On 24.06.19 10:44, Wolfram Sang wrote: > The correct solution is to fix the boot dependency in the affected I2C > client drivers. That definately needs HW and thorough testing. > > It may also need something better than the current deferred probe. Big > topic. So, then the current approach of using subsys_initcall() can't be changed easily, right now. But planned for the future (or at least not introducing new caes). But: how does that conflict w/ just moving the existing redundant pieces into a helper macro ? The logic stays the same - just using a shorter notation. (assuming my patch isn't buggy ;-)). I can add a remark in the function documentation that this shall only by used in rare cases, and maybe something like "that's just legacy - introducing new caller is most certainly wrong" ;-) --mtx -- Enrico Weigelt, metux IT consult Free software and Linux embedded engineering info@xxxxxxxxx -- +49-151-27565287
