Thank you for the review, Dan. Some remarks below. On Mon, Dec 9, 2019 at 9:12 AM Dan Murphy <dmurphy@xxxxxx> wrote: > > > + priv->fwnode = device_get_next_child_node(pdev->dev.parent, NULL); > > Probably need to check for NULL on the return > The driver will work without crashes or oopses even when this returns NULL: - struct led_init_data . fwnode is optional (can be NULL) - fwnode_handle_put() ignores NULL arguments By not checking for NULL here, non-devicetree users can still select led mode through platform data on the parent mfd driver, and things will "just work". Could I persuade you to keep this behaviour? Perhaps I should put a comment in to clarify? > > + ret = regmap_update_bits(tps6105x->regmap, TPS6105X_REG_0, > > + TPS6105X_REG0_MODE_MASK | TPS6105X_REG0_TORCHC_MASK, > > + TPS6105X_REG0_MODE_TORCH << TPS6105X_REG0_MODE_SHIFT); > Checkpatch should have warned about alignment here I used 5.4's checkpatch.pl, but somehow it doesn't warn :( Will fix that up.