| Message ID | 1359944492-23130-1-git-send-email-shawn.guo@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Mon, Feb 04, 2013 at 10:21:32AM +0800, Shawn Guo wrote: > For cpufreq example, it takes 13 steps (25 mV for one step) to increase > vddcore from 0.95 V to 1.275 V, and the time of 64 clock cycles at > 24 MHz for one step is ~2.67 uS, so the total delay time would be > ~34.71 uS. But the current calculation in the driver gives 39 uS. > Change the formula to have the addition of 1 be the last step, so that > we can get a more precise delay time. For example above, the new > formula will give 35 uS. > > Signed-off-by: Shawn Guo <shawn.guo@linaro.org> > --- > drivers/regulator/anatop-regulator.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/regulator/anatop-regulator.c b/drivers/regulator/anatop-regulator.c > index 0df9c6a..0d4a8cc 100644 > --- a/drivers/regulator/anatop-regulator.c > +++ b/drivers/regulator/anatop-regulator.c > @@ -80,8 +80,8 @@ static int anatop_regmap_set_voltage_time_sel(struct regulator_dev *reg, > regmap_read(anatop_reg->anatop, anatop_reg->delay_reg, &val); > val = (val >> anatop_reg->delay_bit_shift) & > ((1 << anatop_reg->delay_bit_width) - 1); > - ret = (new_sel - old_sel) * ((LDO_RAMP_UP_UNIT_IN_CYCLES << > - val) / LDO_RAMP_UP_FREQ_IN_MHZ + 1); > + ret = (new_sel - old_sel) * (LDO_RAMP_UP_UNIT_IN_CYCLES << > + val) / LDO_RAMP_UP_FREQ_IN_MHZ + 1; Agreed that this formula will make the udelay more precise, the previous formula is considering a program model for udelay margin(the biggest margin is about 9us), maybe we don't need so big margin, 1us may be enough, as the hardware design of this step time is already very conservative. > } > > return ret; > -- > 1.7.9.5 >
On Mon, Feb 04, 2013 at 10:21:32AM +0800, Shawn Guo wrote: > For cpufreq example, it takes 13 steps (25 mV for one step) to increase > vddcore from 0.95 V to 1.275 V, and the time of 64 clock cycles at > 24 MHz for one step is ~2.67 uS, so the total delay time would be > ~34.71 uS. But the current calculation in the driver gives 39 uS. > Change the formula to have the addition of 1 be the last step, so that > we can get a more precise delay time. For example above, the new > formula will give 35 uS. Applied, thanks.
diff --git a/drivers/regulator/anatop-regulator.c b/drivers/regulator/anatop-regulator.c index 0df9c6a..0d4a8cc 100644 --- a/drivers/regulator/anatop-regulator.c +++ b/drivers/regulator/anatop-regulator.c @@ -80,8 +80,8 @@ static int anatop_regmap_set_voltage_time_sel(struct regulator_dev *reg, regmap_read(anatop_reg->anatop, anatop_reg->delay_reg, &val); val = (val >> anatop_reg->delay_bit_shift) & ((1 << anatop_reg->delay_bit_width) - 1); - ret = (new_sel - old_sel) * ((LDO_RAMP_UP_UNIT_IN_CYCLES << - val) / LDO_RAMP_UP_FREQ_IN_MHZ + 1); + ret = (new_sel - old_sel) * (LDO_RAMP_UP_UNIT_IN_CYCLES << + val) / LDO_RAMP_UP_FREQ_IN_MHZ + 1; } return ret;
For cpufreq example, it takes 13 steps (25 mV for one step) to increase vddcore from 0.95 V to 1.275 V, and the time of 64 clock cycles at 24 MHz for one step is ~2.67 uS, so the total delay time would be ~34.71 uS. But the current calculation in the driver gives 39 uS. Change the formula to have the addition of 1 be the last step, so that we can get a more precise delay time. For example above, the new formula will give 35 uS. Signed-off-by: Shawn Guo <shawn.guo@linaro.org> --- drivers/regulator/anatop-regulator.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)