Hi Mauro,
Thanks for the review.
On Thu, Nov 05, 2020 at 12:34:13PM +0100, Mauro Carvalho Chehab wrote:
> Em Wed, 7 Oct 2020 11:44:59 +0300
> Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx> escreveu:
...
> > diff --git a/drivers/media/i2c/ccs/ccs-reg-access.c b/drivers/media/i2c/ccs/ccs-reg-access.c
> > index aad2727570ec..9fda16b221a7 100644
> > --- a/drivers/media/i2c/ccs/ccs-reg-access.c
> > +++ b/drivers/media/i2c/ccs/ccs-reg-access.c
> > @@ -236,12 +236,38 @@ int ccs_read_addr_noconv(struct ccs_sensor *sensor, u32 reg, u32 *val)
> > return ccs_read_addr_raw(sensor, reg, val, false, true, false);
> > }
> >
> > +static int ccs_write_retry(struct i2c_client *client, struct i2c_msg *msg)
> > +{
> > + unsigned int retries;
> > + int r;
> > +
> > + for (retries = 0; retries < 5; retries++) {
> > + /*
> > + * Due to unknown reason sensor stops responding. This
> > + * loop is a temporaty solution until the root cause
> > + * is found.
> > + */
> > + r = i2c_transfer(client->adapter, msg, 1);
> > + if (r != 1) {
> > + usleep_range(2000, 2000);
>
> Huh? Does it need to sleep precisely for 2000 us?
I believe this one does not, indeed. I'll make it a range, but add a
separate patch for it as this patch just moves the code form one place to
another.
--
Sakari Ailus
