Skip to contentSkip to conversation

regulator: pwm: Don't warn on probe deferral

PATCH Superseded Jon Hunter 4 participants 7 messages Feb 24, 2020 – Mar 6, 2020

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Jon Hunter

author maintainer
#0 lore raw

PATCH 1/1 regulator: pwm: Don't warn on probe deferral

Deferred probe is an expected return value for devm_pwm_get(). Given that the driver deals with it properly, there's no need to output a warning that may potentially confuse users.

---
 drivers/regulator/pwm-regulator.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/regulator/pwm-regulator.c +2 -1
index e74e11101fc1..fb25777a7d47 100644
@@ -354,7 +354,8 @@ static int pwm_regulator_probe(struct platform_device *pdev)
 	drvdata->pwm = devm_pwm_get(&pdev->dev, NULL);
 	if (IS_ERR(drvdata->pwm)) {
 		ret = PTR_ERR(drvdata->pwm);
-		dev_err(&pdev->dev, "Failed to get PWM: %d\n", ret);
+		if (ret != -EPROBE_DEFER)
+			dev_err(&pdev->dev, "Failed to get PWM: %d\n", ret);
 		return ret;
 	}
 
Signature · 1 line
2.17.1
Reviewed-by on v1 1/1
#1 lore raw
Quoted 8 lines · Jon Hunter

Deferred probe is an expected return value for devm_pwm_get(). Given that the driver deals with it properly, there's no need to output a warning that may potentially confuse users.

---
 drivers/regulator/pwm-regulator.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
maintainer
#2 lore raw
In reply to Jon Hunter

drivers/regulator/pwm-regulator.c static int pwm_regulator_probe(struct platform_device *pdev) · line 358

Quoted 8 lines · Jon Hunter

Deferred probe is an expected return value for devm_pwm_get(). Given that the driver deals with it properly, there's no need to output a warning that may potentially confuse users.

drivers/regulator/pwm-regulator.c +2 -1
 		ret = PTR_ERR(drvdata->pwm);
-		dev_err(&pdev->dev, "Failed to get PWM: %d\n", ret);
+		if (ret != -EPROBE_DEFER)
+			dev_err(&pdev->dev, "Failed to get PWM: %d\n", ret);

This then means that there's no way for users to determine why the driver has failed to instantiate which can be frustrating. It'd be better to at least have some dev_dbg() output when deferring so that there's something for people to go on without having to instrument the code.

#3 lore raw
Quoted 5 lines · Mark Brown
Quoted 3 lines · Jon Hunter

Deferred probe is an expected return value for devm_pwm_get(). Given that the driver deals with it properly, there's no need to output a warning that may potentially confuse users.

Quoted 10 lines

drivers/regulator/pwm-regulator.c static int pwm_regulator_probe(struct platform_device *pdev) · line 358

Quoted 4 lines
drivers/regulator/pwm-regulator.c +2 -1
 		ret = PTR_ERR(drvdata->pwm);
-		dev_err(&pdev->dev, "Failed to get PWM: %d\n", ret);
+		if (ret != -EPROBE_DEFER)
+			dev_err(&pdev->dev, "Failed to get PWM: %d\n", ret);

This then means that there's no way for users to determine why the driver has failed to instantiate which can be frustrating. It'd be better to at least have some dev_dbg() output when deferring so that there's something for people to go on without having to instrument the code.

Not printing an error message is quite usual however. I think a generic approach that for example makes the list of devices that should be retried to probe on the next opportunity inspectable would be nice.

Best regards
Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |
maintainer
#4 lore raw
Quoted 9 lines · Uwe Kleine-König
Quoted 5 lines · Mark Brown

This then means that there's no way for users to determine why the driver has failed to instantiate which can be frustrating. It'd be better to at least have some dev_dbg() output when deferring so that there's something for people to go on without having to instrument the code.

Not printing an error message is quite usual however. I think a generic

Usual yet also frustraing.

Quoted 2 lines

approach that for example makes the list of devices that should be retried to probe on the next opportunity inspectable would be nice.

That's not really the issue, the bigger issue is trying to figure out why things are stuck - what exactly caused things to fail to instantiate.

#5 lore raw
Quoted 11 lines · Mark Brown
Quoted 9 lines · Uwe Kleine-König
Quoted 5 lines · Mark Brown

This then means that there's no way for users to determine why the driver has failed to instantiate which can be frustrating. It'd be better to at least have some dev_dbg() output when deferring so that there's something for people to go on without having to instrument the code.

Not printing an error message is quite usual however. I think a generic

Quoted 8 lines

Usual yet also frustraing.

Quoted 2 lines

approach that for example makes the list of devices that should be retried to probe on the next opportunity inspectable would be nice.

That's not really the issue, the bigger issue is trying to figure out why things are stuck - what exactly caused things to fail to instantiate.

I wonder if we should do something like:

	ret = some_call(some, args);
	if (ret) {
		if (emit_errmsg_for_err(ret))
			dev_err(dev, "some_call failed: %pE\n", ERR_PTR(ret));
		return ret;
	}

and have emit_errmsg_for_err return true if ret != -EPROBE_DEFER or some kernel parameter is given.

Best regards
Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |
maintainer
#6 lore raw
Quoted 11 lines · Uwe Kleine-König

I wonder if we should do something like:

	ret = some_call(some, args);
	if (ret) {
		if (emit_errmsg_for_err(ret))
			dev_err(dev, "some_call failed: %pE\n", ERR_PTR(ret));
		return ret;
	}

and have emit_errmsg_for_err return true if ret != -EPROBE_DEFER or some kernel parameter is given.

There was some effort in the past to have a dev_probe_err() or something which could have a similar implementation but that didn't end up going anywhere I think. I do prefer the debug level log since it's much easier to have the information there without having to ask for it, that design would've supported that.