen.