On Mon, Dec 27, 2021 at 01:19:25PM +0100, Lars-Peter Clausen wrote: > On 12/27/21 10:45 AM, Uwe Kleine-König wrote: > > [...] > > - return devm_counter_register(dev, &priv->counter); > > + err = devm_counter_add(dev, counter); > > + if (err < 0) > > + return dev_err_probe(dev, err, "Failed to add counter\n"); > I wonder if we should put that dev_err_probe into the devm_counter_add since > every driver wants to have it anyway. Personally I'm not a big fan of API functions that emit an error message. Usually the consumer knows best if a certain failing function call is worth an error message. Look at platform_get_irq() that some time ago got a dev_err call. The result was to introduce a silent variant (platform_get_irq_optional()) because emitting an error message isn't the right thing in all cases. Also the API function probably won't call device_set_deferred_probe_reason() and so the consumer has to care for that anyhow (or not which makes -EPROBE_DEFER problems harder to debug). Furthermore the consumer can emit in some cases a better error message than the core. (Well, this doesn't apply here, but in the example of platform_get_irq() the driver knows the irq's purpose and can call it "TX irq" in the error message which the irq core cannot.) Best regards Uwe -- Pengutronix e.K. | Uwe Kleine-König | Industrial Linux Solutions | https://www.pengutronix.de/ |
Attachment:
signature.asc
Description: PGP signature
