lkml.org 
[lkml]   [2022]   [Jul]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH 3/3] media: i2c: ak7375: Add regulator management

On Mon, Jul 11 2022 at 15:34:23 +0200, Jacopo Mondi <jacopo@jmondi.org>
wrote:
> Hello Yassine
>
> On Mon, Jul 11, 2022 at 08:28:39AM +0400, Yassine Oudjana wrote:
>> From: Yassine Oudjana <y.oudjana@protonmail.com>
>>
>> Make the driver get needed regulators on probe and enable/disable
>> them on runtime PM callbacks.
>>
>> Signed-off-by: Yassine Oudjana <y.oudjana@protonmail.com>
>> ---
>> drivers/media/i2c/ak7375.c | 39
>> ++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 39 insertions(+)
>>
>> diff --git a/drivers/media/i2c/ak7375.c b/drivers/media/i2c/ak7375.c
>> index 40b1a4aa846c..59d5cb00e3ba 100644
>> --- a/drivers/media/i2c/ak7375.c
>> +++ b/drivers/media/i2c/ak7375.c
>> @@ -6,6 +6,7 @@
>> #include <linux/i2c.h>
>> #include <linux/module.h>
>> #include <linux/pm_runtime.h>
>> +#include <linux/regulator/consumer.h>
>> #include <media/v4l2-ctrls.h>
>> #include <media/v4l2-device.h>
>>
>> @@ -23,17 +24,32 @@
>> */
>> #define AK7375_CTRL_STEPS 64
>> #define AK7375_CTRL_DELAY_US 1000
>> +/*
>> + * The vcm takes around 3 ms to power on and start taking
>> + * I2C messages. This value was found experimentally due to
>> + * lack of documentation. 2 ms is added as a safety margin.
>> + */
>> +#define AK7375_POWER_DELAY_US 5000
>>
>> #define AK7375_REG_POSITION 0x0
>> #define AK7375_REG_CONT 0x2
>> #define AK7375_MODE_ACTIVE 0x0
>> #define AK7375_MODE_STANDBY 0x40
>>
>> +static const char * const ak7375_supply_names[] = {
>> + "vdd",
>> + "vio",
>> +};
>> +
>> +#define AK7375_NUM_SUPPLIES ARRAY_SIZE(ak7375_supply_names)
>> +
>> /* ak7375 device structure */
>> struct ak7375_device {
>> struct v4l2_ctrl_handler ctrls_vcm;
>> struct v4l2_subdev sd;
>> struct v4l2_ctrl *focus;
>> + struct regulator_bulk_data supplies[AK7375_NUM_SUPPLIES];
>> +
>> /* active or standby mode */
>> bool active;
>> };
>> @@ -132,6 +148,7 @@ static int ak7375_init_controls(struct
>> ak7375_device *dev_vcm)
>> static int ak7375_probe(struct i2c_client *client)
>> {
>> struct ak7375_device *ak7375_dev;
>> + int i;
>
> I would have moved this one down to maintain variable declaration
> in the in-famous reverse-xmas-tree ordering. Up to you.

I'm used to declaring variables in the order of first use,
but I don't really mind it either way. I'll move it down.

>
>> int ret;
>>
>> ak7375_dev = devm_kzalloc(&client->dev, sizeof(*ak7375_dev),
>> @@ -139,6 +156,17 @@ static int ak7375_probe(struct i2c_client
>> *client)
>> if (!ak7375_dev)
>> return -ENOMEM;
>>
>> + for (i = 0; i < AK7375_NUM_SUPPLIES; i++)
>> + ak7375_dev->supplies[i].supply = ak7375_supply_names[i];
>> +
>> + ret = devm_regulator_bulk_get(&client->dev, AK7375_NUM_SUPPLIES,
>> + ak7375_dev->supplies);
>> + if (ret) {
>> + dev_err(&client->dev, "Failed to get regulators: %pe",
>> + ERR_PTR(ret));
>> + return ret;
>> + }
>> +
>> v4l2_i2c_subdev_init(&ak7375_dev->sd, client, &ak7375_ops);
>> ak7375_dev->sd.flags |= V4L2_SUBDEV_FL_HAS_DEVNODE;
>> ak7375_dev->sd.internal_ops = &ak7375_int_ops;
>> @@ -210,6 +238,10 @@ static int __maybe_unused
>> ak7375_vcm_suspend(struct device *dev)
>> if (ret)
>> dev_err(dev, "%s I2C failure: %d\n", __func__, ret);
>>
>> + ret = regulator_bulk_disable(AK7375_NUM_SUPPLIES,
>> ak7375_dev->supplies);
>> + if (ret)
>> + return ret;
>> +
>> ak7375_dev->active = false;
>>
>> return 0;
>> @@ -230,6 +262,13 @@ static int __maybe_unused
>> ak7375_vcm_resume(struct device *dev)
>> if (ak7375_dev->active)
>> return 0;
>>
>> + ret = regulator_bulk_enable(AK7375_NUM_SUPPLIES,
>> ak7375_dev->supplies);
>> + if (ret)
>> + return ret;
>> +
>> + /* Wait for vcm to become ready */
>> + usleep_range(AK7375_POWER_DELAY_US, AK7375_POWER_DELAY_US + 10);
>> +
>
> Isn't 10usec a very small delay to be given to usleep_range() for a
> delay of at least 3msec ? Also assuming 5msec just to be safe seems a
> little arbitrary. Adding 2 milliseconds in the wakeup path introduces
> a non-negligible delay.

I must admit that I didn't give it too much thought. I just
did it similar to the other delay used in this driver
(AK7375_CTRL_DELAY_US). As for adding 2ms, I don't know what
the worst case wake-up time is since I don't have a datasheet
on hand, so I just wanted to stay safe. Also, this driver
doesn't really recover if it fails to resume (which is what
used to happen before adding a delay). Rounding up to 5ms
felt good enough.

>
> It's likely a detail, but according to
> Documentation/timers/timers-howto.rst
>
> Since usleep_range is built on top of hrtimers, the
> wakeup will be very precise (ish), thus a simple
> usleep function would likely introduce a large number
> of undesired interrupts.
>
> With the introduction of a range, the scheduler is
> free to coalesce your wakeup with any other wakeup
> that may have happened for other reasons, or at the
> worst case, fire an interrupt for your upper bound.
>
> The larger a range you supply, the greater a chance
> that you will not trigger an interrupt; this should
> be balanced with what is an acceptable upper bound on
> delay / performance for your specific code path. Exact
> tolerances here are very situation specific, thus it
> is left to the caller to determine a reasonable range.
>
> If you have a min of 3msec I would try with a range of (3000, 3500).
> What do you think ?

Seems good. I haven't yet had it fail to power on within 3ms of
turning on regulators so I guess there is no reason to worry about it.

>>
>> ret = ak7375_i2c_write(ak7375_dev, AK7375_REG_CONT,
>> AK7375_MODE_ACTIVE, 1);
>> if (ret) {
>> --
>> 2.37.0
>

Thanks for the review,
Yassine

\
 
 \ /
  Last update: 2022-07-11 16:09    [W:0.070 / U:0.732 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site