On 01/08/2014 05:16 PM, Arnd Bergmann wrote: > On Wednesday 08 January 2014 16:58:15 Roger Quadros wrote: >> hpriv->phy = devm_phy_get(dev, "sata-phy"); >> if (IS_ERR(hpriv->phy)) { >> if (PTR_ERR(hpriv->phy) == -ENODEV) >> goto continue; >> >> dev_err(dev, "couldn't get sata-phy\n"); >> rc = PTR_ERR(hpriv->phy); >> goto disable_unprepare_clk; >> } >> >> continue: >> >> if (!IS_ERR(hpriv->phy)) { >> rc = phy_init(hpriv->phy); >> if (rc) >> goto disable_unprepare_clk; >> >> rc = phy_power_on(hpriv->phy); >> if (rc) { >> phy_exit(hpriv->phy); >> goto disable_unprepare_clk; >> } >> } > > As I said, I'd prefer to set hpriv->phy to NULL in case of -ENODEV, > but functionally it seems right (with the fixup from your other mail). > Why do you prefer setting hpriv->phy to NULL instead of using IS_ERR() check before hpriv->phy is used? The latter seems to be the norm at least among clock framework users. > One more comment: you shouldn't print an error message before > returning -EPROBE_DEFER, so if you want to keep that message, > you have to check for that return value after all. > Right. cheers, -roger -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html