Hi Tomasz,
Thanks for the review. My replies are as below.
On Thu, 2020-05-21 at 19:51 +0000, Tomasz Figa wrote:
> Hi Dongchun, Sakari,
>
> On Mon, May 18, 2020 at 09:27:31PM +0800, Dongchun Zhu wrote:
> > Add a V4L2 sub-device driver for DW9768 voice coil motor, providing
> > control to set the desired focus via IIC serial interface.
> >
> > Signed-off-by: Dongchun Zhu <dongchun.zhu@xxxxxxxxxxxx>
> > ---
> > MAINTAINERS | 1 +
> > drivers/media/i2c/Kconfig | 13 ++
> > drivers/media/i2c/Makefile | 1 +
> > drivers/media/i2c/dw9768.c | 515 +++++++++++++++++++++++++++++++++++++++++++++
> > 4 files changed, 530 insertions(+)
> > create mode 100644 drivers/media/i2c/dw9768.c
> [snip]
> > +/*
> > + * DW9768_AAC_PRESC_REG & DW9768_AAC_TIME_REG determine VCM operation time.
> > + * If DW9768_AAC_PRESC_REG is 0x41, and DW9768_AAC_TIME_REG is 0x39, VCM mode
> > + * would be AAC3, Operation Time would be 0.70xTvib, that is 8.40ms.
> > + */
> > +#define DW9768_MOVE_DELAY_US 8400
> > +#define DW9768_STABLE_TIME_US 20000
>
> These times are only valid with the specific settings mentioned in the
> comment. If one sets different settings in DT, the driver would apply
> incorrect delays. Rather than hardcoded, they should be computed based
> on the configured values.
>
> That said, I wonder if we're not digging too deep now. Sakari, do you
> think we could take a step back, remove the optional DT properties and
> just support the fixed values for now, so that we can get a basic driver
> upstream first without doubling the effort?
>
Thanks for the reminder.
Yes, here DW9768_MOVE_DELAY_US actually represents Operation Time,
which is dependent upon board-specific settings that defined in DT.
For instance, for one given board, if aac-mode is 2, aac-timing is 0x39,
clock-presc is 1, then Operation Time would be 0.70*(6.3ms+57*0.1ms)*1 =
8.4ms.
> > +
> > +static const char * const dw9768_supply_names[] = {
> > + "vin", /* I2C I/O interface power */
> > + "vdd", /* VCM power */
> > +};
> > +
> > +/* dw9768 device structure */
> > +struct dw9768 {
> > + struct regulator_bulk_data supplies[ARRAY_SIZE(dw9768_supply_names)];
> > + struct v4l2_ctrl_handler ctrls;
> > + struct v4l2_ctrl *focus;
> > + struct v4l2_subdev sd;
> > +
> > + u32 aac_mode;
> > + u32 aac_timing;
> > + u32 clock_dividing_rate;
> > + bool aac_mode_control_enable;
> > + bool aact_cnt_select_enable;
> > + bool clock_dividing_rate_select_enable;
>
> nit: Separate types from names with just 1 space.
>
Fixed in next release.
> > +};
> > +
> > +static inline struct dw9768 *sd_to_dw9768(struct v4l2_subdev *subdev)
> > +{
> > + return container_of(subdev, struct dw9768, sd);
> > +}
> > +
> > +struct regval_list {
> > + u8 reg_num;
> > + u8 value;
> > +};
> > +
> > +static int dw9768_read_smbus(struct dw9768 *dw9768, unsigned char reg,
> > + unsigned char *val)
> > +{
> > + struct i2c_client *client = v4l2_get_subdevdata(&dw9768->sd);
> > + int ret;
> > +
> > + ret = i2c_smbus_read_byte_data(client, reg);
> > +
> > + if (ret < 0)
> > + return ret;
> > +
> > + *val = (unsigned char)ret;
> > +
> > + return 0;
> > +}
>
> Why do we need this function? Couldn't we just call
> i2c_smbus_read_byte_data() directly?
>
Fixed in next release.
> [snip]
> > +static int dw9768_probe(struct i2c_client *client)
> > +{
> > + struct device *dev = &client->dev;
> > + struct dw9768 *dw9768;
> > + unsigned int aac_mode_select;
> > + unsigned int aac_timing_select;
> > + unsigned int clock_dividing_rate_select;
> > + unsigned int i;
> > + int ret;
> > +
> > + dw9768 = devm_kzalloc(dev, sizeof(*dw9768), GFP_KERNEL);
> > + if (!dw9768)
> > + return -ENOMEM;
> > +
> > + v4l2_i2c_subdev_init(&dw9768->sd, client, &dw9768_ops);
> > + dw9768->aac_mode_control_enable = false;
> > + dw9768->aact_cnt_select_enable = false;
> > + dw9768->clock_dividing_rate_select_enable = false;
>
> devm_kzalloc() initializes the memory to zero, so no need to set anything
> to false explicitly.
