i2c: nomadik: Fixup system suspend
authorUlf Hansson <ulf.hansson@linaro.org>
Thu, 10 Apr 2014 13:59:36 +0000 (15:59 +0200)
committerWolfram Sang <wsa@the-dreams.de>
Thu, 22 May 2014 08:09:21 +0000 (10:09 +0200)
For !CONFIG_PM_RUNTIME, the device were never put back into active
state while resuming.

For CONFIG_PM_RUNTIME, we blindly trusted the device to be inactive
while we were about to handle it at suspend late, which is just too
optimistic.

Even if the driver uses pm_runtime_put_sync() after each tranfer to
return it's runtime PM resources, there are no guarantees this will
actually mean the device will inactivated. The reason is that the PM
core will prevent runtime suspend during system suspend, and thus when
a transfer occurs during the early phases of system suspend the device
will be kept active after the transfer.

To handle both issues above, use pm_runtime_force_suspend|resume() from
the system suspend|resume callbacks.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
drivers/i2c/busses/i2c-nomadik.c

index 32c85e9ecdaeb141f0653c3a7e8c85784a1140ce..0e55d85fd4ed9147affa3cf6b7b18adc72c15702 100644 (file)
@@ -879,19 +879,19 @@ static irqreturn_t i2c_irq_handler(int irq, void *arg)
 #ifdef CONFIG_PM_SLEEP
 static int nmk_i2c_suspend_late(struct device *dev)
 {
-       pinctrl_pm_select_sleep_state(dev);
+       int ret;
 
+       ret = pm_runtime_force_suspend(dev);
+       if (ret)
+               return ret;
+
+       pinctrl_pm_select_sleep_state(dev);
        return 0;
 }
 
 static int nmk_i2c_resume_early(struct device *dev)
 {
-       /* First go to the default state */
-       pinctrl_pm_select_default_state(dev);
-       /* Then let's idle the pins until the next transfer happens */
-       pinctrl_pm_select_idle_state(dev);
-
-       return 0;
+       return pm_runtime_force_resume(dev);
 }
 #endif