> 
> 
> very interesting ... drivers are very low level (kernel 
> level). I haven't done any windows driver programming but I 
> have briefly messed about with kernel module programming. I 
> wouldn't have thought it was even a vague possibility ... 
> 
I haven't done any kernel module hacking, but I've written a few
drivers.  For each type of devices (audio, etc) there is usually an
interface that each OS would provide, which the driver hooks to.  Then
the driver can do what it likes.  Assuming that the driver only accesses
memory (ie, a memory mapped device) and responds to interrupts, then you
really only need to write a translation layer to turn the linux driver
hooks into Windows driver hooks, and maybe a few helper functions to map
memory, etc.

Anyhow, on the non-technical side - I personally don't like this idea.
If there is a product (that you pay for) that works seamlessly with all
windows drivers, then there is no incentive for HW manufacturers to
provide any (there is little incentive now, as the market share is low &
OSS people tend to hack up free drivers anyway).
But, honestly it isn't that difficult to have drivers run on many Oses,
it's really just a matter of different function names most of the time.
nVidia do it, why can't everyone else?

Brad

Reply via email to