Skip to contentSkip to conversation

watchdog: imx2_wdt: handle enabled watchdog properly

PATCH New Anatolij Gustschin 4 participants 5 messages Apr 4, 2014 – May 26, 2014

Subsystems (2)
Arm i.MX and Layerscape Watchdog device drivers
lore
Share & tools
lore

Anatolij Gustschin

author
#0 lore raw

PATCH 1/1 watchdog: imx2_wdt: handle enabled watchdog properly

If the watchdog is already enabled (e.g. started by the boot loader), we need to handle this case properly and start reloading the watchdog time-out counter as frequently as needed.

Signed-off-by Anatolij Gustschin <[email protected]>
Cc Wolfram Sang <[email protected]>
Cc Wim Van Sebroeck <[email protected]>
---
 drivers/watchdog/imx2_wdt.c |   24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)
drivers/watchdog/imx2_wdt.c +24 -0
index dd51d95..bef9a9f 100644
@@ -54,6 +54,7 @@
 #define IMX2_WDT_DEFAULT_TIME	60		/* in seconds */
 
 #define WDOG_SEC_TO_COUNT(s)	((s * 2 - 1) << 8)
+#define WDOG_COUNT_TO_SEC(c)	((((c) >> 8) + 1) / 2)
 
 #define IMX2_WDT_STATUS_OPEN	0
 #define IMX2_WDT_STATUS_STARTED	1

@@ -120,6 +121,27 @@ static void imx2_wdt_timer_ping(unsigned long arg)
 	mod_timer(&imx2_wdt.timer, jiffies + imx2_wdt.timeout * HZ / 2);
 }
 
+static inline void imx2_wdt_ping_if_active(void)
+{
+	u16 val = __raw_readw(imx2_wdt.base + IMX2_WDT_WCR);
+
+	if (val & IMX2_WDT_WCR_WDE) {
+		/* Must enable the clock, wdog service doesn't work otherwise */
+		clk_prepare_enable(imx2_wdt.clk);
+
+		set_bit(IMX2_WDT_STATUS_STARTED, &imx2_wdt.status);
+		imx2_wdt.timeout = WDOG_COUNT_TO_SEC(val);
+		if (imx2_wdt.timeout < 2) {
+			imx2_wdt.timeout = 2;
+			/* Set new watchdog time-out value */
+			val &= ~IMX2_WDT_WCR_WT;
+			val |= WDOG_SEC_TO_COUNT(imx2_wdt.timeout);
+			__raw_writew(val, imx2_wdt.base + IMX2_WDT_WCR);
+		}
+		imx2_wdt_timer_ping(0);
+	}
+}
+
 static void imx2_wdt_start(void)
 {
 	if (!test_and_set_bit(IMX2_WDT_STATUS_STARTED, &imx2_wdt.status)) {

@@ -283,6 +305,8 @@ static int __init imx2_wdt_probe(struct platform_device *pdev)
 	if (ret)
 		goto fail;
 
+	imx2_wdt_ping_if_active();
+
 	dev_info(&pdev->dev,
 		"IMX2+ Watchdog Timer enabled. timeout=%ds (nowayout=%d)\n",
 						imx2_wdt.timeout, nowayout);
Signature · 1 line
1.7.9.5
#1 lore raw

PATCH 1/1 watchdog: imx2_wdt: handle enabled watchdog properly

Quoted 7 lines · Anatolij Gustschin

If the watchdog is already enabled (e.g. started by the boot loader), we need to handle this case properly and start reloading the watchdog time-out counter as frequently as needed.

Signed-off-by Anatolij Gustschin <[email protected]>
Cc Wolfram Sang <[email protected]>
Cc Wim Van Sebroeck <[email protected]>

Uh, this driver has not been converted to use the watchdog framework yet? I'd suggest this should be the first step.

@Wim: Does the watchdog core already handle the case that the wdt might have been started by the bootloader?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20140404/e7bcaf92/attachment.sig>

maintainer
#2 lore raw

PATCH 1/1 watchdog: imx2_wdt: handle enabled watchdog properly

Quoted 5 lines · Wolfram Sang
Quoted 4 lines · Anatolij Gustschin

If the watchdog is already enabled (e.g. started by the boot loader), we need to handle this case properly and start reloading the watchdog time-out counter as frequently as needed.

Quoted 7 lines
Quoted 3 lines
Signed-off-by Anatolij Gustschin <[email protected]>
Cc Wolfram Sang <[email protected]>
Cc Wim Van Sebroeck <[email protected]>

Uh, this driver has not been converted to use the watchdog framework yet? I'd suggest this should be the first step.

Yes, that would make sense at this time, especially given the amount of work going into it. There is also a separate patch set pending, converting the driver to use regmap [1]; I am copying the author. Maybe one of the people working with the driver would be willing to do the conversion.

Guenter

[1] http://comments.gmane.org/gmane.linux.kernel/1677735

#3 lore raw
In reply to Wolfram Sang

Hi Wolfram,

Quoted 9 lines · Anatolij Gustschin
Quoted 7 lines

If the watchdog is already enabled (e.g. started by the boot loader), we need to handle this case properly and start reloading the watchdog time-out counter as frequently as needed.

Signed-off-by Anatolij Gustschin <[email protected]>
Cc Wolfram Sang <[email protected]>
Cc Wim Van Sebroeck <[email protected]>
Quoted 5 lines

Uh, this driver has not been converted to use the watchdog framework yet? I'd suggest this should be the first step.

@Wim: Does the watchdog core already handle the case that the wdt might have been started by the bootloader?

There is nothing for that in the core. And I think that it makes no sense to put this in the core. It's something you have to take care of during init/probe and that is driver specific. See also this message: http://www.spinics.net/lists/linux-watchdog/msg04066.html .

Kind regards,
Wim.

#4 lore raw
In reply to Wolfram Sang

Hi Wolfram,

Quoted 9 lines · Anatolij Gustschin
Quoted 7 lines

If the watchdog is already enabled (e.g. started by the boot loader), we need to handle this case properly and start reloading the watchdog time-out counter as frequently as needed.

Signed-off-by Anatolij Gustschin <[email protected]>
Cc Wolfram Sang <[email protected]>
Cc Wim Van Sebroeck <[email protected]>
Quoted 5 lines

Uh, this driver has not been converted to use the watchdog framework yet? I'd suggest this should be the first step.

@Wim: Does the watchdog core already handle the case that the wdt might have been started by the bootloader?

No, not yet.

Kind regards,
Wim.