On Sat, Jun 18, 2022 at 5:16 PM ChiYuan Huang <u0084500@xxxxxxxxx> wrote: > Andy Shevchenko <andy.shevchenko@xxxxxxxxx> 於 2022年6月18日 週六 凌晨1:08寫道： > > On Fri, Jun 17, 2022 at 11:37 AM cy_huang <u0084500@xxxxxxxxx> wrote: ... > > > +KernelVersion: 5.18.2 > > > > Wrong version, this won't be part of a stable kernel. > > > From kernel.org, currently the stable kernel version is 5.18.5. > Change to 5.18.5? Nope. You need to use realistic kernel version, and as I said it can't be a stable one. ... > > But for sure you missed > > types.h > > > Ack in next. But for types.h, i2c.h already include device.h. > And device.h already include types.h. > Is it still needed to declare explicitly for types.h?? Yes. You have to include all headers you are a direct user of, except the ones that are guaranteed to be included by others. The types.h is not guaranteed to be included by listed above. ... > > > + struct device *dev; > > > + struct regmap *regmap; > > > > Swapping these two might give less code in the generated binary. Have > > you run bloat-o-meter? > > > I never know about this tool. > I'll check it before I submit the next revision. > Thanks for the reminding. > > But from Jonathan's reply, I may remove 'struct regmap *regmap'. > If all function need the 'regmap', a local variable 'regmap' need to > be declared. > To use struct regmap *regmap = dev_get_regmap(dev, NULL) is more effective. It's fine, but you may experiment with bloat-o-meter even in that case out of curiosity. -- With Best Regards, Andy Shevchenko
