Current CVS: static build broken?

Daniel Stone daniel at fooishbar.org
Sat Jan 15 14:47:28 PST 2005


On Sat, Jan 15, 2005 at 05:45:07PM -0500, Adam Jackson wrote:
> On Saturday 15 January 2005 16:54, Roland Mainz wrote:
> > Keith Packard wrote:
> > > Around 19 o'clock on Jan 14, Roland Mainz wrote:
> > > > What was the reason for writing a "xf1bpp" driver instead of using
> > > > "mfb" ?
> > >
> > > I believe xf1bpp matches the old PC MGA bitmap format, which is something
> > > like big-endian bits in little endian bytes.
> >
> > Is there anything else in the Xorg tree which supports this kind of
> > cards or will the removal of "xf1bpp" effectively mean we're dropping
> > support for these cards ?
> 
> xf1bpp is used only by the XFree86 DDX.  I did say earlier in the thread that 
> xf1bpp users would be converted to use mfb, so we're not dropping support for 
> anything.  However, I would wager that most of the low color depth support in 
> the drivers - possibly excluding vga and vesa - has rotted.  1 and 4 bpp had 
> been broken in trident since about xf4.2 with no one complaining.  This is 
> not much of a loss; every driver that uses xf1bpp/mfb also supports depths 
> >=8bpp, excluding sunbw2.  Pretty sure most users spend their time at 16bpp 
> or above.
> 
> Also I'm not convinced that Keith's assessment is correct.  For one, sunbw2 
> uses xf1bpp, I doubt it's an MGA card.  Secondly, I don't see any #define 
> hints in the xf1bpp Imakefile indicating anything about endian weirdness.  
> Experimentation would tell me for certain.
> 
> I suspect the motivation was to have a loadable monochrome framebuffer core 
> without needing to touch the "upstream" mfb source.

Shouldn't that be in extras/mfb/, then?

Daniel, with a brutal sense of irony



More information about the xorg mailing list