Re: [MiNT] new drivers?

["Guido Flohr" <gufl0000@stud.uni-sb.de>]

On Mon, May 10, 1999 at 10:32:36PM +0000, johnnie ligotage wrote:
> On Wed, 5 May 1999 t150315@students.cc.tut.fi wrote:
> > First sound driver would be a version of MiNT audiodevice
> > 'just' supporting the same ioctl()s as linux audiodev.
> 
> I'm sure I can do this! Which one's do I need to support?
> (Guido: if I do this, can you incorporate it into the next MiNTLib?)

What do you mean, the ioctl()s?

I can do that but I actually don't like it the way it is currently done.
Currently you have a lot of doubly defined ioctl()s in <sys/ioctl.h> and
<mint/dcntl.h>.  I would actually prefer to have <sys/ioctl.h> #include
<mint/dcntl.h> which in turn should #include <mint/audio.h>,
<mint/mouse.h>, <mint/random.h>, ... to get the device specific ioctl()s.  
But that would require to reorganize the kernel include files a little
bit.

The idea behind it is to keep some include directory in the kernel sources
which is completely integrated into /usr/include.  The <sys/*.h> headers
from the MiNTLib would then simply include these files instead of
providing own defines.  But, again, this would require some changes in the
kernel sources, especially a proper and consequent use of the __KERNEL__
preprocessor macro.  Yet, this would be helpful because not only the
design of <sys/ioctl.h> is likely to call for trouble, there are also
other files that may produce conflicts.

The next question is if the MiNTLib should provide this directory itself
(and update it regularly) or if I should simply expect (like Linux does
it) that I will find these files in /usr/src/mint/include/mint and only
link /usr/include/mint to that directory.  Uhm, that would mean that MagiC
users would have to install a MiNT source tree... ;-)

Ciao

Guido
-- 
http://stud.uni-sb.de/~gufl0000
mailto:gufl0000@stud.uni-sb.de