>
Thanks for the reminder.
Yes, these parameters shall not be needed to initialized as zeros.
> > +
> > + /* Optional indication of AAC mode select */
> > + ret = fwnode_property_read_u32(dev_fwnode(dev), "dongwoon,aac-mode",
> > + &aac_mode_select);
> > +
> > + if (!ret) {
> > + dw9768->aac_mode_control_enable = true;
> > + dw9768->aac_mode = aac_mode_select;
>
> How about making aac_mode a signed int and assigning -1 by
> default? Then we don't need two separate fields in the struct.
>
Good idea.
> > + }
> > +
> > + /* Optional indication of VCM internal clock dividing rate select */
> > + ret = fwnode_property_read_u32(dev_fwnode(dev),
> > + "dongwoon,clock-dividing-rate",
> > + &clock_dividing_rate_select);
> > +
> > + if (!ret) {
> > + dw9768->clock_dividing_rate_select_enable = true;
> > + dw9768->clock_dividing_rate = clock_dividing_rate_select;
>
> Ditto.
>
Got it.
> > + }
> > +
> > + /* Optional indication of AAC Timing */
> > + ret = fwnode_property_read_u32(dev_fwnode(dev), "dongwoon,aac-timing",
> > + &aac_timing_select);
> > +
> > + if (!ret) {
> > + dw9768->aact_cnt_select_enable = true;
> > + dw9768->aac_timing = aac_timing_select;
>
> Ditto.
>
Got it.
> > + }
> > +
> > + for (i = 0; i < ARRAY_SIZE(dw9768_supply_names); i++)
> > + dw9768->supplies[i].supply = dw9768_supply_names[i];
> > +
> > + ret = devm_regulator_bulk_get(dev, ARRAY_SIZE(dw9768_supply_names),
> > + dw9768->supplies);
> > + if (ret) {
> > + dev_err(dev, "failed to get regulators\n");
> > + return ret;
> > + }
> > +
> > + ret = dw9768_init_controls(dw9768);
> > + if (ret)
> > + goto entity_cleanup;
> > +
> > + dw9768->sd.flags |= V4L2_SUBDEV_FL_HAS_DEVNODE;
> > + dw9768->sd.internal_ops = &dw9768_int_ops;
> > +
> > + ret = media_entity_pads_init(&dw9768->sd.entity, 0, NULL);
> > + if (ret < 0)
> > + goto entity_cleanup;
> > +
> > + dw9768->sd.entity.function = MEDIA_ENT_F_LENS;
> > +
> > + pm_runtime_enable(dev);
> > + if (!pm_runtime_enabled(dev)) {
> > + ret = dw9768_runtime_resume(dev);
> > + if (ret < 0) {
> > + dev_err(dev, "failed to power on: %d\n", ret);
> > + goto entity_cleanup;
> > + }
> > + }
> > +
> > + ret = v4l2_async_register_subdev(&dw9768->sd);
> > + if (ret < 0)
> > + goto entity_cleanup;
> > +
> > + return 0;
> > +
> > +entity_cleanup:
>
> Need to power off if the code above powered on.
>
Thanks for the reminder.
If there is something wrong with runtime PM, actuator is to be powered
on via dw9768_runtime_resume() API.
When actuator sub-device is powered on completely and async registered
successfully, we shall power off it afterwards.
> > + v4l2_ctrl_handler_free(&dw9768->ctrls);
> > + media_entity_cleanup(&dw9768->sd.entity);
> > + return ret;
> > +}
> > +
> > +static int dw9768_remove(struct i2c_client *client)
> > +{
> > + struct v4l2_subdev *sd = i2c_get_clientdata(client);
> > + struct dw9768 *dw9768 = sd_to_dw9768(sd);
> > +
> > + pm_runtime_disable(&client->dev);
>
> First the device must be unregistered from the userspace. Otherwise there
> is a race condition that risks the userspace accessing the device while the
> deinitialization is happening.
>
Fixed in next release by adjusting the sequence of unregistering and
runtime PM disable.
> Best regards,
> Tomasz
