Hi Jacopo,
On 02/01/23 18:18, Jacopo Mondi wrote:
Hi Jay
On Fri, Dec 30, 2022 at 02:33:41PM +0530, Luthra, Jai wrote:
Hi Jacopo,
Thank you for the comments.
On 29/12/22 23:37, Jacopo Mondi wrote:
Hi Jai
On Tue, Dec 27, 2022 at 11:06:33PM +0530, Jai Luthra wrote:
Move the register-based reset out of the init_setting[] and into the
powerup_sequence function. The sensor is power cycled and reset using
the gpio pins so the soft reset is not always necessary.
This also ensures that soft reset honors the timing sequence
from the datasheet [1].
[1] https://cdn.sparkfun.com/datasheets/Sensors/LightImaging/OV5640_datasheet.pdf
Fixes: 19a81c1426c1 ("[media] add Omnivision OV5640 sensor driver")
Suggested-by: Jacopo Mondi <jacopo.mondi@xxxxxxxxxxxxxxxx>
Signed-off-by: Jai Luthra <j-luthra@xxxxxx>
---
drivers/media/i2c/ov5640.c | 30 ++++++++++++++++++++----------
1 file changed, 20 insertions(+), 10 deletions(-)
diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
index e6525caa9b8c..5df16fb6f0a0 100644
--- a/drivers/media/i2c/ov5640.c
+++ b/drivers/media/i2c/ov5640.c
@@ -50,6 +50,7 @@
#define OV5640_REG_SYS_CTRL0 0x3008
#define OV5640_REG_SYS_CTRL0_SW_PWDN 0x42
#define OV5640_REG_SYS_CTRL0_SW_PWUP 0x02
+#define OV5640_REG_SYS_CTRL0_SW_RST 0x82
#define OV5640_REG_CHIP_ID 0x300a
#define OV5640_REG_IO_MIPI_CTRL00 0x300e
#define OV5640_REG_PAD_OUTPUT_ENABLE01 0x3017
@@ -532,7 +533,7 @@ static const struct v4l2_mbus_framefmt ov5640_default_fmt = {
};
static const struct reg_value ov5640_init_setting[] = {
- {0x3103, 0x11, 0, 0}, {0x3008, 0x82, 0, 5}, {0x3008, 0x42, 0, 0},
+ {0x3103, 0x11, 0, 0}, {0x3008, 0x42, 0, 0},
have you tried removing the second entry
{0x3008, 0x42, 0, 0},
as well ?
The SW_PWDN state allows programming registers (how can you exit with
SW_PWUP otherwise?), so it's probably right to have it there.
Yes, I initially removed it from here and kept it in the powerup_sequence()
but logic to set SW_PWDN became in that function became a bit complicated:
When we are using HW pins, we want to wait 20ms after de-asserting RESET
before doing any SCCB register writes. But when we are using SW reset, we
were using this register write to de-assert RESET, and then wait 20ms.
I found it cleaner to just keep it here where the registers are being
programmed for both the cases. But I don't have a strong opinion either way.
{0x3103, 0x03, 0, 0}, {0x3630, 0x36, 0, 0},
{0x3631, 0x0e, 0, 0}, {0x3632, 0xe2, 0, 0}, {0x3633, 0x12, 0, 0},
{0x3621, 0xe0, 0, 0}, {0x3704, 0xa0, 0, 0}, {0x3703, 0x5a, 0, 0},
@@ -2440,18 +2441,27 @@ static void ov5640_power(struct ov5640_dev *sensor, bool enable)
*/
static void ov5640_powerup_sequence(struct ov5640_dev *sensor)
{
- gpiod_set_value_cansleep(sensor->reset_gpio, 0);
+ if (sensor->pwdn_gpio) {
+ gpiod_set_value_cansleep(sensor->reset_gpio, 0);
- /* camera power cycle */
- ov5640_power(sensor, false);
- usleep_range(5000, 10000);
- ov5640_power(sensor, true);
- usleep_range(5000, 10000);
+ /* camera power cycle */
+ ov5640_power(sensor, false);
+ usleep_range(5000, 10000);
+ ov5640_power(sensor, true);
+ usleep_range(5000, 10000);
- gpiod_set_value_cansleep(sensor->reset_gpio, 1);
- usleep_range(1000, 2000);
+ gpiod_set_value_cansleep(sensor->reset_gpio, 1);
+ usleep_range(1000, 2000);
- gpiod_set_value_cansleep(sensor->reset_gpio, 0);
+ gpiod_set_value_cansleep(sensor->reset_gpio, 0);
+ } else {
+ /* software reset */
+ ov5640_write_reg(sensor, OV5640_REG_SYS_CTRL0,
+ OV5640_REG_SYS_CTRL0_SW_RST);
+ usleep_range(5000, 10000);
+ ov5640_write_reg(sensor, OV5640_REG_SYS_CTRL0,
+ OV5640_REG_SYS_CTRL0_SW_PWUP);
But now I wonder if this last PWUP is necessary, since init_settings[]
(which programs PWDN) is programmed immediately after.
