On Tue, Jun 28, 2016 at 09:34:34AM +0200, Linus Walleij wrote: > On Mon, Jun 27, 2016 at 8:57 PM, Jonathan Cameron <jic23@xxxxxxxxxx> wrote: > > On 27/06/16 08:42, Vlad Dogaru wrote: > >> On Wed, Jun 22, 2016 at 10:53:39PM +0200, Linus Walleij wrote: > >>> The calibration data is described as coming from an E2PROM and that > >>> means it does not change. Just read it once at probe time and store > >>> it in the device state container. Also toss the calibration data > >>> into the entropy pool since it is device unique. > >> > >> I think my initial thought when writing this was that regmap will take > >> care of the caching and not hit the i2c bus each time. But I don't have > >> an issue with this change. Other than that, series looks good to me. > > > > Good point. I must be half asleep. > > > > Linus, are we missing something? If not, I'd go without this one. > > It's true that regmap does cache the contents of the registers, > cutting the I2C bus traffic bottleneck. But reading it out every time > still incurs a bulk read of 22 bytes from regmap, meaning 22 iterations > in regmap where each need to be checked individually for volatileness > etc in the regmap core, then this happens at every conversion. > > Compared to just doing this once and keeping a copy around > it is wasteful, so still an optimization worth it IMO. > > Don't get me wrong, regmap is optimized > (just check drivers/base/regmap/*) but it can never be as optimized > as reading just once. > > Also Vlad seems to OK it? Yep, it's not a problem for me. You're right about reading once being more efficient and. Thanks, Vlad -- To unsubscribe from this list: send the line "unsubscribe linux-iio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
