On Wed, Nov 05, 2014 at 03:55:26PM +0000, Jonathan Cameron wrote:
> On 31/10/14 18:43, Hartmut Knaack wrote:
> > Vlad Dogaru schrieb am 31.10.2014 12:44:
> >> On Fri, Oct 31, 2014 at 02:23:33AM +0100, Hartmut Knaack wrote:
> >>> The calculations for temperature and pressure compensation were already slightly
> >>> optimized in comparison to the data sheet. So, it makes sense to optimize even a
> >>> bit more, making proper use of C operators:
> >>> - variable t in bmp280_compensate_temp() can be eliminated by directly
> >>> returning the result of the relevant equation.
> >>> - make use of the += operator, eliminate an unnecessary parenthesis level and
> >>> directly return the result of the last equation in
> >>> bmp280_compensate_press().
> >>> When the initialization of the ctrl_meas register fails, the error message will
> >>> now mention the right register name.
> >>> During probe, i2c_set_clientdata() is called, although it is not necessary. Drop
> >>> it.
> >>>
> >>> Signed-off-by: Hartmut Knaack <knaack.h@xxxxxx>
> >>> ---
> >>> diff --git a/drivers/iio/pressure/bmp280.c b/drivers/iio/pressure/bmp280.c
> >>> index 75038da..4f6ae4d 100644
> >>> --- a/drivers/iio/pressure/bmp280.c
> >>> +++ b/drivers/iio/pressure/bmp280.c
> >>> @@ -200,7 +200,7 @@ static s32 bmp280_compensate_temp(struct bmp280_data *data,
> >>> struct bmp280_comp_temp *comp,
> >>> s32 adc_temp)
> >>> {
> >>> - s32 var1, var2, t;
> >>> + s32 var1, var2;
> >>>
> >>> var1 = (((adc_temp >> 3) - ((s32) comp->dig_t1 << 1)) *
> >>> ((s32) comp->dig_t2)) >> 11;
> >>> @@ -209,9 +209,7 @@ static s32 bmp280_compensate_temp(struct bmp280_data *data,
> >>> ((s32) comp->dig_t3)) >> 14;
> >>>
> >>> data->t_fine = var1 + var2;
> >>> - t = (data->t_fine * 5 + 128) >> 8;
> >>> -
> >>> - return t;
> >>> + return (data->t_fine * 5 + 128) >> 8;
> >>
> >> Shouldn't the compiler take care of this?
> > That would be preferable. I just don't see the real benefit in having the extra step of storing the result (and taking care of an extra variable) before returning it. And I am aware, that this calculation is derived from the one in the data sheet (which looks a bit questionable to me with its unnecessary parenthesis and variable). But since you already started optimizing, it seemed legitimate to consolidate it even a bit more.
>
>
> I'm with Hartmut on this, no point in having more actual code / local variables than
> needed... Just a few more lines of code for no gain :)
I guess the problem is I started with the exact code from the datasheet,
then refactored a bit to accomodate the usage of div64_s64 below. Code
does look cleaner now, thanks Hartmut!
Tested-by: Vlad Dogaru <vlad.dogaru@xxxxxxxxx>
> >>
> >>> }
> >>>
> >>> /*
> >>> @@ -229,11 +227,11 @@ static u32 bmp280_compensate_press(struct bmp280_data *data,
> >>>
> >>> var1 = ((s64) data->t_fine) - 128000;
> >>> var2 = var1 * var1 * (s64) comp->dig_p6;
> >>> - var2 = var2 + ((var1 * (s64) comp->dig_p5) << 17);
> >>> - var2 = var2 + (((s64) comp->dig_p4) << 35);
> >>> + var2 += ((var1 * (s64) comp->dig_p5) << 17);
> >>> + var2 += (((s64) comp->dig_p4) << 35);
> >>> var1 = ((var1 * var1 * (s64) comp->dig_p3) >> 8) +
> >>> ((var1 * (s64) comp->dig_p2) << 12);
> >>> - var1 = (((((s64) 1) << 47) + var1)) * ((s64) comp->dig_p1) >> 33;
> >>> + var1 = ((((s64) 1) << 47) + var1) * ((s64) comp->dig_p1) >> 33;
> >>>
> >>> if (var1 == 0)
> >>> return 0;
> >>> @@ -242,9 +240,7 @@ static u32 bmp280_compensate_press(struct bmp280_data *data,
> >>> p = div64_s64(p, var1);
> >>> var1 = (((s64) comp->dig_p9) * (p >> 13) * (p >> 13)) >> 25;
> >>> var2 = (((s64) comp->dig_p8) * p) >> 19;
> >>> - p = ((p + var1 + var2) >> 8) + (((s64) comp->dig_p7) << 4);
> >>> -
> >>> - return (u32) p;
> >>> + return (u32)((p + var1 + var2) >> 8) + (((s64) comp->dig_p7) << 4);
> >>
> >> And this?
> >> --
> >> 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
> >>
> >
> > --
> > 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
> >
--
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
