Re: [PATCH 2/2] rtc: sprd: Add Spreadtrum RTC driver

From: Baolin Wang
Date: Wed Nov 08 2017 - 00:29:28 EST


Hi Alexandre,

On 8 November 2017 at 11:44, Alexandre Belloni
<alexandre.belloni@xxxxxxxxxxxxxxxxxx> wrote:
> Hi,
>
> On 07/11/2017 at 19:34:08 +0800, Baolin Wang wrote:
>> This patch adds the Spreadtrum RTC driver, which embedded in the
>> Spreadtrum SC27xx PMIC series.
>>
>> Signed-off-by: Baolin Wang <baolin.wang@xxxxxxxxxxxxxx>
>> ---
>> drivers/rtc/Kconfig | 7 +
>> drivers/rtc/Makefile | 1 +
>> drivers/rtc/rtc-sprd.c | 673 ++++++++++++++++++++++++++++++++++++++++++++++++
>
> I would prefer that filename to be more specific. What if at some point
> one of your SoC includes an RTC that is not handled by this driver. This
> will become a naming nightmare.

Make sense, I will change the filename to 'rtc-sc27xx.c'.

>
>> 3 files changed, 681 insertions(+)
>> create mode 100644 drivers/rtc/rtc-sprd.c
>>
>> diff --git a/drivers/rtc/Kconfig b/drivers/rtc/Kconfig
>> index e0e58f3..c3059fa 100644
>> --- a/drivers/rtc/Kconfig
>> +++ b/drivers/rtc/Kconfig
>> @@ -1182,6 +1182,13 @@ config RTC_DRV_SPEAR
>> If you say Y here you will get support for the RTC found on
>> spear
>>
>> +config RTC_DRV_SPRD
>> + tristate "Spreadtrum SC27xx RTC"
>> + depends on ARCH_SPRD || COMPILE_TEST
>
> the commit message says this is a PMIC RTC, not a SoC RTC. I don't think
> you need to depend on ARCH_SPRD. But you probably need to depend on the
> MFD parent.

Right. Will fix it in next version.

>
>> + help
>> + If you say Y here you will get support for the RTC found on
>> + Spreadtrum.
>
> Be mor precise here.

OK.

