strange problems after upgrading to kernel 2.6.10rc1
Thu Nov 4 09:20:55 PST 2004
On Wed, Nov 03, 2004 at 07:57:03PM -0800, Jouni Malinen wrote:
> Most, if not all, changes in wireless extensions were backwards
> compatible at least in the source level. From my viewpoint, it's quite
> a bit different to start removing function arguments..
> > Some kernel developpers seem to have a position that backward
> > compatibility should be avoided at all cost and breaking it is good
> > fun. One of the "benefit" is that it punishes driver living outside
> > the kernel, and therefore encourage people to maintain their driver in
> > the kernel. So, I guess we can expect more of it.
> Well.. I can understand the goal of having a clean interface to internal
> kernel functionality. Yes, it will mean things get broken every now and
> then, but then again, one can go the route of taking a distro kernel and
> hope for the best that the distro in question includes all the needed
I'm currently playing with a embedded device that has a
heavily patched 2.4.19 kernel. No way I'm going to forward port thir
patch to 2.6.X. I did appreciate that HostAP compiled and worked with
minimal fuss (once I upgraded the WE).
So, there are people who appreciate your effort of keeping
your driver compatible with multiple kernel version, even if some
kernel developper thing this is bad.
> > Wich brin nicely to the topic of inclusion of HostAP in the
> > kernel...
> I didn't get much (well, none) comments on this when I submitted the
> driver for wireless-2.6 tree. I could try to update wireless-2.6 version
> to match with my current work version and then ask whether it could be
> merged into linux-2.6 tree..
The fact is that nobody is using the wireless-2.6 tree. I
don't, and I don't know anybody who does. It seems to me that Jeff's
effort to coerce wireless effort toward his goals did only get minimal
traction. Maybe it's time to get HostAP in -mm.
Only my personal opinion, of course.
> Jouni Malinen
More information about the Hostap