pmbus_write_smbalert_mask() ignores the errors if the chip can't set
smbalert mask the standard way. It is not necessarily a problem for the irq
support if the chip is otherwise properly setup but it may leave an
uncleared fault behind.

pmbus_core will pick the fault on the next register_check(). The register
check will fails regardless of the actual register support by the chip.

This leads to missing attributes or debugfs entries for chips that should
provide them.

We cannot rely on register_check() as PMBUS_SMBALERT_MASK may be read-only.

Unconditionally clear the page fault after setting PMBUS_SMBALERT_MASK to
avoid the problem.

Suggested-by: Guenter Roeck <linux@roeck-us.net>
Fixes: 221819ca4c36 ("hwmon: (pmbus/core) Add interrupt support")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
---
 drivers/hwmon/pmbus/pmbus_core.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/drivers/hwmon/pmbus/pmbus_core.c b/drivers/hwmon/pmbus/pmbus_core.c
index d355e3fb0d6b7bea392c7dd5551a1c904a05f21b..55167e195e2a22154ae8fee693169d6f0829ffca 100644
--- a/drivers/hwmon/pmbus/pmbus_core.c
+++ b/drivers/hwmon/pmbus/pmbus_core.c
@@ -3346,7 +3346,17 @@ static int pmbus_regulator_notify(struct pmbus_data *data, int page, int event)
 
 static int pmbus_write_smbalert_mask(struct i2c_client *client, u8 page, u8 reg, u8 val)
 {
-	return _pmbus_write_word_data(client, page, PMBUS_SMBALERT_MASK, reg | (val << 8));
+	int ret;
+
+	ret = _pmbus_write_word_data(client, page, PMBUS_SMBALERT_MASK, reg | (val << 8));
+
+	/*
+	 * Clear fault systematically in case writing PMBUS_SMBALERT_MASK
+	 * is not supported by the chip.
+	 */
+	pmbus_clear_fault_page(client, page);
+
+	return ret;
 }
 
 static irqreturn_t pmbus_fault_handler(int irq, void *pdata)

-- 
2.45.2

On Tue, Nov 05, 2024 at 06:58:42PM +0100, Jerome Brunet wrote:
> pmbus_write_smbalert_mask() ignores the errors if the chip can't set
> smbalert mask the standard way. It is not necessarily a problem for the irq
> support if the chip is otherwise properly setup but it may leave an
> uncleared fault behind.
> 
> pmbus_core will pick the fault on the next register_check(). The register
> check will fails regardless of the actual register support by the chip.
> 
> This leads to missing attributes or debugfs entries for chips that should
> provide them.
> 
> We cannot rely on register_check() as PMBUS_SMBALERT_MASK may be read-only.
> 
> Unconditionally clear the page fault after setting PMBUS_SMBALERT_MASK to
> avoid the problem.
> 
> Suggested-by: Guenter Roeck <linux@roeck-us.net>
> Fixes: 221819ca4c36 ("hwmon: (pmbus/core) Add interrupt support")
> Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>

Applied.

Thanks,
Guenter