[PATCH] watchdog/max63xx: be careful when disabling the watchdog

[Marc Zyngier]

Wim,

The attached patch fixes a nasty error with the max63xx driver,
potentially leaving the platform dead (and watchdog disabled...).

Cheers,

        M.
-- 
Who you jivin' with that Cosmik Debris?From 4212d84a08845a6adf3f67367a24d9dbf076f9d6 Mon Sep 17 00:00:00 2001
From: Marc Zyngier <maz@xxxxxxxxxxxxxxx>
Date: Fri, 9 Apr 2010 17:43:33 +0100
Subject: [PATCH] watchdog/max63xx: be careful when disabling the watchdog

When shutting down the watchdog timer, special care must be taken
not to overwrite other bits in the register, as it may be shared
with other peripherals.

For example, on the Arcom Vulcan, the register is shared between
the watchdog and the PCI reset line...

Signed-off-by: Marc Zyngier <maz@xxxxxxxxxxxxxxx>
---
 drivers/watchdog/max63xx_wdt.c |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/drivers/watchdog/max63xx_wdt.c b/drivers/watchdog/max63xx_wdt.c
index 75f3a83..3053ff0 100644
--- a/drivers/watchdog/max63xx_wdt.c
+++ b/drivers/watchdog/max63xx_wdt.c
@@ -154,9 +154,14 @@ static void max63xx_wdt_enable(struct max63xx_timeout *entry)
 
 static void max63xx_wdt_disable(void)
 {
+	u8 val;
+
 	spin_lock(&io_lock);
 
-	__raw_writeb(3, wdt_base);
+	val = __raw_readb(wdt_base);
+	val &= ~MAX6369_WDSET;
+	val |= 3;
+	__raw_writeb(val, wdt_base);
 
 	spin_unlock(&io_lock);
 
-- 
1.7.0