Hi Andy, On Fri, 2020-10-23 at 17:31 +0300, Andy Shevchenko wrote: > On Tue, Oct 13, 2020 at 09:05:03PM +0800, Dongchun Zhu wrote: > > Add a V4L2 sub-device driver for OmniVision OV02A10 image sensor. > > ... > > > +#define OV02A10_ID_MASK 0xffff > > GENMASK() > > (And include bits.h for that) > It seems could build pass without including bits.h, as DW9768 once used. > ... > > > +static int __ov02a10_start_stream(struct ov02a10 *ov02a10) > > +{ > > + struct i2c_client *client = v4l2_get_subdevdata(&ov02a10->subdev); > > + const struct ov02a10_reg_list *reg_list; > > + int ret; > > + > > + /* Apply default values of current mode */ > > + reg_list = &ov02a10->cur_mode->reg_list; > > + ret = ov02a10_write_array(ov02a10, reg_list); > > + if (ret) > > + return ret; > > + > > + /* Apply customized values from user */ > > + ret = __v4l2_ctrl_handler_setup(ov02a10->subdev.ctrl_handler); > > + if (ret) > > + return ret; > > + > > + /* Set orientation to 180 degree */ > > + if (ov02a10->upside_down) { > > + ret = i2c_smbus_write_byte_data(client, REG_MIRROR_FLIP_CONTROL, > > + REG_MIRROR_FLIP_ENABLE); > > + if (ret) { > > Shouldn't you use 'ret < 0' here as well? > Fixed in next release. > > + dev_err(&client->dev, "failed to set orientation\n"); > > + return ret; > > + } > > + ret = i2c_smbus_write_byte_data(client, REG_GLOBAL_EFFECTIVE, > > + REG_ENABLE); > > + if (ret < 0) > > + return ret; > > + } > > + > > + /* Set MIPI TX speed according to DT property */ > > + if (ov02a10->mipi_clock_voltage != OV02A10_MIPI_TX_SPEED_DEFAULT) { > > + ret = i2c_smbus_write_byte_data(client, TX_SPEED_AREA_SEL, > > + ov02a10->mipi_clock_voltage); > > + if (ret < 0) > > + return ret; > > + } > > + > > + /* Set stream on register */ > > + return i2c_smbus_write_byte_data(client, REG_SC_CTRL_MODE, > > + SC_CTRL_MODE_STREAMING); > > +} > > ... > > > +/* > > Was your intention to declare it as a kernel doc? > Removed in next release. > > + * ov02a10_set_exposure - Function called when setting exposure time > > + * @priv: Pointer to device structure > > + * @val: Variable for exposure time, in the unit of micro-second > > + * > > + * Set exposure time based on input value. > > + * > > + * Return: 0 on success > > + */ > > +static int ov02a10_set_exposure(struct ov02a10 *ov02a10, int val) > > ... > > > +static int ov02a10_check_hwcfg(struct device *dev, struct ov02a10 *ov02a10) > > +{ > > + struct fwnode_handle *ep; > > + struct fwnode_handle *fwnode = dev_fwnode(dev); > > + struct v4l2_fwnode_endpoint bus_cfg = { > > + .bus_type = V4L2_MBUS_CSI2_DPHY, > > + }; > > + unsigned int i, j; > > + int ret; > > > + if (!fwnode) > > + return -EINVAL; > > Basically you can avoid this check, but it's up to you. > I'd like to keep it. Thank you. > > + ep = fwnode_graph_get_next_endpoint(fwnode, NULL); > > + if (!ep) > > + return -ENXIO; > > + > > + ret = v4l2_fwnode_endpoint_alloc_parse(ep, &bus_cfg); > > + fwnode_handle_put(ep); > > + if (ret) > > + return ret; > > + > > + for (i = 0; i < ARRAY_SIZE(link_freq_menu_items); i++) { > > + for (j = 0; j < bus_cfg.nr_of_link_frequencies; j++) { > > + if (link_freq_menu_items[i] == > > + bus_cfg.link_frequencies[j]) { > > + ov02a10->freq_index = i; > > + break; > > + } > > + } > > + > > + if (j == bus_cfg.nr_of_link_frequencies) { > > + dev_err(dev, "no link frequency %lld supported\n", > > + link_freq_menu_items[i]); > > + ret = -EINVAL; > > + break; > > + } > > + } > > + > > + v4l2_fwnode_endpoint_free(&bus_cfg); > > + > > + return ret; > > +} > > ... > > > + fwnode_property_read_u32(dev_fwnode(dev), "rotation", &rotation); > > Same Q as per previous reviews. Why device property API can't be used here? > > And everywhere else when you have > fwnode_property_read_*(dev_fwnode(dev), ...) > calls. > Thanks for the reminder. 'fwnode_property_read_u32' and 'fwnode_property_read_u32_array' would be replaced in next release if local test passes.