The wakealarm attribute is currently not exposed in the sysfs interface
as the device has not been set as doing wakealarm when device_register
is called. Changing the order of the calls fixes that problem. Interrupts
are cleared in check_rtc_status prior to requesting the interrupt.

This is only set if an irq is defined. If irq registration fails then
set wakeup_capable to false. With this change the sysfs wakealarm
attribute will be left visible but it is non functional. rtcwake
still returns that the device is not enabled for wakeup.

Signed-off-by: Phil Reid <preid@electromag.com.au>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

Add the max_register  to the regmap_config definition. This allows
dumping of the device's registers via the regmap debugfs interface.

Signed-off-by: Phil Reid <preid@electromag.com.au>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

Whitespace was a combination of spaces and tabs.
Use spaces and align register / bit definitions.

Signed-off-by: Phil Reid <preid@electromag.com.au>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

Since commit d68778b80dd7 ("rtc: initialize output parameter for read
alarm to "uninitialized"") there is no need to explicitly set
unsupported members to -1. So drop the respective assignments from
drivers.

Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

The rtc->ops_lock would be accessed in ds3232_irq() without being
initialized as rtc_device_register() is called too late.

So move devm_rtc_device_register() just before registering irq handler
to initialize rtc->ops_lock earlier.

Signed-off-by: Gong Qianyu <Qianyu.Gong@nxp.com>
Reviewed-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

ds3232->mutex is used to protect for alarm operations which
need to access status and control registers.

But we can use rtc->ops_lock instead.  rtc->ops_lock is held when most
of rtc_class_ops methods are called, so we only need to explicitly
acquire it from irq handler in order to protect form concurrent
accesses.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

ds3232-core requests irq with IRQF_SHARED, so irq can be shared by
several devices.  But the irq handler for ds3232 unconditionally
disables the irq at first and the irq is re-enabled only when the
interrupt source was the ds3232's alarm.  This behaviour breaks the
devices sharing the same irq in the various scenarios.

This converts to use threaded irq and remove outdated code in
suspend/resume paths.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Suggested-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

UIE mode irqs are handled by the generic rtc core now.  But there are
remaining unused code fragments for it.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

Add missing register access error checks and make it return error code
or print error message.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>

The rtctest (tools/testing/selftests/timers/rtctest.c) found that
reading ds3232 rtc device immediately return the value 0x20 (RTC_AF)
without waiting alarm interrupt.

This is because alarm_irq_enable() of ds3232 driver changes RTC_AF
flag in rtc->irq_data.  So calling ioctl with RTC_AIE_ON generates
invalid value in rtc device.

The lower-level driver should not touch rtc->irq_data directly.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>