>
>> +
>> config RTC_DRV_PCF50633
>> depends on MFD_PCF50633
>> tristate "NXP PCF50633 RTC"
>> diff --git a/drivers/rtc/Makefile b/drivers/rtc/Makefile
>> index 7230014..ec5cb75 100644
>> --- a/drivers/rtc/Makefile
>> +++ b/drivers/rtc/Makefile
>> @@ -147,6 +147,7 @@ obj-$(CONFIG_RTC_DRV_SH) += rtc-sh.o
>> obj-$(CONFIG_RTC_DRV_SIRFSOC) += rtc-sirfsoc.o
>> obj-$(CONFIG_RTC_DRV_SNVS) += rtc-snvs.o
>> obj-$(CONFIG_RTC_DRV_SPEAR) += rtc-spear.o
>> +obj-$(CONFIG_RTC_DRV_SPRD) += rtc-sprd.o
>> obj-$(CONFIG_RTC_DRV_STARFIRE) += rtc-starfire.o
>> obj-$(CONFIG_RTC_DRV_STK17TA8) += rtc-stk17ta8.o
>> obj-$(CONFIG_RTC_DRV_STM32) += rtc-stm32.o
>> diff --git a/drivers/rtc/rtc-sprd.c b/drivers/rtc/rtc-sprd.c
>> new file mode 100644
>> index 0000000..6b2477f
>> --- /dev/null
>> +++ b/drivers/rtc/rtc-sprd.c
>> @@ -0,0 +1,673 @@
>> +/*
>> + * Copyright (C) 2017 Spreadtrum Communications Inc.
>> + *
>> + * SPDX-License-Identifier: GPL-2.0
>> + */
>> +
>> +#include <linux/bitops.h>
>> +#include <linux/delay.h>
>> +#include <linux/err.h>
>> +#include <linux/module.h>
>> +#include <linux/of.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/regmap.h>
>> +#include <linux/rtc.h>
>> +
>> +#define SPRD_RTC_SEC_CNT_VALUE 0x0
>> +#define SPRD_RTC_MIN_CNT_VALUE 0x4
>> +#define SPRD_RTC_HOUR_CNT_VALUE 0x8
>> +#define SPRD_RTC_DAY_CNT_VALUE 0xc
>> +#define SPRD_RTC_SEC_CNT_UPD 0x10
>> +#define SPRD_RTC_MIN_CNT_UPD 0x14
>> +#define SPRD_RTC_HOUR_CNT_UPD 0x18
>> +#define SPRD_RTC_DAY_CNT_UPD 0x1c
>> +#define SPRD_RTC_SEC_ALM_UPD 0x20
>> +#define SPRD_RTC_MIN_ALM_UPD 0x24
>> +#define SPRD_RTC_HOUR_ALM_UPD 0x28
>> +#define SPRD_RTC_DAY_ALM_UPD 0x2c
>> +#define SPRD_RTC_INT_EN 0x30
>> +#define SPRD_RTC_INT_RAW_STS 0x34
>> +#define SPRD_RTC_INT_CLR 0x38
>> +#define SPRD_RTC_INT_MASK_STS 0x3C
>> +#define SPRD_RTC_SEC_ALM_VALUE 0x40
>> +#define SPRD_RTC_MIN_ALM_VALUE 0x44
>> +#define SPRD_RTC_HOUR_ALM_VALUE 0x48
>> +#define SPRD_RTC_DAY_ALM_VALUE 0x4c
>> +#define SPRD_RTC_SPG_VALUE 0x50
>> +#define SPRD_RTC_SPG_UPD 0x54
>> +#define SPRD_RTC_SEC_AUXALM_UPD 0x60
>> +#define SPRD_RTC_MIN_AUXALM_UPD 0x64
>> +#define SPRD_RTC_HOUR_AUXALM_UPD 0x68
>> +#define SPRD_RTC_DAY_AUXALM_UPD 0x6c
>> +
>> +/* BIT & MASK definition for SPRD_RTC_INT_* registers */
>> +#define SPRD_RTC_SEC_EN BIT(0)
>> +#define SPRD_RTC_MIN_EN BIT(1)
>> +#define SPRD_RTC_HOUR_EN BIT(2)
>> +#define SPRD_RTC_DAY_EN BIT(3)
>> +#define SPRD_RTC_ALARM_EN BIT(4)
>> +#define SPRD_RTC_HRS_FORMAT_EN BIT(5)
>> +#define SPRD_RTC_AUXALM_EN BIT(6)
>> +#define SPRD_RTC_SPG_UPD_EN BIT(7)
>> +#define SPRD_RTC_SEC_UPD_EN BIT(8)
>> +#define SPRD_RTC_MIN_UPD_EN BIT(9)
>> +#define SPRD_RTC_HOUR_UPD_EN BIT(10)
>> +#define SPRD_RTC_DAY_UPD_EN BIT(11)
>> +#define SPRD_RTC_ALMSEC_UPD_EN BIT(12)
>> +#define SPRD_RTC_ALMMIN_UPD_EN BIT(13)
>> +#define SPRD_RTC_ALMHOUR_UPD_EN BIT(14)
>> +#define SPRD_RTC_ALMDAY_UPD_EN BIT(15)
>> +#define SPRD_RTC_INT_MASK GENMASK(15, 0)
>> +
>> +#define SPRD_RTC_TIME_INT_MASK \
>> + (SPRD_RTC_SEC_UPD_EN | SPRD_RTC_MIN_UPD_EN | \
>> + SPRD_RTC_HOUR_UPD_EN | SPRD_RTC_DAY_UPD_EN)
>> +
>> +#define SPRD_RTC_ALMTIME_INT_MASK \
>> + (SPRD_RTC_ALMSEC_UPD_EN | SPRD_RTC_ALMMIN_UPD_EN | \
>> + SPRD_RTC_ALMHOUR_UPD_EN | SPRD_RTC_ALMDAY_UPD_EN)
>> +
>> +#define SPRD_RTC_ALM_INT_MASK \
>> + (SPRD_RTC_SEC_EN | SPRD_RTC_MIN_EN | \
>> + SPRD_RTC_HOUR_EN | SPRD_RTC_DAY_EN | \
>> + SPRD_RTC_ALARM_EN | SPRD_RTC_AUXALM_EN)
>> +
>> +/* second/minute/hour/day values mask definition */
>> +#define SPRD_RTC_SEC_MASK GENMASK(5, 0)
>> +#define SPRD_RTC_MIN_MASK GENMASK(5, 0)
>> +#define SPRD_RTC_HOUR_MASK GENMASK(4, 0)
>> +#define SPRD_RTC_DAY_MASK GENMASK(15, 0)
>> +
>> +/* alarm lock definition for SPRD_RTC_SPG_UPD register */
>> +#define SPRD_RTC_ALMLOCK_MASK GENMASK(7, 0)
>> +#define SPRD_RTC_ALM_UNLOCK 0xa5
>> +#define SPRD_RTC_ALM_LOCK (~SPRD_RTC_ALM_UNLOCK & SPRD_RTC_ALMLOCK_MASK)
>> +
>> +/* SPG values definition for SPRD_RTC_SPG_UPD register */
>> +#define SPRD_RTC_POWEROFF_ALM_FLAG BIT(8)
>> +#define SPRD_RTC_POWER_RESET_FLAG BIT(9)
>> +
>> +/* timeout of synchronizing time and alarm registers (us) */
>> +#define SPRD_RTC_POLL_TIMEOUT 200000
>> +#define SPRD_RTC_POLL_DELAY_US 20000
>> +#define SPRD_RTC_START_YEAR 2017
>> +
>> +struct sprd_rtc {
>> + struct rtc_device *rtc;
>> + struct regmap *regmap;
>> + struct device *dev;
>> + u32 base;
>> + int irq;
>> +};
>> +
>> +/*
>> + * The Spreadtrum RTC controller has 3 groups registers, including time, normal
>> + * alarm and auxiliary alarm. The time group registers are used to set RTC time,
>> + * the normal alarm registers are used to set normal alarm, and the auxiliary
>> + * alarm registers are used to set auxiliary alarm. Both alarm event and
>> + * auxiliary alarm event can wake up system from deep sleep, but only alarm
>> + * event can power up system from power down status.
>> + */
>> +enum sprd_rtc_reg_types {
>> + SPRD_RTC_TIME,
>> + SPRD_RTC_ALARM,
>> + SPRD_RTC_AUX_ALARM,
>> +};
>> +
>> +static int sprd_rtc_read(struct sprd_rtc *rtc, u32 reg, u32 *val)
>> +{
>> + return regmap_read(rtc->regmap, rtc->base + reg, val);
>> +}
>> +
>> +static int sprd_rtc_write(struct sprd_rtc *rtc, u32 reg, u32 val)
>> +{
>> + return regmap_write(rtc->regmap, rtc->base + reg, val);
>> +}
>> +
>> +static int sprd_rtc_update(struct sprd_rtc *rtc, u32 reg, u32 mask, u32 val)
>> +{
>> + return regmap_update_bits(rtc->regmap, rtc->base + reg, mask, val);
>> +}
>> +
>> +static int sprd_rtc_read_poll(struct sprd_rtc *rtc, u32 reg, u32 mask)
>> +{
>> + u32 val;
>> +
>> + return regmap_read_poll_timeout(rtc->regmap, rtc->base + reg, val,
>> + ((val & mask) == mask),
>> + SPRD_RTC_POLL_DELAY_US,
>> + SPRD_RTC_POLL_TIMEOUT);
>> +}
>> +
>
> All those functions are only wrappers around regmap, they are useless,
> please use regmap_* directly.