Using the PWUP (0x02) as a way to de-assert RST (0x42) here, and not as an
RST is 0x82 if I read it right
Oops made a typo, yes RST is 0x82.
actual PWUP per se.
It is not necessary. The RST cleans up itself when the chip has done
resetting. I tested it with:
@@ -2463,12 +2478,24 @@ static void ov5640_powerup_sequence(struct ov5640_dev *sensor)
gpiod_set_value_cansleep(sensor->reset_gpio, 0);
} else {
+ int ret;
+ u16 sysctrl = 0;
+ u16 counter = 0;
+
/* software reset */
ov5640_write_reg(sensor, OV5640_REG_SYS_CTRL0,
OV5640_REG_SYS_CTRL0_SW_RST);
- usleep_range(5000, 10000);
- ov5640_write_reg(sensor, OV5640_REG_SYS_CTRL0,
- OV5640_REG_SYS_CTRL0_SW_PWUP);
+
+ do {
+ usleep_range(2000, 2500);
+ ret = ov5640_read_reg16(sensor, OV5640_REG_SYS_CTRL0,
+ &sysctrl);
+ if (ret)
+ pr_err("Failed to read sysctrl: %d\n", ret);
+ } while (sysctrl & BIT(7) || ++counter > 100);
+
+ pr_err("Exit from reset in %u loops\n", counter);
}
usleep_range(20000, 25000);
The loop exists after 1 iteration
So I guess you can remove the last SW_PWUP write
Great that simplifies things, thanks for that quick test!
+ }
usleep_range(20000, 25000);
Now that I think of it, we can move the PWDN from the register table to
here, after this sleep - and keep the (0x02) above to de-assert reset. Let
me know your thoughts.
Yes, I think it might be a good idea.
The diff on top of your series should be like
------------------------------------------------------------------------------
-- a/drivers/media/i2c/ov5640.c
+++ b/drivers/media/i2c/ov5640.c
@@ -533,7 +533,7 @@ static const struct v4l2_mbus_framefmt ov5640_default_fmt = {
};
static const struct reg_value ov5640_init_setting[] = {
- {0x3103, 0x11, 0, 0}, {0x3008, 0x42, 0, 0},
+ {0x3103, 0x11, 0, 0},
{0x3103, 0x03, 0, 0}, {0x3630, 0x36, 0, 0},
{0x3631, 0x0e, 0, 0}, {0x3632, 0xe2, 0, 0}, {0x3633, 0x12, 0, 0},
{0x3621, 0xe0, 0, 0}, {0x3704, 0xa0, 0, 0}, {0x3703, 0x5a, 0, 0},
@@ -2466,11 +2466,12 @@ static void ov5640_powerup_sequence(struct ov5640_dev *sensor)
/* software reset */
ov5640_write_reg(sensor, OV5640_REG_SYS_CTRL0,
OV5640_REG_SYS_CTRL0_SW_RST);
- usleep_range(5000, 10000);
- ov5640_write_reg(sensor, OV5640_REG_SYS_CTRL0,
- OV5640_REG_SYS_CTRL0_SW_PWUP);
}
usleep_range(20000, 25000);
+
+ /* Software standby: allows registers programming; exit at s_stream(1) */
+ ov5640_write_reg(sensor, OV5640_REG_SYS_CTRL0,
+ OV5640_REG_SYS_CTRL0_SW_PWDN);
}
static int ov5640_set_power_on(struct ov5640_dev *sensor)
------------------------------------------------------------------------------
Do you agree with this ?
Also, since you have to resend, could you please swap the order of
patches 1 and 2 ? It's more logically correct to plumb the SW reset
handling in, then allow it to happen as you do in patch 1/3.
Makes sense, I will move the SW PWDN and re-order the patches in v4.
}
--
2.17.1
Thanks,
Jai
Thanks,
Jai
