Re: plan to work on user-level device drivers

[Thomas Schwinge]

Hello!

On Fri, Jul 10, 2009 at 07:47:33PM +0200, olafBuddenhagen@gmx.net wrote:
> On Wed, Jul 08, 2009 at 07:21:04PM +0200, Thomas Schwinge wrote:
> > What also could be done, and what I had in mind for a long time
> > already, is a simple driver for a SoundBlaster ISA card (again, no PCI
> > for now -- I don't know if PCI device registration actually is
> > difficult, but let's simply avoid that for now).
> 
> I'm not entirely sure, but my impression thus far is that PCI stuff is
> actually easier to handle than legacy devices with all their quirks...

That may very well be true, but note that I never talked about handling
any quirks at all, or being feature (or whatever) complete.  Only a
prototype thing is what I'm talking about.  (Aren't you usually the
person to encourage such incremental progression?)  And for doing that I
still think a simple SoundBlaster ISA user-space driver would be (a) an
easy device driver to write and (b) one that includes all of the needed
infrastructure: I/O port access, IRQs, DMA.

If I remember correctly, the functionality of this device driver would be
as simple as follows: set up the sample rate etc. by writing to some I/O
ports; create a DMA memory region, register that one with the soundcard,
request IRQs to be sent each time half of the DMA memory had been
handled; handle these IRQs in the device driver by filling one half of
the DMA memory with sample data (while the other half is being output).

But of course, it is up to the person doing the work to decide what to
work on.


Regards,
 Thomas

signature.asc
Description: Digital signature