OK, I will do as you suggested.

>> +static int sprd_rtc_read_aux_alarm(struct device *dev, struct rtc_wkalrm *alrm)
>> +{
>> + struct sprd_rtc *rtc = dev_get_drvdata(dev);
>> + time64_t secs;
>> + u32 val;
>> + int ret;
>> +
>> + ret = sprd_rtc_get_secs(rtc, SPRD_RTC_AUX_ALARM, &secs);
>> + if (ret)
>> + return ret;
>> +
>> + rtc_time64_to_tm(secs, &alrm->time);
>> +
>> + ret = sprd_rtc_read(rtc, SPRD_RTC_INT_EN, &val);
>> + if (ret)
>> + return ret;
>> +
>> + alrm->enabled = !!(val & SPRD_RTC_AUXALM_EN);
>> +
>> + ret = sprd_rtc_read(rtc, SPRD_RTC_INT_RAW_STS, &val);
>> + if (ret)
>> + return ret;
>> +
>> + alrm->pending = !!(val & SPRD_RTC_AUXALM_EN);
>> + return 0;
>> +}
>> +
>> +static int sprd_rtc_set_aux_alarm(struct device *dev, struct rtc_wkalrm *alrm)
>> +{
>> + struct sprd_rtc *rtc = dev_get_drvdata(dev);
>> + int ret;
>> +
>> + /* clear the auxiliary alarm interrupt status */
>> + ret = sprd_rtc_write(rtc, SPRD_RTC_INT_CLR, SPRD_RTC_AUXALM_EN);
>> + if (ret)
>> + return ret;
>> +
>> + if (alrm->enabled) {
>> + time64_t secs = rtc_tm_to_time64(&alrm->time);
>> +
>> + /* enable the auxiliary alarm interrupt */
>> + ret = sprd_rtc_update(rtc, SPRD_RTC_INT_EN, SPRD_RTC_AUXALM_EN,
>> + SPRD_RTC_AUXALM_EN);
>> + if (ret)
>> + return ret;
>> +
>> + ret = sprd_rtc_set_secs(rtc, SPRD_RTC_AUX_ALARM, secs);
>> + if (ret) {
>> + sprd_rtc_update(rtc, SPRD_RTC_INT_EN,
>> + SPRD_RTC_AUXALM_EN, 0);
>> + }
>> + } else {
>> + /* disable the auxiliary alarm interrupt */
>> + ret = sprd_rtc_update(rtc, SPRD_RTC_INT_EN,
>> + SPRD_RTC_AUXALM_EN, 0);
>
> This is not setting the alarm registers properly. What if the alarm is
> enabled at a later time by sprd_rtc_alarm_irq_enable()?

