Re: [PATCH v7 3/5] RTC: RK808: add RTC driver for RK808

[Doug Anderson <dianders@chromium.org>]

Chris,

On Tue, Sep 2, 2014 at 7:01 PM, Chris Zhong <zyw@rock-chips.com> wrote:
>
> On 09/02/2014 11:58 AM, Doug Anderson wrote:
>>>
>>> +static int rk808_rtc_probe(struct platform_device *pdev)
>>> >+{
>>> >+       struct rk808 *rk808 = dev_get_drvdata(pdev->dev.parent);
>>> >+       struct rk808_rtc *rk808_rtc;
>>> >+       struct rtc_time tm;
>>> >+       int ret;
>>> >+
>>> >+       rk808_rtc = devm_kzalloc(&pdev->dev, sizeof(*rk808_rtc),
>>> > GFP_KERNEL);
>>> >+       if (rk808_rtc == NULL)
>>> >+               return -ENOMEM;
>>> >+
>>> >+       platform_set_drvdata(pdev, rk808_rtc);
>>> >+       rk808_rtc->rk808 = rk808;
>>> >+
>>> >+       /* start rtc running by default, and use shadowed timer. */
>>> >+       ret = regmap_update_bits(rk808->regmap, RK808_RTC_CTRL_REG,
>>> >+                                BIT_RTC_CTRL_REG_STOP_RTC_M |
>>> >+                                BIT_RTC_CTRL_REG_RTC_READSEL_M,
>>> >+                                BIT_RTC_CTRL_REG_RTC_READSEL_M);
>>
>> I think this should still be setting to 0, not to
>> BIT_RTC_CTRL_REG_RTC_READSEL_M.  Otherwise the first read of the time
>> will return that time that was frozen at probe time, right?  AKA: if
>> probe happens at 11:00:00 and then we read the time at 11:00:05 we'll
>> still read 11:00:00 the first time.
>>
>  Sorry, I did not describe correctly, in the previous mail.
>  Actually, RK808 has a "GET_TIME" switch bit. When "GET_TIME" bit rising
> thansiton to 1,
>  the current time will save in a shadowed register.
>  If "READSEL" = 1, read rtc time register, return the frozen time.
>  If we need the real time, clr this "READSEL" bit.

Ohhhhh!  I see.  I clearly didn't look closely enough at the whole
register description in the user manual.  I think you've got it right,
then.

I think this would be more obvious if you added a comment and you set
both the high and low bits at the same time.  AKA:

+       /* Force an update of the shadowed registers right now */
+       ret = regmap_update_bits(rk808->regmap, RK808_RTC_CTRL_REG,
+                                BIT_RTC_CTRL_REG_RTC_GET_TIME,
+                                BIT_RTC_CTRL_REG_RTC_GET_TIME);
+       if (ret) {
+               dev_err(dev, "Failed to update bits rtc_ctrl: %d\n", ret);
+               return ret;
+       }
+
+       ret = regmap_update_bits(rk808->regmap, RK808_RTC_CTRL_REG,
+                                BIT_RTC_CTRL_REG_RTC_GET_TIME,
+                                BIT_RTC_CTRL_REG_RTC_GET_TIME);
+       if (ret) {
+               dev_err(dev, "Failed to update bits rtc_ctrl: %d\n", ret);
+               return ret;
+       }

...doing them both at the same time makes it obvious that this is a
one-time copy triggered by the "rising edge" of this bit.  It also
means you don't need any "goto" for error handling.

-Doug