Hi Lee,

On Wed, 18 Feb 2015, Lee Jones wrote:

ST's Low Power Controller (LPC) controls two devices; watchdog and RTC.
Only one of the devices can be used at any one time.  This is enforced
by the correlating MFD driver.  This portion of the driver-set controls
the Real Time Clock.

<snip>

+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&rtc->lock, flags);
+
+	writel(1, rtc->ioaddr + LPC_WDT_OFF);
+
+	writel(msb, rtc->ioaddr + LPC_LPA_MSB_OFF);
+	writel(lsb, rtc->ioaddr + LPC_LPA_LSB_OFF);
+	writel(1, rtc->ioaddr + LPC_LPA_START_OFF);
+
+	writel(0, rtc->ioaddr + LPC_WDT_OFF);

I think all the writel IO accessors in this driver can be replaced
with the *_relaxed variant. This will avoid the overhead of taking a
spinlock in the l2 outer cache part of writel.

<snip>

+#ifdef CONFIG_PM_SLEEP
+static int st_rtc_suspend(struct device *dev)
+{
+	struct st_rtc *rtc = dev_get_drvdata(dev);
+
+	if (device_may_wakeup(dev))
+		return 0;
+
+	writel(1, rtc->ioaddr + LPC_WDT_OFF);
+	writel(0, rtc->ioaddr + LPC_LPA_START_OFF);
+	writel(0, rtc->ioaddr + LPC_WDT_OFF);

Is there a reason for not disabling the IP clock in suspend / resume functions?

regards,

Peter.