On 28/03/2023 15:01:05+0200, Esteban Blanc wrote: > On Tue Mar 28, 2023 at 11:29 AM CEST, Alexandre Belloni wrote: > > Hello, > > > > On 28/03/2023 11:14:46+0200, Esteban Blanc wrote: > > > + /* Start rtc */ > > > + ret = regmap_set_bits(tps->regmap, TPS6594_REG_RTC_CTRL_1, > > > + TPS6594_BIT_STOP_RTC); > > > + if (ret < 0) > > > + return ret; > > > + > > > + mdelay(100); > > > + > > > + /* > > > + * RTC should be running now. Check if this is the case. > > > + * If not it might be a missing oscillator. > > > + */ > > > + ret = regmap_test_bits(tps->regmap, TPS6594_REG_RTC_STATUS, > > > + TPS6594_BIT_RUN); > > > + if (ret < 0) > > > + return ret; > > > + if (ret == 0) > > > + return -ENODEV; > > > + > > > + /* Stop RTC until first call to `tps6594_rtc_set_time */ > > > + ret = regmap_clear_bits(tps->regmap, TPS6594_REG_RTC_CTRL_1, > > > + TPS6594_BIT_STOP_RTC); > > > + if (ret < 0) > > > + return ret; > > > + > > > > This whole block must not be executed when the RTC is already running, > > else, you are stopping a perfectly running RTC. > > I'm not sure to get your point. You mean that during probe, the driver > might encounter an RTC device that is already running with a correct > timestamp? How would this be possible? A previous bootstage or the > driver was removed then re-inserted again? > The whole point of having an RTC is that the time tracking survives a reboot so yes, I would expect the RTC to have a valid timestamp at probe time. -- Alexandre Belloni, co-owner and COO, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
