Thank you for your answer. Ok, even if the hardware vendor decides to open the driver : Will it be included ? In every kernel ? In every distribution ? For only a few (about several thousand) users ? Would is make a difference ? Why is there no really *fix* api *and* abi for all kernels 2.6.x ? For example, you write a kernel driver for e.g. Solaris 9; it will be supported all the livetime of Solaris. I think Linux really lacks such a feature. The main problem is to provide an easy way to distribute such drivers. Thanks for answering. Am Sonntag, 21. November 2004 12:02 schrieb Arjan van de Ven: > On Sun, 2004-11-21 at 10:56 +0100, Stefan Sonnenberg-Carstens wrote: > > The problem I have now, that the driver needs to be > > rebuild with every new kernel version, which is released. > > Is there a way to avoid that ? > > I have developed kernel drivers for Linux for 2.2.x kernels, > > but things seem to have changed .... > > Note that you really should talk to your lawyers about this; it's not > entirely clear if such non-GPL kernel components are legal or not > (personally I don't think they are). > For example, see the statements from Linus Torvalds on this at > http://people.redhat.com/arjanv/COPYING.modules . > > I also wonder if the *driver* or the hardware has the "valuable IP" that > needs to be protected, most of the time it's the hardware after all. > > > Even if your lawyers think you're in the clear, it's a major pain. Well > that's actually an encouragement to make driver vendors consider if the > IP they try to protect really is worth the hassle... because a hassle > it is. ABI and API change with every kernel release. And lots of people > run *lots* of different kernels, including kernels with really weird > patchkits.
