On 4 February 2013 20:25, Linus Walleij [off-list ref] wrote:

Hi Haojian, sorry for taking too long to reply...

On Wed, Jan 30, 2013 at 2:04 AM, Haojian Zhuang
[off-list ref] wrote:

quoted

RTC control register should be enabled in the process of initliazing.

Signed-off-by: Haojian Zhuang <haojian.zhuang@linaro.org>

(...)

quoted

+       data = readl(ldata->base + RTC_CR);
        /* Enable the clockwatch on ST Variants */
        if (vendor->clockwatch)
-               writel(readl(ldata->base + RTC_CR) | RTC_CR_CWEN,
-                      ldata->base + RTC_CR);
+               data |= RTC_CR_CWEN;
+       writel(data | RTC_CR_EN, ldata->base + RTC_CR);

This last line is *not* OK on the ST Variant. In our hardware that bit
is part of the clock divider, which means it will affect our timekeeping.

Do you want me to submit a follow-up patch?

Yours,
Linus Walleij

I prefer you can submit a follow up patch. And I think that you can
use a compatible
name to distinguish from original "arm,rtc-pl031". Then we can get two
branches to
handle the difference in the probe function. What's your opinion?

Regards
Haojian