Ok, I understand your meaning, I will modify the logic in next version.

>> +static int sprd_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
>> +{
>> + struct sprd_rtc *rtc = dev_get_drvdata(dev);
>> + struct rtc_time aie_time =
>> + rtc_ktime_to_tm(rtc->rtc->aie_timer.node.expires);
>> + int ret;
>> +
>> + /*
>> + * We have 2 groups alarms: normal alarm and auxiliary alarm. Since
>> + * both normal alarm event and auxiliary alarm event can wake up system
>> + * from deep sleep, but only alarm event can power up system from power
>> + * down status. Moreover we do not need to poll about 125ms when
>> + * updating auxiliary alarm registers. Thus we usually set auxiliary
>> + * alarm when wake up system from deep sleep, and for other scenarios,
>> + * we should set normal alarm with polling status.
>> + *
>> + * So here we check if the alarm time is set by aie_timer, if yes, we
>> + * should set normal alarm, if not, we should set auxiliary alarm which
>> + * means it is just a wake event.
>> + */
>> + if (!rtc->rtc->aie_timer.enabled || rtc_tm_sub(&aie_time, &alrm->time))
>> + return sprd_rtc_set_aux_alarm(dev, alrm);
>> +
>> + /* clear the alarm interrupt status firstly */
>> + ret = sprd_rtc_write(rtc, SPRD_RTC_INT_CLR, SPRD_RTC_ALARM_EN);
>> + if (ret)
>> + return ret;
>> +
>> + if (alrm->enabled) {
>> + time64_t secs = rtc_tm_to_time64(&alrm->time);
>> +
>> + /* enable the alarm interrupt */
>> + ret = sprd_rtc_update(rtc, SPRD_RTC_INT_EN, SPRD_RTC_ALARM_EN,
>> + SPRD_RTC_ALARM_EN);
>> + if (ret)
>> + return ret;
>> +
>> + ret = sprd_rtc_set_secs(rtc, SPRD_RTC_ALARM, secs);
>> + if (ret)
>> + goto err;
>> +
>> + /* unlock the alarm to enable the alarm function. */
>> + ret = sprd_rtc_unlock_alarm(rtc);
>> + if (ret)
>> + goto err;
>> + } else {
>
> ditto.
>
>> + /* disable the alarm interrupt */
>> + ret = sprd_rtc_update(rtc, SPRD_RTC_INT_EN,
>> + SPRD_RTC_ALARM_EN, 0);
>> + if (ret)
>> + return ret;
>> +
>> + /*
>> + * Lock the alarm function in case fake alarm event will power
>> + * up systems.
>> + */
>> + ret = sprd_rtc_lock_alarm(rtc);
>> + }
>> +
>> + return ret;
>> +
>> +err:
>> + sprd_rtc_update(rtc, SPRD_RTC_INT_EN, SPRD_RTC_ALARM_EN, 0);
>> + return ret;
>> +}
>> +
>> +static int sprd_rtc_set_mmss(struct device *dev, time64_t secs)
>> +{
>> + struct sprd_rtc *rtc = dev_get_drvdata(dev);
>> +
>> + return sprd_rtc_set_secs(rtc, SPRD_RTC_TIME, secs);
>> +}
>
> Because set_time is defined, this function will never be called, please
> remove it.

Ah, yes, I will remove it.

>> +
>> +static int sprd_rtc_init_time(struct sprd_rtc *rtc)
>> +{
>> + time64_t secs = mktime64(SPRD_RTC_START_YEAR, 1, 1, 0, 0, 0);
>> +
>> + dev_dbg(rtc->dev, "RTC power down and reset time.\n");
>> + return sprd_rtc_set_secs(rtc, SPRD_RTC_TIME, secs);
>> +}
>> +
>> +static int sprd_rtc_check_power_down(struct sprd_rtc *rtc)
>> +{
>> + int ret;
>> + u32 val;
>> +
>> + ret = sprd_rtc_read(rtc, SPRD_RTC_SPG_VALUE, &val);
>> + if (ret)
>> + return ret;
>> +
>> + if (val & SPRD_RTC_POWER_RESET_FLAG)
>> + return 0;
>> +
>> + ret = sprd_rtc_init_time(rtc);
>
> Don't do that. This is a valuable information. If you know the time is
> invalid, return -EINVAL in read_time(). What your are doing here is
> confusing the system by making it believe your fake date is the correct
> time.

Usually for mobile device, we should give one reasonable start time if
the RTC powered down. Anyway I can remove this feature now.
Very appreciated for your helpful comments.

--
Baolin.wang
Best Regards