From: Matti Vaittinen <mazziesaccount@gmail.com>

When watchdog is enabled at the probe time, the bd96801 driver retrieves
the timeout configuration from the registers to set-up the heart-beat
values.

As Sashiko pointed out at
https://lore.kernel.org/all/20260722085819.495211F000E9@smtp.kernel.org/
the timeout values are incorrectly computed in driver, resulting wrong
heartbeat. This leads to devere problems if watchdog was enabled at probe
time.

According to the data-sheet, the "too fast" ping limit is configured as
multiple of FASTNG_MIN. Furthermore, the "too slow" ping limit is
configured as multiples of "too fast" timeout. The FASTNG_MIN is set to
11, meaning 1.1 mS and "too fast" and "too slow" limits are computed from
this. Hence, converting the limits to mS should be done by dividing by 10,
not by dividing by USEC_PER_MSEC.

Fix this by dividing the timeout values with correct scaling factor.

Signed-off-by: Matti Vaittinen <mazziesaccount@gmail.com>
Fixes: 09dad69757b6 ("watchdog: ROHM BD96801 PMIC WDG driver")
---

Also, drop an extra whitespace while editing the file - feeling like
that single whitespace fix doesn't warrant own patch. Just let me know if
I feel wrong ;)

 drivers/watchdog/bd96801_wdt.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/watchdog/bd96801_wdt.c b/drivers/watchdog/bd96801_wdt.c
index 12b74fd2bc05..a25b7cf1488b 100644
--- a/drivers/watchdog/bd96801_wdt.c
+++ b/drivers/watchdog/bd96801_wdt.c
@@ -169,7 +169,6 @@ static int bd96801_set_wdt_mode(struct wdtbd96801 *w, unsigned int hw_margin,
 	int fastng, slowng, type, ret, reg, mask;
 	struct device *dev = w->dev;
 
-
 	if (hw_margin_min * 1000 > FASTNG_MAX_US) {
 		dev_err(dev, "Unsupported fast timeout %u uS [max %u]\n",
 			hw_margin_min * 1000, FASTNG_MAX_US);
@@ -258,10 +257,10 @@ static int bd96801_set_heartbeat_from_hw(struct wdtbd96801 *w,
 	fast = FASTNG_MIN << sel;
 
 	sel = (val & BD96801_WD_RATIO_MASK) + 1;
-	w->wdt.max_hw_heartbeat_ms = (fast << sel) / USEC_PER_MSEC;
+	w->wdt.max_hw_heartbeat_ms = (fast << sel) / 10;
 
 	if ((conf_reg & BD96801_WD_TYPE_MASK) == BD96801_WD_TYPE_WIN)
-		w->wdt.min_hw_heartbeat_ms = fast / USEC_PER_MSEC;
+		w->wdt.min_hw_heartbeat_ms = fast / 10;
 
 	return 0;
 }

base-commit: dc59e4fea9d83f03bad6bddf3fa2e52491777482
-- 
2.55.0

On Fri, Jul 31, 2026 at 12:36:28PM +0300, Matti Vaittinen wrote:
> From: Matti Vaittinen <mazziesaccount@gmail.com>
> 
> When watchdog is enabled at the probe time, the bd96801 driver retrieves
> the timeout configuration from the registers to set-up the heart-beat
> values.
> 
> As Sashiko pointed out at
> https://lore.kernel.org/all/20260722085819.495211F000E9@smtp.kernel.org/
> the timeout values are incorrectly computed in driver, resulting wrong
> heartbeat. This leads to devere problems if watchdog was enabled at probe
> time.
> 
> According to the data-sheet, the "too fast" ping limit is configured as
> multiple of FASTNG_MIN. Furthermore, the "too slow" ping limit is
> configured as multiples of "too fast" timeout. The FASTNG_MIN is set to
> 11, meaning 1.1 mS and "too fast" and "too slow" limits are computed from
> this. Hence, converting the limits to mS should be done by dividing by 10,
> not by dividing by USEC_PER_MSEC.
> 
> Fix this by dividing the timeout values with correct scaling factor.
> 
> Signed-off-by: Matti Vaittinen <mazziesaccount@gmail.com>
> Fixes: 09dad69757b6 ("watchdog: ROHM BD96801 PMIC WDG driver")
> ---
> 
> Also, drop an extra whitespace while editing the file - feeling like
> that single whitespace fix doesn't warrant own patch. Just let me know if
> I feel wrong ;)

No problem (since you let me know and didn't do it silently).

Applied.

Thanks,
Guenter