On Sun, Nov 30, 2003 at 12:39:47AM +0200, Ky?sti M?lkki wrote: > > In 2.6.0-test10 i2c-core maintains its own list of adapters and drivers. > I think these are unneccessary as the generic driver-model maintains > those too, and it can also detach/attach clients when necessary. > > The resulting i2c-core does not yet compile nor work, and I have not yet > reflected these changes to header file. But no changes in adapter or > client code is required at this time. Hm, in looking over this, I don't think it will work at all yet :) Want to post something that at least builds? > i2c-core driver-model > > Attempt to fit i2c-core better with driver-model. Reason I didn't fully convert i2c to the full driver model was I was trying to incrementally update the i2c core and ran out of time before the feature freeze of 2.6. I do plan on finishing the whole job in 2.7, unless someone else beats me to it. There are a few things that need to be added to the driver model core in order for this to fully work (think of the i2c-dev driver for why...) > patch-1 > > Remove list of adapters and drivers in i2c-core. Driver-model > maintains these for us. That's nice, but we still need to be able to iterate over them at times, right? > patch-2 > > No functional changes, just make it more readable by moving code > here and there. Hm, any size savings here? Also, some of your documentation updates are nice, but why not use the proper kernel-doc format in order to be able to pick these functions up in the kernel-api documentation? thanks, greg k-h