On Mon, Feb 26, 2018 at 10:43:07PM +0100, Wolfram Sang wrote: > > > It is an alternate and somewhat simpler solution which does not > > attempt to change the kernel API. My primary goal was to find > > a solution for the immediate problem between the watchdog and > > i2c drivers. My solution does not attempt to fix the access problem > > for (initialization) functions with no return code. At the same time, > > I prefer to use functions which _do_ return an error if that is possible > > since it _may_ after all be that there are other users calling > > request_region(). > > > > Having said that, I'll be just as happy if Zoltan's series is accepted > > instead of this patch (except for patch #5; the watchdog driver has > > already been converted to use request_muxed_region). > > Thanks for the explanation! > > So, as I understand: I can apply this patch to fix the immediate (and > long standing) problem and decouple it from the API change. The latter > can then still be worked on and the changes for the piix driver should > then be minimal. Correct? > Correct. Actually, you won't need any further piix driver changes; my understanding is that patches 1-3 from Zoltan's series can be applied independently. Guenter
