Then why don't we ask the WINE or MANE people to join the group for
questions like these. I know atleast the WINE people would have a
really good understanding of what we might be needing to achieve and
to just point us in the right deirect as you suggest Sam. They could
also provide their opinions on our ideas and if they're possible. That
alone would help better than hunting down API's and allow people to
get down to coding. So perhaps we approach them and ask them as
Only an idea,
On Fri, 19 Nov 2004 22:07:51 -0500, Sam Liddicott <sam@...> wrote:
> Henry Nestler wrote:
> > Sam Liddicott wrote:
> > > Henry Nestler wrote:
> > >
> > >> Can we define some of the pseudo physical ram as shared memory area?
> > >>
> > >> Inside the colinux kernel we can remap this space to "virtual video
> > >> buffer". After this the xserver can remap "virtual video buffer" via
> > >> users mmap as frame buffer.
> > >> But question is, how can map this pages on windows side.
> > >
> > > There were two issues, right?
> > >
> > > One was how to make use of agp style mappings to easily tell
> > > which parts of the frame buffer had been updated.
> > > The other was whats good way to do it.
> > >
> > > What if we extend the trick re-mapping that will be done for
> > > the xserver by the linux kernel; how many times can we re-map
> > > a directx (directdraw?) surface?
> > Many. Every page can map to some other virtual memory. Winnt has
> > functions for this, but must be cal inside linux driver (Ring 0).
> > > If the coLinux daemon sets up the directx surface can't this
> > > be mapped to the xserver as a contiguous frame buffer?
> > At end yes.
> > Please can you give a small sample to open a directx surface and write
> > somthing into this surface by using frame buffer?
> Alas, I don't know how; I have never done DirectX programming.
> > After this, we can implement colinux driver and fill from linux side.
> > Than I can replace my SDL sample with directx.
> > I don't know enought about directx. If you can remap a surface under
> > winnt, we can also remap this into linux xserver. Not directly, but
> > step by step.
> > Step 1: Mapping from winnt user into winnt driver, to hold this pages.
> > Step 2: Mapping this pages inside linux kernel, holds page (nocache).
> > Step 3: mmap pages in coFB to xserver
> > After this scenarieo the xserver should directly write into surface.
> > If surface need a "refresh area event", we detect write access to any
> > page via dirty flag inside linux kernel. Numer of dirty pages would be
> > signaled from linux to winnt via message with "list of dirty pages".
> > Inside winnt (driver or daemon) we make page also as dirty or make a
> > simple write access. Directx should than update the area y self.
> > This is my private theory of this all. Don't know, if that all right?
> I think it is possible, its just where to take your questions. There are
> people who can give answers and name APIs within 5 minutes. I guess this
> would include the MAME people, the WINE people for starters.
> > --
> > Henry
> This SF.Net email is sponsored by: InterSystems CACHE
> FREE OODBMS DOWNLOAD - A multidimensional database that combines
> robust object and relational technologies, making it a perfect match
> for Java, C++,COM, XML, ODBC and JDBC. http://www.intersystems.com/match8
> coLinux-devel mailing list