pm selftest: rtc paranoia
David Brownell [Sat, 26 Jul 2008 02:44:38 +0000 (19:44 -0700)]
Cope with a quirk of some RTCs (notably ACPI ones) which aren't guaranteed
to implement oneshot behavior when they woke the system from sleeep:
forcibly disable the alarm, just in case.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/power/main.c

index 95bff23..0b7476f 100644 (file)
@@ -635,6 +635,13 @@ static void __init test_wakealarm(struct rtc_device *rtc, suspend_state_t state)
        }
        if (status < 0)
                printk(err_suspend, status);
+
+       /* Some platforms can't detect that the alarm triggered the
+        * wakeup, or (accordingly) disable it after it afterwards.
+        * It's supposed to give oneshot behavior; cope.
+        */
+       alm.enabled = false;
+       rtc_set_alarm(rtc, &alm);
 }
 
 static int __init has_wakealarm(struct device *dev, void *name_ptr)