On Sun, 31 Jan 2021 18:28:40 +0100, Krzysztof Kozlowski wrote:

On Sat, Jan 30, 2021 at 05:30:14PM +0000, Timon Baetz wrote:

quoted

Get regulator from parent device's node and extcon by name.

Signed-off-by: Timon Baetz <redacted>
---
 drivers/power/supply/max8997_charger.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/drivers/power/supply/max8997_charger.c b/drivers/power/supply/max8997_charger.c
index 321bd6b8ee41..625d8cc4312a 100644
--- a/drivers/power/supply/max8997_charger.c
+++ b/drivers/power/supply/max8997_charger.c

@@ -168,6 +168,7 @@ static int max8997_battery_probe(struct platform_device *pdev)
 	int ret = 0;
 	struct charger_data *charger;
 	struct max8997_dev *iodev = dev_get_drvdata(pdev->dev.parent);
+	struct device_node *np = pdev->dev.of_node;
 	struct i2c_client *i2c = iodev->i2c;
 	struct max8997_platform_data *pdata = iodev->pdata;
 	struct power_supply_config psy_cfg = {};

@@ -237,20 +238,23 @@ static int max8997_battery_probe(struct platform_device *pdev)
 		return PTR_ERR(charger->battery);
 	}

+	// grab regulator from parent device's node
+	pdev->dev.of_node = iodev->dev->of_node;
 	charger->reg = devm_regulator_get_optional(&pdev->dev, "charger");
+	pdev->dev.of_node = np;  

I think the device does not have its own node anymore. Or did I miss
something?

The idea is to reset of_node to whatever it was before (NULL) and basically 
leave the device unchanged. Probe might run again because of deferral.

quoted

 	if (IS_ERR(charger->reg)) {
 		if (PTR_ERR(charger->reg) == -EPROBE_DEFER)
 			return -EPROBE_DEFER;
 		dev_info(&pdev->dev, "couldn't get charger regulator\n");
 	}
-	charger->edev = extcon_get_edev_by_phandle(&pdev->dev, 0);
-	if (IS_ERR(charger->edev)) {
-		if (PTR_ERR(charger->edev) == -EPROBE_DEFER)
+	charger->edev = extcon_get_extcon_dev("max8997-muic");
+	if (IS_ERR_OR_NULL(charger->edev)) {
+		if (!charger->edev)  

Isn't NULL returned when there is simply no extcon? It's different than
deferred probe. Returning here EPROBE_DEFER might lead to infinite probe
tries (on every new device probe) instead of just failing it.

extcon_get_extcon_dev() just loops through all registered extcon devices
and compared names. It will return NULL when "max8997-muic" isn't
registered yet. extcon_get_extcon_dev() never returns EPROBE_DEFER so
checking for NULL seems to be the only way. Other drivers using that
function also do NULL check and return EPROBE_DEFER.

Thanks for reviewing,
Timon


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel