This is the mail archive of the
cygwin-xfree@sourceware.cygnus.com
mailing list for the Cygwin project.
FW: drivers
- To: "Cygwin-Xfree@Sourceware.Cygnus.Com" <cygwin-xfree@sourceware.cygnus.com>
- Subject: FW: drivers
- From: "Suhaib Siddiqi" <Ssiddiqi@InspirePharm.Com>
- Date: Thu, 19 Aug 1999 13:14:52 -0400
following message just came from a friend who has started now working on
devices drivers for cygwin-xfree. it might of interest to others, so they
know what is being done.
Suhaib
> -----Original Message-----
> From: Anton van Straaten [mailto:anton@appsolutions.com]
> Sent: Thursday, August 19, 1999 12:47 PM
> To: Suhaib Siddiqi
> Subject: RE: drivers
>
>
> Suhaib,
>
> The OS-specific code was a big help - I took a look at it this morning.
>
> Question: I assume it's OK if I rewrite the internals of functions like
> xf86ReadBIOS (in cygwin_mmap.c), to use NT API calls (e.g. calls to
> DeviceIOControl)? Since this code is going to be very platform-specific
> anyway, I'm thinking there's not much point in trying to use Unix-style
> functions like device_open to access NT drivers.
>
> It looks to me as though the direct memory access will be fairly
> straightforward, using the DDK sample driver as you suggested (with
> modifications if necessary). I haven't looked at the port access
> functions
> yet, but I'm hoping the situation is similar.
>
> I looked at the idea of using DirectX. For the memory access, I
> don't think
> it will make things any easier, in fact it will probably be more
> complicated. DirectX may help more with port access, and it may eliminate
> the need for another custom driver, so I'll keep that in mind.
>
> How much work is going to be involved when I get to the point of
> wanting to
> test this? Will I need to build the entire system from the full source
> tree? Or am I just going to send the code to you for testing?
>
> > Your help is highly appreciated.
>
> You're welcome. I couldn't have gotten OpenDX running without
> the work you
> did with it and Cygwin, so I'm happy if I can return the favor. Also, I
> quite like the idea of a free X Server for NT (call me perverse!)
>
> Regards,
> Anton
>
>
> > -----Original Message-----
> > From: Suhaib Siddiqi [mailto:Ssiddiqi@InspirePharm.Com]
> > Sent: Wednesday, August 18, 1999 7:29 AM
> > To: Anton van Straaten
> > Subject: RE: drivers
> >
> >
> >
> >
> > Hi Anton,
> > Thanks for your help. It is very much appreciated.
> >
> > > I think that's a good idea, thanks. If it's not a lot of
> > > trouble, go ahead
> > > and send this to me.
> >
> > The different OS specific code is attached. I Tar gzipped it
> because some
> > of the subdirectories contains symlinks to the code in
> os-support/shared.
> > WinZip seems to mess-up the symbolic links.
> >
> > >
> > > I took a brief look at the OS/2 driver code, as well as the sample NT
> > > drivers. The sample NT drivers look fairly familiar to me,
> > > because a lot of
> > > the DDK calls are roughly equivalent to the higher-level Win32 API.
> >
> > I think rewriting OS/2 drivers will be a lot of work, as you mentioned.
> > There is also no guarantee that ported OS/2 driver will work on
> WinNT. As
> > far as I know the OS/2 drivers from Holger Veit can violate NT hardware
> > protection mode. Therefore I thought using NT DDK samples, or something
> > like that would be a better idea. All what is needed; is
> mapping physical
> > memory and IOports to user area. I think it can be either done
> > using NT DDK
> > or
> > DirectX. According to Microsoft Knowledge Base Direct X allows direct
> > accessing to hardware-abstraction layer, reading and mapping BIOS and
> > physical memory of Video Drivers. I am quoting this from MSDN
> > web pages, I
> > have zero knowledge on how to write code for devices using DDK
> or DirectX.
> >
> > > But I
> > > think there could be quite a bit of work in getting it all to
> > > work together
> > > with xfree86. Looking at the xfree86 calling code is
> > definitely the next
> > > step, and I could well imagine that some shortcuts might be
> > > possible at that
> > > level.
> >
> > I assume too, some shortcuts are possible. Therefore I asked
> your help.
> > In the os-support.tar.gz please ignore the subdirectory
> "cygwin." It has
> > some nonsense code in it. I added it intentionally when
> patching XFree86
> > source tree so I can compile rest of the XFree86 source tree.
> >
> > >
> > > If I want to download the full XFree86 source tree, where should
> > > I get that
> > > from?
> >
> > The full Xfree86 source tree is 100 MB. I think what is needed
> > is different
> > OS-support examples. However, if you need the full source tree,
> > I can give
> > you access. At the moment the source tree is at XFree86
> Developers site,
> > which is not accessible to non-members.
> >
> > Please make sure you add your copyright notice at the top of
> each sources
> > file, you write. I suggest this because the your code will
> eventually get
> > merged with XFree86 version 4.x.
> >
> > >
> > > I'm still busy catching up on some other work, but I'll be
> able to spend
> > > more time on this in the next few days.
> > >
> >
> > Your help is highly appreciated.
> >
> > Regards
> > Suhaib
> >
> >
> > > Anton
> > >
> >
>